Thiolate ligands as a double-edged sword for CO oxidation on CeO 2 supported Au25(SCH2CH2Ph) 18 nanoclusters

Article abstract of DOI:10.1021/ja5018706

The effect of thiolate ligands was explored on the catalysis of CeO ₂ rod supported Au₂₅(SR)₁₈ (SR = -SCH ₂CH₂Ph) by using CO oxidation as a probe reaction. Reaction kinetic tests, in situ IR and X-ray absorption spectroscopy, and density functional theory (DFT) were employed to understand how the thiolate ligands affect the nature of active sites, activation of CO and O₂, and reaction mechanism and kinetics. The intact Au₂₅(SR)₁₈ on the CeO₂ rod is found not able to adsorb CO. Only when the thiolate ligands are partially removed, starting from the interface between Au₂₅(SR)₁₈ and CeO₂ at temperatures of 423 K and above, can the adsorption of CO be observed by IR. DFT calculations suggest that CO adsorbs favorably on the exposed gold atoms. Accordingly, the CO oxidation light-off temperature shifts to lower temperature. Several types of Au sites are probed by IR of CO adsorption during the ligand removal process. The cationic Au sites (charged between 0 and +1) are found to play the major role for low-temperature CO oxidation. Similar activation energies and reaction rates are found for CO oxidation on differently treated Au₂₅(SR) ₁₈/CeO₂ rod catalysts, suggesting a simple site-blocking effect of the thiolate ligands in Au nanocluster catalysis. Isotopic labeling experiments clearly indicate that CO oxidation on the Au₂₅(SR) ₁₈/CeO₂ rod catalyst proceeds predominantly via the redox mechanism where CeO₂ activates O₂ while CO is activated on the dethiolated gold sites. These results point to a double-edged sword role played by the thiolate ligands on Au₂₅ nanoclusters for CO oxidation.

Full text of DOI:10.1021/ja5018706

Article
pubs.acs.org/JACS
Thiolate Ligands as a Double-Edged Sword for CO Oxidation on CeO2
Supported Au25(SCH2CH2Ph)18 Nanoclusters
Zili Wu,*,†,‡ De-en Jiang,† Amanda K. P. Mann,† David R. Mullins,† Zhen-An Qiao,† Lawrence F. Allard,§
Chenjie Zeng,∥ Rongchao Jin,∥ and Steven H. Overbury*,†,‡
†Chemical Science Division, ‡Center for Nanophase Materials Sciences, and §Materials Science and Technology Division, Oak Ridge
National Laboratory, Oak Ridge, Tennessee 37831, United States
∥Department of Chemistry, Carnegie Mellon University, 4400 Fifth Avenue, Pittsburgh, Pennsylvania 15213, United States
S
* Supporting Information
ABSTRACT: The effect of thiolate ligands was explored on the
catalysis of CeO2 rod supported Au25(SR)18 (SR =
−SCH2CH2Ph) by using CO oxidation as a probe reaction.
Reaction kinetic tests, in situ IR and X-ray absorption spectros-
copy, and density functional theory (DFT) were employed to
understand how the thiolate ligands affect the nature of active
sites, activation of CO and O2, and reaction mechanism and
kinetics. The intact Au25(SR)18 on the CeO2 rod is found not able
to adsorb CO. Only when the thiolate ligands are partially removed, starting from the interface between Au25(SR)18 and CeO2 at
temperatures of 423 K and above, can the adsorption of CO be observed by IR. DFT calculations suggest that CO adsorbs
favorably on the exposed gold atoms. Accordingly, the CO oxidation light-off temperature shifts to lower temperature. Several
types of Au sites are probed by IR of CO adsorption during the ligand removal process. The cationic Au sites (charged between 0
and +1) are found to play the major role for low-temperature CO oxidation. Similar activation energies and reaction rates are
found for CO oxidation on differently treated Au25(SR)18/CeO2 rod catalysts, suggesting a simple site-blocking effect of the
thiolate ligands in Au nanocluster catalysis. Isotopic labeling experiments clearly indicate that CO oxidation on the Au25(SR)18/
CeO2 rod catalyst proceeds predominantly via the redox mechanism where CeO2 activates O2 while CO is activated on the
dethiolated gold sites. These results point to a double-edged sword role played by the thiolate ligands on Au25 nanoclusters for
CO oxidation.
Aun(SR)m nanoclusters for gas-phase reactions has just started
to emerge.17−20
1. INTRODUCTION
Amidst the already highly active field of heterogeneous gold
catalysis based on nanosized gold particles,1,2 gold nanoclusters
with a specific number of gold atoms have recently added
additional spotlights into this area and opened up a new era for
gold catalysis. It was shown by several research groups3−12 that
gold nanoclusters containing specific numbers of gold atoms
exhibit intrinsically unique and size-specific catalytic behavior
that is different from that of either atomically dispersed gold
species or the traditional nanoparticle counterpart in a variety
of reactions. Among gold nanoclusters, the thiolate ligand-
protected gold nanoclusters (less than 2 nm in diameter),
referred to as Aun(SR)m, where n and m denote the numbers of
gold atoms and thiolate ligands SR (−SCH2CH2Ph),
respectively, have particular potential for shedding light onto
the long puzzling area of gold catalysis thanks to their atomic
precision and monodispersity.13−16 Although research on the
reactivity of Aun(SR)m nanoclusters is still in its infancy,
interesting catalytic properties of Aun(SR)m nanoclusters have
been demonstrated in various types of reactions, including
oxidation, hydrogenation, and coupling.6,11 Most of the studies
were conducted for liquid-phase organic reactions on free or
supported gold nanoclusters, while research utilizing supported
Interestingly, Aun(SR)m nanoclusters are catalytically active
and selective in many reactions in the liquid phase even though
the surface gold atoms are coordinated with thiolate ligands,
which are generally considered to be poisons in traditional
metal catalysis.21−23 Recently it was found that the removal of
thiolate ligands from Au25 supported on carbon was beneficial
for the liquid-phase reduction of 4-nitrophenol.24 Similarly, an
alkanethiol coating of a self-assembled monolayer (SAM) on a
Pd film, closely related to the thiolate ligand protection of metal
clusters, was found to decrease activity by 60% in the selective
hydrogenation of unsaturated epoxides to saturated epoxides.25
Considering that the alkanethiol SAM generally leaves some
fraction of surface metal atoms uncoordinated while all the
surface Au atoms are coordinated with thiolate ligands in the
Aun(SR)m nanoclusters, one has to wonder if there are sites
available on the surface of Aun(SR)m nanoclusters for
adsorption and conversion of reactants. Quintanilla et al.
found that adsorption of a thiol SAM on 10% of the sites of
supported Au catalysts negatively affects the reactivity for
Received: February 21, 2014
Published: April 4, 2014
© 2014 American Chemical Society
6111
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
oxidation or hydrogenation reactions.22 For thiolate ligand
protected Au13 nanoclusters supported on TiO2, the low-
temperature CO oxidation activity was obtained only after
thermal removal of the thiolate ligands.23 Therefore, it is our
primary interest to investigate the thiolate ligand effect on the
catalysis of supported Aun(SR)m nanoclusters with focus on gas-
phase reactions.
Finally the sample was dried in a vacuum oven at room temperature,
giving the as-synthesized catalyst. The Au loading was determined to
be ∼0.15 wt % via X-ray absorption measurement in transmission
mode. Electron micrographs of the bare CeO2 rods and of the
Au25(SR)18/CeO2 rod catalyst were recorded on a JEOL 2200FS FEG
200 kV scanning transmission electron microscope fitted with a CEOS
GmbH (Heidelberg, Germany) aberration corrector on the probe-
forming lenses. With a beam convergence semiangle of 26.5 mrad at a
nominal current of 23 pA, high-angle annular dark-field (HAADF)
images can be acquired at a nominal resolution in the 0.07 nm range. A
standard 1 wt % Au/TiO2 catalyst was obtained from Mintek (catalog
no. 79-0165, AUROlite), designated as Mintek-Au.
2.2. CO Oxidation Test. CO oxidation over the Au25/(SR)18/
CeO2 rod sample was tested on a plug-flow, temperature-controlled
microreactor system (Altamira AMI 200). The sample (ca. 32 mg),
loaded into a U-shaped quartz tube (4 mm i.d.) and supported by
quartz wool, was pretreated sequentially in flowing 5% O2/He (25
mL/min) at different temperatures (423, 448, 473, 523, 573, and 673
K) for 1 h and cooled to room temperature in an O2/He atmosphere.
These treated samples were denoted as AuCe-x, where x represents
the O2-treatment temperature. For example, AuCe-423 refers to the
Au25(SR)18/CeO2 rod sample pretreated at 423 K in O2 for 1 h. The
sample was then exposed to the reaction mixture (5 mL/min 5% O2/
He + 5 mL/min 2% CO/2% Ar/He) at room temperature with a
space velocity of 18 800 cm3/(h·gcat). The system was ramped up to
413 K at a rate of 3 K/min to measure the CO light-off process. The
product gas stream was analyzed with an online quadruple mass
spectrometer (QMS) (OmniStar GSD-301 O2, Pfeiffer Vacuum).
QMS responses were calibrated with known mixtures of CO and CO2
at different ratios. All gases, unless specified, were provided by Air
Liquide with UHP helium as the balance gas, with a moisture content
of less than 1 ppm.
CO−O2 copulse and sequential pulse experiments were manually
conducted in the AMI microreactor system at 353 K using a pulse
volume of 0.5 mL on the AuCe-x samples after different temperature
O2 pretreatments. In the copulse mode, an equal mixture of 2% CO/
2% Ar/He + 5% O2/He was pulsed repeatedly to the sample, with
delays between pulses sufficient for the QMS signals of CO and CO2
to return to baseline. In the sequential pulse mode, a 2% CO/Ar/He
(10 mL/min) + He (10 mL/min) mixture was first pulsed over the
sample and then a 5% O2/He (10 mL/min) + He (10 mL/min)
mixture was pulsed over the sample after the QMS signals of CO and
CO2 traces leveled off. Both pulse modes were repeated nine more
times.
2.3. In Situ IR Spectroscopy. IR spectra of CO adsorption were
collected using a Thermo Nicolet Nexus 670 spectrometer in diffuse
reflectance mode (DRIFTS), while the exiting stream was analyzed by
an OmniStar GSD-301 (Pfeiffer-Balzer) QMS. Since the removal of
thiolate ligands is temperature sensitive, the Au25(SR)18/CeO2 rod
sample was first calcined in a box oven in air at different temperatures
(295, 423, 473, 498, 523, 573, and 673 K) for 1 h to make the
temperature consistent with that used in the reactivity test in CO
oxidation on the AMI system. The box oven calcined sample was then
transferred to a DRIFTS cell (HC-900, Pike Technologies, nominal
cell volume of 6 cm3) and treated in 5% O2/He flow (25 mL/min) at
the same temperatures (the actual IR cell temperature is lower than its
set temperature) for another 30 min to remove adsorbed water and
carbonate species formed during the sample transfer in air. The sample
was then cooled to room temperature before switching to a He stream.
The IR background spectrum was then collected in He flow. CO
adsorption was conducted at room temperature in flow mode on the
different temperature treated samples. After 2% CO/2% Ar/He was
flowed for 10 min, either He or O2/He was switched back to allow
adsorbed CO to desorb or react. IR spectra were recorded
continuously to follow the surface changes during the adsorption
and desorption process. All reported IR spectra are difference spectra
referenced to the background spectrum collected at room temperature
after pretreatment but prior to CO adsorption.
To approach this goal, we select gas-phase CO oxidation as a
probe reaction for two reasons. First, CO oxidation has long
been utilized as a prototype reaction for understanding gold
catalysis, such as the role of the Au particle size, support, and
charge of the Au sites.1,26 Second, recent studies of CO
oxidation over supported Aun(SR)m nanoclusters seem to give
different views on how the ligands affect the reaction. For
example, TiO2 supported Au38(SR)24 nanoclusters did not
show high activity for CO oxidation unless the thiolate ligands
were removed,20 while CeO2 supported Au25(SR)18 and
17,18
Au38(SR)24
were found to be active for low-temperature
CO oxidation without the removal of any thiolate ligands. The
onset of CO oxidation activity was proposed to be dependent
on the O2-pretreatment conditions rather than the thiolate
ligand removal.17,18 The nature of the oxide support (e.g., CeO2
as opposed to TiO2) was also assumed to cause the
discrepancy,18 but the mechanism behind the interesting CO
oxidation behavior of CeO2 supported Aun(SR)m nanoclusters
remains a mystery.
In this study, we select CeO2 supported Au25 nanoclusters as
an example and are motivated to reveal the role of the thiolate
ligands on the CO oxidation reaction. Specifically, we use a
Au25(SR)18/CeO2 (SR = −SCH2CH2Ph; Ph = −C6H5) catalyst
to address the following questions: (i) What are the active sites
for CO and O2 activation? (ii) What is the reaction mechanism
for CO oxidation, i.e., Langmuir−Hinshelwood (L−H) vs
Mars−van Krevelen (MvK)? (iii) How does the presence of
thiolate ligands affect the reaction mechanism? Our results
clearly indicate that the thiolate ligands act as a double-edged
sword for the Au25 nanoclusters in CO oxidation: they stabilize
the structural integrity of Au25, yet they also block the Au sites
that are needed for CO activation. CO oxidation on
Au25(SR)18/CeO2 can take place only when some of the
thiolate ligands are removed from the interface between the Au
nanoclusters and ceria support. With respect to the reaction
mechanism, we show unambiguously that the reaction proceeds
predominantly via the MvK mechanism where CeO2 activates
O2 while CO is activated on the dethiolated gold sites and
oxidized by the CeO2 lattice oxygen.
2. EXPERIMENTAL SECTION
2.1. Materials Synthesis. Au25(SR)18 nanoclusters were synthe-
sized using the method reported previously.27 The synthesized
Au25(SR)18 nanoclusters were characterized by both UV−vis
absorption and matrix-assisted laser desorption ionization mass
spectrometry (MALDI-MS). The UV−vis spectrum was collected on
a Varian Cary 5000 spectrometer in the range of 300−800 nm.
MALDI-MS was performed with a PerSeptive Biosystems Voyager DE
super-STR time-of-flight (TOF) mass spectrometer. Cerium oxide in a
rodlike morphology was chosen as the support for Au25(SR)18
nanoclusters. The synthesis of ceria rods was detailed in our recent
work by using a hydrothermal method, and the rods were calcined at
673 K for 4 h prior to Au cluster impregnation.28,29 For preparation of
the Au25(SR)18/CeO2 rod catalyst, 2 mg of Au25(SR)18 was dissolved
in 5 mL of toluene, and 200 mg of CeO2 rods was added. After the
solution was stirred for 15 h at room temperature, the solid was
collected by centrifugation and washed with toluene three times.
2.4. In Situ Raman Spectroscopy. The Au25(SR)18/CeO2 rod
sample was pretreated in a Raman catalytic reactor (Linkam
6112
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
CCR1000) in flowing 5% O2/He (30 mL/min) at 423 and 473 K for 1
h and cooled to room temperature. Raman spectra were then collected
to look for adsorbed O2 species. Raman scattering was collected via a
customized ellipsoidal mirror and directed by a fiber optics bundle
(Princeton Instruments) to the spectrograph stage of a triple Raman
spectrometer (Princeton Instruments Acton Trivista 555).30 Edge
filters (Semrock) were used in front of the UV−vis fiber optic bundle
to block the laser irradiation. The 532 nm excitation (20 mW at the
sample) was emitted from a solid-state laser (Princeton Scientific,
MSL 532-50). A UV-enhanced liquid N2-cooled charge-coupled device
(CCD) detector (Princeton Instruments) was employed for signal
detection. The Raman reactor sits on an XY stage (Prior Scientific,
OptiScan XY system) and translates in raster mode while the spectrum
is collected. The fast translation has been shown able to eliminate/
minimize any laser damage of the samples. Cyclohexane was used as a
standard for the calibration of the Raman shifts.
Figure 1. (A) UV−vis spectrum and (B) MS pattern of the as-
synthesized Au25(SR)18 nanoclusters. (C) HAADF-STEM image of
the as-synthesized Au25(SR)18/CeO2 rod catalyst. The scale bar
represents 2 nm.
2.5. In Situ X-ray Absorption Spectroscopy (XAS). The XAS
data were recorded at the Au LIII-edge (11 919 eV) at beamline X19a
at the National Synchrotron Light Source, Brookhaven National
Laboratory. A Si(111) double-crystal monochromator was used and
detuned by 20% to reject higher harmonics. The X-ray absorption was
measured simultaneously in transmission and fluorescence. Ion
chambers for measuring Io and It were filled with nitrogen and Ar,
respectively. It should be noted, however, that the ceria support
absorbed more than 99% of the X-ray radiation. Fluorescence was
measured using a four-element silicon drift detector, model Vortex-
ME4, oriented perpendicular to the upcoming beam.31 This
multielement detector not only provided enhanced sensitivity but
was also energy dispersive, which enabled discrimination between the
Au fluorescence and the much more intense Ce background. The Au
absorption was measured out to k = 16.
The starting Au25(SR)18/CeO2 rod sample and the Mintek-Au
reference catalyst were ground to fine powders, mixed with BN, and
pressed into a 13 mm diameter pellet. The typical absorbance of the
analyte, μ(x), was less than 0.1. The pellets were mounted in a
Nashner−Adler reaction cell.32 Both X-ray absorption near edge
structure (XANES) and extended X-ray absorption fine structure
(EXAFS) spectra were collected on the sample as received and after it
was heated in 20% O2/He at 10 K min−1 to a sequence of increasing
temperatures and held for 30 min at each temperature. Following each
heating cycle, the sample was cooled to 173 K for EXAFS
measurements after being purged with He. A liquid nitrogen dewar
was placed on the top of the cell to enable absorption measurements at
low temperature and, consequently, minimize thermal disorder effects.
The programs ATHENA and ARTEMIS (available through the
DEMETER software package, version 0.9.17) were used to reduce and
fit the data, respectively.33 Data reduction consisted of pre-edge
subtraction, background determination, normalization, and spectral
averaging.
previous work.27 The single m/z peak at 7391 in the mass
spectrometry analysis corresponds closely to that of molecular
Au25(SCH2CH2Ph)18 (calculated to be 7394; the deviation of 3
is due to gross calibration of the mass spectrometer),
confirming that the as-prepared nanoclusters are Au25(SR)18
with atomic monodispersity. We conclude from the two
analysis results that atomically precise Au25(SR)18 nanoclusters
were successfully synthesized. An HAADF-STEM (STEM =
scanning tunneling electron microscopy image) of the as-
synthesized Au25(SR)18/CeO2 sample is shown in Figure 1C,
where an isolated Au particle is seen on the CeO2 rod. The size
of ∼1.2 nm is close to the X-ray crystallographically measured
value of a Au25(SR)18 nanocluster. To better illustrate the
successful synthesis and dispersion of the Au25(SR)18 nano-
clusters on CeO2 rods, a high-load (∼1 wt %) Au25(SR)18/
CeO2 sample was prepared and characterized also by HAADF-
STEM imaging. Those results are shown in Figure S1 in the
Supporting Information. Considering the fact that the CeO2
rods have a large amount of defect sites28,29 and the loading of
Au25(SR)18 nanoclusters is very low, we can reasonably expect
that the rod support may be helpful to anchor the gold
nanoclusters. Furthermore, the cluster agglomeration could be
prevented during thermal treatment because the defects in ceria
were found helpful for strong bonding with metal nano-
particles.37 This is supported by the X-ray absorption
spectroscopy results shown in a later section.
3.2. CO Oxidation on the Au25(SR)18/CeO2 Rod
Catalyst. Figure 2A exhibits the light-off curves for CO
oxidation over different temperature O2-treated Au25(SR)18/
CeO2 rod catalysts. The observed CO oxidation activity in the
tested temperature range is due to the presence of gold
nanoclusters since the bare CeO2 rods show activity only at
temperatures above 423 K.28 There is barely any CO oxidation
activity on the as-synthesized sample. The reactivity starts to
emerge only when the sample is O2-treated at elevated
temperatures. The 423 K O2 treatment results in weak CO
oxidation activity. Activity increases dramatically when the
treatment temperature is increased to 448 K. The activity keeps
increasing with the O2-treatment temperature up to 523 K and
then declines with further higher temperature treatment. For
example, the CO conversion at 330 K is 1.7%, 24.9%, 36.3%,
39.1%, 35.3%, and 15.1% for O2-treatment temperatures of 423,
448, 473, 523 K, 573, and 673 K, respectively. The observed
trend is in general agreement with the CeO2 nanoparticle
supported Au25(SR)18 for CO oxidation reported by Nie et al.18
Figure 2B shows the Arrhenius plots from CO oxidation in
the temperature range of 308−363 K with conversion below
2.6. Density Functional Theory (DFT) of CO Adsorption.
Parallel, resolution-of-identity DFT calculations were performed for
CO adsorption on intact and partially dethiolated Au25(SR)18 with the
quantum chemistry program Turbomole, version 6.5.34 Structural
optimization was done with the PBE0 functional35 for electron
exchange and correlation and the def2-SV(P) basis sets. Effective core
potentials which have 19 valence electrons and include scalar
relativistic corrections were used for Au.36 The adsorption energy,
Ead, is defined as Ead = ECO/Au − EAu − ECO, where ECO/Au, EAu, and
ECO are the energies of the adsorbed system, the gold cluster, and the
isolated CO molecule, respectively, so a negative Ead indicates a
favorable interaction.
3. RESULTS AND DISCUSSION
3.1. Characterization of Au25(SR)18 Nanoclusters and
the Au25(SR)18/CeO2 Rod Catalyst. The pure nanoclusters
were characterized by both UV−vis absorption (Figure 1A) and
mass spectrometry (Figure 1B) analysis. The UV−vis spectrum
exhibits three absorption bands at 400, 447, and 683 nm,
similar to that reported for pure Au25(SR)18 nanoclusters in
6113
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
measured by both in situ IR spectroscopy (for monitoring the
C−H moiety of the thiolates) and EXAFS (for monitoring the
interfacial Au−S bond). C−H stretching modes are observed at
2935 and 2844 cm−1 in the IR spectra of the Au25(SR)18/CeO2
rod catalyst pretreated at temperature below 523 K (Figure
3A). These bands are due to the C−H stretching in the
Figure 2. (A) CO oxidation light-off curves for the Au25(SR)18/CeO2
rod catalyst pretreated in O2 at different temperatures. The space
velocity is 18 800 cm3/(h·gcat). (B) Arrhenius plots from the CO
oxidation data in (A) for conversions below 20%. For the AuCe-423
sample, separate CO oxidation tests were run at lower space velocity to
obtain reasonable CO conversion at lower temperatures (see Figure
S2, Supporting Information).
20% on the Au25(SR)18/CeO2 rod catalyst pretreated at
different temperatures in O2. The number of Au sites used
for the turnover frequency (TOF) was calculated in the
following way. The maximum possible number of surface Au
sites was approximated as being 50% of the total number of
gold atoms, since about half of the Au atoms are at the surface
of the core−shell structured Au25(SR)18 nanocluster.6 From
this maximum possible number of sites, the number of
accessible sites for a given sample was then calculated on the
basis of the CO adsorption observed during the IR measure-
ments described below. The activation energy derived from
these plots is 52.5 5 kJ/mol for all the samples pretreated at
different temperatures, indicating that CO oxidation follows a
similar reaction mechanism over the differently treated
catalysts. The activation energy compared well with the value
of ∼50 kJ/mol reported for CO oxidation on ceria supported
Au catalysts.38,39 In general, the plots show that the reaction
rate (TOF) is about the same for the different temperature
Figure 3. (A) IR background spectra of the Au25(SR)18/CeO2 rod
sample pretreated at different temperatures in O2 and (B) EXAFS
spectra of various Au samples and the Au25(SR)18/CeO2 rod sample
pretreated at different temperatures in O2. Spectra from Au foil and
the Mintek Au/TiO2 catalyst are also shown as references.
aliphatic chain of the SCH2CH2Ph ligands. The absence of
aromatic C−H stretching, expected between 3030 and 3100
cm−1, is intriguing but was also recently noticed in the IR
spectrum of unsupported Au25(SR)18 nanoclusters.41 The IR
feature at 2724 cm−1 is likely an overtone band of the C−H
deformation mode of the aliphatic chain and thus can also be
used to monitor the presence of the thiolate ligands.
Interestingly, these characteristic IR features of the thiolate
ligands are not prominent on the room temperature treated
sample (the rather broad features in the 3000−2800 cm−1
region are due to instrument background), probably because
they are obscured by adsorbed water on the ceria surface. These
IR bands are most intense after 423 K treatment, then gradually
decrease in intensity, and finally disappear at a treatment
temperature of 523 K and above, implying complete removal of
the thiolate ligands from the Au nanoclusters at this
temperature. The temperature of 523 K is in good agreement
with that for complete thiolate loss from the thermogravimetric
analysis (TGA) result of bare and supported Au25(SR)18
nanoclusters.18,42
treated samples, e.g., around 0.5
0.1 s−1 at a reaction
temperature of 330 K. For perspective, this value can be
compared with that of the best Au/CeO2 catalyst, where the
TOF at 278 K was calculated to be 0.49 s−1 on the basis of the
number of Au atoms at the interface between Au nanoparticles
and nanocrystalline CeO2.40
3.3. Temperature-Dependent Removal of Thiolate
Ligands. The removal of thiolate ligands from the Au25(SR)18/
CeO2 rod catalyst upon different temperature treatments was
6114
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
The Au−S bond cleavage, a direct indicator of dethiolation,
was followed by in situ EXAFS. The k3-weighted EXAFS
spectra are shown in Figure 3B after the Au25(SR)18/CeO2 rod
catalyst was heated in 20% O2/He at increasing temperature.
EXAFS spectra of Au foil and Mintek Au/TiO2 are shown as
references. The EXAFS spectrum from the as-synthesized
Au25(SR)18/CeO2 rod catalyst is very similar to that of
unsupported Au25(SR)18 nanoclusters.43 The prominent peak
near 0.19 nm corresponds to Au−S nearest neighbors with an
average Au−S distance of 0.231 nm and a coordination number
of ca. 1. A similar average Au−S distance is reported for pure
Au25(SR)18 nanoclusters by Simms et al.43 Theoretically, a
Au25(SR)18 has three distinct Au−Au nearest neighbor
distances associated with the core, surface, and S-bonded
Au.44 Due to the low loading (0.15 wt %) of Au25(SR)18, our
data did not enable us to reliably fit the EXAFS to these three
shells; however, peaks associated with scattering from Au
neighbors are evident between 0.2 and 0.3 nm.
Figure 4. IR spectra of CO adsorption at room temperature on the
Au25(SR)18/CeO2 rod sample pretreated in O2 at different temper-
atures.
After heating to 423 K, the intensity of the Au−S peak barely
changes, implying no obvious loss of the thiolate ligands.
However, Au−Au peaks in the 0.2−0.3 nm range show changes
indicative of a structural rearrangement of the Au25 nano-
clusters. Evidently, the structure of Au25 may change even after
a slight loss of thiolate ligands. After heating at 448 K, the Au−
S peak intensity decreases slightly, followed by a greater
decrease after heating at 473 K. A slight shift of the Au−S peak
to lower distance is evident for the AuCe-448 sample, implying
a stronger Au−S bond compared to that of the as-synthesized
sample. This observation can be well substantiated by a
previous NMR study of the thermal stability of unsupported
thiolated Au25 nanoclusters.45 It was shown that there exist two
types of Au−S binding modes with different thermal stabilities
in thiolated Au25 nanoclusters. The Au−S bonds with lower
thermal stability can be broken at a treatment temperature
above 433 K, a temperature in good agreement with that in our
EXAFS study. The remaining Au−S bonds are thermally more
stable and thus give shorter Au−S distances as implied by the
AuCe-448 EXAFS spectrum. The Au−S peak disappears upon
heating at 498 K, indicating no thiolate ligands left over on the
Au. This is in agreement with the loss of the C−H stretching
modes as observed in the IR spectra after heating to 523 K
(Figure 3A). There is little structure evident in the Au−Au
region, and the Au−Au EXAFS peaks do not show an intensity
increase until the treatment temperature is raised to 573 K.
Interestingly, the Au−Au peaks in the 0.2−0.3 nm range are
quite similar for the 673 K treated sample and the as-
synthesized one. A fitting of the spectrum gives a coordination
number of 8 for Au, corresponding to a Au particle size of 1−
1.5 nm,46 which is much smaller than that of the Mintek Au/
TiO2 reference sample (∼2.9 1.7 nm).47 It appears that the
Au nanoclusters do not increase significantly in size even after
the complete removal of thiolate ligands and high-temperature
treatment. A strong interaction must exist between the Au
nanoclusters and CeO2 rods to prevent their agglomeration at
high temperatures such as 673 K. This point will be elaborated
more in the following IR section.
the presence of thiolate ligands prevents CO from adsorbing on
supported Au25(SR)18, although the computed structural model
of Au25(SR)18 exhibits accessible Au sites.42 This absence of CO
adsorption reasonably explains the negligible CO oxidation
activity on the as-synthesized sample.
O2 treatment at 423 K leads to a very weak IR band at 2150
cm−1 from CO adsorption on the Au25(SR)18/CeO2 rod
catalyst, indicative of slight removal of thiolate ligands. After O2
treatment at 448 K, a new IR band is observed at 2122 cm−1 in
addition to the band at 2150 cm−1. The band at 2122 cm−1
gradually shifts to 2117 cm−1 with increasing O2-pretreatment
temperature and maximizes in intensity after 498 K pretreat-
ment, while a higher wavenumber band is observed at 2157
cm−1. The IR bands in the range of 2150−2117 cm−1 can be
attributed to CO adsorbed on differently charged Au (Auδ+, 0 <
δ < 1) sites,40,48−53 and the CO-band wavenumber decreases
with decreasing positive charge on the Au sites. This red-shift
trend is in line with the continuous dethiolation upon thermal
treatment since the thiolate ligands are electron acceptors that
render the surface Au atoms electron deficient.6 The adsorbed
CO molecules associated with the IR band at 2157 cm−1 are
more resistant to conversion to CO2 at room temperature than
those associated with IR bands below 2150 cm−1 as discussed
below, so we attribute the IR band at 2157 cm−1 to CO
adsorbed on Au+ which was considered active for CO oxidation
only at elevated temperatures.54 A shoulder band observed at
2177 cm−1 is assigned to adsorbed CO on coordinatively
unsaturated Ce4+ sites created from the thermal dehydration of
the ceria support.28 Besides the relatively high wavenumber,
this assignment is based on its increase in intensity after higher
temperature pretreatment (due to further dehydration of the
ceria) and its prompt loss during room temperature helium
purging.55 The weak IR band at 2076 cm−1 can be ascribed to
CO adsorbed on negatively charged Au sites (Auδ−, 0 < δ <
1).52,56 After pretreatment at 573 and 673 K, the IR band at
2117 cm−1 due to CO adsorbed on nearly neutral Au sites
decreases in intensity and shifts to higher wavenumbers. Au+
and a small proportion of Auδ+ (0 < δ < 1) sites seem to exist
after 673 K O2 treatment. Meanwhile the band at 2177 cm−1
from CO adsorbed on Ce4+ sites evolves to dominate the
spectra, due to the increased dehydration degree of CeO2.
The presence of cationic Au sites on the Au25(SR)18/CeO2
rod catalyst is supported by the in situ XANES spectra collected
3.4. Active Sites for CO Adsorption. CO adsorption on
the Au25(SR)18/CeO2 rod catalyst pretreated in O2 at different
temperatures was followed with in situ DRIFTS in an attempt
to reveal the nature of Au sites. As shown in Figure 4, there is
no IR band from CO adsorption on the as-synthesized sample,
an indication that there are no Au sites available for CO
adsorption on an intact Au25(SR)18 nanocluster. It appears that
6115
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
after different temperature O2 treatments (see Figure S3 in the
Supporting Information). The intensity in the “white line”
region between 11 920 and 11 925 eV is higher for all the
Au25(SR)18/CeO2 rod samples than the two gold reference
samples, indicating the presence of positively charged Au
species, consistent with the observation from IR study of CO
adsorption. The white line is more evident by subtracting the
spectrum for Au foil from those of the AuCe-x samples (see
Figure S3B), but overall, the white line intensity is much
weaker than that of Au3+ and Au+ species,4 implying that the
majority of Au sites in the nanoclusters are in the metallic state.
Upon heating at 573 and 673 K, the white line appears to shift
slightly to lower photon energy. This shift could be due to a
change in electronic structure (e.g., the increase of Au+ sites as
indicated by IR of CO adsorption).
To reconcile the seemingly inconsistent results from IR and
XANES studies of the 673 K treated Au25(SR)18/CeO2 rod
sample, i.e., the dominance of Au+ sites from IR spectra vs only
a small amount of cationic Au sites from XANES spectra, we
propose that the surface of the Au nanoclusters is covered by
only a small fraction of oxidized Au (+1 charged) which can
adsorb CO. It appears that the decoration of these Au+ sites
diminishes the CO adsorption capability of the rest of the
metallic Au sites. When the AuCe-673 sample is further treated
with H2 at 473 K, reappearance and dominance of the Auδ+ (0
< δ < 1) sites on the AuCe-673 sample are evidenced by CO
adsorption as exhibited in Figure S4 (Supporting Information).
Thus, the reduction of the decorative Au+ sites can almost
restore the CO adsorption capability of all the Au sites. The
near recovery of the CO adsorption bands also indicates no
apparent aggregation of the Au nanoclusters during 673 K
treatment, consistent with the above EXAFS results.
It is interesting that various types of Au sites, as indicated by
the CO IR, are present on the surface of partially and fully
dethiolated Au25(SR)18/CeO2 rod catalysts since one may
expect a single type of Au site would be created upon
dethiolation of the well-structured Au25(SR)18 nanoclusters. We
believe the presence of these sites is related to the strong
interaction between Au25 nanoclusters and the ceria rod, which
leads to a structural change of the Au25 nanoclusters upon the
loss of thiolate ligands. This is substantiated by the EXAFS
result shown in Figure 3B which shows structural rearrange-
ment of the Au nanoclusters upon dethiolation without obvious
size growth. The strong interaction between the Au nano-
clusters and the ceria rods could result in charge transfer
between the two. This might be responsible for the absence of
completely metallic Au sites for CO adsorption even after full
dethiolation (such as the AuCe-498 and AuCe-523 samples).
The Au-to-ceria charge transfer is not unusual and has been
previously predicted theoretically and observed experimen-
tally.38,57−60
The presence of different types of surface Au sites makes it
difficult to determine the real TOF for the Au25(SR)18/CeO2
rod catalyst in CO oxidation. However, since CO adsorption on
Au is a key step for its subsequent oxidation reaction, via either
the MvK or L−H channel, it is sensible that the calculated TOF
based on surface Au sites accessible for CO activation shall offer
a good unifying scale to compare the activity of differently
dethiolated Au25(SR)18/CeO2 rod catalysts for CO oxidation.
The integrated area of IR bands from CO adsorbed on all
types of Au sites is plotted in Figure S5 (Supporting
Information) as a function of the O2-treatment temperature.
The IR bands’ area reaches its maximum at 498 K treatment
and then decreases at higher treatment temperatures. The
increase in the IR bands’ area is due to the increased exposure
of surface Au sites via thiolate ligand removal at temperatures
below 498 K. At treatment temperatures above 498 K, the
decrease is attributed to the decrease of CO-adsorbing Au sites
because of the decoration by oxidized Au sites. The treatment
temperature leading to the maximized IR band area coincides
nicely with that for the complete removal of the thiolate ligands
on the Au25(SR)18 nanoclusters as observed by in situ IR and
EXAFS spectroscopy data (Figure 3A,B) and previous TGA of
bare Au25(SR)18 nanoclusters.18 Assuming that the IR
absorption coefficients of CO adsorbed on the differently
charged Au sites are similar, the percentage of CO-adsorbing
Au sites exposed at specific treatment temperatures can be
estimated on the basis of the IR bands’ area normalized to that
from 498 K treatment. This is also plotted in Figure S5 as a
function of the pretreatment temperature. Thus, the obtained
Au exposure percentage is used to calculate the turnover
frequency of CO oxidation for different temperature treated
Au25(SR)18/CeO2 rod catalysts, already shown in Figure 2B.
The TOF calculation is based on the assumptions that about
half (13 of 25)6 of the Au atoms are on the surface of Au25
nanoclusters and all the CO-accessible surface Au atoms are
active once dethiolated. We point out that the thus calculated
TOFs are the minimum bound since not all the surface Au
atoms are active.
The total capacity of CO adsorption correlates well with the
CO oxidation activity as depicted in Figure S5 (Supporting
Information). The sample starts to show low-temperature CO
oxidation activity and an IR band of adsorbed CO on Au sites
only after 423 K O2 pretreatment. Furthermore, the larger the
CO−Au IR band intensity (i.e., the percentage of active surface
Au sites exposed), the higher the activity for CO oxidation over
the sample. When the CO−Au band intensity decreases, the
CO oxidation activity also decreases. The nice correlation of
the two observations suggests that the Au25(SR)18/CeO2 rod
catalyst is active for CO oxidation only when some of the Au
atoms can adsorb/activate CO.
The interaction between CO and the Au25(SR)18/CeO2 rod
catalyst was simulated by DFT using Au25(SCH3)18− as a model
without including the CeO2 support for simplicity. For the
intact Au25(SR)18− cluster, we found that the adsorption energy
of CO is about −0.08 eV. This is a rather weak physisorption.
Given the entropy loss during adsorption, the adsorption free
energy is in fact positive at ambient conditions for such a weak
interaction. In other words, DFT predicts that CO would not
−
adsorb on the intact Au25(SCH3)18 cluster, in agreement with
our IR spectrum for the fresh Au25(SR)18/CeO2 rod sample
which shows no sign of CO adsorption (Figure 4). To simulate
the activation of the sample at an intermediate temperature
such as 423−448 K, we hypothesize that a small number of
thiolates are lost and some gold atoms are now exposed. Here
we removed three thiolates from the same dimeric staple motif
and then optimized the structure of the resulting Au25(SCH3)15
cluster. The relaxed geometry is shown in Figure 5A. One can
see that the two dethiolated gold atoms now come together
(with a distance of 2.9 Å); the rest of the cluster maintains the
original structure. Next we examined CO adsorption on one of
the dethiolated Au atoms (Figure 5B) and found that the
interaction is now much stronger with an adsorption energy of
−1.12 eV and the C−Au distance is 1.95 Å, both indicating
chemisorption. In the adsorbed state, the C−O bond length (at
6116
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
more positively charged the Au species are, the less active they
are for low-temperature CO oxidation.48,49,53,54,56 The trend
here is different from that of the previous study of Au
61
nanoparticles supported on nanocrystalline CeO2 where
cationic Au species were found more active than metallic Au
for CO oxidation.
It is also essential to follow the nature of the Au sites during
CO oxidation catalysis. Au25(SR)18/CeO2 rod samples pre-
treated at two different temperatures by O2 were investigated
for CO adsorption before, during, and after CO oxidation.
Figure S6 (Supporting Information) compares such IR spectra
for 423 and 498 K O2-treated samples. In both cases, the
frequency of adsorbed CO does not change before and after
CO oxidation, indicating that the nature of Au sites is not
altered after CO oxidation. During CO oxidation at 373 K, no
obvious CO−Au band can be observed for the 423 K treated
sample while an IR band due to adsorbed CO is observed at a
frequency slightly higher than 2117 cm−1 on the 498 K treated
sample. These observations suggest that the partially charged
Au sites are indeed the active sites for CO oxidation on both
partially and fully dethiolated Au25/CeO2 rod samples.
3.6. Active Sites for O2 Adsorption. Raman spectroscopy
(laser excitation at 532 nm) was employed to detect if there is
any adsorbed O2 species generated from the O2 pretreatment of
the Au25(SR)18/CeO2 rod sample. As shown in Figure S7
(Supporting Information), there are no observable Raman
bands in the spectral region of 800−1600 cm−1 expected from
adsorbed O2 species29,61 at room temperature on the samples
that are O2-pretreated at 423 or 473 K, indicating that either no
special adsorbed O2 species is produced from the O2
pretreatment or its amount is below the detection limit of
Raman spectroscopy. O2 flow or CO oxidation at room
temperature over the two differently treated samples gives
Raman spectra similar to those in Figure S7; i.e., no adsorbed
oxygen species is observable. Thus, we find no evidence for the
presence of adsorbed O2 species for CO oxidation over
dethiolated Au25(SR)18/CeO2 rod samples.
Figure 5. (A) Optimized structure of the Au25(SCH3)15 cluster. (B)
CO adsorption on the Au25(SCH3)15 cluster. Key: Au, orange; S, blue;
C, green; H, light gray; O, pink. The two dethiolated gold atoms are
highlighted in dark gray.
1.134 Å) is slightly increased in comparison with the computed
value for the gaseous CO (at 1.128 Å).
3.5. Active Sites for CO Oxidation. To reveal which type
of Au sites (Au+, Auδ+ (0 < δ < 1), and Auδ− (0 < δ < 1)) are
the active sites for CO oxidation on the O2-pretreated
Au25(SR)18/CeO2 rod catalyst, an IR experiment was done as
follows: CO was adsorbed on the 498 K O2-treated sample at
room temperature, and then desorption was conducted in
flowing O2 to follow the intensity change of the IR bands at
2078 (CO−Auδ−, 0 < δ < 1), 2117 (CO−Auδ+, 0 < δ < 1), and
2158 (CO−Au+) cm−1. As shown in Figure 6, the band at 2117
3.7. CO Oxidation Mechanism. Several isotopic labeling
experiments were carried out to reveal the CO oxidation
mechanism. We choose to illustrate this with a Au25(SR)18/
CeO2 rod sample that was pretreated in O2 at 423 K because
thus treated sample still has most of the thiolate ligands
attached.
3.7.1. CO Oxidation with 18O2. The sample was pretreated
at 423 K and cooled to room temperature in 16O2. Then a CO
+
Figure 6. IR spectra obtained following adsorption of CO at 295 K
onto the AuCe-498 sample. Spectra are shown as a function of time
after introduction of O2 flow at 295 K and then as temperature is
increased in the O2 flow.
18O2 light-off process was carried out after 16O2 was purged
out with He. The CO2 isomers, C16O16O, C16O18O, and
C18O18O, were monitored by the online QMS, and the profiles
are shown in Figure 7A as a function of temperature. The CO2
production is dominated by C16O16O in the temperature range
of 295−420 K. A small contribution from C16O18O is observed
only as the temperature rises above 360 K, while C18O18O
production is barely observable. The growing contribution from
C16O18O at higher temperature is due to the reaction of CO
with lattice 18O replenished by the gas-phase 18O2. The
observation of only C16O16O at the initial stage of the CO
oxidation light-off process and its dominance over the tested
temperature range indicate that CO reacts with either Ce16O2
lattice oxygen or adsorbed 16O2 generated from O2 pretreat-
ment at 423 K. We can rule out the latter, because of the above
Raman result (Figure S7, Supporting Information). The
following experiment also confirms that no adsorbed O2
cm−1 readily decreases in intensity and shifts to higher
wavenumbers with time; meanwhile CO2 production is
observed. The IR bands in the range of 2117−2140 cm−1
disappear after 10 min of O2/He flow at room temperature.
However, the bands at 2078 and 2158 cm−1 stay relatively
unchanged with time on stream at room temperature and only
disappear upon heating. The contrast clearly suggests that the
partially positively charged Au species (Auδ+ (0 < δ < 1)) are
the major active sites for CO oxidation on the Au25(SR)18/
CeO2 rod catalyst at low temperatures while the Au sites in +1
and −δ (0 < δ < 1) charged states can contribute to CO
oxidation only above room temperature. This is in general
agreement with other supported Au nanocatalysts where the
6117
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
Supporting Information). To our knowledge, this is the first
unambiguous experimental evidence for the operation of the
MvK mechanism in low-temperature CO oxidation on Au/CeO2
catalysts. This is well supported by recent DFT studies of CO
oxidation on a ceria supported Au system60,62−65 where the
ceria lattice oxygen is favorably involved in direct CO oxidation.
The MvK mechanism was demonstrated for CO oxidation on
CeO2 rod supported Au nanoparticles, but at much higher
temperature (573 K).66 For low-temperature (<423 K) CO
oxidation, only one study proposed the MvK mechanism for
Au/CeO2 where temporal analysis of products (TAP) was used
to reveal the nature of active oxygen species,67,68 but other
experimental studies suggested otherwise. Through an in situ
Raman spectroscopy study of CO oxidation on nanocrystalline
CeO2 supported Au nanoparticles, Guzman et al.61 concluded
that the gas-phase O2 was activated by the ceria support, i.e.,
adsorbed O2 species, for CO oxidation while the lattice O did
not participate in direct CO oxidation. Lee et al.69 studied Au
nanoparticles supported on ceria nanoshapes for CO oxidation
by monitoring the oxygen vacancy concentration change during
the reaction with Raman spectroscopy. They indicated that the
ceria lattice oxygen does not contribute to CO oxidation below
323 K.
Participation of lattice oxygen may be related to the
uniqueness of Au25 nanoclusters and their interaction with
the CeO2 rods, which leads to the activation of lattice oxygen of
CeO2 rods. CO oxidation on bare CeO2 rods also follows the
MvK mechanism,28 although at a much higher light-off
temperature (>423 K) than for the O2-treated Au25(SR)18/
CeO2 rod samples. The presence of Au25 nanoclusters
apparently promotes the reactivity of the lattice oxygen of
CeO2 rods. This is supported by the results from CO-TPR
(Figure S9, Supporting Information) of the different temper-
ature O2-treated Au25(SR)18/CeO2 rod samples. Although CO2
evolves only above 400 K from bare CeO2 rods in CO-TPR,28
CO2 production starts at 320 K or lower on the Au25(SR)18/
CeO2 rod samples pretreated at 423 K and above. The slight
production of CO2 above 340 K on the as-synthesized sample is
due to decomposition of native carbonate species adsorbed on
the CeO2 rods. In general, the more the thiolate ligands are
removed, the higher the reactivity of the lattice oxygen of ceria
rods. The production temperature of CO2 in the CO-TPR is in
general alignment with the trend of the CO oxidation light-off
process (Figure 2), a good indication that the lattice oxygen
participates in CO oxidation on a Au25(SR)18/CeO2 rod
sample. The increased reactivity of ceria lattice oxygen,
presumably at the interface of Au25 and CeO2, is attributed
to two benefits brought about by Au25(SR)18 nanoclusters:
facile activation of CO and generation of Ce3+ (oxygen
vacancies) via charge transfer from Au to ceria.38,57,60 Both
factors would result in enhanced reducibility of ceria28 and thus
the much lower light-off temperature for CO oxidation on the
Au25(SR)18/CeO2 rod catalyst. This type of Au-assisted MvK
mechanism was also suggested for CO oxidation on other
supported Au catalysts such as Au/TiO2,68,70 Au/FeOx,71 Au/
FePO4,53 etc.
Figure 7. Isotopic exchange experiments during CO oxidation on the
AuCe-423 sample: (A) CO + 18O2 on the 423 K 16O2-treated sample;
(B) CO + 16O2 on the 423 K 18O2-treated sample.
species resulted from the O2 pretreatment of the Au25(SR)18/
CeO2 rod sample at 423 K.
3.7.2. CO Oxidation on the 18O2-Pretreated Sample. The
as-synthesized Au25(SR)18/CeO2 rod sample was pretreated at
423 K and cooled to room temperature in 18O2. Then the CO +
16O2 light-off process was carried out after 18O2 was purged out
with He at room temperature. As shown in Figure 7B, CO2
production is solely the C16O16O isomer. The result clearly
indicates that the 18O2 pretreatment at 423 K does not lead to
any adsorbed oxygen species that would oxidize CO to CO2 at
low temperature on the Au25(SR)18/CeO2 rod catalyst.
3.7.3. Isotopic Switch during Steady-State CO Oxidation.
After the Au25(SR)18/CeO2 rod sample was pretreated at 423 K
with 16O2, CO oxidation was first carried out at 353 K with
18O2. As shown in Figure S8 (Supporting Information), a slight
delay of C16O18O production is observed relative to the
C16O16O production, with the former stabilizing with time on
stream while the latter undergoes a maximum and then
decreases. C18O18O is also seen at a further delayed time with
much less content than the other two isomers, whose
production is probably from the isotopic exchange between
lattice 18O and the other two CO2 isomers. After purging with
He and switching to CO oxidation with 16O2, the production of
all three CO2 isomers is observed simultaneously, with
C16O18O and C18O18O quickly diminishing with time on
stream.
3.8. Effect of Thiolate Ligands on the CO Oxidation
Mechanism. The effect of the thiolate ligands on the CO
oxidation pathways and reaction kinetics over the Au25(SR)18/
CeO2 rod catalyst is illustrated below.
3.8.1. Effect of Thiolate Ligands on CO oxidation
Pathways. To quantify how much the redox route contributes
to the CO oxidation over the Au25(SR)18/CeO2 rod catalyst, a
The observations from the isotopic exchange experiments
clearly imply that CO reacts with lattice oxygen of CeO2,
instead of directly with the cofed O2, to form CO2 starting from
room temperature; namely, CO oxidation proceeds via an MvK
mechanism. The cofed O2 enters into the reaction only after a
delay, due to activation of the O2 at lattice vacancies (Figure S8,
6118
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
CO−O2 copulse and sequential CO and CO2 pulses were
conducted at 353 K over the sample pretreated by O2 at
different temperatures. Due to the transient adsorption of O2,
CO can react only with the lattice oxygen in the sequential
pulse, while in the copulse mode CO can react with both lattice
oxygen and any possible adsorbed O2. Comparison of the two
sets of experiments should give an estimation of the
contribution of the redox route to the total CO oxidation. In
the experiments summarized in Figure 8, CO oxidation was
pulses from the copulse and sequential pulses of CO and O2
over the AuCe-473 sample. The CO2 amount from sequential
pulses is about 80% of that from the copulse, confirming the
dominant role of the redox mechanism in CO oxidation.
Quantification was also done on Au25(SR)18/CeO2 rod samples
pretreated at different temperatures with O2, and the result is
shown in Figure 8B. For all stages of dethiolation, the MvK
mechanism is responsible for about 80% of the CO oxidation
on the Au25(SR)18/CeO2 rod catalyst. Recent DFT stud-
ies60,63−65 of CO oxidation on ceria supported Au clusters
suggested that the MvK mechanism contributes significantly to
CO oxidation activity when the ceria is stepped or nanosized
with vacancies64 or terminated by (110) facets.63 This is
consistent with these experimental results since CeO2 rods are
highly defective and could be partially terminated by (110)
facets.29
It is notable that the contribution from the L−H mechanism
is the largest on the 523 K O2-pretreated sample, ca. 28%. Since
thiolate ligands are completely removed after this temperature
pretreatment and the Au nanoclusters do not change
apparently in size according to EXAFS results (Figure 3B), it
appears that the exposed surface Au sites may be able to
activate O2 and thus contribute to the L−H mechanism for CO
oxidation. This is in line with a recent DFT study of O2
activation on Au25(SR)18 where O2 is found to favorably bind
with Au atoms only when the thiolate ligands are partially
removed.72 On the basis of a DFT+U study of CO oxidation on
Au13 supported on CeO2,65 Kim et al. also showed that an L−H
pathway for CO oxidation can take place by coadsorbed CO
and O2 on the undercoordinated Au atoms. These authors also
suggested a new pathway where CO is oxidized by O2 adsorbed
on the Au−Ce3+ bridge site.65 No evidence is found for the
existence of this mechanism in our catalytic system, but its
contribution, if present, would be much smaller compared to
that of the MvK channel.
In the MvK mechanism, CO adsorbed on the Au sites of the
Au25(SR)18/CeO2 rod catalyst reacts with the lattice oxygen of
CeO2 to form CO2, while the gas-phase O2 replenishes the
consumed lattice oxygen. Considering that only less than 5% of
the surface Au sites are available for CO adsorption on the
sample after 423 K O2 pretreatment (Figure S5, Supporting
Information), these Au sites have to be around the perimeter
area so that the adsorbed CO can have a good chance to react
with the lattice oxygen of CeO2. Likewise, the active lattice
oxygen should be located at the perimeter area. Therefore, the
thiolate ligand removal most likely starts at the interface
between the Au25(SR)18 nanoclusters and the CeO2 support
and then progresses to the top surface of the Au nanoclusters.
This is reasonable because the highly reactive lattice oxygen of
ceria rods28 may assist the removal of the thiolate ligands of the
Au25(SR)18 nanoclusters at temperatures below the thermal
desorption point of the thiolate ligands from the gold
nanoclusters. As schematically shown in Scheme 1, it appears
Figure 8. (A) Amount of CO2 production during the copulse (black
squares) and sequential pulses (red squares) of CO and O2 at 353 K
on the Au25(SR)18/CeO2 rod sample pretreated at 473 K in O2 along
with their ratios (blue squares). (B) Percentage of MvK and L−H
reaction pathways on Au25(SR)18/CeO2 rod catalysts pretreated at
different temperatures with O2.
conducted on the samples at 353 K for an extended period until
the CO2 production was stabilized before the pulse experiments
were conducted. In this way, the catalyst surface is saturated
with carbonate species formed from CO/CO2 interaction with
ceria so that any CO2 produced from the pulses should leave
the catalyst surface and be monitored by the online mass
spectrometer. Figure 8A compares the CO2 production of 10
Scheme 1. CO Oxidation Mechanism on Intact, Partially and Fully Dethiolated Au25(SR)18/CeO2 Rod Catalysts
6119
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
that the removal of thiolate ligands controls the contribution of
the two reaction mechanisms. Initial removal permits the
operation of almost solely the MvK route at the triple-phase
boundary, thiolates, Au, and CeO2. Further removal of thiolates
opens up more surface Au sites which could participate in CO
oxidation via the L−H route65 in addition to the MvK route at
the interface. At higher treatment temperatures, the Au
nanoclusters may wet the ceria rods, and thus, the contribution
of MvK increases again.
adsorbing and further reaction. It would be interesting to
compare the same reaction but in both liquid- and gas-phase
conditions to understand how the thiolate ligands affect the
catalysis by Aun(SR)m nanoclusters. Au nanoclusters with
uncoordinated Au atoms, such as the recently reported
76
diphosphine-stabilized Au22 nanoclusters, are supposed to
be active for catalysis without the need to remove the ligands.
Research on such gold nanoclusters for gas-phase reactions is
under way in our laboratory.
Surface hydroxyls are known to contribute to CO oxidation
on supported Au nanoparticles as described in a most recent
review.73 Considering that the reaction mixtures, CO−O2−He,
contain a very low level of moisture (<1 ppm) in our work, we
believe that the contribution of hydroxyls for CO oxidation is
limited on the Au25(SR)18/CeO2 rod catalyst.
4. CONCLUSION
The effect of thiolate ligands on the catalysis of CeO2 rod
supported Au25(SR)18 has been probed by CO oxidation via a
reaction test, in situ IR and X-ray absorption (EXAFS and
XANES) spectroscopy, isotopic labeling, and DFT calculation.
It was concluded that the thiolate ligands act as a double-edged
sword for the Au25 nanoclusters for CO oxidation in the gas
phase. Specifically, the thiolate ligands maintain the structural
integrity of Au25(SR)18, but disable all Au sites from adsorbing
CO, and thus, the intact catalyst is not active for CO oxidation.
Partial removal of thiolate ligands, starting from the interface
between Au25(SR)18 and CeO2 at treatment temperatures of
423 K and above, is required for CO activation and thus the
onset of low-temperature CO oxidation on the Au25(SR)18/
CeO2 rod catalyst. The removal of thiolate ligands on the
Au25(SR)18/CeO2 rod catalyst results in three types of Au sites:
Auδ+ (0 < δ < 1), Au+, and Auδ− (0 < δ < 1). The former is
active for CO oxidation at low temperature, while the latter two
are active at elevated temperatures. The thiolate ligands seem to
behave simply as site-blocking agents, without impacting the
reactivity of the accessible Au sites. CO oxidation takes place
predominantly at the interface among the thiolate ligands, Au,
and CeO2 on the partially dethiolated Au25(SR)18/CeO2 rod
catalyst via the MvK mechanism. CO is activated by accessible
Au sites, while CeO2 activates O2, and the lattice O participates
in CO oxidation. The removal of thiolate ligands opens up an
additional minor channel for CO oxidation on the Au
nanoclusters supported on CeO2 rods via the L−H mechanism,
where both CO and O2 are activated by the exposed Au sites.
The results here provide fundamental implications for how the
ligand-protected Au nanoclusters can be further investigated as
efficient catalysts for gas-phase reactions.
3.8.2. Effect of Thiolate Ligands on CO Oxidation Kinetics.
Figure 2B shows that the activation energy and reaction rate
(TOF) of CO oxidation are not sensitive to the coverage of
thiolate ligands on the Au25(SR)18/CeO2 rod catalyst. As
described above, the TOF is calculated on the basis of the
assumption that the IR absorption coefficients from adsorbed
CO are similar on all Au sites. Actually, it has been shown that
the absorption coefficient of CO adsorbed on more positively
charged Au species is smaller than that of CO on less positively
charged and metallic Au because the π back-donation in CO−
Au0 enhances the IR band intensity more than the σ bond does
in CO−Auδ+ (δ > 0).51,74 Although there is no quantitative
value for the absorption coefficient of CO on differently
charged Au sites on CeO2, it does not appear the difference is
significant due to the fact that σ contribution is always present
in CO interaction with Au.51 This was shown in the case of Au/
SiO2 where the IR band area increase from CO adsorption on
cationic Au to metallic Au (increase in π bond) is less than
50%.48 Therefore, we expect that even an accurate correction
for the IR coefficients would not significantly affect the
calculation of the numbers of Au sties and therefore the
TOFs for the Au25/CeO2 rod catalyst with/without the
presence of thiolate ligands. It is thus concluded that the
thiolate ligands do not seem to pose an apparent positive or
negative effect on the reaction kinetics of the accessible Au sites
for CO oxidation. The independence of both the activation
energy and reaction rate on the thiolate ligand coverage
suggests that the thiolate ligands have a merely geometric
blocking rather than electronic effect on Au catalysis. The long-
range electronic or the ensemble effect, observed in sulfur
poisoning of other metal catalysts,75 does not seem to work in
the case of thiolate-protected Au nanoclusters for CO
oxidation.
3.8.3. Comment on the Thiolate Ligand Effect for Liquid-
and Gas-Phase Reactions. It is worthwhile to comment that
the intact Au25(SR)18 nanoclusters, either supported or
unsupported, are quite active for a variety of reactions in the
aqueous phase even though the surface Au atoms are
completely coordinated with thiolate ligands.6 This is quite
different from what we found in this study for gas-phase
reaction where the thiolate ligands have to be at least partially
removed to allow Au25 nanoclusters to be active for CO
oxidation. The difference is probably due to the fact that the
thiolate ligands can be flexible or even in an on-and-off dynamic
state in the presence of solvent so that some of the Au sites can
be accessible to reactants.6,42 In the gas phase, the large thiolate
ligands lose their flexibility and may be rigidly bound to the
shell of the Au25 nanoclusters, thus preventing reactant from
ASSOCIATED CONTENT
* Supporting Information
■
S
Additional STEM images, CO oxidation activity data, IR
spectra, XANES spectra, CO-TPR profiles, and isotopic
exchange results (Figure S1−S9). This material is available
free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Authors
wuz1@ornl.gov
■
overburysh@ornl.gov
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by the U.S. Department of Energy,
Office of Science, Office of Basic Energy Sciences, Chemical
Sciences, Geosciences, and Biosciences Division. Part of the
work including the Raman study was conducted at the Center
6120
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
(30) Wu, Z. L.; Dai, S.; Overbury, S. H. J. Phys. Chem. C 2010, 114,
412.
(31) Woicik, J. C.; Ravel, B.; Fischer, D. A.; Newburgh, W. J. J.
Synchrotron Radiat. 2010, 17, 409.
for Nanophase Materials Sciences, which is sponsored at Oak
Ridge National Laboratory (ORNL) by the Scientific User
Facility Division, Office of Basic Energy Sciences, U.S.
Department of Energy. Use of the National Synchrotron
Light Source, Brookhaven National Laboratory, was supported
by the U.S. Department of Energy, Office of Science, Office of
Basic Energy Sciences, under Contract DE-AC02-98CH10886
with additional support through the Synchrotron Catalysis
Consortium under Grant DE-FG02-05ER15688. The research
was supported in part by the appointment for A.K.P.M. and Z.-
A.Q. to the ORNL Postdoctoral Research Associates Program,
administered jointly by ORNL and the Oak Ridge Associated
Universities. R.J. acknowledges research support from the U.S.
Department of Energy, Office of Basic Energy Sciences, Grant
DE-FG02-12ER16354.
(32) Nashner, M. S.; Frenkel, A. I.; Adler, D. L.; Shapley, J. R.;
Nuzzo, R. G. J. Am. Chem. Soc. 1997, 119, 7760.
(33) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537.
(34) Ahlrichs, R.; Bar, M.; Haser, M.; Horn, H.; Kolmel, C. Chem.
Phys. Lett. 1989, 162, 165.
(35) Adamo, C.; Barone, V. J. Chem. Phys. 1999, 110, 6158.
(36) Andrae, D.; Haussermann, U.; Dolg, M.; Stoll, H.; Preuss, H.
Theor. Chim. Acta 1990, 77, 123.
(37) Farmer, J. A.; Campbell, C. T. Science 2010, 329, 933.
(38) Zhou, Z.; Flytzani-Stephanopoulos, M.; Saltsburg, H. J. Catal.
2011, 280, 255.
(39) Aguilar-Guerrero, V.; Gates, B. C. J. Catal. 2008, 260, 351.
(40) Carrettin, S.; Concepcion, P.; Corma, A.; Nieto, J. M. L.; Puntes,
V. F. Angew. Chem., Int. Ed. 2004, 43, 2538.
(41) Farrag, M.; Tschurl, M.; Dass, A.; Heiz, U. Phys. Chem. Chem.
Phys. 2013, 15, 12539.
(42) Li, G.; Jiang, D. E.; Liu, C.; Yu, C. L.; Jin, R. C. J. Catal. 2013,
306, 177.
(43) Simms, G. A.; Padmos, J. D.; Zhang, P. J. Chem. Phys. 2009, 131.
(44) MacDonald, M. A.; Chevrier, D. M.; Zhang, P.; Qian, H. F.; Jin,
R. C. J. Phys. Chem. C 2011, 115, 15282.
(45) Wu, Z. K.; Jin, R. C. ACS Nano 2009, 3, 2036.
(46) Frenkel, A. I.; Hills, C. W.; Nuzzo, R. G. J. Phys. Chem. B 2001,
105, 12689.
(47) Saavedra, J.; Powell, C.; Panthi, B.; Pursell, C. J.; Chandler, B. D.
J. Catal. 2013, 307, 37.
(48) Wu, Z. L.; Zhou, S. H.; Zhu, H. G.; Dai, S.; Overbury, S. H. J.
Phys. Chem. C 2009, 113, 3726.
(49) Wu, Z. L.; Zhou, S. H.; Zhu, H. G.; Dai, S.; Overbury, S. H.
REFERENCES
■
(1) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chem. Lett.
1987, 405.
(2) Hashmi, A. S. K. Science 2012, 338, 1434.
(3) Gates, B. C. Chem. Commun. 2013, 49, 7876.
(4) Aguilar-Guerrero, V.; Lobo-Lapidus, R. J.; Gates, B. C. J. Phys.
Chem. C 2009, 113, 3259.
(5) Flytzani-Stephanopoulos, M.; Gates, B. C. Annu. Rev. Chem.
Biomol. 2012, 3, 545.
(6) Li, G.; Jin, R. C. Acc. Chem. Res. 2013, 46, 1749.
(7) Corma, A. Nat. Chem. 2013, 5, 775.
(8) Xie, S. H.; Tsunoyama, H.; Kurashige, W.; Negishi, Y.; Tsukuda,
T. ACS. Catal. 2012, 2, 1519.
(9) Tsukuda, T. Bull. Chem. Soc. Jpn. 2012, 85, 151.
(10) Oliver-Meseguer, J.; Cabrero-Antonino, J. R.; Dominguez, I.;
Leyva-Perez, A.; Corma, A. Science 2012, 338, 1452.
(11) Zhu, Y.; Qian, H. F.; Das, A.; Jin, R. C. Chin. J. Catal. 2011, 32,
1149.
Chem. Commun. 2008, 3308.
(50) Guan, Y. J.; Ligthart, D. A. J. M.; Pirgon-Galin, O.; Pieterse, J. A.
Z.; van Santen, R. A.; Hensen, E. J. M. Top. Catal. 2011, 54, 424.
(51) Hadjiivanov, K.; Vayssilov, G. Advances in Catalysis; Academic
Press: San Diego, CA, 2002; Vol. 47, p 308.
(52) Manzoli, M.; Boccuzzi, F.; Chiorino, A.; Vindigni, F.; Deng, W.
L.; Flytzani-Stephanopoulos, M. J. Catal. 2007, 245, 308.
(53) Li, M. J.; Wu, Z. L.; Ma, Z.; Schwartz, V.; Mullins, D. R.; Dai, S.;
Overbury, S. H. J. Catal. 2009, 266, 98.
(54) Fierro-Gonzalez, J. C.; Gates, B. C. Catal. Today 2007, 122, 201.
(55) Binet, C.; Daturi, M.; Lavalley, J. C. Catal. Today 1999, 50, 207.
(56) Li, M. J.; Wu, Z. L.; Overbury, S. H. J. Catal. 2011, 278, 133.
(57) Baron, M.; Bondarchuk, O.; Stacchiola, D.; Shaikhutdinov, S.;
Freund, H. J. J. Phys. Chem. C 2009, 113, 6042.
(58) Vayssilov, G. N.; Lykhach, Y.; Migani, A.; Staudt, T.; Petrova, G.
P.; Tsud, N.; Skala, T.; Bruix, A.; Illas, F.; Prince, K. C.; Matolin, V.;
Neyman, K. M.; Libuda, J. Nat. Mater. 2011, 10, 310.
(59) Happel, M.; Myslivecek, J.; Johanek, V.; Dvorak, F.; Stetsovych,
O.; Lykhach, Y.; Matolin, V.; Libuda, J. J. Catal. 2012, 289, 118.
(60) Camellone, M. F.; Fabris, S. J. Am. Chem. Soc. 2009, 131, 10473.
(61) Guzman, J.; Carrettin, S.; Corma, A. J. Am. Chem. Soc. 2005,
127, 3286.
(12) Herzing, A. A.; Kiely, C. J.; Carley, A. F.; Landon, P.; Hutchings,
G. J. Science 2008, 321, 1331.
(13) Zhu, Y.; Qian, H. F.; Jin, R. C. Chem.Eur. J. 2010, 16, 11455.
(14) Jin, R. C.; Qian, H. F.; Wu, Z. K.; Zhu, Y.; Zhu, M. Z.; Mohanty,
A.; Garg, N. J. Phys. Chem. Lett. 2010, 1, 2903.
(15) Maity, P.; Xie, S. H.; Yamauchi, M.; Tsukuda, T. Nanoscale
2012, 4, 4027.
(16) Chaki, N. K.; Negishi, Y.; Tsunoyama, H.; Shichibu, Y.;
Tsukuda, T. J. Am. Chem. Soc. 2008, 130, 8608.
(17) Nie, X. T.; Zeng, C. J.; Ma, X. G.; Qian, H. F.; Ge, Q. J.; Xu, H.
Y.; Jin, R. C. Nanoscale 2013, 5, 5912.
(18) Nie, X. T.; Qian, H. F.; Ge, Q. J.; Xu, H. Y.; Jin, R. C. ACS Nano
2012, 6, 6014.
(19) Ma, G. C.; Binder, A.; Chi, M. F.; Liu, C.; Jin, R. C.; Jiang, D. E.;
Fan, J.; Dai, S. Chem. Commun. 2012, 48, 11413.
(20) Gaur, S.; Miller, J. T.; Stellwagen, D.; Sanampudi, A.; Kumar, C.
S. S. R.; Spivey, J. J. Phys. Chem. Chem. Phys. 2012, 14, 1627.
(21) Somorjai, G. A. J. Catal. 1972, 27, 453.
(22) Haider, P.; Urakawa, A.; Schmidt, E.; Baiker, A. J. Mol. Catal. A
2009, 305, 161.
(62) Shapovalov, V.; Metiu, H. J. Catal. 2007, 245, 205.
(63) Song, W.; Hensen, E. J. M. Catal. Sci. Technol. 2013, 3, 3020.
(64) Kim, H. Y.; Henkelman, G. J. Phys. Chem. Lett. 2013, 4, 216.
(65) Kim, H. Y.; Lee, H. M.; Henkelman, G. J. Am. Chem. Soc. 2012,
134, 1560.
(66) Ta, N.; Liu, J. Y.; Chenna, S.; Crozier, P. A.; Li, Y.; Chen, A. L.;
Shen, W. J. J. Am. Chem. Soc. 2012, 134, 20585.
(67) Widmann, D.; Leppelt, R.; Behm, R. J. J. Catal. 2007, 251, 437.
(68) Widmann, D.; Behm, R. J. Acc. Chem. Res. 2014, 47, 740.
(69) Lee, Y.; He, G. H.; Akey, A. J.; Si, R.; Flytzani-Stephanopoulos,
M.; Herman, I. P. J. Am. Chem. Soc. 2011, 133, 12952.
(70) Widmann, D.; Behm, R. J. Angew. Chem., Int. Ed. 2011, 50,
10241.
(23) Menard, L. D.; Xu, F. T.; Nuzzo, R. G.; Yang, J. C. J. Catal.
2006, 243, 64.
(24) Shivhare, A.; Chevrier, D. M.; Purves, R. W.; Scott, R. W. J. J.
Phys. Chem. C 2013, 117, 20007.
(25) Marshall, S. T.; O’Brien, M.; Oetter, B.; Corpuz, A.; Richards, R.
M.; Schwartz, D. K.; Medlin, J. W. Nat. Mater. 2010, 9, 853.
(26) Green, I. X.; Tang, W. J.; Neurock, M.; Yates, J. T. Science 2011,
333, 736.
(27) Zhu, M.; Lanni, E.; Garg, N.; Bier, M. E.; Jin, R. J. Am. Chem.
Soc. 2008, 130, 1138.
(28) Wu, Z. L.; Li, M. J.; Overbury, S. H. J. Catal. 2012, 285, 61.
(29) Wu, Z. L.; Li, M. J.; Howe, J.; Meyer, H. M.; Overbury, S. H.
Langmuir 2010, 26, 16595.
6121
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122

Journal of the American Chemical Society
Article
(71) Oberhauser, W.; Lavacchi, A.; Vizza, F.; Capozzoli, L.; Lee, H.
M. Appl. Catal., A 2013, 451, 58.
(72) Lopez-Acevedo, O.; Kacprzak, K. A.; Akola, J.; Hakkinen, H.
Nat. Chem. 2010, 2, 329.
(73) Ide, M. S.; Davis, R. J. Acc. Chem. Res. 2014, 47, 825.
(74) Mihaylov, M.; Ivanova, E.; Hao, Y.; Hadjiivanov, K.; Gates, B.
C.; Knozinger, H. Chem. Commun. 2008, 175.
(75) Bartholomew, C. H. Studies in Surface Science and Catalysis;
Elsevier: Amsterdam, 1987; Vol. 34, p 81.
(76) Chen, J.; Zhang, Q. F.; Bonaccorso, T. A.; Williard, P. G.; Wang,
L. S. J. Am. Chem. Soc. 2014, 136, 92.
6122
dx.doi.org/10.1021/ja5018706 | J. Am. Chem. Soc. 2014, 136, 6111−6122