CO oxidation on supported nano-Au catalysts synthesized from a [Au6(PPh3)6](BF4) 2 complex

Article abstract of DOI:10.1006/jcat.2002.3525

Highly active and stable nano-Au catalysts synthesized by the interaction of an Au cluster-phosphine complex with a TiO₂ support were characterized and tested for CO oxidation. The pretreatment conditions showed great influence on the CO oxidation activity of the Au/TiO₂ catalysts. Under optimum pretreatment conditions, these catalysts showed high catalytic activity. Diffuse reflectance IR spectroscopy studies of CO on these supported catalysts were in excellent agreement with the IR absorption spectroscopy data obtained on model Au catalysts. The Au/TiO₂ catalyst prepared by a temperature-programmed reduction-oxidation treatment showed high CO₂ yield for the CO oxidation reaction in the presence of excess hydrogen. The HTR/LTC-treated sample (calcination in 1:2 air + The mixture for 30 min at 673 K) is a promising catalyst for preferential CO oxidation reaction.

Full text of DOI:10.1006/jcat.2002.3525

Journal of Catalysis 207, 247–255 (2002)
doi:10.1006/jcat.2002.3525, available online at http://www.idealibrary.com on
CO Oxidation on Supported Nano-Au Catalysts Synthesized
from a [Au (PPh ) ](BF ) Complex
6
3 6
4 2
∗
∗
∗,1
∗
∗,2
T. V. Choudhary, C. Sivadinarayana, C. C. Chusuei, A. K. Datye,† J. P. Fackler, Jr., and D. W. Goodman
∗
Chemistry Department, Texas A&M University, College Station, Texas 77842-3012; and †Chemical Engineering Department,
University of New Mexico, Albuquerque, New Mexico 87131
Received July 16, 2001; revised December 4, 2001; accepted January 15, 2002
well as other reactions such as methane oxidation (10–12),
Highly active and stable nano-Au catalysts have been synthesized propyleneepoxidation(13, 14), hydrogenation(15–17), and
by the interaction of an Au cluster–phosphine complex with a TiO2 NO reduction (18, 19).
x
support. These catalysts have been characterized by transmission
electron microscopy, X-ray photoelectron spectroscopy and tested
for CO oxidation. The pretreatment conditions had a profound
influence on the CO oxidation activity of the Au/TiO2 catalysts.
Under optimum pretreatment conditions these catalysts showed
orders of magnitude greater than that of Au/TiO2 synthe-
high catalytic activity. This is noteworthy, considering that previ-
ous studies (Y. Yuan, A. P. Kozlova, K. Akasura, H. Wan, K. Tsai,
and Y. Iwasawa, J. Catal. (1997) 170, 191) on similar systems (dif-
ferent pretreatment) showed that CO oxidation activity was one
The CO oxidation activity is strongly influenced by the
catalyst preparation methods. Bamwenda et al. (20) ob-
served that the CO oxidation activity of Au/TiO2 syn-
thesized by the deposition–precipitation method was four
sized by the photodeposition method. In general, copreci-
pitation, deposition–precipitation, and chemical vapor
deposition methods produce highly dispersed gold catalysts
order of magnitude lower for the Au/TiO2 catalysts. The catalyst (particle size < 5 nm) with high activity for CO oxida-
deactivation was very slow, and studies showed that the catalyst tion (21). Although the wet-impregnation method usually
could be successfully regenerated to regain initial activity. Diffuse yields inactive catalysts, studies by Vannice and co-workers
reflectance infrared spectroscopy studies of CO on these supported
catalysts are in excellent agreement with the infrared reflection
absorption spectroscopy data obtained on model Au catalysts in
our laboratory. The Au/TiO2 catalyst prepared by a temperature-
programmed reduction–oxidation treatment showed high CO2
plexes (AuL3NO3 or Au9L8(NO3)3; L = PPh3) onto “as-
yield for the CO oxidation reaction in the presence of excess
hydrogen. ꢀc 2002 Elsevier Science (USA)
showed that these catalysts could be activated by follow-
ing specific reduction–oxidation treatment procedures (22).
Recently, Iwasawa and co-workers obtained highly active
CO oxidation catalysts by grafting Au–phosphine com-
precipitated” Ti(OH)4 (4). In a series of papers (4, 23–27),
the authors strongly emphasized that the “as-precipitated”
Key Words: CO oxidation; Au/TiO2; PROX; stability; phosphine
Ti(OH)4 was essential for obtaining highly dispersed ac-
tive catalysts and that the use of a conventional TiO2 sup-
port resulted in an inactive CO oxidation catalyst (order
of magnitude lower activity) with a particle size greater
than 15 nm. We report here for the first time the success-
complex.
1
. INTRODUCTION
CO oxidation is of considerable environmental concern, ful synthesis of highly dispersed Au catalysts (<5 nm) us-
since even small exposures to CO (ppm) can be lethal. It is ing a wet-impregnation method on a conventional TiO2
therefore of interest to develop highly active CO oxidation support. High-temperature reduction–oxidation pretreat-
catalysts to remove trace amounts of CO. Historically, gold ment of [Au (PPh ) ](BF ) /TiO resulted in highly dis-
6
3 6
4 2
2
was considered to be inactive; however, recently Haruta persed gold catalysts with activities higher than those ob-
and co-workers (1) showed that highly dispersed gold cata- served by Iwasawa and co-workers on the Au/Ti(OH)4
lysts are extremely active for CO oxidation and exhibit high systems (4). The effect of various pretreatment conditions
catalytic activity even at subambient temperatures. Follow- on the CO oxidation activity was highlighted in this inves-
ing this work, there have been a number of investigations tigation. Along with the kinetics of CO oxidation, the cata-
dealing with gold-based catalysts for CO oxidation (2–9) as lyst was also tested for preferential CO oxidation (PROX)
in the presence of hydrogen. PROX is of great interest in
state-of-the-art low-temperature fuel cell systems, as trace
amounts of CO are known to poison fuel cell electrodes
(28–33). PROX can be effectively employed to remove
trace amounts of CO from the hydrogen stream prior to its
1
Present address: Chemistry Department, University of Missouri–
Rolla, Rolla, MO 65401.
2
To whom correspondence should be addressed. E-mail: goodman@
mail.chem.tamu.edu.
247
0
021-9517/02 $35.00
c 2002 Elsevier Science (USA)
All rights reserved.
ꢀ

2
48
CHOUDHARY ET AL.
introduction into the fuel cell. One of the most challenging PHI 560 system using a PHI 25–270AR double-pass cylin-
problems faced by Au catalysts is stability and regeneration; drical mirror analyzer. Survey scans were performed using
this aspect has received little attention (34, 35). This paper a 100-eV pass energy; high-resolution scans were per-
includes an investigation dealing with catalyst stability and formed with a 50-eV pass energy. The signal from adven-
regeneration.
titious carbon at BE = 284.7 eV for the C 1s level was
used to correct for sample charging. Samples were mounted
onto a 1.0 × 1.0 × 0.1 cm support using a double-sided tape
(Scotch 3 M) attached to a probe and introduced into the
UHV via a turbopumped antechamber. The support was
attached to a probe that was differentially pumped using
sliding seals. Curve fitting of the XPS data was performed
using Peakfit Ver. 3.1 B (Jandel Scientific) software. De-
tailed information regarding the XPS instrument is avail-
able elsewhere (39).
2
. EXPERIMENTAL
2
.1. Catalyst Synthesis
The supported gold catalyst was synthesized by grafting
an [Au6(PPh3)6](BF4)2 complex onto TiO2. The synthesis
of [Au6(PPh3)6](BF4)2] is reported elsewhere (36–38). The
complex was dissolved in CH2Cl2 and deposited on TiO2
(
Degussa) at room temperature to obtain a nominal gold
2
.2.3. Diffusereflectanceinfraredspectroscopy(DRIFTS).
loading of 1 wt% with respect to the support. Following
this, the catalyst was mixed thoroughly for ca. 2 h at room
temperature until the solvent had completely evaporated.
The DRIFTS experiments were performed with a Perkin–
Elmer Spectrum 2000 spectrometer equipped with a MCT
detector and a DRIFTS (Harrick) cell used in a flow mode.
The samples under investigation were treated in situ when
pretreatment temperatures were less than 673 K; they were
treated externally in a quartz reactor for treatment tem-
peratures exceeding 673 K. The number of scans employed
varied from 25 to 100 depending on the conditions and the
sample employed. Background signals from gas-phase CO
and CO2 were subtracted before the spectra were reported.
2
TiO2 (Degussa P-25; surface area, 35–65 m /g) was calcined
at 723 K for 4 h prior to the catalyst synthesis. The as-
synthesized catalyst was then pretreated under different
conditions prior to testing for catalytic activity. The follow-
ing pretreatment methods were employed: (i) calcination in
1
: 2 air+He mixture at 673 K for 1 h (LTC); (ii) calcination
in 1 : 2 air+He mixture at 773 K for 1 h (HTC); (iii) heating
in 1 : 2 H2 + He mixture at 773 K for 30 min, flushing with
He for 10 min, and calcination in 1 : 2 air + He mixture for
2
.3. Catalyst Activity Tests
3
0 min at 673 K (HTR/LTC); and (iv) calcination in 1 : 2
The activity tests were carried out in a conventional-flow
air + He mixture at 773 K for 30 min, flushing with He for
reactor system consisting of a quartz reactor, a custom-built
furnace, and an Omega temperature controller which could
be used to ramp the temperature at the desired heating
rate. The gas-analysis system consisted of a Varian 3700 gas
chromatograph equipped with a thermal conductivity de-
tector and a carbosphere column for gas separation. CO
(research purity, Matheson), O2 (UHP, Matheson), and He
1
3
0 min, and heating in 1 : 2 H2 + He mixture at 673 K for
0 min (HTC/LTR). After the pretreatment the sample was
cooled to the desired reaction temperature and a ramp rate
of 5 K/min was employed for sample heating.
2
.2. Catalyst Characterization and
Instrumental Techniques
(
UHP, Matheson) were employed for the study. Prelimi-
2
.2.1. Transmission electron microscopy (TEM). TEM
nary partial dependence studies of CO and O2 on the CO
oxidation activity revealed identical reaction orders for the
HTR/LTC- and the HTC/LTR-treated samples. Detailed
studies (similar to those in Ref. (22)) were carried out on the
HTC/LTR-treated sample, under conditions where there
was very little/no catalyst deactivation, with the conversion
maintained below 20%. The catalytic activity for the dif-
ferent pretreatments was measured at a gas-hourly space
micrographs of the supported Au catalysts were initially
obtained using a Zeiss 10 C electron microscope (Elec-
tron Microscopy Center, Texas A&M University). To con-
firm the results, the TEM study was also conducted on a
high-resolution JEOL 2010 microscope at the University
of New Mexico. Essentially identical results were obtained
on both microscopes. Samples after reaction were ultraso-
nically dispersed in ethanol/acetone and spread over per-
forated carbon grids. On the high-resolution microscope,
several bright-field TEM micrographs of different portions
ofthesamplewereobtainedatmagnificationsupto400,000,
whereas on the Zeiss 10 C the micrographs were obtained
at magnifications up to 100,000. At least 100 particles were
used to obtain a particle size distribution.
3
velocity (GHSV) of 20,000 cm /g/h and CO and O2 partial
pressures of 3.67 kPa. The gas composition for the PROX
reaction was CO : O2 : H2 : He = 1 : 2 : 50 : 25 and the GHSV
3
was 90,000 cm /g/h. The gas-feed concentration was var-
ied using Brooks mass flow controllers. The catalysts were
used in a finely powdered state, and evaluation of the Weisz
criterion (40) showed the absence of any mass-transfer-
2
.2.2. X-ray photoelectron spectroscopy (XPS). XPS related problems. A carbon mass balance of ± 3% was ob-
was performed in an ion-pumped (300 L/s) Perkin–Elmer tained for the activity runs.

CO OXIDATION ON SUPPORTED NANO-Au CATALYSTS
249
3
. RESULTS
35
30
a
Figure 1a shows the TEM image of 1% Au/TiO2 after
the HTR/LTC treatment of the as-synthesized catalyst. It
is apparent that the HTR/LTC procedure results in a high
dispersion of gold particles on TiO2. A major fraction of the
gold particles were in the range of 3–6 nm (shown in Fig. 1b)
and exhibited an average particle size of 4.7 nm. It is note-
worthy that the size of the gold particles did not exceed
25
20
15
8
nm. However, larger gold particles were observed after
the HTC and HTC/LTR treatment procedures (Figs. 2a and
b). In the case of HTC-pretreated catalysts (Fig. 2a), al-
10
5
2
though the particle size varied from 3 to 12 nm, most of the
particles were concentrated in the 7–9 nm region. In con-
trast, a larger range of particle size distribution (with higher
density in the 5–11 nm region) was observed subsequent to
0
2
3
4
5
6
7
8
9
10
11
12
13
Particle size (nm)
20
15
10
5
0
b
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18
Particle size (nm)
FIG. 2. Particle size distribution for the (a) HTC-treated and
(
b) HTC/LTR-treated Au/TiO2 catalyst.
the HTC/LTR treatment procedures. The average particle
sizes were estimated to be 7.9 and 8.5 nm, respectively, for
the HTC- and HTC/LTR-treated samples.
The XP spectra of Au 4 f and P 2p levels under vari-
ous treatment conditions are shown in Figs. 3a and 3b, re-
spectively. Similar to the work of Iwasawa and co-workers
2
2
1
1
5
0
5
0
5
0
b
(
26), a shift from a higher binding energy (84.7 eV) for as-
synthesized Au/TiO2 catalysts to a lower (ca. 83.6–83.9 eV)
binding energy for catalysts treated at high temperatures
(
>673 K) was observed for the Au 4 f7/2 level. This is
in good agreement with the transformation of the Au–
phosphine complex to metallic gold particles as reported
in the literature (26). The absence of appreciable bind-
ing energy shifts for the HTC, HTR/LTC, and HTC/LTR
treatment conditions precludes any appreciable electronic
structural change for the gold particles due to the different
pretreatment procedures. For the as-synthesized samples,
two peaks due to phosphorous were observed at 131.6 and
135.3 eV, respectively (Fig. 3b). The higher binding energy
1
2
3
4
5
6
7
8
9
10
Particle size (nm)
FIG. 1. (a) TEM image and (b) particle size distribution for the
HTR/LTC-treated Au/TiO2 sample.

2
50
CHOUDHARY ET AL.
a
b
a
b
c
a
b
c
d
d
9
5
90
85
80
75
145
140
135
130
125
120
Binding Energy, eV
Binding Energy, eV
FIG. 3. XPS spectra of the (a) Au 4 f and (b) P 2p for the Au/TiO2 catalysts after a, no pretreatment; b, HTC; c, HTR/LTC; and d, HTC/LTR
treatments.
feature disappeared after the high-temperature treatments catalytic activity for the different pretreatment conditions
(
≥673 K), whereas the peak at 131.6 eV shifted to the re- was HTR/LTC ꢁ HTC/LTR > HTC > LTC. While 100%
gion 133.8–134.2 eV. Similarly to a previous study (26), this CO conversion was achieved at 393 K for the HTR/LTC-
shift is attributed to the oxidation of the phosphine species treated catalyst, complete CO conversions were obtained
during the high-temperature treatments.
only at 623 K for the LTC sample. Table 1 compares the
Figure 4 shows the effect of the CO oxidation catalytic ac- CO oxidation catalytic activities for Au/TiO2 samples syn-
tivity (27.5 Torr CO and 27.5 Torr O2) of the Au/TiO2 sam- thesized by the various procedures. The HTR/LTC catalyst
ple as a function of different pretreatment procedures. The clearly shows CO oxidation activity comparable to that of
as-synthesized sample showed no catalytic activity at tem- the most active Au/TiO2 catalysts reported in the literature.
peratures below 373 K (not shown in the figure). A differ-
The catalyst stability/regeneration investigation on a
ence in CO conversion of more than an order of magnitude HTR/LTC-treated sample is depicted in Fig. 5. In these
was observed at room temperature (293 K) for the LTC- studies, a 1 : 2 CO/Oxygen ratio was used to simulate harsh
and the HTR/LTC-pretreated catalysts. The order of the
100
8
0° C
(
III)
100
80
4
0 ° C
80
60
40
20
0
60
40
20
0
(
I)
(II)
LTC
HTC
HTC/LTR
HTR/LTC
0
2
4
6
8
20
22
24
26
28
Time (h)
0
50
100
150
200
250
300
350
400
Temperature (° C)
FIG. 5. Time-on-stream profile of the Au/TiO sample after I, fresh
2
HTR/LTC treatment sample (Trxn = 313 K); II, after first regeneration
FIG. 4. Effect of different pretreatments on the CO conversion as
by HTR/LTC treatment (Trxn = 313 K); and III, after second regener-
3
3
a function of temperature (GHSV = 20,000 cm /g/h; PCO = 3.67 kPa and
ation by HTR/LTC treatment (Trxn = 353 K). (GHSV = 20,000 cm /g/h;
PO = 3.67 kPa).
PCO = 3.67 kPa and PO = 7.34 kPa).
2
2

CO OXIDATION ON SUPPORTED NANO-Au CATALYSTS
251
operating conditions; it is known from the literature that
2
107
long exposures to air cause sintering of the gold particles
andtherebydeactivation(7). Boththepretreatmentandthe
regeneration treatments involved HTR/LTC treatments.
After initial CO oxidation (27 h) at 313 K (profile I), the
catalyst was regenerated and again tested for activity at
e
3
13 K (profile II). Following this, the catalyst was again
regenerated and CO oxidation was carried out at 353 K
Reaction III). The deactivation rate for Reaction I was
relatively small for the initial 9 h and dropped from 59 to
1% over a period of 20 h. The activity remained constant
2
119
d
c
(
2
115
112
2
4
after this period until the reaction was stopped (27 h). Fol-
lowing regeneration, the initial activity was reobtained in
II and a profile similar to that of I was obtained, indicating
excellent regeneration of the catalyst. After the second re-
generation, the catalytic activity was observed to decrease
from ca. 87 to ca. 80% (reaction temperature, 353 K) after
b
a
2
107
2
300 2250 2200 2150 2100 2050 2000 1950 1900 1850
-1
Wave numbers (cm )
2
1 h, subsequent to which stabilized CO oxidation conver-
sion was obtained.
FIG. 7. DRIFT spectra of CO surface species on the HTR/LTC-
Figure 6 shows the partial pressure dependence of the treated sample at 293 K (a) under 5% CO in He after 10 min, (b) under
CO oxidation reaction on CO and O2 at 293 K. For these 5% CO and 5% O2 in He after 10 min, (c) 5 min after stopping CO and
O2, (d) 12 min after stopping CO and O2, and (e) 8 min after reintroducing
studies, the partial pressure of either CO or O2 was main-
5% CO in the He stream.
tained at 27.5 Torr while the other was varied. The slopes
of the linear graphs obtained by plotting the CO turnover
frequency (TOF) versus pressure in logarithmic form were
used to estimate the order of the reaction. The order of the
CO oxidation reaction was found to be 0.2 with respect to
CO and 0.46 with respect to O2. These studies are in ex-
cellent agreement with a recent study on similar systems,
where the reaction orders were found to be 0.25 and 0.41
for CO and O2, respectively (24). On Au/TiO2 catalysts
prepared by the wet-impregnation method, Lin et al. (22)
observed reaction orders of 0.24 (P: 17–187 Torr) and 0.4
for CO and O2, respectively, at 313 K. Their studies indicate
that the CO reaction orders are dependent on temperature
and pressure.
The DRIFTS spectra shown in Fig. 7 were obtained un-
der different reactant feed compositions on the HTR/LTC-
treated Au/TiO2 sample. Ten minutes after the introduc-
tion of CO, an adsorption feature due to CO (shown in
−
1
Fig. 7a) appeared at 2107 cm and is attributed to CO ad-
sorption on gold sites (consistent with previous literature
data (2)). Apart from this, low-intensity nonresolved broad
features (not shown in the figure) were observed in the car-
−
1
bonate/carboxylate region (1200–1700 cm ). In good cor-
-
-
-
-
-
-
-
-
-
1.50
1.55
1.60
1.65
1.70
1.75
1.80
1.85
1.90
respondence to the work by Bollinger and Vannice (2), no
P = 27.5 Torr
PCO= 27.5 Torr
O
2
−1
peak was observed at ∼2183 cm on the Au/TiO2 sample,
indicating the absence of adsorbed CO on TiO2. The intro-
duction of air into the CO + He mixture resulted in a peak
−
1
shift of the adsorbed CO to 2112 cm . Along with this an
additional intense and broad feature from gas-phase CO2
was observed (the gas-phase signal has not been shown).
The peak shift of the CO band to higher binding energy
is consistent with the decrease of the CO coverage on Au
and corroborates recent work in our group on CO probed
by infrared reflection absorption spectroscopy (IRAS) on
the Au(110) surface (41). Spectra c and d, which were ob-
tained after stopping CO and air, showed a decreasing CO
coverage on the gold surface with time and indicated weak
chemisorption of CO on gold (consistent with literature re-
ports (2, 42)). In line with the earlier trend, the peaks shifted
to higher binding energies with decreasing CO coverage.
After the peak due to CO adsorption had completely
disappeared, CO was again introduced into the He stream
x
y
Rate=kPCO PO
2
x=0.2, y=0.46
1
.4
1.6
1.8
2.0
2.2
log(P)
FIG. 6. Partial pressure dependence of CO (at constant PO =
2
2
7.5 Torr) and O2 (at constant PCO = 3.67 kPa) on the CO oxidation ac-
3
tivity at 293 K (HTC/LTR-treated sample: GHSV = 20,000 cm /g/h).

2
52
CHOUDHARY ET AL.
100
followed by a temperature-programmed (4 K/min) calci-
nation treatment; Ti(OH)4 was prepared by the hydrolysis
of titanium–tetraisopropoxide with sodium carbonate and
further washed and filtered as explained in Ref. (26). Under
Yuan et al.’s experimental conditions, the freshly prepared
Ti(OH)4 was found to be essential for the synthesis of highly
dispersed Au catalysts; conventional TiO2 and commercial
Ti(OH)4 were found to be inefficient for the successful syn-
thesis (4, 26).
The present study has demonstrated the successful syn-
thesis (Figs. 1a and 1b) of highly dispersed Au catalysts
(<5 nm) by the temperature-programmed reduction–cal-
cination (HTR/LTC) of Au6(PPh3)6(BF4)2 with a conven-
90
80
70
60
50
40
30
20
10
0
conversion % of CO
selectivity for CO2
20
40
60
80
100
120
140
tional TiO support (Degussa P-25), as predicted by the-
2
Temperature (°C)
FIG. 8. Conversion/selectivity for the PROX reaction on the
oretical calculations (43). Similarly, recent STM studies (44)
in our laboratory have revealed the presence of highly
dispersed Au (<5 nm) on a TiO2 support after a wet
chemistry deposition of Au (PPh ) (BF ) on planar TiO
HTR/LTC-treatedsampleasafunctionofreactiontemperature(GHSV =
3
9
0,000 cm /g/h; CO : H2 : O2 : He = 1 : 50 : 2 : 25).
6
3 6
4 2
2
(
110). In contrast, Iwasawa and co-workers observed an
average particle size of ca. 30 nm after the temperature-
programmed calcination (heating rate: 4 K/min, 4 h in air)
of Au(PPh3)(NO3) and the conventional TiO2 (Degussa P-
25 support). Interestingly, the high-temperature calcination
and spectrum e was obtained. This procedure resulted in a
peak at the same position as that obtained in spectrum a
−
1
(
2107 cm ), showing the reproducibility of the peak shifts
(
HTC) treatment employed in our study also resulted in a
with CO surface coverage.
catalyst with an average particle size of only ca. 8 nm. The
reason for the above is not clearly understood at this point;
however, it should be noted that the solvents for the phos-
phine complex were different (CH2Cl2 in our study as op-
posed to acetone in the study by Iwasawa and co-workers).
Iwasawa and co-workers (27) emphasized the importance
of both the state of the support surface and the choice of the
Au complex in the synthesis process. The solvents, owing to
their difference in polarity, may alter support surfaces in
different ways and hence modify the cluster–support inter-
action. That the HTR/LTC treatment plays a decisive role
in stabilizing the Au clusters on TiO2 is apparent from our
studies.
The results obtained from the preferential CO oxidation
studies on the HTR/LTC-treated Au/TiO2 in the presence
of excess hydrogen are shown in Fig. 8. The CO conver-
sion increased with an increase in temperature, whereas
the selectivity decreased. The Au/TiO2 catalyst showed an
extremely high propensity to oxidize CO preferentially at
3
13K(95%selectivityforCOoxidation). Itshouldbenoted
thatwhileanincreaseintemperaturefrom353to393Konly
marginally increased the conversion, it resulted in a drastic
decrease (73 to 30%) in the CO oxidation selectivity.
4
. DISCUSSION
As discussed earlier, the CO oxidation activity for the
Gold catalysts have received wide attention in recent HTR/LTC sample was comparable to that of the most ac-
years due to their exceptionally high CO oxidation activity. tive Au/TiO catalysts reported in the literature (Table 1).
2
In general, high catalytic activity has been observed only The CO oxidation activity of the HTR/LTC-treated sample
on highly dispersed Au catalysts (particle size <5 nm). The had ca. 100 times greater CO oxidation activity than that
study by Lin and Vannice (22) is an exception in that reaso- of the catalysts (calcined at 673 K) prepared from the Au–
nably high CO oxidation catalytic activity was observed phosphine complex on the conventional Degussa support
on Au/TiO2 (average particle size of 30 nm) after a se- by Iwasawa and co-workers (26). More importantly, the
quential reduction–oxidation–reduction treatment. Cata- HTR/LTC-treated sample showed ca. five times greater ac-
lyst preparation methods such as deposition–precipitation, tivity than that of the most active Au/Ti(OH) samples em-
4
coprecipitation, and chemical vapor deposition have been ployed in Refs. (4, 26). It should be noted (from Table 1)
regularly used to obtain Au catalysts with average particle that different partial pressures of CO and O were emp-
2
sizes of less than 5 nm. Recent studies by Iwasawa and co- loyed in the two studies. This, however, does not pose pro-
workers have shown that highly dispersed Au catalysts can blems for comparison, since the partial pressure depen-
also be obtained by the interaction of an Au–phosphine denceofthereactantsontheCOoxidationactivityissimilar
complex with as-precipitated wet metal hydroxide supports for both studies (see Results). Calculations indicate that a
(
4). In particular, the Au/Ti(OH)4 catalyst was obtained by direct comparison (from Table 1) of the catalysts is valid
grafting Au(PPh3)(NO3) onto the as-precipitated Ti(OH)4 in this case, as the greater activity expected due to the

CO OXIDATION ON SUPPORTED NANO-Au CATALYSTS
253
TABLE 1
Comparison of CO Oxidation Rates on Au/TiO2 Catalysts Prepared by Different Methods
Preparation
methodsa
dAu
(nm)
PCO
(kPa)
PO2
(kPa)
Ea
(kJ/mol)
Rate CO
(mmol/s/gAu )
Au precursor
Catalyst
T (K)
References
HAuCl4
HAuCl4
HAuCl4
HAuCl4
2.3% Au/TiO2
3.1% Au/TiO2
1% Au/TiO2
DP
DP
FD
2.5
2.9
4.6
nr
1
1
1
1
5
1
1
0.25
1
20
20
20
20
4.8
20
20
0.25
20
20
9
300
300
300
300
313
262
293
300
300
313
293
20
27
56
58
13
nr
nr
nr
nr
12
16.3
0.2
0.6
(20)
(20)
(20)
(20)
(22)
(3)
(3)
(9)
(26)
−
−
5
5
2 × 10
1% Au/TiO2
IMP
IMP
CVD
CVD
AuC
IOH
IMP
IMP
2 × 10
b
HAuCl4
2.3% Au/TiO2
4.6% Au/TiO2
2.4% Au/TiO2
1.7% Au/TiO2
3% Au/Ti(OH)4
3% Au/TiO2
30
0.05
Me2Au(acac)
Me2Au(acac)
HAuCl4
AuPPh3NO3
AuPPh3NO3
Au6(PPh3)6(BF4)2
∼2.9
0.05c
0.06d
0.01
0.05
nr
2
3
30
4.7
−
3
1
3.67
6 × 10
(26)
0.3e
f
1% Au/TiO2
a
Abbreviations used: DP, deposition–precipitation; FD, photodecomposition; IMP, impregnation of oxides; IOH, impregnation of as-precipitated
hydroxide; CVD, chemical vapor deposition; nr, not reported.
b
HTR/C/lTR.
amorphous TiO2.
anatase TiO2.
HTR/LTC.
c
d
e
f
Present work.
high partial pressures of CO used in our work is essentially work (average particle size = 30 nm). The difference in dis-
compensated for by the larger O2 partial pressures used persion is attributed to the different decomposition behav-
in the study by Iwasawa and co-workers. The higher activ- ior of the Au precursors used in the two studies; AuCl3 was
ity of our catalyst may be tentatively attributed to the diff- used in their investigation while [Au6(PPh3)6](BF4)2 was
erence in preparation procedure. The beneficial effect of employed in this study.
the reduction–oxidation treatment observed in this study is
The HTR/LTC-treated sample showed very slow initial
similar to that observed by Lin et al. (22) on Au/TiO2 cata- catalyst deactivation at 313 K, followed by stable conver-
lysts prepared by the wet-impregnation method, wherein sions for many hours (Fig. 5, profile I). In contrast, studies
the highest activity (22) was observed after the following on model catalyst systems (46) and catalysts synthesized
sequence: high-temperature reduction at 773 K followed by by the conventional deposition–precipitation method (47)
calcination at 673 K and finally low-temperature reduction showed rapid deactivation during CO oxidation at above
at 473 K (HTR/C/lTR). The authors strongly emphasized ambient temperatures. The deactivation was also found
the high-temperature reduction requirement to obtain high to be more rapid in Au/TiO2 catalysts synthesized by the
CO oxidation catalytic activity. Their studies revealed that conventional wet-impregnation method (22). Most impor-
a HTR treatment induced a low CO oxidation activity in tantly, our studies have shown that the HTR/LTC sample
TiO2, which was otherwise totally inactive for CO oxidation could be successfully regenerated (as seen from the simi-
(
22). However, it is clear that the large increase in activ- larity of the time-on-stream profile to that of the fresh cata-
ity after the HTR/LTC treatment cannot be explained by lyst after regeneration (profile II, Fig. 5)) by a subsequent
the contribution from the support alone. According to the HTR/LTC treatment. Even after the second regeneration
literature, a high-temperature reduction treatment causes (profile III), when the CO oxidation reaction was carried
some transformation from the anatase form to the rutile out at 353 K, the catalyst followed a trend similar to that of
form (45). Based on our results and those in the litera- the fresh sample: very slow initial deactivation followed by
ture, it may be speculated that the HTR/LTC pretreatment stable conversions. These investigations, which were per-
modifies the TiO2 support so that there is an optimized in- formed over periods longer than 24 h, clearly show that
teraction between Au (obtained from [Au6(PPh3)6](BF4)2] the HTR/LTC-treated sample is stable and can easily be
complex) and TiO2; this leads to a higher dispersion and an regenerated. The observed high stability may be attributed
enhanced CO oxidation activity. Further studies are, how- to the stabilization effect of the HTR/LTC treatment of the
ever, required to provide a clearer definitive view of this Au clusters on the TiO2 surface.
process. The CO oxidation activity observed in our work
was much higher than that found in the study by Lin et al. orders for CO oxidation on our Au/TiO2 catalyst are 0.2
22), because of the enhanced Au dispersion (average par- and 0.46 for CO and O2, respectively. These agree very
ticle size = 4.7 nm) in our case as opposed to that in their well with the values obtained by Iwasawa and co-workers
The kinetics experiments (Fig. 6) show that the reaction
(

2
54
CHOUDHARY ET AL.
(
24) on Au/Ti(OH)4, wherein they proposed a Langmuir–
ACKNOWLEDGMENTS
Hinshelwood model which involved the reaction of ad-
sorbed CO molecules (Au surface) with O2 (adsorbed on
oxygen vacancies of the support) as the rate-limiting step.
The activation energy (16.3 kcal/mol) obtained in this study
We acknowledge with pleasure the support of this work by the Depart-
ment of Energy, Office of Basic Energy Sciences, Division of Chemical
Sciences. We are grateful to Professor J. H. Lunsford for giving us access
to the DRIFTS setup, to Professor P. R. Sharp (University of Missouri
for the HTR/LTC sample lies within the range of values at Columbia) for supplying us with the gold complex for these studies,
and to D. C. Meier for helpful discussions on the DRIFTS data. TVC
(
12–21 kcal/mol) observed by Iwasawa and co-workers (23)
gratefully acknowledges the Link Foundation for the Link Energy Fellow-
ship. CCC gratefully acknowledges support from the Associated Western
Universities, Inc., and Pacific Northwest National Laboratories operated
by Battelle Memorial.
on the catalysts synthesized from Au–phosphine complexes
and TiO2. Our DRIFTS studies show weak CO chemisorp-
tion on the Au surface and indicate a blue shift of the CO
band at lower CO coverages. The position of the CO band
(
DRIFTS) and the XPS (Au 4 f ) provide evidence that Au
is present in the metallic form after the HTR/LTC treat-
ment. The XPS also reveals the presence of an oxidized
phosphorus species on the surface; however, at this point,
the role and nature of this species are unknown.
The main requirement for an effective PROX catalyst
is high CO oxidation activity coupled with low hydrogen
oxidation activity. CO oxidation in the presence of excess
hydrogen (Fig. 8) shows promising results for the HTR/
LTC treated sample. A high selectivity of ca. 73% for CO2
was obtained at a high CO oxidation rate of 0.6 mmol/s/gAu
REFERENCES
1
. Haruta, M., Tsubota, S., Kobayashi, T., Kageyama, H., Genet, M. J.,
and Delmon, B., J. Catal. 144, 175 (1993).
. Bollinger, M. A., and Vannice, M. A., Appl. Catal. B 8, 417 (1996).
. Okumura, M., Tanaka, K., Ueda, A., and Haruta, M., Solid State Ionics
2
3
9
5, 143 (1997).
4. Yuan, Y., Kozlova, A. P., Akasura, K., Wan, H., Tsai, K., and Iwasawa,
Y., J. Catal. 170, 191 (1997).
. Tsubota, S., Nakamura, T., Tanaka, K., and Haruta, M., Catal. Lett. 56,
31 (1998).
. Okamura, M., Nakamura, S., Tsubota, S., Nakamura, T., Azuma, M.,
5
1
6
(
CO : H2 : O2 = 1 : 50 : 2) at 353 K, the approximate op-
and Haruta, M., Catal. Lett. 51, 53 (1998).
7. Valden, M., Lai, X., and Goodman, D. W., Science 281, 1647 (1998).
erating temperature of a PEM fuel cell. Comparison with
8
9
. Griesel, R. J. H., and Nieuwenhuys, B. E., J. Catal. 199, 48 (2001).
. Grunwaldt, J.-D., Kiener, C., Wogerbauer, C., and Baiker, A., J. Catal.
81, 223 (1999).
0. Waters, R. D., Weimer, J. J., and Smith, J. E., Catal. Lett. 30, 181
1995).
previous work (Au/MnOx : ∼0.2 mmol/s/g at 353 K (48)
Au
and Au/Al2O3: ∼0.3–1.7 mmol/s/g at 363 K (32) indicates
Au
1
comparable activities for CO oxidation for the PROX re-
action. Operation temperatures above 373 K result only
1
(
in a marginal increase in the CO oxidation activity while 11. Griesel, R. J. H., Kooyman, P. J., and Nieuwenhuys, B. E., J. Catal. 191,
4
30 (2000).
greatly decreasing the selectivity for the desired CO2 prod-
uct, thus implying a favorable pathway for hydrogen oxida-
tion over CO oxidation at higher temperatures. These re-
sults are similar to those of previous studies (6, 29), which
show that highly dispersed Au particles are more active for
CO oxidation as compared to hydrogen oxidation at low
temperatures, while larger Au particles are more active for
hydrogen oxidation.
1
2. Choudhary, T. V., Banerjee, S., and Choudhary, V. R., Submitted for
publication.
1
1
3. Hayashi, T., Tanaka, K., and Haruta, M., J. Catal. 178, 566 (1998).
4. Strangland, E. E., Stavens, K. B., Andres, R. P., and Delgass, W. N.,
J. Catal. 191, 332 (2000).
5. Sakurai, H., and Haruta, M., Appl. Catal. A 127, 93 (1995).
6. Jia, J., Haraki, K., Kondo, J. N., Domen, K., and Tamaru, K., J. Phys.
Chem. B 104, 11,153 (2000).
7. Choudhary, T. V., Sivadinarayana, C., and Goodman, D. W., In prepa-
ration.
8. Salama, T. M., Ohnishi, Y., and Ichikawa, M., J. Chem. Soc., Faraday
Trans. 92, 301 (1996).
9. Salama, T. M., Ohnishi, R., and Ichikawa, M., Chem. Commun. 105
1
1
1
1
1
2
2
5
. CONCLUSION
(
1997).
Highly dispersed and active Au catalysts were synthe-
sized by wet impregnation of [Au6(PPh3)6](BF4)2 on a
conventional TiO2 support. The pretreatment conditions
0. Bamwenda, G. R., Tsubota, S., Nakamura, T., and Haruta, M., Catal.
Lett. 44, 83 (1997).
1. Bond, G. C., and Thompson, D. T., Catal. Rev.–Sci. Eng. 41, 319 (1999).
played a crucial role in defining catalyst dispersion and 22. Lin, S. D., Bollinger, M., and Vannice, M. A., Catal. Lett. 17, 245 (1993).
2
2
2
2
2
2
3. Yuan, Y., Akasura, K., Wan, H., Tsai, K., and Iwasawa, Y., Catal. Lett.
2, 15 (1996).
4. Liu, H., Kozlov, A., Kozlova, A. P., Shido, T., Akasura, K., and
Iwasawa, Y., J. Catal. 185, 252 (1999).
5. Yuan, Y., Akasura, K., Wan, H., Tsai, K., and Iwasawa, Y., Chem. Lett.
755 (1996).
6. Yuan, Y., Akasura, K., Kozlova, A. P., Wan, H., Tsai, K., and Iwasawa,
Y., Catal. Today 44, 333 (1998).
activity. The HTC/LTR treatment produced not only ex-
tremely active CO oxidation catalysts but also highly sta-
ble catalysts that could be easily regenerated. Characteri-
zation studies showed the presence of metallic Au on the
surface along with some oxidized phosphorus species after
the HTR/LTC treatment. DRIFTS experiments revealed
weak chemisorption of CO on the metallic Au particles,
with a blue shift of the CO band at decreasing CO cov-
erages. The HTR/LTC-treated sample was found to be a
promising catalyst for PROX reactions.
4
7. Kozlov, A., Kozlova, A. P., Liu, H., and Iwasawa, Y., Appl. Catal. A.
182, 9 (1999).
8. Choudhary, T. V., and Goodman, D. W., Catal. Lett. 59, 93 (1999).
29. Grisel, R. J. H., and Nieuwenhuys, B. E., J. Catal. 199, 48 (2001).

CO OXIDATION ON SUPPORTED NANO-Au CATALYSTS
255
3
3
0. Choudhary, T. V., Sivadinarayana, C., and Goodman, D. W., Catal.
Lett. 72, 197 (2001).
1. Kahlich, M. J., Gasteiger, H. A., and Behm, R. J., J. Catal. 182, 430
39. Choudhary, T. V., Sivadinarayana, C., Chusuei, C., Klinghoffer, A.,
and Goodman, D. W., J. Catal. 199, 9 (2001).
40. Weisz, P. B., Chem. Eng. Prog. Series 55, 29 (1959).
41. Meier, D. C., Bukhtiyarov, V., and Goodman, D. W., In preparation.
42. Grunwaldt, J.-D., Maciejewski, M., Becker, O. S., Fabrizioli, P., and
Baiker, A., J. Catal. 186, 458 (1999).
43. Omary, M. A., Raevashdeh-Omary, M. A., Chusuei, C. C., Fackler,
J. P., Jr., and Bagus, P. S., J. Chem. Phys. 114, 10695 (2001).
44. Chusuei, C., Lai, X., Davis, K., Bowers, E., Fackler, J. P., Jr., and
Goodman, D. W., Langmuir 17, 4113 (2001).
45. Haller, G. L., and Resasco, D. E., Adv. Catal. 36, 173 (1989).
46. Valden, M., Pak, S., Lai, X., and Goodman, D. W., Catal. Lett. 56, 7
(1998).
(1999).
3
3
3
2. Bethke, G. K., and Kung, H. H., Appl. Catal. A. 194, 43 (2000).
3. Choudhary, T. V., and Goodman, D. W., J. Catal. 192, 316 (2000).
4. Gardner, S. D., Hoflund, G. B., Upchurch, B. T., Schryer, D. R., Kielen,
E. J., and Schryer, J., J. Catal. 129, 114 (1991).
5. Hutchings, G. J., Siddiqui, M. R. H., Burrows, A., Kiely, C. J., and
Whyman, R., J. Chem. Soc., Faraday Trans. 93, 187 (1997).
3
3
3
6. Ramamoorthy, V., Wu, Z., Yi, Y., and Sharp, P. R., J. Am. Chem. Soc.
1
14, 1526 (1992).
7. Dyson, P. J., and Mingos, D. M. P., “Gold: Progress in Chemistry, Bio-
chemistry and Technology” (H. Schmidbaur, Ed.), pp. 511–556. Wiley,
New York, 1999.
47. Choudhary, T. V., Sivadinarayana, C., and Goodman, D. W., Unpub-
lished work.
3
8. Flint, B. W., Yang, Y., and Sharp, P. R., Inorg. Chem. 39, 602
48. Sanchez, R. M. T., Ueda, A., Tanaka, K., and Haruta, M., J. Catal. 168,
125 (1997).
(2000).