Atomic-scale study of in situ metal nanoparticle synthesis in a Ni/TiO 2 system

Article abstract of DOI:10.1021/jp044223d

The nucleation and evolution of Ni nanoparticles during reduction of a Ni(NO₃)₂?·6H₂O precursor supported on a commercial titania substrate have been studied in situ at atomic resolution using environmental transmission electron microscopy. An incipient wetness technique was used to prepare the starting unreduced material (10 wt % Ni precursor on titania). The Ni precursor, before reduction, shows a nonuniform distribution over the titania support. It is observed that upon reduction, the initial Ni seed crystal nucleates within the precursor patch. The distribution and size of the Ni nanoparticles thus generated are influenced by the distribution and size of the precursor patches. In this system, we see no evidence of preferential nucleation of Ni particles on anatase or rutile. At 350 ?°C with CO as the reducing agent, the {111} surface facets of the Ni nanoparticles are predominant during the initial stage of nucleation and growth. However, the {111} facets are partially consumed with time, indicating that they are not thermodynamically favored in the CO atmosphere. In CO and H ₂ atmospheres, Ni particles show a nonwetting morphology on titania, while in a mild oxidizing environment, a thin layer of NiO_x is formed, thus giving rise to a morphology that is indicative of wetting of the support. This work provides fundamental information on understanding and controlling the important parameters involved in the preparation of a well-designed supported Ni catalyst using the incipient wetness technique. ? 2005 American Chemical Society.

Full text of DOI:10.1021/jp044223d

J. Phys. Chem. B 2005, 109, 13883-13890
13883
Atomic-Scale Study of in Situ Metal Nanoparticle Synthesis in a Ni/TiO2 System
,
†
‡
§
†
Peng Li,* Jingyue Liu, Nabin Nag, and Peter A. Crozier
Center for Solid State Science, Arizona State UniVersity, Tempe, Arizona 85287-1704, Monsanto Company,
8
00 North Lindbergh BouleVard, St. Louis, Missouri 63167, Process Technology Group, Engelhard
Corporation, 23800 Mercantile Road, Beachwood, Ohio 44122
ReceiVed: December 20, 2004; In Final Form: May 3, 2005
The nucleation and evolution of Ni nanoparticles during reduction of a Ni(NO
3
)
2
2
‚6H O precursor supported
on a commercial titania substrate have been studied in situ at atomic resolution using environmental transmission
electron microscopy. An incipient wetness technique was used to prepare the starting unreduced material (10
wt % Ni precursor on titania). The Ni precursor, before reduction, shows a nonuniform distribution over the
titania support. It is observed that upon reduction, the initial Ni “seed” crystal nucleates within the precursor
patch. The distribution and size of the Ni nanoparticles thus generated are influenced by the distribution and
size of the precursor patches. In this system, we see no evidence of preferential nucleation of Ni particles on
anatase or rutile. At 350 °C with CO as the reducing agent, the {111} surface facets of the Ni nanoparticles
are predominant during the initial stage of nucleation and growth. However, the {111} facets are partially
consumed with time, indicating that they are not thermodynamically favored in the CO atmosphere. In CO
and H
2
atmospheres, Ni particles show a nonwetting morphology on titania, while in a mild oxidizing
is formed, thus giving rise to a morphology that is indicative of wetting of
environment, a thin layer of NiO
x
the support. This work provides fundamental information on understanding and controlling the important
parameters involved in the preparation of a well-designed supported Ni catalyst using the incipient wetness
technique.
I. Introduction
We have employed this technique to study metal nanoparticle
nucleation and evolution during the reduction of a metal
precursor on a high surface-area oxide support. With such an
approach, we were able to examine some of the fundamental
synthesis processes.
The size, distribution, shape, and structure of supported metal
nanoparticles may strongly influence the catalytic properties in
1
ways that are difficult to understand. For example, Haruta et
2
al. found that the finely dispersed gold particles on a select
We studied the nucleation of Ni nanoparticles on a high
surface-area TiO2 support. The Ni/TiO2 system is suitable for
such a study because it has features similar to many industrial
catalysts. It consists of a catalytically active transition metal
dispersed on a high-surface-area, reducible support. For many
applications, a mixture of anatase and rutile is used because it
gives rise to improved catalytic performance. For example, in
selective oxidation reactions, Ni, supported on TiO2 (i.e.,
normally a mixture of anatase and rutile phases), was found to
have a greater resistance toward carbon contamination and Ni
sintering, as compared with other supports. Unlike other
competitive systems such as Pt or Pd on Al2O3 or SiO2, Ni/
TiO2 catalysts are cheaper and do not suffer as much from
coking and poisoning. Therefore, the latter is an attractive
industrial catalyst system. Being rather simple in composition,
Ni/TiO2 is considered as a good system for studying the genesis
of the Ni nanoparticles in a supported Ni catalyst using
microscopic techniques.
metal oxide support possess an extremely high catalytic activity,
and the catalyst activity is usually attributed to the gold particles
below 5 nm. In Ni/silica systems, Coulter et al. and Goodman
3
4
5
have shown that the activity for hydrogenolysis of ethane is
higher on Ni {100} surfaces, as compared with other surfaces.
These studies demonstrate the importance of precisely control-
ling the size and morphology of nanoparticles in order to
optimize catalytic properties. Nucleation and growth mecha-
nisms for metallic nanoparticles or films deposited on single-
6
-10
crystal oxide substrates have been studied extensively,
and
1
2
they are relatively well understood. However, the typical
structure of many heterogeneous catalysts consists of nanometer
sized metal particles dispersed over a high-surface-area support.
Moreover, in many practical applications, these materials are
synthesized via incipient wetness techniques.11 Very few
fundamental studies have been undertaken at the atomic level
to understand the underlying physical processes taking place
during such synthesis. This is partly due to the complex nature
of the high surface-area supports, which typically consist of
particles with different sizes, crystal facets, and heterogeneous
composition. However, it is possible to follow the synthesis of
such complex nanoparticle systems using in situ atomic resolu-
tion environmental transmission electron microscope (ETEM).
Advanced characterization techniques, such as scanning
tunneling microscopy (STM), have been used to study dynamic
nucleation and growth of Ni nanoparticles (or islands) on single-
7
crystal rutile {110} under ultrahigh vacuum or low gas
-
7
13
pressures (i.e., about 10 Torr). However, modern ETEM
allows gas-solid reactions to be studied at atomic resolution
*
Corresponding author.
Arizona State University.
Monsanto Company.
14-19
in pressures up to 10 Torr.
TEM techniques are not limited
†
‡
§
to single crystal samples and can be used to study the genesis
of metal nanoparticles on high surface-area oxide supports.
Engelhard Corporation.
1
0.1021/jp044223d CCC: $30.25 © 2005 American Chemical Society
Published on Web 07/06/2005

1
3884 J. Phys. Chem. B, Vol. 109, No. 29, 2005
Li et al.
Therefore, the information gathered from such experiments will
be of great value in designing and manufacturing industrially
important catalysts.
evolution of individual Ni nanoparticles. To explore the influ-
ence of the reducing environment on the synthesis process, the
Ni precursor was reduced under both strong (i.e., 0.2 Torr of
5% CO/95%N2, 0.2 Torr of 5% H2/95% N2) and weak (i.e., the
background vacuum of the microscope with 10 Torr) reducing
environments.
Special experimental precautions were taken to ensure that
there was not significant electron-beam induced nucleation and
growth of the nanoparticles. The electron beam was blanked
during most of the experiments and turned on to record data
only for short periods of time in order to avoid irradiation effects.
Comparison between the irradiated and nonirradiated areas
demonstrated that the low electron irradiation did not cause
observable effects on the in situ reduction process of the
samples. This confirms that the observed changes were caused
by thermal and chemical effects only. During the rapid tem-
perature ramping, it was not possible to observe the samples
with atomic resolution because of the severe thermal drift in
the TEM heating holder. For this reason the first observation at
In the present work we report on the mechanism, at the atomic
scale, of the synthesis of Ni nanoparticles from a Ni nitrate
hydrate precursor supported on titania under several different
reducing environments (i.e., CO, H2 and vacuum). The relation-
ship between the size and distribution of the supported precursor
and those of the Ni nanoparticles generated under different
reducing environments is discussed in detail. The wetting
behavior of the Ni nanoparticles formed under different reducing
environments is also discussed.
-7
II. Experimental Section
A. Precursor Preparation. Titanium dioxide (Degussa P-25,
a mixture of about 75% anatase and 25% rutile) was loaded
with the desired amount of Ni by an incipient wetness technique
as described below. A calculated volume of an aqueous solution
(just enough to fill the pores of the support), containing Ni-
3
50 °C was made about 20 min after the start of the heating.
(NO3)2‚6H2O, was added to a certain amount of the titania
C. Characterization of the Titania Support. Additional
support with constant mixing. After the solution was added,
the wet mass was left on an evaporating dish for 30 min in
open air. The material was then dried in air at 120 °C for 16 h
in a drying oven. This was the starting material that was used
for further investigation. For examination by conventional TEM
specimens were prepared by dispersing the precursor over holey
carbon films supported on Cu grids. Samples for the in situ
synthesis study of Ni nanoparticles were prepared by dispersing
the precursor over 50-mesh Pt grids with 99.997% purity.
B. Synthesis of Ni Nanoparticles. In situ synthesis of the
Ni nanoparticles was performed in a Tecnai F-20 field emission
nanoscale characterization of the morphology and structure of
the titania support was carried out with a JEOL 2010F 200kV
Schottky field-emission scanning transmission electron micro-
22
scope (STEM). HREM and Z-contrast images were taken from
0 different places of the titania support. The Gatan Digital-
2
TM
Micrograph 3.7.1
software was used for acquiring both
HREM and Z-contrast images.
III. Results
A. Characterization of Ni Precursor on Titania Support.
The titania (Degussa P-25) support is a mixture of about 75%
anatase and 25% rutile. It is important to distinguish the anatase
phase from the rutile phase in order to determine whether there
are some special sites on the different phases of the titania grains
upon which Ni particles could nucleate preferentially. Even
though both the anatase and rutile phases have tetragonal crystal
structures, the largest lattice spacing is 0.35 nm for anatase,
corresponding to {101} atomic planes, and 0.32 nm for rutile,
corresponding to {110} atomic planes. Thus, it is possible to
distinguish anatase from rutile by measuring the lattice spacings
from HREM images of the titania support.
20
ETEM operating at 200kV with a point resolution of 0.20 nm.
The gas reaction cell (essentially a flow microreactor) of this
instrument allowed us to perform atomic level observations of
gas-solid reactions at pressures up to 8 Torr. High-resolution
electron micrographs (HREM) were acquired with a Gatan video
camera and stored on digital videotape during the in situ ETEM
experiments. Individual frames were extracted from the video-
tape using the IMovie software. HREM images were also
recorded with a Gatan CCD camera using the DigitalMicrograph
3
.1 software. The ETEM was also equipped with a Gatan
Imaging Filter permitting electron energy-loss spectroscopy
EELS) to be employed to perform high spatial resolution nano-
(
Figure 1 shows a representative HREM and a Z-contrast
image of the titania support. In the figure, thick areas of the
sample appear with dark and bright contrast in the HREM and
21
chemical analysis. Nanoanalysis was performed by placing a
nanometer-sized electron probe on the nanoparticles.
2
2
Ni nanoparticles were synthesized by performing in situ
reduction of the titania-supported dry precursor at 350 °C under
different reducing conditions. The samples were loaded into a
Gatan Inconel heating holder and 50 Torr of the reducing gases
was admitted into the environmental cell for 30 min to purge
the microscope before the samples were inserted into the
microscope. This purging procedure reduces the concentration
of background gases in the reaction cell. (It should be noted
that under typical conditions in our cell, the reactive gas pressure
is 7 orders of magnitude higher than the background gas
pressure). The initial morphology and the nanostructure of the
precursor (that is, the unreduced Ni-phase on titania) and the
titania support were characterized at room temperature prior to
the reduction. Reducing gas was then admitted to the reaction
cell and the sample temperature ramped up to 350 °C over a
period of 5 min. A rapid temperature increase was used to
simulate the condition that is generally used for making metal
particle catalysts. The temperature was kept constant for up to
Z-contrast images, respectively. As shown in Figure 1a, the
titania support is composed of separate anatase and rutile
crystallites. A representative Z-contrast image of the titania
(Figure 1b) shows that agglomerated titania support has a grain
size of 10-50 nm.
The initial morphology, structure, and dispersion of the Ni
precursor (i.e., Ni(NO3)2.6(H2O)) on the titania support were
studied prior to reduction and the results are shown in Figure
2. On samples with low metal loading (not shown here), most
of the precursor is in the form of thin layers covering the titania
surface. In this high metal loading sample, we also observe a
large fraction of the precursor in the form of large patches. Many
precursor patches show atomic fringes (Figures 2a-c) with
lattice spacings of 0.51 and 0.47 nm corresponding to ( 1h 01)
and ( 1h 11) atomic planes of a triclinic Ni precursor phase. Notice
that the patches of the precursor are randomly distributed on
the titania support. In Figures 2a,b one observes anatase (lattice
spacing of 0.35 nm) as the support for the precursor, whereas
in Figure 2d rutile (lattice spacing of 0.32 nm) is the support.
5
h and the data were recorded every 30 min to follow the

Metal Nanoparticle Synthesis in a Ni/TiO2 System
J. Phys. Chem. B, Vol. 109, No. 29, 2005 13885
Figure 1. (a) HREM and (b) Z-contrast images of titania P-25.
Figure 2. HREM images of the initial distribution of the Ni precursor on titania support before reduction.
These results indicate that the precursor initially gets fixed on
both the anatase and the rutile phases of the titania support
without showing any preference.
generated Ni particle size is proportional to the size of the
original precursor patch.
After 20 min of reduction (Figure 3b), the {111} Ni surface
facet is dominant at this early stage of the heating and the length
of other surface facets (indicated by white arrows) are com-
paratively short. These facets are generally the low-energy
B. In Situ Synthesis Under Strong Reducing Conditions.
The HREM images of Figure 3 show the time evolution of the
nucleation and growth of typical Ni particles as a result of the
reduction of the Ni(NO3)2‚6(H2O) precursor in a CO atmo-
sphere. Notice in Figure 3a (i.e., HREM image before reduction)
the presence of one patch of the Ni precursor with 0.51 nm
lattice spacing corresponding to ( 1h 01) atomic planes on the
surface of the titania support (as indicated by a white arrow).
Upon reduction, this patch of the precursor directly transforms
into a Ni metal nanoparticle, as shown in Figures 3b-d. The
2
3
surfaces of the metal particles. After 60 min at 350 °C, the
Ni nanoparticle slightly increases in size but the overall shape
of the particle does not change much, as seen in Figure 3c. At
the final stage of the reduction (i.e., 120 min), other surface
facets start to grow at the expense of the {111} surface facet.
Figures 4a-c show another example of the dynamic transforma-
tion of the surface facets as a function of the time of reduction.

13886 J. Phys. Chem. B, Vol. 109, No. 29, 2005
Li et al.
2
Figure 3. A series of in situ HREM images showing the nucleation and growth of Ni particles on titania support in 10 wt % Ni/TiO catalysts:
2
(a) before reduction, and (b-d) after reduced the precursor for 20, 60, and 120 min under 0.2 Torr of 5% CO/N at 350 °C.
Once again, as time passes, new surface facets, marked by white
arrows in Figures 4b,c, start to form and grow at the expense
of the original {111} surface facets. The presence of free {111}
and {100} crystal facets (free means the facet does not touch
support) in a Ni nanoparticle oriented along a low-index 〈110〉
zone axis is clearly shown in Figure 4d. In this figure, it is
observed that the {100} surface facet is 8.0 nm long and the
for 180 min under 5% H2/N2, are given in Figure 6. Notice that
the dramatic size difference between the two particles (i.e., 14
nm in Figure 6b versus 5 nm in Figure 6d) is correlated with
the difference in the initial size of the precursor patches shown
in Figures 6a,c. Also notice that the Ni particles stay at the same
location as the Ni precursors before the reduction. These results
indicate that in this high metal loading case, the size and
distribution of the Ni nanoparticles formed after reduction
correlate with the initial size and distribution of the precursor
patches.
{111} surface facet is 6.7 nm after 300 min heating. The length
of the surface facets from a nanoparticle perfectly oriented along
low index zone axis is directly related to the surface free energy
of these surface facets,24 and this issue will be discussed further
in section IV.
C. In Situ Synthesis Under Weak Reducing Conditions.
HREM images taken during in situ synthesis of Ni nanoparticles
performed under weak reducing conditions (i.e., ambient vacuum
of microscope) are shown in Figure 7. In Figure 7a, the patch
of material, indicated by an arrow, is the Ni(NO3)2‚6(H2O)
precursor. After 60 min of heating in a vacuum, this patch of
the precursor partially transforms into a Ni particle as evidenced
by the presence of Ni metal {111} fringes (see Figure 7b).
Notice that the remaining precursor-like contrast at the surface
of the Ni particle (indicated by a white arrow in Figure 7b)
disappears after longer heating time (Figures 7c,d). The sizes
of the precursor patch in Figure 7a and of the reduced Ni particle
in Figure 7d are about the same (i.e., 10 nm). Moreover, most
Ni nanoparticles still have a wetting or half-cup morphology
after heating for 300 min, as shown in Figures 7d,e. The Ni
particles are coated by an overlayer (indicated by an arrow in
Figure 7e). To identify this overlayer, EELS spectra were
acquired from the surface of the Ni particles using a 0.5 nm
electron probe and the result is shown in Figure 7f. This
nanospectroscopic result shows two edges corresponding to the
On close inspection it is found in Figures 3d and 4c,d that
most of the Ni nanoparticles have a nonwetting morphology
after 300 min reduction. The Ni nanoparticles are considered
to have a nonwetting morphology when the contact angles (i.e.,
the angles between tangent line of the surface of Ni particle
and the Ni/titania interface) are larger than 90°. Notice that there
is a transformation from wetting to nonwetting morphology with
time in Figures 4a-c. Figure 4d also shows that there is no
atomic plane matching of the Ni nanoparticle and anatase
support across the Ni/TiO2 interface in the figure.
Similar results on the morphology of the Ni nanoparticles
are obtained when the reduction is performed in a hydrogen
atmosphere. Figure 5 shows typical HREM images of the Ni
nanoparticles that are generated after the precursor is reduced
in situ for 180 min. These images confirm that the morphology
of the reduced Ni nanoparticles is nonwetting.
To clarify the relationship between the size and distribution
of the precursor and the Ni nanoparticles, HREM images of
two Ni nanoparticles, after having the precursor reduced in situ

Metal Nanoparticle Synthesis in a Ni/TiO2 System
J. Phys. Chem. B, Vol. 109, No. 29, 2005 13887
2
Figure 4. In situ HREM images showing the shape changing of the Ni particles on titania support in 10 wt % Ni/TiO catalysts as the Ni precursor
was reduced in 0.2 Torr of 5% CO/N
panel d after reduced for 300 min.
2
at 350 °C for (a) 90, (b) 102, and (c) 180 min. Most Ni particles have nonwetting morphology as shown in
the Ni nanoparticles on the support in the Ni/TiO2 catalyst
system.
The initial conversion of the Ni nitrate hydrate patches to
the Ni nanoparticles under CO is relatively fast (i.e. nearly
completed during the first 20 min of reduction). The particle
continues to grow at a slower rate as reduction continues and
after 120 min no further growth is observed indicating that the
precursor is completely transformed to Ni. In addition, the Ni
particles do not move during the heating at 350 °C, indicating
that particle coalescence is not responsible for the growth of
the Ni particles at this temperature. This is consistent with other
Figure 5. In situ HREM images showing the nonwetting morphology
of most Ni particles on titania support as the Ni precursor was reduced
in 0.2 Torr of 5% H /N at 350 °C for 180 min.
2 2
works reporting that significant coalescence of the metallic
nanoparticles does not take place until the reduction temperature
reaches 500 °C.
The nucleation and growth rate of Ni nanoparticles from the
precursor under weaker reducing environments (i.e., vacuum)
is much slower than that under CO or H2. Some portion of the
precursor remains unreduced at the surface of the Ni particle
even after 60 min of reduction. This demonstrates that modifica-
tion of the reducing environments can be used to control the
nucleation and growth rates of the Ni nanoparticles on titania.
The slower synthesis rate associated with the weaker reduction
environment also allows the Ni nucleation process to be recorded
at an earlier stage of the reduction process. The HREM images
of Figure 7 show that the growing Ni particle is surrounded by
a partially reduced shell of the precursor, indicating that the Ni
metal nucleation does not take place at the surface of the
precursor particle. The data suggest that the initial metal
nucleation occurs in the large precursor patch either at the
interface with the titania support and/or in the center of the
precursor patch. This initial nanocrystallite acts as the seed
crystal for subsequent metal particle growth.
oxygen K-edge (at 532 eV) and Ni L2,3-edges (at 855 eV),
indicating that this surface overlayer is NiOx.
IV. Discussion
A. Initial Nucleation and Growth of Ni Nanoparticles. In
this high metal loading samples, the nucleation sites of the Ni
nanoparticles depend primarily on the locations of the precursor
patches before reduction, as presented in section III. Ni
nanoparticles are found to nucleate on both anatase and rutile
supports due to the random distribution of Ni precursor on the
titania support. These results are different from those on Au/
TiO225 and Pt/TiO226 systems, where Au and Pt nanoclusters
nucleate preferentially on the rutile support. This difference may
be caused by differences in the interactions between the
precursors and the two forms of titania. For example, there might
exist a preferential chemical bond between Au(OH)3 (i.e., the
25
precursor used in Au/TiO2 system ) and rutile grains, which
results in the preferential nucleation of Au cluster on the rutile
grains. This indicates that a careful selection of the precursor
can probably lead to finer control over the nucleation sites of
It is interesting that for all reduction environments used in
this study, the size of the final Ni particle and the precursor

13888 J. Phys. Chem. B, Vol. 109, No. 29, 2005
Li et al.
2 2
Figure 6. (a, c) HREM images of the precursor before reduction and (b, d) the Ni particles after 180 min reduction under 5% H /N .
patch agree to within a factor of 2. However, when a Ni(NO3)2‚
H2O patch transforms to face-centered-cubic (fcc) Ni metal,
the final particle size should drop by approximately a factor of
. The volume decrease occurs because of the extensive mass
shown that the activity for hydrogenolysis of ethane on Ni {100}
is higher than that on Ni {111}. This is due to the fact that the
5
6
distance between two hollow sites on Ni {100} is much larger
than the C-C bond (0.15 nm); thus, the dissociation of the
ethane molecule is favored on the Ni {100} face than on the
Ni {111} wherein the distance between two hollow sites is 0.14
3
loss that occurs during the reduction and because of the denser
packing of Ni atoms in the fcc metal particle. Since we usually
observe Ni particles sizes within a factor of 2 of the precursor
patches this shows that many of the Ni atoms in the final
nanoparticles were not present in the original precursor patch.
Our measurements suggest that more than 80% of the Ni atoms
have diffused into the growing metal seed crystallite from
adjacent precursor layers on the surface of the titania. The
correlation between the Ni particle size and the precursor patch
implies that large patches are associated with large local surface
coverages and small patches are associated with small local
surface coverages. This shows that the precursor coating the
titania surface is not uniform and is associated with the presence
of patches.
Our results demonstrate the importance of carefully control-
ling the distribution of precursor on support. For high metal
loadings, metal particle size depends on the location of precursor
patches since these are the nucleation centers in this case.
Important parameters for precursor preparation will include the
way the precursor is loaded and fixed on the support (that is,
dry or wet impregnation; the pH of the salt solution used for
impregnation, aging, drying etc.). By carefully studying and
optimizing these parameters, a uniform size and distribution of
the initial precursor on the support may be achieved, which in
turn may yield the desired size and distribution of metal
nanoparticles.
5
nm. One practical approach for modifying the relative distribu-
tion of surface facets is to change the reducing gases. This may
cause a change in the free energies of different surfaces because
of differential adsorption of gas molecules on the surfaces of
17
the metallic particles. Hansen et al. have shown that the {111}
surface facets of Cu transform to {110} and {100} facets in a
Cu/ZnO model catalyst as the reducing gas was changed from
pure H2 to a mixture of H2 and water.
In situ ETEM allows us to follow the evolution of the surface
1
5
structure of individual nanoparticles. During reduction in a
CO environment, Ni {111} surfaces are formed within the first
hour of the nucleation and growth processes. As time passes,
other low-energy surface facets (i.e., most likely {100} and
{110}) develop at the expense of the {111} facets. The size
and shape of the nanoparticles do not change significantly after
120 min of heating implying the attainment of the equilibrium
of particle shape in 120 min.
In general, the {111} surface of fcc metal nanoparticles has
a lower energy than those of {110} and {100} surfaces under
2
8
ultrahigh vacuum conditions. However, the anisotropy of
surface free energy can be inverted by the adsorption of gases
2
3
on the metal surface. For example, the adsorption energy of
CO, calculated by a tight binding approach, increases in the
order of {111}, {100} and {110} surfaces in Pd with similar
2
9,30
B. Evolution of Ni Nanoparticle Surface Facets Under CO
Environment. The modification of the surface facets to achieve
a certain desired structure of the nanoparticles is very important
in the area of heterogeneous catalysis. For example, it has been
fcc crystal structures.
Thus, it is possible that CO might be
adsorbed more strongly on the Ni {100} and {110} surfaces
compared to the Ni {111} surface, resulting in lower free
energies of Ni {100} and {110} surfaces compared to that of

Metal Nanoparticle Synthesis in a Ni/TiO2 System
J. Phys. Chem. B, Vol. 109, No. 29, 2005 13889
Figure 7. A series of in situ HREM images showing the synthesis of 10 wt % Ni/TiO catalysts at 350 °C: (a) before reduction, (b-d) after
2
reduced the precursor for 60, 120, and 300 min, (e) the representative shape of the particles after reduced for 360 min, and (f) the corresponding
EELS spectrum from the surface of the Ni particle.
Ni {111} in a CO atmosphere. We know from the experiment
carried out in a vacuum that the initial Ni nucleation occurs
inside the nitrate precursor. Consequently the initial surface
faceting will be strongly influenced by the presence of unreduced
precursor on the metal surface. This may explain why we
initially see a strong {111} facet which is gradually consumed
by new lower energy surfaces when exposed directly to the CO
environment.
Comparison of the relative lengths of different facets in the
equilibrium shape of the nanocrystals provides strong experi-
mental evidence for the above hypothesis. In the equilibrium
crystal shown in Figure 4d, the length of the {100} surface facet
is 8.0 nm while that of the {111} facet is only 6.7 nm. This
indicates that the surface free energy of Ni {100} is 16% lower
than that of {111} under CO at 350 °C, since the longer surface
facet has lower surface energy.28 Similar phenomenon due to
the stronger absorption of O2 on the {100} surface than on {111}
surface of Pd nanoparticles has also been observed in a Pd/
MgO model catalyst, where the {100} facets of Pd develop at
the expense of the {111} facets before the formation of PdO
3
1
under O2 environment. The modification of the surface Pd
facets involves self-diffusion of atoms from {111} to other low-
energy surfaces. In our measurements, the other facets grow at
the expense of the {111} facet (see Figures 3 and 4) showing
that, in the Ni system, the surface modification also involves
self-diffusion from the {111} surface.
C. Reduction Environments Versus the Wetting Behavior
of Nanoparticles. The wetting behavior of the metallic nano-
particles on the oxide support is mainly described by the
2
7
following equation:
γo ) γm/o + γ cosθ
m
where θ is the contact angle between the metallic particle and
the oxide support, γ and γm are the surface energies of the
o
oxide and the metal, respectively, and γm/o is the interfacial
energy between the metallic particle and the oxide. From this
equation the contact angle θ is larger than 90° giving the
nonwetting behavior showing that γ must be smaller than γ
o
m/o.

13890 J. Phys. Chem. B, Vol. 109, No. 29, 2005
Li et al.
As in our Ni/TiO2 system, most Ni nanoparticles do not wet
the titania support and have well-faceted shape under the CO
and H2 reduction environments. More importantly, there is no
atomic plane matching of the Ni and titania across the interface
we see no evidence for preferential nucleation of Ni particles
on anatase or rutile. At 350 °C in CO, the {111} surface facets
of the Ni nanoparticles are predominant during initial nucleation
and growth. However, the {111} facet is not thermodynamically
favored in the CO atmosphere and is partially consumed with
time. In CO and H2, metal particles show a nonwetting
morphology on titania, while in a more oxidizing environment,
a thin layer of NiOx forms to give rise to a wetting morphology.
This in situ ETEM work provides fundamental information for
understanding and controlling important parameters for syn-
thesizing metal nanoparticle catalyst systems using the incipient
wetness route.
(
as shown in Figure 4(d) and discussed in section III), indicating
2
8,32
that the Ni/TiO2 interface is most likely to be incoherent.
In general, the interfacial energy of an incoherent interface
2
28
ranges from 0.8 to 2.5 J/m . It is also known that the surface
2
33
energy of titania is about 0.7 J/m under vacuum. We expect
the surface energy of titania to be lower in the reducing gases
2
3
because of adsorption of gas molecules onto the surface.
Consideration of these energies and our observations demon-
strate that the nonwetting behavior of the Ni nanoparticles on
a titania support is thermodynamically favored. It is also well-
known that there is a strong metal support interaction (SMSI)
between group VIII metals (such as Ni) and titania [27],
therefore, SMSI might affect the wetting behavior of the titania
supported Ni particles. However, the SMSI is normally signifi-
cant at temperatures higher than 500 °C [27], and we suggest it
has little effect on the wetting behavior of the Ni nanoparticles
reduced at the temperature of 350 °C.
Acknowledgment. This research was conducted using the
facilities in the Center for Solid State Science at Arizona State
University and was supported by Monsanto Company. The
authors thank Mr. K. Weiss for the maintenance of the Tecnai
F-20 ETEM. The authors also thank Dr. R. Sharma and Dr.
R-J Liu for the valuable discussions regarding the ETEM.
References and Notes
(
1) Haruta, M. Catal. Today 1997, 36, 153.
In contrast to nonwetting behavior in strong reducing condi-
tions, the Ni nanoparticles wet the titania support under the
weaker reducing environment (i.e., vacuum). In this case we
observe a very thin NiOx layer with a thickness of about 0.3-
(2) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M.;
Delmon, B. J. Catal. 1993, 144, 175.
3) Wolf, A.; Sch u¨ th, F. Appl. Catal. A 2002, 226, 1.
(4) Coulter, K.; Xu, X.; Goodman, D. W. J. Phys. Chem. 1994, 98,
245.
5) Goodman, D. W. Surf. Sci. 1982, 123, L679.
6) Zhou, J.; Kang, Y. C.; Chen, D. A. J. Phys. Chem. B 2003, 107,
6664.
(
1
0
.4 nm (i.e., one or two monolayers) coating the surface of Ni
(
(
particles. The origin of the oxygen might come from the
decomposition of the Ni nitrate hydrate precursor (i.e.,
Ni(NO3)2.6(H2O)) and/or the background gases in the vacuum
system. Moreover, oxygen easily bonds to the Ni surface
(7) Tanner, R. E.; Goldfarb, I.; Castell, M. R.; Briggs, G. A. D. Surf.
Sci. 2001, 486, 167.
(
8) Diebold, U.; Pan, J.; Madey, T. E. Surf. Sci. 1995, 331-333, 845.
because of a strong Ni-oxygen bonding energy (i.e., 392 kJ/
(9) Wu, M.; Møller, P. J. Surf. Sci. 1992, 279, 23.
mol).34 The NiOx-coated overlayer generates a NiOx/TiO2
(10) Bourgeois, S.; Jomard, F.; Perdereau, M. Surf. Sci. 1991, 249, 194.
(
11) Jos van Dillen, A.; Ter o¨ rde; R. J. A. M.; Lensveld; D. J., Geus, J.
W.; De Jong, K. P. J. Catal. 2003, 216, 257.
12) Vannice, M. A.; Garten, R. L. J. Catal. 1979, 56, 236.
(13) Zhou, J.; Kang, Y. C.; Chen, D. A. Surf. Sci. 2003, 537, L429.
interface. In general, the oxide-oxide interface has a stronger
bond and thus, a lower interfacial energy than the metal-oxide
interface.2 This thin oxide layer causes the nanoparticles to wet
the titania support. The surface oxidation of Ni is inhibited when
synthesis is conducted under strongly reducing environments
such as H2 and CO.
(
7
(
(
(
14) Boyes, E. D.; Gai, P. L. Ultramicroscopy 1997, 67, 219.
15) Gai, P. L. AdV. Mater. 1998, 10, 1259.
16) Gai, P. L. Top. Catal. 1999, 8, 97.
(17) Hansen, P. L.; Wagner, J. B.; Helveg, S.; Rostrup-Nielsen, J. R.;
Clausen, B. S.; Topsøe, H. Science 2002, 295, 2053.
18) Wagner, J. B.; Hansen, P. L.; Molenbroek, A. M.; Topsøe, H.;
Clausen, B. S.; Helveg, S. J. Phys. Chem. B 2003, 107, 7753.
(19) Sharma, R. Microsc. Microanal. 2001, 7, 494.
20) Sharma, R.; Crozier, P. A.; Marx, R.; Weiss, K. Proc. Microsc.
Microanal. 2003, 9 (Supplement 02), 912.
21) Egerton, R. F. Electron Energy-Loss Spectroscopy in the Electron
Microscope; Plenum Press: New York, 1986; pp 255-62.
22) Kirkland, E. J. AdVanced Computing in Electron Microscopy;
Plenum Press: New York, 1998.
23) Henry, C. R. Surf. Sci. Reports 1998, 31, 231.
(24) Howe, J. M.; Mebed, A. M.; Chatterjee, K.; Li, P.; Murayama,
M.; Johnson, W. C. Acta Mater. 2003, 51, 1359.
25) Akita, T.; Lu, P.; Ichikawa, S.; Tanaka, K.; Haruta, M Surf. Interface
Anal. 2001, 31, 73.
26) Iddir, H.; Disko, M. M.; Browning, N. D.; Ogut, S. Proc. Microsc.
Treatment in O2 is recognized as an important means for
redispersion of metallic nanoparticles and is applied in the
industry to rejuvenate spent metallic catalysts.2 A recent study
on the O2 influence on the growth mode of Cu and Ni islands
on titania demonstrates that the 3-D growth of the Cu and Ni
(
7
(
(
13
islands can transform to 2-D growth. This causes a redispersion
of the Cu and Ni islands on the TiO2 support by O2 treatment.
In our study, we find that in a slightly oxidizing environment,
the thin oxide layer formed on the Ni surface causes the
nanoparticle to adopt a wetting morphology. Even though only
the Ni particle surface is oxidized our results demonstrate the
onset of the complex redispersion phenomenon, which involves
oxidizing of the metal particles and wetting and spreading of
(
(
(
(
Microanal. 2004, 10 (Supplement 02), 448.
2
7
(27) Raupp, G. B.; Stevenson, S. A.; Dumesic, J. A.; Tauster, S. J.;
Baker, R. T. K. In Metal-Support Interactions in Catalysis, Sintering, and
Redispersion; Stevenson S. A., Dumesic, J. A., Baker, R. T. K., Ruckenstein,
E., Eds.; Van Nostrand Reinhold Company Inc.: New York, 1987.
the metal oxides under strong oxidizing environment.
V. Conclusions
(
28) Howe, J. M. Interfaces in Materials: Atomic Structure, Kinetics
and Thermodynamics of Solid: Solid, Solid: Liquid and Solid: Vapor
Interfaces; John Wiley and Sons: New York, 1997.
We have performed in situ atomic-resolution studies on Ni
nanoparticle nucleation and evolution during the reduction of a
metal precursor on a high surface-area titania support. Under
high metal loading conditions, the unreduced precursor, prepared
by an incipient wetness technique, shows a nonuniform distribu-
tion over the titania support. Upon reduction, the initial seed
crystals formed in the precursor patches, and the distribution
and size of the synthesized Ni nanoparticles are influenced by
the distribution and size of the precursor patches. In this system,
(29) Mottet, C.; Tr e´ glia, G.; Legrand, B. Surf. Sci. 1996, 352-354, 675.
(30) Sawaya, S.; Goniakowski, J.; Mottet, C.; Saul, A.; Tr e´ glia, G. Phys.
ReV. B 1997, 56, 12161.
(
31) Giorgio, S.; Henry, C. R.; Chapon, C.; Roucau, C. J. Catal. 1994,
1
48, 534.
(32) Li, P.; Howe, J. M.; Reynolds, W. T., Jr. Acta Mater. 2004, 52,
39.
2
(
(
33) Bates, S. P.; Kresse, G.; Gillan, M. J. Surf. Sci. 1997, 385, 386.
34) Lide, D. R., Ed. In CRC Handbook of Chemistry and Physics; VRC
Press: New York, 1997.