Vapor-phase epoxidation of propene using H2 and O2 over Au/Ti-MCM-48

Article abstract of DOI:10.1006/jcat.2002.3642

Vapor-phase epoxidation of propene using H₂ and O₂ over homogeneously dispersed gold particles deposited by a deposition-precipitation (DP) method on the supports Ti-MCM-41 (hexagonal) and Ti-MCM-48 (cubic) is studied at a space velocity of 4000 h^-1·cm^-3/g(cat.) and at reaction temperature of 100 or 150°C. Influences of the various parameters investigated in the case of Ti-MCM-48 support are Si/Ti (30, 50, 75, and 100) ratios, precipitating agents (LiOH, NaOH, KOH, RbOH, or CsOH) for Au deposition, pH (6-8), Au loading (2-25 wt% in solution), and calcination temperature (150-500°C). The supports were characterized by XRD, UV-vis, FT-IR, and surface area measurements, whereas supported gold catalysts were characterized by TEM, ICP, and XPS techniques. Better performance was observed with Ti-MCM-48 (initial conversion = 5.6%, PO selectivity = 92%), due to its three-dimensional pore system, than with Ti-MCM-41 (initial conversion = 5.1%, PO selectivity = 88%) in the propene epoxidation at 150°C. GC-MS investigation of the extracted species using organic solvent from the used catalyst revealed that acidic as well as oligomeric species accumulated on the catalyst surfaces. These species are assumed to cause the catalyst deactivation. Silylation of the catalyst Au/Ti-MCM-48 prohibits it from getting deactivated faster and also helps to improve PO selectivity and decrease H₂ consumption. Based on the above experimental results, a probable reaction mechanism is explained.

Full text of DOI:10.1006/jcat.2002.3642

Journal of Catalysis 209, 331–340 (2002)
doi:10.1006/jcat.2002.3642
Vapor-Phase Epoxidation of Propene Using H2
and O2 over Au/Ti–MCM-48
B. S. Uphade, T. Akita, T. Nakamura, and M. Haruta1,2
Osaka National Research Institute, AIST, Midorigaoka 1-8-31, Ikeda 563-8577, Japan
Received August 16, 2001; revised April 8, 2002; accepted April 23, 2002
recently built plants for the commercial production of PO
are, therefore, based on hydroperoxide processes. The dis-
advantage of the hydroperoxide process is, however, the
production of equimolar amounts of the coproduct, either
styrene or tert-butanol, depending on which variant of hy-
droperoxide is used in the process.
Vapor-phase epoxidation of propene using H2 and O2 over
homogeneously dispersed gold particles deposited by a deposition–
precipitation(DP)methodonthesupportsTi–MCM-41(hexagonal)
and Ti–MCM-48 (cubic) is studied at a space velocity of
4000 h−1 · cm−3/g(cat.) and at reaction temperature of 100 or 150◦C.
Influences of the various parameters investigated in the case of Ti–
MCM-48 support are Si/Ti (30, 50, 75, and 100) ratios, precipitating
agents(LiOH, NaOH, KOH, RbOH, orCsOH)forAudeposition, pH
(6–8), Au loading (2–25 wt% in solution), and calcination temper-
ature (150–500◦C). The supports were characterized by XRD, UV–
vis, FT-IR, and surface area measurements, whereas supported gold
catalysts were characterized by TEM, ICP, and XPS techniques.
Better performance was observed with Ti–MCM-48 (initial conver-
sion = 5.6%, PO selectivity = 92%), due to its three-dimensional
pore system, than with Ti–MCM-41 (initial conversion = 5.1%, PO
selectivity = 88%) in the propene epoxidation at 150◦C. GC–MS
investigation of the extracted species using organic solvent from
the used catalyst revealed that acidic as well as oligomeric species
accumulated on the catalyst surfaces. These species are assumed
to cause the catalyst deactivation. Silylation of the catalyst Au/Ti–
MCM-48 prohibits it from getting deactivated faster and also helps
to improve PO selectivity and decrease H2 consumption. Based on
the above experimental results, a probable reaction mechanism is
In the past, extensive efforts have been made to develop
an alternative direct gas-phase epoxidation process similar
tothatfortheepoxidationofethene(2). However, therehas
so far been no alternative process developed, probably due
to low yields, low raw-material conversion, low productiv-
ities, numerous by-products, and complex catalyst systems.
One of the most important heterogeneous catalysts, CsCl-
promoted Ag/Al2O3, which is used in the epoxidation of
ethylene, could not be used for higher olefins containing
allylic hydrogen. This is mainly due to the easy abstrac-
tion of methyl hydrogen giving allylic intermediates, lead-
ing to direct combustion, thereby lowering PO selectivity.
Recently Enichem utilized TS-1 (MFI) as a catalyst for the
epoxidation of propene in the liquid phase using hydrogen
peroxide (3–5). However, due to the very high production
cost of H2O2 and its handling problems it is desirable to
have the H2O2 produced in situ. Toso Co. Ltd. (6) has de-
veloped Pd/TS-1 catalyst for the in situ generation of H2O2
from H2 and O2. Recently Ho¨lderich and co-workers (7)
modified the catalyst developed by Toso with platinum and
reported that the improved yield of PO could be ascribed
to the maintenance of palladium in its +2 oxidation state.
Our research work on the catalysis of gold (8–10) has
opened a new stage for the direct epoxidation of propene
using hydrogen and oxygen. We have reported recently, in a
series of papers, on the vapor-phase epoxidation of propene
over highly dispersed nanosize Au particles supported on
TiO2 (11, 12), TiO2/SiO2 (11, 12), and titanosilicates such as
TS-1, TS-2, Ti-β, and Ti–MCM-41 (13, 14). This finding is
now being followed by a few other researchers (15–17) and
companies (18, 19). We reported recently on the formation
of propanal from propene, H2, and O2 at high tempera-
tures (≥200◦C) (20). Recent findings from our group (21)
also showed the promoting effect of CsCl when physically
mixed with Au/Ti–MCM-41 for the effective utilization of
hydrogen.
c
explained. ꢀ 2002 Elsevier Science (USA)
Key Words: epoxidation; propene; propene oxide; gold; Ti–MCM-
41; Ti–MCM-48.
1. INTRODUCTION
Propene oxide (PO), which is also known by other
names, such as propylene oxide, methyl oxirane, and 1,2-
epoxypropane, is an important building block for the man-
ufacture of polyurethane, organic intermediates, and sol-
vents. Industrially PO is produced using two commercial
processes. These are the chlorohydrin process and the hy-
droperoxide process (1). The chlorohydrin process is not
environmentally friendly, as it produces chlorinated organic
by-products as well as calcium chloride, CaCl2. All the
1 To whom correspondence should be addressed. Fax: +81-298-61-8240.
E-mail: m.haruta@aist.go.jp.
2 Present address: Research Institute for Green Technology, AIST,
Onogawa 16-1, Tsukuba 305-8569, Japan.
331
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

332
UPHADE ET AL.
Here, we report on the epoxidation of propene over Au for 10 days. The product was collected by centrifugation,
supported on Ti–MCM-48 (cubic). We already have re- washed repeatedly with water, dried at 100◦C overnight,
ported, in a detailed study, on the use of Au/Ti–MCM-41 and finally calcined at 540◦C for 6 h in static air.
in propene epoxidation (22). New results on the epoxida-
tion of propene over Au/Ti–MCM-41 (Si/Ti = 67), which
2.2. Synthesis of Cubic Ti–MCM-48 Support
were additional to and different from those reported ear-
Ti–MCM-48 with various Si/Ti ratios was synthesized un-
lier, are included in this paper for comparison purposes. In
der hydrothermal conditions at 100◦C in a static teflon bot-
the family of M41S materials, first reported by Mobil in 1992
tle for 10 days according to the literature procedure (29).
(23, 24), MCM-41 (hexagonal arrangement of unidirec-
A typical two-step method for its synthesis is as follows. To
tional pores indexed in the space group p6m) and MCM-48
a 44% solution of TEOS in IPA, CTMAOH in methanol
(cubic structure indexed in the space group Ia3d) are very
and water (water/methanol molar ratio = 2) was added
widely studied, both as such and when substituted with re-
to partly hydrolyze TEOS at about 2–5◦C. After 1 h, the
dox metals such as Ti and V (25). Earlier studies on gas-
IPA solution of TBOT was added to the resultant mixture
phase epoxidation of propene over Au/Ti–MCM-41 (22)
very slowly under vigorous stirring. The mixture was then
showed the fast catalyst deactivation, partly due to the
stirred for 1 h, when the aqueous solution of CTMACl
monodimentional array of pores in MCM-41. In accordance
was finally added. The mixture was heated to 50◦C for
with previous reports (26, 27) we are of the opinion that
about 6–8 h to remove IPA. The molar composition of the
MCM-48, due to its three-dimentional, branched-pore sys-
gel subjected to hydrothermal synthesis (100◦C, 10 days)
tem, would be more resistant to clogging than MCM-41,
was as follows: SiO2, xTiO2, 1.2CTMAOH, 0.7CTMACl,
which has straight channels. It is, therefore, the aim of
and 46.6H2O, where x = 0.01–0.03, Si/CTMA = 1.0, and
this study to compare propene epoxidation over Au sup-
CTMACl/OH = 70/30.
ported on Ti–MCM-41 and Ti–MCM-48, with emphasis on
the latter system. Various factors affecting catalyst activ-
ity/selectivity in the support Ti–MCM-48, such as Si/Ti ratio,
2.3. Deposition of Au onto Supports
precipitating agent, and pH for Au deposition, Au loading,
and calcination temperature, are discussed in detail. Possi-
ble causes of catalyst deactivation, use of silylation to avoid
faster catalyst deactivation, and probable reaction mecha-
nisms are also discussed.
Nanoparticles of Au were deposited on the supports
Ti–MCM-41andTi–MCM-48byadeposition–precipitation
(DP) method as described previously (30). In a typical cata-
lyst preparation condition an aqueous solution of HAuCl4 ·
4H2O, 0.17 g (8 wt% Au with respect to support), in 200 ml
H2O was heated to 70◦C and the pH was adjusted to
7.0 0.1 using an aqueous alkali hydroxide solution. The
support, weighing 1.0 g, was suspended to it, the pH was
adjusted once again to the same value, and the solution
was aged at the above temperature for 1 h. The suspen-
sion was washed several times with distilled water, dried at
100◦C for 2 h, and finally calcined in air at 300◦C for 4 h.
2. EXPERIMENTAL
2.1. Synthesis of Hexagonal Ti–MCM-41 Support
Ti–MCM-41 with a Si/Ti atomic ratio of 67 was synthe-
sized according to the patent procedure (28). The molar
composition subjected to hydrothermal synthesis used was
as follows: SiO2, 0.015TiO2, 0.75CTMACl, 0.27TMAOH,
and 71.8H2O. In a typical synthesis, 40.6 g tetraethyl-
orthosilicate (TEOS) and 1.0 g tetrabutylammoniumti-
2.4. Characterization of the Supports and Gold Catalysts
The supports, Ti–MCM-41 and Ti–MCM-48, were char-
tanate (TBOT) were hydrolyzed at 80◦C for 5 h in N2 at- acterized by XRD (Rigaku Rint-2400, Cu-Kα radiation,
mosphere under continuous stirring. After cooling to close 40 kV, 40 mA), UV–vis (Photal Otsuka Electronics, MC-
to room temperature, 47 g of a structure-directing organic 2530 UV/vis light source), FT-IR (Nicolet Model 205, KBr
template, cetyltrimethylammonium chloride (CTMACl), pellets containing 1% finely powdered support pressed at
was added and the mixture was dissolved in 115 ml of 6 tons/cm2), and specific surface area (Quantasorb Jr.)
2-propanol (IPA). That mixture was stirred for 1 h and measurements to confirm the product identity reported
then a mixture of 10 g tetramethylammonium hydroxide in the literature (31–34). Nitrogen adsorption isotherms
(TMAOH, 25% aq.) and 35 ml IPA was added dropwise at −196◦C were obtained using a Micromeritics ASAP
under vigorous stirring. This mixture was stirred overnight. 2010 volumetric adsorption analyzer. Transmission elec-
Finally the remaining 8.9 g TMAOH and 252 g water were tron microscopy (TEM) observations were made using a
added and the mixture was heated to 50◦C and kept at this Hitachi H-9000 to determine Au particle size and distribu-
temperature for 8–10 h or until the disappearance of the tion. XPS (Shimadzu ESCA-KM) was used to determine
IPA. After the water content was adjusted, the resulting theelectronicstateofAuparticlesandthecatalyst’ssurface.
mixture was transferred to a thick teflon bottle and heated Actual Au and Ti contents in the catalysts were analyzed
at 100◦C under autogeneous pressure and static conditions by an inductively coupled plasma (ICP) technique.

EPOXIDATION OF PROPENE OVER Au/Ti–MCM-48
333
2.5. Catalytic Activity Measurements
a) Ti-MCM-48 (Si/Ti=30)
500
400
300
200
100
0
Catalytic tests were carried out in a vertical fixed-bed
U-shaped quartz reactor (i.d., 10 mm) using a feed con-
taining 10 vol% each C3H6, H2, and O2 diluted with Ar
passed over the catalyst (0.5 g) bed at a space velocity of
4000 h−1 · cm−3/g catalyst. The temperature was controlled
and measured using a glass-tube-covered Cr–Al thermo-
couple located in the center of the catalyst bed. Prior to test-
ing, the catalysts were first pretreated at 250◦C for 30 min
in a stream of 10 vol% H2 in Ar, followed by 10 vol% O2 in
Ar. In the catalytic tests for the samples calcined at different
temperatures, from 150–500◦C, the pretreatment was car-
ried out only in a stream of 10 vol% O2 in Ar. The feed and
products were analyzed using three online GCs equipped
with TCD (AC and porapak Q columns) and FID (HR-
20M column) detectors equipped with automatic injection
systems. Since the conversions of propene tended to de-
crease during catalytic activity measurements (see Fig. 6),
in principle, the conversions at 30 min after the start of the
reactant gas feed are presented in this paper.
Trimethylsilylation (35) of Au/Ti–MCM-48 (Si/Ti = 50)
at 150◦C was carried out using methoxytrimethyl silane.
Silylation was performed as per the patent procedure (36).
Methoxytrimethyl silane was kept at 0◦C and argon was
bubbled at the flow rate of 25 ml/min through the silylating
solution for 15 min. The catalyst was then flushed with pure
argon for about 5 h at 200◦C before the catalytic tests were
carried out.
0.0
0.2
0.4
0.6
0.8
1.0
Relative Pressure (P/P0)
b) Ti-MCM-48 (SI/Ti=100)
600
500
400
300
200
100
0
0.0
0.2
0.4
0.6
0.8
1.0
Relative Pressure (P/P0)
FIG. 1. Nitrogen adsorption–desorption isotherm at −196◦C for
Ti–MCM-48. (a) Si/Ti = 30; (b) Si/Ti = 100.
˚
pore diameters in the range of ca. 37–40 A, as shown in
Table 1. Ti–MCM-41 exhibits a well-defined (1 0 0) reflec-
tion; that and Ti–MCM-48 (2 1 1) and (2 2 0) reflections
can be seen in Fig. 2. UV–vis confirmed the absence of a
3. RESULTS AND DISCUSSION
3.1. Characterization of Supports and Supported
Au Catalysts
3.1.1. N2 adsorption isotherms. Nitrogen adsorption
isotherms for Ti–MCM-41 support are explained in a pre-
vious paper (22). For Ti–MCM-48 the isotherms are shown
in Figs. 1a and 1b for Si/Ti ratios of 30 and 100, respectively.
The isotherms of type IV, typical of mesopore materials,
with H2 hysteresis, are observed. At the relative pressure
near P/P0 0.15–0.25, the isotherm exhibits the inflection
characteristic of capillary condensation within the meso-
pores (37). The BJH cumulative surface area, pore volume,
and pore diameter for Si/Ti = 50 were 1649 m2 g−1, 1.3 cm3
(211)
(220)
(100)
Ti-MCM-48
g
−1, and 3.24 nm, respectively, whereas for Si/Ti = 30
the values were 1053 m3 g−1, 0.76 cm3 g−1, and 2.88 nm,
respectively.
3.1.2. XRD, UV–vis, and FT-IR. XRD and UV–vis
spectra for the supports Ti–MCM-41 and Ti–MCM-48 are
compared in Figs. 2 and 3, respectively. Samples give rela-
tively well-defined XRD patterns typical of MCM-41 and
MCM-48 materials, as reported in the literature (31–34).
XRD peaks can be indexed on a hexagonal lattice for
Ti–MCM-41 and on a cubic lattice for Ti–MCM-48 with
(110)
(200)
Ti-MCM-41
8
2
4
6
2θ (degrees)
FIG. 2. XRD spectra of Ti–MCM-41 and Ti–MCM-48.

334
UPHADE ET AL.
e) Si/Ti = 30
d) Si/Ti = 50
c) Si/Ti = 75
c) Si/Ti=50
b) Si/Ti=100
b) Si/Ti = 100
a) Si/Ti=67
300
400
500 600
a) Si/Ti=67
Wavelength ( nm )
FIG. 3. UV–vis spectra of Ti–MCM-41 [(a) Si/Ti = 100/1.5] and
Ti–MCM-48 [(b, c) Si/Ti = 50 and 100].
1600 1400
1200 1000
Wavenumber ( cm-1)
800
600
segregated TiO2 phase at ca. 330 nm for supports with a tita-
nium content ≤2.0 wt%. According to the literature (31, 32)
the band at 210 nm is assigned to Ti in tetrahedral coordina-
tion whereas the band in the range 270–330 nm is assigned
to the Ti in its octahedral coordination. The broad wave-
length peak for curve c in Fig. 3 may be indicative of Ti in a
distorted tetrahedral environment and the presence of Ti in
FIG. 4. FT-IR spectra of Ti–MCM for different Si/Ti ratios. (a) Ti–
MCM-41; (b–e) Ti–MCM-48.
polymeric octahedral coordination is also possible for this
material.
FT-IR spectra (Fig. 4) showed the presence of sharp peak
at 963 cm−1 indicative of Ti in its tetrahedral coordination
TABLE 1
Preparation Conditions and Characterization of Gold Supported on Hexagonal Ti–MCM-41 and Cubic Ti–MCM-48
Loading (wt%)
Porea
Surface
area
Au
(in soln.)
(wt%)
by ICP
b
Si/Ti
ratio
diameter
Ppting
agent
DAU
Support
(A)
(m2 · g−1
)
pH
Au
Ti
(nm)
˚
Ti–MCM-41
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
Ti–MCM-48
67
100
75
50
50
50
50
50
50
50
50
50
50
50
30
37.1
39.1
39.2
39.4
39.4
39.4
39.4
39.4
39.4
39.4
39.4
39.4
39.4
39.4
39.3
970
1117
n.d.
12
8
8
2
8
12
16
16
16
16
24
16
16
16
8
NaOH
NaOH
NaOH
NaOH
NaOH
NaOH
NaOH
NaOH
KOH
NaOH
NaOH
NaOH
LiOH
7.0
7.0
7.0
7.0
7.0
7.0
6.0
7.0
7.0
8.0
7.0
7.0
7.0
7.0
7.0
0.42
0.04
0.06
0.01
0.09
0.11
n.d.
n.d.
n.d.
0.11
0.30
n.d.
0.07
0.35
0.32
1.14
n.d.c
n.d.
1.10
1.12
1.15
n.d.
n.d.
n.d.
1.15
1.11
n.d.
1.00
1.13
n.d.
2.1
2.3
n.d.
n.d.
2.2
1127
1127
1127
1127
1127
1127
1127
1127
1127
1127
1127
n.d.
n.d.
2.6
4.7d
2.4
2.0
n.d.
2.0
2.3
2.3
CsOH
NaOH
n.d.
a By XRD.
b A deviation of 1–3 nm is observed.
c n.d. = not determined.
d Final catalyst calcined at 150◦C.

EPOXIDATION OF PROPENE OVER Au/Ti–MCM-48
335
(22). However, as reported elsewhere (32) the peak at It may be noted from Table 1 that the Ti–MCM-48 with
963 cm−1 could also be assigned to Si–OH groups in ad- a cubic structure deposits less Au loading than hexagonal
dition to Ti–O–Si bonds and, hence, we calcined materi- Ti–MCM-41.
als (MCM-41, MCM-48, Ti–MCM-41, and Ti–MCM-48) at
higher temperature. The peak at 963 cm−1 disappeared for
MCM-41 and MCM-48 but was maintained for Ti–MCM
3.2. Epoxidation of Propene over Au/Ti–MCM Catalysts
The results of propene epoxidation at 150◦C over Au/Ti–
MCM-41 and Au/Ti–MCM-48 are compared in Fig. 7. For
both catalysts the initial propene conversion is about 5%
and it decreases with time. The superior performance of
Au/Ti–MCM-48 in terms of activity, PO selectivity, and
also H2 efficiency over that of Au/Ti–MCM-41 can be at-
tributed to its three-dimensional pore system, which should
be more resistant to blockage by extraneous materials like
oligomers of PO than the one-dimensional pore system of
Ti–MCM-41. Koyano and Tatsumi (29) also have shown in
when calcined at 1000◦C. The disappearance of the band
in pure silicious MCM-41 and MCM-48 when calcined at
higher temperature may be attributed to the loss of silanol
groups, as reported previously (22). Moreover, the inten-
sity of the peak at 963 cm−1, assumed to be due to silanol
groups, clearly shows an increase in its intensity when tita-
nium is present in MCM-41 and MCM-48. Thus, it is clear
from our earlier (22) and present study that the major part
of titanium is present in the tetrahedral coordination.
3.1.3. ICP and TEM. Actual Au loading by ICP ana- the liquid-phase epoxidation of cyclododecene with H2O2
lysiswasverysmallincomparisonwiththeAucontentinthe the higher activity of Ti–MCM-48 over that of Ti–MCM-41.
DP solution especially on Ti–MCM-48 (Table 1). TEM ob- Anpo et al. (37) also reported higher photocatalytic activ-
servations showed a homogeneous dispersion of Au with an ity for Ti–MCM-48 than for Ti–MCM-41 in the reduction of
average particle size of 2.0–2.6 nm for Au/Ti–MCM-41 and CO2 withH2OtoCH3OHandconcludedthatthelargepore
Au/Ti–MCM-48. Typical TEM images of the Au supported size and three-dimentional channel structure of MCM-48
on Ti–MCM-41 and Ti–MCM-48 are compared in Figs. 5a results in better dispersion of titanium dioxide and thus the
and 5b, respectively, whereas the size distributions of Au better results.
particles are shown in Figs. 6a and 6b. Average sizes of Au
The results of the influence of Si/Ti ratio in Ti–MCM-48
particles calculated for various samples are given in Table 1. on the epoxidation of propene over 8% Au/Ti–MCM-48 at
FIG. 5. TEM of (a) Au/Ti–MCM-41 and (b) Au/Ti–MCM-48.

336
UPHADE ET AL.
80
70
60
50
40
30
20
10
0
a) Ti-MCM-41
98
97
Si/Ti=67
96
95
20
16
1.0
2.0
3.0
4.0
5.0
H
O
Diameter ( nm )
2
30
25
12
8
2
b) Ti-MCM-48
Si/Ti=100
20
15
10
5
3.0
2.5
2.0
1.0
0
5.0
1.0
2.0
3.0
4.0
Diameter ( nm )
30
40 50 60 70 80
Si/Ti ratio in support
90 100
FIG. 6. Size distribution of Au particles supported on (a) Ti–MCM-41
and (b) Ti–MCM-48.
FIG. 8. Influence of Si/Ti ratio in Ti–MCM-48 on the epoxidation of
propene over 8 wt% Au/Ti–MCM-48 at 100◦C.
92
Au/Ti-MCM-48
Au/Ti-MCM-41
90
88
86
100◦C are given in Fig. 8. The conversion of H2 and O2 de-
creases whereas PO selectivity increases with an increase in
Si/Ti ratio. This can be ascribed to the decrease in Au load-
ing with an increase in Si/Ti ratio due to lower Ti content, as
shown in Table 1. The propene conversion passes through
a maximum at a Si/Ti ratio of 50. Along similar lines, we re-
ported earlier for Au/Ti–MCM-41 an increase in Au load-
ing on Ti–MCM-41 with a decrease in Si/Ti atomic ratio
and hence poor PO selectivity at very high Ti and Au load-
ings (please see Table 1 and Fig. 11 of Ref. 22). A further
decrease in Si/Ti ratio causes a decrease in propene conver-
sion, which is most probably due to the presence of a poly-
mericanataseTiO2 phase, asobservedbyUV–visandTEM.
We have also found that the best results in propene epoxi-
dation are obtained when NaOH is used as a precipitating
agent for depositing gold on the supports. The results are
presented in Fig. 9. One probable reason for this could be
uniform and high dispersion of Au nanoparticles coupled
with traces of alkali content. The more basic nature of Na
compared with that of Cs may be another reason for higher
PO selectivity. It should also be noted from Table 1 that the
60
50
40
30
5.0
4.5
4.0
3.5
3.0
2.5
0
20
40 60
80 100 120
Time (min)
FIG. 7. Influence of support (Ti–MCM-41 vs Ti–MCM-48) in the
epoxidation of propene. Au in solution, 12 wt%; temperature, 150◦C.

EPOXIDATION OF PROPENE OVER Au/Ti–MCM-48
337
TABLE 2
60
55
50
95
Influence of Catalyst Calcination Temperature
on the Epoxidation of Propene
90
85
80
Selectivity
(%) for
Temp. (◦C) for
Conversion (%) ofc
45
40
Calcinationa
Pretreatmentb
C3H6
H2
O2
PO
CO2
150
200
300
400
500
150
200
250
250
250
1.73
2.86
3.03
1.87
1.50
27.3
26.9
21.8
17.3
15.5
19.5
15.2
13.5
11.6
10.0
89
92
92
93
94
8.0
6.0
5.6
5.0
4.8
5.0
4.5
4.0
3.5
3.0
Note. Catalyst, 16 wt% Au/Ti–MCM-48 (Si/Ti = 50); reaction temper-
ature, 100◦C.
a In air.
b In 10 vol% O2 in Ar.
c After 30 min of reaction.
LiOH NaOH KOH RbOH CsOH
Precipitating agent
92
FIG. 9. Influence of precipitating agent during Au deposition on
Ti–MCM-48 (16 wt% Au in solution, Si/Ti = 50) on the epoxidation of
propene at 150◦C.
90
88
86
Au loading on cubic Ti–MCM-48 increases when the pre-
cipitating agent is changed from LiOH to CsOH. The worst
performance of the catalyst occurring when CsOH is used
to deposit gold may, therefore, be related to the highest
gold loading. To run the reaction at higher temperature the
decrease in both Au and titanium loadings is the most im-
portant parameter. We have also found that a pH of about
7.0 0.1 works best to obtain the better performance of
the catalyst in the propene epoxidation reaction (Fig. 10).
This is most probably due to the higher dispersion of Au
particles and the appropriate amount of Au loading on the
support.
84
82
40
35
H
2
30
25
O
20
2
The results of the influence of Au loading in the catalyst
Au/Ti–MCM-48 (Si/Ti = 50) on the epoxidation of propene
at 150◦C are given in Fig. 11. With an increase in Au load-
ing, propene, H2, and O2 conversions increase whereas PO
selectivity decreases. Propene conversion almost levels off
at the 16% Au loading in the solution. The low Au loading
of 2.0 wt% in the solution, which gave an actual loading of
0.01 wt% (Table 1), is also not good, as it is not sufficient
to obtain a good yield. At the Au loading below 2 wt% in
solution (0.01 wt% actual) formation of propane was ob-
served. We reported earlier (39) that at the lower Au load-
ing of ≤0.2 wt% propane formations occur over Au/TiO2
due to the formation of Au particles smaller than 2 nm in
diameter.
15
4.5
3.5
2.5
1.5
6.0
6.4
6.8
7.2
7.6
8.0
pH of Au solution
The results of the catalyst calcination temperature on the
epoxidation activity are presented in Table 2. Propene con-
FIG. 10. Influence of pH during deposition of Au on Ti–MCM-48
version passes through a maximum for the catalyst calcined (Si/Ti = 50) on the epoxidation activity at 150◦C.

338
UPHADE ET AL.
of the reaction suggests that the catalyst deactivation oc-
98
94
90
curs mainly due to the adsorption of PO, not of propene.
A GC-MS study indicated the presence of a large num-
ber of organic compounds adsorbed on the catalyst sur-
face. These are acetic acid, propanoic acid, 1-acetoxy-2-
propanol, 2-acetoxy-1-propanol, 1-hydroxy-2-propanone,
propylene glycol, propylene carbonate, 2,5-dimethyl-1,4-
dioxane, and so forth. We assume that these species are
formed due to initial PO adsorption on the catalyst sur-
face and its further oligomerization, rearrangement, crack-
ing, coupling, and so forth, on the free titanium and acidic
sites. As in the case of Ti/SiO2 support (17), acidic species
such as acetic acid and propanoic acid may also contribute
to the catalyst deactivation. Earlier studies (40) on liquid-
phase epoxidation of propene with H2O2 over TS-1 using
methanol solvent shows that the major by-products respon-
sible for catalysts deactivation, 1-methoxy-2-propanol and
2-methoxy-1-propanol, are formed from PO by epoxide
ring opening and reaction with a nucleophile. It should be
noted that the catalyst deactivation in liquid- and gas-phase
epoxidation processes are caused by entirely different by-
products, as mentioned above, thus suggesting different
ways of catalyst deactivation in these two methods. Our
attempts to regenerate the deactivated catalysts by ther-
mal treatment at ≥250◦C in the oxygen stream and also by
dissolving organic moieties from the catalyst surface in or-
ganic solvents such as ethanol and acetone failed to recover
the original activity of the catalysts. Nijhuis et al. (15) also
indicated that the sequential oligomerization of propene
oxide leading to species like dioxanes and heavier hydro-
carbons may be the main cause of the lower epoxidation
yield.
H
2
60
40
O
2
20
6.0
4.0
2.0
5
10
15
20
25
Au (wt.%) in solution
FIG. 11. Influence of Au content in DP solution in the catalyst Au/Ti–
MCM-48 (Si/Ti = 50) on the epoxidation of propene at 150◦C.
at 300◦C. The increase in PO selectivity with an increase 3.4. Silylation of Au/Ti–MCM-48
in catalyst calcination temperature is expected due to the
In the silylation process active hydrogen of silanol is re-
stronger contact between the support, Ti–MCM-48, and the
Au particles. The surprising fact is that even the catalyst
calcined at 150◦C produces PO with a propene conversion
of about 1.7%. At such a low calcination temperature the
reduction of Au(OH)3 into its metallic state might not be
completed. Our attempt to detect the presence of Au+ and
Au3+ species by XPS for the catalyst calcined at 150◦C was
unsuccessful due to highly dispersed Au particles, mostly in-
sidethechannels, eventhoughtheAuloadingwasveryhigh.
The results in the epoxidation of propene, however, may in-
dicate that the reduction of Au(OH)3 can take place easily
during pretreatment and/or reaction due to the reducing at-
mosphere. The inferior performance of the catalyst calcined
at 150◦C can be correlated to the somewhat lower disper-
sion and bigger size of Au particles, as given in Table 1.
placed by an alkylsilyl group, making the material more
hydrophobic. Corma and co-workers (40) showed earlier
that the hydrophilic nature of Ti–MCM-41 makes it less
active than Al–Ti–beta in aqueous H2O2 oxidation of alka-
nes and alkenes. Silylation using various volatile and eas-
ily removable silylating agents has improved hydropho-
bicity of Ti–MCM-41, resulting in large enhancement of
epoxide yield, as reported in recent publications (42–44).
In our study, silylation of the Au/Ti–MCM-48 performed
using methoxytrimethyl silane improved PO selectivity and
also substantially decreased H2 consumption, but the initial
propene conversion was also decreased (Fig. 12). The ini-
tial decrease in propene conversion could be related to the
covering of active Au–Ti sites. It would be ideal if hydrogen
was consumed on an equimolar basis to that of propene;
however, due to higher temperature direct hydrogen oxi-
dation takes place preferentially, resulting in low hydrogen
3.3. Catalysts Deactivation and Their Regeneration
A major problem associated with this system is the efficiency. That silylation of the catalyst Au/Ti–MCM-48
fast catalyst deactivation. The experimental results of the results in the improvement of hydrogen consumption ef-
preadsorbed PO or propene on the catalysts before the start ficiency is a good sign and hence critical experiments are

EPOXIDATION OF PROPENE OVER Au/Ti–MCM-48
339
which we proposed for Au/TiO2 (12). Accordingly, hydro-
gen peroxide formed on the surface of gold would form
hydroperoxolike species on the catalyst surface having iso-
lated tetrahedral titanium sites. Hydroperoxolike species
are expected to react with propene preadsorbed on the
SiO2 surface, giving PO formation by the same mechanism
as reported for the liquid-phase epoxidation of propene us-
ing hydrogen peroxide (3). Moulijn and co-workers (15)
proposed three different types of mechanisms and finally
concluded that a hydroperoxidelike compound is more
likely to be formed, creating a necessity for hydrogen. Our
proposed mechanism is very similar to that recently pro-
posed by Moulijn and co-workers (15). Detailed and criti-
cal experiments are needed in order to confirm the above
mechanism.
96
94
92
90
88
4.4
3.6
2.8
2.0
55
4. CONCLUSIONS
1. XRD, UV–vis, FT-IR, and the very high surface area
of above 1000 m2 g−1 confirm the structural integrity and
mesoporous nature of the supports Ti–MCM-41 (cubic)
and Ti–MCM-48 (cubic). No segregated TiO2 phase in the
supports were observed and Ti was present as an isolated
species in its tetrahedral coordination as long as the tita-
nium content Si/Ti ≥ 50.
45
35
25
15
2. Finely dispersed Au with an average particle size of
about 2–5 nm can be deposited homogeneously on both
supports, using the DP method followed by calcination in
air at 300◦C.
0
100
200
300
400 500
Time (min)
FIG. 12. Effect of trimethylsilylation of Au/Ti–MCM-48 catalyst on
the propene epoxidation activity. Au in solution, 8 wt%; Si/Ti = 50;
temperature = 150◦C.
3. The superior performance of Au/Ti–MCM-48 over
that of Au/Ti–MCM-41 in the vapor-phase epoxidation
of propene using H2 and O2 is attributed to its three-
dimensional (cubic) pore system, which can be more re-
sistant to blockage by extraneous materials like oligomers
of PO and bulkier organic compounds, as evidenced in
a GC-MS study, than the unidimensional pore system of
hexagonal Ti–MCM-41.
needed to make the commercial feasibility of this process
more realistic.
3.5. Probable Reaction Mechanism
4. In the case of Ti–MCM-48 support the best results are
obtained when the following parameters are chosen: Si/Ti
mole ratio = 50; precipitating agent = NaOH; Au loading
(in solution) = 16 wt% (0.3 wt% actual); pH of the Au
solution of 7.0 0.1; calcination temperature of the cata-
lyst = 300◦C for 4 h in air; pretreatment conditions =
250◦C for 30 min each, first in a flow of H2 and then in
O2 (10 vol%), both diluted with argon; and reaction
temperature =, 150◦C.
A theoretical prediction of Olivera et al. (44) suggests
that the surface of gold is capable of generating H2O2.
We have, therefore, proposed a probable reaction mech-
anism, as shown in Fig. 13, which is different from that
4+
[Ti -SiO ]
4+
---C H
C H + [Ti -SiO ]
2
2
3 6
3
6
[Au]
H
+ O
H O
2 2
2
2
5. Catalyst deactivation takes place not only due to the
large number of organic compounds on the catalyst surface
but also due to acidic compounds such as acetic acid.
6. Silylation helps improve PO selectivity and H2 effi-
ciency, probably due to an increase in the hydrophobicity
of the catalyst.
4+
H O + [Ti -SiO ]
4+
HOO---[-Ti -SiO ]---H
2
2
2
2
4+
[Ti -SiO ]---C H
3 6
4+
+
HOO---[-Ti -SiO ]---H
2
2
H
4+
+ H O + 2[Ti -SiO ]
H C
C
CH
2
3
2
2
O
7. A new reaction mechanism is proposed. Hydrogen
peroxide formed from H2 and O2 over the gold surface
is first converted into a hydroperoxolike species on the
FIG. 13. Proposed reaction mechanism for propene epoxidation over
Au/Ti–MCM catalysts.

340
UPHADE ET AL.
Ti4+–SiO2 surface, which reacts with propene preadsorbed 21. Uphade, B. S., Okumura, M., Tsubota, S., and Haruta, M., Appl. Catal.
A 190, 43 (2000).
on the SiO2 surface, giving PO.
22. Uphade, B. S., Yamada, N., Nakamura, T., and Haruta, M., Appl. Catal.
A 215, 137 (2001).
23. Kresge, C. T., Leonowicz, M. E., Roth, W. J., Vartuli, J. C., and Beck,
J. S., Nature 359, 710 (1992).
ACKNOWLEDGMENTS
24. Beck, J. S., Vartuli, J. C., Roth, W. J., Leonowicz, M. E., Kresge, C. T.,
Smith, K. D., Chu, C. T.-W., Olson, D. H., Sheppard, E. W., McCullen,
S. B., Higgins, J. B., and Schlenker, J. L., J. Am. Chem. Soc. 114, 10834
(1992).
The authors are greatly indebted to Mr. T. Hayashi and Mr. Wada of
Nippon Shokubai Co., Ltd., Japan, for their discussion and guidance in
silylation technique. B.S. Uphade gratefully acknowledges the financial
support of the Science and Technology Agency of Japan.
25. Morey, M. S., Davidson, A., and Stucky, G. D., J. Porous Mater. 5, 195
(1998).
REFERENCES
26. Zhang, W., and Pinnavaia, T. J., Catal. Lett. 38, 261 (1996).
27. Zhang, S. G., Fujii, Y., Yamashita, H., Tatsumi, T., and Anpo, M.,
Chem. Lett. 659 (1997).
1. Ainsworth, S. J., Chem. Eng. News 9 (1992).
2. van Santen, R. A., and Kuipers, H. P. C. E., Adv. Catal. 35, 265 (1987).
3. Clerici, M. G., Bellussi, G., and Romano, U., J. Catal. 129, 159 (1991).
4. Notari, B., Catal. Today 18, 163 (1993).
28. Watanabe, K., Onoda, Y., Tsuneki, H., and Tatsumi, T., Jpn. Patent
App. H07-300312.
29. Koyano, K. A., and Tatsumi, T., Stud. Surf. Sci. Catal. 105, 93 (1997).
30. Haruta, M., Tsubota, S., Kobayashi, T., Kageyama, H., Genet, M. J.,
and Delmon, B., J. Catal. 144, 175 (1993).
5. Occhiello, E., Chem. Ind. 761 (1997).
6. Sato, A., Miyake, T., and Saito, T., Shokubai 34, 132 (1992).
7. Meiers, R., Dingerdissen, U., and Ho¨lderich, W. F., J. Catal. 176, 376
(1998).
31. Corma, A., Navarro, N. T., and Perez-Pariente, J., J. Chem. Soc. Chem.
Commun. 147 (1994).
8. Haruta, M., Catal. Surv. Jpn. 1, 61 (1997), and references therein.
9. Haruta, M., Catal. Today 36, 153 (1997), and references therein.
10. Haruta, M., Stud. Surf. Sci. Catal. 110, 123 (1997), and references
therein.
32. Alba, M. D., Luan, Z., and Klinowski, J., J. Phys. Chem. 100, 2178
(1996).
33. Koyano, K. A., and Tatsumi, T., J. Chem. Soc. Chem. Commun. 145
(1996).
11. Hayashi, T., Tanaka, K., and Haruta, M., Shokubai 37, 72 (1995).
12. Hayashi, T., Tanaka, K., and Haruta, M., J. Catal. 178, 566 (1998).
13. Kalvachev, Y. A., Hayashi, T., Tsubota, S., and Haruta, M., Stud. Surf.
Sci. Catal. 110, 965 (1997).
14. Haruta, M., Uphade, B. S., Tsubota, S., and Miyamoto, A., Res. Chem.
Intermed. 24, 329 (1998).
34. Koyano, K. A., and Tatsumi, T., Microporous Mater. 10, 259 (1997).
35. Hayashi, T., Wada, M., Haruta, M., and Tsubota, S., Jpn. Pat. Appl.
H10-41833, PCT Pub. WO/JP99/00753.
36. Yuan, Z. Y., Liu, S. K., Chen, T. H., Wang, J. Z., and Li, H. X., J. Chem.
Soc. Chem. Commun. 973 (1995).
37. Anpo, M., Yamashita, H., Ikeue, K., Fuji, Y., Zhang, S. G., Ichihashi,
Y., Park, D. R., Suzuki, Y., Koyano, K., and Tatsumi, T., Catal. Today
44, 327 (1998).
15. Nijhuis, T. A., Huizinga, B. J., Makkee, M., and Moulijn, Ind. Eng.
Chem. Res. 38, 884 (1999).
16. Stangland, E. E., Stavens, K. B., Andres, R. P., and Delgass, W. N.,
J. Catal. 191, 332 (2000).
17. Mul, G., Zwijnenburg, A., van der Linden, B., Makkee, M., and
Moulijn, J. A., J. Catal. 201, 128 (2001).
38. Tanaka, K., Hayashi, T., and Haruta, M., J. Jpn. Inst. Metals 547 (1996).
39. Thiele, G. F., and Roland, E., J. Mol. Catal. A 117, 351 (1997).
40. Blasco, T., Corma, A., Navarro, M. T., and Pariente, J. P., J. Catal. 156,
65 (1995).
18. Bowman, R. G., Clark, H. W., Maj, J. J., and Hartwell, G. E.,
PCT/US97/11414, PCT Pub. WO 98/00413.
41. Tatsumi, T., Koyano, K. A., and Igarashi, N., Chem. Commun. 325
(1998).
19. Hayashi, T., Wada, M., Haruta, M., and Tsubota, S., Jpn. Pat. Pub.
H10-244156, PCT Pub. WO97/00869, U.S. Patent 5,932,750.
20. Uphade, B. S., Tsubota, S., Hayashi, T., and Haruta, M., Chem. Lett.
1273 (1998).
42. D’Amore, M. B., and Schwarz, S., Chem. Commun. 121 (1999).
43. Bu, J., and Rhee, H.-K., Catal. Lett. 65, 141 (2000).
44. Olivera, P. P., Patrito, E. M., and Sellers, H., Surf. Sci. 313, 25
(1994).