Tailoring gold nanoparticle characteristics and the impact on aqueous-phase oxidation of glycerol

Article abstract of DOI:10.1021/acscatal.5b00754

Poly(vinyl alcohol) (PVA)-stabilized Au nanoparticles (NPs) were synthesized by colloidal methods in which temperature variations (-75 to 75 °C) and mixed H₂O/EtOH solvent ratios (0, 50, and 100 vol/vol) were used. The resulting Au NPs were immobilized on TiO₂ (P25), and their catalytic performance was investigated for the liquid phase oxidation of glycerol. For each unique solvent system, there was a systematic increase in the average Au particle diameter as the temperature of the colloidal preparation increased. Generation of the Au NPs in H₂O at 1 °C resulted in a high observed activity compared with current Au/TiO₂ catalysts (turnover frequency = 915 h^-1). Interestingly, Au catalysts with similar average particle sizes but prepared under different conditions had contrasting catalytic performance. For the most active catalyst, aberration-corrected high angle annular dark field scanning transmission electron microscopy analysis identified the presence of isolated Au clusters (from 1 to 5 atoms) for the first time using a modified colloidal method, which was supported by experimental and computational CO adsorption studies. It is proposed that the variations in the populations of these species, in combination with other solvent/PVA effects, is responsible for the contrasting catalytic properties.

Full text of DOI:10.1021/acscatal.5b00754

Research Article
pubs.acs.org/acscatalysis
Tailoring Gold Nanoparticle Characteristics and the Impact on
Aqueous-Phase Oxidation of Glycerol
Scott M. Rogers,†,‡ C. Richard A. Catlow,†,‡ Carine E. Chan-Thaw,§ Diego Gianolio,∥ Emma K. Gibson,†,‡
Anna L. Gould,†,‡ Nan Jian,⊥ Andrew J. Logsdail,‡ Richard E. Palmer,⊥ Laura Prati,§
Nikolaos Dimitratos,*,†,# Alberto Villa,*,§ and Peter P. Wells*,†,‡
†UK Catalysis Hub, Research Complex at Harwell, Rutherford Appleton Laboratory, Harwell Oxon, Didcot, OX11 0FA, United
Kingdom
‡Department of Chemistry, University College London, 20 Gordon Street, London WC1H 0AJ, United Kingdom
§Dipartimento di Chimica, Universita
̀
degli Studi di Milano, via Golgi 19, 20133 Milano, Italy
∥Diamond Light Source Ltd, Harwell Science and Innovation Campus, Chilton, Didcot OX11 0DE, United Kingdom
⊥Nanoscale Physics Research Laboratory, School of Physics and Astronomy, University of Birmingham, Edgbaston, Birmingham, B15
2TT, United Kingdom
#Cardiff Catalysis Institute, School of Chemistry, Cardiff University, Cardiff, CF10 3AT, United Kingdom
S
* Supporting Information
ABSTRACT: Poly(vinyl alcohol) (PVA)-stabilized Au nanoparticles (NPs) were synthesized
by colloidal methods in which temperature variations (−75 to 75 °C) and mixed H2O/EtOH
solvent ratios (0, 50, and 100 vol/vol) were used. The resulting Au NPs were immobilized on
TiO2 (P25), and their catalytic performance was investigated for the liquid phase oxidation of
glycerol. For each unique solvent system, there was a systematic increase in the average Au
particle diameter as the temperature of the colloidal preparation increased. Generation of the
Au NPs in H2O at 1 °C resulted in a high observed activity compared with current Au/TiO2
catalysts (turnover frequency = 915 h−1). Interestingly, Au catalysts with similar average
particle sizes but prepared under different conditions had contrasting catalytic performance.
For the most active catalyst, aberration-corrected high angle annular dark field scanning
transmission electron microscopy analysis identified the presence of isolated Au clusters (from
1 to 5 atoms) for the first time using a modified colloidal method, which was supported by
experimental and computational CO adsorption studies. It is proposed that the variations in the populations of these species, in
combination with other solvent/PVA effects, is responsible for the contrasting catalytic properties.
KEYWORDS: gold nanoparticles, colloids, XAFS, glycerol oxidation, heterogeneous catalysis, aberration-corrected TEM
used.10,18,19 Understanding the effect of temperature and the
surrounding solvent environment on the reduction process, and
INTRODUCTION
■
Tailored preparation of supported metal nanoparticles (NPs) is
consequent particle characteristics, are crucially important.
at the forefront of catalytic research. Characteristics of interest
It has been demonstrated that the catalytic properties of
include metal particle size, shape and structure.1−3 However, it
gold-NPs are highly dependent on the particle size, with
is difficult to achieve fine control over these properties using
minimal activity often observed at sizes >10 nm. At smaller
typical synthesis approaches, that is, impregnation and
particle sizes, Au is extremely active for CO oxidation,20
precipitation.4−6 The sol-immobilization method is a successful
hydrogenations,21,22 and the selective oxidation of alkenes.23−25
route for the synthesis of supported metal nanoparticles with
tailored properties.7,8 Colloidal methods have been studied for
The origin of the reactive gold has been studied and attributed
largely to the population of low-coordinated Au atoms as the
over 60 years and are able to prepare metal particles in a range
active sites.26 It is also well established that gold nanoparticles
of sizes (2−120 nm).9−11 The particle size is dependent on the
choice of stabilizer and the type of reducing agent used. The
use of poly(vinyl alcohol) (PVA) as a stabilizing agent has
are effective for the selective oxidation of many alcohols.16,27−29
Among the possible alcohols, glycerol is an attractive feedstock
in the production of fine chemicals because it can be obtained
as a byproduct from the production of biodiesel. As a highly
become common in the preparation of metal nanoparticles <10
nm.12−17 Colloids incorporating the use of PVA have been
synthesized only as hydrosols, using water as the solvent, at
room temperature. In terms of controlling the nanoparticle size
during the reduction stage, attention has been focused on the
gold concentration and the strength of the reductant
Received: April 10, 2015
Revised: May 21, 2015
© XXXX American Chemical Society
4377
DOI: 10.1021/acscatal.5b00754
ACS Catal. 2015, 5, 4377−4384

ACS Catalysis
Research Article
functionalized molecule, there are many possible oxidation
products; hence, a selective catalyst is required.30 It is
understood that the activity and selectivity toward glyceric
acid is heavily affected by the metal nanoparticle size, with
smaller metal nanoparticles <5 nm generating high turnover
frequencies as a result of their high reaction rate.16,28,29,31
In the present study, we demonstrate how systematic
approaches to designing nanoparticles can achieve both
increased catalytic performance and durability in a manner
that has not been previously reported. The work focuses on
applying two modifications to an established sol-immobilization
protocol, described elsewhere, for the preparation of titania
supported gold nanoparticles:13,16,32 (1) varying the temper-
ature at which the colloid is generated and (2) applying
different solvent systems through varied water/ethanol ratios.
We characterize the Au particle size using transmission electron
microscopy (TEM), aberration-corrected scanning transmission
electron microscopy (STEM) and X-ray absorption fine
structure (XAFS) and investigate the Au surface sites using
CO-chemisorption with infrared (IR), corroborated by density
functional theory (DFT) calculations. We performed catalytic
tests in the liquid phase oxidation of glycerol to investigate the
effects that temperature and the solvent used for nanoparticle
preparation have on performance.
Table 1. Applied Temperature and Solvent Ratios for Au/
TiO2 Catalyst Preparation
temp, °C
H2O/EtOH solvent ratio (vol/vol)
sample abbreviation
1
25
100
100
100
100
50
A1
A2
A3
A4
B1
B2
B3
C1
C2
50
75
−30
0
50
30
50
−75
−30
0
0
before the solvent was evaporated. The samples for TEM and
HAADF STEM were then examined using JEOL JEM 2100 EM
and JEOL JEM 2100F STEM models, respectively. HAADF
STEM analysis was performed in the Nanoscale Physics
Research Laboratory at the University of Birmingham. The
integrated HAADF STEM intensity was used to obtain the size
of ultrasmall clusters, with large Au clusters functioning as a
mass balance.33−35
XAFS. Au L3-edge XAFS studies were carried out on the B18
beamline at the Diamond Light Source, Didcot, U.K.
Measurements were performed in fluorescence mode using a
QEXAFS setup with a fast-scanning Si(111) double crystal
monochromator and a nine-element Ge detector. The time
resolution of the spectra was 2 min/spectrum (kmax = 14). On
average, 25 scans were acquired to improve the signal-to-noise
level of the data. XAS data processing and extended X-ray
absorption fine structure (EXAFS) analysis were performed
using IFEFFIT with the Horae package (Athena and
METHODS
■
Catalyst Preparation. A standard sol-immobilization
method with temperature control was employed in the
preparation of supported Au nanoparticles. HAuCl4·3H2O
was used to prepare solutions with various H2O/EtOH ratios of
the desired gold concentration (1.26 × 10−4 M), to which
solutions of PVA (1 wt % solution, Aldrich, Mw = 9000−10 000
g mol−1, 80% hydrolyzed, PVA/Au (w/w) = 0.65) were added.
Solutions of NaBH4 (0.1 M) (>96%, Aldrich, NaBH4/Au (mol/
mol) = 5) were freshly prepared and added dropwise to each
with stirring to form red-brown sols, depending on the solution
temperature. After the complete reduction of gold species (30
min), the colloidal solutions were individually immobilized
onto titania (P25, Degussa) under vigorous stirring conditions.
The amount of support material required was calculated so as
to give a final metal loading of 1 wt %. The mixture was
acidified to pH 1−2 by sulfuric acid before it was stirred for 60
min. The slurry was filtered, and the resulting solid was washed
thoroughly with distilled water, (2 L of doubly distilled water)
to remove soluble impurities, and dried overnight at room
temperature. The different H2O/EtOH solvent ratios and
temperatures at which each Au/TiO2 catalyst was prepared, as
well as an abbreviated name to be used throughout the paper,
are presented in Table 1.
Artemis).36,37 The amplitude reduction factor, s0 , was derived
2
from EXAFS data analysis of an obtained Au foil spectrum to
be 0.80, which was used as a fixed input parameter. The EXAFS
data were fitted in R-space with a typical fit range of 1.15 < R <
5 Å.
Diffuse Reflectance Infrared Fourier Transform Spec-
troscopy (DRIFTS). FTIR spectra were obtained with an
Agilent Cary 680 spectrometer at a spectral resolution of 2
cm−1 and accumulating up to 64 scans. For each experiment,
the catalyst was placed in the sample holder and smoothed to
leave a flat surface. The cell was purged with helium for 30 min
before CO was introduced using a 10% CO/He mixture at a
flow rate of 30 mL min−1 over a 5 min period. The gas was
switched to helium for 30 min at a flow rate of 30 mL min−1 to
remove gaseous and physisorbed CO from the catalyst surface.
Computer Simulations. DFT calculations were performed
as implemented by the projector-augmented wave method
(PAW).38 Grid-based PAW uses a real-space grid, transforming
the wave functions at the core to a smooth pseudowave
function.39 A converged grid spacing of 0.18 was used to
represent numerically the wave function, and structural
convergence was achieved when the forces on all atoms were
found to be <0.01 eV Å−1.
UV−Visible Spectroscopy. The colloidal hydrosols were
analyzed by UV−vis spectroscopy to determine the position of
the gold surface plasmon resonance band. UV−vis spectra
(200−800 nm, Shimadzu UV-1800 spectrometer) of the Au
sols were recorded in a quartz cuvette after 30 min of sol
generation.
TEM and High Angle Annular Dark Field (HAADF)
STEM Imaging. The surface morphology was investigated
directly by imaging, with Au particle size and distribution the
characteristics of interest. Samples for examination by TEM and
aberration-corrected HAADF STEM were prepared by first
dispersing the catalyst powder in high-purity ethanol using
sonication. A 40 μL portion of the suspension was dropped
onto a holey carbon film supported by a 300-mesh copper grid
The geometries for small Au-CO clusters were taken from
the previous work of Jiang and Xu,40 and geometry
optimizations were performed. The molecules investigated
were Au5CO, Au4CO, Au3CO, Au2CO, Au2(CO)2, Au(CO)2,
and AuCO. Depending on the number of electrons in the
system, spin-polarized (Au5CO, Au3CO, Au(CO)2 and AuCO)
and spin-paired (Au4CO, Au2CO, Au2(CO)2) calculations were
performed for odd- and even-electron systems, respectively,
4378
DOI: 10.1021/acscatal.5b00754
ACS Catal. 2015, 5, 4377−4384

ACS Catalysis
Research Article
with the generalized gradient approximation or Perdew, Burke
and Ernzerhof as the exchange-correlation functional;41 the
residual minimization method, direct inversion in iterative
subspace was used for convergence of the self-consistent cycle.
For the odd-electron systems, a low-spin doublet configuration
(μ = 1) was energetically preferable in all cases; a complete
summary of the different spin states trialled and relative
energies is provided in the Supporting Information (SI). The
IR vibrational frequencies were calculated using finite-difference
methods and compared with CO in the gas phase, and
adsorbed at varying sites on Au (100) and (111) surfaces:
namely, the atop, bridge, and hollow sites on the (100) surface;
and the atop, bridge, hcp, and fcc sites on the (111) surface.
The Au surface was modeled using a 5-layer-thick slab with 2D
periodicity, using 1 k-point per 0.024 Å−1 in the directions
parallel to the surface plane (i.e., x and y). These studies, details
of which are presented in the SI, show that the calculation of
stretching frequencies may be limited to the contributions of
the CO molecule and Au atoms that are directly bound to it,
with all other atoms frozen, thus significantly reducing the
computational load. Finally, we also considered the influence of
charge on the vibrational frequencies to simulate for the effect
of a charge-donating support such as TiO2.
Glycerol Oxidation. A 0.3 M glycerol solution and the Au
catalyst (substrate/total metal = 1000 mol/mol) were mixed in
distilled water and NaOH (4:1 mol/mol NaOH/glycerol, total
volume = 10 mL). The reactor was pressurized at 300 kPa of
nitrogen, and the temperature was set to 50 °C. Once this
temperature was reached, the gas supply was switched to
oxygen, and the monitoring of the reaction started. The
reaction was initiated by stirring. Samples were removed
periodically and analyzed by high-performance liquid chroma-
tography using a column (Alltech OA-10308, 300 mm_7.8
mm) with UV and refractive index detector aqueous H3PO4
solution (0.1 wt %) was used as the eluent. Products were
identified by comparison with original samples. Recycling tests
performed were under the same reaction conditions. The
catalyst was recycled in the subsequent run after filtration
without any further treatment. The recovery of the catalyst was
>95% after each run. The volume of glycerol solution was
adjusted in order to keep constant the substrate/metal ratio
(1000 mol/mol).
ranged between 490 and 538 nm, influenced by both
temperature and solvent during preparation. However, it is
not possible to compare the effect of the different solvent
systems on the position of the plasmon resonance band, as a
result of differing dielectric properties of the solvent media.42
Although each band is broad, it can be noted that there is a
general shift to higher wavelength in the peak position of the
plasmon band as the temperature to which the Au sols are
generated increases. This indicates the formation of larger Au
particles as the temperature of colloidal reduction increases,
with the trend observed for all three solvent systems (H2O
shown in Figure 1, H2O/EtOH and EtOH shown in Figure
S1).
TEM. TEM was performed on all Au/TiO2 catalysts to
determine the particle size distribution. Representative TEM
images and the derived histograms of the Au catalysts are
presented in Figures 2 and S5. The average Au particle size and
deviation for all Au catalysts are presented in Table 2. For the
same solvent of preparation, the particle size analysis confirms
that, as the temperature at which the Au colloid is formed
decreases, a corresponding reduction in Au particle diameter
and a narrowing of the overall distribution range is observed.
This behavior is observed for all solvent systems investigated,
with the water (A1 = 2.0 nm) and water−ethanol (B1 = 1.8
nm) environments yielding the smallest nanoparticles. To
observe how the solvent environment during preparation affects
particle size, it is possible to compare catalysts A1 against B2 as
well as A2 against B3. These samples have comparable
temperatures of colloidal reduction. There is little difference
in average Au particle size when comparing A1 and B2, but
there is an increase of 1.8 nm between the catalysts A2 and B3,
which gives an indication that switching to a higher EtOH/H2O
solvent ratio increases the Au particle size during colloidal
preparation. Switching to a pure EtOH solvent yields larger Au
particles, regardless of the temperature, which could be
attributed to the insoluble nature of PVA in EtOH, with full
solubility achieved only if a few drops of H2O are added during
its preparation.
HAADF STEM. A1 and B1 catalysts were examined with
HAADF STEM microscopy to investigate the presence of Au
clusters that are too small to image with TEM. It was observed
that ultrasmall Au clusters consisting of only a few atoms are
present for the A1 catalyst (Figure 3), which were not,
however, identified for B1. The quantitative analysis of the
clusters’ integrated HAADF STEM intensity shows that the
ultrasmall clusters are of 1−5 Au atoms. For both Au catalysts,
many Au clusters below 1 nm were recorded (Refer to Figures
S2, S3).
RESULTS AND DISCUSSION
■
UV−Visible Characterization. The plasmon resonance
band maximum, indicative of metal nanoparticles <10 nm, is
presented in Table 2 for all Au catalysts. The band position
XAFS Characterization. EXAFS was used to calculate the
average Au particle size, as a representation of the whole
sample. The first-shell EXAFS fitting parameters are shown in
the SI, and the k2-weighted magnitude Fourier transform data
and corresponding fit presented in Figure 4 and Table S2. The
EXAFS data shows that for all catalysts prepared, the Au
environment is dominated by a primary Au coordination shell.
The Au−Au first shell coordination numbers range between 7.7
and 11.2, all of which are lower than that found for bulk Au
(12). The effect of temperature on the coordination number
can be observed in Figure 5. Transforming the coordination
number into a particle size can be performed using the
approach of Beale et al.,43 whose work simulated the average
first shell coordination number versus the number of atoms for
the first shell for each of the packing types. The average particle
Table 2. Au Surface Plasmon Resonance Band Maximum
and the Average Particle Diameter from TEM and EXAFS
Au/TiO2
catalyst
UV maximum,
nm
av TEM particle
size, nm
av EXAFS particle
size, nm
A1
A2
A3
A4
B1
B2
B3
C1
C2
500
2.0 0.6
2.3 0.5
2.8 0.9
3.3 0.8
1.8 0.5
2.0 0.6
4.1 1.8
2.4 1.0
2.9 1.2
1.35
1.42
1.97
490−500
500−510
538
490−500
496
1.18
1.24
535−546
1.63
2.23
516−520
4379
DOI: 10.1021/acscatal.5b00754
ACS Catal. 2015, 5, 4377−4384

ACS Catalysis
Research Article
Figure 1. UV−vis spectra of Au sols prepared at different temperatures in H2O: (A) full spectrum and (B) enlarged region for surface plasmon
resonance band.
Figure 3. (a) 3D surface plot of A1 Au catalyst shows the ultrasmall
Figure 2. Representative TEM images and corresponding histograms
clusters consisting of one and two Au atoms. (b) High resolution
of 1 wt % Au/TiO2 catalysts: (A) A1, (B) B1, and (C) C1.
HAADF STEM images of single Au atom and Au2 cluster. (c) High
resolution HAADF STEM images of Au2 and Au5 clusters.
size from TEM in our work, along with the particle size
obtained from the first shell EXAFS coordination, are presented
in Table 2. The values are based on the assumptions that the
nanoparticles are spherical, face centered cubic, and <3 nm. As
the diameter of the Au particles increases, the primary Au−Au
radial distance (R) extends (2.81−2.84 Å), as is widely
reported.44 The trend for the Au particle diameter derived
from the EXAFS supports the TEM analysis. The particle
diameter obtained from EXAFS is smaller than that obtained
from TEM due to EXAFS accounting for all the atoms that
cannot be observed in microscopy.
DRIFTS. CO-adsorption spectra reveal a difference in surface
sites and electronic properties between the Au samples. Figure
6A details the CO adsorption bands for catalysts A1 and A2,
which have more defined CO adsorption bands compared with
A3 (Figure 6B). With regard to the electronic properties at the
Au/TiO2 interface, previous literature has been used to assign
the observed bands. The small band at ∼2115 cm−1 present in
the CO adsorption spectra of A1 and A2 can be assigned to
CO−Au0 species.45 The broad band ∼2040 cm−1 observed for
4380
DOI: 10.1021/acscatal.5b00754
ACS Catal. 2015, 5, 4377−4384

ACS Catalysis
Research Article
Figure 4. Magnitude component of the k2-weighted Fourier transform
data and corresponding fit for all the Au/TiO2 catalysts prepared in a
H2O solvent (i. A1, ii. A2, iii. A3, iv. A4).
Figure 6. FTIR spectra from CO-adsorption studies onto different
Au/TiO2 catalysts: (A) A1, B1 and (B) C1, B1, and A3.
solvents. The interaction between PVA and the EtOH solvent
appears detrimental to the available Au sites for CO adsorption.
The samples of interest were not pretreated in any way to clean
the catalyst surface or to modify the electronic properties;
hence, in all cases, the stabilizing agent, PVA, may crucially
inhibit the ability of each sample to adsorb CO at the surface.
As a result, facile removal of the stabilizing agent has been
widely investigated, with the aim of removing the protecting
agent without affecting the structure of the catalyst.33
Figure 5. Effect of solvent temperature during Au colloidal formation
in H2O on the first shell EXAFS Au−Au coordination number and
obtained Au particle size.
DFT Calculations. Geometry optimization and vibrational
frequency analysis was performed on small AuxCO clusters and
for CO adsorbed at varying sites on (100) and (111). Our
calculations show that the relevant vibrational modes in the IR
spectra may be restricted to those relating to the CO molecule
and directly bonded Au atoms, as shown in the ESI. The CO
stretching frequencies for selected AuxCO clusters and surface
adsorption sites are given in Table 3.
From the results outlined, we see that the CO stretching
frequencies when adsorbed on small gold clusters are lower
than for gas-phase CO and similar to the slab calculations for
which the CO molecule is adsorbed in an atop position, with a
vibrational frequency of ∼2090 cm−1. However, when CO
binds in a bridging fashion, the vibrational frequency is shifted
to considerably lower wavenumbers on the (100) and (111)
surfaces. Both Au3CO and Au5CO molecules have CO
positioned in a bridging arrangement, yet the CO stretching
frequency for the Au5CO is now much lower, at 1869.9 cm−1.
This result suggests that CO stretching frequencies below 1900
cm−1 may be assigned to bridging positions for the CO
adsorbate on small Au clusters and may therefore correspond to
the unusually distinct peaks highlighted in Figure 6.
each sample is assigned to CO−Auδ− as a result of electronic
transfer from the reducible titania support to very small gold
particles or clusters.46,47 For A1 and A2 samples, it is evident
from the asymmetric nature of this band that it comprises two
different adsorption sites. The bands present below 1900 cm−1
are difficult to assign from the literature, with only clusters of
Au5(CO) being reported with bands at this wavenumber.40 The
broadness associated with the CO adsorption bands for samples
A3, B1, and C1 (Figure 6B) can be a result of the dipole
coupling effect between neighboring CO molecules: when CO
coverage is large, the dipoles between CO molecules couple,
broadening the adsorption band. The sharp absorption bands
observed in A1 and A2 provide evidence that there is limited
coupling, confirming that the sites are more isolated.
Sample B1, in which the smallest average Au particle exists,
did not produce the same CO-adsorption bands as described
from the low temperature preparations in H2O. Instead, the
spectrum was consistent with that of the A3 catalyst, featuring
two broad bands at ∼2040 and ∼1850 cm−1, which is also
observed for the samples prepared in pure EtOH (Figure 6B).
The difference can be ascribed only to the influence of PVA on
the gold species during the catalyst preparation in different
To understand the influence of charge-donating supports,
such as TiO2, on the CO stretching frequencies, we also
4381
DOI: 10.1021/acscatal.5b00754
ACS Catal. 2015, 5, 4377−4384

ACS Catalysis
Research Article
with tetrakis(hydroxymethyl)phosphonium chloride (THPC)
as the stabilizing agent generated TOFs of 721 and 367 h−1,
respectively.48 A2, A3, and A4 have significantly lower activities
than A1 (TOF of 663, 341, and 161 h−1, respectively), with
time online analysis shown in Figure 7.
Table 3. Calculated CO Vibrational Frequencies of Gas-
Phase CO, Small AuxCO Clusters, and CO Adsorbed onto
(100) and (111) Surface Slabs
a
structure
adsorption site
CO vibrational frequency (cm−1
)
CO
gas-phase
atop
2134.6
1998.7
2108.3
2113.5
2084.3
2096.1
1869.9
1767.2
2085.4
1976.7
not stable
2082.6
1998.7
1941.1
1944.9
AuCO
Au2CO
atop
Au2(CO)2
Au3CO
atop
bridge
atop
Au4CO
Au5CO
bridge
bridge
atop
[Au5CO]−
Au (100)
Au (100)
Au (100)
Au (111)
Au (111)
Au (111)
Au (111)
bridge
hollow
atop
bridge
hcp
fcc
a
The (100) hollow site is unstable for CO adsorption.
Figure 7. Reaction profile for Au catalysts prepared in H2O.
negatively charged the Au5CO molecule, reoptimized the
geometry, and recalculated the vibrational frequencies. This
procedure resulted in a significant red shift of the CO stretching
frequency by 102.7 cm−1, from 1869.9 to 1767.2.9 cm−1. Given
also the electronegative nature of Au, such a shift perhaps
indicates that lower CO stretching frequencies in the
experimental IR spectra could also arise from nanoparticle
interactions with the substrate.
Glycerol Oxidation. Au/TiO2 catalysts were evaluated for
the liquid phase oxidation of glycerol under standard
conditions: glycerol 0.3 M, 50 °C, 300 kPa O2, glycerol/Au
1000 mol/mol, NaOH/glycerol 4 mol/mol. In Table 4, the
Correlating the activity data for all Au catalysts with the
average Au particle sizes showed that this is not the only
parameter influencing the catalytic performance. Indeed, A1,
A2, B1, B2, and C1 catalysts showed different catalytic
behavior, despite the similar Au average particle diameter
(1.8−2.4 nm). The difference can be attributed to two factors:
(1) the population of very small Au clusters, as seen in A1 and
(2) the interaction between protecting agent and solvent
system, which is influenced by the temperature and the solvent
in which the Au NPs are generated. It has previously been
demonstrated that PVA affects the catalytic performance, as it is
proposed that PVA can direct the adsorption of glycerol to Au
active sites by interacting with the OH groups of glycerol.49,50
There is an interesting result for the C series, in which the
larger Au particle size catalyst, C2, has a greater activity than
the smaller C1. We also attribute this trend to differences in the
way PVA interacts with the Au colloids as a consequence of the
low solubility of PVA in ethanol, which is exacerbated by the
low temperature synthesis. We suggest that the presence of
EtOH at the NP surface can also influence the adsorption of
glycerol.
To explore the effect of the solvent used for the preparation
of Au catalysts on their stability, we investigated the
recyclability of the catalysts. The most active catalysts for
each solvent (A1, B1, and C2) were chosen. The catalysts
showed good stability after 10 runs, reaching full conversion
after 4 h (Tables 5, 6, and S5). A1 showed a decreasing TOF
between the first and second runs (915 and 652 h−1).
Afterward, the TOF remained constant in the successive
eight runs (Table 5). The selectivity to glycerate decreased
from 73% in the first run to 56% in the second run, during
which a higher amount of tartronate was produced. This result
can be attributed to the change in the average particle size of
the A1 catalyst after the first run, with the aggregation of the
very small Au clusters that were observed in HAADF STEM
analysis. There was, indeed, an increase, from 2.0 to 3.7 nm, in
the average Au particle diameter before and after the recycling
tests by TEM. Au/TiO2 prepared in EtOH/H2O and EtOH
showed an increased TOF after the second run (Tables 6 and
S5), which remained stable in the successive runs. The catalyst
Table 4. Comparison of Au/TiO2 Catalyst Activities in
a
Glycerol Oxidation
c
selectivity (%)
b
Au/TiO2 cat.
TOF (h−1
)
Glyc
Gly
Tartr
Lac
Form
A1
A2
A3
A4
B1
B2
B3
C1
C2
915
663
341
161
202
95
73
74
76
79
76
78
80
78
78
4
4
4
6
4
5
5
4
5
12
12
8
8
6
7
7
6
5
7
6
6
3
1
3
4
3
3
5
3
4
2
11
9
80
3
130
314
9
7
a
Reaction conditions: alcohol/metal 1000/1 (mol/mol), 50 °C, pO2 =
b
300 kPa, 1250 rpm. TOF calculated after 15 min of reaction.
c
Selectivity at 90% conversion. Glyc = glycerate; Gly = glycolate; Tartr
= tartronate; Lac = lactate; Form = formate.
activity and selectivity of Au catalysts prepared using different
solvents and different temperatures are reported. All Au
catalysts showed good selectivity to glyceric acid (73−80%).
The Au/TiO2 catalysts prepared in water showed the best
catalytic performance compared with the alternative solvent
systems, with A1 the most active catalyst (turnover frequency
(TOF) of 915 h−1). This value, compared with previously
reported Au/TiO2 catalysts, is high. Au/TiO2 produced by
deposition precipitation or synthesized using colloidal methods
4382
DOI: 10.1021/acscatal.5b00754
ACS Catal. 2015, 5, 4377−4384

ACS Catalysis
Research Article
a
demonstrated the presence of isolated ultrasmall Au clusters
(1−5 atoms), as shown by HAADF STEM and supported by
CO adsorption studies. The study also shows that catalysts with
similar average Au particle diameters, A1 and B2, as identified
by EXAFS and TEM - gave rise to different catalytic properties,
with TOF values of 915 and 95 h−1, respectively. This insight is
of great importance when trying to elucidate accurate structure
function relationships. This study links performance of Au/
TiO2 catalysts for glycerol oxidation to the population of
specific cluster sizes and also indicates that reduced average
particle diameter alone is not sufficient to promote the reaction.
The interaction between the PVA and the Au NPs should not
be ignored and may also influence the types/number of Au
aggregates formed alongside affecting which sites are available.
Table 5. Recycling Results Using the A1 Au Catalyst
c
selectivity (%)
a
b
A1
TOF (h−1
)
conv. (%) after 4 h Glyc Gly Tartr Lac
run 1
run 2
run 3
run 4
run 5
run 6
run 7
run 8
run 9
run 10
915
652
678
659
649
657
638
652
649
632
98
98
99
98
98
97
98
98
98
99
73
56
57
59
57
56
57
58
56
58
4
6
12
22
25
23
24
23
24
24
22
22
8
9
9
9
6
7
7
6
6
7
6
6
9
11
10
11
10
12
a
Reaction conditions: alcohol/metal 1000/1 (mol/mol), 50 °C, pO2 =
300 kPa, 1250 rpm. TOF calculated after 15 min of reaction.
Selectivity at 90% conversion. Glyc = glycerate; Gly = glycolate; Tartr
= tartronate; Lac = lactate; Form = formate. Prepared in H2O at 1 °C.
b
ASSOCIATED CONTENT
■
c
S
* Supporting Information
a
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acscatal.5b00754.
a
Table 6. Recycling Results Using the B1 Au Catalyst
UV−vis spectra, TEM-HAADF images with supporting
Au particle size distribution histograms, EXAFS fitting
parameters, CO-chemisorption DRIFTS spectra, micro-
wave plasma-atomic emission spectroscopy and compu-
tational information and additional catalytic data for
glycerol oxidation (PDF)
c
selectivity (%)
a
b
B1
TOF (h−1
)
conv. (%) after 4 h Glyc Gly Tartr Lac
run 1
run 2
run 3
run 4
run 5
run 6
run 7
run 8
run 9
run 10
202
256
261
265
263
258
246
269
259
250
94
98
98
95
98
99
94
96
97
93
76
67
62
60
61
60
60
58
56
58
4
11
7
11
12
21
21
19
22
21
22
22
22
6
7
8
9
9
8
8
7
9
7
8
7
AUTHOR INFORMATION
■
7
Corresponding Authors
8
*E-mail: DimitratosN@cardiff.ac.uk,.
*E-mail: alberto.villa@unimi.it.
*E-mail: peter.wells@rc-harwell.ac.uk.
10
10
10
a
Reaction conditions: alcohol/metal 1000/1 (mol/mol), 50 °C, pO2 =
Notes
b
300 kPa, 1250 rpm. TOF calculated after 15 min of reaction.
The authors declare no competing financial interest.
c
Selectivity at 90% conversion. Glyc = glycerate; Gly = glycolate; Tartr
a
= tartronate; Lac = lactate; Form = formate. Prepared in H2O/EtOH
at −30 °C.
ACKNOWLEDGMENTS
■
The authors wish to acknowledge the Diamond Light Source
for provision of beamtime (SP8071). The RCaH are also
acknowledged for use of facilities and support of their staff.
Kristina Penman is thanked for performing the MP-AES
measurements. Dr. A. J. Logsdail is grateful to the UCL
Department of Chemistry and the Ramsay Memorial Trust for
provision of a Ramsay Fellowship. UK Catalysis Hub is kindly
thanked for resources and support provided via our member-
ship of the UK Catalysis Hub Consortium and funded by
EPSRC (Portfolio Grants EP I019693, EP/K014706/1, EP/
K014668/1, EP/K014854/1, and EP/K014714/1). The
authors acknowledge the use of the following high-performance
computing facilities, and associated support services, in the
completion of this work: ARCHER, via our membership of the
UK HPC Materials Chemistry Consortium (EP/L000202);
IRIDIS, provided by the Centre for Innovation; and UCL
Legion.
stability is enhanced by the presence of the protecting agent,
which is consistent with the lower activity observed for catalysts
prepared in solvents containing EtOH, in which the EtOH and
PVA influence the adsorption of glycerol. The catalytic
performance of A1 and A3 catalysts was investigated at lower
reaction temperature (25 °C) to minimize the aggregation of
small clusters. As is shown in Tables S6 and S7, the stability of
the catalysts was good, and selectivities of 81−87% to glycerate
were obtained. These results demonstrate the applicability of
supported gold catalysts in which the presence of gold clusters
can enhance catalytic activity and can also be successfully
reused without significant loss of activity.
CONCLUSIONS
■
This study demonstrates how Au NP characteristics can be
tuned using solvent and temperature variations during colloidal
synthesis and the impact this has on catalytic performance. The
choice of temperature and solvent system proved crucial in
dictating the performance for the liquid phase oxidation of
glycerol, with the colloid generated at 1 °C in a water-only
solvent giving the highest activity. This catalyst had a TOF of
915 h−1, which compares favorably with other Au/TiO2
catalysts in the literature that have been assessed under similar
reaction conditions. The characterization of this catalyst, A1,
REFERENCES
■
(1) Burda, C.; Chen, X.; Narayanan, R.; El-Sayed, M. A. Chem. Rev.
2005, 105, 1025−1102.
(2) Cushing, B. L.; Kolesnichenko, V. L.; O’Connor, C. J. Chem. Rev.
2004, 104, 3893−3946.
(3) Sun, Y.; Xia, Y. Science 2002, 298, 2176−2179.
(4) Schimpf, S.; Lucas, M.; Mohr, C.; Rodemerck, U.; Bruckner, A.;
̈
Radnik, J.; Hofmeister, H.; Claus, P. Catal. Today 2002, 72, 63−78.
4383
DOI: 10.1021/acscatal.5b00754
ACS Catal. 2015, 5, 4377−4384

ACS Catalysis
Research Article
(5) Enache, D. I.; Edwards, J. K.; Landon, P.; Solsona-Espriu, B.;
Carley, A. F.; Herzing, A. A.; Watanabe, M.; Kiely, C. J.; Knight, D. W.;
Hutchings, G. J. Science 2006, 311, 362−365.
(6) Carrettin, S.; McMorn, P.; Johnston, P.; Griffin, K.; Kiely, C. J.;
Hutchings, G. J. Phys. Chem. Chem. Phys. 2003, 5, 1329−1336.
(7) Biella, S.; Prati, L.; Rossi, M. J. Catal. 2002, 206, 242−247.
(8) Prati, L.; Martra, G. Gold Bull. 1999, 32, 96−101.
(9) Roucoux, A.; Schulz, J.; Patin, H. Chem. Rev. 2002, 102, 3757−
3778.
(43) Beale, A. M.; Weckhuysen, B. M. Phys. Chem. Chem. Phys. 2010,
12, 5562−5574.
(44) Szczerba, W.; Riesemeier, H.; Thunemann, A. F. Anal. Bioanal.
Chem. 2010, 398, 1967−1972.
(45) Dekkers, M. A. P.; Lippits, M. J.; Nieuwenhuys, B. E. Catal. Lett.
1998, 56, 195−197.
(46) Boccuzzi, F.; Chiorino, A.; Manzoli, M.; Andreeva, D.;
Tabakova, T. J. Catal. 1999, 188, 176−185.
(47) Panayotov, D. A.; Burrows, S. P.; Yates, J. T.; Morris, J. R. J.
Phys. Chem. C 2011, 115, 22400−22408.
(10) Kimling, J.; Maier, M.; Okenve, B.; Kotaidis, V.; Ballot, H.;
Plech, A. J. Phys. Chem. B 2006, 110, 15700−15707.
(11) Turkevich, J. Gold Bull. 1985, 18, 86−91.
(48) Dimitratos, N.; Villa, A.; Bianchi, C. L.; Prati, L.; Makkee, M.
Appl. Catal., A 2006, 311, 185−192.
(49) Prati, L.; Villa, A. Acc. Chem. Res. 2014, 47, 855−863.
(50) Villa, A.; Wang, D.; Veith, G. M.; Vindigni, F.; Prati, L. Catal.
Sci. Technol. 2013, 3, 3036−3041.
(12) Shi, H.; Xu, N.; Zhao, D.; Xu, B.-Q. Catal. Commun. 2008, 9,
1949−1954.
(13) Porta, F.; Prati, L.; Rossi, M.; Coluccia, S.; Martra, G. Catal.
Today 2000, 61, 165−172.
(14) Longenberger, L.; Mills, G. J. Phys. Chem. 1995, 99, 475−478.
(15) Ketchie, W. C.; Fang, Y. L.; Wong, M. S.; Murayama, M.; Davis,
R. J. J. Catal. 2007, 250, 94−101.
(16) Porta, F.; Prati, L. J. Catal. 2004, 224, 397−403.
(17) Ketchie, W.; Murayama, M.; Davis, R. Top. Catal. 2007, 44,
307−317.
(18) Bond, G. C.; Thompson, D. T. Catal. Rev.: Sci. Eng. 1999, 41,
319−388.
(19) Bonnemann, H.; Richards, R. M. Eur. J. Inorg. Chem. 2001, 2001,
2455−2480.
(20) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chem. Lett.
1987, 16, 405−408.
(21) Sakurai, H.; Haruta, M. Catal. Today 1996, 29, 361−365.
(22) Sakurai, H.; Tsubota, S.; Haruta, M. Appl. Catal., A 1993, 102,
125−136.
(23) Inami, S. H.; Wood, B. J.; Wise, H. J. Catal. 1969, 13, 397−403.
(24) Sinha, A. K.; Seelan, S.; Tsubota, S.; Haruta, M. Angew. Chem.,
Int. Ed. 2004, 43, 1546−1548.
(25) Hughes, M. D.; Xu, Y. J.; Jenkins, P.; McMorn, P.; Landon, P.;
Enache, D. I.; Carley, A. F.; Attard, G. A.; Hutchings, G. J.; King, F.;
Stitt, E. H.; Johnston, P.; Griffin, K.; Kiely, C. J. Nature 2005, 437,
1132−1135.
(26) Janssens, T. W.; Clausen, B.; Hvolbæk, B.; Falsig, H.;
Christensen, C.; Bligaard, T.; Nørskov, J. Top. Catal. 2007, 44, 15−26.
(27) Carrettin, S.; McMorn, P.; Johnston, P.; Griffin, K.; Hutchings,
G. J. Chem. Commun. 2002, 696−697.
(28) Carrettin, S.; McMorn, P.; Johnston, P.; Griffin, K.; Kiely, C.;
Attard, G.; Hutchings, G. Top. Catal. 2004, 27, 131−136.
(29) Demirel-Gulen, S.; Lucas, M.; Claus, P. Catal. Today 2005,
̈
102−103, 166−172.
(30) Sheldon, R. A.; Arends, I. W. C. E.; Dijksman, A. Catal. Today
2000, 57, 157−166.
(31) Dimitratos, N.; Lopez-Sanchez, J.; Lennon, D.; Porta, F.; Prati,
L.; Villa, A. Catal. Lett. 2006, 108, 147−153.
(32) Pritchard, J.; Piccinini, M.; Tiruvalam, R.; He, Q.; Dimitratos,
N.; Lopez-Sanchez, J. A.; Morgan, D. J.; Carley, A. F.; Edwards, J. K.;
Kiely, C. J.; Hutchings, G. J. Catal. Sci. Technol. 2013, 3, 308−317.
(33) Wang, Z. W.; Palmer, R. E. Nano Lett. 2012, 12, 91−95.
(34) Wang, Z. W.; Toikkanen, O.; Yin, F.; Li, Z. Y.; Quinn, B. M.;
Palmer, R. E. J. Am. Chem. Soc. 2010, 132, 2854−2855.
(35) Young, N. P.; Li, Z. Y.; Chen, Y.; Palomba, S.; Di Vece, M.;
Palmer, R. E. Phys. Rev. Lett. 2008, 101, 246103.
(36) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537−
541.
(37) Newville, M. J. Synchrotron Radiat. 2001, 8, 322−324.
(38) Blochl, P. E. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50,
̈
17953−17979.
(39) Mortensen, J. J.; Hansen, L. B.; Jacobsen, K. W. Phys. Rev. B:
Condens. Matter Mater. Phys. 2005, 71, 035109.
(40) Jiang, L.; Xu, Q. J. Phys. Chem. A 2005, 109, 1026−1032.
(41) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
3865−3868.
(42) Underwood, S.; Mulvaney, P. Langmuir 1994, 10, 3427−3430.
4384
DOI: 10.1021/acscatal.5b00754
ACS Catal. 2015, 5, 4377−4384