Dynamic nucleation and growth of Ni nanoparticles on high-surface area titania

Article abstract of DOI:10.1016/j.susc.2005.11.023

Nucleation and growth mechanisms of Ni nanoparticles synthesized via an incipient wetness technique on a high-surface area titania support (i.e., a mixture of anatase and rutile) are studied using environmental transmission electron microscope (ETEM). Most Ni nanoparticles are found to nucleate from the Ni precursor coated on the surface of the titania support. Even though both anatase and rutile supports are the nucleation sites for Ni nanoparticles, it was observed that the particles have different morphologies on the supports, i.e., a non-wetting morphology on the anatase support versus a wetting morphology on the rutile {1 0 1}. This is because the interfacial energy of Ni/rutile is lower than that of Ni/anatase. Titania clusters are found to nucleate on the surface of the Ni particles during in situ ETEM reduction, indicating that the presence of partial titania overlayers is directly related to the synthesis of the Ni/TiO₂ catalysts. The growth mode of the Ni nanoparticles on the titania support is three-dimensional, while that of the rutile cluster on the surface of the Ni is two-dimensional layer-by-layer.

Full text of DOI:10.1016/j.susc.2005.11.023

Surface Science 600 (2006) 693–702
www.elsevier.com/locate/susc
Dynamic nucleation and growth of Ni nanoparticles
on high-surface area titania
a,
b
c
d
P. Li *, J. Liu , N. Nag , P.A. Crozier
a
TEM Laboratory, Department of Earth and Planetary Sciences, University of New Mexico, MSCO3-2040,
Northrop Hall, Albuquerque, NM 87131, USA
b
Monsanto Company, 800 N. Lindbergh Boulevard, St. Louis, MO 63167, USA
Process Technology Group, Engelhard Corporation, 23800 Mercantile Road, Beachwood, OH 44122, USA
c
d
Center for Solid State Science, Arizona State University, Tempe, AZ 85287-1704, USA
Received 22 January 2005; accepted for publication 30 November 2005
Available online 27 December 2005
Abstract
Nucleation and growth mechanisms of Ni nanoparticles synthesized via an incipient wetness technique on a high-surface area titania
support (i.e., a mixture of anatase and rutile) are studied using environmental transmission electron microscope (ETEM). Most Ni nano-
particles are found to nucleate from the Ni precursor coated on the surface of the titania support. Even though both anatase and rutile
supports are the nucleation sites for Ni nanoparticles, it was observed that the particles have different morphologies on the supports, i.e.,
a non-wetting morphology on the anatase support versus a wetting morphology on the rutile {101}. This is because the interfacial energy
of Ni/rutile is lower than that of Ni/anatase. Titania clusters are found to nucleate on the surface of the Ni particles during in situ ETEM
reduction, indicating that the presence of partial titania overlayers is directly related to the synthesis of the Ni/TiO2 catalysts. The growth
mode of the Ni nanoparticles on the titania support is three-dimensional, while that of the rutile cluster on the surface of the Ni is two-
dimensional layer-by-layer.
Ó 2005 Elsevier B.V. All rights reserved.
Keywords: Ni; Titanium oxide; Anatase; Rutile; Nucleation; Growth; Morphology; Reduction; Synthesis; Environmental TEM; Nanoparticles; Catalysts
1. Introduction
fundamental physical processes taking place during such
synthesis. This is partly due to the complex nature of the
high-surface area supports, which may consist of particles
with different sizes, crystal facets and structures. The
problem is compounded by the lack of suitable in situ tech-
niques for studying the transformation process under suit-
ably high gas pressures.
Ni is an important transition metal that is used in many
heterogeneous catalysts. In many applications, Ni is loaded
on a reducible oxide support like titania. There are a signif-
icant number of surface science studies conducted under
UHV conditions on the {110} rutile surface [1–6,8,9]. How-
ever, in catalytic applications, the metal is usually loaded
onto a suitable high-surface area titania. One of the more
commonly used titanias is Degussa P25 consisting of 75%
anatase and 25% rutile grains [10]. It was suggested
that Ni nanoparticles grow three-dimensionally on rutile
A considerable body of work [1–6] has been carried out
on the nucleation and growth of metal clusters on single
crystal oxide surfaces due to the extensive applications of
metal/oxide systems to heterogeneous catalysts. However,
in many practical applications, heterogeneous catalysts
are synthesized via an incipient wetness technique on a
high-surface area oxide support [7]. This approach uses a
combination of wet chemistry and reducing atmospheres
to create a dispersion of metal particles and is very different
from the ultra-high vacuum (UHV) evaporation techniques
used in most of the single crystal work. Very few atomic
level studies have been undertaken to understand the
*
Corresponding author. Tel.: +1 505 2777536; fax: +1 505 2778843.
E-mail address: pengli@unm.edu (P. Li).
0039-6028/$ - see front matter Ó 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2005.11.023

694
P. Li et al. / Surface Science 600 (2006) 693–702
supports [2]. However, the growth mechanism of Ni nano-
particles on anatase grains does not appear to have been
studied. There are many questions surrounding the nucle-
ation and growth of Ni nanoparticles on the surface of these
industrially important high-surface area titanias. Does the
precursor uniformly spread over the surface of the support?
Is the precursor mobile when heated in a reducing environ-
ment? Where are the nucleation sites for metal particle
growth? Does the precursor transform directly to metal or
are intermediate compounds first formed? These question
and many others remain about not just the Ni on titania
but also many other catalytic metal/support combinations.
Environmental transmission electron microscopy
(ETEM) is a suitable technique for studying the nucleation
and growth of metal nanoparticles on a high-surface area
support. The technique allows us to study gas–solid reac-
tions at atomic level with pressures up to a few Torr [10–
15] and temperatures up to 800 °C. It is ideal for studying
reactions on high-surface area supports because the evolu-
tion of individual metal nanoparticles on particular grains
of titania can be followed during in situ reduction of the pre-
cursor. We have already demonstrated that ETEM can be
applied to investigate the in situ synthesis of Ni nanoparti-
cles from heavy Ni precursor loadings (i.e., 10 wt.%) sup-
ported on high-surface area titania [10]. In that study, Ni
precursor, loaded by the incipient wetness technique, was
found to distribute non-uniformly on the titania support
and large patches of Ni precursor acted as nucleation cen-
ters for the Ni nanoparticle growth [10]. Most Ni particles
were found to possess a non-wetting morphology on the
titania support [10]. Modification of the particle equilibrium
shape under different gas conditions was also explored [10].
Here we are interested in studying the nucleation and
growth process under lower, more realistic metal loadings
(3 wt.%). In the present study, we report on the mechanism,
at the atomic scale, of the dynamic nucleation and growth
of Ni nanoparticles on both anatase and rutile grains. The
wetting/non-wetting behavior of the Ni nanoparticles on
rutile/anatase supports is discussed in detail. The growth
mode of the Ni nanoparticles on rutile and anatase supports
is discussed in terms of the surface and interface energies.
The growth mode of the Ni nanoparticles on a high-surface
area rutile support is compared with UHV single crystal
studies. We also observe a double nucleation event in which
rutile simultaneously nucleates on the surface of the grow-
ing Ni particle. This observation provides evidence for the
decoration model proposed [16] to explain the strong metal
support interaction (SMSI) associated with metal nanopar-
ticles on reducible oxide supports. The relevance of this dec-
oration model for SMSI phenomena is discussed.
an incipient wetness technique. An aqueous solution con-
taining Ni(NO3)2 Æ 6H2O was added to fill the pores of a
certain amount of titania support during 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. Precur-
sor specimens for ETEM were prepared by placing the
powders on 50-mesh Pt grids with 99.997% purity.
2.2. In situ synthesis of Ni metal nanoparticles
To study the dynamic nucleation and growth mecha-
nism of the Ni nanoparticles, the Ni(NO3)2 Æ 6(H2O) was
reduced in situ using a Tecnai F-20 field emission ETEM
operating at 200 kV. The microscope has a Scherzer resolu-
tion of 0.24 nm with an information limit of 0.12 nm. Pre-
cursor samples were loaded into an Inconel heating holder
and inserted into the built-in environmental cell between
the objective lens pole pieces of the microscope. This
arrangement allows us to study the chemical reactions be-
tween gas molecules and nanoparticles at reasonably high
pressures and reaction temperatures without compromising
the atomic resolution imaging capability of the transmis-
sion electron microscope (TEM).
In situ metal particle synthesis was carried out by reduc-
ing the Ni(NO3)2 Æ 6(H2O) in 0.2 Torr of CO in the ETEM.
The precursor was rapidly heated to 350 °C over a period
of 5 min and kept at this temperature in the CO atmo-
sphere for 5 h. For this precursor, CO is a stronger reduc-
ing agent than H2 and leads to an easier reduction of
precursor at lower pressures. Comparison of experiments
conducted in H2 and CO show no evidence for the precur-
sor reduction being strongly influenced by the formation of
Ni(CO)4. In addition, the boiling point of Ni(CO)4 is 43 °C
and we anticipated that it rapidly decomposes at the reduc-
tion temperature employed here (i.e., 350 °C).
The dynamic reduction process of the precursor (i.e.,
nucleation and growth of Ni nanoparticles) was recorded
using a high-resolution camera. Since the formation of
individual Ni particles was followed at the atomic level to
understand the dynamic reduction process, we were nor-
mally able to analyze about 20–30 nucleation and growth
events by conducting the experiments under the same
reducing condition twice. In order to avoid the strong elec-
tron beam introduced artifacts to our experiments, very
low electron doses were used to observe the precursor
reduction process with some sacrifices of the quality of
the high-resolution electron microscopy (HREM) images.
HREM images from different areas of the sample were ex-
tracted frame by frame from the digital videotape using the
IMovieTM software. In some cases, a four-frame averaging
process was used to reduce noise. HREM images were also
acquired from a CCD camera using DigitalMicrograph
3.1TM software during the ETEM experiments.
2. Experimental procedures
2.1. Preparation of precursor
The shape of the Ni particles can be clearly determined if
the Ni particles locate on the titania surfaces with edge-on
orientation, i.e., the titania surface paralleled to the elec-
Degussa P-25 (titanium dioxide with a mixture of about
75% anatase and 25% rutile) was loaded with 3 wt.% Ni by

P. Li et al. / Surface Science 600 (2006) 693–702
695
tron beam direction or viewing direction. This is also the
orientation for clearly exploring the interface structures be-
tween the Ni particles and the titania support. We were
able to place a 1 nm electron probe on particles or clusters
extended to the vacuum of the microscope (i.e., particles on
the edge of the supports) to perform precise electron energy
loss spectroscopy (EELS) chemical analysis of the particles.
rutile grain as a result of the reduction of the Ni precursor
(i.e., Ni(NO3)2 Æ 6(H2O)). As shown in Fig. 1(a), 30 min
into the carbon monoxide reduction, the Ni nanoparticle
has nucleated on the rutile {101} surface, where the
0.25 nm and 0.20 nm lattice spacing indicated in the figure
corresponds to {101}rutile and {111}Ni planes, respec-
tively. As time passes (Fig. 1(b)), the area of the Ni/rutile
interface increases with concomitant growth of the Ni par-
ticle along all the directions. The Ni particle tends to align
its {111} atomic planes to the {101} atomic planes of the
rutile support, as demonstrated by the parallel {111}Ni and
{101}support across the interface between the Ni nanoparti-
cle and the rutile support in the figure.
Many Ni nanoparticles have also been found to nucle-
ate on the anatase support, as representatively shown
in Fig. 2. The lattice spacing of 0.35 nm indicated in
Fig. 2(a) corresponds to {101}anatase atomic planes. The
titania support (i.e., anatase) is coated by a surface
overlayer indicated by arrows in the figure. Fig. 2(b)
shows that within 20 min of carbon monoxide contact, a
Ni nanoparticle with lattice spacing of 0.20 nm corre-
sponding to {111}Ni has nucleated at the surface of
the titania support from the coated surface overlayer,
shown in Fig. 2(a). Notice that the Ni particle nucleates
on the rough anatase surface (indicated by a black
arrow in Fig. 2(a)) instead of the flat {101}anatase. As time
passes, the surface overlayer disappears in Fig. 2(c).
This strongly suggests that the surface overlayer in
Fig. 2(a) and (b) is the Ni precursor that is steadily
depleted during reduction with concomitant growth
of Ni particles. Also note from Fig. 2(d) that the Ni
particle is well-faceted and low energy surface facets, such
as the {111} surface facet indicated in the figure, are
distinguishable. Two-dimensional layer-by-layer growth
2.3. TEM imaging of the ex situ Ni/TiO2 catalysts
We also performed ex situ experiments in which samples
were reduced at atmospheric pressure in a mixture of 4%
hydrogen and nitrogen at 250 °C, 500 °C and 800 °C.
TEM samples were prepared by placing the catalyst pow-
ders on holey-carbon films supported on Cu grids. Regular
TEM and Z-contrast imaging of the reduced samples were
performed in a JEOL 2010F 200 kV Schottky field emission
TEM operating in the regular TEM mode and scanning
transmission electron microscope (STEM) mode using an
electron probe size of 0.2 nm. Z-contrast images are formed
by collecting the high-angle electron scattering which has a
dependence on the atomic number, Z, of approximately Z1.7
[17], hence, the image contrast is sensitive to the presence of
heavier nanoparticles (i.e., high-Z) on the lighter support.
This technique was demonstrated to be able to observe sin-
gle atoms and small nanoclusters [18] and is used to explore
the distribution of Ni nanoparticles on the titania support.
The Gatan DigitalMicrograph 3.7.1TM software was used
for acquiring the regular TEM and Z-contrast images.
3. Results
The HREM images in Fig. 1 give the time evolution
of the nucleation and growth of a Ni nanoparticle on a
Fig. 1. In situ HREM images showing the nucleation and growth of Ni particles on rutile support after the reduction of the precursor for (a) 30 min and
(b) 60 min. Note the superimposed EELS spectrum in Fig. 1(b) acquired from the rutile cluster contains a Ti peak.

696
P. Li et al. / Surface Science 600 (2006) 693–702
Fig. 2. A series of in situ HREM images showing the nucleation and growth of Ni particles on anatase support: (a) before the reduction, and (b)–(d) after
the reduction of the precursor for 20 min, 52 min, and 96 min.
of the Ni nanoparticles has not been observed through
the whole synthesis process and the Ni particles grow as
three-dimensional nanoislands.
at the original location of the Ni precursor. It is also found
that the Ni nanoparticle does not move around at this
reduction temperature (i.e., 350 °C) and the dynamic parti-
cle sintering or coalescence has not been observed.
The presence of patches of precursor on some areas of
the titania support indicate that the precursor loading
was not perfectly uniform. Even though most Ni nanopar-
ticles (i.e., 22 in 24 observations) nucleate from the thin
layer of Ni precursor that uniformly coats the titania sup-
port, a small number of the Ni nanoparticles (i.e., 2 in 24
observations) transformed directly from the patches of
the precursor. In this case, there is a direct relationship be-
tween the location of the Ni nanoparticle (indicated by a
white arrow in Fig. 3(b)) and the precursor (indicated by
a white arrow in Fig. 3(a)), i.e., the Ni nanoparticle stays
As demonstrated in Fig. 1, an additional cluster some-
times nucleates at the surface of the Ni nanoparticles sup-
ported on both anatase and rutile supports (i.e., in this
case, the fuzzy materials indicated by a white arrow in
Fig. 1(a)). This phenomenon occurred on the surface of
about 20% of the Ni nanoparticles observed (i.e., 5 in 24
observations). The cluster grows with time, as shown in
Fig. 1(b). The lattice fringe from this cluster has exactly
the same spacing as {101} planes of the rutile support
(i.e., 0.25 nm), which is clearly exposed in Fig. 1(b) suggest-

P. Li et al. / Surface Science 600 (2006) 693–702
697
Fig. 3. In situ HREM images showing the direct nucleation of Ni particles from one patch of the precursor on titania support: (a) before the reduction and
(b) after the reduction of precursor for 50 min.
ing that this additional cluster is rutile. Furthermore, EELS
spectrum acquired from this tiny cluster (see Fig. 1(b)) by
precisely placing a 1 nm electron nanoprobe in the center
of it, although noisy, shows a Ti peak confirming the rutile
interpretation. More importantly, this rutile cluster also
tends to align its {101} atomic planes to the {111} atomic
planes of the Ni nanoparticles, giving a parallel orientation
relationship among the rutile support, Ni nanoparticle and
the rutile cluster (i.e., {101}supportk{111}Nik{101}cluster).
The rutile cluster in Fig. 1(a) is already five or six mono-
layers high and occupies the entire Ni{111} surface. In this
case the initial growth of the first one or two monolayers of
the cluster was not recorded due to the thermal drifting of
the specimen holder during the first 20 min of heating. How-
ever, the dynamic growth mode of this rutile cluster on the
surface of the Ni nanoparticle was observed at a later time
and the result is shown in Fig. 4. Notice in Fig. 4(a), an atom-
ic ledge with height about two or three {101}rutile atomic
planes forms at the surface of the titania cluster (indicated
by a white arrow), and as time passes (Fig. 4(b)), the surface
steps on the titania cluster grows to a complete {101}rutile
terrace. This growth process repeats and the titania cluster
continuously grows as seen in Fig. 4(c) and (d).
Another important phenomenon is the different shapes
of the Ni nanoparticles on different types of titania sup-
ports. In general, the morphology of the supported metallic
particles has two types of shape, i.e., non-wetting and
wetting. The morphology of the particle is defined to be
non-wetting when the contact angle between the metallic
particles and support are larger than 90° [16]. As demon-
strated in Fig. 2(d), all Ni particles observed in this study
(i.e., all eight observations of Ni particles locating on the
anatase surfaces with edge-on orientation) possess a non-
wetting morphology on anatase surfaces, while the shape
of the Ni particle was wetting in Fig. 1(b) when supported
on the rutile {101}, which holds for the two observations
of Ni particles locating on rutile {101} with edge-on orien-
tation. Also notice that the {111} Ni atomic planes are not
oriented with the {101} anatase planes in Fig. 2(d), while
there is a one-dimensional matching of the atomic plane
orientation across the Ni/rutile interface in Fig. 1(b). The
shape of the Ni particles is directly related to the Ni/titania
interface structures, which is discussed in the next section.
The results from our ex situ analysis performed on the
hydrogen-reduced samples are in qualitative agreement
with our in situ measurements. Regular TEM and Z-con-
trast imaging of the ex situ samples (in Fig. 5) show average
Ni particle sizes of 2 nm, 59 nm and 81 nm corresponding
to reduction at 250 °C, 500 °C and 800 °C respectively.
The average size on the Ni particles obtained from in situ
reduction in CO at 350 °C is 10 nm. Our in situ measure-
ments show that Ni particles nucleate on both anatase
and rutile supports (shown in Figs. 1 and 2) during reduc-
tion in CO at 350 °C. The ex situ data also shows this
trend. Fig. 5(a) and (b) are two representative TEM images
of Ni/TiO2 system after ex situ reduction at 250 °C. In
these images, the size and distribution of the 1–2 nm Ni
particles are uniform over all titania grains (i.e., 75% ana-
tase grains mixing with 25% rutile grains) demonstrating
that the particles nucleate on both anatase and rutile.
4. Discussions
4.1. Nucleation and growth mechanism of Ni nanoparticles
on titania support
Most Ni nanoparticles nucleate from the Ni precursor
uniformly coating the titania support, even though a small

698
P. Li et al. / Surface Science 600 (2006) 693–702
Fig. 4. In situ HREM images showing growth of a titania crystallite on a Ni surface after the precursor was reduced for (a) 180 min, (b) 200 min,
(c) 210 min, and (d) 220 min.
number of Ni nanoparticles are directly transformed from
the patches of the precursor. This is in contrast to the nucle-
ation process that we observed on samples with higher metal
loading where the precursor had a non-uniform distribution
on the titania support and the large precursor patches were
the nucleation centers for the metal particles [10].
As presented in Section 3, Ni nanoparticles nucleate on
both anatase and rutile support in the CO environment.
Regular TEM and Z-contrast imaging on the ex situ
reduced Ni/TiO2 catalysts under H2/N2 also demonstrates
a uniform distribution of the Ni nanoparticles on the tita-
nia supports. Even though the particle size distributions are
different, the results indicate that Ni particles do not pref-
erentially nucleate on either anatase or rutile under both
CO and H2 environments. This is different from that in
Au/TiO2 catalyst system, where Au was found to preferen-
tially nucleate only on the rutile support [19]. Ni nanopar-
ticles were also found to nucleate on both anatase and
rutile in the high Ni loading sample [10].
Even though Ni particles were occasionally observed to
nucleate on the flat anatase {101} surface [10], most Ni
nanoparticles nucleate on rough anatase surfaces instead
of the low energy flat anatase {101} surface, as shown in
Fig. 2(b). This is probably because the surface defects
(i.e., the ledges and kinks of the surface) can hold the Ni
species for a longer time so that the Ni atoms can meet
and grow to nuclei [2,6]. We also observe two instances
of Ni nucleating on the {101} surface of rutile (i.e., a rel-
atively higher energy surface than {110}rutile). Density
functional calculations suggest that, for the equilibrium
crystal shapes, 56% of the rutile surface should terminate
with the low energy {110} facets and 94% of the anatase
surface should terminate with the low energy {101} facets
[20,21]. The number of observations of Ni nucleation

P. Li et al. / Surface Science 600 (2006) 693–702
699
Fig. 5. Regular TEM and Z-contrast images of the 3 wt.% Ni/TiO2 catalysts ex situ reduced at (a) and (b) 250 °C, (c) 500 °C and (d) 800 °C.
events in which the surface facet can be unambiguously
identified is rather small because of the random orienta-
tions of the high-surface area support. However, even
though the low energy surface of rutile (i.e., {110}) and
anatase (i.e., {101}) should compose the majority of the
crystal facets, these surfaces do not appear to be the pref-
erential nucleation sites for Ni particles.
The different morphologies of the Ni nanoparticles on
anatase and rutile {101} are directly related to the differ-
ences in titania surface energies and the Ni/TiO2 interfacial
energies. Typical surface energies for Ni, anatase and rutile
along with typical interfacial energies are listed in Table 1.
The surface energy of the metal is a function of tempera-
ture [22,23]. The experimental value of the Ni surface
energy at 0 K was extrapolated from that at high tempera-
ture and the surface energy of Ni solid at the melting point
was calculated from that of Ni liquid [22].
The Ni nanoparticles possess a non-wetting or wetting
morphology depending on whether co (i.e., the surface en-
ergy of the titania) is smaller or larger than cm/o (i.e., the
Ni/TiO2 interfacial energy) [16]. As is also well known,
the interfacial energy is closely related to the interface
structure (i.e., atomic matching at the interface) [24], which
is qualitatively studied in this work. For all the Ni nano-
particles observed on the anatase support (as shown in
Fig. 2(d)), there is no one-dimensional matching of the
atomic planes across the Ni/anatase interface. This indi-
cates that the atomic matching may be poor at the interface
suggesting that this interface is likely to be incoherent
[24,25]. In general, the average surface energy of anatase

700
P. Li et al. / Surface Science 600 (2006) 693–702
Table 1
Ni nanoparticles since the Ni particles did not move during
the observation at 350 °C. This indicates that the growth of
the Ni nanoparticles involves the diffusion of the Ni con-
taining species to the surface of the growing metal particles.
The mobile Ni species could be either Ni containing precur-
sor molecules or Ni metal atoms produced as a result of the
reduction of the precursor. The growth process of Ni nano-
particles can be compared to that in the deposition process
of the metallic adatoms on the oxide support [1–4,8,26].
These works show that Ni atoms can diffuse on the titania
support at 350 °C, since it has been found that the Ni is-
lands deposited via evaporation at 127–177 °C on rutile
{110} were substantially bigger than those deposited at
room temperature, indicating the Ni adatoms have suffi-
cient energy to diffuse on the titania support at even lower
temperature ranges of 127–177 °C [3]. The discussions from
the next section suggest that, in our case, a significant frac-
tion of the mobile Ni species is unreduced or partially
reduced precursor.
Surface energies of Ni, rutile and anatase and interfacial energies
Free energy
(J/m2)
Reference and
comments
Rutile (average)
(110)
(100)
(101)
Anatase (average)
(101)
1.1
Calculations using local
density approximation
[20,21]
0.84
1.05
1.36
0.9
0.84
0.96
(100)
Ni (average)
(111)
2.241–2.082
Data at 10 k and
2melting point [22]
Calculation using EAM
potentials [23]
2.04
Incoherent interface
Semicoherent interface
Coherent interface
0.8–2.5
0.2–0.8
0.05–0.2
Generally calculated
values [24]
is about ꢀ0.9 J/m2 [20] and is probably lower than the
incoherent interfacial energy (generally ranging from 0.8
to 2.5 J/m2 [24]) thus generating the non-wetting morphol-
ogy of the Ni nanoparticles on the anatase support. The
non-wetting morphology of Ni particles on flat anatase sur-
face was also demonstrated to be energetically favored by
Li et al. [10], where the shape of the Ni particle transforms
from wetting to non-wetting on a flat anatase {101} as the
reducing time. For the Ni nanoparticles supported on the
{101}rutile, the surface energy is much larger (ꢀ1.36 J/m2
[20]) compared to the corresponding surfaces for anatase.
Moreover, the alignment of the {111}Ni and {101}rutile
planes make it more likely for atoms inside both planes
to match at the interface thus lowering the interfacial en-
ergy. The comparatively high {101}rutile surface energy
and low interfacial energy makes it thermodynamically
favorable for the Ni nanoparticles to possess a wetting
morphology on the {101}rutile. It has been found that Cu
nanoparticles possess a similar wetting morphology on
the ZnO support under a gas mixture of 95% H2 and 5%
CO, where the parallel atomic planes across the Cu/ZnO
interface has been observed [12].
The three-dimensional island growth is favored when
co < cm + cm/o, where co and cm are the surface energies
of titania and Ni nanoparticles, respectively, and cm/o is
again the interfacial energy between the Ni nanoparticle
and titania [2,3]. Even the lowest surface energy of Ni
(i.e., lowest {111} surface energy of ꢀ2.04 J/m2 [23]) is
higher than the surface energies of the most probable tita-
nia surfaces (see Table 1). Consequently, the Ni nanoparti-
cles usually grows three-dimensionally on the titania
support, even though the Ni/titania interfacial energy is
low in some specific cases (i.e., Ni on rutile {101} in this
study).
4.2. Growth mechanism of titania clusters on the surface
of Ni nanoparticles and its relationship to SMSI
Titania clusters were found to nucleate on the surface of
about 20% Ni nanoparticles (i.e., 5 out of 24 observations)
during the reduction process. These titania clusters may
originate from titania moieties which dissolve in the Ni
solution as a result of the constant mixing that takes place
during the precursor impregnation process [16]. After the
subsequent drying procedure, this wet solution transforms
to a dried precursor coating on the surface of the titania
supports. As the reduction takes place, Ni precipitates
out of the precursor mixture resulting in an increase in
the local concentration of titania species around the Ni
particle. More precursor species diffuse into the vicinity
of the Ni particle and get reduced as they come into contact
with the growing Ni particle. This process leads to a build
up in the concentration of titanium moieties on the surface
of the growing Ni particles and eventually may result in the
nucleation of a titania particle on the Ni surface.
We were not able to observe the onset of the growth of
the first one or two monolayers of the rutile cluster on
Ni{111} surface. However, Fig. 4 indicates that the growth
mechanism of the rutile cluster involves two major steps:
(1) the nucleation of the ledge at the surface of the rutile
cluster, and (2) the diffusion of the rutile moieties (or
atoms) to the ledge to grow to a complete terrace. The
growth mechanism of the rutile cluster on Ni{111} surface
is similar but slightly different to the traditionally defined
two-dimensional layer-on-layer growth mode in the deposi-
tion process of metallic particles on large single crystal
oxide surfaces [6]. The two-dimensional layer-on-layer
growth mode requires that the growth of the next mono-
layer starts after the first monolayer completes [6,27], while
in this study, the growth of the rutile cluster on Ni{111}
surface involves the completion of two or three monolayers
of rutile {101}. We define the growth mode of the
The Ni precursor coated on the titania surface near the
initially nucleated Ni nanoparticles is consumed as the
metal particle grows. Furthermore, dynamic particle sinter-
ing or coalescence is not responsible for the growth of the

P. Li et al. / Surface Science 600 (2006) 693–702
701
Fig. 6. Illustration of the nucleation and growth of the Ni particle on titania support and the titania cluster on the Ni surface.
rutile cluster on Ni{111} surface to be quasi two-dimen-
sional layer-by-layer. A so-called comparable terrace-ledge
growth mechanism has also been frequently observed at
heterogeneous solid–solid interfaces and grain boundaries
[24]. The dynamic nucleation and growth of the titania
cluster on the surface of Ni particles is illustrated in Fig. 6.
The energetics of the growth of rutile clusters on the sur-
faces of Ni particles is the reverse case of that discussed in
Section 4.1 for Ni growing on titania. Two-dimensional
layer-by-layer growth of rutile clusters on the surface
to the {111} atomic planes of Ni. This indicates that the
parallel orientation relationship (i.e., {111}Nik{101}rutile
)
is favorable. Parallel atomic plane matching has not been
found for any Ni nanoparticles supported on anatase.
The nucleation of the titania cluster on the Ni surface is
probably related to the SMSI phenomenon in Ni/titania
system. SMSI was used to describe the drastic changes in
the chemisorption properties of group VIII noble metals
supported on reducible oxide supports [28]. One common
explanation for the SMSI phenomenon is that reduced tita-
nia from the support migrates onto the surface of the metal
particles [29–32]. Regarding this interpretation, HREM on
ex situ reduced catalyst samples have clearly shown the
occurrence of titania surface coatings on metal nanoparticle
surfaces [33–36]. Our ex situ experiments at 500 °C and
800 °C also show a titania coating on Ni surfaces. In our
in situ experiment, significant titania reduction is unlikely
to occur at the temperature of 350 °C used in this study
(titania reduction normally occurs at temperatures above
500 °C [16]). However, it has already been suggested that
small amounts of titania species may dissolve in the impreg-
nation solution and be deposited onto the surface of the
metal particle during reduction [16]. Our results support
this hypothesis and show that nucleation of at least some
titania clusters on the surface of the Ni particles takes place
during catalyst synthesis. However, our results also show
that only a small number of Ni particles have titania parti-
cles on their surfaces suggesting that this mechanism may
make only a small contribution to the SMSI phenomena.
of the Ni nanoparticles is favored while cNi > ccluster
+
c
cluster/Ni, where cNi and ccluster are the surface energies
of Ni nanoparticle and rutile cluster, respectively, and
ccluster/Ni is the interfacial energy between the Ni nanopar-
ticle and rutile cluster. In this case, the Ni{111} surface
energy is much higher than that of rutile {101} surface
(i.e. ꢀ2.04 J/m2 [23] versus ꢀ1.36 J/m2 [20]) and the Ni/
rutile interfacial energy can be low due to the parallel
{111}Ni and {101}rutile at the interface as discussed in
Section 4.1. With this configuration of energies, the condi-
tion for two-dimensional growth of the rutile cluster on the
{111} surface of the Ni nanoparticles can be satisfied. No-
tice that in this case, the substrate (Ni) has higher surface
energy whereas for the Ni on titania case the substrate
(titania) has lower surface energy.
It is found that not only the Ni nanoparticle aligns its
{111} atomic planes to the {101} atomic planes of the
rutile support, but also the rutile cluster on the surface of
the Ni nanoparticles tends to align its {101} atomic planes

702
P. Li et al. / Surface Science 600 (2006) 693–702
[6] C.T. Campbell, Surf. Sci. Rep. 27 (1997) 1.
5. Conclusions
[7] A.J. Van Dillen, R.J. Tero¨rde, D.J. Lensveld, J.W. Geus, K.P. Jong,
J. Catal. 216 (2003) 257.
[8] M. Wu, P.J. Møller, Surf. Sci. 279 (1992) 23.
[9] U. Diebold, J. Pan, T.E. Madey, Surf. Sci. 331–333 (1995) 845.
[10] P. Li, J. Liu, N. Nag, P.A. Crozier, J. Phys. Chem. B 109 (2005)
13883.
[11] P.L. Gai, Adv. Mater. 10 (1998) 1259.
[12] P.L. Hansen, J.B. Wagner, S. Helveg, J.R. Rostrup-Nielsen, B.S.
Clausen, H. Topsøe, Science 295 (2002) 2053.
Nucleation and growth mechanisms for synthesis of Ni
nanoparticles on a high-surface area titania support (i.e.,
a mixture of anatase and rutile) have been studied at atom-
ic resolution. It was found that most Ni nanoparticles
nucleate from a thin uniform Ni precursor coating on the
surface of the titania support. A minority of Ni particles
nucleated directly from irregular patches of the Ni precur-
sor which were also present on the titania support. Ni par-
ticles nucleated on both anatase and rutile grains although
low energy oxide surfaces do not appear to be preferential
nucleation sites. Ni nanoparticles showed a non-wetting
morphology when nucleated on the anatase support, while
it was observed that Ni nanoparticles possess a wetting
morphology when nucleated on the rutile {101}. This is
due to the one-dimensional atomic plane matching across
the Ni/rutile interface causing a lowering of the interfacial
energy. The Ni nanoparticles grow as three-dimensional is-
lands and growth involves the surface diffusion of the Ni
species. Titania clusters have also been found to nucleate
on the surface of the growing Ni particles, indicating that
titania overlayers on metal particles can be directly related
to the synthesis of the Ni/TiO2 catalysts. The growth of the
rutile cluster on the surface of the Ni particle follows the
two-dimensional layer-by-layer growth mode.
[13] P.L. Gai, Top. Catal. 8 (1999) 97.
[14] E.D. Boyes, P.L. Gai, Ultramicroscopy 67 (1997) 219.
[15] R. Sharma, Microsc. Microanal. 7 (2001) 494.
[16] G.B. Raupp, S.A. Stevenson, J.A. Dumesic, S.J. Tauster, R.T.K.
Baker, Metal–support interactions, in: S.A. Stevenson, J.A. Dumesic,
R.T.K. Baker, E. Ruckenstein (Eds.), Catalysis, Sintering, and
Redispersion, Van Nostrand Reinhold Company Inc., New York,
NY, 1987.
[17] E.J. Kirkland, Advanced Computing in Electron Microscopy, Plenum
Press, New York, NY, 1998.
[18] P.D. Nellist, S.J. Pennycook, Science 274 (1996) 413.
[19] T. Akita, P. Lu, S. Ichikawa, K. Tanaka, M. Haruta, Surf. Interface
Anal. 31 (2001) 73.
[20] M. Lazzeri, A. Vittadini, A. Selloni, Phys. Rev. B 63 (2001) 155409.
[21] M. Ramamoorthy, D. Vanderbilt, R.D. King-Smith, Phys. Rev. B 49
(1994) 16721.
[22] W.R. Tyson, W.A. Miller, Surf. Sci. 62 (1977) 267.
[23] M.I. Baskes, Phys. Rev. B 46 (1992) 2727.
[24] J.M. Howe, Interfaces in Materials:Atomic Structure, Kinetics and
Thermodynamics of Solid:Solid, Solid:Liquid and Solid:Vapor Inter-
faces, John Wiley and Sons, New York, NY, 1997.
[25] P. Li, J.M. Howe, W.T. Reynolds Jr., Acta Mater. 52 (2004) 239.
[26] Q. Zhu, A. Zhang, E.D. Williams, R.L. Park, Surf. Sci. 172 (1986)
433.
Acknowledgements
[27] F.C. Frank, J.H. van der Merwe, Proc. Roy. Soc. (London) A198
(1984) 205.
[28] S.J. Tauster, S.C. Fung, R.L. Garten, J. Am. Chem. Soc. 100 (1978)
170.
[29] R.T.K. Baker, J.J. Chludzinski, J.A. Dumesic, J. Catal. 93 (1985) 312.
[30] J.A. Dumesic, S.A. Stevenson, R.D. Sherwood, R.T.K. Baker, J.
Catal. 99 (1985) 79.
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.
[31] S. Bernal, J.J. Calvino, M.A. Cauqui, J.M. Gatica, C. Lpez, J.A.
´
Perez Omil, J.M. Pintado, Catal. Today 77 (2003) 385.
[32] D. Olga, H. Wilhelm, D. Ulrike, Phys. Rev. Lett. 84 (2000) 3646.
[33] A.D. Logan, E.J. Braunschweig, A.K. Datye, D.J. Smith, Langmuir 4
(1988) 827.
References
[34] E.J. Braunschweig, A.D. Logan, A.K. Datye, D.J. Smith, J. Catal.
118 (1989) 227.
[1] A. Kolmakov, D.W. Goodman, The Chemical Record 2 (2002) 446.
[2] R.E. Tanner, I. Goldfarb, M.R. Castell, G.A.D. Briggs, Surf. Sci. 486
(2001) 167.
´
´
[35] S. Bernal, F.J. Botana, J.J. Calvino, C. Lopez Cartes, J.A. Perez
´
Omil, J.M. Rodrıguez-Izquierdo, J. Chem. Soc., Faraday Trans. 92
[3] J. Zhou, Y.C. Kang, D.A. Chen, Surf. Sci. 537 (2003) L429.
[4] C. Xu, X. Lai, G.W. Zajac, D.W. Goodman, Phys. Rev. B 56 (1997)
13464.
(1996) 2799.
[36] S. Bernal, J.J. Calvino, M.A. Cauqui, J.M. Gatica, C. Lo´pez
´
Cartes, J.A. Perez Omil, J.M. Pintado, Catal. Today 77 (2003)
[5] C.R. Henry, Surf. Sci. Rep. 31 (1998) 235.
385.