Surface Science 603 (2009) 2210–2217

Contents lists available at ScienceDirect

Surface Science
journal homepage: www.elsevier.com/locate/susc

Nickel deposition on c-Al2O3 model catalysts: An experimental

and theoretical investigation
Francois Loviat, Izabela Czekaj *, Jörg Wambach, Alexander Wokaun
General Energy Department, Paul Scherrer Institut, 5232 Villigen PSI, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: Recently, surface modifications on a commercial Ni/c-Al2O3 catalyst during the production of methane
Received 19 January 2009 from synthesis gas were investigated by quasi in situ X-ray photoelectron spectroscopy (XPS) [I. Czekaj,
Accepted for publication 27 April 2009 F. Loviat, F. Raimondi, J. Wambach, S. Biollaz, A. Wokaun, Appl. Catal. A: Gen. 329 (2007) 68]. The conclu-
Available online 6 May 2009
sion was that the reactivity and the observed reaction mechanisms on the different Ni particles are influ-
enced directly by both the size and the composition of the particles on the c-Al2O3 support.
Keywords: In this investigation, Ni deposition and cluster growth on model catalyst samples (10 nm thick, poly-
Nickel catalyst
crystalline c-Al2O3 on Si(100)) were investigated by XPS. Several steps in the binding energy during Ni
Alumina support
XPS
deposition indicate changes in the cluster growth. The molecular structure of the catalyst was investi-
c-Al2O3 gated using Density Functional Theory calculations (StoBe) with a cluster model and non-local functional
DFT (RPBE) approach. An Al15O40H35 cluster was selected to represent the c-Al2O3(100) surface. Ni clusters of
Methanation different size were cut from a Ni(100) surface and deposited on the Al15O40H35 cluster in order to validate
Metal deposition the deposition model determined by XPS.
Metal–support interactions Ó 2009 Elsevier B.V. All rights reserved.

1. Introduction
The activity and selectivity of supported metal catalysts are
strongly influenced by the amount of metal employed, the size of
the dispersed metal particles, the composition of the support,
and metal–support interactions. In the case of supported metal cat-
alysts, attention has focused on the effect of dispersion, which in-
volves both a surface size effect and a relation between reactivity
and the electronic properties of various systems. Nickel-based cat-
alysts supported on c-Al2O3, TiO2, Pt, zeolites or SiO2 [2–9] are
interesting examples due to their usage in numerous industrial
processes [2–5]. Previously, it was shown that nickel particles
change their morphology during catalytic reactions by cluster
growth processes, and that part of the active clusters are lifted
from the support due to carbon deposition and carbon whisker for-
mation [1]. The industrial application of these types of catalysts
[2,10] makes the investigation of the role of metal–support interac-
tions for the Ni particle growth and detachment during methana-
tion an important issue. However the complex structure of these
catalysts demands further exploration on model catalysts.
The structure, the electronic properties and the reactivity of
supported model catalysts have been studied applying a large
number of surface science techniques [1]. It was found that the
electronic properties of small metal particles depend not only on
* Corresponding author. Tel.: +41 056 310 4464; fax: +41 056 310 2199.
E-mail address: izabela.czekaj@psi.ch (I. Czekaj).
their size but also on their shape. Thus, the surface structure of
the particles is closely related to the chemisorption properties.
The presence of the support was recognised to play an important
role in the control of the particle morphology. The intrinsic heter-
ogeneity of the supported model catalysts has to be taken into ac-
count to understand in detail the catalytic reactions.
Three different theoretical growth modes during metal deposi-
tion have been suggested [6–9]: ‘‘Volmer–Weber” mode, where the
deposited metal forms clusters immediately on the support;
‘‘Frank van der Merwe” mode describes a layer-by-layer growth;
and ‘‘Stranski–Krastanov” mode, which includes transition from
an initial layer-by-layer to a consecutive three-dimensional cluster
growth at a critical layer thickness. An additional model for the
growth mechanism was suggested by Jacobs et al. [4], who de-
scribe a complex growth mechanism of Ni on Al2O3 when atomic
layer epitaxy is applied. Determining the surface composition with
LEIS and XPS, the authors conclude that in the initial stage nickel
adsorbs as well dispersed atoms and interacts with the strongest
binding sites on the alumina support. These Ni centres are the base
for the subsequent growth of nickel particles. Some literature data
[3,11–13] suggest the possibility that Ni ions migrate from the sur-
face into the bulk forming nickel aluminate, NiAl2O4, where nickel
appears in NiO form. The latter is extremely stable and difficult to
reduce to metallic Ni, which is the catalytically active form. Ni0
particles grow on top of the NiO interface later [1]. Furthermore,
charge transfer processes accompanying the metal–support inter-
actions are important additionally as they strongly influence the

0039-6028/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2009.04.032
 F. Loviat et al. / Surface Science 603 (2009) 2210–2217
electronic structure of the Ni surface and therefore the chemical
activity of Ni-based catalysts. Applying XPS, Sarapatka [5] studied
the interactions between nickel particles and the Al2O3/Al support
as well as the charge transfer from Al2O3 to the deposited Ni. The
observed chemical shift of the nickel core levels are attributed to
a transfer of electrons from anions of the oxide surface to the dis-
persed Ni particles. However, there are still many open questions
concerning the modification of the electronic structure of Ni-parti-
cles and the support (e.g. Al2O3).
Several theoretical investigations (mostly DFT calculations) of
the pure Ni system [14–16] focussed on the stability and diffusivity
of surface species, possible mechanisms for the methane activation
and also try to explain graphite/graphene formation on the surface
of nickel particles, which could lead to the formation of carbon
whiskers. Numerous theoretical studies about pure and modified
c-Al2O3 systems were performed previously [17–19]. However to
our knowledge, theoretical studies about the combined Ni/c-
Al2O3 system are still missing in the literature. A promising study
of combined systems – Mo-based catalysts, mainly Mo-methyli-
dene over c-Al2O3 – was already achieved by Handzlik et al. [17]
using a cluster model. Following the strategy outlined in this com-
bined model, the effect of metal–support interactions, the role of
edges as well as their effect on catalytic reactivity can be studied
for the combined Ni/c-Al2O3 system. It is well known from the lit-
erature [2] that hydrogenation processes take place on metallic Ni0
sites, only. Therefore, any other Ni particles, like oxides, hydroxides
or carbides are not active for the methanation reaction. However,
the co-existence of these inactive Ni compounds either as separate
clusters or even inside an otherwise catalytically active metallic
cluster are observed over c-Al2O3 support, as well [1]. The latter
is likely due to better structural compatibility of these kind of
not-reactive nickel compounds (Ni3C, NiO or Ni(OH)2) with the
c-Al2O3 support (see Table 1) thus forming the interface of the
otherwise metallic nickel particles.
In contrast to the Ni case, theoretical investigations of the me-
tal–support interactions for the adsorption of Pd on c-Al2O3 have
been published already [21,22]. The conclusions are that metal
atoms prefer tetrahedral Al sites and create localised atom–surface
bonding. Metal–support interface formation is determined by the
acid properties of the cationic sites available at support. However,
these studies include only single metal adsorption on the support,
which correspond only to a very low metal coverage.
When atoms or small metal particles are deposited on a surface,
they may create geometrical and electronic structures different
from their bulk. This is due to differences in the lattice constants
between metal and support. By starting from a small cluster-level,
orbitals still have discrete levels. By increasing the number of
atoms in the particles so that the particles reach several nm diam-
eters, a band structure appears [23]. In this sense, a model system
must consider several surface scenarios, namely the presence of
small clusters and large agglomerates, as well as different surface
Table 1
Comparison of structure parameters.
Compound M and Y definition M–M distance (Å)
(a) NiYx compounds and support [24–26].
Ni Nimetal 2.49
NiO Ni, O 2.95
Ni(OH)2 Ni, OH 2.71
NiAl2O4 Ni or Al, O 2.85
c-Al2O3 Al, O 2.78
Compound M–M lateral distance (Å) M–M distance (Å)
(b) Si, SiO2 and c-Al2O3 compounds with cubic structure [32,33]
Si 3.80/5.38 2.33/3.80
SiO2 7.15/5.05 3.09/5.92
c-Al2O3 7.91/2.78 3.28/6.25
2211
compounds. Furthermore, small particles require studies including
support interactions, while in the case of large particles (with sizes
of several nm) on the support just a (pure) metal model will be va-
lid for the description.
The goal of the theoretical part of this work is to define the
geometry and favourable localisation of metal particles on the
support (disregarding steps or kink sites), metal–support interac-
tions and the role of the support in the modification of the geomet-
rical/electronic structure of metal particles and active sites. For the
experiments, model catalysts have been chosen due to the fact that
they (a) possess flat surfaces, which allow the application of AFM,
SE, etc., and (b) allow a sufficient electron tunnelling through the
(10 nm thin) c-Al2O3 layer, making XPS and AES investigations
possible. This ‘‘ideal” definition of the surface allows the compari-
son of experimental data with theoretical modelling, leading to
additional insights into the topology of metal particles on the
support. The interactions of different nickel clusters are presented
in details as a prerequisite to understanding the role of metal–
support interactions in catalytic activity and stability of metal
particles.
This paper consists of the following parts: information about
the crystal structure of the material used in our theoretical calcu-
lations is presented in Section 2. Experimental and computational
details are described in Section 3. The results are presented in Sec-
tion 4. In Section 4.1, THE results obtained from XPS experiments
are presented and the role of the deposition time with regard to
changes of the chemical state of surface is shown. The DFT model-
ling of Ni/Al2O3 interface is described in Section 4.2. Finally, the
conclusions are summarised in Section 5.
2. Crystal structure of Al2O3 and nickel clusters
The cubic phase [24,25] of defective alumina spinel, Al2O3, is de-
scribed by the space group Fd-3m (No. 227) with lattice constants
a = b = c = 7.911 Å. The crystal unit cell of defective alumina spinel
contains 56 atoms, where Al occupies two different types of posi-
tions – 8a, 16d, and oxygen 32e sites.
Inside the bulk both AlO6 octahedra and AlO4 tetrahedra are
present with Al–O distances equal 1.94 Å and 1.78 Å, respectively.
The bulk is built by sixfold and fourfold coordinated aluminium
and threefold and fourfold coordinated oxygen. Fig. 1 shows the
structure of the c-Al2O3(100) surface. Two different aluminium
sites with fourfold and fivefold coordinated aluminium, AlO4 and
AlO5, as well as two different threefold coordinated oxygen sites,
O(3) and O0 (3), have been distinguished. It is important to note that
Al(4) is always located below the Al(5) centres and that the rows of
AlO5 pyramids are separated from each others by rows of AlO4
tetrahedra.
Two surface planes are predominantly detected in the diffrac-
tion patterns of c-Al2O3, namely the (1 1 0) and (1 0 0) surfaces.
Following some previous studies of alumina by Handzlik et al.
Differences in respect to Ni/c-Al2O3 (%) M-Y distance ((Å) Crystal structure
–/10.4 – Cubic
+18.5/+6.1 2.08 Cubic
+8.8/2.5 2.01, 2.39 Trigonal
+14.5/+2.5 1.83, 1.96 Cubic
+11.6/– 1.78, 1.93 Cubic
Differences in respect to c-Al2O3 (%) M–O/O–O distance (Å)
28.9/39.2 –
5.8/5.3 1.55/2.53
– 1.78, 1.94/2.78
2212 F. Loviat et al. / Surface Science 603 (2009) 2210–2217
Fig. 1. Crystal structure of the c-Al2O3(1 0 0) surface: (a) top view and (b) side view. Al (grey spheres) and O (red spheres). Note: the radii of the spheres do not correspond to
the real radii. This is made for better visibility of active centres. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of
this article.)
[17], the (1 0 0) surface was chosen for our studies. Three different
distances are observed between the aluminium centres on the
(1 0 0) surface of alumina: (i) 3.28 Å between Al(4)–Al(5) centres,
(ii) 2.79 Å between Al(5)–Al(5) centres in the AlO5 row and (iii)
5.59 Å between Al(4)–Al(4) centres. The surface exhibits specific
holes with a distance of 2.90 Å and 2.79 Å between the O(3) cen-
tres. All these surface features are important, since sufficient com-
patibility of the (1 0 0) surface with the Ni particles is necessary in
order to obtain stable Ni-clusters on the c-Al2O3 support.
Concerning nickel particles, the cubic phase [26] of metallic
nickel is described by the space group Fm-3m (No. 225) with lat-
tice constants a = b = c = 3.5239 Å. The crystal unit cell of nickel
contains four atoms, where Ni occupies 4a positions. The distance
between Ni–Ni is equal to 2.49 Å. Each Ni atoms has six neighbours
at the surface as well as three in the sub-surface.
3. Technical details
3.1. Experimental details
The XPS measurements were carried out in a VG ESCALAB 220i
XL (VG Scientific) electron spectrometer. XPS measurements were
made using non-monochromatic Mg Ka (1253.6 eV) radiation.
While measuring, the chamber pressure was always lower than
1  109 Torr. The electron energy analyzer was operated in the
constant pass energy mode with a pass energy of 50 eV for survey
scans and 20 eV for detailed scans of selected spectral regions. The
conductivity of the samples was high enough to keep sample
charging below a few eV. For charge compensation, the binding
energy (BE) scale was adjusted by setting the main C1s peak to
284.5 eV. Quantification of the XP spectra was carried out using
CasaXPS [27] and applying the transmission function of the elec-
tron energy analyzer and cross-sections calculated by Scofield.
The spectra were deconvoluted by applying Gaussian–Lorentzian
line-shapes with a Shirley-type background.
Attached via a distribution chamber is a preparation chamber,
which is equipped with a home-made Metal Evaporator (ME).
The latter can evaporate thermally two metals (Ni and Pd), stored
and heated independently in separate chambers. The evaporation
rate is determined using a quartz crystal microbalance (QCM).
For a detailed description of the set-up see [28]. The systems base
pressure is below 1  1010 Torr and during metal evaporation the
pressure in the preparation chamber was always below
1  109 Torr. XPS characterisation was performed before and after
any Ni evaporation and without exposure of the catalyst to air.
The c-Al2O3/Si model substrates were prepared in the Labora-
tory for Neutron Scattering at the Paul Scherrer Institut by M. Hor-
isberger. Thick films of aluminium oxide (10 nm) were deposited
on a Si(1 0 0) wafers by argon sputtering of an aluminium target
in a Leybold Z600 DC magnetron sputtering apparatus. The partial
pressures of argon and oxygen during the deposition were
3.2  103 mbar. The aluminium oxide deposition rate of
0.25 nm/s was determined with a QCM. The structure, morphology
and chemical composition of c-Al2O3/Si substrates were investi-
gated by the combined application of XPS, AES, SE, and AFM.
Merckling et al. [29,30] reported that c-Al2O3 grows epitaxially
on Si(0 0 1) in spite of the large apparent large lattice mismatch
(>30%) between c-Al2O3 and Si. This is possible when c-Al2O3
 F. Loviat et al. / Surface Science 603 (2009) 2210–2217
grows with its oxide lattice being rotated by 45° with respect to the
silicon lattice. This reduces the resultant lattice mismatch to 3.5%.
Further, their studies showed that Al2O3 films grow pseudomor-
phically on Si(1 0 0) up to thickness of 2 nm, and that for higher
thicknesses, a cubic-to-hexagonal surface phase transition occurs.
Another possibility of overcoming the above-mentioned lattice
mismatch is described by Jung et al. [31]. The authors investigated
the hetero-epitaxial growth of c-Al2O3 films on Si substrates.
Atomically smooth single-crystalline epitaxial c-Al2O3 layers were
successfully grown on Si(1 1 1) at growth temperatures between
650 °C and 750 °C. When growing Al2O3 films at temperatures be-
low 650 °C, initially a SiO2 layer is being formed on Si, on which
polycrystalline Al2O3 films are growing consecutively. This SiO2
interlayer thus forms a ‘‘matching” interface between the silicon
and the Al2O3 helping to overcome the lattice mismatch (see Table
1b).
Introduced into vacuum, the samples were cleaned by an initial
flash to 573 K and a subsequent sputtering (at 1 keV, 10 lA ion cur-
rent) in 105 mbar of Ar, annealed in 5  106 mbar oxygen, and fi-
nally cooled down maintaining the oxygen pressure until the
substrate temperature decreased below 373 K. After two cleaning
cycles, the surface was considered as free of any external contam-
ination. The images of the c-Al2O3/Si substrates obtained by AFM
did not allow distinguishing any structure on the surface, indicat-
ing therefore a very smooth morphology. SE and XPS depth profil-
ing suggested an additional layer of SiO2 with a few nm thickness
at the interface between Si and Al2O3. This SiO2 interface helps
overcoming the above-mentioned lattice mismatch.
The Ni/c-Al2O3/Si model catalysts were prepared by depositing
various amount of Ni from the metal evaporator onto the c-Al2O3/
Si substrates at substrate temperatures of 623 K. The growth
mechanism of nickel over c-Al2O3 has been investigated by a com-
bination of XPS and AFM analysis.
3.2. Computational details
In our studies the Ni and c-Al2O3 support surfaces are modelled
by clusters of different size and geometry, which reflect local sec-
tions of the ideal surface. Clusters are constructed by a successive
addition of neighbouring shells to a small core structure. This pro-
cedure is repeated until convergence of the electronic properties of
the clusters is obtained.
For the c-Al2O3 cluster representing the support surface, formal
valence charge saturation (based on Al3+ and O2) is imposed and
cluster neutrality with respect to the surface is achieved by satu-
rating the peripheral oxygen centres with hydrogen atoms placed
at the standard OH distance, ROH = 0.97 Å, in the direction of the
Fig. 2. Top view (a) and side view (b) of the Al15O40H35 model cluster describing the c-Al2O3(100) surface.
2213
appropriate broken Al–O bonds. The thus obtained cluster
Al15O40H35 has been selected for the investigation (Fig. 2). The
Al15O40H35 cluster consists of all characteristic centres observed
on the Al2O3 (1 0 0) surface (for comparison see Fig. 1). For our
investigations we have chosen the c-Al2O3 surface with a (1 0 0)
termination based on previous investigations [17,18] as well as
experimental results [28], which show in details the structure of
a model catalyst with specific ‘‘valley-hill” rowed structure.
The clusters representing nickel (Ni2, Ni7, and Ni9 cluster) were
cut from the ideal bulk structure without saturation by hydrogen
or point charges to realistically simulate small surface particles
(see Fig. 3a).
Fig. 3b–d shows the results of a geometric optimization of the
chosen Ni clusters on c-Al2O3 cluster representing the Nix/
Al15O40H35 system (x = 2,7,9) used for the theoretical modelling.
The electronic structure of all clusters was calculated by ab ini-
tio density functional theory (DFT) methods (program code StoBe
[34]) using the non-local generalized gradient corrected function-
als according to Perdew, Burke, and Ernzerhof (RPBE) [35,36], in
order to account for electron exchange and correlation. All Kohn–
Sham orbitals are represented by linear combinations of atomic
orbitals (LCAO’s) using extended basis sets of contracted Gaussians
from atom optimizations [37,38]. Detailed analyses of the elec-
tronic structure in the clusters are carried out using Mulliken pop-
ulations [39] and Mayer bond order indices [40,41]. During
relaxation, the nickel atoms in the supported clusters had allowed
to move in 3D space.
4. Results
4.1. XPS results of Ni deposition on Al2O3: the role of deposition time
with regard to changes in the chemical state of surface
Fig. 4 gives the Ni 2p3/2/Al 2p ratio versus the deposition (ex-
pressed in seconds) of nickel onto the c-Al2O3 support. Dots repre-
sent the experimentally obtained data. The deposition was
performed with a deposition rate corresponding (theoretically) to
0.75 ML/min. The red line symbolizes the theoretically expected
Ni 2p3/2/Al 2p ratio for a pure layer-by-layer growth. For this esti-
mation, we presumed that this (theoretical) Ni monolayer would
have the dense geometry of a Ni(1 0 0) monolayer on the c-Al2O3.
Up to a deposition time t  45 s, our experimental data follow al-
most the expected layer-by-layer growth line. This suggests that
in the initial phase Ni does not form clusters immediately, but 1D
(or possibly 2D) agglomerates. In a second stage, the Ni 2p3/2/Al
2s ratio is below the ‘‘layer-by-layer” line. This makes it likely that
in this phase 3D Ni clusters are growing on the surface. We
2214 F. Loviat et al. / Surface Science 603 (2009) 2210–2217

Fig. 3. Models of Ni clusters: (a) cut from Ni(100) surface; as well as deposited + relaxed on the c-Al2O3(100) surface represented by Al15O40H35 cluster: (b) Ni2 cluster, (c) Ni7
cluster, and (d) Ni9 cluster.

Fig. 4. Changes of Ni 2p3/2/Al 2p ratio during deposition of nickel onto the Al2O3
support. The red line corresponds to the theoretical prediction of the Ni 2p3/2/Al 2p
ratio for a layer-by-layer growth. The theoretical depositions times for 0.5 ML and
1.0 ML of Ni (layer-by-layer growth) are indicated, too. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of
this article.)

conclude that Ni deposition on c-Al2O3 support follows a kind of

‘‘modified” Stranski–Krastanov growth mode under the applied
experimental conditions, which is in accordance to the findings of
Jacobs et al. [4].
A support for the suggested change in the growth mode can be
found by XPS. In Fig. 5, the spectra of the Ni 2p region for several Ni
deposition times are displayed, and in Fig. 6 the summarised
changes of Ni 2p3/2 binding energies during deposition are given.
The change of the binding energy can be separated into three parts,
as indicated by the two vertical lines. In the initial part, the first
found BE value (854.0 eV) is close to published Ni oxide (NiO) BE
data [12,13,42,43]. This can be interpreted as single Ni atoms being
deposited and chemically modified by being bonded strongly to
the oxygen of the support. The further initial part up to about 45
s is characterised by a fast decrease of the BE down to circa
853.3 eV. In the second part, between 50 s and about 150 s deposi-
tion times, the slope of the BE change is decreased and the BE value
is lowered to 853.0 eV. The third part, i.e. deposition times more
than 170 s, is characterised by a slow approach to 852.75 eV (ob-
tained for depositions times >350 s (not shown)), the value of
metallic Ni as indicated by the horizontal dash-dotted line.
Fig. 5. XP spectra of the Ni 2p region. The different traces correspond to various
deposition times: (a) 10 s; (b) 50 s; (c) 100 s; (d) 150 s; (e) 210 s; and (f) 380 s. The
two bottom traces show spectra of bulk Ni and NiO, and are included as reference.
Interestingly, the time of the transition from the first to the sec-
ond part (45 s) is comparable with the time, where the deviation
of the experimentally derived Ni 2p3/2/Al 2p ratio from the ‘‘layer-
by-layer” ratio starts, as shown in Fig. 4 (see the vertical line). This
makes it likely that in the second stage additional Ni atoms are
added to the already existing ones forming 3D Ni agglomerates.
Finally, the third part indicates that continual Ni deposition and
cluster growth processes leads to extended three-dimensional Ni
clusters, exhibiting a band structure typical for bulk Ni. This behav-
iour is in agreement with a model described by Wertheim [44]. A
 F. Loviat et al. / Surface Science 603 (2009) 2210–2217 2215

clusters Al15O40H35 and Nix/Al15O40H35 (x = 2,7,9). The data are bas-

Fig. 6. Changes of XPS binding energies of the Ni 2p3/2 peak during deposition of
nickel on c-Al2O3. (a) single atoms, (b) 1D (or 2D) growth, and (c) 3D growth.
comparable behaviour and BE shifts as function of the aggregate
size have been measured by XPS for different metals deposited
on alumina and other substrates [45,46].
This behaviour can be explained with the drastic changes of the
electronic structure of the deposits as the cluster size increases.
The size of the cluster determine the final state in the photoemis-
sion process, because after the photoemission the positive charge
left on the aggregate can be less screened respectively delocalised
when compared with an extended metallic system. This results in a
shift of the XP spectrum and corresponds to the Coulomb energy of
the localised charge. It was found that the shift is proportional to
the reciprocal particle diameter [44,47]. Consequently, the binding
energies of small metal aggregates should be found at higher
values as is indeed observed in many cases.
4.2. DFT modelling of Ni/Al2O3 interface
Table 2 presents information about atomic charges (Mulliken
analysis), distances (Å) and bond orders (Mayer bond analysis) of
ing on optimization calculations described in Section 3.2. Geomet-
ric structures of all ideal clusters are presented in Figs. 2 and 3a.
The charge of aluminium centres on the (1 0 0) surface
increases in the order q[Al(5)] < q[Al(6)] < q[Al(4)]. The Al(5) centre
P
is stronger bound to the surface ( b.o. = 3.1) than Al(4)
P
( b.o. = 2.3), but is not fully coordinated. Therefore, the Al(5) cen-
tre can be easily oxidized or hydratized.
The two types of surface three-coordinated oxygen centres, O(3)
P
and O0 (3), are similarly bound to the surface ( b.o.  2.0), but
0
show a different ionic character. The O (3) centres, connected with
five and four coordinated aluminium centres, have higher charge
and are supposed to be more reactive than O(3).
The Al15O40H35 cluster was found to be a good representation
for the electronic states of both oxygen O0 (3) and O(3) as well as
three types of aluminium centres, Al(4), Al(5) and Al(6), present
on the Al2O3(100) surface. Taking into account computational time
(including optimization and metal deposition at the surface) as
well as accuracy of the electronic structure, the Al15O40H35 cluster
seems to be a good enough compromise of cluster convergence cri-
teria and therefore was chosen for all further considerations in me-
tal deposition studies.
The stabilization energy of nickel particles deposition on
Al15O40H35 cluster is calculated as:
Edep ðM n =Al2 O3 Þ ¼ Etot ðMn =Al15 O40 H35 Þ  Etot ðAl15 O40 H35 Þ
 nEtot ðMÞ;
where Etot(Mn/Al15O40H35) is the total energy of the metal deposited
at the Al2O3 surface. Etot(Al15O40H35) and Etot(M) are the total ener-
gies of pure Al15O40H35 and the M atom, respectively. In all cases,
the first layer of nickel deposits on the Al2O3(100) in positions clo-
ser to O(3) centres with different stabilization energy per Ni atom
(for Ni2: 0.82 eV, for Ni7: 0.90 eV and for Ni9: 0.71 eV). The
determined stabilization energies of the different Ni clusters are
in the range obtained for the adsorption of different metals, mainly
Pd, on c-Al2O3 [19,20]. The deposited nickel influences the elec-
tronic structure of c-Al2O3 and is strongly bound to octahedral
O(3), as can be seen from the bond order analysis shown in Table
2. Interesting results are obtained for the aluminium ions Al(4),

Table 2
Results obtained by DFT calculations summarising atomic charges, distances (Å) and bond orders (Mayer bond analysis) of (a) the Al15O40H35 cluster, (b) the Ni2 cluster at
Al15O40H35 cluster, (c) the Ni7 cluster at Al15O40H35 cluster, and (d) the Ni9 cluster at Al15O40H35 cluster.

(a) (b) (c) (d)

Cluster Al15O40H35 Ni2/Al15O40H35 Ni7/Al15O40H35 Ni9/Al15O40H35
Centre Charge [eV]
Al(4) +1.58 +1.87 +2.20 +2.16
Al(5) +0.99 +0.91 +0.80 +1.01
Al(6) +1.11 +1.13 +1.13 +1.15
O(3) 0.68 0.82 0.89 0.92
O0 (3) 0.75 0.78 0.84 0.89
Ni1st – +0.28 +0.38 +0.35
Bond Distance [Å] Bond Order
O(3)–Al(5) 1.94 0.73 0.61 0.66 0.66
O(3)–Al(5) 1.94 0.66 0.58 0.45 0.51
O(3)–Al(6) 1.94 0.62 0.52 0.54 0.53
O(3)–Ni (see diff. Nix) – 0.21 (2.16 Å) 0.17 (2.04 Å) 0.15 (2.09 Å)
P
b.o. 2.01 1.92 1.82 1.85
O’(3)–Al(5) 1.94 0.66 0.61 0.58 0.57
O’(3)–Al(5) 1.94 0.66 0.61 0.58 0.57
O’(3)–Al(4) 1.78 0.63 0.43 0.48 0.38
O’(3)–Ni (see diff. Nix) – 0.08 (2.57 Å) 0.01 (2.96 Å) 0.004 (3.10 Å)
P
b.o. 1.95 1.73 1.65 1.50
Ni1st–Ni1st (see diff. Nix) – 0.34 (3.53 Å) 0.08 (4.52 Å) 0.09 (4.77 Å)
EHOMO/LUMO (eV) 4.51/4.46 3.80/3.70 4.32/4.14 4.12/4.09
ENi_deposition/cluster (eV) – 1.64 6.27 6.41
ENi_deposition/Niatom (eV) 0.82 0.90 0.71
2216 F. Loviat et al. / Surface Science 603 (2009) 2210–2217
Fig. 7. Topography of first (interface) Ni atoms (green spheres) for the relaxed Ni9 cluster from an Al15O40H35 cluster: (a) all Ni interface atoms included (hNi  0.4 ML); (b)
only Ni atoms in hole positions included (hNi  0.2 ML). Black dots symbolize the position of Ni atoms in an ideal Ni(1 0 0) monolayer. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)
for which the positive charge is found to be increased after a higher
Ni deposition. Together with the changes of the charges of the oxy-
gen centres O(3) and O0 (3) as well as of the neighbouring nickel
atoms, Ni1st, an increasing polarity at surface has to be stated. This
is in agreement with a decrease of the oxygen bond orders, which
describe a covalent character of the system, also. As a consequence,
due to the deposition of metallic particles the surface of Ni/Al2O3
system becomes more ionic.
The largest investigated nickel cluster (Ni9) creates many inter-
esting structures at the support surface. Fig. 7 shows the interface
atoms of the Ni9 cluster after relaxation (the additional atoms have
been removed; the entire cluster can be seen in Fig. 3d). The figure
was created by multiplying the calculated Ni/Al2O3 cluster in x and
y directions. The interface nickel atoms are important due to the
fact that the direct Ni–Al2O3 interaction can be directly responsible
for the stability of growing metal clusters on alumina. The black
dots, added for comparison with the relaxed structure, symbolize
the mismatch of an ‘‘ideal” Ni(1 0 0) monolayer with the c-Al2O3
surface. Green spheres give the positions of Ni atoms with coverage
h  0.75 ML (Fig. 7a) and h  0.25 ML (Fig. 7b). As visible in Fig. 7a,
strong vertical and lateral re-arrangements of the interface Ni
atoms in the Ni9 clusters with respect to the position on the
Ni(1 0 0) surface are indicated by DFT. Part of the nickel atoms
seem to prefer the ‘‘valley regions” between AlO5 rows and close
to O(3) centres. This is visible more clearly in Fig. 7b. The nickel
centres (‘‘Ni1st”) in that row increase their respective distance from
3.5 to 4.8Å, and approach almost the O(3)–O(3) distance along
the AlO5 rows (= x-axis), which is about 5.6 Å. The neighbouring
nickel atoms (‘‘Ni2nd”) are three displaced in three dimensions: ver-
tically (along the x-axis) and additionally by a slight relocation
along the y-axis. The lateral movement of the ‘‘Ni2nd” atoms is
more prominent, most likely due to the repulsion of nickel and
the aluminium Al(5) centres, and results in a lateral distance be-
tween ‘‘Ni2nd”–‘‘Ni1st” layers of about 0.54 Å.
The Al(6) centres, being located in subsurface region, are not
influenced by the Ni deposition. This suggests that, after deposition
of a very low number of metal atoms on c-Al2O3, only the local sur-
face structure is influenced, in particular the neighbouring centres
of Ni, such as: Al(4), Al(5), O(3) and O0 (3). From the results pre-
sented in Table 2 for three different nickel clusters, we can derive
that the O(3) centres are mostly influenced creating strong bonds
with the first (interface layer) nickel atoms. This finding is in good
accordance with our experimental results (see Fig. 6), where we
observe the existence of electronically strongly altered Ni being ad-
sorbed on c-Al2O3 in the first part of the deposition. As mentioned
above, we presume that initially single Ni atoms are deposited
(first 25 s of deposition), and that further Ni deposition (up to
120 s) leads to the formation of either 1D Ni chains or 2D Ni
agglomerates.
A further important question is the suggested existence of nick-
el aluminate in the interface and its influence on cluster stability at
alumina. Studies going in that direction are presently running in
our laboratory and will be presented elsewhere [48].
5. Conclusions
Nickel is stabilized on the c-Al2O3 surface influencing the elec-
tronic properties of the newly formed surface. Our DFT data
suggest that at low coverages (60.2 ML) Ni prefers being
localised in AlO4 tetrahedra between rows of AlO5. The DFT
results correspond well with the experimental data of the initial
stage of Ni deposition, where the formation of a partial Ni mono-
layer is suggested. Additional Ni deposition leads first to
non-metallic three-dimensional agglomerates, which by continual
deposition are finally transferred to metallic Ni clusters on
the surface. For Ni7 and Ni9 clusters, the initially deposited Ni
atoms, which represent the interface nickel atoms (‘‘Ni1st”), are
bound strongly to the oxygen of the support and are located in
positions closer to O(3) centres as well as between rows of
AlO5 with adsorption energy, which varies with the size of the
cluster.
 F. Loviat et al. / Surface Science 603 (2009) 2210–2217 2217

Acknowledgements [21] A.M. Marquez, J.F. Sanz, Appl. Surf. Sci. 238 (2004) 82.
[22] M.C. Valero, P. Raybaud, P. Sautet, J. Phys. Chem. B 110 (2006) 1759.
[23] H.J. Freund, M. Bäumer, H. Kuhlenbeck, Adv. Catal. 45 (2000) 333.
The calculations were done partially using the unix farm as well [24] R.-S. Zhou, R.L. Snyder, Acta Crystallogr. B 47 (1991) 617.
as the linux farm at the Paul Scherrer Institut. We would like to ex- [25] E.J.W. Verwey, Z. Kristallograp. 91 (1935) 317.
[26] R.W.G. Wyckoff, Crystal Structures, vol. I, Interscience Publishers John Wiley &
press our thanks to M. Horisberger for preparation of the c-alu-
Sons Inc., New York, London, Sydney, 1965.
mina model support as well as to R. Kötz for Spectroscopic [27] N. Fairley, CasaXPS Software, Version 2.3.12, 2006.
Ellipsometry. [28] F. Loviat, Photoassisted activation of methane with a xenon excimer lamp, PhD
Thesisß ETH 2008.
[29] C. Merckling, M. El-Kazz, V. Favre-Nicolin, M. Gendry, Y. Robach, G. Grene, G.
References Hollinger, Thin Solid Films 515 (2007) 6479.
[30] C. Merckling, M. El-Kazzi, G. Delhaye, M. Gendry, G. Saint-Girons, G. Hollinger,
[1] I. Czekaj, F. Loviat, F. Raimondi, J. Wambach, S. Biollaz, A. Wokaun, Appl. Catal. L. Largeau, G. Patriarche, J Appl. Phys. 102 (2007) 024101.
A: Gen. 329 (2007) 68. [31] Y.-C. Jung, H. Miura, M. Ishida, J. Crystal Growth 201/202 (1999) 648.
[2] G. Ertl, H. Knözinger, J. Weitkamp, Handbook of Heterogeneous Catalysis, [32] D.N. Batchelder, R.O. Simmons, J. Chem. Phys. 41 (1964) 2324.
Wiley Company, Weinheim, 1997. [33] W.W. Schmahl, I.P. Swainson, M.T. Dove, A. Graeme-Barber, Z. Kristallogr. 201
[3] D. Nazimek, A. Machocki, T. Borowiecki, Adsorption Sci. Technol. 16 (1998) (1992) 125.
747. [34] The program package StoBe is a modified version of the DFT-LCGTO program
[4] J.P. Jacobs, L.P. Lindfors, J.G.H. Reintjes, O. Jylhä, H.H. Brongersma, Catal. Lett. package DeMon, originally developed by A. St.-Amant and D. Salahub
25 (1994) 315. (University of Montreal), with extensions by L.G.M. Pettersson and K.
[5] T.J. Sarapatka, Chem. Phys. Lett. 212 (1993) 37. Hermann.
[6] C.S. Shern, J.S. Tsay, T. Fu, Appl. Surf. Sci. 92 (1996) 74. [35] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865.
[7] R.E. Tanner, I. Goldfarb, M.R. Castell, G.A.D. Briggs, Surf. Sci. 486 (2001) 167. [36] B. Hammer, L.B. Hansen, J.K. Nørskov, Phys. Rev. B 59 (1999) 7413.
[8] J.D. Carey, L.L. Ong, S.R.P. Silva, Nanotechnology 14 (2003) 1223. [37] J.K. Labanowski, J.W. Anzelm (Eds.), Density Functional Methods in Chemistry,
[9] M.A. Karolewski, Surf. Sci. 517 (2002) 138. Springer-Verlag, New York, 1991.
[10] M.C. Seemann, T.J. Schildhauer, S.M.A. Biollaz, S. Stucki, A. Wokaun, Appl. Catal. [38] N. Godbout, D.R. Salahub, J. Andzelm, E. Wimmer, Can. J. Phys. 70 (1992) 560.
A: Gen. 313 (2006) 14. [39] R.S. Mulliken, J. Chem. Phys. 23 (1955) 1833. 1841, 2388, 2343.
[11] R. Lamber, G. Schulz-Ekloff, Surf. Sci. 258 (1991) 107. [40] I. Mayer, Chem. Phys. Lett. 97 (1983) 270.
[12] K. Shih, J.O. Leckie, J. Euro. Ceramic Soc. 27 (2007) 91. [41] I. Mayer, J. Mol. Struct. (Theochem) 34 (1987) 81.
[13] R. Lamber, G. Schulz-Ekloff, Surf. Sci. 258 (1991) A595. [42] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, J. Chastain (Eds.), Handbook
[14] J. Sehested, J.A.P. Gelten, I.N. Remediakis, H. Bengaard, J.K. Nørskov, J. Catal. of X-ray Photoelectron Spectroscopy, Perkin–Elmer Corporation, Physical
223 (2004) 432. Electronics Division, Eden Prairie, MN, USA, 1992.
[15] H.S. Bengaard, J. Nørskov, J. Sehested, B.S. Clausen, L.P. Nielsen, A.M. [43] C.D. Wagner, A.V. Naumkin, A. Kraut-Vass, J.W. Allison, Cedric J. Powell, J.R.
Molenbroek, J.R. Rostrup-Nielsen, J. Catal. 209 (2002) 365. Rumble Jr., NIST X-ray Photoelectron Spectroscopy Database, NIST Standard
[16] F. Abild-Pedersen, O. Lytken, J. Engbaek, G. Nielsen, I. Chorkendorff, J.K. Reference Database 20, Version 3.4 (Web Version), National Institute of
Nørskov, Surf. Sci. 590 (2005) 127. Standards and Technology, Gaithersburg, USA, 2006.
[17] J. Handzlik, J. Ogonowski, R. Tokarz-Sobieraj, Catal. Today 101 (2005) 163. [44] G.K. Wertheim, Z. Phys. B 66 (1987) 53.
[18] M. Digne, P. Sautet, P. Raybaud, P. Euzen, H. Toulhoat, J. Catal. 226 (2004) 54. [45] M. Bäumer, H.-J. Freund, Prog. Surf. Sci. 61 (1999) 127. and literature therein.
[19] A. Ionescu, A. Allouche, J.P. Aycard, M. Rajzmann, F. Hutschka, J. Phys. Chem. B [46] D.-Q. Yang, E. Sacher, Appl. Surf. Sci. 195 (2002) 187.
106 (2002) 9359. [47] G.K. Wertheim, Z. Phys. D 12 (1989) 319.
[20] C. Arrouvel, M. Breysse, H. Toulhoat, P. Raybaud, J. Catal. 232 (2005) 161. [48] I. Czekaj, F. Loviat, J. Wambach, A. Wokaun, Chimia 63 (2009) 193.