Surface-functionalized TUD-1 mesoporous molecular sieve supported palladium for solvent-free aerobic oxidation of benzyl alcohol

Article abstract of DOI:10.1016/j.jcat.2010.07.006

Palladium catalysts supported on TUD-1 mesoporous molecular sieves functionalized with various organosilanes were successfully prepared using a post-synthesis grafting method combined with a metal adsorption-reduction procedure. The correlation between catalyst structure, surface chemistry, metal loading, and catalytic activity was explored in the solvent-free selective oxidation of benzyl alcohol using molecular oxygen. TUD-1 outperformed other mesoporous silicas due to its unique open 3-D sponge-like mesostructure, which can effectively confine Pd nanoparticles and suppress the mass diffusion resistance. The type and content of grafted functional groups greatly affected the catalytic performance by tuning the surface basicity, metal particle size and size distribution, and metal-support interaction. 3-Aminopropyl triethoxysilane functionalization displayed the largest improvement in catalytic activity, showing a dramatically high quasi-turnover frequency of 18,571 h ^-1 for benzyl alcohol conversion.

Full text of DOI:10.1016/j.jcat.2010.07.006

Journal of Catalysis 275 (2010) 11–24
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Surface-functionalized TUD-1 mesoporous molecular sieve supported palladium
for solvent-free aerobic oxidation of benzyl alcohol
Yuanting Chen, Zhen Guo, Tao Chen, Yanhui Yang *
School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 637459, Singapore
a r t i c l e i n f o
a b s t r a c t
Article history:
Palladium catalysts supported on TUD-1 mesoporous molecular sieves functionalized with various orga-
nosilanes were successfully prepared using a post-synthesis grafting method combined with a metal
adsorption–reduction procedure. The correlation between catalyst structure, surface chemistry, metal
loading, and catalytic activity was explored in the solvent-free selective oxidation of benzyl alcohol using
molecular oxygen. TUD-1 outperformed other mesoporous silicas due to its unique open 3-D sponge-like
mesostructure, which can effectively confine Pd nanoparticles and suppress the mass diffusion resistance.
The type and content of grafted functional groups greatly affected the catalytic performance by tuning
the surface basicity, metal particle size and size distribution, and metal–support interaction. 3-Amino-
propyl triethoxysilane functionalization displayed the largest improvement in catalytic activity, showing
Received 5 April 2010
Revised 5 June 2010
Accepted 3 July 2010
Available online 16 August 2010
Keywords:
Benzyl alcohol oxidation
Palladium
TUD-1
Surface functionalization
ꢀ1
a dramatically high quasi-turnover frequency of 18,571 h for benzyl alcohol conversion.
Ó 2010 Elsevier Inc. All rights reserved.
1
. Introduction
the reaction solution is vital to enhance the reaction by facilitating
the reactant adsorption and product desorption, leading to a good-
to-excellent selectivity [6]. Alternatively, base can be added to a
heterogeneous catalyst as the promoter, enhancing the hydrogen
abstraction from –OH groups which is proposed as the rate-limit-
ing step [7]. Klitgaard et al. also suggested that base additives sup-
pressed the catalyst deactivation and restored the catalyst activity
in benzyl alcohol oxidation by diminishing the free benzoic acid le-
vel in the reaction solution [8]. Zhu et al. identified that the role of
base was to provide OH anions to form Au–OH sites for the H-
abstraction step, which was difficult by using only gold as the ac-
tive sites [9]. Nevertheless, adding alkaline agents (either tuning
the pH in reaction solution or adding promoters to the catalyst)
brings several drawbacks: the metal catalyst may dissolve at high
pH and be leached out [6]; the base promoter is also susceptible to
leaching during the reaction, and side reactions could be provoked,
e.g., keto-enol equilibration, Cannizzaro reaction, and oxidative
decarbonylation [10].
Nowadays, chemical functionalization of catalyst support using
a base modifier via the post-synthesis grafting method has been
employed as a promising alternative to the addition of alkaline me-
tal hydroxide and cations. Many silane modifiers can be selected to
functionalize the support surface to obtain the ‘‘customized” sur-
face properties. In addition, these immobilized organic functional
groups, such as thiol and amino groups, act as anchoring sites to
facilitate the stabilization of metal nanoparticles, which are mobile
and prone to sinter on the support surface [11]. Development of
surface-functionalized catalysts for selective alcohol oxidation
has been reported [12]. The localized basicity on the support
Making aldehydes or ketones from the selective oxidation of
corresponding alcohols, in particular, benzaldehyde from benzyl
alcohol is regarded as one of the most versatile and powerful reac-
tions in organic synthesis on both the laboratory and industrial
scales. The catalytic roles of heterogenized noble metal (Pd, Au,
Pt, Ru) catalysts in the aerobic oxidation of benzyl alcohol have
been extensively investigated due to their superior catalytic per-
formance relative to non-noble metals [1]. Among these noble
metals, palladium catalysts have attracted considerable attention
and been intensively studied. Numerous authors have reported
the investigations into Pd-heterogeneous catalysts since the first
successful Pd catalyzed aerobic oxidation of secondary alcohols
ꢀ
ꢀ
[
2]. Mori et al. showed that supported Pd catalysts are highly effec-
tive for the oxidation of 1-phenylethanol under mild solvent-free
ꢀ1
conditions, showing a turnover frequency (TOF) of 9800 h for
Pd supported on hydroxyapatite (Pd/HAP) [3]. Li et al. reported that
zeolite-supported Pd nanoparticles with a mean diameter of
ꢀ1
2
.8 nm exhibited an extraordinary TOF of 18,800 h for 1-phenyl-
ethanol oxidation [4].
A strong alkaline medium is beneficial for certain reactions, and
it has been demonstrated that gold alone cannot catalyze alcohol
dehydrogenation in the absence of a strong base [5]. Basicity also
plays a crucial role in the selective oxidation of alcohols, especially
for primary alcohols. Precisely, controlling the alkaline pH level in
*
Corresponding author. Fax: +65 6794 7553.
E-mail address: yhyang@ntu.edu.sg (Y. Yang).
0
021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.07.006

1
2
Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
surface is able to accelerate the alcohol dehydrogenation, suppress
side reactions, and enhance the desorption of products without sig-
nificantly elevating the solution pH [13]. Hu et al. successfully con-
fined highly dispersed Au nanoparticles on the surface of
functionalized silica materials, showing a 100% conversion in the
benzyl alcohol oxidation [14].
Our laboratory has previously reported both highly dispersed
Au monometallic and Au–Pd bimetallic nanoparticles confined in
surface-functionalized SBA-16 as good catalytic materials for the
selective oxidation of benzyl alcohol [15,16]. Nevertheless, limited
enhancement of catalytic activity was found and attributed to the
mass transfer restriction. A recent addition to the mesoporous
material family is the so-called TUD-1 mesoporous silica [17]. In
the synthesis of TUD-1, surfactant and/or block co-polymer are re-
placed by inexpensive triethanolamine molecules as the structure-
directing agent. Interestingly, the unique open 3-D sponge-like
wide mesoporous structure of TUD-1 can effectively decrease the
pore diffusion resistance as we previously demonstrated [18].
Herein, we synthesize and characterize a series of Pd catalysts sup-
ported on TUD-1 mesoporous silicas functionalized with several
organic functional groups, especially amino groups. The aerobic
oxidation of benzyl alcohol in the absence of any inorganic base
is employed as a model reaction to evaluate these as-synthesized
Pd catalysts. The catalytic activity will be benchmarked against
Pd catalysts supported on other mesoporous silicas, e.g., MCM-
Scheme 1. Preparation procedures and catalytic evaluation of surface-functional-
ized TUD-1-supported Pd catalysts.
toluene, and dried at 80 °C to remove the remaining solvent. The
corresponding samples were denoted as APS-TUD, ATMS-TUD,
HMDS-TUD, and MPTMS-TUD.
Palladium catalysts supported on surface-functionalized TUD-1
were prepared by an adsorption–reduction method reported by
4
1, SBA-15, and SBA-16. The correlation between catalyst struc-
ture, surface chemistry, metal loading, and catalytic activity will
be explored and discussed in detail.
2
. Experimental
2
Chi et al. [19]. A volume of 375.9 lL of 0.05 M PdCl aqueous solu-
tion was added to 0.2 g of surface-functionalized TUD-1, e.g., APS-
TUD, suspended in 20 mL deionized water, followed by vigorous
stirring at 80 °C for 5 h. The catalyst, donated as Pd/APS-TUD,
was obtained by filtering and washing with deionized water, and
2
.1. Chemicals
Tetraethyl orthosilicate (TEOS, 98%, Aldrich), triethanolamine
(
3
TEA, >98.5%, Fluka), tetraethyl ammonium hydroxide (TEAOH,
5%, Aldrich), (3-aminopropyl)triethoxysilane (APS, P98%, Sig-
ma–Aldrich), [3-(2-aminoethylamino)propyl]trimethoxysilane
ATMS, 97%, Aldrich), hexamethyldisilazane (HMDS, Sigma), (3-
mercaptopropyl)trimethoxysilane (MPTMS, Aldrich), palladium
chloride (PdCl , Sigma–Aldrich), benzyl alcohol (98%, Sigma) were
used as received without any further pretreatment.
drying at 80 °C overnight. The catalyst was pretreated in a H flow
of 20 mL min at 400 °C for 2 h to reduce Pd cations to Pd metal
nanoparticles.
2
ꢀ
1
(
2.3. Characterizations
2
Nitrogen physisorption isotherms were measured at ꢀ196 °C on
a static volumetric instrument Autosorb-6b (Quanta Chrome).
Prior to each measurement, the sample was degassed at 250 °C
for 12 h under high vacuum. The specific surface area was esti-
mated by the Brunauer–Emmett–Teller (BET) method [21], and
the pore size distribution was calculated by the Barrett–Joyner–
Halenda (BJH) method [22] using the desorption isotherm branch.
Powder X-ray diffraction (XRD) patterns were recorded on a Bruker
AXS D8 diffractometer (under ambient conditions) using filtered
2
.2. Synthesis
TUD-1 was synthesized following the hydrothermal method re-
ported by Quek et al. [18], using TEA and TEOS as template and sil-
ica precursor, respectively. The preparation procedures were as
follows: 7.2 g of TEA and 1.8 g of deionized water were added drop
wise to 10.0 g of TEOS under vigorous stirring. After 30 min, 10.1 g
of TEAOH was added. The mixture was aged at room temperature
for 24 h, dried at 100 °C for 24 h, and hydrothermally treated in a
Teflon-lined stainless steel autoclave at 180 °C for 8 h. The final
product was calcined at 600 °C for 10 h in air to remove the tem-
plate. MCM-41 and SBA-15 were synthesized following the meth-
ods reported by Chi et al. [19]; and SBA-16 was prepared following
the procedure reported by Kleitz et al. [20].
TUD-1, as well as other mesoporous silica materials surface
functionalized with organosilanes (APS, ATMS, HMDS, and MPT-
MS), was successfully prepared via a post-synthesis grafting meth-
od, as illustrated in Scheme 1. In a typical preparation as
exemplified by APS-modified TUD-1, 1.0 g of TUD-1 was dispersed
Cu Ka radiation (k = 0.15406 nm) operated at 40 kV and 40 mA.
Diffraction data were collected from 30 to 80° with a resolution
of 0.01° (2h). Prior to a test, sample was dried at 100 °C overnight.
Transmission electron microscopy (TEM) was performed on a JEOL
JEM-2100F, operated at 200 kV. The sample was suspended in eth-
anol and dried on holey carbon-coated Cu grids. The metal content
was measured by inductively coupled plasma (ICP), 40% hydroflu-
oric acid was used to dissolve the sample. The pH of pristine and
functionalized TUD-1 supports with different organic groups was
measured following the method reported by Tian et al. [23]. A con-
centration of 0.2 mg of dry sample was added to 10 mL of hot fresh
distilled water, boiled for 5 min, cooled to room temperature, and
stirred overnight to reach equilibrium. The suspension was filtered
and the pH of the filtrate was measured by a pH meter.
2
in 30 mL of toluene and refluxed at 110 °C for 12 h under a N flow
ꢀ1
of 50 mL min . Then, 1.2 mmol of APS (or an appropriate amount
of other silane modifiers) was added to the suspension and stirred
for another 5 h. The resulting powders were filtered, washed with
UV–vis–NIR diffuse reflectance spectra were collected on a
Varian-Cary 5000 UV–vis–NIR spectrophotometer equipped with

Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
13
a diffuse reflectance accessory. The spectra were recorded in the
range of 1000–2500 nm at room temperature with BaSO as a ref-
analyzed using an Agilent gas chromatograph 6890 equipped with
a HP-5 capillary column. Dodecane was the internal standard to
calculate benzyl alcohol conversion and benzaldehyde selectivity.
The conversion of benzyl alcohol and the selectivity toward benz-
aldehyde and quasi-turnover frequency (qTOF) are defined as
follows:
4
erence. All samples were dried at 100 °C overnight before conduct-
ing the test. Fourier Transform Infrared (FTIR) spectra were
recorded on a PerkinElmer Spectrum One FT-IR spectrometer at
ꢀ1
room temperature with KBr pellets (4000–450 cm , resolution
ꢀ
1
of 1 cm ). The in situ infrared reflection absorption spectroscopy
IRAS) was collected on the same equipment using CO as probe
molecules. The sample was pressed into pellet with KBr and placed
into an in situ IR cell with CaF windows. After alignment, the sam-
moles of reactant converted
conversion ð%Þ ¼ 100% ꢁ
(
moles of reactant in feed
moles of product formed
moles of reactant converted
2
selectivity ð%Þ ¼ 100% ꢁ
ple compartment was subjected to a thermal pretreatment in He
flow at 250 °C for 2 h to remove the moisture. The sample cell
was cooled to room temperature and switched to pure CO flow
qTOF ðhꢀ1Þ ¼
moles of reactant converted
moles of total active sites ꢁ reaction time
(
99.5%) for 0.5 h. The absorption spectra were recorded at room
ꢀ1
ꢀ1
temperature (4000–900 cm , resolution of 1 cm , and scan for
min). X-ray photoelectron spectroscopy (XPS) was measured on
a VG Escalab 250 spectrometer equipped with an Al anode (Al
= 1846.6 eV). The background pressure in the analysis chamber
1
3. Results
K
a
3.1. Effect of functional groups
ꢀ7
was lower than 1 ꢁ 10 Pa. Measurements were performed using
0 eV pass energy, 0.1 eV step, and 0.15 dwelling time. Energy cor-
rection was carried out using the C1s peak of adventitious C at
84.6 eV.
2
The bulk average information of TUD-1 mesoporous silica and
the effect of surface functionalization with various organosilanes
were assessed by N physisorption. The amount of different orga-
2
2
nosilanes was kept as 1.2 mmol/g TUD-1. The nitrogen adsorp-
tion–desorption isotherms and the corresponding pore size
distributions are illustrated in Fig. 1a. The plots are typical type
IV isotherms with a pronounced H3 hysteresis loop characteristic
for mesostructured feature with a large pore diameter. Siliceous
2
.4. Catalytic reaction
The solvent-free aerobic oxidation of benzyl alcohol using
molecular O
atmospheric conditions: a three-necked glass flask (capacity:
5 mL) precharged with certain amount of reactant and catalyst
2
was carried out in a bath-type reactor operated under
TUD-1 shows a sharp step increase at P/P
illary condensation of N inside the mesopores. The step increase
shifts to lower P/P , particularly for ATMS-TUD, suggesting a smal-
0
= 0.6–0.8 due to the cap-
2
2
0
as well as a stirring bar was heated in a silicon oil bath, where a
thermocouple was applied to control the reaction temperature. A
reflux condenser was employed to condense the hot vapor of prod-
ucts. The amount of reactant and catalyst was constant at alcohol/
Pd = 250 mol/g. In each reaction run, the mixture was heated to the
desired reaction temperature under vigorous stirring (stirring rate:
ler pore size after surface functionalization with organosilanes,
which can be directly observed from the pore size distributions
in the inset of Fig. 1a. A summary of surface area, pore volume,
and pore diameter is listed in Table 1 (entry 1–4, 7). Pristine
2
ꢀ1
TUD-1 exhibits a BET surface area of 529 m g , pore volume of
3
ꢀ1
0.99 cm g , and a narrow pore size distribution centered at
7.5 nm, which are in good agreement with the previous report
[18]. Surface functionalization results in an apparent decrease
in these textural parameters, which can be attributed to the
immobilization of organic functional groups on the inner wall of
1
200 rpm). Oxygen flow was bubbled at certain flow rate con-
trolled by a mass flow controller into the mixture to initiate the
reaction. After the allowed reaction time duration, the catalyst
powder was filtered off, and the liquid organic products were
TUD-1
TUD-1
APS-TUD
ATMS-TUD
HMDS-TUD
MPTMS-TUD
0.3APS-TUD
0.6APS-TUD
1.2APS-TUD
2.4APS-TUD
3.6APS-TUD
0
10
20
0
10
20
pore diameter,nm
pore diameter,nm
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
P/P0
P/P0
Fig. 1. (a) N
2 2
physisorption isotherms and inset pore size distribution of pristine and surface-functionalized TUD-1; (b) N physisorption isotherms and inset pore size
distribution of TUD-1 surface functionalized with different APS contents.

1
4
Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
Table 1
Textural properties of pristine and surface-functionalized TUD-1 supports.
for pristine TUD-1 assignable to the stretching vibrations of termi-
nal silanol (Si–OH) groups disappears upon the surface modifica-
tion [25], implying the successful replacement of terminal silanol
groups with organosilane modifiers. Luan et al. suggested that
these abundant silanol groups on the inner surface of mesopores
serve as the sites for grafting functional groups during the surface
modification [27]. In the hydroxyl region, the weak band at
Entry
Sample
pH
Surface area
Pore volume
(cm g )
Pore diameter
(nm)
2
ꢀ1
3
ꢀ1
(
m
g
)
1
2
3
4
5
6
7
8
9
TUD-1
6.3
9.7
7.4
4.8
7.8
8.2
9.1
9.1
9.2
529
396
403
394
515
502
420
416
362
0.99
0.62
0.74
0.72
0.99
0.88
0.79
0.76
0.64
7.5
4.7
6.3
6.0
6.8
6.3
6.8
5.9
5.5
ATMS-TUD
HMDS-TUD
MPTMS-TUD
0.3APS-TUD
0.6APS-TUD
1.2APS-TUD
2.4APS-TUD
3.6APS-TUD
ꢀ1
ꢀ1
1
633 cm and the broad band at 3450 cm can be attributed to
a combination of the stretching vibration of silanol groups or sila-
nol ‘‘nests” with cross hydrogen-bonding interactions and the H–
O–H stretching mode of physisorbed water [27]. The absence of
these two bands for HMDS-TUD suggests the successful silylation
of isolated silanol groups with HMDS, leading to the formation of
a hydrophobic surface due to Si(CH
broad inconspicuous band at 2960 cm , which is characteristic
of the asymmetric vibration of the CH groups of silylating agents,
3 3
) groups. The presence of a
ꢀ
1
mesopores. It is worth mentioning that APS-TUD exhibits a de-
crease of 0.7 nm in pore diameter. Considering the organic group
dimension (ca. 0.6 nm) and assuming these groups adopt a folded
conformation [24], this diameter decrease can be suggested as a
monolayer of aminopropyl groups covering the inner wall of mes-
opores. ATMS possesses diamine groups with the longest chain,
resulting in a more extended conformation structure; ATMS-TUD
shows the most significant variation in pore volume and pore
diameter, which is consistent with the lowest onset condensation
2
further confirms the successful silylation [28].
XRD and TEM characterization techniques were introduced to
examine the Pd catalysts supported on these surface-functional-
ized TUD-1 silicas. The XRD patterns of 1 wt.% Pd-containing cata-
lysts are shown in the Supporting information Fig. S1. No distinct
diffraction peak can be observed except for 1Pd/TUD and 1Pd/
HMDS-TUD, implying the highly dispersed nature of these metallic
nanoparticles. For 1Pd/TUD and 1Pd/HMDS-TUD, a diffraction peak
indexed to the (1 1 1) facet of face-centered cubic (FCC) lattice
structure of Pd metal is discernable, suggesting the formation of
metallic Pd nanocrystals with a relatively larger particle size for
these two samples. These XRD analyses can be further confirmed
by the direct TEM microscopic observation. The TEM images and
the corresponding particle size distributions derived from counting
ca. 300 particles are illustrated in Fig. 3. TUD-1 shows a sponge-like
porous structure, suggesting the successful synthesis of TUD-1
mesostructure and good stability after metal adsorption and reduc-
tion. In agreement with nitrogen physisorption results, the lack of
long-range ordered mesostructure in TUD-1 after metal adsorption
and reduction is also reflected in the TEM image. 1Pd/TUD exhibits
a broad palladium particle size distribution centered at 7.0 nm,
which is consistent with the pore diameter of pristine TUD-1
(7.5 nm), implying the encapsulation of Pd nanoparticles in the
mesopores of TUD-1. Upon surface functionalization, 1Pd/APS-
TUD, 1Pd/ATMS-TUD, and 1Pd/MPTMS-TUD show narrow particle
size distributions with mean particle diameters of 1.9, 3.2, and
1.6 nm, respectively. Nonetheless, 1Pd/HMDS-TUD exhibits a con-
siderably wide particle size distribution with a mean particle size
of 17.0 nm.
2
step in the N physisorption isotherms. As mentioned earlier, the
average pH of support surface, which reflects the surface chemistry
to certain extent, plays a crucial role in the selective oxidation of
benzyl alcohol. The results of pH measurement of pristine and sur-
face-modified TUD-1 supports are also summarized in Table 1. It is
noteworthy that the pH varies significantly with surface function-
alizations. The surface of pristine TUD-1 is mildly acidic, showing a
pH of 6.3. Surface functionalization with both APS and ATMS in-
creases the pH to 9.1 and 9.7, respectively, showing that silylation
with amino groups provides the effective raise in the surface basi-
city of silica support. HMDS-functionalization exhibits a neutral
surface as expected. Silylation with MPTMS notably reduces the
pH to 4.8 due to the presence of thiol group. All these changes in
pH validate the successful surface functionalizations and the effec-
tively tuned surface basicity of TUD-1 support.
Fig. 2 displays the FTIR spectra of pristine and surface-function-
alized TUD-1. All TUD-1 samples, regardless of surface functional-
ꢀ
1
ization, show typical bands at 1090 and 798 cm in the skeletal
region of framework vibrations, which are assigned to the asym-
metric and symmetric stretching vibrations of Si–O–Si bridges,
ꢀ1
respectively [25]. The absorbance at 1090 cm has a slight blue
shift after surface functionalization, which can be attributed to
ꢀ
1
the shrinkage of pore structure [26]. The absorbance at 965 cm
After surface functionalization via the silylation of TUD-1 silica
matrix with various organosilanes, these immobilized functional
groups bonded onto the inner pore wall surface of TUD-1 act as an-
2
chors for the metal adsorption. Hydrophilic amino groups (–NH )
7
98
9
65
and thiol groups (–SH) show great affinity to the aqueous palla-
dium chloride and therefore facilitate the deposition of metal pre-
cursor and the formation of small and highly dispersed Pd
nanoparticles embedded on TUD-1. Furthermore, 1Pd/HMDS-TUD
with hydrophobic trimethyl groups shows a barrier for the adsorp-
tion of Pd precursor and leads to the formation of large Pd particles
with board particle diameter distribution. 1Pd/APS-TUD displays a
narrower Pd particle size distribution and a smaller mean particle
diameter than 1Pd/ATMS-TUD (see Fig. 3b and c). Lee et al. also re-
ported a similar result, and they attributed it to the strong interac-
tion between Pd nanoparticles and the grafted diamino groups
1
633
TUD-1
APS-TUD
ATMS-TUD
HMDS-TUD
MPTMS-TUD
2
960
1
090
3
450
[
1
29]. The substantially small pore diameter and pore volume of
Pd/ATMS-TUD may also play a role in forming the Pd nanoparti-
4
000 3500 3000 2500 2000 1500 1000
500
cles with an irregular size distribution, which has been indicated
in our previous report [15].
Wavenumber, cm-1
We introduced the solvent-free selective oxidation of benzyl
alcohol using molecular oxygen as a model reaction to examine
Fig. 2. FTIR spectra of pristine and surface-functionalized TUD-1 with different
modifiers.

Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
15
40
3
0
1Pd/TUD
mean size
=7.0±1.4 nm
20
10
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20
(a)
(b)
(c)
particle size, nm
1
9
8
7
6
5
4
3
00
0
0
0
0
0
0
0
1Pd/APS-TUD
mean size=1.9±0.2 nm
2
1
0
0
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
particle size, nm
5
0
4
0
1Pd/ATMS-TUD
mean size=3.2±0.9 nm
3
0
2
0
1
0
0
0
1
2
3
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
particle size, nm
40
30
1Pd/HMDS-TUD
mean size=17.0±10.0 nm
20
10
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
(
d)
e)
particle size, nm
8
0
70
1
Pd/MPTMS-TUD
60
mean size=1.6 0.3 nm
±
50
40
30
20
10
0
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
(
particle size, nm
Fig. 3. TEM micrographs and particles size distributions of 1 wt.% Pd supported on TUD-1 surface functionalized with different modifiers: (a) 1Pd/TUD; (b) 1Pd/1.2APS-TUD;
c) 1Pd/ATMS-TUD; (d) 1Pd/HMDS-TUD; (e) 1Pd/MPTMS-TUD.
(

1
6
Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
Table 2
Catalytic properties of Pd-containing catalysts supported on pristine and surface-functionalized TUD-1.
a
Catalyst
Pd contentb (wt.%)
Conversion (%)
Selectivity (%)
Benzaldehyde
qTOFc (hꢀ1
)
Toluene
Benzoic acid
1
1
1
1
1
Pd/TUD
0.68
0.99
0.97
0.41
0.77
13.2
22.3
17.0
11.5
26.3
91.7
95.2
93.5
91.5
81.5
1.5
0
0
0
17.7
6.8
4.8
6.5
8.5
0.8
12,910
18,571
13,179
16,802
29,456
Pd/APS-TUD
Pd/ATMS-TUD
Pd/HMDS-TUD
Pd/MPTMS-TUD
a
b
c
ꢀ1
Reaction conditions: benzyl alcohol/Pd = 250 mol/g; O
Pd content was tested by ICP.
2
, 20 mL min ; temperature, 160 °C; reaction time, 1 h; stirring rate, 1200 rpm.
qTOF is calculated by using the Pd content obtained from ICP and subtracting the non-catalytic effect (conversion ꢂ5%)
these as-synthesized Pd catalysts, aiming to elucidate the effect of
surface functionalization on catalytic performance. The catalytic
results are listed in Table 2, where qTOF was determined by using
the Pd content obtained from ICP tests and subtracting the non-
catalytic effect (benzyl alcohol conversion of ꢂ5% at 160 °C). Benz-
aldehyde is produced as the main product along with a trace
amount of toluene and benzoic acid as by-products. The catalytic
activity is remarkably enhanced upon the surface functionalization
after we rule out the effect of actual metal loadings. Among all the
catalysts, 1Pd/MPTMS-TUD shows the best catalytic activity with a
benzaldehyde, APS was selected as the best modifier in the follow-
ing study.
3.2. Effect of APS amount
The nitrogen physisorption patterns of TUD-1 surface function-
alized with different APS contents varying from 0.3 to 3.6 mmol/g
TUD-1, as shown in Fig. 1b, exhibit irreversible type IV isotherms,
which are identical to that of pristine TUD-1. Increasing the APS
loading amount notably shifts the hysteresis toward a lower rela-
ꢀ1
conversion of 26.3% and an exceptionally high qTOF of 29,456 h
,
0
tive pressure (P/P ) and slightly decreases the overall nitrogen
which is twice that of 1Pd/TUD. Nevertheless, this high activity is
accompanied by the formation of a large amount of toluene and
benzoic acid, resulting in consequently low benzaldehyde selectiv-
ity (81.5%). No toluene was detected for other surface-modified
catalysts, and benzoic acid is the only by-product.
adsorption volume, implying the successful grafting of APS on
the inner wall of TUD-1 mesopores. The narrow pore size distribu-
tions are shown in the inset of Fig. 1b. The calculated textural
parameters of TUD-1 modified with various APS amounts are listed
in Table 1 (entry 1, 5–9). The APS-modified TUD-1 samples show
comparable structural parameters to the parent TUD-1, implying
that the integrity of pristine mesoporous structure is retained after
the surface-grafting procedure. A trend of decreasing surface area,
pore volume, and pore diameter with increasing APS content can
be observed, which is attributed to the self-assembly and incorpo-
ration of functional groups into the framework of mesoporous
materials. The results of pH measurement of TUD-1 supports sur-
face functionalized with different APS amounts are also listed in
Table 1. Increasing the APS amount notably increases the pH level
from 6.3 for pristine TUD-1 to 9.1 for 1.2APS-TUD, indicating that
the local surface basicity can be remarkably elevated after silyla-
tion with an appropriate amount of APS. However, further increas-
ing the APS content does not substantially increase the pH value,
which is probably due to that a maximum grafting amount of
APS exists as suggested by Ramila et al. [31].
According to Xu et al., well-isolated amino group and hydroxyl
group absorption bands cannot be resolved using FTIR because of
the bending vibration overlap between N–H and hydroxyl or sila-
nol groups in IR range [32]. Near-infrared (NIR) is superior to IR
for characterizing amino groups because amino groups exhibit
well-resolved absorption bands in the NIR region (1000–
2500 nm). As illustrated in Fig. 4a. All the APS-modified TUD-1
samples display two major absorption bands: 1897 nm assignable
to the combination of stretching and deformation vibrations for
adsorbed water and 1403 nm ascribable to the first overtone of
the stretching frequencies of silanol groups and adsorbed water
[33]. Upon grafting APS onto TUD-1 surface, the reflection bands
at 1528 and 2020 nm appear and become more and more intense
as increasing the APS loading amount. These two bands are in-
Recalling the TEM results in Fig. 3, uniform Pd nanoparticles
with a narrow size distribution centered at 1.6 nm on 1Pd/MPT-
MS-TUD explain the remarkably high activity of this particular cat-
alyst, implying that the Pd nanoparticle size plays a vital role in
controlling the catalytic activity, which is coincident with the re-
sults reported by Li et al. [4]. Nevertheless, the catalytic activity
of noble metal catalysts in the selective oxidation of alcohols heav-
ily relies on the metal–support interaction in addition to the mean
metal particle size. Over 1Pd/MPTMS-TUD, a large amount of tolu-
ene as by-product is induced by the acidic surface due to the
hydrolysis of –SH groups, as suggested by Enache et al. [30]. Fur-
thermore, the generation of benzoic acid is suppressed, and thus
the improvement in benzyl alcohol conversion over 1Pd/MPTMS-
TUD is mainly attributed to the formation of toluene by dispropor-
tionation of benzyl alcohols. The extremely low Pd content in 1Pd/
HMDS-TUD is due to the adsorption barrier of aqueous palladium
precursor onto the hydrophobic surface in the presence of tri-
methyl groups. This particular catalyst exhibits the lowest benzyl
alcohol conversion (11.5%), which probably resulted from the lack
of active sites due to the extremely low Pd content as well as the
large and irregular Pd nanoparticles as indicated in XRD and TEM.
For amino group-functionalized catalysts, i.e., both 1Pd/APS-
TUD and 1Pd/ATMS-TUD, the surface is basic due to the hydrolysis
of amino groups, which slightly promotes the formation of benzoic
acid. The elevated surface basicity is suggested to enhance the
dehydrogenation of benzyl alcohol, which is the rate-controlling
step during the reaction, suppress the formation of by-product tol-
uene, as well as facilitate the reactant adsorption and product
desorption from the Pd active sites. Although APS and ATMS both
contain amino groups, 1Pd/APS-TUD exhibits larger enhancement
in the catalytic activity as well as the selectivity. This is mainly
due to the formation of larger Pd nanoparticles with fewer active
sites for the reaction in 1Pd/ATMS-TUD as suggested by TEM be-
cause diamine groups in ATMS show stronger interaction with Pd
nanoparticles than that of monoamine groups in APS. After consid-
ering both excellent catalytic activity and high selectivity toward
dexed to the combination band of asymmetry stretching (
and symmetry stretching ( sym) modes and the combination band
of stretching vibration ( ) and bending (d) modes of the amino
groups, respectively [34]. Moreover, the reflection bands between
1600 and 860 nm are assigned to the stretching vibration of CH
asym
m )
m
m
2
moiety in the propyl chain of the APS silane modifier [35]. The
UV–vis–NIR spectra provide further evidence that after surface

Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
17
derived from this variation of absorption bands. The decreased
Pd nanoparticle size can lead to the broadening and blue shift of
CO stretching band in IRA spectra, which is due to the increased
number of defect sites (such as edges and kinks) and disorder on
a small Pd nanoparticle, as suggested by Ozensoy and Goodman
a
NH (ν+δ)
2
1
897
1
403
2
020
NH (νasym+νsym
)
2
1
528
TUD-1
[
36]. Hollins reported that the increase in CO stretching frequency
is also attributed to a weaker CO–metal interaction, resulting in a
reduced occupation of CO 2 orbital when CO is bound at a high-
0
.3APS-TUD
0
.6APS-TUD
p
1
.2APS-TUD
coordinated metal atom [37]. With increased grafting content of
aminopropyl groups, Pd atoms have higher coordination with the
surrounding amino groups; the interaction between Pd nanoparti-
cles and the APS-modified TUD-1 becomes stronger. Hence, the
interaction between Pd and adsorbed CO gets weaker and less CO
2
.4APS-TUD
.6APS-TUD
3
2
p
orbit is occupied, leading to the blue shift of the CO stretching
band.
To further investigate the structural and electronic properties of
1400
1600
Wavelength, nm
1800
2000
these Pd nanoclusters, Pd 3d XPS measurements were conducted,
and the results of four selected 1Pd/APS-TUD samples are illus-
trated in Fig. 5. All the samples show a typical doublet of 3d core
level bands centered around 340 and 335 eV, which are assigned
2
114
b
2057
2
172
1873
1
Pd/TUD
Pd/0.3APS-TUD
Pd/0.6APS-TUD
2058
1
0
0
to Pd 3d3/2 and Pd 3d5/2, respectively [38]. Recently, Radkevich
2059
1
et al. have demonstrated the amino groups possessing electron-
0
donor properties can increase the proportion of Pd and the resis-
2069
1
1
Pd/1.2APS-TUD
Pd/2.4APS-TUD
0
tance against re-oxidation of Pd due to the electronic interactions
2057
between nitrogen atoms and Pd metallic nanoparticles, which are
active centers for hydrogenation–dehydrogenation reactions [39].
We cannot draw the same conclusion because no oxide peak is de-
tected throughout the whole Pd 3d region, implying the formation
of Pd metallic nanoparticles by completely reducing the palladium
precursors. With increasing grafted amount of APS, both bands un-
dergo a negative shift to lower binding energies. The maximum
shift (1 eV) when compared to the binding energy of Pd supported
on pristine TUD-1 can be observed on 1Pd/1.2APS-TUD. With
2057
1
Pd/3.6APS-TUD
2
400
2200
2000
1800
-
1
Wavenumber, cm
Fig. 4. (a) The UV–vis–NIR spectra of pristine and APS-functionalized TUD-1 with
different APS contents; (b) IRA spectra for CO adsorption on 1Pd/APS-TUD with
different APS contents. CO dosage and measurement performed at room
temperature.
0
Pd 3d
0
Pd 3d5
/2
Pd 3d3
/2
modification with APS, the amino functional groups are indeed
grafted onto the silica surface and covalently bonded to the inner
wall of mesopores.
1
Pd/
2
.4APS-TUD
Vibrational spectroscopy techniques, e.g., IRAS, have been re-
garded as exceptionally useful tools for elucidating the heteroge-
neous catalytic reaction mechanisms, indentifying the active
species on a catalyst surface with sub-monolayer accuracy, and
addressing the particle size and morphology effects and particle–
support interactions. In this study, IRAS was employed using CO
as a probe molecule to identify specific interaction between Pd
nanoparticles and amino groups. All samples show the characteris-
tic features for CO adsorption on Pd nanoparticles, see Fig. 4b. The
1
Pd/
1
.2APS-TUD
1Pd/
.3APS-TUD
ꢀ
1
0
bands at 2172 and 2114 cm are assigned to the absorption of
residual gaseous CO. These two characteristic bands do not change
significantly with increasing the APS amount. The C–O stretching
ꢀ1
bands at around 2057 and 1873 cm correspond to the CO mole-
cules adsorbed on top and threefold hollow sites on a (1 1 1) facet
of Pd nanoclusters, respectively, suggesting that TUD-1-supported
Pd nanoparticles in this study exhibit predominantly (1 1 1) facets,
which in agreement with previous reports [36].
1
Pd/TUD
With increasing grafted amount of aminopropyl groups, the
ꢀ1
absorption band at 2057 cm becomes boarder and slightly blue
shifts to high wave number. In particular, this band shifts to
ꢀ
1
350 348 346 344 342 340 338 336 334 332 330
Binding Engergy,eV
2
069 cm
and becomes an imperceptible shoulder for 1Pd/
1
.2APS-TUD. When the APS contents further increase, this band be-
ꢀ1
comes narrower and shifts back to 2057 cm as that of Pd sup-
ported on pristine TUD-1. The information of Pd nanoparticle size
and interaction between Pd and amino groups can be indirectly
Fig. 5. TEM micrographs and particles size distributions of 1Pd/APS-TUD with
different APS contents: (a) 1Pd/0.3APS-TUD; (b) 1Pd/0.6APS-TUD; (c) 1Pd/1.2APS-
TUD; (d) 1Pd/2.4APS-TUD; (e) 1Pd/3.6APS-TUD.

1
8
Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
further increase in the APS content, both bands exhibit a shift back
to high binding energy. Fox et al. suggested that the shift to low
binding energy indicates the high dispersion of Pd species [40]. It
is worthwhile noticing that this trend of binding energy shift is
4
0
(
(
a)
b)
1
Pd/0.3APS-TUD
mean size=5.8±2.3 nm
30
20
10
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Particle size, nm
60
1Pd/0.6APS-TUD
50
mean size=2.40.5 nm0.5 nm
40
30
20
10
0
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Particle size, nm
(
(
c)
d)
100
9
0
1Pd/1.2APS-TUD
mean size=1.9±0.2 nm
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Particle size, nm
100
90
1
Pd/2.4APS-TUD
8
70
0
mean size=2.9±0.4 nm
6
0
5
0
4
0
3
0
2
0
1
0
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Particle size, nm
(
e)
80
70
1Pd/3.6APS-TUD
mean size=3.2±0.5 nm
60
50
40
30
20
10
0 0
1
2
3
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Particle size, nm
Fig. 6. Pd 3d XP spectra of 1Pd/APS-TUD catalysts with various APS contents.

Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
19
identical to that of CO adsorption peak from IRAS and can be
understood as complementary evidence of the APS amount effect
on the interaction between Pd nanoparticles and amino groups.
The XRD patterns of 1Pd/APS-TUD catalysts with different APS
amounts are shown in the Supporting information Fig. S2. There
is no obvious discrepancy in the Pd nanoparticles dispersion
among all 1Pd/APS-TUD catalysts because no noticeable diffraction
peak can be observed in the XRD patterns. The TEM images are
shown in Fig. 6. The mean Pd particle size is reduced from 7.0 to
mized amount of APS functional groups to substitute the silanol
groups and occupy the pore volume, which was also demonstrated
by other groups [27,31]. In this study, the effect of APS content is
pronounced between 1.2APS-TUD substrate and Pd species.
To examine the effect of grafted APS content on the catalytic
activity, benzyl alcohol oxidation over 1Pd/APS-TUD catalysts with
different APS amounts was investigated and illustrated in Fig. 7.
The catalytic activity is improved when APS is grafted. 1Pd/
1.2APS-TUD shows the highest conversion and selectivity toward
benzaldehyde of 22.3% and 95.2%, respectively. Further increasing
the APS loading slightly decreases the catalytic activity and selec-
tivity. The trend of benzyl alcohol conversion implies the activity
of supported Pd catalysts is mainly dependent on the particle size
and size distribution [4] as well as the local surface basicity. The
outstanding catalytic behavior of 1Pd/1.2APS-TUD resulted from
the small and uniform Pd nanoparticles due to the enhanced me-
tal–support interaction and the appropriate surface basicity. 1Pd/
1.2APS-TUD was selected as the optimal catalyst in the following
work.
5
.8 nm upon increasing surface modification with APS. It decreases
to 1.9 nm with a narrow size distribution for 1Pd/1.2APS-TUD.
Nevertheless, larger Pd nanoclusters are formed with a boarder
particle size distribution for 1Pd/3.6APS-TUD due to the excess
amount of APS. This trend is consistent with IRAS and XPS results.
It can be verified that the grafting amount of amino groups plays
an important role in controlling the size and morphology of Pd
nanoparticles, i.e., small and homogeneous Pd nanoparticles
formed upon surface functionalized with APS, whereas further
loading excess APS results in large and irregular nanoparticles. A
specific interaction between Pd nanoclusters and APS immobilized
TUD-1 support is also evidenced.
3.3. Effect of Pd loading
Romanmartinez et al. suggested that the distribution of metal
precursors on the support and the specific metal–support interac-
tion are strongly dependent on the support surface chemistry [41].
This APS effect on the size control of Pd nanoparticles can be ex-
plained by the nature of the support surface, which results in dif-
ferent interactions between metal and support. Due to the
hydrolysis in the metal adsorption procedure, the silica surface
without APS group is mildly acidic and negatively charged, which
hinders the diffusion and adsorption of negatively charged
The XRD patterns of Pd/1.2APS-TUD with different Pd loadings
(0.5–3 wt.%) illustrate that only 3Pd/1.2APS-TUD shows a weak
and broad diffraction peak assigned to the (1 1 1) facet of face-cen-
tered cubic (FCC) lattice structure for Pd nanoparticles, suggesting
the formation of large and irregular Pd nanoclusters in this partic-
ular sample. It is further demonstrated by the direct TEM micro-
scopic images as displayed in Fig. 8. Increasing Pd loading results
in a larger mean particle diameter, varied between 1.3 nm for
0.5Pd/1.2APS-TUD and 6.5 nm for 3Pd/1.2APS-TUD. Moreover, the
nanoparticles on 3Pd/1.2APS-TUD show a wide size distribution.
This discrepancy is mainly caused by the agglomeration of Pd
nanoparticles due to high Pd content.
2ꢀ
PdCl
groups with amino groups, the surface is positively charged due
to the hydrolysis of –NH groups, exhibiting great affinity to the
2
(OH
2
)
species. After substituting the terminal silanol
2
palladium precursors, thus the palladium precursors can be easily
adsorbed into the mesopores. In addition, these immobilized ami-
no groups are suggested to act as anchors to stabilize the Pd nano-
particles by binding them through covalent interactions, resulting
in highly dispersed and uniformly distributed Pd nanoparticles
The catalytic results of benzyl alcohol oxidation over Pd/1.2APS-
TUD catalysts with different Pd loadings are shown in Fig. 9. All
catalysts exhibit a high and constant selectivity around 96% toward
benzaldehyde. The benzyl alcohol conversion increases as the Pd
loading rises from 0.5 to 1 wt.%, which is due to the increased num-
ber of active sites available participating in the reaction. The con-
version slightly decreases upon further adding Pd, which can be
attributed to the agglomeration of nanoparticles which decreases
the available number of active sites for the reaction at a high Pd
content [4]. Among all the samples, 1Pd/1.2APS-TUD shows the
[
11]. Nevertheless, further adding APS may result in the oligomer-
ization and polymerization of organosilanes into the form of a
cross-linked monolayer of alkanolamine attached on the surface;
partially, blocking the mesopores occurs as suggested in nitrogen
physisorption (see Table 1) [42]. More Pd nanoparticles on the
external surface of the support are susceptible to agglomeration
due to the lack of structural confinement [15]. Thus, there is a opti-
ꢀ
1
best catalytic result with a remarkably high qTOF of 18,571 h
and is regarded as the best catalyst in this study.
The inset table in Fig. 9 lists the activation energies of these cat-
alysts in aerobic oxidation of benzyl alcohol determined from the
Arrhenius plots. These activation energy values are comparable
100
25000
2
to that of TiO -supported noble metal catalysts reported by Enache
benzaldehyde
selectivity
80
60
40
20
0
2
0000
et al. (45.8 kJ/mol) [30] and much higher than the Pd catalyst sup-
ported on SBA-16 (12.3 kJ/mol) [16]. It was reported that low acti-
vation energy in a liquid phase reaction indicates the alcohol
oxidation is likely to be mass diffusion limited and controlled by
the access of reactants to active sites [43]. This suggests that the
unique 3-D sponge-like mesoporous structure of TUD-1 can effec-
tively eliminate the diffusion limitation.
qTOF
15000
10000
5000
0
benzyl alcohol
conversion
3.4. Studies on the 1Pd/1.2APS-TUD catalyst
1
Pd/1.2APS-TUD, which has been verified possessing the best
0
0.3
0.6
1.2
2.4
3.6
catalytic performance in the solvent-free oxidation of benzyl
alcohol, was investigated in detail. The time course of 1Pd/
APS amount,mmol APS/g TUD-1
1
.2APS-TUD for benzyl alcohol oxidation was monitored periodi-
Fig. 7. Catalytic performance of 1Pd/APS-TUD catalysts with different APS contents.
ꢀ
1
cally, as depicted in Fig. 10. Benzyl alcohol conversion monotoni-
cally increases with the reaction time duration along with a
Reaction conditions: benzyl alcohol/Pd = 250 mol/g; O
60 °C; reaction time, 1 h; stirring rate, 1200 rpm.
2
, 20 mL min ; temperature,
1

2
0
Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
8
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
0
(a)
(b)
(c)
(d)
0
.5Pd/1.2APS-TUD
mean size=1.3±0.3 nm
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
particle size,nm
1
00
9
8
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
0
0
1
Pd/1.2APS-TUD
mean size=1.9 0.2 nm
±
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
particle size,nm
8
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
0
2
Pd/1.2APS-TUD
mean size=3.3 0.7 nm
±
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
particle size,nm
4
3
2
1
0
0
0
0
0
3
Pd/1.2APS-TUD
mean size=6.5 1.4 nm
±
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
particle size,nm
Fig. 8. TEM micrographs and particles size distributions of Pd/1.2APS-TUD with various Pd loading: (a) 0.5Pd/1.2APS-TUD; (b) 1Pd/1.2APS-TUD; (c) 2Pd/1.2APS-TUD; (d) 3Pd/
.2APS-TUD.
1
rapidly declined qTOF. The selectivity toward benzaldehyde is
rather low (81.5%) at first 0.5 h of reaction; it rises to 95.2% at
well maintained above 95% when the temperature increases from
80 to 160 °C (entry 1–5). Benzyl alcohol conversion is extremely
low at low reaction temperatures, and it remarkably increases
when the temperature is elevated from 120 to 160 °C. The benzyl
alcohol conversion displays a tenfold enhancement when the reac-
tion temperature increases from 100 to 160 °C. This trend is consis-
tent with the activation energies listed in Fig. 9. The high activation
energy implies the absence of mass transport limitation, as well as
the large energy barrier for the reaction. Therefore, high tempera-
1
0
h and gradually decreases afterward. The poor selectivity at first
.5 h is contributed by the formation of a large amount of toluene
(
15.8%), which is produced from the hydrogenolysis of benzyl alco-
hol facilitated by hydrogen adsorbed on Pd surface formed in the
dehydrogenation step at the beginning of reaction [44]. The decay
of selectivity at long reaction time is ascribed to the generation of
benzoic acid due to the further oxidation of benzaldehyde.
The effects of reaction temperature and oxygen flow rate were
also examined and summarized in Table 3. Neither TUD-1 nor sur-
face-functionalized TUD-1 displayed noticeable catalytic activity at
a low reaction temperature. The selectivity toward benzaldehyde is
2
ture is beneficial to initiate the reaction. The O flow also shows
impact on the catalytic performance. The selectivity to benzalde-
hyde is low accompanied by the formation of a large amount of tol-
2
uene at low O flow rate, and the selectivity gradually improves

Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
21
1
00
20000
of external diffusion effect, which can be eliminated by increasing
the O flow rate.
2
benzaldehyde selectivity
Several other aromatic alcohols were employed to examine the
generality of 1Pd/1.2APS-TUD for alcohol oxidation, and the results
are listed in Table 3 (entry 8–12). The selectivity toward the corre-
sponding aldehyde or ketone is remarkably high for all the alcohols
8
6
4
2
0
0
0
0
0
catalyst
.5Pd/1.2APS-TU
Pd/1.2APS-TUD
Pd/1.2APS-TUD
Pd/1.2APS-TUD
Ea(KJ/mol)
15000
10000
5000
0
0
52.4
52.1
54.1
43.7
1
2
3
ꢀ
1
tested. High efficiency (qTOF of 29,576 h ) was obtained for 1-
phenylethanol oxidation, implying the higher reactivity of second-
ary alcohol compared to primary alcohol. 4-methylbenzyl alcohol
shows a low conversion but non-ignorable. Poor activities are ob-
served for the oxidation of 4-nitrobenzyl alcohol and 4-bromoben-
zyl alcohol, suggesting that this 1Pd/1.2APS-TUD catalyst shows
higher catalytic activity for substituted aromatic alcohols contain-
qTOF
benzyl alcohol conversion
0.5
1.0
1.5
2.0
2.5
3.0
ing electron-donating group (e.g., –CH
electro-withdrawing groups (such as –NO
) than those containing
and –Br) [46]. Addition-
3
Pd loading,wt.%
2
ally, this catalyst also displays a good conversion for the oxidation
of cinnamyl alcohol (entry 12), suggesting the effective catalytic
activity for oxidizing allylic alcohols.
Fig. 9. Catalytic performance of Pd/1.2APS-TUD with various Pd loadings and inset
table of activation energy. Reaction conditions: benzyl alcohol/Pd = 250 mol/g; O
2
,
ꢀ
1
2
0 mL min ; temperature, 160 °C; reaction time, 1 h; stirring rate, 1200 rpm.
The recyclability has been taken as a curial factor in evaluating a
heterogeneous catalyst participating in a multiphase system. The
metallic components with weak stability and poor recyclability
can leach out during the course of reaction, resulting in the forma-
tion of an active homogeneous catalyst and a loss of catalytic
activity on subsequent consecutive runs. The recyclability of
1Pd/1.2APS-TUD catalyst in the solvent-free selective oxidation of
benzyl alcohol with molecular oxygen was examined. The catalyst
1
00
35000
3
0000
5000
8
6
4
2
0
0
0
0
0
benzaldehyde
selectivity
2
20000
qTOF
benzyl alcohol
conversion
15000
10000
5000
0
1
00
80
60
35000
0000
25000
3
benzaldehyde selectivity
benzyl alcohol conversion
qTOF
2
0000
5000
0
2
4
6
8
10
reaction time,h
1
4
0
0
0
Fig. 10. Time course of 1Pd/1.2APS-TUD for benzyl alcohol oxidation. Reaction
conditions: benzyl alcohol/Pd = 250 mol/g; O
stirring rate, 1200 rpm.
ꢀ
1
10000
2
, 20 mL min ; temperature, 160 °C;
2
5000
with the increased flow rate (entry 5–7). Toluene has been verified
as the major by-product under anaerobic conditions due to the
deficiency of surface oxygen, which therefore results in either a
concentrated nucleophilic attack of s surface hydride to the ben-
zylic carbon of adsorbed benzyl alcohol or the recombination of
0
1
2
3
4
5
Number of batches
Fig. 11. Recyclability of 1Pd/1.2APS-TUD catalyst for solvent-free oxidation of
benzyl alcohol with molecular oxygen. Reaction conditions: benzyl alcohol/
[
PhCH
conversion at a low O
2
(ad)] with [H(ad)] [45]. Moreover, the slight decrease in
ꢀ1
Pd = 250 mol/g; O
rate, 1200 rpm.
2
, 20 mL min ; temperature, 160 °C; reaction time, 1 h; stirring
2
flow rate also implies the possible presence
Table 3
a
Effect of reaction conditions for 1Pd/1.2APS-TUD on alcohol oxidation.
flow rate (mL min 1
ꢀ
)
Conversion (%)
Selectivityb (%)
qTOF (h
ꢀ1
)
Entry
Substrate
Temperature (°C)
O
2
1
2
3
4
5
6
7
8
9
Benzyl alcohol
Benzyl alcohol
Benzyl alcohol
Benzyl alcohol
Benzyl alcohol
Benzyl alcohol
Benzyl alcohol
1-Phenylethanol
4-Methylbenzyl alcohol
4-Nitrobenzyl alcohol
4-Bromobenzyl alcohol
Cinnamyl alcohol
80
100
120
140
160
160
160
160
160
160
160
160
20
20
20
20
20
10
5
20
20
20
20
20
0.8
2.2
3.0
10.7
22.3
18.8
14.4
27.5
6.0
96.2
96.6
96.2
94.5
95.2
83.7
79.9
98.2
97.2
100
908
2316
3251
11,493
18,571
14,847
13,329
29,576
6437
1
1
1
0
1
2
1.0
0.8
14.4
1025
874
15,493
100
100
a
Reaction conditions: substrate/Pd = 250 mol/g; reaction time, 1 h; stirring rate, 1200 rpm.
Selectivity refers to the corresponding aldehyde or ketone; for benzyl alcohol, by-products are toluene and benzoic acid; for 1-phenylethanol, by-product is ethylbenzene;
b
for 4-methylbenzyl alcohol, by-product is p-xylene.

2
2
Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
was recovered after each reaction run, washed with acetone, dried
at 60 °C, and reused in a new reaction. Fig. 11 shows the recyclabil-
ity of 1Pd/1.2APS-TUD for 5 consecutive cycles. The catalytic activ-
ity and selectivity only undergo a moderate decrease, which is
attributed to the loss of catalyst during recovery. The Pd content
after 5 consecutive reaction cycles was almost the same as that
of fresh catalyst, and Pd cannot be detected by ICP analysis in the
filtrate after the reaction. Moreover, no noticeable conversion of
benzyl alcohol was observed when the liquid filtrate was employed
instead of the solid catalyst for further reaction, further confirming
that there was no leaching of Pd. Therefore, APS-modified TUD-1-
supported Pd catalysts not only dramatically improve the catalytic
performance but also show superiority in enhancing the resistance
against deactivation.
by our group [15,16], the mass diffusion limitation is severe and its
inferior catalytic activity is due to the restriction by the diffusion of
reactants to active sites during the reaction.
For TUD-1, the mesopores have larger diameter than that of
SBA-16 and are randomly interconnected. The tortuous pore
arrangements can help to provide sufficient defects to encapsulate
and restrict the mobility of nanoparticles that lead to thermal sin-
tering and growth of particles, and prevent the active centers from
reactant poisoning. In addition, according to our previous report,
TUD-1 with open 3-D mesopore structure and large pore size can
effectively decrease the pore diffusion resistance of the reactants.
Therefore, it can provide greater access of the reactants to the ac-
tive sites embedded on the pore wall surface [18]. The excellent
catalytic performance of TUD-1-supported Pd catalysts herein
again verifies the superiority of the 3-D sponge-like mesostructure
of TUD-1. Hence, upon the APS-modification, palladium precursors
can be easily adsorbed, and Pd nanoparticles are uniformly embed-
ded in the mesopores of TUD-1 with a specific size distribution.
The suppressed mass transfer limitation, enhanced metal–support
interaction, and a finely tuned local basicity surrounding the
Pd nanoparticles result in a remarkably improved catalytic
performance.
4
. Discussion
To demonstrate the advantage of TUD-1’s unique mesostructure
in the liquid phase reaction, we investigated the catalytic perfor-
mances of Pd catalysts supported on various surface-functionalized
mesoporous silica materials, including TUD-1, MCM-41 and SBA-
1
5 with 2-D hexagonal structure and SBA-16 with 3-D cubic
It is clear that local surface basicity and Pd nanoparticle size are
two vital factors controlling both the catalytic activity (qTOF) and
selectivity. Nevertheless, the surface basicity also affects the parti-
cle size and size distribution, as well as the dehydrogenation step
in the benzyl alcohol oxidation, resulting in the complex and
ambiguous explanation of the effect of individual factors and their
interaction on the catalytic performance. To facilitate more rigor-
ous interpretation, we employed statistical techniques to develop
a second-order polynomial regression model including both linear
cage-like structure. As depicted in Fig. 12, TUD-1 shows superior
catalytic activity relative to other silica materials, regardless of
the surface functionalization. SBA-16 shows a moderate perfor-
mance, whereas MCM-41 and SBA-15 exhibit a similar but rather
low activity. This can be attributed to the effect of different meso-
porous structures since the formation of metal nanoparticles is
mainly controlled by the architecture of the host material. On
MCM-41 and SBA-15, Pd nanoparticles show a high degree of ther-
mal sintering within the straight mesochannels, leading to the
migration and coalescence of Pd nanoparticles and thereby the
ineffective catalytic performance. The inferior performance of
SBA-15 when compared to MCM-41 is due to the interconnected
micropores in SBA-15 where Pd nanoparticles may be completely
buried and inactive. Although the ‘‘super-cage” porosity of SBA-
and quadratic terms to fit the experimental data of two factors (x
Pd size, x : pH value) and two responses (y : qTOF, y : selectivity)
see Supporting information). The contour plots of qTOF and selec-
1
:
2
1
2
(
tivity provide a rational examination of the influence of each
experimental variable on the responses, as illustrated in Fig. 13.
Both contour plots are elliptically shaped, implying the significant
interaction between pH and Pd size [47]. It is obvious in Fig. 9a that
1
6 is superior for confining metal nanoparticles with high disper-
sion in a ‘‘ship in a bottle” way, which has been previously proven
3
2
0000
5000
1
00
25
20
15
10
5
0
9
9
8
8
7
5
0
5
0
5
20000
15000
q
TUD-1
MCM-41
SBA-15
SBA-16
1
0000
70
HMDS none ATMS APS MPTMS
HMDS none ATMS APS MPTMS
HMDS none ATMS APS MPTMS
modifier
modifier
modifier
Fig. 12. Catalytic performance of 1 wt.% Pd-containing catalysts supported on pristine and surface-functionalized mesoporous silica supports. Reaction conditions: benzyl
ꢀ
1
2
alcohol/Pd = 250 mol/g; O , 20 mL min ; temperature, 160 °C; reaction time, 1 h; stirring rate, 1200 rpm.

Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
23
with a metal adsorption–reduction procedure. The catalytic activ-
ities of these as-prepared catalysts were explored in the solvent-
free selective oxidation of benzyl alcohol using molecular oxygen.
The catalytic results proved that the type and amount of functional
group for functionalizing the TUD-1 support showed great impact
on the catalytic performance. APS immobilized on TUD-1 with an
appropriate amount of APS coverage displayed the best improve-
ment owing to the enhanced metal–support interaction and finely
tuned local surface basicity surrounding the Pd nanoparticles.
Among all the catalysts, 1Pd/1.2APS-TUD exhibited a dramatically
ꢀ1
high qTOF of 18,571 h for benzyl alcohol conversion. TUD-1 is
superior to other mesoporous silicas due to its unique open 3-D
sponge-like wide mesostructure, which can effectively confine
the nanoparticles and suppress the mass diffusion resistance.
Acknowledgements
We are grateful to AcRF tier 2 (ARC 13/07) for funding support.
Authors would also like to acknowledge the great help on discus-
sion and assistance with editing from Professor Gary L. Haller, Yale
University.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2010.07.006.
Reference
[
[
[
1] G.J. Hutchings, Chem. Commun. (2008) 1148.
2] T.F. Blackburn, J. Schwartz, J. Chem. Soc., Chem. Commun. (1977) 157.
3] K. Mori, T. Hara, T. Mizugaki, K. Ebitani, K. Kaneda, J. Am. Chem. Soc. 126 (2004)
10657.
[
[
[
[
4] F. Li, Q.H. Zhang, Y. Wang, Appl. Catal. A: Gen. 334 (2008) 217.
5] A.S.K. Hashmi, G.J. Hutchings, Angew. Chem. Int. Ed. 45 (2006) 7896.
6] T. Mallat, A. Baiker, Chem. Rev. 104 (2004) 3037.
7] P.J.M. Dijkgraaf, M.J.M. Rijk, J. Meuldijk, K. Vanderwiele, J. Catal. 112 (1988)
329.
[
[
8] S.K. Klitgaard, A.T. DeLa Riva, S. Helveg, R.M. Werchmeister, C.H. Christensen,
Catal. Lett. 126 (2008) 213.
9] J.J. Zhu, J.L. Figueiredo, J.L. Faria, Catal. Commun. 9 (2008) 2395.
Fig. 13. Contour plots of (a) qTOF and (b) selectivity as a function of Pd size and pH
of support.
[
[
[
10] C. Bianchi, F. Porta, L. Prati, M. Rossi, Top. Catal. 13 (2000) 231.
11] D.V. Leff, L. Brandt, J.R. Heath, Langmuir 12 (1996) 4723.
12] C. Aprile, A. Abad, G.A. Hermenegildo, A. Corma, J. Mater. Chem. 15 (2005)
4
408.
[13] C. Bronnimann, T. Mallat, A. Baiker, J. Chem. Soc., Chem. Commun. (1995)
377.
14] J.C. Hu, L.F. Chen, K.K. Zhu, A. Suchopar, R. Richards, Catal. Today 122 (2007)
77.
[15] H. Sun, Q.H. Tang, Y. Du, X.B. Liu, Y. Chen, Y.H. Yang, J. Colloid Interface Sci. 333
2009) 317.
16] Y. Chen, H. Lim, Q. Tang, Y. Gao, T. Sun, Q. Yan, Y. Yang, Appl. Catal. A: Gen.
2010), doi:10.1016/j.apcata.2010.03.026.
to obtain high qTOF at small Pd size, the catalyst surface has to be
mildly alkaline. Similarly, high selectivity can be achieved at small
Pd size and high pH level, and vice versa. Both contour plots reflect
similar trend; and therefore, the adjustment of Pd size and surface
basicity will affect the catalytic results remarkably. Based on the
normalized variables and the improved model (see Supporting
1
[
2
(
[
(
information), the prediction formulas of y
ity) are as follows:
1
(qTOF) and y
2
(selectiv-
[17] J.C. Jansen, Z. Shan, L. Marchese, W. Zhou, N. von der Puil, T. Maschmeyer,
Chem. Commun. (2001) 713.
[
[
18] X.Y. Quek, Q.H. Tang, S.Q. Hu, Y.H. Yang, Appl. Catal. A: Gen. 361 (2009) 130.
19] Y.S. Chi, H.P. Lin, C.Y. Mou, Appl. Catal. A: Gen. 284 (2005) 199.
2
1
y ¼ 19753:3 þ 5787:1x
1
þ 8196:8x ꢀ 21677:9x
2
1
[20] F. Kleitz, T.W. Kim, R. Ryoo, Langmuir 22 (2006) 440.
[21] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309.
2
ꢀ
43109:2x ꢀ 60288:2x
1
x
2
2
[
22] E.P. Barrett, L.G. Joyner, P.P. Halenda, J. Am. Chem. Soc. 73 (1951) 373.
2
1
y2 ¼ 85:94 ꢀ 2:17x
2
ꢀ 10:55x ꢀ 18:22x
1 2
x
[23] S.L. Tian, H. Mo, R.F. Zhang, P. Ning, T.H. Zhou, Adsorption 15 (2009) 477.
[
[
24] P. Horcajada, A. Ramila, F. Gerard, M. Vallet-Regi, Solid State Sci. 8 (2006) 1243.
25] M.A. Camblor, A. Corma, J. Perezpariente, J. Chem. Soc., Chem. Commun. (1993)
These quadratic polynomial equations, which were developed
to predict the response as a function of the variables and their
interactions variables, can be used to determine the catalytic re-
sults (qTOF and selectivity) quantitatively and facilitate precise
optimization of the reaction parameters.
557.
[
26] C.A. Koh, R. Nooney, S. Tahir, Catal. Lett. 47 (1997) 199.
[27] Z.H. Luan, J.A. Fournier, J.B. Wooten, D.E. Miser, Micropor. Mesopor. Mater. 83
(2005) 150.
[
[
[
[
[
28] T. Joseph, K.V. Kumar, A.V. Ramaswamy, S.B. Halligudi, Catal. Commun. 8
2007) 629.
29] B. Lee, Z. Ma, Z.T. Zhang, C. Park, S. Dai, Micropor. Mesopor. Mater. 122 (2009)
60.
30] D.I. Enache, J.K. Edwards, P. Landon, B. Solsona-Espriu, A.F. Carley, A.A. Herzing,
M. Watanabe, C.J. Kiely, D.W. Knight, G.J. Hutchings, Science 311 (2006) 362.
31] A. Ramila, B. Munoz, J. Perez-Pariente, M. Vallet-Regi, J. Sol–Gel Sci. Technol. 26
(
1
5
. Conclusions
A series of Pd catalysts supported on TUD-1 functionalized with
various organosilanes (APS, ATMS, HMDS, and MPTMS) were suc-
cessfully prepared using a surface-modification scheme followed
(
2003) 1199.
32] L. Xu, J.H. Fu, J.R. Schlup, J. Am. Chem. Soc. 116 (1994) 2821.
[33] C.C. Perry, X.C. Li, J. Chem. Soc., Faraday Trans. 87 (1991) 761.

2
4
Y. Chen et al. / Journal of Catalysis 275 (2010) 11–24
[
[
34] X.B. Liu, Y. Du, Z. Guo, S. Gunasekaran, C.B. Ching, Y. Chen, S.S.J. Leong, Y.H.
Yang, Micropor. Mesopor. Mater. 122 (2009) 114.
35] Sujandi, S.E. Park, D.S. Han, S.C. Han, M.J. Jin, T. Ohsuna, Chem. Commun.
[40] E.B. Fox, S. Velu, M.H. Engelhard, Y.H. Chin, J.T. Miller, J. Kropf, C.S. Song, J.
Catal. 260 (2008) 358.
[41] M.C. Romanmartinez, D. Cazorlaamoros, A. Linaressolano, C.S.M. Delecea, H.
Yamashita, M. Anpo, Carbon 33 (1995) 3.
(
2006) 4131.
[
[
[
36] E. Ozensoy, D.W. Goodman, Phys. Chem. Chem. Phys. 6 (2004) 3765.
37] P. Hollins, Surf. Sci. Rep. 16 (1992) 51.
38] Z. Hou, N. Theyssen, A. Brinkmann, K.V. Klementiev, W. Grunert, M. Buhl, W.
Schmidt, B. Spliethoff, B. Tesche, C. Weidenthaler, W. Leitner, J. Catal. 258
[42] X. Feng, G.E. Fryxell, L.Q. Wang, A.Y. Kim, J. Liu, K.M. Kemner, Science 276
(1997) 923.
[43] Q.H. Tang, Q.H. Zhang, H.L. Wu, Y. Wang, J. Catal. 230 (2005) 384.
[44] C. Keresszegi, D. Ferri, T. Mallat, A. Baiker, J. Phys. Chem. B 109 (2005) 958.
[45] D.M. Meier, A. Urakawa, A. Baiker, J. Phys. Chem. C 113 (2009) 21849.
[46] Q.H. Tang, T. Liu, Y.H. Yang, Catal. Commun. 9 (2008) 2570.
[47] Q.H. Tang, Y.T. Chen, C. Zhou, T. Chen, Y.H. Yang, Catal. Lett. 128 (2009) 210.
(
2008) 315.
[
39] V.Z. Radkevich, T.L. Senko, K. Wilson, L.M. Grishenko, A.N. Zaderko, V.Y. Diyuk,
Appl. Catal. A: Gen. 335 (2008) 241.