Influence of the cobalt particle size in the CO hydrogenation reaction studied by in situ X-ray absorption spectroscopy

Article abstract of DOI:10.1021/jp901602s

The influence of particle size in the carbon monoxide hydrogenation reaction has been studied using cobalt nanoparticles (NPs) with narrow size distribution prepared from colloidal chemistry. The surfactant covering the NPs after synthesis could be remove

Full text of DOI:10.1021/jp901602s

J. Phys. Chem. B 2009, 113, 10721–10727
10721
Influence of the Cobalt Particle Size in the CO Hydrogenation Reaction Studied by In Situ
X-Ray Absorption Spectroscopy
†
†
†
‡
,†,§
Tirma Herranz, Xingyi Deng, Andreu Cabot, Jingua Guo, and Miquel Salmeron*
Materials Sciences DiVision, Lawrence Berkeley National Laboratory, AdVance Light Source, Lawrence
Berkeley National Laboratory, Materials Science and Engineering Department, UniVersity of California,
Berkeley, California 94720
ReceiVed: February 20, 2009; ReVised Manuscript ReceiVed: April 23, 2009
The influence of particle size in the carbon monoxide hydrogenation reaction has been studied using cobalt
nanoparticles (NPs) with narrow size distribution prepared from colloidal chemistry. The surfactant covering
the NPs after synthesis could be removed by heating to 200-270 °C in H
2
. Soft X-ray absorption spectroscopy
was performed using a gas flow cell under reaction conditions of H and CO at atmospheric pressure. Flow
2
of pure hydrogen at 350 °C removed the protecting surfactant layer and reduced the NPs from oxidized to
metallic. The NPs remained metallic during the methanation reaction with their surface covered by CO. The
methanation turnover frequency of silica-supported NPs was found to decrease with diameter below 10 nm,
whereas the reaction activation energy was found to be independent of NP size. H-D exchange experiments
indicated that it is the dissociation of H
2
that is responsible for the observed decrease in activity with size.
1
. Introduction
Cobalt nanoparticles (NPs) have found applications in data
are masked when the size distribution of the Co particles is
broad, as typically obtained when prepared by impregnation
techniques.
storage devices and sensors, thanks to their magnetic properties.1
,2
Knowledge of the chemical state of the NP and of the species
adsorbed during the FT reaction is also lacking, thus preventing
a complete understanding of the catalytic process. This is due
to a lack of suitable characterization techniques that are able to
characterize the chemical and electronic structure of the catalyst
during reaction. To remedy this, we use X-ray absorption
spectroscopy (XAS), a synchrotron-based technique for studying
the electronic structure of matter. The X-rays can easily penetrate
through gases at atmospheric pressure over distances of mil-
limeters and excite electrons from core levels of atoms and
molecules in the catalyst. Soft X-ray absorption edges have the
important advantage of being sharp, due to the long lifetime of
the core states. XAS is sensitive to details of the electronic
structure, such as the valence level structure, symmetry, spin
state, etc. The use of soft X-rays for the study in situ of cobalt
catalysts was first reported by Bazin et al.2 In these studies,
Cobalt is also a well-known catalyst for many reactions,
3
including growth of carbon nanotubes, oxidation of carbon
4
5
monoxide, selective reduction of nitrogen oxides and Fischer-
6
,7
Tropsch (FT) synthesis. Its usefulness in FT synthesis is due
to its low water-gas shift activity so that it is able to produce
8
liquid hydrocarbons with high carbon efficiency.
To increase catalytic performance, the catalyst metal disper-
9
sion should be maximized. Typical cobalt catalysts are prepared
from a metallic salt deposited by impregnation over an oxide
support (Al
2
O
3
or SiO
2
), followed by several steps of drying
), which
and calcination. The precursor is a cobalt spinel (Co
3 4
O
is inactive for Fischer-Tropsch synthesis. This oxide is reduced
to the active metallic cobalt phase prior to the reaction.10 The
reducibility of cobalt is a key factor in its catalytic performance.
The Co
3 4
O particles are easy to reduce completely, but in
0,21
supported catalysts, reduction of the cobalt is difficult when the
1
1
spectra of a powder Co/SiO
2
sample were obtained in the
diameter is below 6 nm. If the reduction step is not complete,
-
4
1
2
presence of hydrogen. However, the partial pressure was 10
the catalytic performance decreases. The use of NPs instead
of precursor salts results in improved catalytic activity due to
the lower interaction of the active phase with the support, as
-
6
Pa (<10 Torr), far from the conditions used in industrial
systems, and the catalysts could not be reduced, even after
heating at 650 °C for 30 min. In the present work, we used a
specially designed gas flow reaction cell that made it possible
to acquire XAS spectra at atmospheric pressure and at temper-
atures as high as 360 °C, which is similar to the temperature
used in real industrial conditions. However, the pressure is still
lower than the 20-30 bar used in industrial reactors.
1
3-16
described previously.
Several papers have reported that the turnover frequency or
activity per surface site decreases dramatically when the catalyst
1
3,17,18
particle size reaches a critical point,
a behavior that is not
well understood. Recently, synthesis of Co nanoparticles with
diameters between 3 and 12 nm and a narrow size distribution
was successfully accomplished using colloidal methods.19 The
availability of NPs with such narrow size distributions makes
it possible to explore size and shape effects that might be
important in affecting reactivity and selectivity. These effects
2
. Experimental Section
The synthesis of the cobalt nanoparticles is described else-
where.22 Briefly, in an inert atmosphere, a carbonyl precursor,
Co
perature near its boiling point (∼182 °C).
of organometallic reagents in the hot solvent produces discrete
and homogeneous nucleation sites. Oleic acid, CH (CH CHd
CH(CH CO H, and trioctylphosphine oxide, (CH (CH2) Pd
2 8
(CO) , was decomposed in 1,2-dichlorobenzene at a tem-
23,24
*
To whom correspondence should be addressed. E-mail: MBSalmeron@
Rapid injection
lbl.gov.
†
Materials Sciences Division, Lawrence Berkeley National Laboratory.
Advance Light Source, Lawrence Berkeley National Laboratory.
University of California, Berkeley.
‡
§
2 7
)
3
)
2 7
2
3
7 3
)
1
0.1021/jp901602s CCC: $40.75  2009 American Chemical Society
Published on Web 07/14/2009

1
0722 J. Phys. Chem. B, Vol. 113, No. 31, 2009
Herranz et al.
2
00 mg. The conditions were chosen to maximize the methane
yield in the CO + 3H
2
f CH
4
2
+ H O reaction.
To avoid diffusion limitations, the powder samples were
pelletized, crushed, and sieved to obtain a pellet size of
7
0
.25-0.30 mm. The reactor was placed in an oven connected
to a PID temperature controller (Watlow, St. Louis, MO). The
gas flow was regulated by mass flow controllers (Porter
Instruments, Hatfield, PA). To remove the surfactant and reduce
the cobalt oxide to metal, the samples were heated from room
temperature to 650 °C at a heating rate of 5 °C/min in a 10%
Figure 1. Left: transmission electron micrograph (top) of 5.3 nm cobalt
particles. Right: plot of the particle size distribution.
2
H /Ar atmosphere (total flow rate 50 mL/min). The CO
hydrogenation reaction was subsequently measured at increasing
temperatures from 200 to 270 °C at 0.1 MPa total pressure in
2
a 3 CO/10 H /20 Ar gas flow of 33 mL/min. The concentration
TABLE 1: Catalysts Used in This Studya
particle
mass of catalyst
used (mg)
of the outlet gases was monitored using selected m/z fragment
peaks measured with a quadrupole mass spectrometer connected
to the reactor. The reactor used for measuring the catalytic
activity is equipped with a bypass valve that allows us to first
calculate the CO conversion by looking at the decrease in mass
28 (corresponding to CO). The CO conversion was also
calculated using methane production (mass 15 fragment,
representing 85% of the maximum signal of methane in the
spectrometer) and assuming a 100% selectivity. Because we
use Ar as internal standard, the calculation of the CO conversion
is not affected by the differences in gas flow or by changes in
the pressure of the chamber connected to the mass spectrometer.
In all the catalysts tested, the conversion values calculated using
the decrease in mass 28 and mass 15 were the same within 5%.
name
diameter (nm) loading (% wt)
CoSiO
2
2
2
2
2
2
2
-3
3.0
4.2
4.6
5.3
10.0
8
7.0
3.3
2.5
2.9
1.3
70
74
180
70
230
100
200
CoSiO
CoSiO
CoSiO
CoSiO
CoSiO
CoSiO
-4.2
-4.6
-5.3
-10
-St
-PM
12.0
3.7
25
a
They include Co nanoparticles on SiO
2
powder, as a function of
diameter (in nm), a standard catalyst prepared by impregnation (St),
and a physical mixture (PM).
For comparison, the catalytic activity of two reference
samples, a standard one and a physical mixture, was also
measured. The standard sample (labeled Co/SiO
prepared by incipient wetness impregnation25 using Co(NO
98+%, Sigma-Aldrich, St. Louis, MO) as precursor. After
2
-St) was
3 2
)
(
impregnation, the sample was dried for 16 h at 110 °C and
calcined in air for 2 h at 300 °C.
Another type of catalyst, in the form of a physical mixture
(
labeled PM) was prepared by mixing cobalt spinel (Co
obtained from Co(NO after a thermal treatment in air (2 h at
00 °C), and silicon oxide.
In situ XAS experiments were performed in the total electron
3 4
O ),
Figure 2. Transmission electron micrograph of the 5.3 nm Co particles
supported on SiO powder.
2
3 2
)
2
O, were used to control the growth and to provide a protective
surfactant capping layer. Monodispersed Co NPs with diameters
between 3 and 10 nm ((0.5 nm) were prepared by adjusting
the amount of oleic acid used in the synthesis. Following
synthesis, the samples were cleaned several times by precipita-
tion with alcohols and redispersion in chloroform. The size of
the NP was determined by transmission electron microscopy
yield mode using a specially designed gas flow cell at the
undulator beamline 7.0.1 of the Advanced Light Source
(
Berkeley, CA).26 Details of the design, shown schematically
22
in Figure 3, are described elsewhere. The NPs were deposited
on a gold foil using the Langmuir-Blodgett (LB) technique.
Before the spectroscopy measurements, the samples were
characterized by scanning electron microscopy (Zeiss Gemini
Ultra-55) and X-ray photoelectron spectroscopy (PHI 5400),
both instruments located in the Molecular Foundry (Berkeley,
CA).
27,28
(TEM) using a JEOL 2100-F 200 kV instrument. Figure 1 shows
a typical TEM micrograph of 5.3 nm NPs along with the
measured size distribution.
The NPs dispersed in chloroform were supported over a high
The gold support was placed on a stainless steel holder
attached to a button heater. A k-type thermocouple located under
the sample was used to measure the temperature, and a shielded
cable spot-welded to the sample holder was used for electron
yield measurements. The sample and reaction chamber are
located inside a UHV chamber connected to the synchrotron
2
area commercial SiO (Cab-O-Sil H-5 Cabot Corporation, 300
2
m /g) by wet impregnation. We chose silica because its
7
interaction with cobalt is weak. After impregnation, the samples
were dried at 110 °C for 16 h. The loading was measured by
inductively coupled plasma atomic emission spectroscopy
(
1
Galbraith laboratories) and varied from 1.5 to 7 wt %. Table
lists the names, loading, and mass of all the catalysts used in
3 4
ring and separated from it by a Si N membrane window 0.5
2
this study. Figure 2 shows a TEM micrograph of 5.3 nm
supported Co nanoparticles. As can be seen, the nanoparticles
have roughly spherical shapes and are well-dispersed over the
silica support.
Catalytic hydrogenation of CO was performed in a U-shaped
quartz flow reactor using an amount of sample between 70 and
mm in area and ∼100 nm thick. XAS at the Co-L and O-K
edges were recorded with an energy resolution of 0.3 eV. A
cobalt foil was used for energy calibration. The use of total
electron yield to record XAS makes the method surface-sensitive
(to a depth in the nanometer range), as compared with the bulk
sensitive fluorescence yield detection mode.

Co Particle Size and the CO Hydrogenation Reaction
J. Phys. Chem. B, Vol. 113, No. 31, 2009 10723
Figure 5. Catalytic activity for CO hydrogenation in a 3 CO/10 H
2
/
20 Ar gas flow mixture at a pressure of 0.1 MPa. (a) Turnover frequency
at 240 °C (the line is a visual guide only) and (b) activation energy as
a function of particle diameter.
display more surface area. This point was also confirmed by
thermogravimetric analysis.
Figure 3. Drawing and schematic diagram of the gas-flow cell used
After reduction, the catalytic activity for CO hydrogenation
was measured at ambient pressure and different temperatures
in the 200-270 °C range. In addition to methane, mass
fragments corresponding to products such as ethylene, ethane,
propylene, and propane were also monitored. Specifically, the
fragments corresponding to m/z ) 27, 29, and 41 were
monitored using QMS during all reaction experiments. The peak
at m/z ) 27 corresponds to fragmentation of ethylene and ethene
for in situ X-ray absorption measurements.
(
60 and 70% of the maximum signal in mass spectrometry),
and the fragment at m/z ) 41 corresponds to the maximun signal
in the fragmentation spectrum of propene. Because the products
of the Fischer-Tropsch synthesis in the absence of promoters
follow the Anderson-Schulz-Flory distribution, the main
products (by weight) will correspond to the C1-C3 fraction in
the reaction conditions studied (low alpha, relatively high
temperature, and high hydrogen-to-carbon monoxide ratio). We
did not observe any increment of the cited fragments during
the reaction, implying that hydrocarbons above C1 were
produced in concentrations below the detection limit of our
QMS. In the samples prepared using NPs, the production of
methane was stable at every temperature studied, with no
decrease in activity over time. However, the sample prepared
Figure 4. Products evolved from the 10 nm Co/SiO
2
samples during
temperature-programmed treatment in a H /Ar stream.
2
2
by impregnation (Co/SiO -St) showed a strong deactivation at
higher temperatures (over 250 °C), a behavior attributed to
3
. Results and Discussion
.1. Catalytic Activity of SiO
ment of the samples in 10% H
12
sintering of the particles or to reactions with the support that
3
2
Supported Catalysts. Treat-
in He prior to the catalytic
produce cobalt silicates.30
2
Figure 5a shows the turnover frequency (number of CO
molecules converted per exposed Co atom per second) measured
at 240 °C as a function of the NP diameter. The turnover
frequency increased with diameter, in agreement with previous
reaction yielded various products as the temperature increased,
as shown in Figure 4. First, between 100 and 200 °C, carbon
dioxide, water, and methane evolved due to decomposition of
the surfactant. Between 300 and 450 °C, a broad water peak
appears that corresponds to the reduction of CoO to metallic
cobalt.29 At 550-650 °C, methane evolution was observed,
which we assign to the hydrogenation of carbonaceous species
on the surface of the catalyst. The area of the peaks, normalized
by the total amount of cobalt in the samples, indicated that the
amount of methane and carbon dioxide was larger for the smaller
particles, as expected from the fact that the smaller particles
17
reports, reaching a stable value for diameters larger than 10 nm.
The value of 25 nm in the graph corresponds to samples
prepared by physical mixture, which showed the highest
turnover, although the activity per gram of catalyst was very
low. When the activity is expressed as the number of CO
molecules converted per gram of cobalt, it is larger for the
smaller sizes because of the higher fraction of exposed cobalt
atoms. The activation energy (kcal/mol) calculated from Ar-

1
0724 J. Phys. Chem. B, Vol. 113, No. 31, 2009
Herranz et al.
Figure 7. Co L-edge spectra of cobalt nanoparticles deposited over
Au foils acquired in situ during reduction in pure hydrogen. Particle
diameters are (a) 3 and (b) 10 nm.
cell. In Figure 7, we show the evolution of 2p3/2 and 2p1/2 core
levels of selected Co NPs during reduction of the samples at
increasing temperatures in a 40 mL/min stream of pure H
2
at
0
.1 MPa. For all particle sizes, the spectra at room temperature
20
2+
correspond to CoO, which contains Co ions octahedrically
coordinated with oxygen anions. The presence of this cobalt
oxide is evidenced by the extra features at 776.5 and 780 eV.
During the heating treatment in hydrogen from 250 to 330 °C,
the features in the X-ray absorption spectra changed to those
of the metallic state characterized by two asymmetric absorption
white lines at 778 and 794 eV, as it was described by Morales
et al.34 In this earlier work, the influence of promoters and the
support on the Co reducibility was studied using in situ XAS
with a hydrogen pressure of 2 mbar. However, the influence of
the particle size was not analyzed because the samples were
prepared by homogeneous deposition precipitation from a cobalt
nitrate precursor solution. Unlike the case of cobalt supported
Figure 6. Scanning electron micrographs of cobalt nanoparticles (NPs)
transferred onto Au foils using the Langmuir-Blodgett techniques: (a)
3
nm NPs and (b) 10 nm NPs.
rhenius plots of activity versus 1/T for each particle size is
displayed in Figure 5b. The values are very similar for all
diameters, indicating that the rate-determining step is the same
for all the particle sizes.
7
on oxide powders, our experiments revealed that the smaller
nanoparticles were easier to reduce than the larger ones. The 3
nm sample was completely reduced after a flash heating at 330
°C. When the size increased to 4.2 nm, it was necessary to heat
for 2 min at 330 °C to reach complete reduction. For the 10
nm nanoparticles, the reduction was not completed until after
8 min.
3
.2. Preparation of 2-D Model Catalysts. A single layer
of nanoparticles was deposited over a conductive flat gold
substrate using the LB technique. The particles were first washed
to remove the excess surfactant. Because of its low vapor
pressure, the original dichlorobenzene solvent was substituted
by chloroform for LB preparation. A few droplets of the NP
dispersed in chloroform were spread over a trough filled with
distilled water at room temperature. The surface pressure of the
film was controlled with two Teflon barriers that compress it.
Several samples were prepared at increasing surface pressures.
As reported before,31 the nanoparticles are distributed homo-
geneously over the support, even at low pressure, with a loading
that can be increased with surface pressure. Although the as-
2
After reduction, the H atmosphere was switched to pure
carbon monoxide, and the changes in the electronic structure
of the nanoparticles were followed both at room temperature
and after heating. As shown in Figure 8, the Co L-edge spectra
of the cobalt foil and 3 nm NP did not change after flowing
CO at room temperature or heating in CO or H + CO mixtures.
2
Similar results were obtained with other NP sizes. To obtain
spectra of the O K-edge, the CO gas flow was stopped, and the
line was purged with hydrogen to avoid X-ray absorption by
the gas phase. As expected, the reduced samples show no
features in the O K-edge spectra (Figure 9a). After exposure to
the CO gas flow at room temperature, a large absorption peak
at 534.2 eV appeared. This peak was still present after purging
the line with hydrogen, as shown in Figure 9a for the 4.2 nm
NP sample. We ascribe this peak to CO adsorbed on the cobalt
NPs, because in the conditions studied (room temperature and
ambient pressure), carbon monoxide did not react with the cobalt
surface.35 However, when the sample was heated in a flowing
hydrogen atmosphere at 240 °C for 5 min, the area of the peak
decreased substantially due to the reaction with the flowing gas
to form methane. Heating the sample for 10 min removed the
1
9
prepared nanoparticles are in the metallic state and show a
sharp size distribution (Figure 1), the cleaning process and the
transfer to the gold support caused oxidation of the particles to
CoO,3 as confirmed by electron diffraction and XPS. Figure
2,33
6
shows SEM pictures of two different samples prepared in this
way. The SEM pictures were acquired with low electron energy
1 kV) to minimize beam damage (the particles are still covered
by surfactant). To reduce contamination, after preparation, the
samples were kept in a drybox with N flowing.
.3. In Situ XAS Experiments. Model catalysts consisting
(
2
3
of Co nanoparticles supported on a Au foil were prepared using
the Langmuir-Blodgett technique and studied in situ (i.e., in
reaction conditions of gas pressure and temperature) in the XAS

Co Particle Size and the CO Hydrogenation Reaction
J. Phys. Chem. B, Vol. 113, No. 31, 2009 10725
Figure 10. Products evolved during temperature-programmed treatment
in a H /Ar stream of 10 nm Co/SiO sample. The samples had been
2 2
used in the catalytic CO hydrogenatation reaction.
methanation reaction, with one molecule formed for every
methane molecule. In all cases, irrespective of the particle
diameter, the electronic structure of the cobalt L-edge did not
change during reaction.
Figure 8. XAS of the Co L-edge for a cobalt foil (a) and 3 nm Co
NPs (b) during treatment with carbon monoxide (CO) and syngas (H
CO) at different temperatures.
2
+
These results show that the active phase of Co catalysts in
Fischer-Tropsch synthesis is the metallic state and that it
remains unchanged during the reaction. Some authors have
reported oxidation of the cobalt catalysts due to the water vapor.
There are, indeed, predictions based on theoretical calculations
that support these models.37 However, in Co/SiO
NP catalysts
of 4-5 nm, oxidation was not observed until very high
temperatures (more than 400 °C), even with a H /H O ratio of
This is in agreement with our XAS results, which did not
2
2
2
38
1
.
show any peak assigned to cobalt oxide in the O K-edge region,
and where the Co L-edge peak structure was always that of Co
in the metallic state. To further confirm this result, we performed
another experiment treating the reduced cobalt samples with
2 2 2
H /H O (1:1) and H O/He (PH O ) 23 Torr) in a temperature
2
range of 25-250 °C. After exposing the reduced cobalt
nanoparticles to water vapor, the Co L-edge showed none of
the features characteristic of the oxide (data not shown).
3
.4. Characterization of Samples after Reaction. After
testing the catalytic activity of the Co/SiO powder samples,
the temperature was lowered to 25 °C, and the gas flow was
switched to H /Ar to perform a thermal treatment (TPD-H ) to
2
2
2
characterize the possible carbonaceous species adsorbed on the
surface of the catalysts. Figure 10 shows the evolution of the
H
2
O, CO, CO
nm Co/SiO NP catalyst. As can be seen in the graph, there is
no trace of desorbed products in the temperature range studied
2 4
, and CH signals with temperature for the 10
2
(
50-650 °C), indicating that no detectable amount of carbon-
aceous species was deposited on the catalyst during the CO
hydrogenation.
Figure 9. O K-edge of 4.2 nm cobalt NP after exposure to CO and
after heating in H and He. First, CO was flown over the sample, and
then the sample was heated at 240 °C either in H or He in flow mode
a) or in H in batch mode (b).
2
Figure 11 shows selected TEM pictures obtained from used
powder samples of 3 nm (top), physical mixture (middle), and
the standard sample (bottom). The physical mixture displays
large cobalt particles of irregular shape whose electron diffrac-
tion patterns corresponds to metallic cobalt and CoO. The
standard sample shows a bimodal distribution of particle sizes,
most of the cobalt being present as nanoparticles of 5-6 nm,
but large clusters are also present. In contrast, on the sample
prepared using NPs of 3 nm diameter, the cobalt remains spread
over the support with no signs of agglomeration.
2
(
2
peak completely. If the same heating treatment was performed
in a flowing He atmosphere, the CO-related peak in the O
K-edge at 534.2 eV decreased due to partial desorption but was
still present. It was necessary to pump the line to completely
remove it.
The experiment was also performed in batch mode (Figure
9
b). After flowing pure CO, the cell was filled with hydrogen,
Figure 12 shows a high-resolution SEM micrograph of the
and then the inlet and outlet valves were closed. In this case,
after heating, the 534.2 eV peak decreased, although much more
slowly than under flowing conditions. At the same time, new
features appeared at 536 and 537.2 eV, which we assign to water
10 nm 2D model catalysts after reduction in H
2
and treatment
with H and CO at high temperature. The NPs retained their
2
morphology and size even after the high-temperature treatment.
3.5. H-D Exchange Experiments. The differences in cata-
lytic activity as a function of particle diameter cannot be
3
6
adsorbed on the Co surface. Water is a byproduct of the

1
0726 J. Phys. Chem. B, Vol. 113, No. 31, 2009
Herranz et al.
Figure 13. HD production per exposed cobalt atom as a function of
temperature for various particle diameters.
discarded, as shown by the SEM and TEM micrographs of the
used samples that show no agglomeration of the NP.
It is clear that in any case, the dissociative adsorption of
hydrogen molecules plays a major role in the reaction mecha-
nism. Although it is believed that at saturation, hydrogen
chemisorption involves one hydrogen atom per cobalt surface
site,39 the dissociative step might require a particular geometry
of contiguous empty Co sites with a minimum of two or more
sites. Indeed, it has been shown that on Pd, Pt, and Ru, hydrogen
dissociation requires at least three adjacent sites4
0-42
so that the
dissociative probability, p, will not follow a simple Langmuir
n
law p ∝ (1 - x) function, with n > 2, or even a more
complicated dependence depending on site geometry. In this
model, hydrogenation of CO would require three or more
adjacent sites to dissociate hydrogen and at least one more for
CO adsorption, as in the H-assisted CO dissociation mechanism
43
suggested by King et al. The probability of finding these larger
ensemble sites should decrease rapidly for the smaller particles
and explain at least qualitatively the drop in the turnover
frequency with lower particle diameters.
2
Figure 11. Transmission electron micrographs of Co/SiO catalysts
after reaction: (top) 3 nm NPs, (middle) standard sample, and (bottom)
physical mixture sample.
To test this model, we performed H-D exchange experiments
on selected samples of reduced cobalt NP supported on SiO
2
powder. After reducing the samples with hydrogen at 500 °C,
they were cooled down to room temperature and then exposed
2 2
to a flowing mixture of H /D /Ar (20:20:60, 50 mL/min). The
temperature was then increased slowly (1 °C/min), and the
concentration of HD was measured at 10 °C intervals in the
4
0-90 °C range. The signals of hydrogen (m/z ) 2), deuterium
(m/z ) 4), and HD (m/z ) 3) were monitored online using mass
spectrometry. Figure 13 shows the turnover frequency for HD
production as a function of temperature and particle diameter.
The specific activity for HD production increases with temper-
ature as expected, but more significant, the HD turnover
frequency decreases with particle size. The behavior with
temperature is also different in the cobalt nanoparticles with
diameters below 10 nm. HD production in those samples reaches
a steady state at temperatures below 100 °C. These results
demonstrate that the hydrogen dissociation is structure-sensitive
for particles with size below 10 nm. Although at present we do
not have a detailed model to explain the evolution of the H-D
exchange TOF with size and temperature and that more
experiments are needed for a full understanding, we believe the
observations point strongly to the adsorption and dissociation
Figure 12. Scanning electron micrograph of 10 nm cobalt NPs
deposited on a Au foil after reduction in hydrogen (330-350 °C) and
2
treatment in CO and CO + H mixtures.
attributed to oxidation of the catalysts during reaction, as shown
by the in situ XAS results. The lack of deactivation with time
of the NP catalysts is also consistent with these observations.
Deactivation by carbon deposition can also be ruled out, as
demonstrated by the fact that temperature-programmed treatment
in hydrogen of all the samples after reaction did not produce
carbon-containing products. Finally, sintering of the NP can be
of H
dependence of the methanation reaction. We further propose
that the limiting factor in H adsorption is the availability of
large ensembles of Co atoms on the surface of the nanoparticles.
2
as the mechanism responsible for the particle size
2

Co Particle Size and the CO Hydrogenation Reaction
J. Phys. Chem. B, Vol. 113, No. 31, 2009 10727
4
. Conclusions
(9) Soled, S. L.; Iglesia, E.; Fiato, R. A.; Baumgartner, J. E.; Vroman,
H.; Miseo, S. Top. Catal. 2003, 26, 101.
Monodispersed cobalt nanoparticles prepared from colloidal
(10) Iglesia, E. Appl. Cat. A: Gen. 1997, 161, 59.
(
(
11) Khodakov, A. Y. J. Catal. 1997, 68, 16.
12) Niemel a¨ , M. K.; Backman, L.; Krause, A. O. I.; Vaara, T. Appl.
chemistry methods have been used to study the structure
sensitivity of the CO hydrogenation reaction. Powder samples
supported on silicon oxide and on gold have been prepared using
Co nanoparticles from the same batch. The chemical state of
the nanoparticles and the changes occurring during reaction were
investigated using in situ soft X-ray absorption spectroscopy at
atmospheric pressure and temperature up to 330 °C. We have
shown that the surfactant ligands can be readily removed by
Catal. A: Gen. 1997, 156, 319.
(13) Martinez, A.; Prieto, G. J. Catal. 2007, 245, 470.
(
14) Rioux, R. M.; Song, H.; Hoefelmeyer, J. D.; Yang, P.; Somorjai,
G. A. J. Phys. Chem. B 2005, 109, 2192.
(15) Zheng, N.; Stucky, G. D. J. Am. Chem. Soc. 2006, 128, 14278.
(16) Herranz, T.; Rojas, S.; P e´ rez-Alonso, F. J.; Ojeda, M.; Terreros,
P.; Fierro, J. L. G. Appl. Catal. A: Gen. 2006, 308, 19.
17) Bezemer, G. L.; Bitter, J. H.; Kuipers, H. P. C. E.; Oosterbeek, H.;
(
Holewijn, J. E.; Xu, X.; Kapteijn, F.; van Dillen, A. J.; de Jong, K. P.
J. Am. Chem. Soc. 2006, 128, 3956.
2
heating in H and that the nanoparticles are in the metallic state
and covered with CO during methanation. Our electron micros-
copy studies have shown that the particles do not sinter as a
result of the heating treatments to remove surfactants, or during
the reaction.
We have also shown that the methanation activity decreases
with particle size below 10 nm and that this reduction is not
due to carbon deposition, sintering or oxidation by water vapor.
Notably, we have shown that the activation energy for metha-
nation remains constant with particle size, indicating that the
rate-limiting step is unchanged. Hydrogen-deuterium exchange
experiments show that the specific activity in the HD production
decreases also when the particle size decreases below 10 nm
(18) Yu, Z.; Borg, O.; Chen, D.; Enger, B. C.; Froseth, V.; Rytter, E.;
Wigum, H.; Holmen, A. Catal. Lett. 2006, 109, 43.
(
19) Puntes, V. F.; Krishnan, K. M.; Alivisatos, A. P. Science 2001,
2
91, 2115.
(20) Bazin, D.; Kovacs, I.; Guczi, L.; Parents, P.; Laffon, C.; de Groot,
F.; Ducreaux, O.; Lynch, J. J. Catal. 2000, 189, 456.
(21) Bazin, D.; Guczi, L. Appl. Catal. A: Gen. 2001, 213, 162.
(22) Herranz, T.; Deng, X.; Guo, J.; Cabot, A.; Alivisatio, P.; Salmeron,
M.; not yet published, 2009.
(23) Puntes, V. F.; Gorostiza, P.; Aruguete, D. M.; Bastus, N. G.;
Alivisatos, A. P. Nat. Mater. 2004, 3, 263.
(
24) Puntes, V. F.; Krishnan, K. M.; Alivisatos, A. P. Appl. Phys. Lett.
2
001, 78, 2187.
(25) Ojeda, M.; P e´ rez-Alonso, F. J.; Terreros, P.; Rojas, S.; Herranz,
T.; L o´ pez Granados, M.; Fierro, J. L. G. Langmuir 2006, 22, 3131.
26) Nordgren, J.; Guo, J. H. J. Electron Spectrosc. Relat. Phenom. 2001,
10-111, 1.
27) Heath, J. R.; Knobler, C. M.; Leff, D. V. J. Phys. Chem. B 1997,
101, 189.
(
2
and demonstrates that the dissociative adsorption of H is the
1
important factor responsible for the size effect. We interpret
our results with a model in which large ensembles (more than
four sites) are required for the hydrogen molecule dissociative
adsorption, a key starting step in any model mechanism of CO
methanation.
(
(
28) Rioux, R. M.; Song, H.; Grass, M.; Habas, S.; Niesz, K.;
Hoefelmeyer, J. D.; Yang, P.; Somorjai, G. A. Top. Catal. 2006, 39, 167.
(29) Jacobs, G.; Ji, Y.; Davis, B. H.; Cronauer, D.; Kropf, A. J.; Marshall,
C. L. Appl. Cat. A: Gen. 2007, 333, 177.
30) Puskasa, I.; Fleischa, T. H.; Fulla, P. R.; Kaduka, J. A.; Marshall,
C. L.; Meyers, B. L. Appl. Catal. A: Gen. 2006, 311, 146.
31) Song, H.; Kim, F.; Connor, S.; Somorjai, G. A.; Yang, P. J. Phys.
(
Acknowledgment. We thank Prof. Paul Alivisatos for his
support in the synthesis of nanoparticles and careful reading of
this paper. This work was supported by the Director, Office of
Science, Office of Basic Energy Sciences, Chemical Sciences,
Geosciences, and Biosciences Division, under the Department
of Energy Contract No. DE-AC02-05CH11231. Synthesis and
characterization of the nanoparticles was performed in the
Molecular Foundry and at the Advanced Light Source. T.H.
acknowledges also financial support from the Ramon Areces
Foundation from Spain.
(
Chem. B 2005, 109, 188.
(32) Yang, H. T.; Shen, C. M.; Wang, Y. G.; Su, Y. K.; Yang, T. Z.;
Gao, H. J. Nanotechnology 2004, 15, 70.
(
33) Bao, Y. P.; Beerman, M.; Pakhomov, A. B.; Krishnan, K. M. J.
Phys. Chem. B 2005, 109, 7220.
(34) Morales, F.; de Groot, F. M. F.; Glatzel, P.; Kleimenov, E.; Bluhm,
H.; H a¨ vecker, M.; Knop-Gericke, A.; Weckhuysen, B. M. J. Phys. Chem.
B 2004, 108, 16201.
(
35) Beitel, G. A.; Laskov, A.; Oosterbeek, H.; Kuipers, E. W. J. Phys.
Chem. 1996, 100, 12494.
(36) St o¨ hr, J. NEXAFS Spectroscopy. Springer Series in Surface Sci-
ences, 2nd print; Springer-Verlag: New York, 2003.
(
37) van Steen, E.; Clayes, M.; Dry, M. E.; van der Loosdrecht, J.;
Vilkoen, E. L.; Visagie, J. L. J. Phys. Chem. B 2005, 109, 3575.
38) Saib, A. M.; Borgna, A.; van de Loosdrecht, J.; van Berge, P. J.;
References and Notes
(
(
(
(
(
1) Qiang, Y. J. Phys. ReV. B 2002, 66, 064404.
Niemantsverdriet, J. M. J. Phys. Chem. B 2006, 110, 8657.
(
(
2) Petit, C. J. Phys. Chem. B. 2003, 107, 10333.
39) Reuel, R. C.; Bartholomew, C. H. J. Catal. 1984, 85, 63.
40) Mitsui, T.; Rose, M. K.; Fomin, E.; Ogletree, D. F.; Salmeron, M.
3) Terrado, E. Mater. Sci. Eng., C 2006, 26, 1185.
4) Yang, W. H.; Kim, M. H.; Ham, S. H. Catal. Today 2007, 123,
Nature 2003, 422, 705.
41) Montano, M.; Bratlie, K.; Salmeron, M.; Somorjai, G. A. J. Am.
Chem. Soc. 2006, 128, 13229.
42) Rose, F.; Tartakhanov, M.; Fomin, E.; Salmeron, M. J. Phys. Chem.
C. 2007, 111, 19052.
43) Inderwildi, O. R.; Jenking, S. J.; King, D. A. J. Phys. Chem. C,
008, 112, 1305.
9
4.
(
(
(
5) Armor, J. N. Catal. Today 1995, 26, 147.
6) Dry, M. E., Catalysis: Science and Technologies; Anderson, J. R.,
(
Boudart, M., Eds.; Springer-Verlag: Berlin, 1981; Vol. 1 p 159.
7) Khodakov, A. Y.; Chu, W.; Fongarland, P. Chem. ReV. 2007, 107,
692.
8) Jacobs, G.; Das, T.; Zhang, Y.; Li, J.; Racoillet, G.; Davis, B. H.
Appl. Catal. A: Gen. 2002, 233, 263.
(
(
1
2
(
JP901602S