Effect of supporting surface layers on catalytic activities of gold nanoparticles in CO oxidation

Article abstract of DOI:10.1021/jp052006m

The surfaces of fumed silica materials were modified with a surface sol-gel process for catalysis applications. This surface-modification approach allows not only a monolayer growth of TiO₂ or Al₂O₃ but also a stepwise double-layer growth of TiO₂/TiO₂, Al₂O₃/Al₂O₃, TiO₂/Al ₂O₃, or Al₂O₃/TiO₂ on the surfaces of the silica materials with a monolayer precision. XRD analyses revealed that the coated monolayers and double layers of TiO₂ and Al₂O₃ were amorphous. Gold nanoparticles were successfully deposited on the above six surface-modified silica materials via a deposition-precipitation method. The catalytic activities of these six gold catalysts for CO oxidation are highly dependent on the structures of their surface monolayers or double layers. The gold catalyst supported on the silica material functionalized with a TiO₂ monolayer (Au/TiO₂) is the most active in both as-synthesized and oxidized forms, while the gold catalyst supported on the silica material functionalized with an Al ₂O₃/TiO₂ double layer (Au/Al2O ₃/TiO₂/SiO₂) is the most active in the reduced form among the six catalysts. Surprisingly, the gold catalyst supported on the silica material functionalized with a TiO₂/Al₂O ₃ double layer (Au/TiO₂/Al₂O ₃/SiO₂) has much less activity than Au/Al ₂O₃/TiO₂/SiO₂ under all various treatments, underscoring the sensitivity of the catalytic activity to the structure of the supporting surfaces. ? 2005 American Chemical Society.

Full text of DOI:10.1021/jp052006m

J. Phys. Chem. B 2005, 109, 15489-15496
15489
Effect of Supporting Surface Layers on Catalytic Activities of Gold Nanoparticles in CO
Oxidation
Wenfu Yan, Shannon M. Mahurin, Bei Chen, Steven H. Overbury, and Sheng Dai*
Chemical Sciences DiVision, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831
ReceiVed: April 18, 2005; In Final Form: June 8, 2005
The surfaces of fumed silica materials were modified with a surface sol-gel process for catalysis applications.
This surface-modification approach allows not only a monolayer growth of TiO
double-layer growth of TiO /TiO , Al O /Al O , TiO /Al O , or Al O /TiO
2 2 2 3 2 3 2 2 3 2 3 2
2
or Al
on the surfaces of the silica
materials with a monolayer precision. XRD analyses revealed that the coated monolayers and double layers
of TiO and Al were amorphous. Gold nanoparticles were successfully deposited on the above six surface-
2 3
O but also a stepwise
2
2 3
O
modified silica materials via a deposition-precipitation method. The catalytic activities of these six gold catalysts
for CO oxidation are highly dependent on the structures of their surface monolayers or double layers. The
gold catalyst supported on the silica material functionalized with a TiO
active in both as-synthesized and oxidized forms, while the gold catalyst supported on the silica material
functionalized with an Al /TiO double layer (Au/Al /TiO /SiO ) is the most active in the reduced form
among the six catalysts. Surprisingly, the gold catalyst supported on the silica material functionalized with a
TiO /Al double layer (Au/TiO /Al /SiO ) has much less activity than Au/Al /TiO /SiO under all
2 2
monolayer (Au/TiO ) is the most
2
O
3
2
2
O
3
2
2
2
O
2 3
2
2
O
3
2
O
2 3
2
2
various treatments, underscoring the sensitivity of the catalytic activity to the structure of the supporting
surfaces.
Introduction
they concluded that the nature of the support directly affects
the activity of gold.
Although gold in bulk has often been regarded to be poorly
active as a catalyst, Haruta and co-workers have found that the
gold nanoparticles deposited on selected metal oxides exhibited
surprisingly high catalytic activity for low-temperature CO
Corma and co-workers have recently explored an interesting
aspect of the support promotion effects through controlled
27
variation of the particle size of a CeO2 support. The gold
nanoparticles deposited on the nanocrystals of cerium oxide via
coprecipitation exhibited an increased activity for CO oxidation
by 2 orders of magnitude compared with the gold nanoparticles
deposited on a regular cerium oxide. This observation has
opened up a new dimension in the support studies, indicating
that not only the size of gold nanoparticles but also the size of
the support oxide has a profound influence on the catalysis
properties of gold nanoparticles.
1
oxidation. An extensive literature now exists concerning CO
oxidation and selective propylene oxidation catalyzed by gold
2
,3
nanoparticles supported on various metal oxides, such as
4
-14
1,15-17
Co3O4,1 NiO,1 SiO2,
18-20
21
TiO2,
Fe2O3,
ZrO2,
MgO,2
2-24
and Al2O3.25 Even though the catalytic mechanism
of the gold nanoparticles is not well understood, the nature of
the supports is commonly considered to be one of the key factors
affecting the catalytic activity and stability of supported gold
catalysts. To understand the contribution of oxide supports to
the catalytic activity of gold nanoparticles, intensive studies have
been initiated in the past several years, primarily using different
metal oxide supports. Notably, Grunwaldt et al. prepared two
gold model catalysts on the basis of deposition of “size-
controlled” gold colloids (∼2 nm) on TiO2 and ZrO2 at a pH
In the present study, we report a new support effect induced
by the tailored nanostructure of supporting surface layers on
the catalytic activity of CO oxidation by gold nanoparticles.
The supporting surface layers were a single surface monolayer
of TiO (or Al O ) on silica or surface double layers grown on
2
2
3
silica in all four possible combinations of TiO and Al O . These
2
2
3
supporting surface layers were synthesized on silica surfaces
by a surface sol-gel (SSG) process.
9
lower than the isoelectric points (IEPs) of both metal oxides.
The significant support effect induced by the structural differ-
ence between TiO2 and ZrO2 was observed, showing that the
former is a much more active support than the latter in
controlling the catalytic activity of gold catalysts. More recently,
Arrii et al. directly deposited metallic Au nanoparticles (∼3 nm)
generated via a laser ablation on high-surface-area metal oxides
of γ-Al2O3, ZrO2, and TiO2 to determine the influence of the
supports on catalytic CO oxidation without the hindrance
imposed by the differences in particle size or oxidized gold
species.26 By comparing the activities of the different catalysts,
The surface sol-gel (SSG) process was originally developed
28,29
by Kunitake and co-workers.
This novel technology enables
a molecular-scale control of film thickness over a large 2D
substrate area and can be viewed as a solution-based methodol-
3
0
ogy for atomic layer deposition synthesis. Compared with
conventional deposition methods, the surface sol-gel method
has an advantage in producing very thin, conformal films with
control of the thickness and composition of the films possible
at the atomic level, which is critical for many applications in
catalysis for uniformity of supports. The SSG technique
generally consists of two half-reactions as shown in Scheme 1:
(i) nonaqueous condensation of metal alkoxide precursor
*
Author to whom correspondence should be addressed. E-mail: dais@
ornl.gov.
molecules with surface hydroxyl groups followed by careful
1
0.1021/jp052006m CCC: $30.25 © 2005 American Chemical Society
Published on Web 07/15/2005

15490 J. Phys. Chem. B, Vol. 109, No. 32, 2005
Yan et al.
SCHEME 1. The Schematic Diagram for the Basic Coating Protocol for Ultrathin Metal Oxide Layers over a Large 2D
Substrate Area
TABLE 1: The Details on the Amount of Chemicals Used in the Growth of One-Layer or Multilayers of Titanium Oxide and
Aluminum Oxide on the Surface of Cab-O-Sil SiO
2
precursor (g)
Ti(OBun) Al(OBus)
composites
/SiO
substrate
(g)
4
3
toluene (mL)
methanol (mL)
Al
TiO
Al
TiO
Al
TiO
2
O
3
2
SiO
SiO
TiO
2
9.8
10.2
2.0
2.0
2.0
2.0
N/A
7.9
4.9
N/A
N/A
5.0
7.74
N/A
N/A
4.8
70
70
20
20
20
20
120
70
20
20
20
20
2
/SiO
2
2
2
O
3
/TiO
O
2
/SiO
/SiO
2
2
/SiO
2
2
/Al
2
3
2
Al
Al
TiO
2
O
O
3
/SiO
/SiO
2
2
O
3
/Al
2
3
/SiO
2
2
3
2
5.0
2
/TiO
2
2
/SiO
2
N/A
nonaqueous washing to remove the unreacted alkoxide precursor
and (ii) aqueous hydrolysis of the adsorbed metal alkoxide
species to regenerate surface hydroxyls. The iteration of the
above sequential condensation and hydrolysis reactions allows
the layer-by-layer coating of a selected metal oxide on a
hydroxyl-terminated surface. Although the original technique
was developed to functionalize 2D oxide surfaces, Huang and
Kunitake have recently succeeded in using this technique for
fossilization of cellulosic substances.31 We have recently
extended this methodology to prepare mesoporous materials with
continually stirred for 2 h. The precipitates were separated by
centrifugation and were washed several times with deionized
water. The yellow product was dried at 40 °C overnight to get
the as-synthesized sample.
Catalysis Test. The CO oxidation reaction was carried out
in an AMI 200 (Altamira Instruments). Typically, 50 mg of
catalyst was packed into a 4-mm ID quartz U-tube, supported
by quartz wool. Sample treatments were carried out on the same
2
instrument using either premixed 8% O -He for oxidations or
using 50% H mixed with Ar for reductions. During reactions,
2
13
tailored pore sizes and unique SERS substrates with dielectric
a gas stream of 1% CO balanced with dry air (<4 ppm water)
3
2
overlayers.
was flowed at ambient pressure through the catalyst at a rate
that was adjusted from sample to sample to maintain a constant
3
Experimental Section
space velocity of 44 000 mL/(h‚gcat) or about 37 cm /min. Gas
exiting the reactor was analyzed by a Buck Scientific 910 gas
chromatograph equipped with dual molecular sieve/porous
polymer column (Alltech CTR1) using a thermal conductivity
detector. The reaction temperature was varied using an oven or
by immersing the U-tube in a dewar of ice water or of cooled
acetone by liquid nitrogen, which slowly warmed throughout
the approximately 10-20 h taken to measure a light-off curve.
Characterization. Both transmission electron microscopy
Synthesis of Monolayers and Double Layers. The silica
material used in this investigation is a nonporous fumed silica
(
Cab-O-Sil, CABOT Co. Tuscola, IL), which has a surface area
2
-1
2 33
of 216.5 m g and a silanol density of 4-8 mmol SiOH/m .
The procedure for functionalization of monolayers or double
layers of titanium oxide and aluminum oxide on the silica
surfaces of Cab-O-Sil was developed according to the method
described by Kunitake and co-workers.29 Typically, a pre-
weighed SiO2 powder sample was loaded into a reflux bottle
and dried at 125 °C for 16 h. The bottle was then sealed with
a predried rubber septum. Subsequently, the metal oxide
precursor and the anhydrous mixture of toluene and methanol
were transferred into the bottle through a syringe. The resulting
solution was refluxed for 3 h. The final product was filtered,
washed several times with anhydrous ethanol, hydrolyzed with
deionized water, and dried at 80 °C overnight. The iteration of
the above sequential condensation and hydrolysis reactions
allows the coating of the second monolayer on the one generated
in the previous deposition cycle. The details on the amount of
the chemicals used are summarized in Table 1.
(
TEM) and Z-contrast scanning electron microscopy (STEM)
were carried out using a HITACH HD-2000 STEM operated at
00 kV. Powder X-ray diffraction (XRD) data were collected
via a Siemens D5005 diffractometer with Cu KR radiation (λ
1.5418 Å). Inductively coupled plasma (ICP) analysis was
2
)
performed on IRIS Intrepid II XSP spectrometer (Thermo
Electron Corporation), and each measurement was repeated three
times. Nitrogen gas adsorption measurements (Micromeritics
Gemini) were used to determine the surface areas of surface-
modified Cab-O-Sil samples.
Results and Discussions
Precipitation Deposition of Gold. The deposition-precipita-
tion (DP) procedure used here was based on the method
originally developed by Haruta and co-workers.34 First, 3.0 g
of hydrogen tetrachloroaurate(III) trihydrate (HAuCl4‚3H2O,
The main surfaces discussed here consist of two sequential
combinations of TiO2 and Al2O3 monolayers on amorphous
silica: (1) Al2O3/TiO2/SiO2 (SiO2 coated with a first layer of
TiO2 and then with a second layer of Al2O3) and (2) TiO2/Al2O3/
SiO2 (SiO2 coated with a first layer of Al2O3 and then a second
layer of TiO2). For comparison, we have also prepared the same
silica sample functionalized by TiO2 (TiO2/SiO2) or Al2O3
(Al2O3/SiO2) monolayer or by two sequential layers (double
layer) of TiO2 (TiO2/TiO2/SiO2) or Al2O3 (Al2O3/Al2O3/SiO2).
The schematic diagram for the various combinations of ultrathin
TiO2 and Al2O3 layers on fumed SiO2 are shown in Scheme 2.
9
9.9+%, Aldrich) was dissolved into 500 mL deionized water
to form a gold precursor solution. Typically, the pH value of
the preweighed gold precursor solution (20 mL) was adjusted
to about 10 with vigorous stirring using a solution of 1.0 M
KOH at room temperature. The solution was then heated with
an 80 °C water bath, and a selected surface-modified SiO2 (0.4
g) was added with stirring. The resulting cloudy solution was

Supporting Surface Layers on Catalytic Activities
J. Phys. Chem. B, Vol. 109, No. 32, 2005 15491
, Au/TiO /SiO , Au/Al /Al /SiO , Au/TiO
TABLE 2: The Experimental ICP Results (Metal Loading) for Au/Al
2
O
3
/SiO
2
2
2
2
O
3
2
O
3
2
2
/
a
TiO
2
/SiO
2
, Au/Al
2
O
3
/TiO
2
/SiO
2
2 2 3 2
, and Au/TiO /Al O /SiO
catalyst
Au wt%
Al wt%
Ti wt%
first:second layer (mol)
coverage (first layer)
Au/Al
Au/Al
Au/TiO
Au/TiO
Au/TiO
2
O
O
3
/SiO
/Al
2
7.9
10.3
11.7
5.1
6.8
10.3
4.6
14.2
3.5
N/A
N/A
17.7
N/A
N/A
16.3
0.8
3.1
0.7
N/A
160%
N/A
N/A
13.5%
N/A
N/A
2
3
2
O
3
3
/SiO
2
1:2.1
1:2.6
N/A
1:3.0
1:47.6
2
/Al
2
/SiO
2
/TiO
/TiO
2
O
/SiO
2
2
2
/SiO
/SiO
2
Au/Al
2
O
3
2
2
a
2
2
The BET surface area of Cab-O-Sil is 216.5 m /g. The average surface Si-OH density is 6 µmol/m .
SCHEME 2. The Schematic Diagram for the Basic
Coating Protocol for Ultrathin TiO2 and Al2O3 Layers on
Fumed SiO2
The total titanium content of the double-layer sample TiO2/
TiO2/SiO2 is 3.1 wt %. Accordingly, the mole ratio of the
second-layer titanium content to the first-layer titanium content
is close to 3.0:1.0 on the basis of the ICP analyses. This result
can be explained as follows. The low coverage of the first-
layer functionalization indicates that most of the TiO2 precursor
species are attached to Si-OH groups in the form of -Si-O-
i
Ti(OPr )3 and are highly dispersed on silica surfaces without
cross-linking. Each functionalized surface site [-Si-O-Ti-
i
(
OPr )3] can generate three titanol sites during subsequent
hydrolysis, available for the second-layer functionalization,
which is in perfect agreement with the observed 3:1 ratio. In
this model, the growth of the second layer can be viewed as a
dendritic growth process occurring primarily on the titania layers
and is consistent with the layer-by-layer characteristics of the
2
9
SSG process. However, we cannot eliminate the possibility
that additional growth occurs on the unfunctionalized silanols
during the second-layer depositions.
The elemental compositions of the supporting surfaces were
determined through ICP analyses of the digestion solutions
obtained via dissolution of the surface-modified silica samples
in HF/HCl/HNO3. Table 2 shows the measured weight percents
In the case of the double-layer sample Al2O3/Al2O3/SiO2, the
mole ratio of the second-layer aluminum content to the first-
layer aluminum content is 2.1:1.0, suggesting that some of the
aluminum alkoxide sites generated by the first-layer function-
alization are cross-linked during the hydrolysis step. Therefore,
a limited dendritic growth is expected so that the mole ratio of
the second-layer aluminum content to the first-layer aluminum
content deviates from 3:1. In the case of Al2O3/TiO2/SiO2, the
mole ratio of Al to Ti is as high as 47.6:1 on the basis on ICP
analysis. This high mole ratio indicates that the aluminum
alkoxide not only undergoes the oligomeric attachment but also
reacts with the titanols and the unreacted silanols leftover from
the first-layer functionalization. Hence, the reactivity of Al-
(wt %) of Au, Ti, and Al obtained from ICP results for the six
surface-modified SiO2 supports. It is of interest to determine
the equivalent number of the surface monolayers obtained for
each deposition to provide an indication of the extent to which
the layer-by-layer deposition is achieved by the SSG process
For both single- and double-layer depositions, the Ti and Al
species are assumed to be present as -Ti(OH)3 and -Al(OH)2,
respectively. On the basis of this approximation, the first-layer
coverages were computed in units of the surface silanol density
2
33
(
6 µmole/m ) and are given in Table 2 along with the ratio of
s
n
the second layer to the first layer.
(OBu )3 with silica surfaces is higher than that of Ti(OBu )4
with silica surfaces, which has been observed previously by
Kunitake and co-workers for planar silica surfaces. In the reverse
case in which titania is deposited onto alumina layer (TiO2/
Al2O3/SiO2), the titania reacts very facilely, giving rise to high
Ti loading. The second-to-first layer ratio of 2.6:1 suggests that
the titania functionalizes a high fraction of the alumina hydroxyl
surface species.
The measured titanium content of the TiO2/SiO2 support is
0
.8 wt %, giving rise to a 13% coverage. The low surface
coverage of titanium species indicates that most of the surface
silanols are not functionalized by titanium species via formation
of Si-O-Ti linkages. Either a high fraction of silanols are
unavailable for functionalization or the functionalization is
kinetically limited. In contrast, the aluminum content of the
Al2O3/SiO2 support is equivalent to a 160% coverage. This
overcoverage of SiO2 surfaces by Al2O3 via the SSG process
can be attributed to two factors. First, the reactivity of aluminum
alkoxide with SiOH groups may be higher than that of titanium
alkoxide, which allows most of the SiOH sites to react with
aluminum alkoxide to form Si-O-Al linkages. Second, dimeric
and oligomeric aluminum species were attached to a single SiOH
site. The existence of dimeric and tetrameric aluminum alkoxides
has been well documented in the literature. For example, both
The elemental analyses of samples Au/Al2O3/SiO2, Au/TiO2/
SiO2, Au/Al2O3/Al2O3/SiO2, Au/TiO2/TiO2/SiO2, Au/Al2O3/
TiO2/SiO2, and Au/TiO2/Al2O3/SiO2 for gold contents showed
that the gold-weight loadings of these six samples were 7.9,
5.1, 10.3, 6.8, 10.3, and 11.7%, respectively. The above
differences in the gold loadings introduced via DP can be
qualitatively attributed to the surface-charging properties of these
six samples. For example, the isoelectric points (IEPs) of SiO2,
TiO2, and Al2O3 are 2.2, 6.6, and 9.1, respectively. Accordingly,
the surface modified with Al2O3 is more positively charged than
the unmodified silica or the surface modified with TiO2 under
similar aqueous conditions (pH 10). Therefore, the alumina-
modified surface should interact more favorably with negatively
charged gold precursors during the DP process. This explains
why the gold loading of Au/Al2O3/SiO2 is greater than that of
35
spectroscopical investigations and single-crystal X-ray struc-
ture determinations36 have shown the tetrametric character of
aluminum isoproxide in both solution and solid states. The
t
37
dimeric species of [Al(OBu )3]2 has also been reported.
Therefore, a single SiOH site can attach two or more aluminum
species, leading to the overcoverage of silica surfaces.

15492 J. Phys. Chem. B, Vol. 109, No. 32, 2005
Yan et al.
Figure 1. The simulated XRD pattern of Au and experimental XRD patterns for both as-synthesized and after running the light-off curve of the
as-synthesized catalysts without pretreatment (reaction I).
Au/TiO2/SiO2 and why the gold loadings on the titania only
surfaces are lower than those of the alumina-terminated surfaces.
Similarly, because the gold precursor species interacts weakly
or even repulsively with the silica surface, the low Ti loading
is also a factor in causing low Au loadings for the samples
functionalized with a titania single layer or double layer. In this
case, it is expected that the titania moieties provide the platform
for Au hydroxide nucleation. Finally, the Au/TiO2/Al2O3/SiO2
sample has a high loading because in this case, a high, probably
uniform Ti loading is achieved during the second-layer func-
tionalization. The XRD analyses shown in Figure 1 revealed
that both monolayers and double layers on the surfaces of the
fumed silica were amorphous. For the as-synthesized catalysts,
only a very weak Au(111) peak was observed in the XRD
patterns, which indicated that most of the Au species in the
as-synthesized catalysts are in a cationic oxidation state
dark-field TEM image of the as-synthesized Au/Al2O3/Al2O3/
SiO2. In Z-contrast TEM, heavy atoms (such as gold) stand out
very clearly on a light background of silicon, aluminum,
titanium, and oxygen. The tiny, highly uniform bright spots
(0.8-4.0 nm diameter) in Figure 2a correspond to the deposited
gold precursor species.
For all catalysts, cationic gold species were reduced by the
reactant mixture during the catalytic measurements, as indicated
by the color change from yellow to black. The XRD analyses
of the samples after catalytic measurements (Figure 1) indicate
the appearance of very sharp XRD peaks corresponding to
reduced Au for the catalysts containing the aluminum-oxide
overlayers. However, only very broad Au XRD peaks are
observed for the catalysts containing only titanium-oxide
overlayers. These observations reveal that the gold nanoparticles
on Al2O3 surfaces are more susceptible to aggregation than those
on TiO2 surfaces even under ambient conditions. The Z-contrast
TEM investigation shown in Figure 2b also revealed that there
was a significant aggregation during the reduction of the as-
synthesized Au/Al2O3/Al2O3/SiO2. However, considerable num-
(hydroxide) and are highly dispersed. The high dispersion of
the cationic gold precursors for the as-synthesized catalysts was
also confirmed by a high-angle, annular dark field STEM (Z-
contrast TEM) investigation. For example, Figure 2 a shows a

Supporting Surface Layers on Catalytic Activities
J. Phys. Chem. B, Vol. 109, No. 32, 2005 15493
2 3 2 3 2
Figure 2. The Z-contrast TEM images of the as-synthesized (a) and reduced (b) Au/Al O /Al O /SiO catalysts.
Figure 3. The light-off curves of Au/Al
Al /SiO . These light-off curves were measured from the samples of both as-synthesized and pretreated under the following conditions: 150 °C
in H (50% H /He) and 150 °C in O (8% O /He).
2 3 2 2 2 2 3 2 3 2 2 2 2 2 3 2 2 2
O /SiO , Au/TiO /SiO , Au/Al O /Al O /SiO , Au/TiO /TiO /SiO , Au/Al O /TiO /SiO , and Au/TiO /
O
2 3
2
2
2
2
2
bers of gold nanoparticles with size <3 nm still exist in the
reduced sample.
O2/He). Activities can be qualitatively assessed by the temper-
atures (T50) at which a 50% conversion of CO to CO2 is
achieved. The values for T50 are collected in Table 3.
As a comparison, a sample of unmodified fumed silica was
prepared by the identical DP method. This sample shows no
measurable CO conversion even at temperatures as high as 160
°C (not shown). This observation is consistent with the results
Figure 3 shows the catalytic light-off curves for Au catalysts
derived from the six layered supports tested for CO oxidation.
These light-off curves were measured for both as-synthesized
and posttreated samples. Two posttreatment conditions were
used: 150 °C in H2 (50% H2/He) and 150 °C in O2 (8%

15494 J. Phys. Chem. B, Vol. 109, No. 32, 2005
Yan et al.
TABLE 3: Activity (T50 °C) for Catalysts after Various
For the as-synthesized catalyst, the T50 value is -44 °C.
Treatment at 150 °C in H2 decreased the activity slightly
(T50 ) -32 °C), while the treatment at the same temperature
in O2 increased the activity of this catalyst (T50 ) -54 °C). A
single, fractional (0.13) monolayer of titania, grafted to the silica,
can completely alter the behavior of the resulting Au catalysts
compared to the unmodified silica support. Its activity is higher
than the Au catalysts derived from either Al2O3-modified silica
despite lower Au loadings. The further surface modification of
the TiO2-modified fumed silica by a second TiO2 monolayer
has little further effect upon the activity. The as-synthesized
samples of single-layer and double-layer derived catalysts have
virtually the same T50 values, as they do after treatment at 150
Pretreatments
Au loading
wt%
150 °C 150 °C
in H in O
catalyst
as-syn.
unstable N/A
7.8
-22
-44
-45
-31
2
2
Au/Al
Au/Al
2
O
O
3
/SiO
/Al
2
7.9
10.3
11.7
5.1
6.8
10.3
N/A
43
2
3
2
O
3
/SiO
2
12
Au/TiO
Au/TiO
Au/TiO
2
2
2
/Al
/SiO
/TiO
/TiO
2
O
3
/SiO
2
-11
-32
-95
-18
-54
-55
-32
2
2
/SiO
/SiO
2
Au/Al
2
O
3
2
2
<-98
previously reported.13 Because the IEP of SiO2 is only ∼2.2,
the interaction of SiO2 with gold precursor species is very weak
during DP (pH ) 6-10), resulting in very small gold loading
but large gold particles. Furthermore, the color of the sample
obtained via DP of the gold species on the fumed silica is purple,
°
C in O2. Clearly, the catalytic activity depends weakly on the
number of the monolayers under these reaction and treatment
conditions. Treating in H2 at 150 °C dramatically altered the
light-off curve of catalyst Au/TiO2/TiO2/SiO2, giving rise to an
anomalously slow-temperature variation in the conversion below
-50 °C. This observation is consistent with those reported
which indicates that the gold species on the SiO2 surface are
reduced.38 The TEM imaging of the gold-deposited sample
reveals that the size of the gold nanoparticles is too big (>8
nm) to be catalytically active.13
39
previously. Haruta attributed this anomalous catalytic activity
to the formation of carbonate species condensed on the catalyst
No clear improvement of the catalytic activities was observed
for the gold nanoparticles on the fumed silica whose surface
was modified with an Al2O3 monolayer through the SSG
process. The as-synthesized Au/Al2O3/SiO2 sample had a very
low conversion at high temperature but was quickly deactivated.
The initial apparent conversion can be attributed to stoichio-
metric CO consumption by the reduction of the gold cationic
species, which disappeared after the gold cationic species were
totally reduced. This conjuncture is also consistent with the
observations that the catalysts pretreated at 150 °C both in H2
and in O2 exhibit no activities. Sharp Au lines in the XRD
measurements in Figure 1 prove that large Au particles are
formed under the reaction conditions, and this large particle size
is the probable explanation of the poor activity. Compared with
the fumed silica modified with a single Al2O3 monolayer, the
substrate modified with an Al2O3 double layer had a higher gold
loading, 10.3 wt % versus 7.9 wt %, and it exhibited moderate,
stable activity for CO oxidation. The as-synthesized catalyst
reached 50% conversion at 7.8 °C, but T50 was increased by
the treatments at 150 °C in either H2 or O2. The higher activity
obtained for the double-layer catalysts compared to the single
layer cannot be explained by the 30% increase in Au loading.
Therefore, the second layer of Al2O3 must exhibit some
additional effect upon the catalytic activity.
3
9
surfaces at low temperature.
The deposition of gold nanoparticles on the mixed double-
layer-coated supports (Au/TiO2/Al2O3/SiO2 and Au/Al2O3/TiO2/
SiO2) also gave rise to two active catalysts for CO oxidation.
Of the two mixed layer supported catalysts in either the as-
synthesized state or after treatment at 150 °C in O2, the sample
with the outer alumina termination (Au/Al2O3/TiO2/SiO2)
provides slightly lower T50 and therefore higher activity.
Similarly as for the catalyst functionalized with a TiO2 double
layer, treating in H2 at 150 °C dramatically altered the light-off
curve of catalyst Au/Al2O3/TiO2/SiO2, giving rise to an anoma-
lously slow-temperature variation in the conversion below -50
°
C but apparently very high activity.
The differences between Au/TiO2/Al2O3/SiO2 and Au/Al2O3/
TiO2/SiO2 in the catalytic activities indicate the sensitivity of
gold catalysts to the local surface structures. Figure 4 shows
the high-magnification Z-contrast TEM images of Au/Al2O3/
TiO2/SiO2 (a) and Au/TiO2/Al2O3/SiO2 (b) after the treatments
at 150 °C in H2. The Z-contrast images recorded at identical
magnification show a clear difference in size distribution (see
Supporting Materials, Figure s1, the particle size distribution
calculated from TEM images). More small gold nanoparticles
(<2 nm) were found on the surface of the reduced Au/Al2O3/
TiO2/SiO2, while the population of small gold nanoparticles was
considerably less on the surface of the reduced Au/TiO2/Al2O3/
Compared to the Al2O3-modified fumed silica, the single-
layer TiO2-modified fumed silica (TiO2/SiO2) is a better
substrate for the preparation of highly active gold nanocatalysts.
2 3 2 2 2 2 3 2
Figure 4. The Z-contrast transmission electron microscopy (TEM) images of the reduced Au/Al O /TiO /SiO (a) and Au/TiO /Al O /SiO (b) at
1
50 °C in H
2
.

Supporting Surface Layers on Catalytic Activities
J. Phys. Chem. B, Vol. 109, No. 32, 2005 15495
TABLE 4: Au Particle Diameters in Reduced Samples
Calculated from Au(111) Peak
In conclusion, a surface-modification methodology based on
the surface sol-gel synthesis has been developed to tailor the
surfaces of amorphous silica materials for catalysis applications.
The essence of this methodology is to generate the amorphous
oxide layers with controlled composition and comformity on
silica surfaces. Both surface compositions and postsynthesis
treatments of the catalyst supports have been demonstrated to
play important roles in determining the activities and stabilities
of the gold nanocatalysts. The activity of Au/Al2O3/TiO2/SiO2
is higher than that of Au/TiO2/Al2O3/SiO2 for CO oxidation.
This observation indicates that the surface-functionalization
sequence is an important factor in determining the catalytic
activities of the gold nanoparticles deposited on the surface-
modified fumed silica supports.
Au/Al
SiO
2
O
2
3
/
Au/Al
2
O
3
/
2
Au/TiO
2
/
catalyst
Al /SiO
2
O
3
Al /SiO
2
O
3
2
diameter (nm)
24.5
8.0
21.3
Au/TiO
2
/
Au/TiO
2
/
Au/Al
2 3
O /
catalyst
SiO
2
TiO /SiO
2
2
TiO /SiO
2
2
diameter (nm)
2.7
3.8
7.7
SiO2. This observation was consistent with the analysis results
of the corresponding XRD patterns of metallic gold in Figure
1
(see the discussions in the next paragraph). Accordingly, a
stronger interaction of gold nanoparticles with the Al2O3/TiO2/
SiO2 surface than that with the TiO2/Al2O3/SiO2 surface is
indicated, evidently because of a higher barrier for particle
coarsening on the former surface. This conclusion is supported
by addition treatments performed at 300 and 500 °C. The activity
remains higher for the Au/Al2O3/TiO2/SiO2 sample following
these treatments. (see Figure s2, Supporting Materials).
Acknowledgment. This work was conducted at the Oak
Ridge National Laboratory and was supported by the Division
of Chemical Sciences, Office of Basic Energy Sciences, U.S.
Department of Energy, under contract No. DE-AC05-00OR22725
with UT-Battelle, LLC. This research was supported in part by
an appointment for W.Y. to the Oak Ridge National Laboratory
Postdoctoral Research Associates Program administered jointly
by the Oak Ridge Institute for Science and Education and Oak
Ridge National Laboratory.
To correlate metallic gold particle sizes to catalytic activities,
the diffraction peak of Au(111) in each reduced XRD pattern
shown in Figure 1 was fitted to a Lorentzian distribution using
2
a Levenberg-Marquardt ø minimization procedure to obtain
an estimation of average particle sizes on the basis of the
corresponding full width at half-maximum (fwhm). The gold-
particle diameters were subsequently calculated from the fwhm’s
of the XRD peaks using Sherrer’s equation:40
Supporting Information Available: Au particle size dis-
tribution in both reduced Au/Al O /TiO /SiO and Au/TiO /
2
3
2
2
2
Al O /SiO calculated from their TEM images, Figure s1. The
2
3
2
light-off curves of Au/Al2O3/TiO2/SiO2 and Au/TiO2/Al2O3/SiO2
treated at 300 °C and 500 °C in O2, Figure s2. This material is
available free of charge via the Internet at http://pubs.acs.org.
kλ
â cos θ
D )
,
where k is a constant (0.94), λ is the wavelength of the radiation
References and Notes
(
Cu KR ) 1.5418 Å), â is the fwhm of the peak in radians, and
(
1) Haruta, M.; Yamada, N.; Kobayashi, T.; Iijima, S. J. Catal. 1989,
15, 301.
(2) Bond, G. C.; Thompson, D. T. Catal. ReV.sSci. Eng. 1999, 41,
19.
θ is the Bragg angle. The analyses results are summarized in
Table 4. As seen from Table 4, the average Au particle size of
the reduced Au/TiO2/SiO2 sample is the smallest among the
six catalysts. The high catalytic activity of this sample (Table
1
3
(3) Hutchings, G. J. Gold Bull. 2004, 37, 3.
(4) Moreau, F.; Bond, G. C.; Taylor, A. O. Chem. Commun. 2004,
3
) indicates that the high dispersion and stabilization of small
1
642.
(5) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647.
(6) Chen, M. S.; Goodman, D. W. Science 2004, 306, 252.
7) Stangland, E. E.; Stavens, K. B.; Andres, R. P.; Delgass, W. N. J.
Catal. 2000, 191, 332.
8) Lin, S. D.; Bollinger, M.; Vannice, M. A. Catal. Lett. 1993, 17,
gold particles is a key factor in determining catalytic activities
of gold catalysts. The Au particle size of the reduced Au/Al2O3/
TiO2/SiO2 sample is about 3 times smaller than that of the
reduced Au/TiO2/Al2O3/SiO2 sample, which seems to be
inconsistent with the particle-size distribution obtained from the
high magnification Z-contrast TEM images (Figure 4). To
investigate this inconsistency, the Z-contrast TEM imaging of
both samples with a lower magnification was conducted to
survey larger sample areas. At the lower magnification, many
particles distributed in the range of 10-40 nm are visible on
reduced Au/TiO2/Al2O3/SiO2, which are not observed on
reduced Au/Al2O3/TiO2/SiO2 (see Supporting Information,
Figure S3). This observation is qualitatively consistent with the
result from the XRD analysis and is in a qualitative agreement
with the results of the catalysis activities showing that the latter
catalyst is more active than the former. The average Au-particle
size of the reduced Au/Al2O3/Al2O3/SiO2 sample is comparable
with that of the reduced Au/Al2O3/TiO2/SiO2 sample but is about
(
(
2
45.
(9) Grunwaldt, J. D.; Kiener, C.; Wogerbauer, C.; Baiker, A. J. Catal.
999, 181, 223.
10) Grunwaldt, J. D.; Baiker, A. J. Phys. Chem. B 1999, 103, 1002.
11) Dekkers, M. A. P.; Lippits, M. J.; Nieuwenhuys, B. E. Catal. Lett.
1
(
(
1
998, 56, 195.
(12) Yan, W. F.; Chen, B.; Mahurin, S. M.; Dai, S.; Overbury, S. H.
Chem. Commun. 2004, 1918.
13) (a) Yan, W. F.; Chen, B.; Mahurin, S. M.; Hagaman, E. W.; Dai,
S.; Overbury, S. H. J. Phys. Chem. B 2004, 108, 2793. (b) Yan, W. F.;
Petkov, V.; Mahurin, S. M.; Overbury, S. H.; Dai, S. Catal. Commun. 2005,
6, 404.
14) Yan, W. F.; Chen, B.; Mahurin, S. M.; Schwartz, V.; Mullins, D.
R.; Lupini, A. R.; Pennycook, J.; Dai, S.; Overbury, S. H. J. Phys. Chem.
B 2005, 109, 10676.
15) Kahlich, M. J.; Gasteiger, H. A.; Behm, R. J. J. Catal. 1999, 182,
430.
16) Schubert, M. M.; Hackenberg, S.; van Veen, A. C.; Muhler, M.;
Plzak, V.; Behm, R. J. J. Catal. 2001, 197, 113.
17) Boccuzzi, F.; Chiorino, A.; Manzoli, M.; Andreeva, D.; Tabakova,
T. J. Catal. 1999, 188, 176.
(18) Liu, J. H.; Wang, A. Q.; Chi, Y. S.; Lin, H. P.; Mou, C. Y. J.
Phys. Chem. B 2005, 109, 40.
19) Date, M.; Okumura, M.; Tsubota, S.; Haruta, M. Angew. Chem.,
Int. Ed. 2004, 43, 2129.
20) Overbury, S. H.; Ortiz-Soto, L.; Zhu, H. G.; Lee, B.; Amiridis, M.
D.; Dai, S. Catal. Lett. 2004, 95, 99.
(
(
(
(
3
times smaller than that of the reduced Au/ TiO2/Al2O3/SiO2
sample. Surprisingly, the catalytic activity of the reduced Au/
Al2O3/Al2O3/SiO2 sample is much worse than those of the
reduced Au/TiO2/Al2O3/SiO2 sample and the reduced Au/Al2O3/
TiO2/SiO2 sample. This observation indicates that the Au particle
size is not the only factor in determining the catalytic activity
of gold catalysts. The support compositions also have a profound
influence.
(
(
(

15496 J. Phys. Chem. B, Vol. 109, No. 32, 2005
Yan et al.
(
21) Knell, A.; Barnickel, P.; Baiker, A.; Wokaun, A. J. Catal. 1992,
37, 306.
22) Yoon, B.; Hakkinen, H.; Landman, U.; Worz, A. S.; Antonietti, J.
M.; Abbet, S.; Judai, K.; Heiz, U. Science 2005, 307, 403.
(32) Bao, L. L.; Mahurin, S. M.; Dai, S. Anal. Chem. 2004, 76, 4531.
(33) Nawrocki, J. J. Chromatogr., A 1997, 779, 29.
(34) Tsubota, S.; Haruta, M.; Kobayashi, T.; Ueda, A.; Nakahara, Y.
Stud. Surf. Sci. Catal. 1991, 63, 695.
1
(
(
(
(
(
23) Guzman, J.; Gates, B. C. J. Am. Chem. Soc. 2004, 126, 2672.
24) Molina, L. M.; Hammer, B. Phys. ReV. Lett. 2003, 90, 206102.
25) Bethke, G. K.; Kung, H. H. Appl. Catal., A 2000, 194, 43.
26) Arrii, S.; Morfin, F.; Renouprez, A. J.; Rousset, J. L. J. Am. Chem.
(35) Mehrotra, R. C.; Rothwell, I.; Singh, A.; Bradley, D. C. Alkoxo
and Aryloxo DeriVatiVes of Metals; Academic Press: San Diego, CA, 2002.
(36) Folting, K.; Streib, W. E.; Caulton, K. G.; Poncelet, O.; Hubert-
pfalzgraf, L. G. Polyhedron 1991, 10, 1639.
(37) Cayton, R. H.; Chisholm, M. H.; Davidson, E. R.; Distasi, V. F.;
Du, P.; Huffman, J. C. Inorg. Chem. 1991, 30, 1020.
(38) Mukherjee, P.; Patra, C. R.; Ghosh, A.; Kumar, R.; Sastry, M. Chem.
Mater. 2002, 14, 1678.
Soc. 2004, 126, 1199.
27) Carrettin, S.; Concepcion, P.; Corma, A.; Nieto, J. M. L.; Puntes,
V. F. Angew. Chem., Int. Ed. 2004, 43, 2538.
(
(28) Ichinose, I.; Senzu, H.; Kunitake, T. Chem. Lett. 1996, 831.
(29) Ichinose, I.; Senzu, H.; Kunitake, T. Chem. Mater. 1997, 9, 1296.
(30) Leskela, M.; Ritala, M. Angew. Chem., Int. Ed. 2003, 42, 5548.
(31) Huang, J.; Kunitake, T. J. Am. Chem. Soc. 2003, 125, 11834.
(39) Haruta, M. Cattech 2002, 6, 102.
(40) Suryanarayana, C.; Norton, M. G. X-ray diffraction; Plenum
Press: New York, 1998.