Intermolecular condensation of ethylenediamine to 1,4-diazabicyclo[2,2,2]octane over TS-1 catalysts

Article abstract of DOI:10.1016/j.jcat.2009.06.016

The intermolecular condensation of ethylenediamine (EDA) to 1,4-diazabicyclo[2.2.2]octane or triethylenediamine (TEDA) has been carried out over various titanosilicate catalysts. Superior to Ti-MWW, Ti-Beta, Ti-FER, and Ti-MOR, TS-1 showed higher EDA conversion and TEDA selectivity. The effects of reaction parameters, Ti content, and crystal size on the EDA condensation over TS-1 have been investigated. The mechanism for the TS-1-catalyzed condensation of EDA has also been considered. The acid sites, originated from the Si-OH groups adjacent to the "open" Ti sites, were assumed to contribute to the intermolecular condensation of EDA, whereas the Lewis acid sites directly related to Ti(IV) ions were not the true active sites. The primary intermolecular condensation of EDA to 1,4-diazacyclohexane or piperazine (PIP) took place mainly inside the micropores of the MFI structure, while the secondary condensation of PIP with EDA to TEDA was favored by the acid sites located near the pore entrance and on the outer surface of crystals.

Full text of DOI:10.1016/j.jcat.2009.06.016

Journal of Catalysis 266 (2009) 258–267
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Intermolecular condensation of ethylenediamine to 1,4-diazabicyclo[2,2,2]octane
over TS-1 catalysts
*
*
Yong Wang, Yueming Liu , Xiaohong Li, Haihong Wu, 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:
The intermolecular condensation of ethylenediamine (EDA) to 1,4-diazabicyclo[2.2.2]octane or triethy-
lenediamine (TEDA) has been carried out over various titanosilicate catalysts. Superior to Ti-MWW, Ti-
Beta, Ti-FER, and Ti-MOR, TS-1 showed higher EDA conversion and TEDA selectivity. The effects of reac-
tion parameters, Ti content, and crystal size on the EDA condensation over TS-1 have been investigated.
The mechanism for the TS-1-catalyzed condensation of EDA has also been considered. The acid sites, orig-
inated from the Si–OH groups adjacent to the ‘‘open” Ti sites, were assumed to contribute to the intermo-
lecular condensation of EDA, whereas the Lewis acid sites directly related to Ti(IV) ions were not the true
active sites. The primary intermolecular condensation of EDA to 1,4-diazacyclohexane or piperazine (PIP)
took place mainly inside the micropores of the MFI structure, while the secondary condensation of PIP
with EDA to TEDA was favored by the acid sites located near the pore entrance and on the outer surface
of crystals.
Received 1 February 2009
Revised 21 May 2009
Accepted 14 June 2009
Available online 16 July 2009
Keywords:
Ethylenediamine
Triethylenediamine
Piperazine
Condensation
Titanosilicate
TS-1
Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction
their unique shape selectivity and suitable acidic properties. Nev-
ertheless, it is still not well addressed for the reaction mechanism
1,4-Diazabicyclo[2.2.2]octane, known also as triethylenedi-
amine (TEDA), is one of the principal catalysts used for producing
polyurethanes from polyisocyanates and polyols [1]. It also acts
as anti-fade reagent that scavenges the free radicals produced by
excitation of fluorochromes. It is also usually added to the mount-
ing medium in fluorescence microscopy to retard photobleaching
of fluorescein and other fluorescent dyes. In addition, TEDA serves
as useful structure-directing agent in zeolite synthesis [2].
The known processes for the preparation of TEDA essentially
differ in the nature of the starting materials and the catalysts. Early
routes to synthesize TEDA involve the vapor-phase condensation of
substituted piperazines (PIP) such as aminoethylpiperazine, N-(2-
hydroxyethyl)piperazine, and N,N-di(2-hydroxyethyl)piperazine
over alumina catalysts [3,4], and the liquid-phase reaction in the
presence of aromatic carboxylic acid catalysts [5]. Recently, the re-
searches are reported mainly for the conversion of monoethanola-
mine, ethylenediamine (EDA), and PIP or their mixture over
modified ZSM-5 zeolites [6–8]. The synthesis of TEDA from mono-
ethanolamine over modified MCM-41 catalyst has also been re-
ported [9]. In addition, the preparation of TEDA via the addition
of EDA and PIP using TS-1 catalyst has been patented [10]. The pre-
vious studies disclose that the MFI type zeolites are suitable cata-
lysts for intermolecular condensation of EDA to TEDA because of
on zeolites, and the parameters, particularly the nature of zeolite
acidity, which determine the TEDA selectivity remain unclear.
TS-1, being able to catalyze efficiently the liquid-phase oxida-
tion of a variety of organic compounds to oxygenated products,
has brought about a breakthrough in the field of zeolite catalysis
in the past two decades [11–13]. However, the application of TS-
1 as an acid catalyst is seldom reported. The gas-phase Beckmann
rearrangement is one of the acid-catalyzed reactions catalyzed by
TS-1 [14]. In addition, TS-1 is shown to catalyze efficiently the
transesterification of linear esters such as ethylacetoacetate and
diethylmalonate [15].
Several studies have mentioned the acid properties of TS-1, but
two different points of view have been debated for a long time [16].
Earlier study reported that TS-1 contained neither Brønsted nor Le-
wis acid sites [17]. On the other hand, TS-1 was inferred to be more
acidic in comparison to Ti-free Silicalite-1 [18]. In addition, the sig-
nificant activity given by TS-1 in the cycloaddition of CO2 with
epoxides to cyclic carbonates [19], a reaction typically catalyzed
by Lewis acids such as AlCl3, implies the Lewis acidity of TS-1. With
respect to the Brønsted acidity of TS-1, very few reports give clear
evidence for the generation of new and detectable OH groups
exhibiting protic acidity after incorporating Ti in the silicalite lat-
tice. 31P and 1H MAS NMR spectra of adsorbed trimethylphosphine
verified that TS-1 contained also Brønsted acid sites in addition to
Lewis acid sites [20], and also suggested that the incorporation of
Ti into the framework led to the formation of new OH groups, tita-
nols, which may be even more acidic than the silanols of Silicalite-1.
* Corresponding authors. Fax: +86 21 6223 2292.
E-mail addresses: ymliu@chem.ecnu.edu.cn (Y. Liu), pwu@chem.ecnu.edu.cn (P.
Wu).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.06.016

Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
259
Because of indetermination of its acidity, it is still encouraged to
2.2. Catalytic reactions
carry out further study on the acidic nature of TS-1 and on how to
make full use of its functions in practical acid-catalyzed reactions.
In this study, we have applied TS-1 to the intermolecular con-
densation of EDA to investigate its effectiveness in producing
chemicals of PIP and TEDA. After comparing the catalytic perfor-
mance of TS-1 with that of other titanosilicates and optimizing
the reaction parameters for selective synthesis of TEDA, the active
sites for the EDA conversion have been correlated to TS-1’s acidity.
The issue of reaction space’s contributions to the consecutive con-
densation of EDA to PIP and finally to TEDA has also been
addressed.
The intermolecular condensation of EDA was carried out in a
down-flow fixed-bed reactor at atmospheric pressure. In a typical
run, 1 g of catalyst (18–30 mesh size) was activated at 673 K for
3 h in a flow of N2. After cooling to the reaction temperature, the
reaction was started by feeding an aqueous EDA solution (50 wt%
EDA in H2O) into the reactor at a flow rate of 5 mL hꢀ1 together
with a nitrogen flow of 30 mL minꢀ1. The products were collected
in a receiver cooled with an ice-salt bath. The amounts of the prod-
ucts were quantified on a Shimadzu 14B gas chromatograph
equipped with an OV-1 column (30 m) and an FID detector, and
they were identified using authentic samples or by GC–MS (Agilent
6890).
2. Experimental
3. Results
2.1. Catalyst preparation and characterizations
3.1. Characterization of various titanosilicate catalysts
TS-1 with different Si/Ti ratios was hydrothermally synthesized
using tetrabutyl orthotitanate (TBOT), tetraethylorthosilicate
(TEOS), and tetrapropylammonium hydroxide (TPAOH, 20% aque-
ous solution) by modifying the procedures reported elsewhere
[21]. Typically, TEOS was mixed and stirred vigorously with a
desirable amount of TBOT in a polyethylene flask at ambient tem-
perature for 15 min. The clear solution obtained was then added
dropwise to the alkali-free aqueous solution of TPAOH under stir-
ring, which may avoid the formation of TiO2 phase such as anatase.
The mixture was stirred at room temperature for about 3 hours to
hydrolyze TEOS and TBOT completely. And it was further heated at
353 K under stirring to evaporate the alcohols, resulting in the
clear gels having the molar compositions of 1 SiO2:(0.005 –
0.033) TiO2:0.15 TPAOH:15 H2O. The gels were charged into Tef-
lon-lined stainless-steel autoclaves and crystallized at 443 K for 2
days under autogenous pressure and static conditions. The solid
product was separated by centrifugation, washed thoroughly with
distilled water, dried at 383 K overnight, and calcined in air at 823
K for 10 h to burn off the occluded organic species. For control
experiment, Ti-free Silicalite-1 and other titanosilicates such as
Ti-FER [22], Ti-MOR [23,24], Al-free Ti-Beta [25], and Ti-MWW
[26,27] were also prepared. The TS-1 samples with larger crystal
sizes were hydrothermally synthesized using fumed silica (Cab-
o-sil) or colloidal silica and tetrapropylammonium bromide
(TPABr) as silicon source and structure-directing agent (SDA) at
Si/Ti of 60 according to the Refs. [28,29].
All titanosilicates had objective zeolite structures and relatively
high crystallinity, and contained no impurity of other zeolite phase
as detected by XRD patterns (Fig. 1). In UV–visible and IR spectra,
the samples showed the characteristic adsorption bands at 210 nm
and 960 cmꢀ1, respectively, which are assigned to the tetrahedral
Ti species isolated in the zeolite framework [11,12,26]. These tita-
nosilicates had reasonably high surface area (Langmuir) in the
range of 430–653 m2 gꢀ1 as determined by N2 adsorption (Table 1).
The TS-1 samples synthesized using TEOS and TPAOH at various
Si/Ti ratios were further characterized. The samples contained
essentially the isolated Ti species in the framework as evidenced
by the predominant band at 210 nm in UV–visible spectra when
the Si/Ti ratio was higher than 60 (Fig. 2a–d). Nevertheless, the
band due to the anatase-like Ti species, although not intense, was
also observed in addition to the 210-nm band for the samples syn-
thesized at Si/Ti ratios of 40 and 30 (Fig. 2e and f), which is indica-
tive of the presence of a small amount of extraframework Ti. The IR
spectra of TS-1 in the framework vibration region showed the char-
acteristic band at 960 cmꢀ1 assigned to the framework Ti, and the
band intensity increased with increasing Ti content (not shown).
Thus, both IR and UV–visible spectra verified that the TS-1 catalysts
were of good quality in terms of incorporating the tetrahedrally
Sodium ion-exchanged samples were prepared by stirring 1 g of
calcined TS-1 in 100 mL of 1 M NaNO3 or 1 M Na2CO3 solution at
room temperature for 24 h. And the removal of sodium ions was
carried out by washing with 1 M HNO3 solution at room
temperature.
All of the catalysts were characterized by inductively coupled
plasma (ICP) on a Thermo IRIS Intrepid II XSP atomic emission
spectrometer, X-ray diffraction (XRD) on a Bruker D8 ADVANCE
e
d
diffractometer (Cu-Ka), N2 adsorption on an Autosorb Quanca-
chrome (02108-KR-1), scanning electron microscopy (SEM) on a
JEOL JSM-T220 (Hitachi S-4800), and UV–visible (Shimadzu UV-
2400PC) and IR (Nicolet NEXUS-FTIR-670) spectroscopies. The
amount of sodium was determined by atomic adsorption spec-
troscopy (AAS). IR spectra of pyridine adsorption were recorded
as follows: a self-supported wafer (9.6 mg cmꢀ2 thickness and
£ 2 cm) was set in a quartz IR cell sealed with CaF2 windows,
where it was evacuated at 773 K for 2 h before the pyridine
adsorption. The adsorption was carried out by exposing the wafer
to a pyridine vapor (1.3 kPa) at 323 K for 0.5 h. The physisorbed
pyridine was then removed by evacuation at different tempera-
tures (323–470 K) for 1 h. All the spectra were collected at room
temperature.
c
b
a
5 10 15 20 25 30 35
2Theta (deg.)
Fig. 1. Representative XRD patterns of TS-1 (60) (a), Ti-Beta (37) (b), Ti-MWW (55)
(c), Ti-MOR (90) (d), and Ti-FER (52) (e). The number in the parentheses represents
the Si/Ti molar ratio.

260
Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
Table 1
The result of intermolecular condensation of EDA over different titanosilicate catalysts.a
No.
Catalyst
Si/Tib ratio
SAc (m2 gꢀ1
)
Main pores
EDA conv. (%)
Product sel. (%)
PIP
TEDA
Othersd
1
TS-1
60
37
37
90
55
52
530
653
653
553
610
430
3D 10-MR
2D 12-MR
2D 12-MR
1D 12-MR
2D 10-MR (12-MR cup, supercage)
1D 10-MR
84.4
41.1
80.5
30.5
35.0
25.0
31
45
10
48
58
43
51
15
45
18
7
18
40
45
34
35
20
2
Ti-Beta
Ti-Beta
Ti-MOR
Ti-MWW
Ti-FER
3e
4
5
6
37
a
b
c
Reaction conditions: catalyst, 1 g; pressure, 1 atm; temp.: 623 K; EDA feed (50 wt% aqueous solution), 5 mL hꢀ1; N2, 30 mL minꢀ1; TOS: 3 h.
Given by ICP.
Specific surface area (Langmuir) given by N2 adsorption at 77 K.
Mainly lower aliphatic amines and the derivatives of pyrazine.
Reaction conditions: N2, 10 mL minꢀ1; others, same as a.
d
e
coordinated Ti ions. The scanning electron microscopy (SEM) re-
vealed that TS-1 was composed of 300–400 nm spherical crystals,
the size of which was independent of the Ti content (Fig. 3a and
b). On the other hand, TS-1 with larger crystals was synthesized
when using other silicon source or SDA. The TS-1 sample obtained
from fumed silica and TPAOH consisted of the spherical crystals of
210
f
e
d
ca. 20
lm (Fig. 3c), while the sample prepared using colloidal silica
c
and TPABr as SDA showed a morphology of 30 ꢁ 8 ꢁ 2
l
m plate
crystals (Fig. 3d). Nevertheless, UV–visible spectra showed that
both large crystal samples contained the isolated tetrahedral Ti spe-
cie but no anatase phase.
330
b
a
On the basis of the crystallinity, surface area, and the nature of
Ti species, these titanosilicates qualified as the catalysts for inter-
molecular condensation of EDA.
3.2. Intermolecular condensation of EDA over various titanosilicates
200
300
400
500
According to the mechanism reported previously [6] and the
products actually obtained, a general reaction pathway showing
the product distribution of intermolecular condensation of EDA
was obtained (Scheme 1). The major products were PIP and TEDA,
Wavelength (nm)
Fig. 2. UV–visible spectra of TS-1 synthesized at Si/Ti ratio of 1 (a), 200 (b), 100 (c),
60 (d), 40 (e), and 30 (f).
Fig. 3. SEM images of TS-1 samples synthesized with TEOS and TPAOH at Si/Ti of 60 (a), 40 (b), synthesized with fumed silica and TPAOH (c), and with silica sol and TPABr (d)
both at Si/Ti of 60.

Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
261
N
N
N
N
N
Pyrazine
derivatives
+
+
N
-NH3
N
N
Pyrazine
H2N
CH3CH2NH2 + CH3NH2
Degradation
-3H2
H
N
N
NH2
NH2
H2N
H2N
-2NH3
H2N
+
-2NH3
N
N
H
EDA
TEDA
PIP
Scheme 1.
while the byproducts were lower aliphatic amines (ethylamine and
methylamine) and the derivatives of pyrazine. Intermolecular dea-
mino-condensation of two EDA molecules led to the formation of
PIP, which further condensed with one more EDA molecule to give
TEDA. Meanwhile, PIP partially underwent dehydrogenation to
form pyrazine that may further react with the products from
EDA degradation to produce the corresponding derivatives, alkyl
pyrazines.
Table 1 shows the comparison of the results of intermolecular
condensation of EDA among various titanosilicate catalysts. The
products were mainly PIP and TEDA together with some lower ali-
phatic amines and the pyrazine derivatives. The EDA conversion
and product selectivity remained almost unchangeable during
10 h of time on stream (TOS) investigated. Under optimized condi-
tions (the details will be given in later sections), TS-1 showed the
highest EDA conversion and more importantly, the highest total
selectivity to PIP and TEDA (Table 1, No. 1). Other titanosilicates
were not only less active, but also less selective to PIP and TEDA
even when compared at equal conversion (Table 1, Nos. 2–6).
Therefore, the catalytic properties of TS-1 were then further stud-
ied by investigating the effects of various reaction parameters.
100
80
60
40
20
0
EDA conv.
TEDA
PIP
Others
580 600 620 640 660
Temp. (K)
Fig. 4. Effect of reaction temperature on the intermolecular condensation of EDA on
TS-1 (Si/Ti = 60). Other reaction conditions: see Table 1.
3.3. Effects of reaction parameters on the intermolecular condensation
of EDA over TS-1
3.3.1. Effect of reaction temperature
crease in selectivity to nearly zero, and the formation of byprod-
ucts was enhanced obviously. At high temperatures, the Lewis
acidic sites may be formed due to the dehydroxylation, which then
contributes to the formation of pyrazine derivatives [30]. More-
over, the experiment indicated that the degradation of EDA to
the aliphatic amines occurs more easily at higher temperatures.
This may also make the selectivity to PIP and TEDA decrease. The
reaction at the temperatures below 623 K does not favor the
formation of pyrazine and its derivatives, but is suitable for the for-
mation of the main products of PIP and TEDA [7]. However, too low
The reaction temperature exhibited a great effect on the inter-
molecular condensation of EDA over TS-1(60). As shown in Fig. 4,
the conversion of EDA increased with increasing temperature,
whereas the selectivity to PIP decreased monotonously and the
TEDA selectivity increased first and then decreased slightly. The
TEDA selectivity reached a maximum value of 51% at 623 K, where
the EDA conversion and the PIP selectivity were 84% and 31%,
respectively. When the temperature was further raised to 673 K,
EDA was 100% converted, whereas PIP showed an exponential de-

262
Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
temperatures are energetically inefficient for converting EDA to PIP
and further to TEDA. Consequently, the foregoing experiments
showed that the intermolecular condensation of EDA to TEDA pro-
ceeded most effectively at an optimum reaction temperature of
623 K at catalyst loading of 1 g, EDA feeding rate of 5 mL hꢀ1 and
80
60
40
20
0
EDA conv.
N2 flow rate of 30 mL minꢀ1
.
TEDA
3.3.2. Effect of concentration of EDA
Fig. 5 shows the effect of the concentration of EDA on the reac-
tion at 623 K. TS-1 showed the highest EDA conversion at 10 wt%
of aqueous EDA. The conversion was comparably high in the range
of 20–90% concentration, but the product selectivity varied greatly.
At 50% concentration of EDA, the catalyst gave the highest selectiv-
ity to TEDA (51%) as well as the highest total selectivity to PIP and
TEDA (82%). Lowering the EDA concentration, the selectivity of PIP
and TEDA decreased greatly owing to the formation of byproducts.
At higher concentration of EDA, the conversion of EDA was still
higher than 80%, the TEDA selectivity decreased, and the catalyst
turned to deactivate at 3 h of TOS. Meanwhile, the effluent became
very thick because of high concentration of heavy products having
high boiling points, and the catalyst deactivated more rapidly.
Steam is assumed to have the function of refreshing the catalytic ac-
tive sites by washing out the basic product molecules adsorbed
strongly on the acid sites inside zeolite channels or on crystal sur-
face. This is then helpful for reducing the coke formation. Thus, a
moderate concentration of EDA is preferable to achieve a high
EDA conversion and high selectivity to PIP and TEDA.
PIP
Others
0.0
0.4
0.8
1.2
1.6
ln(1+W/F)
Fig. 6. Effect of contact time on the intermolecular condensation of EDA on TS-1 (Si/
Ti = 60). Other reaction conditions: see in Table 1.
100
EDA conv.
80
3.3.3. Effect of contact time
The contact time expressed as catalyst weight to flow (W/F) ra-
tio was varied in a wide range to determine what were the primary
products in the intermolecular condensation of EDA (Fig. 6). The
reactions were carried out by varying the catalyst amount at
623 K with an EDA (50 wt%) feed rate of 5 mL hꢀ1. The conversion
of EDA and selectivity to TEDA increased with increasing W/F ratio,
while the selectivity to PIP decreased. Apparently, the intermolec-
ular condensation of EDA was a typical consecutive reaction,
namely, PIP was primarily produced immediately after the contact
of reactant molecules with the catalyst bed, and it was successively
converted to TEDA as a result of secondary condensation with EDA
in deep catalyst bed. Meanwhile, a long contact time made the side
reactions take place easily, leading to a high selectivity for byprod-
ucts such as lower aliphatic amines and pyrazine derivatives.
TEDA
60
PIP
40
20
Others
0
0
10
20
30
Ti/(Si+Ti) molar ratio (*10-3
)
100
Fig. 7. Dependence of EDA conversion and product selectivity on the Ti content of
TS-1. Other reaction conditions: see Table 1.
EDA conv.
80
3.3.4. Effect of Ti content
A series of TS-1 catalysts with Si/Ti ratios of 30–200 were pre-
pared, and applied to the intermolecular condensation of EDA to-
gether with Silicalite-1 (Fig. 7). Silicalite-1 free of Ti but
containing silanol groups was not active and selective to the reac-
tion, giving only 19.5% EDA conversion and 7.5% TEDA selectivity.
Both the EDA conversion and TEDA selectivity increased reason-
ably with increasing Ti content, but reached a ceiling at a Ti con-
tent of 0.25 mmol gꢀ1 (corresponding to Si/Ti = 40). This is
probably due to the presence of a part of anatase phase in the high
Ti content samples as evidenced by UV–visible spectra (Fig. 2).
60
40
20
0
TEDA
PIP
Others
4. Discussion
20
40
60
80
4.1. The effect of zeolite structure on the EDA conversion
EDA concentration (wt%)
Table 1 shows that the catalytic performance depended greatly
on the titanosilicates. Except for Ti-MOR, the titanosilicates
Fig. 5. Effect of concentration of EDA on the intermolecular condensation of EDA on
TS-1 (Si/Ti = 60). Reaction conditions: see Table 1.

Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
263
employed in the reaction had comparable Si/Ti ratio. The Ti content
3741
3520
is thus presumed not to account for the greatly different catalytic
properties observed between TS-1 and the other catalysts. The
advantage of TS-1 must originate from the medium porosity of
10-MR windows and the unique hydrophobic nature of the MFI
topology. These results are fully consistent with the findings ob-
served on the aluminosilicate zeolites, among which modified
ZSM-5 was reported to be the most shape-selective and effective
catalyst for the conversion of PIP to TEDA [8].
e
d
c
In the case of Ti-Beta and Ti-MOR, the selectivity to TEDA de-
creased, while the alkylpyrazine byproducts were more copro-
duced. This is because the 12-MR large pores of the *BEA and
MOR topologies are not narrow enough to suppress the alkylation
of pyrazine to the corresponding derivatives with large molecular
sizes (Table 1, Nos. 2 and 4). Ti-Beta, containing a large quantity
of stacking faults related to its polymorphic crystalline structure
and then exhibiting a highly hydrophilic feature, reasonably
showed a lower EDA conversion with the reaction operated under
steaming conditions. The low activity of Ti-MOR is considered to be
related to its one-dimensional 12-MR channels not suitable for the
diffusion and desorption of bulky molecules such as TEDA. Hence,
both Ti-MOR and Ti-Beta were not active and selective for the syn-
thesis of TEDA. Ti-FER imposed more significant diffusion limita-
tion for the product molecules practically because of its one-
dimensional 10-MR channels with an elliptical shape and smaller
window than that of MFI, and then showed the lowest conversion
(Table 1, No. 6). In addition, Ti-MWW was composed of two inde-
pendent interlayer and intralayer 10-MR channels both of which
are two-dimensional but slightly smaller than that of TS-1. The
pore entrance of the MWW structure may impose a serious steric
hindrance to the cyclic products, which made Ti-MWW less active
than TS-1 (Table 1, No. 5). Moreover, the open reaction spaces of
the MWW structure, namely, the 12-MR side cups on the exterior
surface and the 12-MR supercages embedded in interlayer 10-MR
channels, made Ti-MWW less shape-selective either. Above results
indicated that only TS-1 with the MFI topology and medium poros-
ity is possible to give an outstanding catalytic activity and TEDA
selectivity not shared by the other zeolite structure in the intermo-
lecular condensation of EDA.
b
a
3725 3505
3800 3600 3400 3200
Wavenumber (cm-1
)
Fig. 8. IR spectra in hydroxyl stretching region of dehydrated TS-1 synthesized at
Si/Ti ratio of 1 (a), 200 (b), 60 (c), 40 (d), and 30 (e).
lite crystals. The observed behavior is in agreement with previous
studies which report that Silicalite-1 contains many structural de-
fects while the incorporation of tetrahedral Ti ions tends to mend
these defects to reduce their number [32]. In addition, the
3676 cmꢀ1 band, once ascribed to the titanol groups [33], was
not distinctive probably because the stretching of Ti–OH group
was far less intense than silanol groups, or its vibration mode did
not give a sharp enough band to be observable.
Pyridine was then used as a probe molecule to provide more de-
tailed informations of acid site’s types, strength, and amount. Fig. 9
shows the IR spectra of adsorbed pyridine in the range of pyridine
ring-stretching modes after desorption at various temperatures.
Since the pyridine adsorption did not cause apparent change in
the region of hydroxyl stretching vibration, the corresponding
spectra thus are not shown for concision. As shown in Fig. 9A, Sil-
icalite-1 showed the vibrations of stretching modes of hydrogen-
bonded (hb) and physically (ph) adsorbed pyridine at 1599 cmꢀ1
(hb, mode 8a), 1589 cmꢀ1 (ph, mode 8a), 1581 cmꢀ1 (hb, ph, mode
8 b), 1483 cmꢀ1 (hb, mode 19a), 1446 cmꢀ1 (hb, mode 19b), and
1440 cmꢀ1 (ph, mode 19b) [34–36]. The bands at 1589, 1483,
and 1440 cmꢀ1 associated with physically adsorbed pyridine
nearly diminished after desorption at 323 K. The bands at 1599
and 1446 cmꢀ1 associated with hydrogen-bonded pyridine were
more resistant against evacuation, but decreased in intensity with
rising desorption temperature, and disappeared totally at 473 K.
Thus, the acidity related to the silanols in Silicalite-1 was not
strong. On the other hand, the band at 1490 cmꢀ1 characteristic
of both Brønsted and Lewis acids was absent in the spectra of Sil-
icalite-1 regardless of desorption temperature, suggesting it was
almost free of Lewis acid sites.
4.2. Investigation into the active sites for EDA condensation on TS-1
Fig. 7 shows that the EDA conversion and the TEDA selectivity
depended closely on the Ti content of TS-1. The results implied that
the catalytic active sites for the EDA condensation to TEDA related
to the framework Ti species. In solid acid-catalyzed reactions, the
framework Ti species of TS-1 can serve as Lewis acid sites. For in-
stance, TS-1 shows a significant activity for the cycloaddition of
CO2 to epoxides to give cyclic carbonates [19]. On the other hand,
the neural, moderately acidic, isolated surface silanol groups are
reported to be responsible for the gas-phase Beckmann arrange-
ment of cyclohexanone oxime [14].
IR spectroscopy has been adopted to investigate the acidity of
TS-1. Fig. 8 shows the IR spectra in the region of hydroxyl stretch-
ing vibration of TS-1 and Silicalite-1 after dehydration at 773 K. Sil-
icalite-1 showed a sharp band at 3725 cmꢀ1 together with a broad
one at 3500 cmꢀ1 (Fig. 8a), which are attributed to the location of
terminal silanol internal to zeolite pores and the hydrogen-bonded
silanol nests, respectively [31], suggesting the presence of a rela-
tively high concentration of defect sites in Silicalite-1. With
increasing Ti content, the TS-1 samples showed a progressive de-
crease in the intensity of the broad band which simultaneously
shifted from 3505 to 3520 cmꢀ1. The main band in the left side,
on the other hand, gradually became sharper and projected at
3741 cmꢀ1 when the Si/Ti ratio was below 60. The 3741 cmꢀ1 band
is attributed to the terminal silanol on the external surface of zeo-
In the spectra of TS-1 (Fig. 9B), we observed two distinct bands
at 1604 and 1490 cmꢀ1 and a shoulder one at 1450 cm
(super-
ꢀ1
imposed to those of hydrogen-bonded pyridine) in addition to sim-
ilar bands shown by Silicalite-1. These new components absent for
Silicalite-1 were more evacuation temperature resistant, and re-
mained even after desorption at 473 K. The bands at 1604 and
1490 cmꢀ1, apparently related to the Ti species incorporated into
the matrix of silicalite, are assigned to the pyridine-ring vibration
modes of 8a and 19a, respectively [36,37]. The C5H5Nꢂ ꢂ ꢂTi(IV) ad-
duct with pyridine molecule (ligand) coordinated to the Ti(IV)
ion (center) may contribute to these bands. These two bands can
be taken as the evidence for the presence of Lewis acid sites with

264
Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
1446
B
A
1446
1440
1604
1599
1599
1440
1450
1490
1589
1581
1581
1483
1483
a
a
b
c
d
e
b
c
d
e
1600 1550 1500 1450 1400
Wavenumber (cm-1
1600 1550 1500 1450 1400
Wavenumber (cm -1
)
)
Fig. 9. IR spectra of Silicalite-1 (Si/Ti = 1) (A) and TS-1 (Si/Ti = 40) (B) after pyridine adsorption without evacuation (a), and after evacuation at 323 K (b), 373 K (c), 423 K (d),
and 473 K (e) for 1 h, respectively.
a relatively high strength in TS-1. The spectra did not show an
obvious band at 1550 cmꢀ1 which is assigned to the vibration of
pyridium ion formed by protonation of pyridine ring, and com-
monly used as an evidence for the presence of Brønsted acid sites.
Nevertheless, the 1446 cmꢀ1 band due to hydrogen-bonded pyri-
dine was still visible after desorption at 473 K (Fig. 9B, e). TS-1 is
thus presumed to possess relatively stronger acid sites than Silica-
lite-1. These acid sites are related to the hydroxyl groups adjacent
to the framework Ti.
Pyridine adsorption was further carried out on the TS-1 samples
with various Ti contents. Fig. 10 shows the comparison of the spec-
tra of the samples after removing weakly adsorbed species by
evacuation at 473 K. The bands at 1604, 1490, and 1446 cmꢀ1, ab-
sent for Silicalite-1, increased in intensity with increasing Ti con-
tent. Thus, the introduction of tetrahedrally coordinated Ti
created in TS-1 the Lewis acid sites that exhibited the former
two bands ascribed to coordinated pyridine species (the contribu-
tion to the 1490 cmꢀ1 band of the pyridium ions generated from
chemisorbed pyridine on Brønsted acid sites would be negligible
for weak solid acid of TS-1), and also developed the hydroxyl
groups showing the latter one due to hydrogen-bonded species.
It is deduced that the TS-1 catalysts contained somewhat strong
Lewis acid sites and hydroxyl groups with a moderate acidity (at
least tolerant to the pyridine desorption at 473 K), whereas Silica-
lite-1 contained only the neutral silanols.
The above results are in agreement to the characterizations and
catalysis reported for TS-1. Silanols as well as titanols possibly
present in TS-1 are reported to act potentially as weak acid sites
[38]. 31P MAS NMR spectroscopy using trimethylphosphine as ba-
sic probe molecule showed that the amount of acid sites of TS-1 in-
creased with increasing Ti content [39]. Therefore, TS-1 is an active
and selective catalyst for the gas-phase Beckmann rearrangement
of cyclohexanone oxime, a reaction believed to take place on the
hydroxyl groups of zeolites [14]. In addition, the framework Ti(IV)
ions in the vicinity of the silanol groups could change the electron
density around Si owing to different electronegativity of Ti to Si, or
due to a local structure deformation caused by the introduction of
bulky Ti ions into the framework. The strength of the SiO–H bond
then may become weakened, making it donate the proton easily.
This is presumed to be one of the possibilities that TS-1 shows a
relatively stronger acidity than Silicalite-1 [40,41].
In order to differ the active sites between the acidity relative to
hydroxyl groups and Lewis acidity for the intermolecular condensa-
tion of EDA on TS-1, we have carried out the removal of hydroxyl-re-
lated active sites with Na+ ion-exchange and the regeneration of
acidity by acid retreatment, and investigated the effect of these post
treatments on the catalytic activity. The alkali cations incorporated
by either direct hydrothermal synthesis or post ion-exchange are
shown to poison the active sites of TS-1 [42]. As shown in Scheme
2, the Na+ ion-exchange occurred mainly on the Si–OH group adja-
cent to the ‘‘open” Ti site but not on the SiOHꢂ ꢂ ꢂHOSi sites or the iso-
lated terminal SiOH sites. Since the isoelectric points of SiO2 and
TiO2 are in the pH range of 2–3, and 5–6, respectively, the SiO–H
groups should be more easily ion-exchanged than the TiO–H
groups.
1446
1605
1490
e
d
c
b
a
1600 1550 1500 1450 1400
Fig. 11 shows the IR spectra in the structure vibration region of
TS-1 before and after various ion-exchanges. When TS-1 was ex-
changed with NaNO3 solution, the characteristic band at 963 cmꢀ1
due to the framework Ti ions remained almost intact (Fig. 11a and
b). However, the 963-cmꢀ1 band disappeared completely and
Wavenumber (cm-1
)
Fig. 10. Pyridine-adsorbed IR spectra after evacuation at 473 K for 1 h of TS-1
synthesized at Si/Ti ratio of 1 (a), 200 (b), 60 (c), 40 (d), and 30 (e).

Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
265
SiO
SiO
SiO
SiO
OSi
OH
ONa
OSi
SiO
OSi
Na2CO3
HNO3
SiO
OSi
O
Ti
Si
Ti
Si
Ti
Si
SiO
SiO
OH
OH
SiO
SiO
or
SiO
OSi
OSi
SiO
Tetrahedral Ti site
Na+-exchanged form
Scheme 2.
while the exchange with Na2CO3 resulted in a much higher Na con-
tent than that of Ti (Table 2, Nos. 1, 2 and 4). In agreement to pre-
vious observations [21,42], the results indicate that the exchange
of sodium for the protons in TS-1 takes place more readily at high
pH. The sodium ions were removed to a very low level after a fur-
ther washing with 1 M HNO3 at room temperature for 24 h,
whereas the Ti content was influenced rarely (Nos. 3 and 5). The
Na+ ion-exchanged and further acid-treated samples showed very
similar UV–visible spectra to the parent TS-1 (see Supporting infor-
mation, Fig. S1). All samples contained mainly the isolated Ti spe-
cies in the framework since the 220-nm band was the predominant
adsorption band in their UV spectra. These indicate that the state of
Ti species was intact by these treatments.
d
c
Table 2 lists also the catalytic activity of the ion-exchanged TS-1
samples for intermolecular condensation of EDA. Compared with
the parent TS-1, the sample exchanged with NaNO3 decreased
slightly in EDA conversion, and showed very similar EDTA selectiv-
ity but a lower selectivity for byproducts. The removal of sodium
by subsequent acid washing recovered the activity and the product
selectivities (Nos. 1–3) almost totally. However, the ion-exchange
with Na2CO3 caused the EDA conversion to decrease dramatically
from 92.1% to 6.5% (No. 4). The acid washing with HNO3 recovered
the catalytic performance greatly (No. 5). The incomplete regener-
ation may be due to an insufficient removal of the poisoning so-
dium ions.
Fig. 12A shows the IR spectra of TS-1 before and after ion-ex-
change in the region of hydroxyl stretching vibration. Compared
with the parent sample, the bands at 3741 and 3725 cmꢀ1 varied
little when exchanged with NaNO3, whereas the 3505-cmꢀ1 band,
ascribed to hydrogen-bonded silanol groups probably on internal
defect sites such as hydroxyl nests, decreased in intensity appar-
ently (Fig. 12A, a and b). These hydroxyl groups were regenerated
after acid treatment (Fig. 12A, c). On the other hand, the ion-ex-
change with Na2CO3 removed various OH groups significantly,
b
a
963
1000
900
800
700
Wavenumber (cm-1
)
Fig. 11. IR spectra in framework vibration region of TS-1 (Si/Ti = 40) (a), (a) treated
with NaNO3 (b), (a) treated with Na2CO3 (c), and (c) washed with HNO3 (d).
turned to a shoulder at 985 cmꢀ1 when the sample was ion-ex-
changed in a more basic solution of Na2CO3 (Fig. 10c). A subsequent
washing of Na–TS-1 with 1 M HNO3 solution at room temperature
restored the 963-cmꢀ1 band (Fig. 11d), the slight broadness of
which suggested that the sample adsorbed a large number of water
molecules on the defect sites developed by desilication, since we ob-
served that the sample lost weight by 25 wt% after Na2CO3
treatment.
The Na content analyses indicate that the ion-exchange with
NaNO3 incorporated a relatively low amount of sodium into TS-1,
leading to
a spectrum consisting of only a weak band at
Table 2
The results of intermolecular condensation of EDA over the TS-1 catalysts after various post-treatments.a
No.
Catalyst
Si/Tib
Na2Oc (wt%)
Na/Ti ratio
EDA conv. (%)
Product sel. (%)
PIP
TEDA
Othersd
1
2
3
4
5
TS-1
39
40
40
36
41
Trace
0.77
0.08
4.05
0.18
–
92.1
76.7
88.5
6.5
24
29
27
42
24
58
62
60
40
53
18
9
13
18
23
e
TS-1-NaNO3
0.60
0.06
2.82
0.14
TS-1-Na2CO3-AWf
e
TS-1-Na2CO3
TS-1-Na2CO3-AWf
81.7
a
b
c
d
e
f
Reaction conditions: see Table 1.
Determined by ICP.
Determined by AAS.
Mainly lower aliphatic amines and the derivatives of pyrazine.
The ion-exchange was carried out in 1 M NaNO3 or Na2CO3 solution at room temperature, respectively.
AW, acid washing with 1 M HNO3 at room temperature.

266
Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
1446
1490
3741
B
1605
1595
A
1442
e
3510
e
d
c
d
c
b
a
b
a
3725
3800 3600 3400 3200
Wavenumber (cm-1
1600 1550 1500 1450 1400
Wavenumber (cm-1
)
)
Fig. 12. IR spectra of TS-1 (Si/Ti = 40) before (A) and after pyridine adsorption (B). (a) Parent; (b) as a after NaNO3 treatment; (c) as b after HNO3 washing; (d) as a after Na2CO3
treatment, and (e) as d after HNO3 washing (e). The pyridine desorption was carried out by evacuation at 423 K for 1 h.
3741 cmꢀ1 due to terminal silanols (Fig. 12A, d). The acid washing
then regenerated the hydroxyl groups by removing the counterpart
sodium ions (Fig. 12A, e). Thus, the hydroxyl groups-related acidity
in TS-1 depends greatly on the form of ion-exchange.
4.3. Investigation into the reaction spaces of TS-1 for EDA
condensation
The reports reported so far deal rarely with the issue of reaction
spaces for the intermolecular condensation of EDA which involves
the reaction steps of small reactant molecules to monocyclic inter-
mediate and further to more bulky multicyclic product. The TS-1
samples with varying crystal sizes were thus synthesized using dif-
ferent silicon and SDA sources to investigate the contribution of
intracrystal channels and the outer surface to the reaction (Table
3). The catalytic properties were strongly influenced by the size of
the crystals of TS-1. When the TS-1 samples with similar spherical
morphology increased in crystal size from an average size of 0.3–
The sodium ion-exchange developed two additional bands at
1592 and 1442 cmꢀ1 in pyridine-adsorbed IR spectra in compari-
son to the parent TS-1 (Fig. 12B). The phenomena were particu-
larly obvious in the case of extensively ion-exchanged sample
with Na2CO3 solution (Fig. 12B, d). These two bands (mode 8a
and 19b of the pyridine ring) are attributed to pyridine adsorbed
on alkali ions [43–45]. The still existence of the bands at 1605
and 1490 cmꢀ1 implies that the sodium ion-exchange did not re-
place all the protons in TS-1, although it removed most hydroxyl
groups. The removal of sodium ions by acid washing made the lo-
cal chemical environment around the Ti site reversible to go back
to its original state (Scheme 2), and then the spectra of adsorbed
pyridine turned to be the same as the parent TS-1 (Fig. 12B, c
and e). In particular, the band at 1442 cmꢀ1 changed back to the
hydrogen-bonded one at 1446 cmꢀ1. These results would deny
the contribution to the EDA conversion of the Lewis acidity di-
rectly related to the framework Ti ions in TS-1, but support that
the actual active sites for intermolecular condensation of EDA
are those Si–OH groups adjacent to the so-called ‘‘open” Ti sites,
which show a moderate acidity as evidenced by the hydroxyl
stretching vibration as well as the hydrogen-bonded pyridine
20 lm, the EDA conversion decreased from 84.4% to 79.5% (Nos. 1
and 2). At similar EDA conversion, the selectivity to TEDA dropped
from 53% to 39% (No. 3). On the other hand, the plate-shaped TS-1
zeolite with a much larger crystal size of 30 ꢁ 8 ꢁ 2
lm gave the
lowest EDA conversion (No. 4), and the lowest selectivity to TEDA
even at similar EDA conversion (No. 5). The orientation of the pore
system is presumed to affect the diffusion rate of the molecules
for this catalyst which had greatly different crystal lengths or
widths on three crystal faces. The zigzag 10-MR channels
(0.51 ꢁ 0.55 nm) with a relatively smaller pore size are parallel to
the plate crystals. They are considered to provide a longer diffusion
pathway for the molecules than the straight 10-MR channels
(0.53 ꢁ 0.56 nm) perpendicular to the plate crystal (the thinnest
vibration at 1446 cmꢀ1
.
Table 3
The result of intermolecular condensation of EDA over TS-1 catalysts with various crystal size.a
No.
Si/Ti ratiob
Crystalc
Size ( m)
EDA conv. (%)
Product sel. (%)
PIP
l
Morphology
TEDA
Othersd
1
60
61
61
58
58
0.3–0.4
20
20
spherical
spherical
spherical
plate shaped
plate shaped
84.4
79.5
84.0
71.1
82.1
31
52
41
64
51
53
32
39
21
28
16
16
20
15
21
2
3e
4
30 ꢁ 8 ꢁ 2
5e
30 ꢁ 8 ꢁ 2
a
b
c
Reaction conditions: see Table 1.
Determined by ICP.
Determined by SEM.
d
e
Mainly lower aliphatic amines and the derivatives of pyrazine.
Reaction conditions: N2, 10 mL minꢀ1
.

Y. Wang et al. / Journal of Catalysis 266 (2009) 258–267
267
direction), and then make the molecules suffer more diffusion lim-
References
itations. This may correspond to the lowest EDA conversion as well
as the lowest selectivity to bulky product of TEDA on this TS-1 with a
large crystal size. It is thus favorable to reduce the crystal size and
then increase the external surface area to make the primary conden-
sation of EDA to PIP and the secondary condensation of PIP to TEDA
both moving smoothly. The average molecular sizes are 0.31 ꢁ 0.54
nm for EDA, 0.27 ꢁ 0.48 nm for PIP and 0.46 ꢁ 0.48 ꢁ 0.48 nm for
TEDA as given by Gaussian2003 software. Therefore, we assume
that the first intermolecular condensation of EDA to PIP takes place
mainly inside the medium pores of the MFI structure. With respect
to secondary condensation of PIP with EDA to TEDA, since the TEDA
molecules with a polycyclic shape are relatively rigid, the 10-MR
channels of TS-1 would propose mass transfer limitations inside
for these molecules. Moreover, TEDA is formed via a bulky interme-
diate of aminoethylpiperazine (0.27 ꢁ 0.48 ꢁ 0.7 nm), which is
hardly accommodated inside the pores of the MFI structure. Thus,
the condensation of PIP with EDA is considered to take place mainly
on the acid sites near the pore entrance and on the external surface.
[1] Ullmann’s Encyclopedia of Industrial Chemistry, vol. A2, VCH, 1998, p. 16.
[2] A. Belhekar, T.K. Das, K. Chaudhari, S.G. Hegde, A.J. Chandwadka, Stud. Surf. Sci.
Catal. 113 (1998) 195.
[3] H.G. Bosche, K.Baer, K. Schneider, DE Patent 2, 442, 929, 1976.
[4] I. Hudea, M. Biemel Werner, R. Czifra, RO Patent 85, 563, 1984.
[5] G. Engemann, G. Spielberger, US Patent 3, 080, 371, 1963.
[6] N. Srinivas, D. Venu Gopal, B. Srinivas, S.J. Kulkarni, M. Subrahmanyam,
Micropor. Mesopor. Mater. 51 (2002) 43.
[7] Y. Bhat, J. Das, S. Ali, B.D. Bhatt, A.B. Halgeri, Appl. Catal. A 148 (1996) L1.
[8] R. Anand, B.S. Rao, Catal. Commun. 3 (2002) 479.
[9] M. Selvaraj, B.R. Min, Y.G. Shul, T.G. Lee, Micropor. Mesopor. Mater. 74 (2004)
157.
[10] M. Frauenkron, B. Stein, US Patent 6, 562, 971 B2, 2003.
[11] G. Bellussi, M.S. Rigguto, Stud. Surf. Sci. Catal. 85 (1994) 177.
[12] B. Notari, Adv. Catal. 41 (1996) 253.
[13] H. Ichihashi, Catal. Catal. (Shyokubai) 47 (2005) 190.
[14] A. Thangaraj, S. Sivasanker, P. Ratnasamy, J. Catal. 137 (1992) 252.
[15] D. Srinivas, R. Srivastava, P. Ratnasamy, Catal. Today 96 (2004) 127.
[16] R.S. Drago, S.C. Dias, J.M. McGilvray, A.L.M.L. Mateus, J. Phys. Chem. B 102
(1998) 1508.
[17] G. Deo, A.M. Turek, I.E. Wachs, Zeolites 13 (1993) 365.
[18] A. Auroux, A. Gervasini, E. Jorda, A. Tuel, Stud. Surf. Sci. Catal. 84 (1994)
653.
[19] R. Srivastava, D. Srinivas, P. Ratnasamy, Catal. Lett. 91 (2003) 133.
[20] J.Q. Zhuang, Z.M. Yan, X.M. Liu, X.C. Liu, X.W. Han, X.H. Bao, U. Mueller, Catal.
Lett. 83 (2002) 87.
5. Conclusions
[21] T. Tatsumi, K.A. Koyano, Y. Shimizu, Appl. Catal. A 200 (2000) 125.
[22] A. Corma, U. Diaz, E.M. Domine, V. Fornés, J. Am. Chem. Soc. 122 (2000)
2804.
[23] P. Wu, T. Komatsu, T. Yashima, J. Phys. Chem. 100 (1996) 10316.
[24] P. Wu, T. Komatsu, T. Yashima, J. Catal. 168 (1997) 400.
[25] M.A. Camblor, A. Corma, A. Martínez, J. Pérez-Pariente, J. Chem. Soc. Chem.
Commun. (1992) 589.
[26] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, J. Phys. Chem. B 105 (2001) 2897.
[27] P. Wu, T. Tatsumi, Catal. Surv. Asia 8 (2004) 137.
[28] A. Tuel, Zeolites 16 (1996) 108.
[29] X.W. Guo, G. Li, X.F. Zhang, X.S. Wang, Stud. Surf. Sci. Catal. 112 (1997)
499.
[30] P.R. Reddy, M. Subrahmanyam, S.J. Kulkarni, Catal. Lett. 54 (1998) 95.
[31] E. Astorino, J.B. Peri, R.J. Willey, G. Busca, J. Catal. 157 (1995) 482.
[32] D. Scarano, A. Zecchina, S. Bordiga, F. Geobaldo, G. Spoto, G. Petrini, G. Leofanti,
M. Padovan, G. Tozzola, J. Chem. Soc. Faraday Trans. 89 (1993) 412.
[33] W. Lin, H. Frei, J. Am. Chem. Soc. 124 (2002) 9292.
[34] R. Buzzoni, S. Bordiga, G. Ricchiardi, C. Lamberti, A. Zecchina, G. Bellussi,
Langmuir 12 (1996) 930.
TS-1 is superior to other titanosilicates in the intermolecular
condensation of EDA to TEDA, giving EDA conversion to PIP and
TEDA as high as 95% and a total selectivity of ca. 85%, when synthe-
sized to have small a crystal size (0.3–0.4 lm) as well as a high Ti
content corresponding to Si/Ti ratio of 30. The incorporation of Ti
ions into the framework generates both acidic hydroxyl groups
and Lewis acid sites in TS-1. The acidic hydroxyl groups with a
moderate acidity is originated from the internal Si–OH groups
adjacent to the ‘‘open” Ti sites, while the Lewis acidity is due to
the framework Ti ions. It is the internal silanols in TS-1 that con-
tribute to the condensation of EDA rather than its Lewis acid sites.
A small crystal size of TS-1 benefits more the secondary condensa-
tion of PIP to TEDA, a bulky reaction which requires open reaction
spaces such as pore entrance and external surface, compared to the
primary condensation of EDA to PIP.
[35] C. Pazé, S. Bordiga, C. Lamberti, M. Salvalaggio, A. Zecchina, G. Bellussi, J. Phys.
Chem. B 101 (1997) 4740.
[36] F. Bonino, A. Damin, S. Bordiga, C. Lamberti, A. Zecchina, Langmuir 19 (2003)
2155.
Acknowledgments
[37] C. Ngamcharussrivichai, P. Wu, T. Tatsumi, J. Catal. 235 (2005) 139.
[38] C. Lamberti, S. Bordiga, D. Arduuino, A. Zecchina, F. Geobaldo, G. Spano, F.
Genoni, G. Petrini, A. Carati, F. Villain, G. Vlaic, J. Phys. Chem. B 102 (1998)
6382.
[39] G. Yang, X.J. Lan, J.Q. Zhuang, D. Ma, L.J. Zhou, X.C. Liu, X.W. Han, X.H. Bao, Appl.
Catal. A 337 (2008) 58.
[40] L. Chen, L. Norena, J. Navarrete, J. Wang, Mater. Chem. Phys. 97 (2006) 236.
[41] J. Anderson, C. Fergusson, I. Rodriguez-Ramos, A. Gueerero-Ruiz, J. Catal. 192
(2000) 344.
We gratefully acknowledge NSFC of China (20673038,
20873043), Science and Technology Commission of Shanghai
Municipality (09XD1401500, 08JC1408700, 07QA14017), 973
Program (2006CB202508), 863 Program (2007AA03Z34), and
Shanghai Leading Academic Discipline Project (B409). Y.W. thanks
the PhD Program Scholarship Fund of ECNU 2008.
[42] C.B. Khouw, M.E. Davis, J. Catal. 151 (1995) 77.
[43] J.W. Ward, Am. Chem. Soc. Monogr. 171 (1979) 118.
[44] E. Dumitriu, V. Hulea, S. Kaliaguine, M.M. Huang, Appl. Catal. A 135 (1996)
57.
Appendix A. Supplementary material
[45] H. Bludau, H.G. Karge, W. Niessen, Micropor. Mesopor. Mater. 22 (1998)
297.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2009.06.016.