Au/MxOy/TiO2 catalysts for CO oxidation: Promotional effect of main-group, transition, and rare-earth metal oxide additives

Article abstract of DOI:10.1016/j.molcata.2007.04.007

Au/TiO₂ catalysts are active for CO oxidation, but they suffer from high-temperature sintering of the gold particles, and few attempts have been made to promote or stabilize Au/TiO₂. Our recent communication addressed these issues by loading gold onto Al₂O₃/TiO₂ prepared via surface-sol-gel processing of Al(sec-OC₄H₉)₃ on TiO₂ [W.F. Yan, S.M. Mahurin, Z.W. Pan, S.H. Overbury, S. Dai, J. Am. Chem. Soc. 127 (2005) 10480-10481]. In our current full paper, Au/Al₂O₃/TiO₂ catalysts were prepared alternatively by thermal decomposition of Al(NO₃)₃ on TiO₂ followed by loading gold, and the influences of the decomposition temperature and Al₂O₃ content were systematically surveyed. This facile method was subsequently extended to the preparation of a battery of metal oxide-modified Au/TiO₂ catalysts virtually not reported. It was found that Au/TiO₂ modified by CaO, NiO, ZnO, Ga₂O₃, Y₂O₃, ZrO₂, La₂O₃, Pr₂O₃, Nd₂O₃, Sm₂O₃, Eu₂O₃, Gd₂O₃, Dy₂O₃, Ho₂O₃, Er₂O₃, or Yb₂O₃ could retain significant activity at ambient temperature even after aging in O₂-He at 500 °C, whereas unmodified Au/TiO₂ lost its activity. Moreover, some 200 °C-calcined promoted catalysts showed high activity even at about -100 °C. The deactivation and regeneration of some of these new catalysts were studied. This work furnished novel catalysts for further fundamental and applied research.

Full text of DOI:10.1016/j.molcata.2007.04.007

Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
Au/M O /TiO catalysts for CO oxidation: Promotional effect of
x y
2
main-group, transition, and rare-earth metal oxide additives
∗
Zhen Ma, Steven H. Overbury, Sheng Dai
Chemical Sciences Division, Oak Ridge National Laboratory, TN 37831, USA
Received 20 March 2007; received in revised form 2 April 2007; accepted 4 April 2007
Available online 8 April 2007
Abstract
Au/TiO
2
catalysts are active for CO oxidation, but they suffer from high-temperature sintering of the gold particles, and few attempts have
been made to promote or stabilize Au/TiO
2
. Our recent communication addressed these issues by loading gold onto Al
on TiO [W.F. Yan, S.M. Mahurin, Z.W. Pan, S.H. Overbury, S. Dai, J. Am. Chem. Soc. 127
2005) 10480–10481]. In our current full paper, Au/Al /TiO catalysts were prepared alternatively by thermal decomposition of Al(NO
followed by loading gold, and the influences of the decomposition temperature and Al content were systematically surveyed. This
catalysts virtually not reported. It was
, Gd , Dy , Ho , Er , or
lost its activity.
2 3 2
O /TiO prepared via
surface-sol–gel processing of Al(sec-OC
4
H
9
)
3
2
(
2
O
3
2
3 3
)
on TiO
2
2 3
O
facile method was subsequently extended to the preparation of a battery of metal oxide-modified Au/TiO
2
found that Au/TiO
2
modified by CaO, NiO, ZnO, Ga
2
O
3
, Y
2
O
3
, ZrO
2
, La
2
O
3
, Pr
2
O
3
, Nd
2
O
3
, Sm
2
O
3
, Eu
2
O
3
2
O
3
2
O
3
2
O
3
2 3
O
◦
◦
Yb
2
O
3
could retain significant activity at ambient temperature even after aging in O
2
–He at 500 C, whereas unmodified Au/TiO
2
◦
Moreover, some 200 C-calcined promoted catalysts showed high activity even at about −100 C. The deactivation and regeneration of some of
these new catalysts were studied. This work furnished novel catalysts for further fundamental and applied research.
Published by Elsevier B.V.
Keywords: Gold catalysis; CO oxidation; Titania; Metal oxide; Additive; Sintering
1
. Introduction
stabilized on Al2O3/TiO2 synthesized via surface-sol–gel pro-
cessing of Al(sec-OC4H9)3 on TiO2 [22,28]. Tai et al. [29] and
Venezia et al. [30] reported parallel findings with TiO2/SiO2
supports prepared from Ti(iso-OC3H7)4 precursor.
Heterogeneous catalysis by gold, previously uncharted [1,2],
has become an emerging field recently [3–5]. Supported gold
catalysts have many applications in environmental control [6–9],
chemical synthesis [10–13], energy generation [14,15], and
materials processing [16,17]. In particular, Haruta et al. initially
reported that supported gold nanoparticles were active for low-
temperatureCOoxidation[3,7]. However, duetothelowmelting
Although the performance of Au/Al2O3/TiO2 is impressive
[22,28], several questions remain unclear: (1) is the surface-
sol–gel method necessary, or put it another way, is it possible
to prepare Au/Al2O3/TiO2 without involving non-aqueous con-
dition, organic solvents, and aluminum alkoxide, while still
achieving comparable performance? (2) Is “monolayer” [19,23]
Al2O3 required, or what is the effect of the Al2O3 content?
(3) Is Al2O3 unique, or will other metal oxide additives exhibit
similar, or perhaps even better, promotional effect? (4) Are
these modified gold catalysts stable on stream? As commented
by Bond and Thompson, one of the unsatisfactory aspects of
gold catalysis research is that possible changes in conversion
with time on stream typically are not reported or only qual-
itatively mentioned [6]. In the literature, Au/Al2O3 is known
to suffer from deactivation on stream [31,32], but the tem-
poral stability of Au/Al2O3/TiO2 has not been investigated
point of gold nanoparticles, the sintering of gold catalysts above
◦
4
00 C is a drawback that may constrain their practical appli-
cations [7,18,19]. To address this issue, our group has used CO
oxidation as a sensitive probe reaction [3–9], and studied differ-
ent supports such as polymorphs of TiO2 [20,21], Al2O3/TiO2
22], Al2O3/TiO2/SiO2 [23], PO43 /TiO2 [24], mesoporous
25] or fumed [26] SiO2, and nano-crystalline LaPO4 [27]. Par-
−
[
[
ticularly interesting is the finding that gold particles could be
∗
Corresponding author. Tel.: +1 865 576 7307; fax: +1 865 576 5235.
[22,28].
E-mail address: dais@ornl.gov (S. Dai).
1
381-1169/$ – see front matter. Published by Elsevier B.V.
doi:10.1016/j.molcata.2007.04.007

Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
187
The objective of the current work is not only to clarify these
important issues, but also to explore novel gold catalysts pro-
moted by metal oxides. In particular, we explored the promotion
of Au/TiO2 because few attempts have been made to promote
it [3–9]. After demonstrating that Au/Al2O3/TiO2 prepared via
decomposition of Al(NO3)3 on TiO2 could show comparable
performance in CO oxidation as those made via surface-sol–gel
processing of Al(sec-OC4H9)3 [22], we extended this sponta-
neous dispersion method [33,34] to the preparation of a number
of surface-modified TiO2 supports. The key finding is that
Au/TiO2 modified by CaO, NiO, ZnO, Ga2O3, Y2O3, ZrO2,
La2O3, Pr2O3, Nd2O3, Sm2O3, Eu2O3, Gd2O3, Dy2O3, Ho2O3,
Er2O3, or Yb2O3 could retain smaller gold particles and exhibit
by a glass dish, and the slurry magnetically stirred for 2 h.
The final pH value after deposition–precipitation was measured
using a pH meter and recorded in Tables 2–4. The precipitate
was centrifuged, washed three times with water and once with
ethanol [20,22]. In each washing-centrifugation cycle, the solid
deposited at the bottom of the centrifuge tube was re-dispersed
using a vortexer, and then re-centrifuged. The final washing with
ethanol[20,22]wasfoundtoreducethesupportedgoldprecursor
to metallic gold particles, analogous to the reduction of sup-
+
2+
ported Ag by formaldehyde [35] and that of Pd by methanol
◦
[36]. The product was dried at 40 C for 2 days, and referred
to as “as-synthesized” catalyst. The catalyst was denoted as
Au/c-MxOy-d/TiO2, where c means the temperature used to cal-
cine the MxOy/TiO2 precursor and d the loading in terms of d g
metal oxide per g TiO2, and was simplified as Au/MxOy/TiO2 if
c = 500 and d = 0.05 (which is usually the case for the majority
of catalysts reported here).
significant activity at ambient temperature even after aging at
◦
5
00 C, whereas unmodified Au/TiO2 lost its activity. More-
over, some catalysts showed extraordinarily high activity as
◦
low as −100 C. To the best of our knowledge, the promotion
of Au/TiO2 by these metal oxide additives for CO oxidation
has virtually not been reported in the open literature. In our
recent patent on surface-stabilized gold nanocatalysts [28], these
freshly obtained findings were not claimed, either. Hence, this
work furnished a novel gold catalyst system for further funda-
mental and applied research.
Catalytic CO oxidation was tested in a plug-flow microreac-
tor (Altamira AMI 200). Typically, 50 mg catalyst was loaded
into a U-shaped quartz tube (4 mm i.d.), except that in some sta-
bility tests, the catalyst load was less than 50 mg if the regular
load (50 mg) would lead to 100% conversion. The exact catalyst
load under such circumstances will be specified in Figs. 7–9.
The catalyst was pretreated in flowing 5% O2 (balance He) at
◦
2
. Experimental
200 or 500 C for 2.5 h. The catalyst was cooled down, the gas
stream switched to 1% CO (balanced air) and the reaction tem-
◦
HAuCl4·3H2O, TiO2 (Degussa P25), Al(NO3)3·9H2O,
perature ramped using a furnace (at a rate of 0.5 C/min above
Ca(NO3)2·4H2O, Fe(NO3)3·9H2O (EM), Ni(NO3)2·6H2O, Cu
ambient temperature) or by immersing the U-shaped tube in ice-
water or acetone–liquid nitrogen mixture to record the light-off
(
NO3)2·2.5H2O, Zn(NO3)2·6H2O, Ga(NO3)3·xH2O, Y(NO3)3·
3
4
H2O, ZrO(NO3)2·xH2O, (NH4) Mo7O24·4H2O, La(NO3)3·
curve. The flow rate of reactant stream was 37 cm /min, and the
6
3
xH2O, Ce(NO3)3·6H2O, Pr(NO3)3·6H2O, Nd(NO3)3·6H2O,
space velocity was 44,400 cm /(h·gcat). A portion of the product
Sm(NO3)3·6H2O, Eu(NO3)3·5H2O, Gd(NO3)3·6H2O, Dy
stream was extracted periodically with an automatic sampling
valve, and analyzed using a dual-column gas chromatograph
with a thermal conductivity detector.
(
5
NO3)3·5H2O, Ho(NO3)3·5H2O, Er(NO3)3·5H2O, Yb(NO3)3·
H2O, (NH4) H2W12O40·xH2O, H3PW12O40·xH2O, H4SiW12
6
O40·xH2O, and Bi(NO3)3·5H2O were obtained from Aldrich
X-ray diffraction (XRD) data were collected on a Siemens
D5005 diffractometer with Cu K␣ radiation. Scans were per-
formed at a rate of 0.01 /s, typically in the range of 2θ = 20–85 .
The average gold particle sizes were estimated from X-ray line
broadening analysis applying the Debye–Scherrer equation on
unless otherwise indicated.
◦
◦
Metal oxide-modified TiO2 supports were prepared via
excess-solution impregnation followed by spontaneous disper-
sion upon calcination [33,34]. A calculated amount of metal
oxide precursor, usually a soluble metal nitrate, was dissolved
in deionized water (typically 10 ml water per gram of TiO2),
and TiO2 was subsequently added. The intended loading was
◦
the (2 2 0) diffraction (2θ = 44 ) of gold. Elemental analysis was
performed using inductivity coupled plasma-optical emission
spectrometry (ICP-OES) on a Thermo IRIS Intrepid II spec-
trometer. Typically, accurately weighed 30–100 mg of sample
0
0
.05 g MxOy per g TiO2, except that for Al2O3/TiO2, it was
.01–1.0 g Al2O3 per gram of TiO2. The slurry was magnetically
◦
was dissolved into HCl–HNO3–HF acids at 70 C, diluted by
stirred at room temperature until dry. The product was dried at
deionized water to 200 ml, and analyzed along with known
solutions. BET surface areas of samples were measured by N2
adsorption–desorption at 77 K using a Micromeritics Gemini
instrument.
◦
8
5 C overnight, collected and ground into fine powders, and
◦
further calcined at 500 C for 3 h (except that for the prepara-
tion of Al2O3/TiO2, the Al(NO3)3/TiO2 precursor was either not
calcined or calcined at 200–700 C). Al2O3 was prepared by cal-
◦
◦
cining Al(NO3)3·9H2O at 500 C for 3 h (except that in control
3. Results
experiments reported in Table S1 and Fig. S1 in the Support-
◦
ing Information, it was not calcined or calcined at 100–800 C).
3.1. Au/Al2O3/TiO2 catalysts
Goldwasloadedviaadeposition–precipitationmethod[3–9].
First, 50 ml HAuCl4 solution containing 0.3 g HAuCl4·3H2O
was poured into a beaker, and the pH value adjusted to 10.0
by 1.0 M KOH with vigorous stirring. The resulting solution
3.1.1. General effects of Al2O3 surface modification
Fig. 1 depicts the light-off curves of Au/TiO2, Au/Al2O3,
◦
and Au/Al2O3/TiO2. The 200 C-pretreated catalysts exhib-
◦
was heated at 80 C, 1.0 g support added, the beaker covered
ited high activities, achieving T50 (temperature required for

1
88
Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
Fig. 1. CO light-off curves of Au/TiO2 [24], Au/Al2O3, and Au/Al2O3/TiO2
◦
◦
pretreated at 200 C (A) or 500 C (B).
◦
5
(
0% conversion) values of −27, 20, and −25 C, respectively
◦
Fig. 1A). After 500 C-aging, the T values of Au/TiO2 and
50
◦
◦
◦
Fig. 2. XRD patterns of TiO2, Al2O3, and Al2O3/TiO2 supports, as well as as-
synthesized Au/TiO2 [24], Au/Al2O3, and Au/Al2O3/TiO2 and those after 200
Au/Al2O3 increased dramatically to 126 C [24] and 237 C,
respectively (Fig. 1B). The 126 C value associated with 500 C-
aged Au/TiO2 is consistent with the 125 C value of 500 C-aged
Au/TiO2 independently made and tested by Yan et al. [22].
Nevertheless, the T50 of 500 C-aged Au/Al2O3/TiO2 was main-
tained at −10 C (Fig. 1B), comparable to the −17 C value
independently obtained by Yan in our group on Au/Al2O3/TiO2
synthesized alternatively [22,28]. These data not only confirm
the anti-aging effect previously observed by Yan et al. [22,28],
but also suggest that the surface-sol–gel method is not necessary
to get active Au/Al2O3/TiO2.
◦
◦
or 500 C-pretreatments and light-off curve measurements.
◦
◦
◦
Fig. 2B, Al2O3 derived from decomposition of Al(NO3)3·9H2O
◦
◦
◦
at 500 C showed broad features ascribed to amorphous Al2O3
[38]. In control experiments, nanocrystalline ␥-Al2O3 was not
◦
formed until calcining Al(NO3)3·9H2O at 800 C (Fig. S1).
After loading gold onto the amorphous Al2O3, what appeared
◦
were not only the expected gold peaks, but also new peaks at 28 ,
◦
◦
49 , and 72 , indicating the formation of boehmite (AlOOH,
also known as Al2O3·H2O) [39] after deposition–precipitation.
The basic aqueous environment (pH ∼ 10) and the high solution
Fig. 2 collects the XRD data of relevant supports and gold
catalysts before and after reaction. In general, gold peaks
◦
◦
◦
◦
◦
◦
appeared at 2θ = 38 , 44 , 65 , 78 , and 82 , corresponding
temperature (80 C) of deposition–precipitation may facilitate
to Au (1 1 1), (2 0 0), (2 2 0), (3 1 1), and (2 2 2), respectively
the formation of beohmite, as detected by XRD (Fig. 2B). The
◦
[
37]. In Fig. 2A, both TiO2 (anatase and rutile) and gold
boehmite phase was maintained after 200 C-treatment, but van-
◦
were identified for Au/TiO2 [24]. The gold peaks of the as-
synthesized Au/TiO2 were already obvious, due to the use of
ethanol [20,22] to wash catalysts during the final stage of prepa-
ration (which is not absolutely necessary, but was consistently
carried out in this work following the initial practice reported
ished after 500 C-aging due to the release of crystal water [1],
and the Al2O3 phase was again amorphous at that point. In
Fig. 2C, only TiO2 and gold phases could be clearly detected
for Au/Al2O3/TiO2.
After the structural and compositional analysis, the effect
of Al2O3 surface modification on gold particle sizes was
+
in Refs. [20,22]), analogous to the reduction of supported Ag
by formaldehyde [35] and that of Pd2 by methanol [36]. In
+
◦
assessed. The Au (2 0 0) peaks (2θ = 44 ) of Au/Al2O3/TiO2

Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
189
Table 1
Surface areas and elemental analysis results of Al2O3/TiO2 samples before and after loading gold
2
Entry
Support
Surface area (m /g)
Al content (wt%)
Before loading Au
Before loading Au
After loading Au
After loading Au
1
2
3
4
5
6
7
8
9
0
1
2
3
4
TiO2
48.0
26.9
32.9
40.1
42.4
42.2
43.5
43.4
47.0
46.8
38.3
64.6
86.4
18.3
47.3
47.7
40.5
40.5
42.4
42.2
43.9
44.0
44.2
–
–
dry-Al2O3–0.05/TiO2
200-Al2O3–0.05/TiO2
300-Al2O3–0.05/TiO2
400-Al2O3–0.05/TiO2
500-Al2O3–0.05/TiO2
600-Al2O3–0.05/TiO2
700-Al2O3–0.05/TiO2
500-Al2O3–0.01/TiO2
500-Al2O3–0.04/TiO2
500-Al2O3–0.2/TiO2
500-Al2O3–0.5/TiO2
500-Al2O3–1.0/TiO2
500-Al2O3–from Al(NO3)3
1.89
1.98
2.11
2.21
2.21
2.21
2.23
0.24
1.94
8.60
17.1
23.9
45.5
1.96
1.94
2.11
2.16
2.18
2.16
2.16
0.21
1.78
8.39
16.2
20.1
37.7
1
1
1
1
1
50.0
102.0
186.7
172.0
112.9
(
Fig. 2C) were much broader than those of Au/TiO2 (Fig. 2A)
Ref. [22], and provide a possible explanation for the anti-aging
effect observed in Fig. 1B.
and Au/Al2O3 (Fig. 2B), suggesting smaller gold particles for
Au/Al2O3/TiO2 [22]. Indeed, the average gold particle sizes
of 500 C-aged Au/TiO2, Au/Al2O3, and Au/Al2O3/TiO2 were
◦
3.1.2. Effects of the calcination temperature of Al O /TiO
2
2
3
estimated by XRD line-broadening method as 15.5, 10.9, and
Yan et al. loaded gold onto surface-sol–gel-derived
Al O /TiO that was dried but not calcined [22,28]. Herein,
◦
6
.9 nm, respectively. The average gold particle of 500 C-aged
2
3
2
Au/TiO2 (15.5 nm), as estimated by X-ray line-broadening
method, seems to be consistent with Yan et al.’s TEM study
which showed gold particles generally larger than 5 nm, and
up to 35 nm [22]. Although the X-ray line-broadening method
can allow for the characterization of a sufficient amount of
catalyst sample, gold particles smaller than 3 nm can hardly
be detected because the peak is too broad, and the estimated
average particle size is biased by the larger particles [40,41].
to study the influence of synthesis conditions, a first series
of supports were prepared with or without calcining the
Al(NO ) /TiO precursor. Their XRD patterns were identical
3
3
2
to that of TiO (Fig. S2).
2
Table 1 lists the measured surface areas and Al contents
2
of various supports. The surface area of TiO was 48.0 m /g.
2
After impregnating of TiO by Al(NO ) salt followed by dry-
2
3 3
◦
ing at 85 C, the surface area of the material decreased to
26.9 m /g, possibly due to the diluting effect of low-surface-area
Al(NO ) (14.5 m /g after drying at 85 C) and the plugging of
◦
2
Nevertheless, our XRD data of 500 C-aged catalysts may
show the general trend, are in harmony with the trend seen in
2
◦
3 3
Table 2
Characterization results and catalysis data of Au/Al2O3/TiO2 catalysts before and after reaction
◦
◦
Entry
Catalyst
Final pH
As synthesized
Au (wt%)
200 C pretreated
500 C pretreated
◦
◦
Au size (nm)
T50 ( C)
Au size (nm)
T50 ( C)
Au size (nm)
1
2
3
4
5
6
7
8
9
0
1
2
3
4
Au/TiO2
9.1
8.4
8.5
8.4
8.9
8.9
9.2
9.0
9.0
8.9
9.5
9.4
9.2
9.1
5.6
4.6
5.8
2.1
5.8
4.1
3.9
3.2
3.4
3.2
2.9
3.7
5.1
5.8
11.6
7.6
7.3
7.4
7.3
6.9
6.8
6.4
8.6
8.3
8.4
9.1
9.3
10.4
−27
−31
−23
−33
−21
−25
−22
−21
−22
−16
−3
11.7
7.5
7.3
7.4
7.5
7.0
7.0
6.7
8.7
8.1
8.5
9.1
9.1
10.4
126
−4
−4
9
15.5, 16.2a
7.5
7.3
7.4
7.4
6.9, 6.9a
6.9
7.5
10.0
8.7
8.6
9.3
Au/dry-Al2O3–0.05/TiO2
Au/200-Al2O3–0.05/TiO2
Au/300-Al2O3–0.05/TiO2
Au/400-Al2O3–0.05/TiO2
Au/500-Al2O3–0.05/TiO2
Au/600-Al2O3–0.05/TiO2
Au/700-Al2O3–0.05/TiO2
Au/500-Al2O3–0.01/TiO2
Au/500-Al2O3–0.04/TiO2
Au/500-Al2O3–0.2/TiO2
Au/500-Al2O3–0.5/TiO2
Au/500-Al2O3–1.0/TiO2
Au/500-Al2O3–from Al(NO3)3
11
−10
14
4
22
−8
−23
−2
11
1
1
1
1
1
79
61
20
9.7
10.9
237
Final pH means the pH value of the aqueous phase after deposition–precipitation of gold, even though the initial pH value before deposition–precipitation was adjusted
to 10.0. Gold contents were measured by ICP method. Average gold particle sizes were estimated using XRD line broadening method. The T50 (temperature-required
for 50% conversion) values were read out from the corresponding light-off curves.
a
Average gold particle sizes of the used catalysts collected after 120 h stability tests.

1
90
Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
the pore structure of TiO2 by the impregnated Al(NO3)3 salt. In
highly dispersed on TiO2 [33] or it might form a composite mate-
rial with crystalline TiO2 when the Al2O3 content was higher
than the dispersion capacity of Al2O3 on TiO2.
a control experiment, Al(NO3)3·9H2O and TiO2 (0.368 g/1 g)
◦
were mechanically mixed and dried at 85 C overnight. The
2
surface area of the mixture was measured to be 42.0 m /g,
consistent with the 42.2 m /g value calculated assuming solely
Table 1 compares the surface areas and Al contents of
2
2
supports. The surface area of Al2O3 was 18.3 m /g, but that
2
the diluting effect, so the low surface area of Al(NO3)3/TiO2
of Au/Al2O3 was 112.9 m /g. The bigger surface area of
2
(
26.9 m /g) should be due to both the diluting and plugging
the latter is consistent with the formation of boehmite dur-
ing deposition–precipitation preparation (Fig. 2B). Boehmite is
known to exhibit bigger surface areas than amorphous Al2O3
[39]. Likewise, when the Al2O3 contents were larger than
0.2 g Al2O3 per g TiO2, the surface areas of as-synthesized
Au/Al2O3/TiO2 were obviously bigger than those of corre-
sponding Al2O3/TiO2 supports (Table 1), possibly because the
surface-area effect of boehmite became more apparent when the
Al content was higher.
effects. Both the surface areas and Al contents of supports
increased with the calcination temperature and leveled off at
◦
4
00–700 C (Table 1), indicating the more obvious decom-
◦
position of Al(NO3)3 below 400 C. The decomposition of
Al(NO3)·9H2O at elevated temperature to form Al2O3 is a well
known process (Table S1 and Fig. S1) [38].
Table 1 also lists the surface areas and Al contents
of as-synthesized catalysts. The surface area of uncalcined
2
Al(NO3)3/TiO2 was 26.9 m /g, but after loading gold, the
The catalytic data with respect to Al2O3 content was sum-
2
◦
surface area increased to 47.7 m /g (Table 1, entry 2). One pos-
marized in Table 2. The T50 values of these 200 C-pretreated
◦
sible explanation is that the impregnated Al(NO3)3 on TiO2
got leached during deposition–precipitation, resulting in gold
on bare TiO2 (note that the surface area of bare TiO2 was
Au/Al2O3/TiO2 catalysts (−25 to 79 C) were comparable to
◦
◦
that of 200 C-pretreated Au/TiO2 (−27 C) only when the
Al2O3 contents were low (0.01–0.05 g Al2O3 per g TiO2), and
2
2
◦
4
8.0 m /g, close to the 47.7 m /g value mentioned above). How-
increased to 70 or 61 C when the Al2O3 content was 0.5 or 1.0 g
Al2O3 per g TiO2 (Table 2). The T50 values of these 500 C-aged
◦
ever, the elemental analysis data indicated that the Al content
was retained (Table 1, entry 2). Possibly, Al(NO3)3/TiO2 was
transformed to Al(OH)3/TiO2 in the basic media (pH ∼ 10)
of deposition–precipitation. Note that Al3 is transformed to
Al(OH)3 at pH 9–10 [42].
◦
catalysts were in the range of −23 to 22 C, and the effect of
Al2O3 modification on stabilizing gold particles was apparent
evenwhen the Al2O3 contentwas 0.01 gAl2O3 perg TiO2. Inter-
+
Table 2 summarizes the catalytic performance. The T50 val-
◦
ues of these 200 C-pretreated Au/Al2O3/TiO2 were in the range
◦
◦
of −33 to −21 C, and those of 500 C-aged catalysts −10 to
◦
1
4 C, showing neither big differences nor clear-cut trends. As
represented by the left trace in Fig. 1B, there were often dips
or plateaus in the conversion curves, which certainly biased
the precise determination of T50 values. These dips or plateaus
were virtually not seen with Au/PO43 /TiO2 [24], but were seen
with some of the catalysts reported here (Figs. 1, 3, 5 and 6) or
elsewhere [25,43,44], explainable by the influence of carbonate
−
[
45–48]. In any case, under our experimental conditions, the cat-
alytic performance of Au/Al2O3/TiO2 is not significantly depen-
dant on the calcination temperature of the modified support.
3
.1.3. Effects of Al2O3 content of Al2O3/TiO2
A second series of Al2O3/TiO2 supports were prepared
by tuning the Al2O3 content. When the Al2O3 content was
increased from 0.01 to 1.0 g Al2O3 per g TiO2, their XRD pat-
terns were identical to that of TiO2, except that the TiO2 peaks
were attenuated when the Al2O3 content was high (Fig. S2). The
absence of Al2O3 peaks could be either due to the possibility that
Al2O3 was not loaded, or due to the possibility that Al2O3 was
amorphous. The elemental analysis results excluded the possi-
bility that Al2O3 was not loaded (Table 1). This is expected,
because according to the principle of impregnation, all of the
Al in the Al(NO3)3·9H2O precursor should be confined in the
Al2O3/TiO2 product. As for the second possibility, in fact, the
◦
Al2O3 derived from calcining Al(NO3)3·9H2O at 500 C was
amorphous according to a XRD experiment (Fig. S1), and Yan et
al. also did not observe any crystalline Al2O3 on Au/Al2O3/TiO2
Fig. 3. CO light-off curves of unmodified [24] and CaO, NiO, ZnO, Ga2O3,
◦
Y2O3, ZrO2, or SiO2–WO3 modified Au/TiO2 catalysts pretreated at 200 C
◦
[
22] or Au/Al2O3/Al2O3/TiO2 [23]. Thus, Al2O3 could be
(A) or 500 C (B).

Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
191
Table 3
Characterization results and catalysis data of Au/MxOy/TiO2 catalysts before and after reaction
◦
◦
Entry
Catalyst
Final pH Support
As synthesized
200 C pretreated
500 C pretreated
2
2
◦
◦
BET (m /g) BET (m /g) Au (wt%) Au size (nm)
T50 ( C)
Au size (nm)
T50 ( C)
Au size (nm)
1
2
3
4
5
6
7
8
9
0
1
2
3
4
Au/TiO2
9.1
9.1
9.5
9.0
10.2
10.0
8.4
9.1
9.6
7.2
8.9
7.9
8.5
9.5
48.0
34.8
44.4
41.7
39.4
38.8
47.1
46.4
48.2
53.0
50.3
48.2
50.5
47.2
47.3
37.3
41.3
41.6
37.1
41.3
44.6
43.6
42.6
45.4
51.8
45.5
47.1
42.8
5.6
11.4
3.7
4.3
2.6
3.0
3.7
5.7
2.7
8.3
2.0
3.9
2.8
5.2
11.6
6.0
5.6
6.9
4.9
5.7
9.1
5.2
5.6
10.5
9.1
9.2
8.4
8.1
−27
<−100
−26
<−100
4
<−100
−32
<−100
−30
25
31
43
34
−31
11.7
6.0
6.0
7.3
5.2
5.6
9.0
5.4
5.6
10.9
9.1
9.2
8.9
8.1
126
−63
25
29
31
15.5 (16.2a)
Au/CaO/TiO2
Au/Fe2O3/TiO2
Au/NiO/TiO2
6.2
6.2
8.1
6.4
6.5 (6.9 )
9.3 (9.3 )
Au/CuO/TiO2
Au/ZnO/TiO2
Au/Ga2O3/TiO2
Au/Y2O3/TiO2
Au/ZrO2/TiO2
Au/MoO3/TiO2
Au/WO3/TiO2
Au/SiO2–WO3/TiO2
Au/P2O5–WO3/TiO2
Au/Bi2O3/TiO2
a
−31
a
9
a
−48
14
6.4 (6.2 )
6.3 (6.2 )
a
1
1
1
1
1
202
124
154
146
156
16.9
9.2
11.0
10.1
9.5
a
Average gold particle sizes of the used catalysts collected after 120 h stability tests.
◦
estingly, some 500 C-aged Au/Al2O3/TiO2 catalysts were even
Among these additives, Ga2O3 and ZrO2 virtually did not
change the activity, CaO, NiO, ZnO, and Y2O3 significantly
promoted Au/TiO2, and SiO2–WO3 reduced the activity.
In particular, Au/CaO/TiO2, Au/NiO/TiO2, Au/ZnO/TiO2,
and Au/Y2O3/TiO2 exhibited anomalously high activity
◦
more active than the corresponding 200 C-aged ones when the
Al2O3 contents were in the range of 0.2–1.0 g Al2O3/TiO2. This
thermal activation is quite surprising but resembles our previous
finding that 500 C-aged Au/LaPO4 and Au/TiO2/AlOOH could
be more active than the as-synthesized ones [27,28].
◦
◦
about −100 C, even though these catalysts were sufficiently
◦
cooled at −100 C. In the literature, virtually no supported
◦
3
.2. Extension to other metal oxide additives
gold catalysts could be so active at −100 C [3–9]. Even
for the highly active Au/LaPO4, its activity was less than
◦
The key idea of modifying TiO2 with Al2O3 for supporting
20% about −100 C [27,28]. Nevertheless, the similar anoma-
goldwasextendedtoanumberofmaingroup, transition, andrare
earth metal oxide additives. The XRD patterns of these modified
supports were identical to that of TiO2 (Figs. S3 and S4), and
their surface areas, in most cases, were close to that of TiO2
lous activity was observed in our previous investigation
of CO oxidation on H2-reduced Au/TiO2/TiO2/SiO2 and
Au/Al2O3/TiO2/SiO2, but no high activity was found after
calcination of these catalysts in O2–He [23]. It should be
◦
(Tables 3 and 4).
mentioned that the high activity at −100 C may not be
3
+
due to the stoichiometric reaction between Au and CO:
◦
after measuring the light-off curve of 200 C-pretreated
3
.2.1. Effects of main-group and transition metal oxide
◦
Au/Y2O3/TiO2, we cooled down the catalyst to −100 C,
additives
◦
and found that the high activity was still there (data not
shown).
Fig. 3A shows the light-off curves of 200 C-pretreated
Au/MxOy/TiO2 (M = Ca, Ni, Zn, Ga, Y, Zr, and Si–W).
Table 4
Characterization results and catalysis data of Au/RExOy/TiO2 catalysts before and after reaction
◦
◦
Entry
Catalyst
Final pH
Support
As synthesized
200 C pretreated
500 C pretreated
2
2
◦
◦
BET (m /g)
BET (m /g)
Au (wt%)
Au size (nm)
T50 ( C)
Au size (nm)
T50 ( C)
Au size (nm)
a
1
2
3
4
5
6
7
8
9
Au/TiO2
9.1
10.4
9.4
9.8
9.6
9.7
9.8
9.7
9.7
9.1
9.4
9.6
48.0
43.7
45.5
45.4
44.0
45.8
44.7
45.0
44.6
45.0
45.8
46.3
47.3
42.5
44.3
43.4
42.3
44.2
47.6
43.0
43.5
45.8
44.6
42.1
5.6
3.5
3.6
5.2
4.8
4.0
4.4
3.9
4.2
2.6
4.8
4.3
11.6
5.6
9.0
6.7
7.6
6.7
6.3
6.7
6.6
7.3
6.1
7.0
−27
−75
11.7
5.6
9.1
6.7
7.4
6.9
6.1
6.9
6.8
7.2
6.1
7.0
126
−39
49
15.5 (16.2 )
a
Au/La2O3/TiO2
Au/CeO2/TiO2
Au/Pr2O3/TiO2
Au/Nd2O3/TiO2
Au/Sm2O3/TiO2
Au/Eu2O3/TiO2
Au/Gd2O3/TiO2
Au/Dy2O3/TiO2
Au/Ho2O3/TiO2
Au/Er2O3/TiO2
Au/Yb2O3/TiO2
6.1 (6.1 )
−19
9.2
6.6
7.9
7.3
6.3
7.1
7.0
8.0
7.0
7.3
<−100
<−100
<−100
<−100
−78
16
−42
−31
−46
−12
−17
−25
−33
−29
−88
1
1
1
0
1
2
<−100
<−100
<−100
a
Average gold particle sizes of the used catalysts collected after 120 h stability tests.

1
92
Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
◦
Fig. 3B shows the light-off curves of 500 C-pretreated
Au/SiO2–WO3/TiO2 were sharp, consistent with its low activ-
ity. Nevertheless, it should be kept in mind that gold particle
size is only one of the factors that may influence the catalytic
performance [3–9]. In fact, from Table 2, it is observed that cata-
lysts with similar average gold particle sizes could have different
activity. This should be attributed to the influence of factors other
than gold sizes (e.g., the nature of the support is different when
using this additive compared to another [3,43,49,50]) as well as
the limitation of XRD methods [40,41].
Au/MxOy/TiO2 (M = Ca, Ni, Zn, Ga, Y, Zr, and Si–W). The
T50 values of 500 C-aged Au/TiO2 and Au/SiO2–WO3/TiO2
increased dramatically to 126 and 154 C, respectively,
whereas those of Au/CaO/TiO2, Au/NiO/TiO2, Au/ZnO/TiO2,
◦
◦
Au/Ga2O3/TiO2, Au/Y2O3/TiO2, and Au/ZrO2/TiO2 were −63,
◦
2
5
(
9, −31, 9, −48, and 14 C, respectively. In particular,
◦
00 C-aged Au/CaO/TiO2, Au/Y2O3/TiO2, and Au/ZnO/TiO2
◦
Fig. 3B) exhibited even higher activities than 500 C-aged
Au/Al2O3/TiO2 (Fig. 1B) at low temperature. These T50 values
◦
were even lower than the −17 C value of Au/Al2O3/TiO2 syn-
3
.2.2. Effects of rare-earth metal oxide additives
Although some neat rare earth metal oxides have been used to
thesized using the surface-sol–gel method [22,28]. There were
again dips in their light-off curves, implying the influence of
carbonate [45–48].
support gold [51–53], to the best of our knowledge, the promo-
tion of Au/TiO2 by rare earth metal oxides for CO oxidation was
virtually not reported. Figs. 5 and 6 show that, many rare earth
metal oxides could promote Au/TiO2 for CO oxidation. They,
The drastic difference in T5 value can certainly show the
0
activity trend, but because the gold loadings of catalysts are dif-
ferent (Table 2), specific rates were calculated (Table S2). To
compare the activity using specific rates of different catalysts, it
is important to make sure that the specific reaction temperature
chosen is not so high that the conversions all reach 100%. Thus,
◦
except for CeO2, again exhibited anomalous activity at −100 C
◦
when they were pretreated at 200 C, and they were very active
◦
◦
below 0 C even after aging at 500 C. The specific rates were
calculated, and the promotional effect of the beneficial metal
oxide additives mentioned above is obvious (Table S3). The
average gold particle sizes of these catalysts, estimated by XRD
experiments, were all smaller than those of Au/TiO2 (Table 4).
◦
◦
the CO conversion at −80 C for 200 C-pretreated gold cata-
◦
◦
lysts and that at −25 C for 500 C-aged catalysts were used for
calculating specific rates. The majority of the catalysts reported
here have activity at these temperatures under the respective
pretreatment conditions. As shown in Table S2, the promotional
effects of the beneficial metal oxide additives reported above are
still apparent.
Fig. 4 summarizes the XRD patterns of Au/MxOy/TiO2
◦
(
M = Ca, Ni, Zn, Ga, Y, Zr, and Si–W) after 500 C-aging
◦
and light-off curve measurements. The 500 C-aged Au/TiO2
exhibited relatively sharp gold peaks, corresponding to big
gold particles and poor activity. However, the gold peaks
◦
of 500 C-aged Au/CaO/TiO2, Au/NiO/TiO2, Au/ZnO/TiO2,
Au/Ga2O3/TiO2, Au/Y2O3/TiO2, and Au/ZrO2/TiO2 were
much broader, indicating the stabilizing of gold particles by
these metal oxide additives. This is similar to the effect of Al2O3
additive [22] (Fig. 2C), and consistent with the high activities
◦
in Fig. 3B. On the other hand, the gold peaks of 500 C-aged
Fig. 4. XRD patterns of unmodified [24] and CaO, NiO, ZnO, Ga2O3, Y2O3,
ZrO2, or SiO2–WO3 modified Au/TiO2 catalysts after 500 C-treatments and
light-off curve measurements.
Fig. 5. CO light-off curves of unmodified [24] and La2O3, CeO2, Pr2O3, Nd2O3,
Sm2O3, orEu2O3-modifiedAu/TiO2 catalystspretreatedat200 C(A)or500 C
(B).
◦
◦
◦

Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
193
Fig. 7. Effects of the load of Au/ZnO/TiO2 catalyst in the reactor on the CO
◦
conversion with time on stream. The catalyst was aged at 500 C, and tested for
3
120 h. The flow rate of reactant stream was 37 cm /min.
to about 60%. This is expected, assuming that the number of
active sites is doubled when doubling the amount of catalyst
loaded in the reactor [5]. Thus, conceptually, the steady-state
conversion should increase with the catalyst load until reaching
1
1
00%. Indeed, we found that the conversion was always close to
00% if we put 200 mg catalyst in the reactor (Fig. 7). However,
the fact that no deactivation was observed (always 100% con-
version) should not be used as grounds to claim that a catalyst
has high stability on stream. This pitfall, sometimes overlooked,
was critically discussed in recent reviews [5,61].
Fig. 6. CO light-off curves of unmodified [24] and Gd2O3, Dy2O3, Ho2O3,
◦
◦
Er2O3, or Yb2O3–modified Au/TiO2 catalysts pretreated at 200 C (A) or 500 C
B).
With these considerations, we properly tested the stability
(
◦
of typical 500 C-calcined Au/MxOy/TiO2 catalysts, loading
less catalyst if the regular load (50 mg) would lead to 100%
conversion at room temperature [5], and testing the stability
for sufficient duration of time if several hours are deemed
insufficient [34,60,62]. As shown in Fig. 8, the CO conver-
However, it should be mentioned again that gold particle size is
not the only factor that determines activity [3–9].
◦
◦
3
.3. Stability tests as a function of reaction time and
sion of 500 C-aged Au/TiO2 at 170 C dropped from ca. 93
to 89% during 5 days [24], consistent with the trend seen in
the literature [50,62,63]. However, the temporal stability of
Au/MxOy/TiO2 catalysts was poorer. They often underwent ini-
tial faster deactivation, and the conversions leveled off later,
implying the formation of less reactive intermediates on catalyst
surfaces [45–48]. In particular, the activity of Au/Al2O3/TiO2
was decreasing on stream, consistent with others’ finding that
Au/Al2O3 suffered from deactivation on stream because of the
influence of carbonate [31,32]. For comparison purposes, the
catalytic stability of our Au/Al2O3/TiO2 catalysts previously
synthesized via the surface-sol–gel method [22,28] was inves-
tigated. These catalysts also suffered from some deactivation
on stream (Fig. S5) [22,28]. The deactivation of Au/ZnO/TiO2
seems to contradict the observation by Zhang et al., who claimed
that the conversion of CO on a large quantity of Au/ZnO could
be maintained at 100% for months followed by observable deac-
tivation [57]. Nevertheless, our data are in line with their another
observation that some less active Au/ZnO with initial CO con-
versions lower than 100% deactivated quickly [58], and with the
finding by Scurrell et al. that Au/ZnO/TiO2 deactivated more
regeneration
Stability means the ability of the catalyst to maintain a
constant level of performance during the course of reaction
[
2]. Some research did not report stability [20,22,23,27,54], or
mentioned vaguely that no deactivation was detected [55,56].
Sometimes, the catalyst load or reaction temperature was higher
than just needed for complete conversion, resulting in 100% con-
version for a long time [57–59]. However, others cautioned that
in order to clarify the stability trends, it is crucial to make sure
that the conversion would not reach 100% [5,60]. Indeed, our
control experiments indicated that a misleading picture of stabil-
ity could be obtained if the catalyst load or reaction temperature
exceeded more than needed (Fig. 7), although such practice may
be useful for practical purposes to compensate the loss in activity
[
61].
Specifically, in Fig. 7, when 50 mg Au/ZnO/TiO2 was loaded,
the steady-state conversion was about 30%. When 100 mg cat-
alyst was loaded and the flow rate of the reactant stream was
the same (37 cm /min), the steady-state conversion increased
3

1
94
Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
Fig. 9. The deactivation of Au/ZnO/TiO2 in the absence or presence of H2. The
◦
catalysts were aged at 500 C before testing. In one run, 50 mg catalyst was
loaded, and no H2 was added. In other two runs, 35 mg catalyst was loaded, and
the H2 flow rate was 2 or 5 ml/min, with the 1% CO rate 37 ml/min.
and found that many of these catalysts could be regenerated that
◦
way, by treating the used catalysts in O2–He at 400 C (data
not shown). The regeneration of gold catalysts is a topic worth
further research [9,66].
Concerning the suppressing of deactivation, Yan et al. in our
group have recently found that the deactivation of Au/AlPO4
on stream could be eliminated in the presence of H2, arguably
via suppressing the build-up of carbonate [28]. In fact, this
interesting phenomenon has already been noticed by others
[67,68]. In the current study, we found that the deactivation
of Au/ZnO/TiO2 was suppressed to some extent (Fig. 9), if
Fig. 8. CO conversion as a function of time on stream. The catalysts were aged
◦
at 500 C, and tested for 120 h. Unless otherwise indicated, 50 mg catalyst was
3
loaded, and the flow rate of reactant stream was 37 cm /min. The stability data
◦
of 500 C-aged Au/TiO2 [24] was plotted for comparison.
2 ml/min pure H was co-fed with the regular reactant stream
2
(1% CO, 37 ml/min). During the tested period, around 30% of
the H2 gas was consumed. In another run, 5 ml/min H2 was
added, and the deactivation was further suppressed (Fig. 9). Dur-
ing that period, around 20% of the H2 gas was consumed. It
was proposed in the literature that the formation of water from
H2 oxidation may clean up the surface carbonate species, thus
suppressing the deactivation of gold catalysts [9,67,68]. Further
spectroscopic experiments could be very worthwhile.
quickly than Au/TiO2 [64]. Others have found deactivation in
Au/ZrO2 [45,48], whereasthedeactivationofAu/TiO2 wasoften
less obvious [50,63,65]. These exist some references mention-
ing that the deactivation of some gold catalysts are due to the
accumulation of carbonate, although the growth of gold parti-
cles under high-reaction-temperature conditions is also possible
[
3,45–48]. Nevertheless, we noted that under our low-reaction-
temperature conditions, the growth of gold particles was not
obvious (Tables 2, 3 and 4).
4. Discussion
Given the fact of deactivation on stream, the next step is to
regenerate the catalyst after deactivation and to develop methods
for suppressing the temporal deactivation. Concerning the first
idea, we carried out a preliminary regeneration experiment, in
Gold catalysis is a hot topic, with tens of papers published
weekly [4]. However, most of the gold catalysts were synthe-
sized by loading gold onto neat, but not surface-modified, metal
oxides [3–9]. This situation is in contrast to the research on
mesoporous materials [69], solid strong acids [70], and semi-
conductor photocatalysts [71], in which the use of metal oxide
promoters has become very popular.
◦
which 100 mg 500 C-aged Au/ZnO/TiO2 catalyst, subject to the
1
20 h stability test in Fig. 7 and stored in the U-type tube after-
wards at ambient temperature for weeks, was treated in O2–He
◦
at 400 C for 1 h, and cooled down to ambient temperature for
reaction. The initial activity of the regenerated catalyst at ambi-
ent temperature returned to 100%, indicating that the activity
of the deactivated catalyst could at least be partially restored.
In fact, we recently studied the regeneration of Au/Al2O3 and
Au/TiO2/Al2O3 catalysts initially reported in our patent [28],
as well as the regeneration of other catalysts not reported yet,
To address such imbalance, our group recently communi-
cated that Al2O3 additive on TiO2 could stabilize gold particles
and suppress the thermal aging [22], but it was not clear from
thatcommunicationwhethertheuseofanadditivesuchasAl2O3
[22] to stabilize gold particles is a serendipitous choice, or repre-
sents a general methodology. The key contribution of our current

Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
195
work is not only to confirm the effects of Al2O3, but also to
demonstrateforthefirsttimethatanarrayofotheradditives, such
as CaO, NiO, ZnO, Ga2O3, Y2O3, ZrO2, La2O3, Pr2O3, Nd2O3,
Sm2O3, Eu2O3, Gd2O3, Dy2O3, Ho2O3, Er2O3, and Yb2O3 on
TiO2 could exhibit similar or even better effects. The key advan-
tages of the synthesis method used here [33,34], compared to the
surface-sol–gel method [22,28], are that it adopts simple steps,
avoids non-aqueous environments, uses water instead of organic
solvents, and avoids the use of expensive metal alkoxides.
To place our work in proper perspective, the use of modified
metal oxides on gold catalysis is summarized. Since Au/SiO2
and Au/Al2O3 are inherently less active than Au/TiO2 for CO
oxidation, some attempts have been made to modify the under-
lying SiO2 and Al2O3 supports in order to make them more
active. For instance, SiO2 was modified by Al2O3 [23,72],
TiO2 [29,30,54,73–75], Fe2O3 [62], CoOx [41,72,76,77], LaOx
metallic nickel particles for CO methanation [33]. A sim-
ilar effect was recently reported with Cu/FeOx/SiO2 [92],
Cu–B/Cr2O3/Al2O3 [93], Ni/ZrO2/Al2O3 [94], Ni/La2O3/
Al2O3 [95], Ni/CeO2/Al2O3 [95], and Pt/CeO2/ZrO2 [96] for
various reactions. Recall that in the traditional synthesis of
ammonia, Al2O3, CaO, and MgO promoters were used to min-
imize the sintering of metallic iron catalysts [1,2].
Thus, from the above literature survey, it could be argued
that the effect of Al2O3 modification on stabilizing gold parti-
cles [22,28] is in fact not alone, and our current work involving
25 different surface additives apparently detected many modi-
fied gold catalysts not previously reported. Among the catalysts
we surveyed, Au/MxOy/TiO2 (M = Al, Ca, Ni, Zn, Ga, Y, Zr,
La, Pr, Nd, Sm, Eu, Gd, Dy, Ho, Er, or Yb) exhibited small
◦
gold particle sizes even after 500 C-aging, consistent with their
◦
high activities after 500 C-aging. It is generally recognized that
[
76], and CeOx [76], and then loaded with gold for CO oxi-
small gold particles are necessary for achieving high activity in
CO oxidation, although gold particle size is not the only fac-
tor that determines the activity [3–9]. Our data are different
from those of Scurrell et al. [64], who found that Au/ZnO/TiO2
was less active than Au/TiO2 in CO oxidation, probably due
to different experimental conditions. Nevertheless, our data on
Au/ZnO/TiO2 and Au/ZrO2/TiO2 are in line with others’ find-
ing that ZnO/Al2O3 and ZrOx/Al2O3 stabilized gold particles
[81], although the supports and studied reactions are different.
Our data are also in line with the finding that ZrO2/Al2O3 [94]
and La2O3/Al2O3 [33,95] stabilized nickel particles, although
the underlying substrates, active metal, and catalyzed reactions
are all different.
dation [23,29,30,41,54,62,72,76,77] and propylene epoxidation
[
[
73–75]. Al2O3 wasmodifiedbyMgO[78], MnOx [79,80], FeOx
79], CoOx [79], NiOx [79], ZnOx [79,81], ZrOx [81], BaO [82],
and CeO2 [83,84], and then loaded with gold for CO oxida-
tion [78–80,82,84], CH4 oxidation [81], and organic combustion
[
83]. On the other hand, because Au/TiO2 is already quite active
in CO oxidation, the further modification of the TiO2 support
was often overlooked [4]. For example, TiO2 was modified by
Al2O3 [22], MnO2 [85], Fe2O3 [86,87], and ZnO [64], and then
loaded with gold for CO oxidation [22,64,85,86] and methanol
oxidation [87].
The complexity in assessing the previous literature lies in the
fact that many of the reports mentioned above focused on the
effect of additives on catalytic performance, but the gold cata-
lysts were not often systematically treated at high temperatures
to see whether they remained active or not. In addition, the fact
that an additive was tried does not imply that it showed positive
effect. Nevertheless, based on the limited and sometime contra-
dictory results in the literature, it could be generalized that: (1)
TiO2/SiO2 [29,30,54], MgO/Al2O3 [78], BaO/Al2O3 [82], and
Al2O3/TiO2 [22] could both stabilize gold particles and promote
CO oxidation; (2) ZnO/Al2O3 [81] and ZrOx/Al2O3 [81] could
stabilize gold particles, but could not promote CH4 oxidation;
and (3) MnOx/Al2O3 may or may not stabilize gold particles,
but the MnOx component could promote CO oxidation and CH4
oxidationmainlyasaco-catalystorelectronicpromoter[78–81].
In fact, the stabilization of gold particles on supports is not
limited to certain surface-modified oxides, but is also seen on
certain mixed oxides or solid solutions. For instance, Goodman
and coworkers found the stabilizing effect of the defect sites
of TiO2–SiO2 by comparing the evolution of gold particles on
TiO2–SiO2 and SiO2 surfaces under vacuum [88]. Mou et al.
The question arises as to how certain additives or mixed
oxides could stabilize metal particles. In principle, small metal
particles possess extra surface free energy and higher chemical
potential than those of the bulk metal [5,97]. Thus, metal atoms
can move around even below the melting point of bulk metal,
causing sintering [5,97]. Certain modified or mixed metal oxides
may have surface defects [22,88–90] and so-called “spacers”
[95] to hinder the movement of metal particles. A recent first-
principle calculation proposed another model for Au/IrO2/TiO2,
in which IrO2 islands sit on top of TiO2 surface, and Au parti-
cles sit on top of IrO2 islands [98], so the migration of gold to
an adjacent gold particle on another IrO2 island is hindered by
the large energy barrier across the TiO2 surface [99]. Note that
the stabilizing effect appears to be dependant on the nature of
the combination of additive and support. For instance, although
Al2O3 additive on TiO2 [22] and Al3 in Al2O3–SiO2 mixed
oxide [89,90] could stabilize gold particles, no significant pro-
motional effect was seen on Al2O3 coatings dispersed on SiO2
[23,26,72] or Al2O3–CeO2 mixed oxide [100]. In addition,
not all the additives are beneficial, as seen here with MoO3,
+
[
89] and Wan et al. [90] reported that Al3+ in Al-SBA-15 [89] or
WO3, SiO2–WO3, P2O –WO3, and Bi2O3 additives. Moreau
5
zeolite Y [90] could stabilize gold particles. Zheng and cowork-
ers found that Fe2O3–ZrO2 could stabilize gold particles, albeit
Fe2O3–MnOx, Fe2O3–CeO2, Fe2O3–Co2O3, and Fe2O3–MoO3
accelerated the sintering of gold [91].
A more thorough literature survey indicated that the sta-
bilizing effect could go beyond the supported gold system.
A 1990 review introduced that La2O3/Al2O3 could stabilize
and Bond also mentioned briefly that the addition of MoOx to
Au/TiO2 significantly decreased the activity [62]. Further stud-
ies of the “support effects” [3,43,49,50] using surface science
[88] and first-principle [99] methods should shed new light in
these controversies.
In closing, the finding that a certain surface modifier such
as Al2O3 on TiO2 [22,28] could stabilize gold catalysts against

1
96
Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
sinteringactuallypointstoageneralmethodologyofusingtextu-
ral promoters in making metal catalysts [1,2]. Additional effects
may include the change in electronic structure of the active metal
and the supply of reactive oxygen caused by certain additives.
Further research can be carried out based on the new catalyst sys-
tem developed here. In particular, the nature of the active sites of
gold catalysts has been highly debated [3–9], but virtually none
of the catalysts used in these fundamental studies could achieve
treatment, and the catalyst stability could be improved in
the presence of H2.
Acknowledgements
This work was supported by the Office of Basic Energy Sci-
ences, U.S. Department of Energy. The Oak Ridge National
Laboratory is managed by UT-Battelle, LLC for the U.S. DOE
under Contract DE-AC05-00OR22725. This research was sup-
ported in part by the appointment for Z. Ma to the ORNL
Research Associates Program, administered jointly by ORNL
and the Oak Ridge Associated Universities.
◦
superior activity at −100 C. Thus, our newly developed highly
active gold catalysts could be used as model catalysts in subse-
quent model studies on the nature of the active sites. In addition,
considering the high activity of some promoted catalysts in CO
oxidation, the selective CO oxidation in the presence of H2 for
fuel cell purposes could be possible and worth further research.
From a historical perspective of the popularization of metal
Appendix A. Supplementary data
oxide-promoted mesoporous SiO2 [69], SO42 /ZrO2 [70] and
TiO2 photocatalysts [71], it is predicted that the use of modified
supports for gold catalysis may represent a promising direction
for further research and development. In addition, considering
the emerging applications of supported copper [101,102] and
silver [103,104] nanoparticles, it is expected that this doping
methodology may be extended to these systems to address the
common sintering problem.
−
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.molcata.2007.04.007.
References
[1] C.N. Satterfield, Heterogeneous Catalysis in Industrial Practice, 2nd ed.,
Krieger Publishing, Malabar, 1996.
2] Z. Ma, F. Zaera, in:R.B. King(Ed.), EncyclopediaofInorganicChemistry,
[
2
nd ed., John Wiley & Sons, Chichester, 2005, p. 1768.
5
. Conclusions
[
[
3] M. Haruta, Cattech 6 (2002) 102.
4] A.S.K. Hashmi, G.J. Hutchings, Angew. Chem. Int. Ed. 45 (2006) 7896.
Au/TiO2 is perhaps the most studied catalyst in gold catalysis
[5] G.C. Bond, C. Louis, D.T. Thompson, CatalysisbyGold, ImperialCollege
Press, London, 2006.
research, but few attempts have been made to use surface-
modified TiO2 as support [3–9]. We have recently shown that
Al2O3/TiO2 synthesized by a surface sol–gel method could sta-
bilize gold particles and improve the anti-aging ability [22,28].
Based on this key finding, in the present work we first attempted
to synthesize Au/Al2O3/TiO2 alternatively, and then developed
a number of metal oxide-modified Au/TiO2 catalysts. It was
found that:
[6] G.C. Bond, D.T. Thompson, Gold Bull. 33 (2000) 41.
[
7] M. Haruta, M. Dat e´ , Appl. Catal. A 222 (2001) 427.
[8] T.V. Choudhary, D.W. Goodman, Top. Catal. 21 (2002) 25.
[9] H.H. Kung, M.C. Kung, C.K. Costello, J. Catal. 216 (2003) 425.
[10] S. Biella, G.L. Castiglioni, C. Fumagalli, L. Prati, M. Rossi, Catal. Today
2 (2002) 43.
7
[
[
11] P. Claus, Appl. Catal. A 291 (2005) 222.
12] M.D. Hughes, Y.-J. Xu, P. Jenkins, P. McMorn, P. Landon, D.I. Enache,
A.F. Carley, G.A. Attard, G.J. Hutchings, F. King, E.H. Stitt, P. Johnston,
K. Griffin, C.J. Kiely, Nature 437 (2005) 1132.
(
(
(
(
1) The surface-sol–gel method [22,28] is not necessary
to prepare highly active surface-modified gold cata-
lysts. Au/Al2O3/TiO2 prepared by the decomposition of
Al(NO3)3 on TiO2 followed by loading gold could show
comparable performance in CO oxidation.
2) Monolayer Al2O3 [22] is preferable to realize high activ-
ity in CO oxidation, but not absolutely necessary. Gold
deposited on Al2O3/TiO2 supports prepared with a range
of calcination temperatures and Al2O3 contents could show
significant activity.
[13] A. Corma, P. Serna, Science 313 (2006) 332.
[
[
14] D. Andreeva, Gold Bull. 35 (2002) 82.
15] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopoulos, Science 301 (2003)
935.
[
[
16] N. Sakulchaicharoen, D.E. Resasco, Chem. Phys. Lett. 377 (2003) 377.
17] M. Yamada, M. Kawana, M. Miyake, Appl. Catal. A 302 (2006) 201.
[18] G. Pattrick, E. van der Lingen, C.W. Corti, R.J. Holliday, D.T. Thompson,
Top. Catal. 30/31 (2004) 273.
[
19] W.F. Yan, S.M. Mahurin, S.H. Overbury, S. Dai, Top. Catal. 39 (2006)
99.
1
[
20] W.F. Yan, B. Chen, S.M. Mahurin, S. Dai, S.H. Overbury, Chem. Com-
mun. (2004) 1918.
3) Al2O3 additive is not unique. Au/TiO2 promoted by CaO,
NiO, ZnO, Ga2O3, Y2O3, ZrO2, La2O3, Pr2O3, Nd2O3,
Sm2O3, Eu2O3, Gd2O3, Dy2O3, Ho2O3, Er2O3, or Yb2O3
[21] W.F. Yan, B. Chen, S.M. Mahurin, V. Schwartz, D.R. Mullins, A.R.
Lupini, S.J. Pennycook, S. Dai, S.H. Overbury, J. Phys. Chem. B 109
(
2005) 10676.
[
[
[
[
[
22] W.F. Yan, S.M. Mahurin, Z.W. Pan, S.H. Overbury, S. Dai, J. Am. Chem.
Soc. 127 (2005) 10480.
23] W.F. Yan, S.M. Mahurin, B. Chen, S.H. Overbury, S. Dai, J. Phys. Chem.
B 109 (2005) 15489.
24] Z. Ma, S. Brown, S.H. Overbury, S. Dai, Appl. Catal. A, submitted for
publication.
25] H.G. Zhu, C.D. Liang, W.F. Yan, S.H. Overbury, S. Dai, J. Phys. Chem.
B 110 (2006) 10842.
could show significant activity at ambient temperature after
◦
5
00 C-aging.
4) The promoted gold catalysts can undergo deactivation on
stream. Although the sintering-resistant properties of gold
on oxide supports were considerably improved via sur-
face modification, no improvement on the corresponding
temporal stability on stream was observed. However, the
deactivated catalyst could be regenerated through thermal
26] H.G. Zhu, Z. Ma, J.C. Clark, Z.W. Pan, S.H. Overbury, S. Dai, Appl.
Catal. A 326 (2007) 89–99.

Z. Ma et al. / Journal of Molecular Catalysis A: Chemical 273 (2007) 186–197
197
[
[
[
[
27] W.F. Yan, S. Brown, Z.W. Pan, S.M. Mahurin, S.H. Overbury, S. Dai,
Angew. Chem. Int. Ed. 45 (2006) 3614.
28] S. Dai, W.F. Yan, Surface-Stabilized Gold Nanocatalysts, US Patent
Application Publication, 2006/0293175 A1, 2006.
29] Y. Tai, J. Murakami, K. Tajiri, F. Ohashi, M. Dat e´ , S. Tsubota, Appl.
Catal. A 268 (2004) 183.
30] A.M. Venezia, F.L. Liotta, G. Pantaleo, A. Beck, A. Horvath, O. Geszti,
A. Kocsonya, L. Guczi, Appl. Catal. A 310 (2006) 114.
31] S.-J. Lee, A. Gavriilidis, J. Catal. 206 (2002) 305.
32] C.K. Costello, M.C. Kung, H.-S. Oh, Y. Wang, H.H. Kung, Appl. Catal.
A 232 (2002) 159.
[69] D. Trong On, D. Desplantier-Giscard, C. Danumah, S. Kaliaguine, Appl.
Catal. A 253 (2003) 545.
[70] X.M. Song, A. Sayari, Catal. Rev. -Sci. Eng. 38 (1996) 329.
[71] M.A. Fox, M.T. Dulay, Chem. Rev. 93 (1993) 341.
[72] M.T. Bore, M.P. Mokhonoana, T.L. Ward, N.J. Coville, A.K. Datye,
Micropor. Mesopor. Mater. 95 (2006) 118.
[73] T.A. Nijhuis, B.J. Huizinga, M. Makkee, J.A. Moulijn, Ind. Eng. Chem.
Res. 38 (1999) 884.
[74] E.E. Stangland, K.B. Stavens, R.P. Andres, W.N. Delgass, J. Catal. 191
(2000) 332.
[
[
[75] C.X. Qi, T. Akita, M. Okumura, M. Haruta, Appl. Catal. A 218 (2001)
81.
[76] M.A.P. Dekkers, M.J. Lippits, B.E. Nieuwenhuys, Catal. Today 54 (1999)
381.
[
[
[
33] Y.C. Xie, Y.Q. Tang, Adv. Catal. 37 (1990) 1.
34] Z. Ma, W.M. Hua, Y. Tang, Z. Gao, J. Mol. Catal. A 159 (2000) 335.
35] X.G. Zhao, J.L. Shi, B. Hu, L.X. Zhang, Z.L. Hua, Mater. Lett. 58 (2004)
2
152.
[77] X.Y. Xu, J.J. Li, Z.P. Hao, W. Zhao, C. Hu, Mater. Res. Bull. 41 (2006)
406.
[78] R.J.H. Grisel, B.E. Nieuwenhuys, J. Catal. 199 (2001) 48.
[79] R.J.H. Grisel, B.E. Neiuwenhuys, Catal. Today 64 (2001)
69.
[
[
[
[
36] Z.Y. Sun, Z.M. Liu, B.X. Han, S.D. Miao, Z.J. Miao, G.M. An, J. Colloid
Interf. Sci. 304 (2006) 323.
37] H. Araki, A. Fukuoka, Y. Sakamoto, S. Inagaki, N. Sugimoto, Y.
Fukushima, M. Ichikawa, J. Mol. Catal. A 199 (2003) 95.
38] O.V. Mokhnachuk, S.O. Soloviev, A.Y. Kapran, Catal. Today 119 (2007)
[80] D.H. Wang, Z.P. Hao, D.Y. Cheng, X.C. Shi, C. Hu, J. Mol. Catal. A 200
(2003) 229.
145.
39] X.-F. Xu, J.Y. Jeon, M.H. Choi, H.Y. Kim, W.C. Choi, Y.-K. Park, J. Mol.
Catal. A 266 (2006) 131.
40] S.Y. Lai, H.X. Zhang, C.F. Ng, Catal. Lett. 92 (2004) 107.
41] K. Qian, W.X. Huang, Z.Q. Jiang, H.X. Sun, J. Catal. 248 (2007) 137.
42] W.M. Hua, Y.D. Xia, Y.H. Yue, Z. Gao, J. Catal. 196 (2000) 104.
43] M. Comotti, W.C. Li, B. Spliethoff, F. Sch u¨ th, J. Am. Chem. Soc. 128
[81] A.C. Gluhoi, B.E. Nieuwenhuys, Catal. Today 119 (2007) 305.
[82] A.C. Gluhoi, X. Tang, P. Marginean, B.E. Nieuwenhuys, Top. Catal. 39
(2006) 101.
[83] M.A. Centeno, M. Paulis, M. Montes, J.A. Odriozola, Appl. Catal. A 234
(2002) 65.
[
[
[
[
[84] M.A. Centeno, C. Portales, I. Carrizosa, J.A. Odriozola, Catal. Lett. 102
(2005) 289.
(
2006) 917.
[
[
[
44] W.-C. Li, M. Comotti, A.H. Lu, F. Sch u¨ th, Chem. Commun. (2006) 1772.
45] A. Knell, P. Barnickel, A. Baiker, A. Wokaun, J. Catal. 137 (1992) 306.
46] B. Schumacher, V. Plzak, M. Kinne, R.J. Behm, Catal. Lett. 89 (2003)
[85] L.-H. Chang, N. Sasirekha, Y.-W. Chen, Catal. Commun.,
doi:10.1016/j.catcom.2006.08.050, in press.
[86] M. Shou, H. Takekawa, D.-Y. Ju, T. Hagiwara, D.-L. Lu, K. Tanaka, Catal.
Lett. 108 (2006) 119.
109.
[
[
[
[
[
[
47] P. Konova, A. Naydenov, C. Venkov, D. Mehandjiev, D. Andreeva, T.
Tabakova, J. Mol. Catal. A 213 (2004) 235.
48] P. Konova, A. Naydenov, T. Tabakova, D. Mehandjiev, Catal. Commun.
[87] F.-W. Chang, H.-Y. Yu, L.S. Roselin, H.-C. Yang, T.-C. Ou, Appl. Catal.
A 302 (2006) 157.
[88] B.K. Min, W.T. Wallace, D.W. Goodman, J. Phys. Chem. B 108 (2004)
14609.
[89] C.-W. Chiang, A.Q. Wang, B.-Z. Wan, C.-Y. Mou, J. Phys. Chem. B 109
(2005) 18042.
[90] J.-H. Chen, J.-N. Lin, Y.-M. Kang, W.-Y. Yu, C.-N. Kuo, B.-Z. Wan, Appl.
Catal. A 291 (2005) 162.
[91] J.M. Hua, Q. Zheng, Y.H. Zheng, K.M. Wei, X.Y. Lin, Catal. Lett. 102
(2005) 99.
[92] C.-S. Chen, W.-H. Cheng, S.-S. Lin, Appl. Catal. A 257 (2004)
97.
5
(2004) 537.
49] M.M. Schubert, S. Hackenberg, A.C. van Veen, M. Muhler, V. Plzak, R.J.
Behm, J. Catal. 197 (2001) 113.
50] S. Arrii, F. Morfin, A.J. Renouprez, J.L. Rousset, J. Am. Chem. Soc. 126
(
2004) 1199.
51] S. Carrettin, P. Concepci o´ n, A. Corma, J.M.L. Nieto, V.F. Puntes, Angew.
Chem. Int. Ed. 43 (2004) 2538.
52] N.S. Patil, B.S. Uphade, P. Jana, S.K. Bhargava, V.R. Choudhary, Chem.
Lett. 33 (2004) 400.
[
[
53] U.R. Pillai, S. Deevi, Appl. Catal. A 299 (2006) 266.
54] W.F. Yan, B. Chen, S.M. Mahurin, E.W. Hagaman, S. Dai, S.H. Overbury,
J. Phys. Chem. B 108 (2004) 2793.
55] W.F. Yan, V. Petkov, S.M. Mahurin, S.H. Overbury, S. Dai, Catal. Com-
mun. 6 (2005) 404.
56] B.L. Zhu, Q. Guo, X.L. Huang, S.R. Wang, S.M. Zhang, S.H. Wu, W.P.
Huang, J. Mol. Catal. A 249 (2006) 211.
57] G.Y. Wang, W.X. Zhang, H.L. Lian, Q.S. Liu, D.Z. Jiang, T.H. Wu, React.
Kinet. Catal. Lett. 75 (2002) 343.
[93] P.J. Guo, L.F. Chen, S.R. Yan, W.L. Dai, M.H. Qiao, H.L. Xu, K.N. Fan,
J. Mol. Catal. A 256 (2006) 164.
[94] J.G. Seo, M.H. Youn, I.K. Song, J. Mol. Catal. A 268 (2007) 9.
[95] S. Natesakhawat, R.B. Watson, X.Q. Wang, U.S. Ozkan, J. Catal. 234
(2005) 496.
[96] S.M. Stagg-Williams, F.B. Noronha, G. Fendley, D.E. Resasco, J. Catal.
194 (2000) 240.
[97] A.W. Adamson, A.P. Gast, Physical Chemistry of Surfaces, 6th ed., John
Wiley & Sons, New York, 1997.
[
[
[
[
58] G.Y. Wang, W.X. Zhang, H.L. Lian, D.Z. Jiang, T.H. Wu, Appl. Catal. A
[98] T. Akita, M. Okumura, K. Tanaka, S. Tsubota, M. Haruta, J. Electron.
Microsc. 52 (2003) 119.
239 (2003) 1.
[
[
[
[
[
[
[
[
59] B.T. Qiao, Y.Q. Deng, Appl. Catal. B 66 (2006) 241.
[99] Z.-P. Liu, S.J. Jenkins, D.A. King, Phys. Rev. Lett. 93 (2004)
156102.
[100] L. Ilieva, G. Pantaleo, I. Ivanov, A.M. Venezia, D. Andreeva, Appl. Catal.
B 65 (2006) 101.
[101] O.P.H. Vaughan, G. Kyriakou, N. Macleod, M. Tikhov, R.M. Lambert, J.
Catal. 236 (2005) 401.
[102] C.-H. Tu, A.-Q. Wang, M.-Y. Zheng, X.-D. Wang, T. Zhang, Appl. Catal.
A 297 (2006) 40.
60] X.Y. Deng, Z. Ma, Y.H. Yue, Z. Gao, J. Catal. 204 (2001) 200.
61] M.S. Spencer, M.V. Twigg, Annu. Rev. Mater. Res. 35 (2005) 427.
62] F. Moreau, G.C. Bond, Catal. Today 114 (2006) 362.
63] B. Schumacher, V. Plzak, J. Cai, R.J. Behm, Catal. Lett. 101 (2005) 215.
64] K. Mallick, M.S. Scurrell, Appl. Catal. A 253 (2003) 527.
65] F. Moreau, G.C. Bond, Appl. Catal. A 302 (2006) 110.
66] X.-H. Zou, S.-X. Qi, Z.-H. Suo, L.-D. An, F. Li, Catal. Commun. 8 (2007)
7
84.
[103] Y.Y. Chen, C.A. Wang, H.Y. Liu, J.S. Qiu, X.H. Bao, Chem. Commun.
(2005) 5298.
[104] J. Shen, W. Shan, Y.H. Zhang, J.M. Du, H.L. Xu, K.N. Fan, W. Shen, Y.
Tang, J. Catal. 237 (2006) 94.
[
[
67] M. Manzoli, A. Chiorino, F. Boccuzzi, Appl. Catal. B 52 (2004) 259.
68] M. Azar, V. Caps, F. Morfin, J.L. Rousset, A. Piednoir, J.C. Bertolini, L.
Piccolo, J. Catal. 239 (2006) 307.