A novel titanosilicate with MWW structure III. Highly efficient and selective production of glycidol through epoxidation of allyl alcohol with H2O2

Article abstract of DOI:10.1016/S0021-9517(02)00170-7

The catalytic properties of Ti-MWW in the epoxidation of allyl alcohol (AAL) with hydrogen peroxide to glycidol (GLY) have been studied in detail by a comparison with those of TS-1 and pure silica Ti-Beta, and mechanical considerations have been given to the relation between the catalytic performance and the structural, acidic, and hydrophilic/hydrophobic nature of titanosilicates. Ti-MWW catalyzed the AAL epoxidation more actively and selectively than TS-1 and Ti-Beta in the presence of H₂O or MeCN, and exhibited a conversion of 95% for AAL and a selectivity of 99% for GLY when the AAL epoxidation was carried out at 333 K for 30 min and at 12 wt% of catalyst to substrate. Ti-MWW proved to be a reusable and sustainable catalyst as it stood up to Ti leaching and maintained the catalytic activity and the product selectivity in the reaction-regeneration cycles. The acidic character due to the boron framework was very weak, and thus contributed negligibly to the solvolysis of GLY. The AAL epoxidation proceeded mainly within the intralayer sinusoid 10-MR channels which supplied more steric fitness to the substrate molecules than the tunnel-like channels of TS-1 ad Ti-Beta. Ti-MWW was more hydrophilic than TS-1, but much more hydrophobic than Ti-Beta. The hydrophilicity of Ti-MWW was presumed to derive mainly from the defect sites due to the incomplete dehydroxylation between the layers and partially as a result of deboronation. The sinusoidal 10-MR channels serving as the reaction space for the AAL epoxidation were considered to be hydrophobic, thus rendering the Ti-MWW catalyst applicable to the substrates and solvents, both of a polar nature.

Full text of DOI:10.1016/S0021-9517(02)00170-7

Journal of Catalysis 214 (2003) 317–326
www.elsevier.com/locate/jcat
A novel titanosilicate with MWW structure
III. Highly efficient and selective production of glycidol through
epoxidation of allyl alcohol with H O
2
2
∗
Peng Wu and Takashi Tatsumi
Division of Materials Science & Chemical Engineering, Graduate School of Engineering, Yokohama National University, 79-5 Tokiwadai,
Hodogaya-ku, Yokohama 240-8501, Japan
Received 19 August 2002; revised 9 October 2002; accepted 9 October 2002
Abstract
The catalytic properties of Ti-MWW in the epoxidation of allyl alcohol (AAL) with hydrogen peroxide to glycidol (GLY) have been
studied in detail by a comparison with those of TS-1 and pure silica Ti-Beta, and mechanical considerations have been given to the relation
between the catalytic performance and the structural, acidic, and hydrophilic/hydrophobic nature of titanosilicates. Ti-MWW catalyzed the
AAL epoxidation more actively and selectively than TS-1 and Ti-Beta in the presence of H O or MeCN, and exhibited a conversion of 95%
2
for AAL and a selectivity of 99% for GLY when the AAL epoxidation was carried out at 333 K for 30 min and at 12 wt% of catalyst to
substrate. Ti-MWW proved to be a reusable and sustainable catalyst as it stood up to Ti leaching and maintained the catalytic activity and the
product selectivity in the reaction-regeneration cycles. The acidic character due to the boron framework was very weak, and thus contributed
negligibly to the solvolysis of GLY. The AAL epoxidation proceeded mainly within the intralayer sinusoid 10-MR channels which supplied
more steric fitness to the substrate molecules than the tunnel-like channels of TS-1 ad Ti-Beta. Ti-MWW was more hydrophilic than TS-1,
but much more hydrophobic than Ti-Beta. The hydrophilicity of Ti-MWW was presumed to derive mainly from the defect sites due to the
incomplete dehydroxylation between the layers and partially as a result of deboronation. The sinusoidal 10-MR channels serving as the
reaction space for the AAL epoxidation were considered to be hydrophobic, thus rendering the Ti-MWW catalyst applicable to the substrates
and solvents, both of a polar nature.

2003 Elsevier Science (USA). All rights reserved.
Keywords: Ti-MWW; Allyl alcohol; Glycidol; Hydrogen peroxide; Liquid-phase epoxidation
1
. Introduction
Nevertheless, TS-1 has a major problems in that the
medium pores of a 10-membered ring (MR) limit its ap-
plication to the substrates with relatively small molecular
sizes and that its intrinsic activity per Ti site is still not
high enough for the epoxidation of functionalized alkenes.
Other titanosilicates with larger pores developed there-
after by hydrothermal synthesis or postsynthesis methods,
such as Ti-Beta [3,4], Ti-MOR [5], Ti-ITQ-7 [6], and
Ti-MCM-41 [7], have partially solved the problems of steric
constraints encountered in the oxidation of bulky substrates.
Ti-Beta especially proves to be a very attractive catalyst
according to its high activity for the cyclic and branched
alkenes and alkanes [3,4]. However, it seems that the above
titanosilicates are intrinsically less active than TS-1 for the
oxidation of small molecules and have hardly reached the
level required for industrialization in terms of both catalytic
activity and selectivity. Much effort has been made to pre-
MFI-type titanosilicate, TS-1, being able to actively
catalyze the liquid-phase oxidation 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 [1,2]. The unique activity of TS-1 is mostly
due to the isolated Ti species of tetrahedral coordination
in the hydrophobic framework. The discovery of TS-1 has
led to an industrialized process of hydroxylation of phenol
to hydroquinone and catecol, and a promising process of
cyclohexanone ammoximation to oxime. These successes
imply potential applications of titanosilicates to the clean
production of fine chemicals.
*
Corresponding author.
E-mail address: ttatsumi@ynu.ac.jp (T. Tatsumi).
0021-9517/03/$ – see front matter  2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0021-9517(02)00170-7

318
P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
pare titanosilicates that are more applicable to the fields of
fine chemicals or pharmaceuticals.
AAL epoxidation catalyst by considering the pore structure,
acidity, and hydrophilicity/hydrophobicity.
Recently, we succeeded in synthesizing a novel titanosil-
icate with the MWW structure, Ti-MWW [8]. Ti-MWW
consists not only of the open reaction spaces of side cups
and supercages but also of the medium reaction space of
two independent 10-MR channels. Taking advantages of this
structural diversity, Ti-MWW proves to be active for the
epoxidation of linear alkenes and cyclic alkenes as well.
Ti-MWW is more active in the epoxidation of linear alkenes
than TS-1, the titanosilicate showing the highest intrinsic ac-
tivity so far. The unusual catalytic performance of Ti-MWW
suggests that it may serve as a potential catalyst for the oxi-
dation of functionalized alkenes.
2
. Experimental
2
.1. Preparation of titanosilicate catalysts
Ti-MWW catalysts were prepared in two steps according
to procedures previously reported [8b]. The first step was
to hydrothermally synthesize Ti-containing MWW lamel-
lar precursors using piperidine as a structure-directing agent
(
SDA) and boric acid as a crystallization-supporting agent,
from synthetic gels with molar compositions of 1 SiO2:
.01–0.05 TiO2:0.67 B2O3:1.4 SDA:19 H2O. The second
0
Among the oxidation of functionalized alkenes, the epox-
idation of allyl alcohol (AAL) to glycidol (GLY) is probably
one of the most demanded reactions. GLY is industrially
produced by epoxidation of AAL with hydrogen peroxide,
hypochlorous acid, or peracetic acid [9,10]. The most en-
vironmentally benign oxidant is hydrogen peroxide, which
is to be catalytically activated; however, the conventional
tungstate catalyst suffers from problems in activity, regener-
ation, and reusability. Several studies concentrated on AAL
epoxidation with hydrogen peroxide on TS-1. Nevertheless,
the initial reaction rate for the epoxidation of AAL was re-
ported to be about 30 times slower than the rate of 1-butene
epoxidation [11]. The selectivity to GLY was not so suffi-
cient; only glycerol was observed when dilute H2O2 solution
step was to treat the precursors in HNO3 or H2SO4 solutions
under reflux for the purpose of removing the extra frame-
work Ti species together with a part of the framework of
boron. Subsequent calcination on the acid-treated sample re-
sulted in the objective Ti-MWW catalysts.
TS-1 catalysts, with Si/Ti ratios of 100-36, were syn-
thesized following conventional procedures [16], and were
washed with 1 M HCl solution at ambient temperature to
reduce the residual alkali cations contaminating the SDA
solution of tetrapropylammonium hydroxide because al-
kali cations may partially poison the Ti active sites [17].
Titanosilicates of the BEA structure, Al-containing Ti,
Al-Beta, (Si/Ti = 40, Si/Al = 70), and Al-free Ti-Beta
(
Si/Ti = 42, Si/Al > 1000), were synthesized according to
(
5 wt% in water) was used [12]. The low GLY selectivity
conventional [3] and seeding methods [18], respectively.
The catalysts were characterized by inductively cou-
pled plasma (ICP, Rigaku JY38S), X-ray diffraction (XRD,
Rigaku Rint 2000), N2 adsorption (Coulter SA 3100), and
scanning electron microscopy (SEM, JEOL JSM-T220), and
the spectroscopies of UV–visible (Shimadzu UV-2400PC)
and IR (Shimadzu FTIR-8100). IR spectra in the region of
hydroxyl stretching vibration were recorded after the self-
supported disk (30 mg with 2 cm diameter) was evacuated
at 773 K for 1 h. The acidic properties were measured with
NH3-TPD (BEL Japan Inc.). The sample was dehydrated un-
der He at 773 K for 1 h and then treated with NH3 at 373 K
for 0.5 h. The TPD spectra were recorded at a heating rate
is mainly due to the consecutive hydrolysis or solvolysis of
GLY on the residual acid sites of TS-1 as reported by Hutch-
ings et al. [13]. The GLY selectivity dramatically declined
from 100 to 2% when the AAL conversion increased from
3
to 100% especially in the presence of alcoholic solvents
of methanol or ethanol preferred by TS-1. Control of crystal
size and poisoning the acid sites with sodium cations have
been attempted to improve the performance of TS-1 in AAL
epoxidation [13], but the former seems to have limitations,
while the latter deactivates the Ti active sites simultaneously.
Therefore, TS-1 catalysts cannot maintain a favorable bal-
ance between AAL conversion and GLY selectivity.
Surveying the literature on AAL epoxidation, it seems to
us that high yields of GLY depend predominantly on the in-
trinsic activity of titanosilicates with low acidity. What is
also important is that the hydrophilicity/hydrophobicity of
titanosilicates may play a significant role in the epoxidation
−1
of 10 K min using m/z of 16 to avoid the influence of any
contaminating water.
2
.2. Catalytic reaction
[
4,14]. Particularly, different from unfunctionalized alkenes,
The epoxidation of allyl alcohol with hydrogen peroxide
AAL is hydrophilic and then could exhibit different behav-
ior. The hydrophilicity/hydrophobicity may also influence
the activity and selectivity through the solvent effect [15].
In this study, we have applied Ti-MWW catalysts to
the AAL epoxidation for the purpose of producing GLY
efficiently and selectively. The effects of solvent and the
reaction conditions on the AAL epoxidation have been
investigated by comparing Ti-MWW with TS-1 and Ti-Beta.
The main aim is to investigate why Ti-MWW is an excellent
was carried out in a 20-ml glass flask with a condenser under
vigorous stirring. In a typical run, 10 mmol of AAL, 5 ml
of solvent, and a certain amount of catalyst were mixed in
the flask and heated to desired temperatures under agitation.
H2O2 (10 mmol, 30 wt%) was then added to the mixture to
start the reaction. The products were separated by filtration
and analyzed with a gas chromatograph(Shimadzu GC-14B)
using a OV-1 capillary column of 50 m and an FID detector.
The amount of unconverted H2O2 was determined by the

P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
319
standard titration method using a 0.1 M Ce(SO4)2 solution.
The hydrolysis of glycidol and the epoxidation of 1-hexene
and cyclohexene were conducted under conditions similar to
AAL epoxidation.
3
. Results and discussion
3
.1. Preparation and characterization of various
titanosilicate catalysts
MWW-type titanosilicate, Ti-MWW, seemed to be hardly
hydrothermally synthesized up to date directly from the
sources of silicon and titanium in the presence of or-
ganic structure-directing agents. Nevertheless, we recently
developed new procedures to prepare this novel titanosili-
cate [8]. The procedures involve the hydrothermal synthesis
of Ti-containing lamellar precursor with boric acid as a
crystallization-supporting agent and the removal of extra
framework Ti species and framework B by subsequent acid
treatment of the precursor. Table 1 shows the physicochem-
ical properties of Ti-MWW catalysts prepared by following
the above procedures. Ti-containing lamellar precursors of
MWW were synthesized, which proved to have Ti contents
nearly stoichiometrically equivalent to those of the synthetic
gels. The treatment of the precursors with HNO3 or H2SO4
solutions partially removed the Ti and B species. Further cal-
cination of the acid-treated samples resulted in Ti-MWW
catalysts with Si/Ti ratios varying in a wide range. Acid-
treated and subsequently calcined samples showed the XRD
patterns corresponding to the MWW topology (not shown)
and extremely high specific surface areas measured by N2
adsorption, consistent with their high crystallinity.
Fig. 1. UV–visible spectra of Ti-MWW with Si/Ti ratios of ∞ (a), 146 (b),
76 (c), 64 (d), 46 (e), and 38 (f).
extra framework Ti species, although not a large amount,
remained in Ti-MWW with a Si/Ti ratio of 38 (Fig. 1f);
it was obtained from the precursor of Si/Ti = 20 which
contained too high a concentration of octahedral Ti [8b]. The
IR spectra of Ti-MWW in the framework vibration region
−1
showed the characteristic band at 960 cm assigned to the
framework Ti; the band intensity increased with increasing
Ti content [8b]. Both IR and UV–visible data verified
that Ti-MWW catalysts thus obtained had good quality in
terms of incorporation of tetrahedral Ti coordination. The
scanning electron micrographs of Ti-MWW revealed thin
platelet crystals with a hexagonal morphology (0.2–0.5 µm
in length and 0.05–0.1 µm in thickness). The relatively small
crystal size, particularly, in thickness, is therefore considered
to make Ti-MWW a highly active liquid-phase oxidation
catalyst, which may place less restrictions on the diffusion
of the substrates. The physicochemical properties of other
titanosilicates, TS-1, Ti, Al-Beta, and Ti-Beta, matched well
those given in the literature.
Acid-treated Ti-MWW samples contained only isolated
Ti species in the framework as evidenced by the predominant
band at 220 nm in their UV–visible spectra when the Si/Ti
ratio was higher than 46 (Fig. 1, a–e). However, anatase-like
Table 1
a
Preparation of Ti-MWW catalysts with various Ti contents
No.
Gel
Ti-containing
precursor
Ti-MWW catalyst
c
b
Si/Ti Si/B
Acid treatment Si/Ti Si/B SA
b
Si/Ti Si/B SA
3
.2. Effects of reaction conditions on the epoxidation
1
2
3
4
5
6
7
8
9
∞
0.75
∞
12 616
6 M HNO
6 M HNO
2 M HNO
∞
170
146
107
76
94
64
49
46
102 611
37
34 628
67
41 622
51
36 624
36
3
3
3
of allyl alcohol
100 0.75 120
100 0.75 120
70 0.75 63
70 0.75 63
50 0.75 51
50 0.75 51
30 0.75 31
30 0.75 31
20 0.75 21
13 625
13 625
12 612
12 612
12 621
12 621
11 623
11 623
13 540
–
3
.2.1. Influence of solvents
4 M H SO
–
2
4
4
As depicted in Scheme 1, the epoxidation of allyl al-
1 M H SO
2
6 M HNO
2 M HNO
–
cohol produces the main product, glycidol, which may be
secondarily hydrolyzed to glycerol through the ring opening
reaction because aqueous H2O2 solution is used as an ox-
idant. GLY also successively undergoes solvolysis to form
the by-products of glycerol ethers especially when alcohols
are used as a solvent. Both the hydrolysis and the solvoly-
sis of GLY are due to the acidic character of titanosilicates
derived from the silanol groups and the additional bridg-
ing hydroxyls of trivalent cations coexisting with Ti in the
3
3
2 M H SO
–
2
4
2 M HNO
2 M HNO
49 624
54 556
3
1
0
38
3
a
Crystallization: stirred at 403 K for 1 day, at 423 for 1 day, and then at
4
43 K for 5 days.
b
2
−1
.
Surface area (Langmuir) in m g
c
−1
;
Acid treatment: ratio of acid solution to solid precursor, 20 ml g
refluxed for 20 h.

320
P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
Scheme 1.
framework. The solvent, generally having a great influence
on the intrinsic activity of Ti species in titanosilicates, thus
also affects the product distribution in AAL epoxidation.
Table 2 compares the catalytic performance of Ti-MWW
with that of conventional titanosilicates of TS-1 and Ti-Beta
for AAL epoxidation in various solvents. A clear solvent
effect was observed for Ti-MWW. The most favorable
solvents were MeCN and H2O, which gave comparable
conversions of 82–87% for AAL, produced almost only
GLY, and exhibited high efficiency for H2O2 utilization.
The epoxidation of AAL was greatly retarded in the protic
solvents of alcohols and also in the aprotic solvents of
acetone and dioxane. The role of alcohol solvents in the
oxidation activity of TS-1 is usually explained by assuming
the formation of a stable five-membered cyclic intermediate
species (Scheme 2), which is formed by the coordination
of the protic molecule, ROH, to the Ti center and by
the stabilization of the Ti-hydroperoxo complex through
hydrogen bonding [19]. This species formed in the channels
would gradually impose a steric hindrance on the substrate
molecules when ROH becomes bulky. Indeed, the AAL
conversion slightly decreased in the order MeOH > EtOH >
1
-PrOH. Besides the influence on the catalytic activity, the
protic alcohols caused the opening of the oxirane ring to
reduce greatly the selectivity for GLY.
In the case of TS-1, very comparable AAL conversion
was observed for the protic solvents with relatively small
molecular size, H2O and MeOH, and aprotic acetone. Apro-
tic but basic MeCN, a well-known unsuitable solvent for
the oxidation of simple alkenes on TS-1, showed some-
what lower conversion in the epoxidation of AAL. 1-PrOH
retarded greatly the reactivity of TS-1, which is simply
assumed to be a result of forming bulky and sterically re-
stricted species in Scheme 2. When the epoxidation was
suppressed, nonproductive decomposition of H2O2 was ob-
served to increase as reflected by the lower H2O2 efficiency.
It is obvious that TS-1, despite higher Ti content, is much
Scheme 2.

P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
321
less effective in epoxidizing AAL than Ti-MWW for both
the activity and the selectivity to GLY.
Ti-Beta catalyzed the epoxidation of AAL to a certain
extent in the presence of organic solvents, but was less ef-
fective even than TS-1. In agreement with the proposition
that basic MeCN may poison the Brønsted acid sites of Ti,
Al-Beta to suppress the ring opening of epoxide [15], the
selectivity for GLY was much higher in MeCN than in the
protic alcohols. Nevertheless, this positive effect of MeCN
seemed to be limited since the GLY selectivity was only
Fig. 3. Effect of reaction temperature on the AAL epoxidation over
7
5%, probably because the Beta zeolite contains too high
Ti-MWW. Ti-MWW (Si/Ti = 38, Si/B = 54), 70 mg; AAL, 10 mmol;
a concentration of acidic silanol groups at structural de-
fect sites. An important observation about Ti-Beta was that
H2O, a good solvent for both Ti-MWW and TS-1 in AAL
epoxidation, appeared to be unsuitable. This proposes the
importance of hydrophilic/hydrophobicfeatures of titanosil-
icates. Lower activity of Ti-Beta should be ascribed to its
hydrophilic character despite the Al-free form. Concerning
this issue, a detailed consideration will be given in a later
section.
The above investigation on the solvent effect proves that
Ti-MWW is a promising catalyst for the selective formation
of GLY in the AAL epoxidation if MeCN or H2O is chosen
as a solvent.
H O , 10 mmol; MeCN, 5 ml; time, 60 min.
2
2
short time. The consecutive hydrolysis or solvolysis can then
be limited to a low level. Thus Ti-MWW proves to be an
excellent catalyst.
3
.2.3. Influence of reaction temperature
The epoxidation of AAL depended greatly on the reaction
temperature (Fig. 3). The AAL conversion reached about
5% with increasing temperatures up to 333 K. The GLY
9
selectivity was always maintained above 99% irrespective of
the temperature. Ti-MWW showed a significant conversion
of 42% even at 273 K, indicating its high activity and
usability at low temperatures.
3
.2.2. Influence of Ti content
To further ensure that Ti-MWW has superiority over
3
.2.4. Influence of reaction time
TS-1 in AAL epoxidation, a series of Ti-MWW and TS-1
with various Ti contents was prepared and compared for
their catalytic performance (Fig. 2). The AAL conversion
reasonably increased with increasing Ti content, but Ti-
MWW obviously showed much higher catalytic activity than
TS-1. Ti-MWW also gave much higher GLY selectivity than
Fig. 4 shows the time courses for the ALL epoxidation at
33 K over Ti-MWW in the presence of H2O and MeCN,
3
the most suitable solvents as shown in Table 1. The AAL
conversion increased rapidly with time up to 30 min and
showed no further increase at 1 h because nearly all the
substrate and the oxidant were consumed within 30 min.
The products were predominantly GLY together with a
small amount of glycerol (< 1%), indicating that the ring
−1
TS-1; when the Ti content was more than 0.2 mmol g
corresponding to an Si/Ti ratio of ca. 80, the GLY selectivity
was 99%. AAL epoxidation predominantly produced GLY,
indicating that the reaction rate for GLY formation is much
faster than that of the hydrolysis of GLY to glycerol.
A gradual increase in GLY selectivity with increasing Ti
content strongly indicates that highly efficient formation of
GLY only can be achieved on those catalysts which are so
active as to promote the epoxidation extensively within a
Fig. 2. Dependence of AAL conversion (A) and GLY selectivity (B) on the
Ti content of Ti-MWW and TS-1. Cat., 50 mg; AAL, 10 mmol; H O ,
Fig. 4. Effect of reaction time on the AAL epoxidation over Ti-MWW.
Ti-MWW (Si/Ti = 38, Si/B = 54), 70 mg; AAL, 10 mmol; H O ,
2
2
2 2
10 mmol; MeCN, 5 ml; temperature, 333 K; time, 30 min.
10 mmol; MeCN or H O, 5 ml; temperature, 333 K.
2

322
P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
opening of GLY occurred at a negligible level. What is
more important is that no obvious difference in both the
activity and the GLY selectivity was observed between
H2O and MeCN. This possibly makes Ti-MWW-catalyzed
epoxidation of AAL an environmentally benign process by
using H2O as a solvent.
The identical time dependence of AAL conversion ob-
served for these two characteristically different solvents is
presumably because the AAL epoxidation in water does not
suffer the heterogenization problems of reaction mixtures in
the case of epoxidation of other alkene substrates since both
AAL and glycidol products are soluble in water. A gradual
decrease in the reaction rate with time is mainly due to the
consumption of AAL and partially due to the deactivation of
catalyst as shown below.
lack the stability of Ti species under hydrothermal con-
ditions [20,21]. In fact, the reaction mixture of the AAL
epoxidation over Ti-Beta changed into bright yellow due to
the soluble Ti species after the removal of the catalyst, in-
dicating leaching of Ti species out of the framework sites.
Ti-MWW has proved to be stable against Ti leaching in the
epoxidation of cyclohexene [8e]. However, it has been re-
ported recently that even TS-1, especially crystallized in a
short period, is not so stable to the leaching of Ti species
in the epoxidation of crotyl alcohol, because the triol by-
products may chelate the Ti sites, thereby cleaving easily
the Si–O–Ti bonds [22]. Since the present AAL epoxidation
gave rise to glycerol, although not in a large amount, and
AAL itself is a highly hydrophilic substrate, the stability of
Ti-MWW may be different from what was investigated pre-
viously in the epoxidation of simple alkenes.
The stability against Ti leaching of Ti-MWW has been
thus investigated together with its reusability in AAL epox-
idation. The experiments were conducted starting with an
enlarged reaction scale using 1.2 g of Ti-MWW. The used
Ti-MWW catalyst was regenerated by washing with acetone
or by further calcination in air at 773 K. After taking out a
part of the sample for IPC analyses, the regenerated catalyst
was subjected to repeated epoxidation of AAL while keep-
ing the catalyst-substrate-solvent composition constant. The
AAL conversion decreased dramatically with the reaction-
regeneration cycles if the Ti-MWW used was only washed
with acetone and then dried (Fig. 6A, 2). In this case, the
selectivity to GLY also declined gradually due to the sup-
pression of epoxidation (Fig. 6B, 2). Nevertheless, when the
catalyst collected after 5 reaction-regeneration cycles was
calcined at 773 K, both the AAL conversion and the GLY se-
lectivity were restored. This suggests that the decrease in the
catalytic activity of acetone-washed catalyst is due only to
the deposition of heavy organic compounds of high boiling
points within the channels of Ti-MWW. Indeed, when the Ti-
MWW used was regenerated by directly burning off residual
organic species at 773 K, the AAL conversion increased dur-
ing the first regenerationcycles and then maintained the level
over 90% (Fig. 6A, "); the GLY selectivity was higher than
3
.2.5. Influence of amount of Ti-MWW catalyst
The above investigations suggest that highly selective
formation of GLY should be performed on the titanosilicate
intrinsically active for the epoxidation of AAL. This allows
the epoxidation reaction to reach an extensive level within a
short reaction time, and then would suppress the consecutive
solvolysis of GLY to by-products. Although it is verified
that Ti-MWW is active enough to serve as such a promising
catalyst, the relative amount of Ti-MWW to the substrate is
an issue. As shown in Fig. 5, the AAL conversion increased
rapidly with increasing amount of Ti-MWW, and 12 wt% of
Ti-MWW relative to AAL was enough to convert AAL up to
9
2% at 333 K in 30 min, producing almost only GLY with
efficiency for H2O2 consumption higher than 90%.
3
.3. Stability and reusability of Ti-MWW
Besides the catalytic activity and selectivity, one of the
other important properties that titanosilicate should have is
its stability against Ti leaching, a notable problem com-
monly encountered in the liquid-phase reactions. Ti species
of TS-1 are generally believed to be stable enough in the
liquid-phase oxidation of simple alkenes, but Ti-Beta and
Ti-MCM-41, both containing a large amount of hydrophilic
silanol groups on the defect sites and the external surface,
Fig. 6. Changes of AAL conversion (A) and GLY selectivity (B)
with the epoxidation-regeneration cycles. AAL epoxidation conditions:
Ti-MWW (Si/Ti = 49, Si/B = 36):AAL:H O :MeCN = 50 mg:
Fig. 5. Dependence of AAL conversion and GLY selectivity (A), and
2
2
H O conversion and efficiency (B) on the relative amount of Ti-MWW.
10 mmol:10 mmol:5 ml; temperature, 333 K; time, 60 min. Regeneration:
used catalyst was washed with acetone and dried at 393 K (2), and further
calcined at 773 K in air for 4 h (").
2
2
Ti-MWW (Si/Ti = 38, Si/B = 54), 5–100 mg; AAL, 10 mmol; H O ,
2
2
10 mmol; MeCN, 5 ml; temperature, 333 K; time, 30 min.

P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
323
supplies relatively complicated reaction spaces, side cups
on the exterior crystal surface and two independent 10-MR
channels, one of which comprises supercages [23]. Selective
poisoning of the Ti active sites with amines of different
molecular sizes has been carried out to determine which
part of Ti sites contributes more to the AAL epoxidation
(
Table 3). The amount of amine (2 mmol) was about 70 times
as large as the Ti amount in Ti-MWW catalysts, and thus
should be sufficient to poison the Ti sites where the amine
molecules diffused and approached. The poisoning reagents
used were tripropylamine (TPA) and triphenylamine (TPhA)
with relatively small and large molecular sizes, respectively.
The former, believed to reach all the Ti sites, deactivated Ti-
MWW completely as expected. The coexistence of TPhA,
on the other hand, had little influence on the AAL conversion
when compared to the nonpoisoning epoxidation. However,
TPhA totally retarded the epoxidation of cyclohexene with
tert-butylhydroperoxide that requires open reaction spaces
Fig. 7. Changes of the content of Ti and B during the repeated epoxidation
of AAL over Ti-MWW. For epoxidation conditions, see Fig. 6.
(
not shown). Bulky molecules of TPhA are considered to
9
9% (Fig. 6, "). Therefore, it can be safely assumed that
hardly penetrate the pore mouth of 10-MR channels and then
selectively poison the Ti sites within the cups on the external
surface. Since Ti-MWW was prepared through severe acid
treatment on the lamellar precursor, the acid treatment would
selectively remove the Ti species on the external surface
of the layers and then deeply lower the Ti concentration
within the side cups and the supercages consisting of two
cups. Thus it is assumed from the above results that the
extremely high activity for the AAL epoxidation mainly
originates from the intralayer 10-MR channels containing no
supercages.
The intralayer 10-MR channels within the MWW struc-
ture running along the ab plane have a slightly restricted el-
liptical aperture (4.0×5.9 A) compared to the MFI structure,
exhibiting a special tortuosity of two-dimensional sinusoidal
form (Scheme 3). This kind of channel system not only is fa-
vorable for the diffusion of linear molecules but might also
supply “steric fitness” between the Ti active sites and the
substrate molecules. It is then proposed that the intralayer
Ti-MWW can be employed as a reusable catalyst for the se-
lective epoxidation of AAL to GLY.
This has been further confirmed by quantifying the Ti and
B contents in the used Ti-MWW catalyst (Fig. 7). The B
species were gradually leached out of the framework during
the reaction probably because of its too small ionic radius
compared with that of Si. The loss of B may decrease the
framework electronegativity to vary the electronic density
in the vicinity of Ti sites, a usual phenomena reported
for Al-containing Ti-Beta [3] and Ti-MOR [5b], and may
also increase the hydrophobicity of the framework. Thus
the lose of B is presumed to account for a slight increase
in the AAL conversion observed after the first reaction-
regeneration cycle (Fig. 6, "). On the other hand, the
amount of Ti leached was negligible, which matches well
the high conversion obtained for the catalyst regenerated by
the calcination.
3
.4. Factors governing the unusual catalytic activity and
1
0-MR channels of MWW structure lead to a rather high
selectivity of Ti-MWW in AAL epoxidation
3
.4.1. High activity of Ti-MWW
The extraordinarily high activity of Ti-MWW in the AAL
epoxidation must be closely related to the unique crystalline
structure of the MWW topology. The MWW structure
Table 3
a
Poisoning epoxidation of AAL over Ti-MWW
Poisoning
reagent
Conversion
(mol%)
Product selectivity
(mol%)
H O (mol%)
2 2
Conversion
Efficiency
GLY Glycerol
None
TPA
TPhA
92.6
1.3
91.4
99.5
87.7
99.6
0.5
12.3
0.4
93.7
–
94.8
98.8
–
96.5
a
Ti-MWW (Si/Ti = 38), 70 mg; AAL, 10 mmol; H O , 10 mmol;
2
2
MeCN, 5 ml; amine, 2 mmol; temperature, 333 K; time, 0.5 h.
Scheme 3.

324
P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
Table 4
intermediate (Scheme 2), proposed to act as a kind of weak
a
Hydrolysis of GLY over various titanosilicates
acid site [17], does not promote the ring opening of GLY
under our reaction conditions. The GLY conversion was
slightly high for Ti-Beta and further for Ti,Al-Beta owing
to high concentration of acid sites.
Catalyst
H O
2
GLY
conversion
Product selectivity H O conversion
2
2
2
(mol%)
(mol%)
(
mol%)
Glycerol Others
Ti-MWW
Ti-MWW
TS-1
TS-1
Ti-Beta
Ti, Al-Beta
No
Yes
No
Yes
No
no
3.5
3.4
3.6
3.1
5.7
100
100
100
100
100
100
0
0
0
0
0
0
–
0.4
–
2.3
–
Nearly defect-free TS-1 reasonably contained no acid
sites (Fig. 8a), while Ti,Al-Beta had both weak and strong
acid sites as evidenced by the desorption peaks at 452
and 580 K (Fig. 8f). These two kinds of acid sites are
presumably attributed to the silanols at defect sites and
the framework Si(OH)Al groups, respectively. The high-
temperature peak disappeared for pure silica Ti-Beta, but
the low one still remained, although the intensity was
slightly reduced (Fig. 8e). Ti-MWW contained some acid
sites, weaker in strength and lesser in amount than Ti-
Beta (Figs. 8b–d). The acid sites of Ti-MWW are mainly
related to the framework B since the amount decreased from
the directly calcined precursor to acid-treated sample, and
further to the catalyst repeatedly used in the actual AAL
epoxidation.
13.0
–
a
Cat., 70 mg; GLY, 10 mmol; H O, 5 ml; H O , 10 mmol (if added);
2
2 2
333 K; time, 0.5 h.
reaction rate for functionalized alkenes not observed within
the tunnel-like channels of TS-1. Indeed, taking advantage
of this channel structure, Ti-MWW has been proved to oxi-
dize more actively and selectively the trans-isomer from the
mixture of linear cis/trans alkenes than other titanosilicates
[
8c,8d].
Combining the results of Table 3 and Fig. 8, it is
concluded that the acid sites of Ti-MWW are too weak
to promote the hydrolysis or solvolysis of GLY effectively
to yield corresponding glycerol and its derivatives. It is
possible to remove completely the acidity related to B by
repeated acid treatment on Ti-MWW [8b], but it seems not
necessary for AAL epoxidation.
3
.4.2. High GLY selectivity of Ti-MWW
High GLY selectivity of Ti-MWW may be due to its
outstanding oxidation activity which makes the epoxidation
progress to a high level within a short time before significant
solvolysis occurs. Alternatively, Ti-MWW may contribute
little to the acid-catalyzed solvolysis. Concerning this issue,
the hydrolysis of GLY has been carried out (Table 4), and the
acidity of Ti-MWW together with TS-1, Ti-Beta and Ti,Al-
Beta has been characterized with NH3-TPD measurements
3
.4.3. Hydrophilic/hydrophobic properties of Ti-MWW
Besides the dimensional suitability between the sub-
(
Fig. 8). Both TS-1 and Ti-MWW showed comparably low
strates and the zeolite channels, the polarity and solubility
of reactants and products, and the diffusion problems, the
hydrophilicity/hydrophobicityof titanosilicates may play an
important role in the liquid-phase oxidation especially in the
case of AAL, which is protic and hydrophilic unlike simple
alkenes. We investigated this subject by comparing the epox-
idation of AAL with that of simple alkenes. As shown in Ta-
ble 5, Ti-MWW showed the highest conversion for 1-hexene,
while Ti-Beta, less effective than Ti-MWW and TS-1 in
conversion of GLY to glycerol, irrespective of whether
H2O2 is present or not, suggesting the Ti hydroperoxo
Table 5
Epoxidation of simple alkenes with H O over various titanosilicates
a
2
2
Catalyst
Substrate
Conversion Product selectivity
H O
2 2
(mol%)
(mol%)
(mol%)
b
Epoxide Others Conv. Eff.
Ti-MWW 1-Hexene
Ti-MWW Cyclohexene
44.8
7.1
9.7
0.8
8.5
99
63
98
78
96
52
0
1
37
2
12
4
48.2 93.0
15.0 95.0
11.5 84.2
TS-1
1-Hexene
TS-1
Cyclohexene
1-Hexene
Cyclohexene
Cyclohexene
–
–
Ti-Beta
Ti-Beta
Ti-Beta
9.1 63.7
35.8 75.4
58.3 86.0
16.2
24.7
48
100
c
Fig. 8. NH -TPD profiles of TS-1 (a), reused Ti-MWW after 4 reac-
3
a
Cat., 50 mg; alkene, 10 mmol; MeCN, 10 ml; H O , 10 mmol for 1-
2
2
tion-regeneration cycles in AAL epoxidation (Si/Ti = 39, Si/B = 149) (b),
hexene and 5 mmol for cyclohexene; 333 K; time, 2 h.
b
fresh Ti-MWW (Si/Ti = 38, Si/B = 54) (c), directly calcined precursor
Glycols, monomethyl glycol ether, 2-cyclohexen-1-ol, and 2-cyclo-
(
Si/Ti = 21, Si/B = 13) (d), Ti-Beta (Si/Ti = 42, Si/Al > 1000) (e), and
hexen-1-one.
Ti, Al-Beta (Si/Ti = 40, Si/Al = 70) (f).
c MeOH was used as a solvent.

P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
325
area or intensity in 3200–3800 cm 1 corresponds to the to-
tal amount of silanols that possibly adsorb water molecules,
and consequently can be taken as a measure for evaluating
the hydrophilicity. Obviously, TS-1 and Ti-Beta contained
the lowest and the highest concentration of silanol groups,
respectively (Figs. 9a and e), which are consistent with the
well-known fact that TS-1 is much more hydrophobic than
Ti-Beta. Careful consideration then should be taken for Ti-
MWW because it was prepared by the acid treatment of a
lamellar precursor. The removal of B and Ti during the acid
treatment would produce hydroxyl nests at the framework
vacancy of tetrahedral cation, as shown in Scheme 4 for the
removal of B. Actually, directly calcined lamellar precur-
sors already had relatively high concentrations of silanols
probably due to the formation of defect sites between the
layers when the interlayer dehydroxylation upon calcination
is incomplete (Fig. 9b). Silanols increased further in con-
centration for the acid-treated and calcined Ti-MWW as a
result of deboronation (Fig. 9c). Nevertheless, the amount
of silanols on MWW samples was not as large as that on
the hydrophilic Ti-Beta. The band intensity for Ti-MWW
decreased after the reaction-regeneration cycles (Fig. 9d),
despite a progressive deboronation in the AAL epoxida-
tion as shown earlier (Fig. 7). The silanols on Ti-MWW,
unlike those on Ti-Beta formed in the crystallization, were
therefore presumed to be in the pseudo-stable states and par-
tially disappeared probably through the dehydroxylation to
form Si–O–Si bridging bonds, or by the migration of Si
into the vacancy at elevated temperatures (Scheme 4). This
rearrangement or so-called framework recrystallization is
commonly observed in the dealuminaion of zeolites [24–26],
and may explain why Ti-MWW is more hydrophobic than
Ti-Beta although it is slightly hydrophilic compared to TS-
−
AAL epoxidation (Table 2), catalyzed the epoxidation of 1-
hexene and simultaneously showed the highest activity in
cyclohexene epoxidation. The more effective Ti-Beta for the
bulky substrates is simply owing to its large pores impos-
ing less steric restriction than those of Ti-MWW and TS-1.
Considering the extremely low activity of Ti-Beta in AAL
epoxidation especially in the presence of H2O (Table 2) and
in the epoxidation of 1-hexene compared to Ti-MWW and
TS-1 (Table 5), the issue of hydrophilicity/hydrophobicityis
raised. Beta zeolite generally contains high concentration of
hydrophilic silanol groups at the defect sites formed within
the complex and severely intergrown crystallites. High hy-
drophilicity thereafter originating from the silanols would
favor the adsorption of protic molecules of solvent and sub-
strate, which could make these molecules hardly accessible
to the Ti sites. This probably accounts for the low specific
activity per Ti site in Ti-Beta. This then raises the question
of the hydrophilicity/hydrophobicity of Ti-MWW.
IR spectra of hydroxyl stretching vibration have been
measured to investigate the hydrophilicity of titanosilicates.
It has been verified that this indirect method is useful for
Ti-Beta synthesized by the dry-gel conversion method [4c].
Fig. 9 shows the IR spectra of various titanosilicates. The
−
1
band at 3740 cm
is due to the surface silanols, while
−1
the broad one around 3550 cm is attributed to the defect
site silanols of a hydrogen-bonded nature [4c,5a]. The band
1
. Considering Ti-MWW without deboronation contained a
certain amount of defect sites between the layers, the amount
of silanols, consequently the hydrophilicity, is conceivably
not so high within the sinusoidal 10-MR channels, i.e., the
main reaction space for the AAL epoxidation (see above).
The hydrophobicity of the intralayer 10 MR channels of Ti-
MWW would allow the protic molecules of AAL to diffuse
and reach the Ti active sites to cause epoxidation even in
H2O solvent. In contrast, AAL and H2O molecules prefer to
be adsorbed on the silanols within the hydrophilic channels
Fig. 9. IR spectra in the region of hydroxyl stretching vibration for TS-1 (a),
directly calcined precursor (Si/Ti = 21, Si/B = 13) (b), fresh Ti-MWW
(
Si/Ti = 38, Si/B = 54) (c), reused Ti-MWW after 4 reaction-regeneration
cycles in AAL epoxidation (Si/Ti = 39, Si/B = 149) (d), and Ti-Beta
Si/Ti = 42, Si/Al > 1000) (e). The spectra were recorded at room
temperature after the evacuation at 773 K for 1 h.
(
Scheme 4.

326
P. Wu, T. Tatsumi / Journal of Catalysis 214 (2003) 317–326
of Ti-Beta, which makes AAL hardly accessible to Ti sites,
retarding the epoxidation.
[7] A. Corma, Chem. Rev. 97 (1997) 2373.
[8] (a) P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, Chem. Lett. (2000)
74;
7
(
(
(
(
(
b) P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, J. Phys. Chem. B 105
2001) 2897;
c) P. Wu, T. Tatsumi, Chem. Commun. (2001) 897;
d) P. Wu, T. Tatsumi, J. Phys. Chem. B 106 (2001) 748;
e) P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, J. Catal. 202 (2001)
4
. Conclusion
Ti-MWW catalyzes the epoxidation of AAL to GLY
more effectively and selectively than TS-1 and Ti-Beta in
the solvents of H2O or MeCN. It is conceivable that the
unique tortuosity makes the sinusoidal 10-MR channels of
Ti-MWW more suitable for molecular diffusion, and also
allows the Ti sites to be more accessible to the carbon–
carbon double bonds of substrate molecules than those in the
tunnel-like channels of TS-1. This explains the higher intrin-
sic activity of Ti-MWW when compared to TS-1. Ti-MWW
is more hydrophilic than TS-1 due to the silanols on the ex-
terior surface and on the defect sites formed between the
layers, but the space within the sinusoidal 10-MR channels
where the AAL epoxidation may mainly occur is essentially
hydrophobic. The present study shows that the steric fitness
between the substrate molecules and the Ti active sites and
the hydrophobic environment of Ti sites is simultaneously
important for a titanosilicate when it is employed as a cata-
lyst for the liquid-phase oxidation of hydrophilic substrates
especially in solvents of a protic nature.
2
45.
[
9] K. Weissermel, H.J. Arpe, Industrielle Organische Chemie, VCH,
Weinheim, 1994.
[10] H.A. Wittcoff, B.G. Reuben, in: Industrial Organic Chemicals, Wiley,
New York, 1996, p. 189.
11] M.G. Clerici, P. Ingallina, J. Catal. 140 (1993) 71.
12] M. Taramasso, G. Perego, B. Notari, US Patent 4410501, 1983.
13] (a) G.J. Hutchings, D.F. Lee, Chem. Commun. (1994) 1095;
(b) G.J. Hutchings, D.F. Lee, A.R. Minihan, Catal. Lett. 33 (1995) 369;
(c) G.J. Hutchings, D.F. Lee, A.R. Minihan, Catal. Lett. 39 (1996) 83.
14] (a) T. Tatsumi, K.A. Koyano, Y. Tanaka, S. Nakata, Stud. Surf. Sci.
Catal. 117 (1998) 143;
[
[
[
[
(b) T. Tatsumi, K.A. Koyano, N. Igarashi, Chem. Commun. (1998)
3
25.
[15] A. Corma, P. Esteve, A. Martínez, J. Catal. 161 (1996) 11.
[16] T. Tatsumi, K.A. Koyano, Y. Shimizu, Appl. Catal. A 200 (2000) 125.
17] C.B. Khouw, M.E. Davis, J. Catal. 151 (1995) 77.
18] (a) M.A. Camblor, M. Costantini, A. Corma, L. Gilbert, P. Esteve, A.
Martínez, S. Valencia, Chem. Commun. (1996) 1339;
[
[
(
b) T. Blasco, M.A. Camblor, A. Corma, P. Esteve, J.M. Guil, A.
Martínez, J.A. Perdigón-Melón, S. Valencia, J. Phys. Chem. B 102
1998) 75.
(
[19] (a) G. Bellussi, A. Carati, M.G. Clerici, G. Maddinelli, R. Millini,
References
J. Catal. 133 (1992) 220;
(b) M.G. Clerici, P. Ingallina, J. Catal. 140 (1993) 71;
[
[
[
1] G. Bellussi, M.S. Rigguto, Stud. Surf. Sci. Catal. 85 (1994) 177.
2] B. Notari, Adv. Catal. 41 (1996) 253.
3] (a) M.A. Camblor, A. Corma, A. Martínez, J. Pérez-Pariente, J. Chem.
Soc. Chem. Commun. (1992) 589;
(
(c) M.G. Clerici, P. Ingallina, R. Millini, in: R. von Ballmoos, et
al. (Eds.), Proceedings of the 9th International Zeolite Conference,
Butterworth–Heinemann, Woburn, MA, 1993, p. 445.
[20] A. Carati, C. Flego, E. Previde Massara, R. Millini, L. Carluccio, G.
Bellussi, Micropor. Mesopor. Mater. 30 (1999) 137.
b) A. Corma, M.A. Camblor, P.A. Esteve, A. Martínez, J. Pérez-
Pariente, J. Catal. 145 (1994) 151;
c) A. Corma, P. Esteve, A. Martínez, S. Valencia, J. Catal. 152 (1995)
8.
4] (a) T. Tatsumi, Q.H. Xia, N. Jappar, Chem. Lett. (1997) 677;
[21] L.Y. Chen, G.K. Chuah, S. Jaenicke, Catal. Lett. 50 (1998) 107.
[22] (a) L. Davies, P. McMorn, D. Bethell, P.C.B. Page, F. King, F.E.
Hancock, G.J. Hutchings, Chem. Commun. (2000) 1807;
(b) L. Davies, P. McMorn, D. Bethell, P.C.B. Page, F. King, F.E.
Hancock, G.J. Hutchings, Phys. Chem. Chem. Phys. 3 (2001) 632.
[23] (a) M.E. Leonowicz, J.A. Lawton, S.L. Lawton, M.K. Rubin, Sci-
ence 264 (1994) 1910;
(
1
[
(
(
b) N. Jappar, Q.H. Xia, T. Tatsumi, J. Catal. 180 (1998) 132;
c) T. Tatsumi, N. Jappar, J. Phys. Chem. 102 (1998) 7126.
[5] (a) P. Wu, T. Komatsu, T. Yashima, J. Phys. Chem. 100 (1996) 10316;
(
b) P. Wu, T. Komatsu, T. Yashima, J. Catal. 168 (1997) 400;
(b) S.L. Lawton, M.E. Leonowicz, P.D. Partridge, P. Chu, M.K. Rubin,
Micropor. Mesopor. Mater. 23 (1998) 109.
(c) P. Wu, T. Komatsu, T. Yashima, Stud. Surf. Sci. Catal. 105 (1997)
6
(
63;
d) P. Wu, T. Komatsu, T. Yashima, J. Phys. Chem. B 102 (1998) 9297.
6] M. Diaz-Cabanas, L.A. Villaescusa, M.A. Camblor, Chem. Commun.
2000) 761.
[24] G.T. Kerr, J. Phys. Chem. 73 (1969) 2780.
[25] P.K. Maher, F.D. Hunter, J. Scherzer, J. Adv. Chem. Ser. 101 (1971)
266.
[
(
[26] P. Wu, T. Komatsu, T. Yashima, J. Phys. Chem. 99 (1995) 10923.