Metal particle growth during glucose hydrogenation over Ru/SiO2 evaluated by X-ray absorption spectroscopy and electron microscopy

Article abstract of DOI:10.1021/jp057022y

Biorenewable resources such as carbohydrates are considered alternative feedstocks for oxygenated chemicals. This work investigates the stability of silica-supported Ru catalysts in the aqueous phase conversion of glucose to sorbitol. In situ X-ray absorption spectroscopy at the Ru K edge revealed that air-exposed silica-supported Ru was in an oxidized state but was subsequently reduced in aqueous solutions saturated with 40 bar H₂ at 373 K. Furthermore, exposure to aqueous phase conditions resulted in the sintering of Ru particles on the silica surface. However, the presence of glucose in the aqueous phase stabilized the growth of the Ru particles. Batchwise hydrogenation of glucose at 373 K and 80 bar H2 over a Ru/SiO₂ (2.67 wt %) catalyst is nearly 100% selective to sugar alcohol with an average turnover frequency of 0.21 ± 0.04 s^-1. The hydrogenation reaction was not mass transfer limited according to the Madon-Boudart criterion.

Full text of DOI:10.1021/jp057022y

J. Phys. Chem. B 2006, 110, 7869-7876
7869
Metal Particle Growth during Glucose Hydrogenation over Ru/SiO2 Evaluated by X-ray
Absorption Spectroscopy and Electron Microscopy
Erin P. Maris,† William C. Ketchie,† Vladimir Oleshko,‡ and Robert J. Davis*,†
Department of Chemical Engineering and Department of Materials Science and Engineering, UniVersity of
Virginia, CharlottesVille, Virginia 22904-4741
ReceiVed: December 2, 2005; In Final Form: February 21, 2006
Biorenewable resources such as carbohydrates are considered alternative feedstocks for oxygenated chemicals.
This work investigates the stability of silica-supported Ru catalysts in the aqueous phase conversion of glucose
to sorbitol. In situ X-ray absorption spectroscopy at the Ru K edge revealed that air-exposed silica-supported
Ru was in an oxidized state but was subsequently reduced in aqueous solutions saturated with 40 bar H2 at
373 K. Furthermore, exposure to aqueous phase conditions resulted in the sintering of Ru particles on the
silica surface. However, the presence of glucose in the aqueous phase stabilized the growth of the Ru particles.
Batchwise hydrogenation of glucose at 373 K and 80 bar H2 over a Ru/SiO2 (2.67 wt %) catalyst is nearly
100% selective to sugar alcohol with an average turnover frequency of 0.21 ( 0.04 s-1. The hydrogenation
reaction was not mass transfer limited according to the Madon-Boudart criterion.
Introduction
of Ru particles. Thus, the choice of catalyst support is crucial
in the hydrogenation of glucose. Because of the inert nature of
The production of fuels and chemicals relies heavily upon
fossil-fuel-derived precursors that are finite in supply. Increased
effort to find alternatives to these fossil-based resources has
focused on species such as carbohydrates that are derived from
renewable biomass. Carbohydrates can be converted to dihy-
drogen, alkanes, oxygenated chemicals, and other industrially
important chemical intermediates.1-12 As a first step, the simple
sugar glucose can be hydrogenated to the sugar alcohol sorbitol
over a transition metal catalyst (Figure 1). Sorbitol is an
important intermediate used in the production of vitamin C.12
Sorbitol can also be converted to fuels such as H2 and alkanes
by aqueous phase reforming and transformed into lower
molecular weight polyols and acids through hydrogenoly-
sis.1,4,7,13 Hydrogenolysis products can be further converted to
fuels or commodity oxygenated chemicals.2,7
Traditionally, the hydrogenation of glucose to sorbitol utilized
nickel-based catalysts because of their low cost.14-20 Unfortu-
nately, leaching of nickel from these catalysts decreased hydro-
genation activity and posed a separation problem in downstream
processing. Efforts to find alternatives to nickel-based catalysts
have resulted in the exploration of cobalt-, platinum-, palladium-,
rhodium-, and ruthenium-based catalysts.15,21-23 Materials con-
taining Ru appear to be the most active hydrogenation catalysts,
and unlike nickel, they do not appear to leach into the reaction
medium.15,21,24-26 Deactivation of the catalysts, however, can
often be an issue.25 For example, Arena investigated the hydro-
genation of glucose over Ru/Al2O3, whereas Gallezot et al.
studied a Ru/C catalyst. Both groups concluded that deactivation
of the catalyst could be attributed, in part, to poisoning of the
Ru surface by metal impurities in solution. Arena also found
that byproduct formation and increased levels of alumina
crystallinity contributed to catalyst deactivation. Furthermore,
Gallezot et al. postulated that deactivation also involved sintering
silica, ruthenium catalysts supported on silica appear to be
superior to their titania-, alumina-, and carbon-supported
analogues for minimizing byproduct formation.15 However, the
stability of the silica can be compromised under aqueous reaction
conditions.27 This paper focuses on the use of in situ catalyst
characterization used to probe the stability of silica-supported
ruthenium catalysts for the conversion of glucose to sorbitol.
Experimental Methods
Materials Preparation. Various weight loadings of ruthe-
nium were supported on silica (Cab-O-Sil M-5, BET surface
area 196 m2 g-1) by incipient wetness impregnation involving
an aqueous solution of Ru(NO)(NO3)3 (Alfa-Aesar, 1.5% w/v).
Approximately 4.7 mL of the impregnating solution was used
per gram of silica. After impregnation, the catalysts were dried
overnight under ambient conditions, followed by drying in air
at 393 K for 2 h.
A carbon-supported sample (Ru/C) was prepared to be used
during X-ray absorption studies evaluating the effect of support
on the stability of Ru particles in an aqueous environment. The
Ru/C samples were prepared by cation exchange according to
the method published by Gallezot et al.26 Prior to the ion
exchange, 20 g of carbon powder (Alfa Aesar, BET surface
area 665 m2 g-1) was chemically oxidized by treatment with
380 mL of aqueous sodium hypochlorite solution (Acros, 5%
active chlorine) at ambient temperature. The NaClO solution
was added dropwise to the carbon. The resulting slurry was
stirred for approximately 24 h under ambient conditions. The
mixture was subsequently filtered and washed with 250 mL of
1 M HCl (Acros) to remove Na+. The oxidized carbon was
then washed repeatedly with distilled deionized water until the
filtrate tested negative for Cl- using a 0.1 M AgNO3 solution.
To prepare the Ru/C sample, the oxidized carbon support was
washed with 1 M aqueous ammonia solution (Alfa Aesar, 50%
v/v) while bubbling N2 through the mixture at ambient tem-
perature. An appropriate amount of Ru(NH3)6Cl3 (Acros) in 1
* Corresponding author. E-mail: rjd4f@virginia.edu.
† Department of Chemical Engineering.
‡ Department of Materials Science and Engineering.
10.1021/jp057022y CCC: $30.25 © 2006 American Chemical Society
Published on Web 03/25/2006

7870 J. Phys. Chem. B, Vol. 110, No. 15, 2006
Maris et al.
Figure 1. Reaction scheme of glucose hydrogenation to sorbitol.
Figure 3. Schematic of the flow system and sample cell for XAS
studies. “PG” denotes pressure gauge.
Figure 2. Schematic of the XAS sample cell.
M aqueous ammonia solution was added dropwise to the carbon
slurry. The resulting suspension was stirred at ambient temper-
ature in the presence of bubbling N2 for 24 h. The suspension
was filtered and washed with distilled deionized water until no
Cl- could be detected with a 0.1 M AgNO3 test. The sample
was dried overnight under ambient conditions, followed by
drying in air at 383 K for 2 h.
pressed into the 1.5 mm × 20 mm × 3 mm (path length) sample
holder. A 20 µm stainless steel frit retained the catalyst in the
flowing reaction mixture. The cell windows consisted of 1 mm
thick poly-ether-ether-ketone (PEEK) disks. A schematic of the
flow system and sample cell setup is shown in Figure 3. The
aqueous solution used as the reaction medium was first purged
with N2 and then pressurized with either N2 or H2 in a 75 mL
vessel. A reciprocating piston pump was used to circulate the
aqueous solution at a rate of 1.5 mL min-1. The solution was
first pumped through a recycle loop that bypassed the sample
cell for 90 min to allow the solution to reach H2 saturation before
introduction to the sample. The cell was heated and maintained
at 373 K using heating tape and a temperature controller. All
aqueous treatments were conducted for 3 h prior to data
collection.
The extended X-ray absorption fine structure (EXAFS) data
analysis was performed using the WinXAS software package.29
An average of four to six scans was used for analysis. The first-
shell Ru-Ru amplitude function and phase shift were extracted
from the Ru/Al2O3 (17.6 wt % Ru) sample. The average Ru
particle size was found to be greater than 10 nm, as determined
by H2 chemisorption and X-ray diffraction (XRD). The reference
sample was assumed to have the bulk crystallographic values
of the Ru hexagonal close-packed (hcp) structure with a first-
shell coordination number of 12 and nearest neighbor distance
of 2.68 Å. The EXAFS data were k1 and k3 weighted and then
Fourier transformed from 3 to 15.5-16.0 Å-1. Next, the Fourier
transforms were back-transformed over the first Ru-Ru shell
in R-space from 1.3 to 3.1 Å. Curve fitting was performed in
both R-space and k-space with k1 and k3 weighting, and the
average of these four values is reported herein. This procedure
results in an estimated accuracy within 10% for the coordination
number, (10% for the Debye-Waller factor, and (0.01 Å for
the interatomic distance.30
These materials were heated at 1 K min-1 in H2 (BOC,
99.999%, purified using a Matheson 8371V palladium hydrogen
purifier) flowing at 84 mL min-1 to 648 K and held at that
temperature for 3 h. The reduced samples were then cooled
under H2 and slowly exposed to air. These will be referred to
as the “as-prepared” samples. Ruthenium loading was evaluated
by Galbraith Laboratories (Knoxville, TN).
An alumina-supported Ru sample was prepared for use as a
reference in the X-ray absorption studies. This sample was
prepared by incipient wetness impregnation of a γ-alumina
support (Alfa Aesar, 99.97%, 80-120 m2 g-1) with an aqueous
solution of RuCl3‚xH2O (Alfa Aesar, 41.6 wt % Ru). Following
impregnation, the sample was dried in air overnight, followed
by drying in air at 393 K for 2 h. Subsequently, the sample
was heated in flowing H2 to 673 K at a rate of 4 K min-1 and
then held at 673 K for 4 h. The sample was cooled under H2
followed by a slow exposure to air.
X-ray Absorption Spectroscopy. In situ X-ray absorption
spectroscopy (XAS) studies were performed on beamline X-18B
at the National Synchrotron Light Source (2.8 GeV, ring currents
of 160-300 mA) at Brookhaven National Laboratory, Upton,
NY. All XAS data were collected in transmission mode. A slit
size of 0.3 mm × 10 mm was used which allowed for 3.5 eV
resolution at the Ru K edge (22.118 keV).
An XAS sample cell similar to that found in ref 28 was
constructed to allow for liquid flow across the catalyst at the
elevated pressures and temperatures required for glucose
hydrogenation. A schematic of the cell is shown in Figure 2.
The catalyst samples (particle size 53-75 µm) were hand-
Transmission Electron Microscopy. The catalyst particle
size and morphology were examined by bright-field and dark-

Metal Particle Growth during Glucose Hydrogenation
J. Phys. Chem. B, Vol. 110, No. 15, 2006 7871
TABLE 1: Comparison of the Reaction Rates Observed
after the Various Pretreatments to the Standard Rate of
Hydrogenation over a 4.47 wt % Ru/SiO2 Catalyst
path
rate (molsorbitol produced(moltotal Ru) )
-1 s-1
standard reaction
0.17 ( 0.02
0.09 ( 0.03
0.13 ( 0.01
0.08 ( 0.01
1a
1b
2
During the reaction, consumption of H2 resulted in a decrease
in pressure of up to 20 bar; however, following the periodic
sampling of the reactor contents, the pressure was returned to
80 bar using H2. Upon completion of the reaction, the vessel
was allowed to cool to ambient temperature and the pressure
was released. Product solutions were filtered through a series
of Pall Gelman 0.45 and 0.20 µm syringe filters. The filtrate
was analyzed by either a Waters, model 501, High Performance
Liquid Chromatograph (HPLC) or a Waters, model 717, plus
Autosampler HPLC equipped with a Waters, model R401,
differential refractometer. Both an Aminex fermentation moni-
toring column (Bio-Rad) and an Aminex HPX-87H column
(Bio-Rad) were used for separation of the products. The columns
and refractive index detector were maintained at 333 and 304
K, respectively. The mobile phase consisted of 5 mM sulfuric
Figure 4. Schematic diagram representing kinetic studies performed
under various pretreatments.
field transmission electron microcopy (TEM), selected-area
electron diffraction (SAED), and high-resolution TEM (HR-
TEM) using a JEOL 2010F analytical electron microscope
(AEM) equipped with a Schottky field-emission (FE) gun and
a high-tilt pole piece (spherical aberration coefficient of Cs )
1.0 mm and 0.23 nm point-to-point resolution) and operated at
200 kV. Energy-dispersive X-ray spectroscopic (EDXS) analy-
ses were performed with a JEOL 2010F FE-AEM, which has
an Oxford Instruments ultrathin-window EDX detector with a
136 eV resolution at the Mn KR line (5.9 keV) and nanoprobe
capabilities with a probe diameter varying from 2.4 to 0.5 nm.
Samples were tilted 15° toward the detector. The samples were
supported on copper mesh microscopy grids with holey carbon
films by dipping the grid into the catalyst powder and shaking
the excess. Ruthenium particle size distributions both for the
as-prepared 4.47 wt % Ru/SiO2 catalyst and the same catalyst
recovered after hydrogenation were determined using the
ImagePro software, version 5.1 (Media Cybernetics, Inc., Silver
Spring, MD).
Adsorption of H2. The ruthenium dispersion and particle size
of the Ru/SiO2 catalysts were determined by H2 chemisorption
performed on a Coulter Omnisorp3 100CX instrument at 303
K in the pressure range 0-150 Torr. Prior to chemisorption,
the catalysts were heated from ambient conditions to 648 K at
4 K min-1 in flowing H2 (BOC 99.999%). The catalysts were
reduced in H2 at 648 K for 75 min prior to cooling to 303 K.
Surface Ru was evaluated by the total amount of H2 adsorbed
at 303 K extrapolated to zero pressure, assuming a stoichiometry
ratio (H/Ru) equal to unity.
Hydrogenation Procedure. The hydrogenation of glucose
to sorbitol was carried out in a 50 mL Parr Instrument Company
4592 batch reactor. In the standard reaction, 30 mL of a 50 wt
% aqueous glucose (Alfa Aesar, 99%) stock solution was loaded
into the reactor along with 0.05 g (particle size 53-180 µm) of
the as-prepared catalyst. The reactor was fitted with a dip tube
and a 10 µm stainless steel filter to allow for periodic liquid
sampling of the reaction mixture. To remove air in the
headspace, N2 (BOC, 99.998%) was bubbled through the reactor
for 30 min at ambient temperature. The vessel was then sealed
and heated at a rate of 2 K min-1 to 373 K. After reaching 373
K, the vessel was flushed three times with 5 bar H2 to remove
the N2. A liquid sample was removed from the vessel im-
mediately after flushing to mark the start of the reaction. After
sampling, H2 was added to a total pressure of 80 bar and the
reactor was stirred for 3 h. Samples of the reaction mixture were
removed periodically to track the progress of the hydrogenation.
acid at a flow rate of 0.5 mL min-1
.
Results and Discussion
To explore the stability of Ru/SiO2 catalysts during the
aqueous phase hydrogenation of glucose, three different tech-
niques were employed. In the first, a 4.47 wt % Ru/SiO2 catalyst
was subjected to various pretreatments after which the pretreated
catalysts were used for a hydrogenation reaction. The global
rate of hydrogenation was measured over these pretreated
catalysts and compared to the rate over an untreated catalyst
during the standard hydrogenation reaction described in the
Experimental Section. In a second technique, in situ X-ray
absorption spectroscopy was used to monitor how the pretreat-
ment conditions employed during the kinetic studies affect the
Ru oxidation state and particle size. Last, transmission electron
microscopy was used to visually track changes in the catalyst
before and after use in a standard hydrogenation reaction.
Pretreatment Studies. As summarized in Figure 4, these
pretreatments probed the influence of initial Ru oxidation state
and aqueous environment on the rate of hydrogenation (moles
of sorbitol produced(total moles of Ru)-1 s-1). Comparison of
paths 1a and 1b to path 2 examines the effect of initial Ru
oxidation state on the rate of glucose hydrogenation. Comparison
of paths 1b and 2 to path 1a tests the effect of glucose in the
aqueous medium during pretreatment. In paths 1b and 2, the
catalyst was pretreated in a glucose/water solution for 3 h prior
to hydrogenation. In path 1a, the aqueous pretreatment involves
only H2-saturated water. The reaction rates observed after the
various pretreatments are compared to that observed during the
standard hydrogenation reaction, which was 0.17 ( 0.02
-1
molsorbitol produced(moltotal Ru
)
s-1 over a 4.47 wt % Ru/SiO2
catalyst. As reported in Table 1, the various pretreatments
lowered the rate of glucose hydrogenation per Ru metal atom.
The rates of reaction after pretreating the catalysts along paths
1a and 2 were nearly half that of the standard reaction, which
suggests that the pretreatment conditions lowered the availability
of active Ru by half.
In some cases, an induction period was observed in the
conversion of glucose with time, as depicted in Figure 5. The
induction period was clearly observed for the pretreatment
described in path 2. As the pretreatment conditions appear to
alter the availability of Ru on the silica surface, which may

7872 J. Phys. Chem. B, Vol. 110, No. 15, 2006
Maris et al.
Figure 5. Conversion of glucose to sorbitol over 4.47 wt % Ru/SiO2
at 373 K and 80 bar H2 following the pretreatment paths shown in
Figure 4. The error bars represent 95% confidence limits.
result in the observed induction periods, detailed XAS studies
were used to explore the influence of pretreatments on the state
of the Ru.
X-ray Absorption Spectroscopy. In situ XAS studies were
performed in a high-pressure cell to mimic the conditions of
the pretreatment paths in Figure 4. Figure 6 presents the X-ray
absorption near edge structure (XANES) of the 4.47% Ru/SiO2
catalyst before and after various pretreatments. In addition,
spectra of RuO2 and large clusters of Ru/Al2O3 (17.6 wt %)
were used as references.
The X-ray absorption near edge spectra in Figure 6 confirm
that Ru is oxidized on the as-prepared 4.47 wt % Ru/SiO2
sample exposed to air. Treatment of the sample in H2 at 373 K
reduced the Ru to metal particles in the presence or absence of
liquid water. Reduction was also accomplished by H2 in a
glucose/water mixture. Thus, the chemical state of Ru during
the hydrogenation reaction is metallic, regardless of the pre-
treatment conditions. Nevertheless, the induction period ob-
served after pretreatment path 2 suggests a slow reduction
process after prolonged exposure to the glucose/water solution.
The extended X-ray absorption fine structure (EXAFS)
associated with each of the spectra shown in Figure 6 was
analyzed to determine the Ru coordination number (N), Ru-
Ru interatomic distance (R), change in Debye-Waller factor
(∆σ2), and shift in edge energy (∆E0). The Fourier transforms
of the EXAFS are given in Figure 7. EXAFS parameters derived
from curve fitting are provided in Table 2. The ∆σ2 value was
calculated relative to the metallic Ru/Al2O3 reference.
A plausible explanation for the decrease in the global
hydrogenation rate with treatment in water at 373 K is a decrease
in the availability of surface Ru. Analysis of the Ru K edge
EXAFS of 4.47 wt % Ru/SiO2 reduced by H2(g) at 373 K
revealed a Ru-Ru coordination number of 8.9 (Table 2), which
corresponds to an average spherical particle size of 1.5-2.5
nm. However, when H2-saturated water at 373 K and 40 bar
was added to the reduced catalyst, the coordination number
increased to 11, indicating substantial particle growth. The
EXAFS results summarized in Table 2 show that particle growth
in water occurred regardless of the initial Ru oxidation state.
Interestingly, the presence of glucose in the water inhibited
particle growth. The as-prepared catalyst treated at 373 K with
H2-saturated glucose/water solution revealed a coordination
number of 8.6, which is similar to the sample reduced by H2(g)
at 373 K. However, a sample prereduced with H2(g) and then
Figure 6. X-ray absorption near edge spectra of 4.47 wt % Ru/SiO2
at the Ru K edge: (R1) RuO2 reference; (A) as-prepared catalyst; (B)
reduced in H2(g); (C) reduced in H2(g) followed by the addition of
H2-saturated H2O; (D) as-prepared catalyst reduced by H2-saturated
H2O; (E) as-prepared catalyst reduced by the addition of H2-saturated
glucose/H2O; (F) reduced in H2(g) followed by the addition of H2-
saturated glucose/H2O; (R2) Ru/Al2O3 reference. The spectra are offset
for clarity. The energies are defined relative to the first inflection point
of the Ru/Al2O3 (R2) sample.
exposed to H2-saturated glucose/water solution had a coordina-
tion number of 10, which is consistent with some particle growth
that is not as severe as a treatment in H2-saturated water.
Evidently, aqueous phase glucose helps stabilize the oxidized
Ru from sintering into large metal particles upon reduction.
During the standard hydrogenation procedure, the as-prepared
catalyst was not prereduced prior to the addition of glucose/
water solution. Therefore, the highest global reaction rate
obtained with this procedure is the direct result of higher Ru
dispersion on the support. The various pretreatment conditions
facilitate particle growth and decreased availability of surface
Ru, which leads to lower global reaction rates.
Transmission Electron Microscopy. To complement the
XAS studies, TEM was used to monitor the size and distribution
of Ru particles on one sample before and after use in the
hydrogenation reaction. Parts a and b of Figure 8 show
micrographs of the 4.47 wt % Ru/SiO2 catalyst as prepared and
recovered by filtration after a standard hydrogenation reaction,

Metal Particle Growth during Glucose Hydrogenation
J. Phys. Chem. B, Vol. 110, No. 15, 2006 7873
right inset in Figure 8b shows 0.23 nm (100) hcp Ru lattice
fringes of the aggregated metal particles. EDXS analyses of
individual 3-5 nm Ru particles performed using 0.5-2.4 nm
diameter electron probes in randomly chosen areas of both the
as-prepared catalyst and the catalyst after hydrogenation showed
that the particles were highly pure, with no impurities or
admixtures.
Although Ru particle growth was not observed in the EXAFS
for an as-prepared catalyst treated with H2-saturated glucose/
water solution, severe restructuring of the catalyst was observed
by TEM. Figure 8a shows the as-prepared 4.47 wt % Ru/SiO2
sample with well-distributed Ru particles on the silica support
with a mean particle size of about 3.0 ( 0.1 nm. After reaction,
the mean Ru particle size was essentially unchanged, but the
distribution of Ru was severely altered. The sample appears to
contain large aggregate regions of high Ru particle density.
Although glucose/water solution inhibited particle growth as
determined by EXAFS, the metal particles were clearly mobile
on the support, which suggests a poor long-term stability of
the catalyst under these reaction conditions.
Sintering of supported metal catalysts is usually viewed as
occurring by one of the following mechanisms: (1) migration
of crystallites on the surface of the support, (2) migration of
atomic species on the support, and (3) dissolution and redepo-
sition of metal ions.31 The first two mechanisms are commonly
considered with high-temperature reactions, while the third
mechanism occurs in liquid phase reactions at relatively low
temperatures. The results obtained in this work support the
findings of Hoang et al., who showed that sintering of silica-
supported Ru catalysts occurred in liquid water under 60 bar
H2 at 373 K in an autoclave reactor.27 They found that Ru
particles of varying initial dispersions (15-86%) sintered to a
final dispersion of ∼10% after 4 h in liquid water and speculated
that the mechanism for agglomeration was the formation of a
soluble Ru(OH)x species that readsorbed and reduced on larger
Ru particles. Because of the flow-through design of the XAS
cell used in our work, it is expected that a percentage of the
soluble species would eventually be washed out of the catalyst
bed before readsorbing to form larger particles. By tracking the
X-ray absorption coefficient (which is directly proportional to
the amount of Ru present) during the XAS experiments, it is
possible to monitor the Ru content. Analysis of all of the
progressive scans during each aqueous treatment showed no
decrease in the ∆µx value at the Ru K edge. This finding
suggests that sintering did not occur via a soluble species but
more likely occurred by migration of the Ru during hydrolysis
of the silica. Indeed, Hoang et al. report a loss of silica surface
area by treatment in water.27 Inspection of the histograms shown
in Figure 8 reveals that the average Ru particle size did not
change over the course of the reaction. Instead, Ru particles
appeared to migrate on the surface of the silica to regions of
higher particle density. This migration occurred over the course
of a 3 h hydrogenation reaction. One could imagine that, given
additional exposure to the glucose/water solution, the Ru
particles would coalesce to form larger particles. This theory is
supported by the behavior of path 2 in Figure 4. After the initial
induction period, the rate of hydrogenation appears to be
comparable to that of the standard hydrogenation reaction. This
suggests that Ru particles coalesced during the extended glucose/
water treatment, subsequently taking a longer time to reduce.
In fact, when the hydrogenation reaction was conducted over a
period of 6 h using half the amount of catalyst compared to the
Figure 7. Magnitude of the radial distribution function (not corrected
for phase shifts) after Fourier transformation (k3 weighting) for the 4.47
wt % Ru/SiO2 under various aqueous pretreatments: (A) as-prepared
catalyst; (B) reduced in H2(g); (C) reduced in H2(g) followed by the
addition of H2-saturated H2O; (D) as-prepared followed by the addition
of H2-saturated H2O; (E) reduced in H2(g) followed by the addition of
H2-saturated glucose/H2O; (F) as-prepared reduced by the addition of
H2-saturated glucose/H2O. The transforms are offset for clarity.
TABLE 2: Summary of the First-Shell Ru-Ru
Coordination Number (N), Ru-Ru Interatomic Distance (R),
Change in Debye-Waller Factor (∆σ2) Relative to Metallic
Ru/Al2O3, and Shift in Edge Energy (∆E0) Obtained for the
Catalysts under Various Treatmentsa
treatment
N
R (Å)
∆σ2 (Å2)
∆E0 (eV)
A
B
C
D
E
F
G
H
3.7
8.9
11
11
10
8.6
2.68
2.66
2.68
2.68
2.68
2.67
2.66
2.65
0.0042
0.0032
0.0018
0.0014
0.0019
0.0031
0.0021
0.0032
2.8
-1.4
0.9
0.5
-1.6
-2.2
-2.7
-1.0
9.5
9.3
a (A-F) 4.47 wt % Ru/SiO2; (G and H) 12.98 wt % Ru/Cex. (A)
As-prepared catalyst; (B) reduced in H2(g); (C) reduced in H2(g)
followed by the addition of H2-saturated H2O; (D) as-prepared reduced
by the addition of H2-saturated H2O; (E) reduced in H2(g) followed by
the addition of H2-saturated glucose/H2O; (F) as-prepared reduced by
the addition of H2-saturated glucose/H2O; (G) as-prepared Ru/C reduced
in H2(g); (H) as-prepared Ru/C reduced in 40 bar H2-saturated H2O.
All XAS data were collected at 373 K, except for samples A and G,
which were collected at 298 K.
respectively. Shown beside the micrographs are the correspond-
ing differential histograms of Ru particle size, which match well
to a normal distribution curve. For the as-prepared catalyst, the
size of fairly well dispersed rounded particles varied from 1.6
to 4.8 nm with a mean roundness of 1.18. In a typical aggregate
formed during hydrogenation, as shown in Figure 8b, the particle
sizes were slightly higher, varying from 1.9 to 5.4 nm with a
mean roundness of 1.12. However, in both cases, the mean of
the particle size was estimated as 3.0 ( 0.1 nm. Although the
SiO2 support could mask to some extent the contrast arising
from ultrafine Ru particles, dark-field TEM observations (dif-
fraction contrast) in conjunction with SAED and HRTEM (phase
contrast) indicated their crystallinity. Thus, the enlarged upper
standard reaction, the resulting overall rate of reaction was 0.08
-1
( 0.01 molsorbitol produced(moltotal Ru
)
s-1. This overall rate

7874 J. Phys. Chem. B, Vol. 110, No. 15, 2006
Maris et al.
Figure 8. High-magnification TEM images of 4.47 wt % Ru/SiO2 with the corresponding histograms of Ru particle size: (a) as-prepared catalyst;
(b) after standard hydrogenation reaction, demonstrating changes in the surface distribution and aggregation of 3 nm Ru particles. The aggregate
contains about 90 particles. The enlarged upper right inset in part b shows (100) hcp Ru lattice fringes. Continuous lines in histograms denote
normal distribution curves.
matches that observed after the pretreatment of path 2 in Figure
4, further supporting the conclusion that coalescence of Ru
particles can eventually lead to sintering of these particles. The
TEM and XAS studies in combination with the kinetics studies
suggest that sintering of Ru particles occurred via the migration
of crystallites on the support.
samples. This also supports the conclusion that there was no
dissolution and redeposition of Ru ions under the aqueous
conditions. Furthermore, kinetic studies on a separate com-
mercial Ru/C (Acros, 5 wt % Ru) catalyst show that, contrary
to the Ru/SiO2 catalyst, the rate of hydrogenation does not
decrease with extended reaction time. The overall rate of the
To explore the effect of support on the stability of Ru particles
under aqueous phase reaction conditions, a carbon-supported
Ru sample was also examined by XAS. Table 2 contains the
structural properties of Ru/C that was reduced in either gaseous
H2 at 373 K or H2-saturated liquid water at 373 K and 40 bar
total pressure. The Ru-Ru coordination number remained
essentially unchanged during these treatments, indicating that
Ru particle growth did not occur on the carbon-supported
standard hydrogenation was 0.12 ( 0.02 and 0.11 ( 0.01
-1
molsorbitol produced(moltotal Ru
)
s-1 after 3 and 6 h of reaction,
respectively. This indicates that Ru particles are more stable
on a carbon support and less likely to sinter under aqueous phase
reaction conditions when compared to the silica-supported
counterparts.
As sintering of Ru particles occurred on the silica-supported
sample but not on the carbon-supported sample, we concluded

Metal Particle Growth during Glucose Hydrogenation
J. Phys. Chem. B, Vol. 110, No. 15, 2006 7875
TABLE 3: Results of H2 Chemisorption Providing an
Estimate of Ru Dispersion and Particle Size
catalyst Ru wt % dispersion (%) particle sizea (nm) TOFb (s-1
)
Ru/SiO2
Ru/SiO2
Ru/SiO2
Ru/Cc
0.56
2.67
4.47
5d
91
78
53
43
1.1
1.3
1.9
2.4
0.28 ( 0.02
0.21 ( 0.04
0.32 ( 0.13
0.25 ( 0.03
a Estimated from H2 chemisorption (1/dispersion). b TOFs given with
95% confidence intervals. c Commercial catalyst purchased from Acros
Organics. d Ru loading provided by Acros Organics.
that particle growth depends on catalyst support, not catalyst
metal. Similar results have been found for supported Pt catalysts.
Douidah et al. compared the stability of silica- and alumina-
supported Pt catalysts under H2 in the aqueous phase at varying
pH.32 They found a decrease in Pt dispersion on the silica but
not on the alumina following an aqueous treatment. It was
concluded that, on the silica-supported catalysts, the loss of
metal dispersion was due to Pt particle growth. In a study of
small Ru particles deposited on a variety of supports, it was
found that the strength of the metal-support interaction was
inversely proportional to the electronegativity of the support
metal ion.33 A strong interaction between the support metal and
support oxygen results in a weak interaction between the Ru
metal and the support oxygen, resulting in a weakly anchored
catalyst. In a similar study, small Ru particles were prepared
on a range of supports that were analyzed with XAS.34 Analysis
of the distance of the Ru to support oxygen showed that the
Ru-O distance for silica was longer than those for Ru supported
on alumina and magnesia. Treating the bond distance as being
inversely proportional to the bond strength would indicate that
the silica-supported catalyst is more weakly anchored on the
silica support compared to the alumina and magnesia supports.
The results presented in this work are in agreement with the
idea that Ru particles are weakly anchored to the silica support.
The XAS studies have shown that the addition of glucose to
the reaction media slowed the rate of particle growth on the
Ru/SiO2 catalyst. The mechanism by which glucose stabilizes
the catalyst is not known but is the subject of further study.
As the presence of glucose stabilizes the Ru particles on the
as-prepared Ru/SiO2 catalyst, a true kinetic study of the use of
these catalysts in glucose hydrogenation was feasible. A series
of three silica-supported catalysts was prepared with different
Ru loadings; Table 3 summarizes the Ru loading and dispersion
determined from H2 chemisorption for each of the samples. The
inverse of the dispersion provides a simple estimate for the mean
Ru particle size.35 As shown in Table 3, an increase in Ru
loading corresponded to a decrease in Ru dispersion.
Figure 9. Percent conversion of glucose to sorbitol over Ru/SiO2 (2.67
wt %) as a function of time at 373 K and 80 bar H2. The line represents
a linear regression of the data. The error bars represent 95% confidence
limits.
the same TOF as Ru/C (Acros), which is among the highest
reported in the literature. Gallezot et al. studied the hydrogena-
tion of glucose to sorbitol over a 1.6 wt % Ru/C catalyst in a
trickle-bed reactor under similar reaction conditions of 373 K
and 80 bar H2 pressure.26 Assuming 100% dispersion of Ru
based on TEM results, the authors reported a TOF of 0.03 s-1
.
The lower TOF obtained by Gallezot et al. may be due to
internal mass transfer limitations. Kusserow et al. studied the
batchwise hydrogenation of glucose over a 2.75 wt % Ru/Al2O3
catalyst at 393 K and 120 bar H2 pressure.15 On the basis of a
Ru particle size of 1.0 nm, as determined by TEM (100%
dispersion), the TOF for glucose hydrogenation was ap-
proximately 0.08 s-1
.
The Madon-Boudart criterion was applied in our study to
determine if mass transfer artifacts in the kinetic data were
significant during the hydrogenation reaction.37 Thus, glucose
hydrogenation was performed over a series of silica-supported
catalysts with various Ru loadings. As shown in Table 3, the
concentration of Ru varied by nearly an order of magnitude,
while the Ru dispersion changed by only a factor of 2. The
turnover frequency of the hydrogenation reaction was nearly
constant over this series of catalysts, which, according to the
Madon-Boudart criterion, indicates that mass transfer artifacts
are negligible in this system. Thus, the rate of hydrogenation
should be a good measure of available surface Ru atoms in the
reactor.
The primary tool for characterization of these catalyst samples
was H2 chemisorption, as shown in Table 3. However, EXAFS
analysis and TEM studies were used to corroborate the H2
chemisorption results for the 4.47 wt % Ru/SiO2 catalyst. A
Ru particle size of 1.9 nm was estimated by H2 chemisorption,
which is in fair agreement with the estimate of 3.0 ( 0.1 nm
obtained by TEM. Following reduction of the catalyst in H2-
(g), the EXAFS analysis yielded a Ru coordination number of
8.9 (Table 2) which corresponds to an average spherical particle
size of about 1.5-2.5 nm.36 Agreement between the results of
H2 chemisorption, TEM, and EXAFS analysis indicates that H2
chemisorption is an appropriate tool for the estimate of Ru
particle size.
Conclusions
This work has shown that, in the presence of H2-saturated
water, silica-supported Ru catalysts were unstable due to
sintering of the Ru particles. This sintering occurred regardless
of the initial Ru oxidation state. When the catalyst was reduced
prior to treatment with a H2-saturated glucose/water solution,
there was also growth of Ru particles. However, the extent of
particle growth did not appear as significant as that in a purely
aqueous environment. The presence of glucose in a H2-saturated
aqueous solution appeared to stabilize initially oxidized Ru
particles against sintering. These findings were important, as
they closely mimicked the conditions for the standard hydro-
genation reaction. Transmission electron microscopy confirmed
that Ru particles did not sinter during glucose hydrogenation;
however, there was migration of the Ru, forming regions of
higher particle density. Comparison of the XAS and TEM
As shown in Figure 9, the conversion of glucose to sorbitol
was linear with time over the first 3 h. The average turnover
frequency (TOF) for glucose hydrogenation is presented in Table
3 for each of the catalysts. The Ru/SiO2 tested in our work had

7876 J. Phys. Chem. B, Vol. 110, No. 15, 2006
Maris et al.
(10) Ritter, S. K. Chem. Eng. News 2004, 82, 31.
studies with the kinetic studies suggests that glucose can only
stabilize Ru against sintering for a limited period of time (∼3
h as performed in these studies). Extended exposure (∼6 h) of
the Ru/SiO2 catalyst to aqueous conditions resulted in growth
of Ru particles and consequently a reduced rate of glucose
hydrogenation. It is likely that the mechanism of particle growth
involves migration of Ru species during the hydrolysis of the
silica surface. While Ru-based catalysts are effective hydrogena-
tion catalysts, supports other than silica should be considered
for the aqueous phase hydrogenation of glucose.
(11) Lichtenthaler, F. W.; Mondel, S. Pure Appl. Chem. 1997, 69, 1853.
(12) Crezee, E.; Hoffer, B. W.; Berger, R. J.; Makkee, M.; Kapteijn,
F.; Moulijn, J. A. Appl. Catal., A 2003, 251, 1.
(13) Tronconi, E.; Ferlazzo, N.; Forzatti, P.; Pasquon, I.; Casale, B.;
Marini, L. Chem. Eng. Sci. 1992, 47, 2451.
(14) Phillips, M. A. Br. Chem. Eng. 1963, 8, 767.
(15) Kusserow, B.; Schimpf, S.; Claus, P. AdV. Synth. Catal. 2003, 345,
289.
(16) Turek, F.; Chakrabarti, R. K.; Lange, R.; Geike, R.; Flock, W.
Chem. Eng. Sci. 1983, 38, 275.
(17) Dechamp, N.; Gamez, A.; Perrard, A.; Gallezot, P. Catal. Today
1995, 24, 29.
(18) Gallezot, P.; Cerino, P. J.; Blanc, B.; Fleche, G.; Fuertes, P. J. Catal.
1994, 146, 93.
Acknowledgment. This work was supported by the National
Science Foundation (CTS-0313484). Research was carried out
in part at the National Synchotron Light Source, Brookhaven
National Laboratory, which is supported by the U.S. Department
of Energy, Division of Materials Sciences and Division of
Chemical Sciences, under Contract No. DE-AC02-98CH10886.
V.O. gratefully acknowledges the support by the Director, Office
of Science, Division of Materials Science & Engineering, U.S.
Department of Energy, under Contract No. DE-FG02-01ER45918.
(19) Li, H.; Wang, W.; Fa Deng, J. J. Catal. 2000, 191, 257.
(20) Li, H.; Li, H.; Deng, J. F. Catal. Today 2002, 74, 53.
(21) Wisniak, J.; Simon, R. Ind. Eng. Chem. Prod. Des. DeV. 1979, 18,
50.
(22) Li, H.; Li, H.; Wang, M. Appl. Catal., A 2001, 207, 129.
(23) Li, H.; Deng, J.-F. J. Chem. Technol. Biotechnol. 2001, 76, 985.
(24) van Gorp, K.; Boerman, E.; Cavenaghi, C. V.; Berben, P. H. Catal.
Today 1999, 52, 349.
(25) Arena, B. J. Appl. Catal., A 1992, 87, 219.
(26) Gallezot, P.; Nicolaus, N.; Fleche, G.; Fuertes, P.; Perrard, A. J.
Catal. 1998, 180, 51.
(27) Hoang, L. C.; Menezo, J. C.; Montassier, C.; Barbier, J. Bull. Soc.
Chim. Fr. 1991, 128, 491.
References and Notes
(28) Bowron, D. T.; Weigel, R.; Filipponi, A.; Roberts, M. A.; Finney,
J. L. Mol. Phys. 2001, 99, 761.
(29) Ressler, T. WinXAS, version 2.33, 2001; www.winxas.de.
(30) Koningsberger, D. C.; Mojet, B. L.; Dorssen, G. E. v.; Ramaker,
D. E. Top. Catal. 2000, 10, 143.
(31) Besson, M.; Gallezot, P. Catal. Today 2003, 81, 547.
(32) Douidah, A.; Marecot, P.; Barbier, J. Appl. Catal., A 2002, 225,
11.
(33) Asakura, K.; Iwasawa, Y. J. Chem. Soc., Faraday Trans. 1990,
86, 2657.
(34) Mizushima, T.; T., K.; Udagawa, Y.; Ueno, A. J. Am. Chem. Soc.
1990, 112, 7887.
(1) Cortright, R. D.; Davda, R. R.; Dumesic, J. A. Nature 2002, 418,
964.
(2) Davda, R. R.; Dumesic, J. A. Angew. Chem., Int. Ed. 2003, 42,
4068.
(3) Davda, R. R.; Shabaker, J. W.; Huber, G. W.; Cortright, R. D.;
Dumesic, J. A. Appl. Catal., B 2005, 56, 171.
(4) Huber, G. W.; Cortright, R. D.; Dumesic, J. A. Angew. Chem., Int.
Ed. 2004, 43, 1549.
(5) Huber, G. W.; Shabaker, J. W.; Dumesic, J. A. Science 2003, 300,
2075.
(6) Embree, H. D.; Chen, T.; Payne, G. F. Chem. Eng. J. 2001, 84,
133.
(35) Boudart, M.; Djega-Mariadassou, G. Kinetics of Heterogeneous
Catalytic Reactions; Princeton University Press: Princeton, NJ, 1984.
(36) Greegor, R. B.; Lytle, F. W. J. Catal. 1980, 63, 476.
(37) Madon, R. J.; Boudart, M. Ind. Eng. Chem. Fundam. 1982, 21,
438.
(7) Cortright, R. D.; Sanchez-Castillo, M.; Dumesic, J. A. Appl. Catal.,
B 2002, 39, 353.
(8) Chen, N. Y.; Degnan, T. F.; Koenig, L. R. CHEMTECH 1986,
506.
(9) Hollingsworth, R. I. Chem. Eng. 2001, 66.