Fu 2007

Surface Science Reports 62 (2007) 431498
www.elsevier.com/locate/surfrep
Interaction of nanostructured metal overlayers with oxide surfaces

Qiang Fu a,b,1 , Thomas Wagner a,
a Max-Planck-Institut fur Metallforschung, Heisenbergstrasse 3, D-70569 Stuttgart, Germany
b State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, The Chinese Academy of Sciences, Zhongshan Road 457, Dalian 116023, PR China
Accepted 29 July 2007
Abstract
Interactions between metals and oxides are key factors to determine the performance of metal/oxide heterojunctions, particularly in
nanotechnology, where the miniaturization of devices down to the nanoregime leads to an enormous increase in the density of interfaces. One
central issue of concern in engineering metal/oxide interfaces is to understand and control the interactions which consist of two fundamental
aspects: (i) interfacial charge redistribution electronic interaction, and (ii) interfacial atom transport chemical interaction. The present paper
focuses on recent advances in both electronic and atomic level understanding of the metaloxide interactions at temperatures below 1000 C,
with special emphasis on model systems like ultrathin metal overlayers or metal nanoclusters supported on well-defined oxide surfaces. The
important factors determining the metaloxide interactions are provided. Guidelines are given in order to predict the interactions in such systems,
and methods to desirably tune them are suggested.
The review starts with a brief summary of the physics and chemistry of heterophase interface contacts. Basic concepts for quantifying the
electronic interaction at metal/oxide interfaces are compared to well-developed contact theories and calculation methods. The chemical interaction
between metals and oxides, i.e., the interface chemical reaction, is described in terms of its thermodynamics and kinetics. We review the different
chemical driving forces and the influence of kinetics on interface reactions, proposing a strong interplay between the chemical interaction and
electronic interaction, which is decisive for the final interfacial reactivity. In addition, a brief review of solidgas interface reactions (oxidation of
metal surfaces and etching of semiconductor surfaces) is given, in addition to a comparison of a similar mechanism dominating in solidsolid and
solidgas interface reactions.
The main body of the paper reviews experimental and theoretical results from the literature concerning the interactions between metals and
oxides (TiO2 , SrTiO3 , Al2 O3 , MgO, SiO2 , etc.). Chemical reactions, e.g., redox reactions, encapsulation reactions, and alloy formation reactions,
are highlighted for metals in contact with mixed conducting oxides of TiO2 and SrTiO3 . The dependence of the chemical interactions on the
electronic structure of the contacting metal and oxide phases is demonstrated. This dependence originates from the interplay between interfacial
space charge transfer and diffusion of ionic defects across interfaces. Interactions between metals and insulating oxides, such as Al2 O3 , MgO, and
SiO2 , are strongly confined to the interfaces. Literature results are cited which discuss how the metal/oxide interactions vary with oxide surface
properties (surface defects, surface termination, surface hydroxylation, etc.). However, on the surfaces of thin oxide films grown on conducting
supports, the effect of the conducting substrates on metaloxide interactions should be carefully considered.
In the summary, we conclude how variations in the electronic structure of the metal/oxide junctions enable one to tune the interfacial
reactivity and, furthermore, control the macroscopic properties of the interfaces. This includes strong metalsupport interactions (SMSI), catalytic
performance, electrical, and mechanical properties.
c 2007 Elsevier B.V. All rights reserved.

Keywords: Oxide surfaces; Metal films; Oxide films; Model systems; Metalsupport interaction; Interface reaction; Charge transfer; Titanium oxide; Strontium
titanate; Aluminum oxide; Magnesium oxide; Silica; Catalysis; Growth; Epitaxy
Corresponding author. Tel.: +49 711 6891470; fax: +49 711 6891472.
E-mail addresses: qfu@dicp.ac.cn (Q. Fu), t.wagner@fkf.mpg.de (T. Wagner).

1 Current address: Dalian Institute of Chemical Physics, The Chinese Academy of Sciences, Zhongshan Road 457, Dalian 116023, PR China. Tel.: +86 411
84379976; fax: +86 411 84694447.
c 2007 Elsevier B.V. All rights reserved.
0167-5729/$ - see front matter
doi:10.1016/j.surfrep.2007.07.001
432
Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498
Contents
1.
2.
3.
4.
5.
6.
Introduction............................................................................................................................................................................ 432
Fundamental aspects of metaloxide interaction ......................................................................................................................... 434
2.1. Physics and chemistry of heterophase interface contacts .................................................................................................... 434
2.1.1. Physics at metalsemiconductor interfaces .......................................................................................................... 434
2.1.2. Chemistry at metalsemiconductor interfaces....................................................................................................... 435
2.2. Electronic interaction of metals with oxides charge redistribution ................................................................................... 436
2.2.1. Local charge redistribution ................................................................................................................................ 436
2.2.2. Space charge transfer ........................................................................................................................................ 437
2.3. Chemical interaction of metals with oxides mass transport............................................................................................. 438
2.3.1. Chemical interactions at metal/oxide interfaces .................................................................................................... 438
2.3.2. Thermodynamic considerations .......................................................................................................................... 440
2.3.3. Kinetic consideration ........................................................................................................................................ 441
2.4. Interplay between electronic and chemical interactions ...................................................................................................... 442
2.4.1. CabreraMott theory in solidgas interface reactions ............................................................................................ 442
2.4.2. CabreraMott theory in solidsolid interface reactions .......................................................................................... 443
2.4.3. Generalized CabreraMott theory in interface reactions ........................................................................................ 445
Experimental methods ............................................................................................................................................................. 445
3.1. Preparation of model systems ......................................................................................................................................... 446
3.1.1. Oxide surfaces ................................................................................................................................................. 446
3.1.2. Metal overlayers on oxide surfaces ..................................................................................................................... 447
3.1.3. Inverse model systems....................................................................................................................................... 449
3.2. Characterization techniques............................................................................................................................................ 449
3.2.1. Electron-based spectroscopy .............................................................................................................................. 449
3.2.2. Scanning probe techniques ................................................................................................................................ 450
3.2.3. Transmission electron microscopy ...................................................................................................................... 451
3.2.4. Other techniques .............................................................................................................................................. 452
Interaction of metals with mixed conducting oxides .................................................................................................................... 453
4.1. Metals on TiO2 ............................................................................................................................................................. 453
4.1.1. TiO2 surfaces ................................................................................................................................................... 453
4.1.2. TiO2 bulk defect chemistry ................................................................................................................................ 455
4.1.3. MetalTiO2 interactions .................................................................................................................................... 457
4.2. Metals on SrTiO3 .......................................................................................................................................................... 463
4.2.1. SrTiO3 surfaces................................................................................................................................................ 463
4.2.2. SrTiO3 bulk defect chemistry............................................................................................................................. 465
4.2.3. MetalSrTiO3 interactions................................................................................................................................. 467
Interaction of metals with insulating oxides ................................................................................................................................ 471
5.1. Metals on Al2 O3 ........................................................................................................................................................... 472
5.1.1. -Al2 O3 surfaces ............................................................................................................................................. 472
5.1.2. Metal interactions with bulk Al2 O3 .................................................................................................................... 475
5.1.3. Metal interactions with alumina films.................................................................................................................. 476
5.2. Metals on MgO............................................................................................................................................................. 477
5.2.1. Metal interactions with bulk MgO ...................................................................................................................... 477
5.2.2. Metal interactions with MgO films ..................................................................................................................... 480
5.3. Metals on SiO2 ............................................................................................................................................................. 480
5.3.1. Metal interactions with bulk SiO2 ....................................................................................................................... 480
5.3.2. Metal interactions with silica films ..................................................................................................................... 482
Summary ............................................................................................................................................................................... 484
Acknowledgements ................................................................................................................................................................. 485
References ............................................................................................................................................................................. 485
1. Introduction
Metal/oxide interfaces play critical roles in many applications including materials science, microelectronics, and chemical applications. Metaloxide interactions, which consist of interfacial charge redistribution and/or mass transport upon interface formation, are important factors determining properties
and performances of the heterojunctions.
For materials science, metal/oxide interfaces can be found in

many technological materials, such as functional ceramics with
metals, oxide dispersion-strengthened alloys, oxide coatings
on metals functioning as thermal barriers or natural corrosion
protection layers, etc. Fundamental problems with these
systems include the adhesion strength, mechanical stability,
and fracture behavior of the interfaces, which are all closely
related to metaloxide interactions [1,2]. To further understand
433
Fig. 1. Schematic of a metal island on a flat oxide support in thermodynamic

equilibrium.
this, consider Fig. 1 which shows a schematic of a metal

island supported on a flat oxide surface. In the simplified
case of absence of surface energy anisotropies of the metal,
a mechanical equilibrium for this system can be written by
Youngs equation:
oxide = metal cos + interface .
(1)
Here, oxide is the surface free energy of the oxide substrate,

metal is the surface free energy of the metal overlayer,
interface is the interface free energy including any metaloxide
interaction, and is the contact angle between metal and
interface . Experimentally, the work of adhesion (Wad ) rather
than the interface energy can be measured. Wad is expressed
by the following formula:
Wad = metal + oxide interface .
(2)
Thus, the mechanical strength at the interfaces is closely

related to the metaloxide interactions. Furthermore, according
to (1), interface strongly influences the growth behavior of metal
overlayers, determining wetting or non-wetting of the metal,
layer growth or island growth of the metal, etc.
In the field of microelectronics, metaloxidesemiconductor
(MOS) field-effect transistor (FET) devices are core components. As shown in Fig. 2, two interfaces are very important
in the devices: the upper interface of gate electrode with dielectric oxide layer and the lower interface of Si with dielectric
oxide layer [3]. In the near future, the actual gate lengths of the
devices will be scaled down to 10 nm and the thicknesses of
dielectric oxide layers will be a few atomic layers [4]. The performance of the devices sensitively depends on properties of the
two interfaces, and they have to be stable in order to be successful. Chemical reactions and atom diffusion across interfaces
must be avoided to prevent any failure of the devices, and the
density of interface electronic states should be well-controlled
to obtain stable transport properties at the junctions [5].
Finally, chemical applications will be highlighted. Many
catalytic systems consist of nanosized metal catalysts supported
on oxides. It has been found that the interaction between metals
and oxide supports, so-called metalsupport interactions, are
of great importance in heterogeneous catalysis [613]. In
particular, the strong metalsupport interaction (SMSI) was
first suggested by Tauster et al. to explain the suppression
of both H2 and CO chemisorption capacity of metal clusters
Fig. 2. Schematic of important regions of a MOS field-effect transistor gate

stack. From [3].
supported on TiO2 which are reduced at high temperatures [14,

15]. Later, SMSI was widely observed in many metal/oxide
catalytic systems [813]. Two major factors contribute to
the SMSI states, an electronic factor and a geometric factor.
The electronic factor is determined by a perturbation of the
electronic structure of the metal catalyst, which originates
from charge transfer between the metal and the oxide, while
the geometric factor results from a thin layer of reduced
oxide support physically covering the metal particles (called
the encapsulation or decoration model), which blocks active
catalytic sites at the metal surface. Fig. 3 is used as an
example that illustrates the encapsulation of a Ru particle by
its amorphous titania support [16].
Presently, the sizes of many devices and technological
materials are rapidly decreasing to the nanoregime. As we
approach the nano limit, the density of interfaces increases
substantially such that the effect of metaloxide interactions
becomes more and more significant. In past decades, much
progress has been made thanks to modern surface science
techniques and advanced calculation methods. Model systems
were developed, which consist of nanometer scale metal
clusters or ultrathin metal films supported on well-defined oxide
surfaces under the conditions of ultrahigh vacuum (UHV).
In this manner, studies of charge redistribution and/or mass
transport at metal/oxide interfaces have been significantly
simplified and many of the modern surface science techniques
can be applied to study the interfacial atomic and electronic
structures. Moreover, many sophisticated calculation methods
allow for the derivation of the atomic structure and charge
density distribution at metal/oxide interfaces. This theoretical
work contributes to the understanding of the nature of
metal/oxide interfaces in many aspects, which helps spur
further experimental development to catch up with the advances
in the theory.
434
Fig. 3. (a) A Ru particle partially covered by amorphous titania; (b) A Ru particle completely covered by amorphous titania. From [16].
The knowledge of metal/oxide interfaces has been discussed

in some excellent books, conference proceedings, and
reviews [1,2,1729]. Many of the fundamental questions
about these systems have been addressed. In the present
paper, we concentrate on the understanding of metaloxide
interactions at low temperatures (below 1000 C), with special
emphasis on experimental and theoretical results obtained
in the past decades using the model systems consisting of
nanostructured metal overlayers on well-defined oxide surfaces.
Two aspects of the interaction are particularly examined: the
charge redistribution and the mass transport. It is demonstrated
how these two aspects are influenced by different factors.
In addition, generalized guidelines are given to predict the
interaction in the metal/oxide systems and the procedures to
tune them in a desirable way.
2. Fundamental aspects of metaloxide interaction
2.1. Physics and chemistry of heterophase interface contacts
Metalsemiconductor (MS) interfaces are the most wellstudied heterophase junctions. The knowledge of both the
physical and chemical aspects of the interface contributes
much to the understanding of other heterophase interfaces,
e.g., metal/oxide interfaces. A central issue of concern for the
MS interface is the electronic properties of the device, which
can be characterized by the Schottky barrier height (SBH). In
past decades, various theories have been developed to describe
the basic mechanism of the barrier formation [3032], which is
now discussed below.
2.1.1. Physics at metalsemiconductor interfaces
Schottky [33] and Mott [34] suggested that energy bands
of a metal and a semiconductor in contact adjust so as to
align both vacuum levels. They considered long-range charge
transfer, but ignored any local interaction at the interface. Due
to the readjustment of the vacuum levels, it was necessary for
the Fermi energy levels (E F ) to be bent to higher (or lower)
energy. This forms a barrier at the interface that originates from
a space charge layer at the semiconductor surface. The barrier
height, B,n , is simply the difference between the metal work
function M and the electron affinity of the semiconductor S :
B,n = M S .
(3)
Fig. 4. Energy band diagram of a metaln-type semiconductor interface at the

Schottky limit. VBB , band bending at semiconductor surface.
This is the Schottky limit for the band bending (Fig. 4). In the
Schottky approximation, the charge transfer process and the
form of space charge layers are governed by Poissons equation
d2 V
(r )
,
=
2
0
d r
(4)
where (r ) is the space charge density, is the static

dielectric constant of the semiconductor, and 0 is the
vacuum permittivity in free space. This has been extensively
documented and can be found in the classic Schottky contact
theory [35,36].
Bardeen [37] first considered that there may be interface
states located in the semiconductor band gap. In case
of sufficiently high density of the interface states, the
semiconductor is completely screened by the interface states
and the SBH is given by
B,n = E g o
(5)
where E g is the band gap energy of the semiconductor. o

is defined as the energy measured from the valence band
maximum (VBM) at the semiconductor surface to the level
below which all interface states must be filled for charge
neutrality at the surface. This phenomenon is known as Fermi
level pinning. It should be noted that Eq. (5) only considers the
local charge transfer between the metal and the interface states.
As shown in Fig. 5, Cowley and Sze [38] assumed a
continuum of interface states with a constant density of states,
Dis , and proposed a hybrid approach, where they combined the
435
a metal comes into contact with a semiconductor, is due to

the formation of polarized chemical bonds at the interface.
The transferred charge and the interface dipole can be treated
by the electrochemical potential equalization [47,48] method
employed in molecular physics. It has been shown that the
polarization of interface bonds leads to a weak dependence of
the SBH on the metal work function.
Fig. 5. Energy band diagram of a metaln-type semiconductor contact with an

interfacial layer after Cowley and Sze [38]. 0 : energy level measured from
the VBM at the semiconductor surface to the level below which all interface
states must be filled for charge neutrality at the surface, 1n = image force
barrier lowering; 1: potential across interfacial layer; it : dielectric constant
of interfacial layer; : thickness of interfacial layer; VBB : band bending at
semiconductor surface.
Schottky limit and the Bardeen limit in order to analyze charge

balance at the interface. They derived the following relation:

B,n = (m ) + (1 ) E g o
(6)
where , the interface parameter, is defined by
1

= 1 + e2 Dis /it
(7)
with defined as the thickness of the interfacial layer, and it is

the permittivity of the interface layer. can be compared to the
experimentally determined slope parameter S = B,n /m .
For a high density of interface states (Dis , 0), the
barrier height becomes independent of the metal work function,
i.e. the Bardeen limit; if Dis = 0 and = 1, it results in
the Schottky limit. Covalent solids, e.g. Si, present smaller ,
indicating the dominant role of the local charge transfer and
the strong Fermi level pinning; while ionic semiconductors
with wide band gap, e.g., SrTiO3 and SiO2 , have larger ,
which suggests that interface states do not affect the interface
formation significantly [3941].
Concerning the physical origin of the interface states, an
important theoretical approach is based on the concept of metalinduced gap states (MIGS) [4244]. Heine [42] suggested
that at a MS interface, the metal electronic wave functions
tail into the semiconductor in the energy range in which the
conduction band (CB) of the metal overlaps the band gap of the
semiconductor. These tails give rise to a continuum of MIGS in
the semiconductor. The charge neutrality level (CNL), CNL , of
the MIGS is defined such that the tail of the metal wave function
penetrating into the semiconductor will carry no charge when it
coincides with E F . The charge transfer is thus governed by the
position of E F relative to CNL .
Tung [32,45,46] has used another way to account for
the experimentally observed strength of Fermi level pinning
on different semiconductors. His bond polarization theory
assumes that the most significant charge rearrangement, when
2.1.2. Chemistry at metalsemiconductor interfaces

Besides the physical explanations, there is much experimental evidence showing the importance of chemical mechanisms
in the charge transfer and Schottky barrier formation at MS interfaces.
Andrews and Phillips [49] attempted to correlate the SBHs
with the heats of formation of transition metal silicides, 1H f
(tmSi). They showed a linear relation between the experimental
B,n and 1H f (tmSi) indicating the strong effect of interface
chemical bonding on the barrier formation. At transition
metalSi interfaces, Ottaviani et al. [50] also identified the
strong relationship between SBHs and interface reactions.
The reaction between the two contacting solids produces an
interfacial layer, which dominates the SBH. The relationship
is manifested through the correlation between the SBHs
and eutectic temperatures for the transition metalsilicideSi
systems. Brillson [51] revealed a systematic dependence of
SBHs on interface reactivity. The SBH of metals on individual
compound semiconductors, such as ZnO, ZnS, CdS, and GaP,
demonstrates a sharp transition as a function of heat of
reaction 1H R (from the most stable known metalanion bulk
compound). The local charge redistribution, which is made
evident by new interface compounds for reactive metals and by
discrete interface states for non-reactive metals, determines the
barrier formation [52].
On the other hand, the chemical trends in SBHs were
made to manifest with the concept of electronegativity.
Originally, Kurtin [39] used the electronegativity difference,
1X , between the constituents of the compound semiconductor
e.g. 1X GaAs = X As X Ga to describe the semiconductor
ionicity, and showed a correlation between the slope parameter,
S, and 1X . Monch [31,53] applied Paulings electronegativity
model to explain the charge transfer across the MS interface.
It is proposed that the transferred charge varies proportionally
to X m X S , which is simply the difference between the
electronegativities of the metal, X m , and the semiconductor,
X S . In case of X m X S = 0, the metal E F will coincide with
the CNL of MIGS and no charge transfer occurs.
Thus, the Schottky barrier formation should be attributed
to a two-fold charge transfer upon the interface formation: the
local charge transfer and the long-range charge transfer [54].
The long-range charge transfer between the metal band and
semiconductor band occurs over the semiconductor space
charge region with a scale of screening length L [35]. The
process can be described by the Schottky contact theory.
The local charge rearrangement happens at the interface with
The local interaction may be characterized
dimensions of A.
in the band picture using the MIGS concept [4244,55] or in
436
the bond picture which considers interface chemical bonding

between the metal and semiconductor atoms [45,54,56].
It should be mentioned that there is a strong interplay
between the physical and chemical aspects of the interface. On
the one hand, charge transfer, formation of interface dipole,
and SBHs are strongly correlated with the chemical interaction
(e.g., chemical reaction and atom diffusion) at the interfaces.
On the other hand, charge transfer produces strong electric
fields at the interface, which further affects the interface
reaction or diffusion dynamics [30].
2.2. Electronic interaction of metals with oxides charge
redistribution
Interfacial contact between a metal and an oxide can result
in charge redistribution at the junction similar to the situation at
the MS interfaces discussed above. The electronic interactions
are simply driven by the principles of the system energy
minimization and the continuity of the electric potential in a
solid [32].
Electron transfers always occur at reactive interfaces, where
already existing chemical bonds are broken and new ones are
formed resulting in interdiffusion over length scales of more
than a single monolayer and formation of new phases with a
thickness of at least one monolayer [29]. These systems will
be considered in Section 2.3. Here, the discussion of electronic
interactions is limited only to non-reactive interfaces, where
the localized charge redistribution occurs among atoms at the
interface and/or the delocalized charge transfers between the
metal and the space charge region of the oxide. Many theories
and methodologies have been developed to reveal the various
contributions to the electronic interactions.
2.2.1. Local charge redistribution
The local charge redistribution is the electron rearrangement
involving a few atomic layers at the interface. Depending on the
metal/oxide system, different mechanisms may be dominant in
the electronic interaction [22,57]. The following sections detail
the relevant mechanisms.
2.2.1.1. Empirical correlations from dispersion force. At a
metalinsulator interface, the mutual polarization of the
two media gives rise to dispersion forces. The dispersion
contribution, i.e. the van der Waals interaction, is quite
weak compared to other interactions, such as the electrostatic
interaction [22,58]. Therefore, it is believed that this interaction
is important only in the systems consisting of a noble metal
deposited on a wide band gap oxide [59]. Under this condition,
the van der Waals contribution has been shown to qualitatively
account for some of the interaction energy experimental results.
From this, an empirical correlation has been derived showing
that the adhesion energies at metal/oxide interfaces increase as
the plasmon energy of the metal increases and/or as the band
gap of the oxide narrows [59,60]. Stoneham [61] found another
empirical correlation where the optic dielectric constant, , of
the oxide provides a simple classifying rule. An higher than
a critical value leads to wetting; indeed oxides with sufficiently
high , such as TiO2 and Nb2 O5 , exhibit SMSI effects in
catalysis.
Fig. 6. Image charge interaction. (a) Interaction between a point charge (q) and
a metal surface. (b) Image charges at a metal/oxide interface. From [2].
2.2.1.2. Image charge theory. Stoneham and Tasker [62]

first considered that many phenomena associated with
metal/nonmetal interfaces with a large dielectric constant
mismatch can be understood in terms of image interactions due
to charges in the nonmetal. Based on classical electrodynamics,
they suggested an image charge theory which describes the
interaction between a metal and an ionic crystal by electrostatic
forces [62,63].
Positioning a charge, q, at distance, z 0 , from a metal surface
will cause rearrangement of the metal charge and induce a
quasi-free image charge on the metal surface. Such a charge
distribution results in a potential of the form
q
V (r, z) =
1/2
2
r + (z 0 + z)2
(8)
where r is the distance parallel to the surface. At a metalionic

oxide interface, anions and cations in the oxide induce image
charges in the metal. The image theory assumes that the
metaloxide interaction energy originates from the attractive
Coulomb force between ions in the oxide and their images in
the metal (Fig. 6). The interaction may give adhesion energy of
the order of joules per square meter [22].
It is known that the valence band (VB) of metal atoms
supported on an ionic oxide can be distorted by the electrostatic
field of the oxide. This phenomenon is called the polarization
effect, and results in electron redistribution among the different
components of the metal band [6467]. As an example, upon
adsorption of Pd or Ru on the surface oxygen of sapphire,
electrons are depleted from the d2z orbital and move to the
lateral d orbitals. The polarization of these atoms makes the
adsorbate positive above the O ions and negative between the
O ions [66]. In the case of ultrathin metal overlayers, e.g. one
monolayer (ML) of adsorbate, the binding of metals to oxides
is dominated by the polarization effect, which mimics the
macroscopic image charge effect.
2.2.1.3. Metal-induced gap states. Noguera and Bordier [57]
tried to understand the nature of bonding between a metal and
an oxide using the concept of MIGS, which was introduced
by Heine to model MS junctions [42]. They treat the metal
as a jellium, whose eigenstates are plane waves, and the
oxide as a tight-binding system with local orbitals. MIGS are
produced by matching the delocalized metal wave functions
with exponentially decaying oxide states in the band gap energy
range at a defectless interface. The effect of the metal jellium
seems to allow some states originally in the valence and

conduction bands of an oxide to be pulled into the band gap
of the oxide. The MIGS are associated with a smooth density
of states in the band gap and their density decays from the
[68]. This
interface into the oxide over a length of several A
length correlates with that of the ionicity of the oxide where
larger band gap energy leads to smaller length.
The gap states are derived from the bulk bands of the oxide.
Thus, they may have donor-like or acceptor-like character
depending on the position of the states in the band gap. Like
the CNL of MIGS at MS interfaces, Noguera and Bordier state
that charge neutrality occurs if the MIGS are filled up to E ZCP ,
the zero-charge point. The significance of E ZCP is that it can be
used to predict if charge transfer will occur when a metal is in
contact with an oxide. If, for example, E F of a metal is above
E ZCP , a transfer of electrons from the metal to the oxide would
take place, creating an interface dipole.
2.2.1.4. Interface bonding. Formation of chemical bonds is
an important mechanism for charge transfer at metal/oxide
interfaces. It is known that either covalent bond or ionic bond
can be established between metal atoms and surface oxygen
ions or cations of oxides.
Polar covalent bonds, in which the metal and oxide
electronic orbitals are strongly hybridized, are partly ionic.
As proposed by Pauling [69], the ionic character of a polar
covalent bond between atoms A and B is determined by their
electronegativities, X A and X B . The bonding charge shifted to
the more electronegative atom is expressed by the relation [70]
1q/e = 0.16 |X A X B | + 0.035 |X A X B |2 .
(9)
In case of a compound solid, the electronegativity is given by

the geometric mean of the atomic values of the constituents,
e.g. X AB = (X A X B )1/2 [31]. Using the simple concept
of Paulings electronegativity, one can predict the direction
and amount of charge transfer at metal/oxide interfaces.
An alternative way to describe the charge transfer through
metaloxide bonding can be adopted from the bond polarization
theory of MS interfaces suggested by Tung [45]. The
conglomerate of the entire metal/oxide interface is regarded
as a giant molecule and the charge redistribution in the
molecule can be treated using methods applied in molecular
physics. The transferred charge per interface bond is estimated
to be

M M Ox E g /2

(10)
q =

Eg +
where M is the metal work function, M Ox the electron affinity
of the oxide, E g is the band gap energy of the oxide, and the
parameter related to interface characters [45].
An ionic bond is formed via electron donation from one
atom to another without orbital mixing. At a metal/oxide
interface, the ionic bonding is accompanied by charge transfer
between interfacial atoms, leading to a redox reaction.
Experimentally, ionic bonds are often observed during the
initial stage of deposition of reactive metals on oxides. An
437
example of this is the oxidation of metal adatoms and reduction

of topmost cations which take place via metal-to-oxygen and
oxygen-to-cation charge transfer [71,72]. The bonding and the
resultant reaction concern atoms at the interface, and, therefore,
may not necessitate atomic diffusion over more than one lattice
distance.
It is worth mentioning that formation of interface bonding
strongly depends on oxide surface properties including, but not
limited to, surface termination, surface reconstruction, surface
impurities, and surface point defects. A good example can
be found for the Ag/MgO interface where chemical bond
formation is demonstrated to be not important [73,74]. On
a defective surface, however, a Ag atom sitting on a neutral
Mg vacancy, Vs0 , donates two valence electrons to the four O
atoms surrounding the vacancy which forms strong ionic bonds
between the Ag and O atoms [74]. As we will show below,
defects on oxide surfaces, such as Al2 O3 (0001), MgO(100),
SiO2 , and TiO2 (110), greatly affect metal adsorption and the
atomic interaction energy.
2.2.1.5. Theoretical calculations. Many sophisticated calculation methods allow for the derivation of the charge density distribution at metal/oxide interfaces. Interface electronic effects,
such as polarization, bonding, charge transfer, etc., can be explicitly derived from these computational results.
To model metal/oxide interfaces, two main approaches are
used in electronic structure calculations, a cluster model and a
slab modeling method [7578]. In the cluster model the solid
is replaced by a finite cluster of atoms and is based on the
assumption that all interactions are locally well-described. It is
important to carefully choose the cluster such that all important
aspects of the electronic structure of the infinite system can
be reproduced. The slab model is to take advantage of the
translational symmetry of the system and model the interface
by slabs. In this model, a slab consisting of a finite number of
layers mimics the semi-infinite system, with a 2D translational
periodicity. In this model, a band structure calculation is
usually performed in the reciprocal space using delocalized
plane waves (PW) for expanding the wave functions. The
cluster approaches often use the real-space molecular orbitals,
e.g. the linear combination of atomic orbitals (LCAO) of the
surface atoms. These calculations can be performed using semiempirical methods, density functional theory (DFT), or from ab
initio principles. The advantage of these quantum mechanical
calculations is that they contribute to the understanding of
the nature of metal/oxide interfaces in many systems where
technology has not advanced to the point where the necessary
experiments can be done.
2.2.2. Space charge transfer
Oxides, such as TiO2 , SrTiO3 , ZnO, etc., have relatively
small band gap energies, (usually <3.5 eV) [17]. Intrinsic or
extrinsic dopants may produce free electronic carriers inside the
solids causing them to become equivalent to n-type or p-type
semiconductors. Thus, for these systems, MS contact theory
is applicable to describe interface contact phenomena between
metals and the semiconducting oxides.
438
Recall from 2.1.1 that ionic semiconductors, e.g., SrTiO3 ,

ZnO, and SiO2 , present a larger and S [3941], which
suggests that interface states on their surfaces do not control
the interface formation. It is therefore reasonable to assume
that charge transfer happens primarily between metals and the
space charge regions at oxide surfaces. The relative position
of E F of the metal and the oxide determines the direction and
magnitude of the charge transfer. This process can be described
using the Schottky contact theory (see Fig. 4). It is now useful
to discuss two cases describing supported metal overlayers on
oxides, those with continuous metal overlayers on oxides and
those with metal clusters or discontinuous overlayers on oxides.
(1) For an infinite interface, (continuous metal overlayers
on oxides) under the condition of strong depletion of an
n-type semiconducting oxide, the space charge density is
approximated by a step function with
(z) = eN D
0 z d,
(11)
where N D is the donor concentration in the bulk semiconductor,

and d is the thickness of the depletion layer. Poissons equation
(4) becomes
d2 V
eN D
=
,
2
0
d z
(12)
with the relevant boundary conditions, V (z = 0) = VBB =

(m sc )/e the band bending at the surface, and V (z = d) =
0. The thickness of the space charge layer is calculated to be

d=
20 |VBB |
eN D
1/2
(13)
The transferred charge per unit area from the semiconductor to

the metal, Q, is
Q = (20 eN D |VBB |)1/2 .
(14)
(2) The finite interface (metal clusters or discontinuous

overlayers on oxides) model is used to describe the situation
of small clusters on semiconducting oxides. Verykios and
coworker [79] suggested a physical model consisting of
spherical metal particles of radius, r M , embedded in the
semiconductor bulk (Fig. 7). Within approximation (11),
Poissons equation in 3D form can be analytically solved. The
thickness of the depletion region, d, and band bending, VBB
have a relation of
!
2
eNd d 2 r M
d3
VBB =
.
(15)
0
2
6
3r M
The number of electrons transferred at the interface is now
given by
Q=

4 Nd 3
3
d rM
.
3
(16)
As seen in the above equations, both d and Q are functions of

the cluster size, r M . For example, on doped TiO2 , the amount of
charge transferred to the metal is dependent on the metal cluster
size. They report 0.5 electrons per metal atom for a cluster size
Fig. 7. Physical model used to simulate the contact of a metal cluster with a
semiconducting oxide support. From [79].
of 1.5 nm and only 0.01 for crystallites larger than 10 nm [79].

Eq. (16) well illustrates the effect of size on charging the metal
clusters.
The long-range charge transfer between a metal and an oxide
produces space charges and band bending at the oxide surface.
Experimentally, band bending with a few tenths eV has been
observed when various metals are deposited on ZnO [80,81],
TiO2 [8284], and SrTiO3 [85,86]. In these systems, there exist
interfacial dipole fields due to the space charges in the devices.
Typical space layers extend for 10100 nm, and lead to an
energy difference of 0.2 eV and fields of 106 107 V/m [87].
The built-in electric fields clearly have a strong effect on
reaction kinetics as we will discuss below.
2.3. Chemical interaction of metals with oxides mass
transport
In this case chemical interaction between metals and oxides
is regarded as an interaction which results in chemical reactions
at the interfaces. The term chemical reaction is used for
those cases where interdiffusion occurs over length scales of
more than a single monolayer or where new phases with a
thickness of at least one monolayer are formed [29]. The
chemical interaction involves mass transport over more than
one lattice distance. The driving force behind the mass transport
in the solids is a gradient in the electrochemical potential,
consisting of a gradient in the electrical potential and/or
chemical potential.
2.3.1. Chemical interactions at metal/oxide interfaces
Due to the nature of the products formed at metal/oxide
interfaces, the chemical interaction can be generally classified
into four different groups, redox reaction, alloy formation,
encapsulation, and interdiffusion [20,23]. Each case is now
discussed and the mass transport processes during every
reaction are to be demonstrated.
Case 1: Redox reaction. A redox reaction at a metal/oxide
interface (MeI k MeII Ox ) will occur by oxidizing the metal
overlayer and reducing the oxide substrate. The reaction can be
schematically written as:
II
MeI k MeII Ox MeIO
y k Me Oxy .
Of all the reactions occurring at metal/oxide interfaces,

these are the most frequently observed, especially for reactive
metals on oxides such as TiO2 and SrTiO3 [18,29,88,89].
For example, thin epitaxial Nb films on TiO2 (110) have
439
Fig. 8. (a) HRTEM image of Nb/TiO2 (110) interface formed at room temperature. A distorted interface between Nb and TiO2 crystal is present [90]. (b) HRTEM
image obtained on the 5% Pt/CeTbOx catalyst reduced at 1173 K, resulting in the formation of CePt5 alloy [93]. (c) HRTEM image of a 0.5% Rh/Ce0.8 Tb0.2 O2x
catalyst reduced at 1173 K showing the encapsulation of Rh particle by support [94]. (d) HRTEM image of Cr/TiO2 (110) interface formed at 400 C. The interface
layer of Tix Cry Oz is from the interdiffusion between Cr overlayers and TiO2 substrate.
been prepared by molecular beam epitaxy (MBE) at room

temperature. Investigation of the interfaces of these systems
by high resolution transmission electron microscopy (HRTEM)
and electron energy loss spectroscopy (EELS) confirmed that
the first two monolayers of Nb were oxidized, resulting in
the partial reduction of a 2 nm thick TiO2 layer near the
interface [90] (Fig. 8(a)). Mass transport in the form of oxygen
(i.e. oxygen vacancy) diffusion, commonly occurs during the
reaction process (e.g., [91,92]).
Case 2: Alloy formation. At some interfaces, stable
intermetallic compounds may be formed according to
MeI k MeII Ox MeI MeIIy k MeII Oz ,
or
MeI k MeII Ox MeI O y k MeI MeIIz k MeII Ox .
The former case, where MeII is reduced during alloy formation,

occurs for noble metals supported on oxides of CeO2 [9597],
SiO2 [98100], Al2 O3 [96], and SnO2 [101]. Pt nanoparticles
supported by thin films of silica, alumina, and ceria were
subjected to hydrogen reduction at temperatures up to 1073 K.
Formation of Pt-rich Pt3 Me (Me = Si, Al, Ce) alloy phases
were identified in all three systems after the reactions [96].
Similar reactions of the alloy formation were observed for other
noble metals, such as Pd, Rh, Ni, and Cu, on these oxides.
Fig. 8(b) gives one example of formation of an intermetallic
compound at a Pt/CeTbOx interface.
In cases of reactive metals deposited onto Al2 O3 and SiO2
substrates, the second case may happen. The reaction leads to
formation of a thin layer of aluminide or silicide sandwiched
between the substrate and a top layer of metal oxide [102,103].
440
Table 1
Summary of interface reactions at metal/oxide interfaces
Chemical interaction at MeI k MeII Ox
Interfacial products
Main mass transport process
Typical systems
Redox reaction
MeI O y k MeII Oxy
Oxygen
RMa on TiO2 & SrTiO3
Alloy formation
II
MeI MeII
y k Me Oz
I
I
II
Me O y k Me MeII
z k Me Ox
Cations
NMb on CeO2 & SiO2

RMa on Al2 O3 & SiO2
Encapsulation
MeII Ox k MeI k MeII Ox
Cations
NMb on TiO2 & CeO2
Interdiffusion
MeI MeII O y
Metal atoms and/

or support atoms
Ni/Al2 O3
Al/spinel
a RM: reactive metals.

b NM: noble metals.
In alloy formation processes, it is mostly expected that

cations are physically extracted from oxide substrates and
incorporated in metal overlayers.
Case 3: Encapsulation. Also called decoration, the
encapsulation reaction involves mass transport from the oxide
support onto the surface of metal particles. It necessitates
the cations outward diffusion from the bulk to the interface
(e.g.[13]). Encapsulation is a special interface process whereby
metal particles are physically covered by a thin layer of
reduced oxide support (Fig. 8(c)) according to the following
relationship:
Me k Me Ox Me Ox k Me k Me Ox .
I
II
II
II
This reaction results in blocking active catalytic sites on metal

surfaces and contributes to the SMSI state [8,11,13,14]. Noble
metals, including Pt, Pd, and Rh on TiO2 are good examples of
encapsulation reactions. Ultrathin layers of TiO2x (0 < x <
1), which have a thickness of several atomic layers, have been
confirmed to emigrate onto surfaces of metal particles (see [13,
94] and references therein). Similar results can be found for
noble metals on CeO2 [95].
Case 4: Interdiffusion. It is known that metals may diffuse
into their oxide supports and/or substrate atoms may diffuse to
the metal surface. Such interdiffusion leads to the formation of
interdiffusion zones or mixed oxides (e.g. ternary oxides and
oxide solid solutions) at the interfaces (Fig. 8(d)). They can be
expressed as
MeI k MeII Ox MeI MeII O y .
One example is that NiAl2 O4 has been confirmed to form
at the Ni/Al2 O3 interface [104]. Raj et al. have also shown
that an ion exchange reaction between Al and spinel ceramic
of (MgO)(1.25Al2 O3 ) produces interdiffusion zones at the
interface [105,106].
The four reactions at metal/oxide interfaces are summarized
in Table 1 and show the typical reaction systems and mass
transport processes. We now turn to a discussion of the
thermodynamic and kinetic aspects of these interface reactions
in the following sections.
2.3.2. Thermodynamic considerations
Thermodynamics can be used to find out if a chemical
reaction is favorable. The classic thermodynamics may consider
the bulk thermodynamic data. For interface reactions, the

influences of interface and surface should be taken into
consideration. These two cases are being discussed here.
(1) Bulk thermodynamics: As a first approximation, simple
bulk thermodynamic calculations for solid state reactions are
often used to predict interface reactions in metal/oxide systems.
For example, a redox reaction at a metal/oxide interface may be
expressed as follows:
MeI + MeII Ox MeI O y + MeII Oxy .
The Gibbs free energy change for this reaction, 1G R = 1H R
T 1S R , will suggest the feasibility of the reaction. For solid
state reactions, changes in entropy are negligible such that
the enthalpy changes for the reactions can be simply used for
thermodynamic criteria. As discussed by Campbell [23], the
heat of formation of oxides (1H of ) is a decisive thermodynamic
parameter for description of the interface reactions. The heats
of oxide formation per mole of oxygen (1H of in kJ/mol O)
were compiled in his review paper. He showed that 1H of can be
successfully applied to explain the experimental results in many
metal/oxide systems. The trend in metal overlayers reactivity
to a specific oxide, e.g. TiO2 , can be explained very well by the
variation in 1H of of the metals [28].
An alternative way to describe reactivity of metals to
oxides is to introduce the concept of oxygen affinity of metals
( pO) [107]. pO is defined as
pO = log pO2 = 1G f /RT,
(17)
at 1000 K for the equilibrium between the metal and its lowest
oxide:
1G f is given for the reaction of a metal with one mole of
O2 :
2
2
Me + O2 MeOn .
n
n
pO is often used to compare the reactivity of metals to
oxides [20,108].
To obtain a rough estimate if a reaction occurs, the Gibbs
free-energy-changes for reactions, such as alloy formation, can
be calculated using the corresponding reaction formula.
(2) Thermodynamics including interface terms: In the above
thermodynamic calculations, energies associated with surfaces
and interfaces were not included. In fact, the contribution from
441
case of a redox reaction at a metal/oxide interface, where an

oxidized metal (MeI Oy ) and reduced oxide (MeII Oxy ) that
possess different stoichiometry, may form complex phases at
the interfaces. Accurate phase determination is complicated,
since the resulting phases are often approximately a few
atomic layers or unit cells thick. Thus, calculations can only
be performed assuming specific reaction phases. Moreover,
equilibrium conditions are assumed in these calculations, which
may be very wrong since solid state reactions are often
kinetically limited, especially at relatively low temperatures
(e.g. <1000 C). Therefore, nonequilibrium thermodynamics
and reaction kinetics must be considered as discussed below.
Fig. 9. Schematics showing the mass migration of TiOx (x < 2) onto metal
clusters driven by the minimization of surface energy of the whole system [13].
the surface and interface energies to the total free energy could
be significant, particularly for nanostructured metal overlayers
on oxides where the ratio of interface to bulk is very large. At a
metal/oxide interface, interface energies include metal surface
energy metal , oxide surface energy oxide , and interface energy
interface (recall Fig. 1). These factors may play a critical role in
some reaction processes, such as oxidation and encapsulation
reactions.
Oxidation reaction: Metals generally have a larger surface
energy than an oxide such that 3D metal islands are favored
to grow on oxide surfaces due to oxide < metal + interface .
If the metal islands are transformed into oxides, the decrease
in surface energy may reverse the above inequality and drive
the conversion of 3D islands to 2D structures. This can be
seen in the Cr/SrTiO3 (100) system, where we have observed
the flattening of Cr oxide islands after Cr oxidation above
600 C [91]. Also, STM investigations show that V clusters
supported on a thin alumina film become flatter in case of
oxidation at 800 K in UHV [109]. Finally, room temperature
adsorption of oxygen to Cu islands supported on TiO2 (110)
allows 2D Cu islands to form on the surface at the expense of
the existing 3D islands [110].
Encapsulation: It has been proposed that minimization of
the surface energy of a system is one of the main driving forces
for an encapsulation reaction [13,111,112]. Only in systems
where metals have high surface energies and oxides have low
surface energies, can supported metal clusters be encapsulated
by oxide support layers (Fig. 9). The surface energetic factor
explains why Pt and Pd but not Au and Ag (both metals have
lower surface energies) have been subjected to the reactions.
Similarly, oxides with low surface energy, e.g. TiO2 and V2 O5 ,
undergo encapsulation reactions more easily than oxides with
relatively high surface energies, such as SiO2 and Al2 O3 .
Restrictions of thermodynamic calculations are now being
addressed. Thermodynamic data of interfacial reactants and
products are often unavailable so, as an approximation,
bulk thermodynamic data are typically used to calculate the
dynamics. These bulk data are frequently different from that
of the interface phase. Secondly, the chemical composition of
the interface phases is often unknown. This can be seen in the
2.3.3. Kinetic consideration

Interface reaction kinetics can be controlled by the reaction
itself, the mass transport process, or both. If the interface
reaction controls the kinetics, the growth rate of newly
formed layer is linearly dependent on the reaction time.
Whereas parabolic reaction kinetics, where the reaction rate is
proportional to the square root of time, is observed in case of
diffusion-controlled reactions (e.g. [113]). At many metal/oxide
systems, as we will see below, interface reactions at relatively
low temperatures are often limited by mass transport, e.g. the
transport processes shown in Table 1.
Using the hopping model for ionic diffusion in solids,
Fromhold and Cook [114] derived an equation for the ionic
diffusion current J in the steady-state approximation
J = 4a exp(W/k B T ) sinh(Z eE 0 a/k B T )

C(L) C(0) exp(Z eE 0 L/k B T )
.
1 exp(Z eE 0 L/k B T )
(18)
W is the activation energy; is the ionic attempt frequency;

2a is the ionic jump distance; E 0 is the electric field applied
externally or from space charges in solids; Z e is the effective
charge per particle of the ionic species undergoing transport
through the lattice; C(L) and C(0) are the bulk defect
concentrations of the diffusing ionic species at x = L and
x = 0. Eq. (18) is the nonlinear diffusion equation in the
homogeneous field limit, which can be simplified by various
approximations.
Approximation (1). In case of E 0 0, which would occur
if the effect of electric field is negligible or it is diffusion of
neutral particles in a solid, Eq. (18) changes to
J = 4a 2 exp(W/k B T ) [C(L) C(0)] /L ,
(19)
which can be written as

J = D [C(L) C(0)] /L
(20)
with a diffusion coefficient D = 4a 2 exp(W/k B T ). This is

the form of Ficks first law showing that the particle current is
linearly related to the spatial concentration gradient C/ x.
Approximation (2). Eq. (18) under a low electric field and for
the limiting case of a high temperature, i.e., E 0 |k B T /Z ea|,
results in

C(L) C(0) exp(Z eE 0 L/k B T )
.
(21)
J = E 0
1 exp(Z eE 0 L/k B T )
442
Fig. 10. Schematic diagrams showing the potential energy of an interstitial

metal ion as a function of position near the metal/oxide interface. The electric
field at oxide surface, which is generated by the interfacial charge transfer,
lowers the energy barrier for ions moving away from the metal/oxide interface.
After [116,117].
Here, = (Z e/k B T )4a 2 exp(W/k B T ) is the mobility

of the diffusing particle, which is related to the diffusion
coefficient, D, through the well-known Einstein relation
k B T = Z eD. The above equation is the integral form of
the ordinary linear diffusion equation in the homogeneous field
limit
C(x)
+ E 0 C(x),
(22)
x
where the ionic diffusion current is linearly dependent on the
electric field.
Approximation (3). In the case where the driving force
due to the concentration gradient is insignificant compared to
that from the electric field, we have C(L)/C(0) 1 and
exp(Z eE 0 L/k B T ) 1. This is the high electric field and low
temperature limiting case. Eq. (18) becomes
J = D
J = 4aC exp(W/k B T ) sinh(Z eE 0 a/k B T ),
(23)
or
J = 2n exp(W/k B T ) sinh(Z eaVM /k B T L),
(24)
where n is the number of ions per unit area which are in a

position to jump the barrier W , n = 2aC. This is the equation
given by Mott, which is based on the assumption that nonlinear
diffusion under the aiding potential VM (Mott potential) is ratelimiting [115,116]. It can be seen from Eq. (23) that an electric
field E 0 modifies the activation energy for the ionic motion
from W to W Z eE 0 a as shown by Fig. 10. The additional
term describes either an extra driving or retarding force for the
transport of ionic defects. The ionic diffusion current shows
an exponential dependence on the electric field such that the
macroscopic electric field contributes largely to ionic diffusion.
At relatively low temperatures, k B T W , ordinary diffusion
given by (20) is negligible and the diffusion of ionic defects is
often thermally limited. Therefore, electric fields from surface
or space charge layers may play an important role in the defect
diffusion and the reaction kinetics. Such a field-driven effect
is quite general and critical in many interface reactions. We
discuss this in more detail in the next section.
2.4. Interplay between electronic and chemical interactions
At low temperatures, e.g. room temperature, defect diffusion
in solids is often thermally limited, whereas a large electric field
could lower the energy barrier thereby influencing the ionic

diffusion and chemical reaction. As discussed in Section 2.2,
the electronic interaction, in particular, the long-range charge
transfer at metal/oxide interfaces, produces space charges and
consequently, induces an electric field. It is therefore expected
that there is interplay between the electronic interaction and the
chemical interaction in some cases. The coupling between the
defect diffusion and charge transfer enables interface reactions
to be dependent on the electronic structure of the interfaces.
The process is explained in the framework of a Generalized
CabreraMott theory [13,85,89], which is now discussed in
detail.
2.4.1. CabreraMott theory in solidgas interface reactions
The CabreraMott theory was first suggested to explain the
low temperature oxidation of metal surfaces. Later, it was found
that the etching of semiconductor surfaces can be understood
using the same theory. Here, we give a brief overview of the two
different surface reactions. Special attention is given to charge
transfer and mass transport processes at the surfaces.
2.4.1.1. Low temperature oxidation of metal surfaces. Many
theories have been developed to describe the oxidation of metal
surfaces ([117,118] and references therein). Cabrera and Mott
suggested a theory of formation of thin films by metal oxidation
at low temperatures [115,116]. In this so-called CabreraMott
theory, they considered two important points, (1) electron
transfer occurs between surface adsorbed oxygen and metal and
(2) the resulting electric field in the oxide layer can lower the
barrier for ionic mobility. Both of these cases are discussed in
some detail.
To begin the discussion of the CabreraMott theory, we
consider the fact that a layer of atomic oxygen is adsorbed
at the gassolid interface during the oxidation reaction. The
central assumption of Cabrera and Mott is that electrons can
pass through the oxide layer from the metal to the oxygen atom.
This charge transfer is facilitated by thermionic excitation from
the metal into the conduction levels of the oxide or by electron
tunneling through the oxide barrier [119,120]. This electronic
motion is considered to be rapid compared to the ionic motion
such that an electric field across the oxide layer forms and
a quasi-equilibrium state is set up between the metal and the
adsorbed oxygen. A contact potential VM , known as the Mott
potential, forms and is defined by the initial difference in E F
of the metal and O2p level. The Mott potential VM is defined as
(see Fig. 11)
VM = (0 L )/e,
(25)
where 0 is the work function of the metal, and L is the energy

difference between the vacuum energy level and O2p energy
level.
As stated above, an electric field, E 0 = VM /L(t), may
occur both in the oxide layer with thickness of L(t) and at
the metal/oxide interface. It may lower the energy barriers for
the initiation of ionic motion and enable the metal cations or
oxygen anions to move through the oxide layer without much
help from temperature. For very thin films, the field is large
443
Fig. 11. Electronic levels in the metal, oxide, and adsorbed oxygen (after Cabrera and Mott [116]): (a) before electrons have passed through the oxide, (b) when
equilibrium is reached. The thickness of grown oxide layer is L(t). VM is the Mott potential producing an electric field E 0 at the metal surface. 0 is the work
function of the metal, and L is the energy difference between the vacuum energy level and O2p energy level. 0 is the energy difference between the conduction
band minimum (CBM) in the oxide and E F of the metal and L the energy difference between the CBM in the oxide and the O2p energy level.
such that ion motion may show an exponential dependence on

the electric field. The ion diffusion current can be expressed
using the formula (24).
The CabreraMott theory predicts a very rapid initial growth
rate followed by a sharp leveling off to a very slow growth stage
This reaction mechanism has been
at thickness of 50150 A.
found to be active in many oxidation experiments [116118,
121124].
2.4.1.2. Etching of semiconductor surfaces. Reactions of
semiconductor surfaces with reactive gases, which produce
volatile products, are used for surface etching. One important
and well-studied example is the etching of Si via exposure of
a Si surface to atomic F generated in fluorocarbon plasma or
from dissociation of XeF2 molecules. The interaction of the
Si surface with F atoms leads to growth of a fluorosilyl SiFx
(x < 4) layer on the surface. The layer is composed of SiF,
SiF2 , and mostly SiF3 . In a quasi-steady etching condition,
the thickness of the layer (l) is nearly constant with values
Etching occurs from the surface of this
between 10 and 30 A.
layer, where volatile SiF4 is formed. The structure of the Si
surface is schematically shown in Fig. 12. The etching rate has
been found to be strongly influenced by Si-doping [125132].
The following facts were well-established in those papers: (1)
heavily doped n-Si (n+ -Si) etches much faster than undoped Si
and lightly n- or p-doped Si; (2) lightly doped p-type and n-type
Si have similar etching rates; (3) heavily doped p-Si (p+ -Si)
exhibits the lowest etching rate.
Winters et al. [125,126] suggested that the Si etching
reactions can be described on the basis of the CabreraMott
theory (Fig. 12):
(1) Atomic F adsorbed on a Si surface readily becomes
negatively charged. This is because the electron affinity level
of F lies below the E F of Si when the atom is close to
the surface [126,131,133]. An electron is transferred from Si
substrate to a surface F-atom in order to equilibrate the energy
levels of Si and F. The process proceeds via electron tunneling
through the SiFx layer, which is less than 3 nm thick, and thus
feasible for this process to occur.
(2) There is also an electric field similar to what was

mentioned in Section 2.4.1.1. Now, the presence of surface
anions F forms an electric field, E 0 , and bends the surface
Si bands upwards [126,131]. The concentration of surface F
(n) and E 0 at the SiFx layer are related to Si-doping by
(E F E a )surf
.
(26)
el
Here, SiFx is the dielectric constant of the SiFx layer, E F
the Fermi level of the Si surface, E a is the affinity level of
F at the surface, and l is the thickness of the SiFx layer.
The expression (26) indicates that an increase in n and E 0 is
expected if Si changes from p+ -doping to to n+ -doping. As we
will discuss below, it is n and E 0 which influence the etching
rate significantly.
(3) The reaction rate is determined by F diffusion through
the SiFx layer toward the SiFx Si interface as well as the
reaction of F with Si and SiFx . The large electric field E 0 can
drive the diffusion of F from the gasSiFx interface toward
the SiFx Si interface; moreover, the high concentration of F
adsorbed on the surface (n) favors the reaction of F + SiFn
SiFn+1 (n < 4). Accordingly, n+ -Si has a higher etching
rate than lightly doped Si and p+ -Si. The doping effect in the
etching of semiconductor surfaces can be mostly attributed to
the differences in charge transfer process between surface F and
Si as well as the field-assisted anion diffusion process.
E 0 n = SiFx
2.4.2. CabreraMott theory in solidsolid interface reactions

In previous papers, we have systematically studied reactions
at solidsolid interfaces using metal/oxide model systems. The
reaction kinetics of metal oxidation and metal encapsulation
on oxide surfaces was found to be controlled, besides other
materials parameters, by E F of both solid phases, and the
reaction results can be explained in the framework of the
CabreraMott theory [13,85,89]. We now discuss the metal
oxidation and metal encapsulation reactions on oxide surfaces,
respectively.
2.4.2.1. Oxidation of metals on oxide surfaces. The redox
reaction at metal/oxide interfaces was investigated, with an
444
Fig. 12. (a) Schematic diagram to illustrate the spatial structure of Si surface during etching reactions. The thickness of the interface reaction layer SiFx is l. (b)
Schematic diagram to illustrate the band structure at FSi interfaces. The charge transfer between F adatoms and Si crystals produces electric field E 0 at Si surface.
After Winters and Haarer [126].
emphasis on the dependence of the reaction kinetics on

electronic structure of both metal and oxide. We studied
oxidation of ultrathin metal overlayers on SrTiO3 (100) and
TiO2 (110) surfaces. The experimentally determined metal
oxidation rate was found to closely correlate with E F of both
the metal and oxide phases. We concluded that (1) p-type doped
oxide favors metal oxidation, (2) metals with low surface work
function have high oxidation rate, and 3) the coupling between
interface charge transfer and O2 diffusion at the interface is
a critical factor for the reaction kinetics, which is given below
[85,89].
To begin with, for the oxidation of identical metals, e.g. Cr
on different SrTiO3 (100) and TiO2 (110) crystals, the oxidation
rate is dependent on the doping state, i.e., free electron carrier
concentration [e0 ] of SrTiO3 crystals or TiO2 crystals, but is
virtually independent on the oxygen vacancy concentration
[VO ] in the crystals. The reaction rate monotonically decreases
with the increase of the free electron carriers in the oxide
crystals. For a p-type oxide, this means that the system favors
metal oxidation.
Second, for the oxidation of different metals on identical
oxide crystals, it is found that the reactions show a systematic
dependence on the work function of the metal films. The
oxidation rate here increases with decreasing surface work
function of the metal.
Third, long-range charge transfer after contact between
metals and SrTiO3 (or TiO2 ) crystals produces space charges
at the oxide surfaces and induces band bending. The electronic
interaction is controlled by the relative position of E F of the
metal and that of the oxide (See Section 2.2.2). The oxidation
of reactive metals on SrTiO3 or TiO2 substrates requires O2
transport from the oxide bulk to the interface or equivalently an
inward diffusion of VO . The electric field resulting from space
charge transfer dominates the O2 diffusion at the interface,
and the corresponding interface reactivity and oxidation rates.
In case that E F of the metal is greater than E F of the oxide,
charge is transferred from the metal to the oxide and negative
space charges form in the oxide. The resulting electric field
at the interface promotes outward diffusion of O2 and metal
oxidation. This is schematically shown in Fig. 13.
2.4.2.2. Encapsulation of metals on oxide surfaces. In a
previous work, the model system Pd/TiO2 (110) was used
to evaluate the correlation between metal encapsulation and
Fig. 13. Upper panel: schematic diagram showing the energy bands of a
metal/oxide interface in the case of E F (MeI ) > E F (MeII Ox ). Lower panel:
negative space charges at oxide surface regions and the electric field E 0
produced by the interfacial charge transfer process, promoting the outward
diffusion of O2 . The arrangement of E F of metals and oxides as E F (MeI ) >
E F (MeII Ox ) favors oxidation reactions. See [85,89].
electronic structure of TiO2 crystals. Some previous results of

the encapsulation reactions at metal/oxide interfaces were also
carefully revisited therein. The following points were derived
from such investigations [13]:
(1) The studies in the interaction of Pd clusters with
differently doped TiO2 (110) crystals show that n-type doped
TiO2 crystals favor the encapsulation of Pd clusters. More
generally, we proved that encapsulation reactions occur only for
metals with large work function, e.g. Pt, Pd, and Rh, supported
on n-type doped oxides. The prerequisite to the onset of the
encapsulation is that E F of the metal phase should be much
lower than that of the oxide, E F (metal) < E F (oxide).
(2) The encapsulation reactions at metal/TiO2 interfaces
involve the dominant outward diffusion of interstitial titanium
cations, Tiin+ (n = 3 or 4), from the bulk to the TiO2 surface.
At relatively low temperature, the outward diffusion of Tiin+
must be facilitated by the space charges as shown in Fig. 14.
This requires E F (metal) < E F (oxide). In other words, the
system must have a metal with a large M and a strongly
n-type doped TiO2 . The equilibration of both E F results in
a charge transfer from the occupied donor states in TiO2 to
the metal and, consequently, an upward bending of the TiO2
bands. Positive space charges form at TiO2 surface such that
445

Table 2
Summary of reactions at solidsolid and solidgas interfaces
Fig. 14. Upper panel: Schematic diagram showing the energy bands of a
metalTiO2 interface in the case of E F (Metal) < E F (TiO2 ). Lower panel:
Positive space charges at TiO2 surface regions and the electric field E 0
produced by the interfacial charge transfer process, promoting the outward
diffusion of Ti interstitial ions, Tin+ (n < 4). The arrangement of E F of
metals and oxides as E F (MeI ) < E F (MeII Ox ) favors encapsulation reactions.
See [13].
interstitial titanium cations are driven to diffuse toward the

interface (Fig. 14).
2.4.3. Generalized CabreraMott theory in interface reactions
We have shown that the four reactions at both solidgas and
solidsolid interfaces demonstrate strong interplay between the
interfacial electronic interaction and the chemical interaction.
Such interplay originates from the strong coupling between the
interfacial mass transport and charge transfer. Table 2 lists the
transport processes in the reaction types discussed above. The
coupling between the mass transport and charge transfer may
require different electronic configurations of the two contacting
phases in order to favor the interface reaction (see Table 2). The
classic CabreraMott theory can be generalized to apply to both
solidgas and solidsolid interface reactions.
The results introduced above demonstrate that interface
reactions can be uniquely controlled by the relative positions
of E F in both phases before establishing contact. A variation
of these positions is directly coupled to changes in the electric
field at the interface and modifies the corresponding reaction
rates. These results can be combined in a unique picture by
calculating the surface electric field E S , i.e. electric field at
the interface, as a function of the E F positions in both phases
at the interface. An analytic expression of E S was derived by
Kingston and Neustadter [134]:

kT
F(s , b ),
(27)
ES =
qLD
where

qb qb
qs
F(s , b ) = 2 sinh
kT
kT
kT

qb
qs 1/2
cosh
cosh
kT
kT
(+ for s > b , for s < b ).
Interface
reactions
Contacting
phases A k B
Diffusing species
& direction
E F arrangements
favoring reactions
Redox
reaction
Metal k Oxide
O2 , A B
E F (A) > E F (B)
Encapsulation
Metal k Oxide
Tin+
i ,AB
E F (A) < E F (B)
Surface
oxidation
Metal k Gas
O2 , A B or
Min+ , A B
E F (A) > E F (B)
Surface
etching
Gas k Si
F , A B
E F (A) < E F (B)
This equation shows how the surface electric field varies

with E F of two contacting interface phases. At a metal/oxide
interface, b is the energy difference between E F and E i
(intrinsic energy level) of the bulk oxide, which reflects the
influence of the oxide and can be changed by varying the E F
of the oxide (see schematic inset in Fig. 15(a)), s is the energy
difference between E F of the metal and E i of the oxide, varying
with the E F of the metal (schematic inset in Fig. 15(b)), and
L D is the Debye length [36]. From a plot of E S as a function
of b and s one can predict the condition under which metals
will react with oxides. Fig. 15(a) shows E S as a function of
b for the case where s = 1 eV, E g = 3 eV (band gap
energy) and T = 290 K. The upward bending (b > 1 eV),
flat band (b = 1 eV), and downward bending (b < 1 eV)
will be experienced, respectively, as the oxides E F shifts down
from the VB to the CB. This illustrates that a change of the
oxide E F leads to a distinct modification of the direction and
magnitude of the electric field. A plot of E S as a function
of s with b = 1 eV is depicted in Fig. 15(b). Here, s
decreases with increasing M , which shows that E S changes
from positive values (downward bending) to negative values
(upward bending).
The calculation demonstrates that E S changes with E F of
the metal (Fig. 15(b)) or the oxide (Fig. 15(a)) monotonically
and this is one way to predict the thermal stability of
the metal/oxide interface semi-quantitatively. After contact
between the metal and the oxide, E S can be established by
charge transfer across the interface, subsequently affecting
ion transport in the vicinity of the interface as illustrated
by Eq. (23). At relatively low temperatures, E S is a critical
parameter that determines transport processes at the interface in
cases including the initiation of metal oxidation at metal/oxide
interfaces, the surface reactions in reactive Si etching, the
oxidation of metal surfaces in gas, and the encapsulation
of metal nanoclusters by oxide supports. The systematic
dependence of these interface reactions on E F is consistent
with the monotonic change of E S with E F in the materials.
3. Experimental methods
(28)
As we have discussed in the introduction, a deeper

understanding of the nature of metaloxide interactions relies
on the development of better theoretical models and advances
in experimental techniques. In the model systems, ultrathin
446
Fig. 15. Dependence of the surface electric field E S on b and s . (a): s is

constant (1 eV) for a distinct metal; E F of the oxide shifts from CB toward
VB, varying b from positive to negative (small inset). (b): For a distinct oxide,
b is regarded as constant (1 eV); E F of the metal moves down, resulting in
different values for s (small inset). The constant K is a function of the oxides
material parameters and T . For a given oxide and constant temperature, K is
constant.
metal overlayers or metal nanoclusters are grown on welldefined oxide surfaces. These simplified systems significantly
reduce the complexity in real systems in order to study
fundamental interface processes. In this chapter, we present
important methods and strategies to prepare these systems. The
most frequently used techniques to characterize metal/oxide
interfaces are subsequently introduced, with an emphasis on the
benefits and limitations of these experiments.
3.1. Preparation of model systems
As stated above, model systems consist of nanostructured
metal overlayers supported on well-defined oxide surfaces.
These systems necessitate the preparation of well-defined oxide
surfaces and well-controlled deposition of metal overlayers,
which is discussed in detail.
3.1.1. Oxide surfaces
3.1.1.1. Single crystal oxides. The simplest way to prepare
clean and well-ordered single crystal surfaces is by cleaving
in UHV [36]. The surfaces obtained in this way are generally

stoichiometric and exhibit a low defect density. There is a
limitation to the cleaving of a single crystal surface. First,
it may only be used for brittle materials, such as MgO and
ZnO crystals. Second, cleavage is only possible along certain
crystallographic directions, which results in the formation of
low surface energy surfaces with distinct cleavage planes. For
example, MgO with the rock salt structure cleaves along the
{001} surfaces, and crystals with the wurtzite structure such as
ZnO cleave along the non-polar faces {1010}. Therefore, only a
few surfaces can be studied in this way. MgO(100) is the most
used cleaved surface [135140]. The cleavage of some polar
surfaces, such as MgO(111), ZnO(0001), and ZnO(0001), has
been also reported ([24,76] and references therein).
It is also possible to prepare oxide surfaces by mechanical
cutting and polishing in air. This method is applied to the most
currently used oxide crystals, including TiO2 , Al2 O3 , SrTiO3 ,
SiO2 , as well as the cleavable MgO and ZnO crystals. Different
orientation surfaces (of a polar or non-polar nature) can be
obtained by the process. Ion sputtering is then used to remove
surface impurities, which may originate from the polishing
process, contaminations from exposure to air, segregation of
bulk impurities to the surface, or from previous experiments
carried out on the surface. Surface order, which may have been
disrupted by this harsh treatment, can be achieved by annealing
in oxygen or vacuum. Sometimes, cycles of sputtering and
annealing are required to get clean and well-ordered surfaces.
When preparing oxide surfaces one should note the
following problems.
(1) Surface stoichiometry: For compound surfaces, in
particular oxides, adequate attention must be paid to the
change in surface stoichiometry. On many oxide surfaces, ion
sputtering preferentially removes oxygen from the surfaces,
resulting in formation of oxygen vacancies and oxygendeficient surfaces. Also, vacuum heating may cause desorption
of surface oxygen or metal atoms and, therefore, deviation from
the surface stoichiometry.
(2) Surface reconstruction: During the surface preparation
processes, oxide surfaces may undergo surface reconstructions,
which result from rearrangement of surface atoms, desorption
of surface atoms, or the addition of atoms to the surface from
deposition. The most commonly observed reconstruction on
TiO2 (110) surfaces has a (1 2) symmetry with a doubling
of the periodicity along the [110] direction [28]. SrTiO3 (100)
has various surface structures,
such
as (1 1), (2 1), (2
2),
c(4
2),
c(6
2),
(
5
5)R26.6
, etc [141146].
( 3 3)R30 , (3 3 3 3)R30 , and ( 31 31)R

9 reconstructions have been found on the -Al2 O3 (0001)
surface [147150]. ZnO(0001)-(1 1) can be transformed into
a (2 2) surface structure [151].
(3) Surface termination: Another important character of
oxide surfaces is the termination. For example, SrTiO3 (100)
surfaces have two kinds of terminations: SrO- and TiO2 terminated surfaces [152,153] depending on how they are
prepared. This is similar to the case of cutting a ZnO crystal
perpendicular to the c axis creating a Zn-terminated (0001)
surface on one side and an O-terminated (0001) surface on
the other side [154,155]. For the bulk truncated -Al2 O3 (0001)
surface there are three different terminations: O layer
termination, single Al layer termination, and double Al layers
termination [25,156]. Oxide surfaces can be hydroxylated under
certain conditions, e.g. exposure to water, which leads to
a fully (or partially) covered hydroxyl surface with surface
OH-termination. These surface terminations vary sensitively
with the surface preparation conditions, for example heating
temperature, heating atmosphere, and surface sputtering.
(4) Bulk defect chemistry: For a variety of oxide crystals,
prolonged and/or high temperature heating may cause mass
exchange between the surface and bulk [157,158]. The
concentration of defects in the bulk will change during the
surface preparation process, which is particularly important for
the mixed conducting oxides, such as TiO2 and SrTiO3 , since
both the ionic and electronic conductivities in the bulk solids
will change during the surface treatment process. As we will
discuss in Section 4, defect chemistry of the oxides plays a
critical role in metaloxide interactions. The variation of the
oxide defect chemistry as a function of surface preparation
conditions has to be known in detail.
The above discussion suggests that the chemical stoichiometry, surface composition, surface atomic structure, and morphology sensitively depend on the preparation conditions. It is,
therefore, important to be aware of the dependence of surface
properties on the treatment conditions, such as the sputtering
energy and sputtering time, heating temperature and time it is
applied, and oxygen partial pressure during preparation of the
surfaces.
3.1.1.2. Supported thin oxide films. Another important method
to prepare model oxide surfaces is to grow thin films supported
on other solids. The main advantage of the method is that
adequate conductivity can be obtained by preparation of an
ultrathin oxide film on a conducting substrate or by growing
a doped thin oxide film while these thin films could mimic the
situation of the bulk oxide materials [159]. On the surfaces of
the thin oxide films, electron/ion spectroscopy (photoemission
electron spectroscopy (PES), Auger electron spectroscopy
(AES), and ion scattering spectroscopy (ISS)) and scanning
tunneling microscopy (STM) measurements become feasible.
Oxide films can be synthesized using various advanced thin
film growth techniques, in particular MBE. These surfaces
can be prepared by direct vapor evaporation of bulk oxides
using a high energy laser or, alternatively, thermal heating.
Most often, metals are evaporated onto clean well-ordered
single crystal surfaces under molecular O2 atmosphere or using
atomic oxygen [160162]. In some cases, metal deposition
is separated from the oxidation process: metal evaporation is
conducted in UHV to form epitaxial metal films, which are
subsequently introduced to an oxygen atmosphere for growth
of oxide films. This technique has been applied to the growth
of oxides of Ti [163165], Ce [166], and Fe [167,168]. For
example, Fig. 16 shows atomically flat iron oxide films grown
on Pt(111) using different preparation processes [168]. Thanks
to the well-developed MBE technique, a wide range of thin
447
Fig. 16. Low energy electron diffraction (LEED) patterns, STM images, and
structural schematics of iron oxide films grown on Pt(111). From [168].
oxide films have been deposited on refractory metals or single

crystal oxides and the oxide surfaces can be well-controlled.
An alternative method to produce thin oxide films is
to oxidize a metal or an alloy single crystal surface. The
thickness and structure of the oxide layers depend on oxygen
partial pressure, oxidation temperature, and oxidation time.
This method has been successfully used for preparation of
systems of SiO2 /Si(111) [169,170], Al2 O3 /Al(111) [171,172],
and Al2 O3 /NiAl(110) [26,27]. For example, epitaxial Al2 O3 films grown on NiAl(110) show a high degree of crystallinity,
very low surface roughness, and good reproducibility.
Experimentally, Al2 O3 /NiAl(110) has been extensively used
for model oxide surfaces [26,173].
3.1.2. Metal overlayers on oxide surfaces
There are many methods to prepare metal films, such as
physical vapor deposition (PVD), chemical vapor deposition
(CVD), and chemical solution deposition (CSD), which have
been described extensively, see [174]. Here, we give a short
description of the most commonly used preparation routes
for ultrathin metal films and metal nanoclusters supported on
model oxide surfaces.
3.1.2.1. Metal deposition by evaporation. Vapor evaporation
under UHV is by far the most applied method to deposit
metal overlayers on oxide surfaces. When a metal is vapor
deposited onto an oxide support, various atomic processes
take place at the surface. Fig. 17 shows a schematic diagram
of the elementary atomic processes and the corresponding
characteristic energies for metal adatoms on an oxide
surface [175,176].
Atoms which arrive at the surface may reside on the surface
for a certain length of time before returning to vacuum. This
is known as the residence time a . a is determined by the
448
Fig. 17. Schematic of elementary steps and characteristic energies taking place
for deposition of a metal onto an oxide surface. Z represents the dimension
perpendicular to the surface. Three basic atomic processes are critical in the
interface formation: the in-plane surface atom diffusion (Z = 0), the out-ofplane surface atom diffusion (Z > 0) including the down-step diffusion and
the up-step diffusion, and the atom interdiffusion (Z < 0). E a is the adsorption
energy; E d is the surface diffusion energy;E 1 is the up-step diffusion energy;
E 1 is the down-step diffusion energy.
adsorption energy E a ,
a1 = a exp(E a /kT ),
(29)
where a is the atomic vibration frequency. It is expected that

higher surface temperature results in a shorter residence time
and induces re-evaporation (i.e. desorption) of some adatoms.
This makes the net sticking coefficient smaller than one. For
many transition metals on oxides at room temperature, the
sticking coefficient is near unity ([26] and references therein),
indicating that this temperature is not sufficiently large to cause
desorption of the particles or the particles are caught at defects
during their residence on the oxide surface.
Adsorbed atoms can also move on the surface. This diffusion
process depends on the diffusion activation energy, E d . For
heterogeneous nucleation, the diffusing atoms may find surface
defect sites, e.g., steps and vacancy sites, and get trapped there
forming nuclei for subsequent growth process. In the process
of homogeneous nucleation, a stable nucleus is formed by
aggregation of n (n > 1) adatoms on a regular surface site.
After the saturation density of islands has been reached, it is
in the stage of island growth. Upon further metal deposition,
coalescence of islands happens and finally leads to continuous
film formation. During the growth and coalescence processes,
surface diffusion between the island and the oxide substrate
level becomes possible, as shown in the right side of Fig. 17.
Up-step (which occurs for 3D growth) or down-step diffusion
(which occurs for 2D growth) of adatoms influences the metal
film growth mode and the island shape [177,178]. The diffusion
is controlled by the activation energies for up-step and downstep diffusion, E 1 and E 1 [81].
The energies shown in Fig. 17 vary with the metal/oxide
system, and thus are strongly dependent on the metaloxide
interaction. For example, a strong interaction between a metal
and an oxide results in large activation energies of E a , E d , and
E 1 . Weak desorption, homogeneous nucleation, and layer-like
growth are thus expected. Studies of the surface processes in
the initial stage of metal deposition allow for the determination
of these activation energies and yield insight into the strength
of the interaction.
3.1.2.2. Size-selected deposition of metal clusters. An elegant

method to deposit metal nanoclusters on oxide surfaces is
the use of size-selected molecular beams and softlanding onto
substrates [179184]. Clusters generated by laser evaporation,
magnetron sputtering, or thermal evaporation sources are
passed through an ion optics system, where mass selection and
controlling of cluster energy are conducted by quadrupole mass
spectrometer and electrostatic lenses. Subsequently, the massselected clusters are deposited under UHV with low kinetic
energy onto the oxide surfaces. The softlanding of the cluster
onto the substrate avoids the fragmentation or fission of the
cluster. The method allows for the preparation of supported,
monodispersed metal nanoclusters with a very narrow size
distribution. Since the bonding between mass-selected clusters
and the support should not be strong, thermal stability of the
model systems is quite low. Reliable experiments may only be
performed below room temperature.
3.1.2.3. Chemical vapor deposition. The CVD route to metal
overlayers is carried out by exposing organometallic precursors
to a substrate in vacuum. Heat is then applied to the systems to
remove the ligands and convert the metal into the zero-valence
state. This method is sometimes referred as metal-organic vapor
phase epitaxy (MOVPE).
A special modification of CVD is the atomic layer deposition
(ALD) or atomic layer epitaxy (ALE) [185190]. In this
method, film growth takes place in a cyclic manner. One growth
cycle normally produces a single monolayer or a fraction of
it. These cycles can be repeated until thicker films are formed.
The basic characteristic of an ALD process is the self-limiting
nature of the surface reactions. Because of this, ALD enables
simple and accurate control of film growth process at an atomic
layer level. The ALD technique has been widely used to
deposit metal oxide films. Nevertheless, Ta [189,190], Ru [191,
192], Pt [193], and other transition metals [194] have been
successfully synthesized by the ALD process. The thickness
of metal overlayers or the size of metal particles can be
precisely controlled by the number of cycles used. The ability
to control the growth of metal overlayers at the atomic level
makes ALD a promising method for preparation of metal model
systems.
3.1.2.4. Solution chemical deposition. Various solution chemical methods have been utilized to synthesize metal overlayers. SolGel is one of the most used solution methods. In this
process, metal precursors are dispersed in aqueous or organic
solutions. The metal containing solutions are spin-coated or
dip-coated onto substrate surfaces, which are subsequently subjected to further treatments to get pure metal overlayers. Sometimes, colloidal chemistry is applied to control the size of colloids in the precursor solutions and, thus, the size of the deposited metal particles. Wet chemical impregnation is another
common method of metal catalyst preparation in real catalytic
systems. The method is occasionally used for the model systems. Further details about the synthesis of metal on model supports can be found in Ref. [195].
3.1.3. Inverse model systems

Most studies of metaloxide interactions involve the model
systems of metal overlayers supported on well-defined oxide
surfaces. An alternative route towards the fabrication of
model systems is the deposition of nanostructured oxide
layers on well-defined metal surfaces. This process forms an
inverse model system, which has been useful to explore the
metaloxide interactions [7,196,197]. In case of electronic
interaction between metals and oxide supports, the ratio of the
amount of metal to that of semiconducting oxide support must
be very small (e.g., very small metal clusters on large oxide)
such that the transferred charge to or from the metal is sufficient
to affect the electronic structure of the metal, such as the
metal Fermi Energy. In an inverse model system, the very large
number of electrons in the metal support can easily modify
the electronic structure of the supported semiconducting oxide
layer. The electronic interaction between the two phases
becomes more significant such that the interaction can be easily
probed [7].
The oxide-on-metal system also allows one to model the
promotion effect of oxides on metal catalysts [198,199] because
it has been found that the promotion of some catalytic reactions
may happen near the metal/oxide phase boundary. In inverse
model systems, the perimeter of the oxide islands offers an
active catalytic site, which can be well-studied by surface
science techniques.
An inverse model system consists of an oxide layer
supported on a well-defined metal surface. The growth of oxide
films has been described in Section 3.1.1.2. Systems of titania
and vanadia on Pt and Rh [197], ceria on Rh [200202], vanadia
on Pd(111) [198,203205], and vanadia on Rh(111) [199,
206] have been synthesized and studied in detail. It has been
shown that the metaloxide interactions lead to the formation
of reduced oxide phases with lower oxidation state in the
vicinity of the metal interface. The lower oxidation state derives
from the chemical interactions with the substrate atoms at the
interface and/or from the geometry-related effects, i.e., lattice
matching.
3.2. Characterization techniques
3.2.1. Electron-based spectroscopy
Electron-based spectroscopy, which includes X-ray photoelectron spectroscopy (XPS), ultraviolet photoelectron spectroscopy (UPS), AES, and EELS, are extensively used to
study metaloxide interactions. They yield information about
interface electronic structures and enable the understanding
of charge transfer processes at metal/oxide interfaces. These
techniques are also surface-sensitive. As a consequence, small
changes in surface chemical composition, e.g., surface diffusion
and interdiffusion, can be detected such that atom movements
at interfaces are able to be in situ monitored under UHV.
3.2.1.1. Photoemission spectroscopy. XPS has been applied
to investigate the changes in core level binding energy (BE).
For supported metal overlayers, in particular metal clusters,
449
core level BE shifts contain two contributions: the initial

state [207212] and final state effects [211217].
The initial state effects arise from the electronic structure
factors which are present in the neutral atoms before
photoemission. The mechanisms for this state include the
following factors [211]:
(i) Charge transfer toward or from the core-ionized atom due
to metaloxide interaction (interatomic charge transfer).
This can be seen in the oxidation of transition metals on
oxides which often produces large positive shifts of metal
core level BEs, typically >1.5 eV [18].
(ii) Electric field that arises from the effective charges in
supports or metal overlayers, for example, charging effects
on insulating supports and electric field effects from
interface space charges.
(iii) Surface core level shifts including contributions of reduced
metalmetal coordination number and rehybridization of
valence levels (intra-atomic charge transfer). The two
contributions have opposite signs, such that the surface
core levels typically present small negative BE shifts
(300 meV) [208,210]. In case that the initial state
effects are dominant in BE shifts, the shifts can be used
to determine interfacial charge transfer, bonding state of
interface metal atoms, and interface reactions.
The final state effects arise from the charge rearrangement
or relaxation which screens the core holes created after photoemission. The screening effects depend on the surrounding environment, i.e., coordination number, ligands from atmosphere,
and substrate. In small metal clusters supported on weakly interacting substrates, the screening of core holes in the particles is limited compared to the bulk metal. The extra energy of
the unscreened charge should be proportional to the Coulomb
energy e2 /2R (R: radius of clusters), which results in a positive shift of the core level BE by a similar amount [214217].
If the final state effects are dominant, the shift of the metal
core level BEs is inversely dependent on the cluster diameters. This can be seen in Fig. 18 which shows BE position
of Cr2p spectra as a function of Cr nominal thickness for Cr
on a SrTiO3 (100) grown at room temperature [218]. Positive
BE shifts originating from final state effects can also be found
in other systems with magnitudes up to 1 eV, e.g., 0.7 eV in
Pt/TiO2 (110) [219], 1.0 eV in Ag/TiO2 (110) [220], and 0.8 eV
in Pt/SrTiO3 (100) [221].
In most systems, both the initial state and final state effects
contribute to the metal core level BE shifts. The Auger
parameter has been introduced which helps us to separate the
two effects. If we reference all the BE values with respect to
VBM the BE shifts can be given by
1BE = 1E 1R
(30)
where 1E is energy shift due to the initial state effects and 1R

is the relaxation energy due to the final state effects. The Auger
electron energy is influenced by the same effects as core level
BE. The shift of the Auger electron kinetic energy (1KE) is
written as
1KE = 1E + 31R.
(31)
450
Fig. 18. Cr2p3/2 core level BE as a function of the thickness of Cr overlayers

on SrTiO3 (100) grown at room temperature. The insert figure is the dependence
of BE shift (1E) on the reciprocal of Cr cluster radius (1/R). From [218].
The Auger parameter 0 is defined as [222224]:

0 = KE + BE.
(32)
Based on certain assumptions and the three equations above,

the relaxation energy, 1R is simply estimated to be half of the
shift of the Auger parameter (1 0 ) which means that the initial
state contribution is equal to 1BE + 12 1 0 [225,226]:
1
1
1 0 ;
1E = 1BE + 1 0 .
(33)
2
2
Therefore, the initial state and final state effects can be
distinguished in this way [221,227].
For core level spectra recorded from oxide supports, changes
in the spectra can be simply associated with the initial
state effects. The charge transfer between metal adatoms and
substrate surface atoms causes BE shifts or line shape changes
in the corresponding substrate core levels. For example,
redox and encapsulation reactions at metalTiO2 interfaces
often result in reduction of surface Ti. Ti2p spectra become
asymmetric due to the presence of shoulder peaks from Tin+
(n < 4), e.g. see Fig. 40 in Ref. [28]. On the other hand, space
charge transfer between the two contacting phases leads to the
band bending and the rigid shifts of all substrate core levels.
Bending of oxide bands can be monitored as a function of metal
coverage by measuring the rigid shift.
UPS provides information on the VB structure. The VB of
supported metal clusters varies with the metal cluster size or
the metal coverage. With a decreasing cluster size, the metal
VB shows a similar shift to that of the core levels. The strong
localization of electrons in small clusters causes a narrowing of
the VB. The shift of oxide VB spectra, however, reflects the
band bending at the oxide surface. In particular, the change
of oxide VBM positions after metal adsorption can be used
to determine direction of space charge transfer between the
support and the metal. UPS can often be used to measure the
surface work function as a function of metal coverage [228].
1R =
3.2.1.2. Electron-excited spectroscopy. AES is widely used to

monitor the metal film growth process due to the simplicity
of the technique and the speed at which a measurement can

be taken. Quantitative or qualitative analysis of AES intensitycoverage data allows extracting film growth modes and the
strength of the interaction [177,178,229]. Combining AES
data acquisition with ion beam sputtering, one can conduct
a depth profile analysis [230]. The depth profiling of a
metal/oxide interface gives chemical composition distribution
from the surface to the bulk, which is powerful in investigation
of interface reactions, especially encapsulation or decoration
reactions [231].
As an Auger transition includes three ionization processes,
AES measures a convolution of three densities of states
which makes it hard to get direct information about the
surface electronic structure. Nevertheless, Auger peak shifts
occasionally allow one to identify the change in chemical states.
Bernath et al. [232] studied Ti deposition onto Al2 O3 (0001)
surface by AES. With increasing Ti coverage, the initial Al LVV
Auger peak position at 61 eV (from Al3+ in Al2 O3 ) shifts to
66 eV, which is characteristic of metallic Al, i.e., metallic Al
LVV transition. This shift is due to the reduction of the sapphire
surface by absorbed Ti.
EELS is another electron promoted technique that consists
of inelastic scattering of low energy electrons by surface
species. The application of EELS in metal/oxide interfaces
makes use of the interaction of incoming electrons with surface
phonons, bulk and surface plasmons, and interband transitions.
The features related to electronic structures of oxide surfaces
and metal overlayers are in the range 150 eV. The primary
energies (E 0 ) are in the 100 eV range.
HREELS, a high resolution form of EELS, is performed at
low primary energies, with E 0 < 10 eV. It is used to reveal
the surface optical phonons or FuchsKliewer modes [233]
on oxide surfaces. The intensities of oxide surface phonons
can be attenuated by formation of a metallic free carrier layer
atop an oxide surface or by charge transfer from the metal
overlayers to the substrate. Additionally, free carriers in the
metallic overlayers would give rise to a plasmon-like mode
ranging from zero to higher energies. The elastic peak can
be broadened asymmetrically toward the positive side [234].
Changes in intensity of surface phonons, the full width at half
maximum (FWHM) of the elastic peak, and loss-peak positions
can give information about electronic interaction between
metal overlayers and oxide surfaces. Petrie and Vohs applied
HREELS to study Pt films on ZnO surfaces. The HREELS
results indicate that there are only weak interactions at the
Pt/ZnO (0001) interface, while charge transfer and Schottky
barrier formation occurs at the Pt/ZnO(0001) interface [235].
3.2.2. Scanning probe techniques
STM can deliver images of solid surfaces down to atomic
scales, which helps us to identify metaloxide interactions, such
as the SMSI state, adsorption sites, and metal interaction with
surface defects. The tunneling characteristics of the technique
limits STM to oxide crystals with relatively small band gaps,
e.g., TiO2 , SrTiO3 , and ZnO. The nucleation of Pd on a
TiO2 (110) surface was investigated through STM. The images
show the presence of dimer and tetramer Pd clusters but
451
Fig. 19. (a) Low energy ion scattering (LEIS) spectra of (bottom) the clean TiO2 (110) surface, (center) after evaporation of 25 monolayers Pt at room temperature,
Most
and (top) after the high temperature treatment causing encapsulation. (b)(e) STM results after the high temperature treatment. (b) Overview (2000 A2000
A).
500 A),
clusters show hexagonal shape elongated along the substrate [1] direction (type-A). A few square clusters (type-B) are seen. (c) Small-scale image (500 A
filtered to show the structure of the encapsulation layer on type-A clusters. (d) Atomic-resolution image of an encapsulated type-A cluster. (e) Atomic-resolution
image of a square type-B cluster, showing an amorphous overlayer. (f) STS of the different surfaces. From [11].
no single Pd atoms at the nucleation stage. The nucleation

of Pd clusters is preferred at step edges. On terraces, Pd
atoms adsorb on the five-fold coordinated Ti cations between
two bridging oxygen rows [236]. Dulub et al. succeeded in
recording atomically resolved STM images on Pt clusters
grown on a TiO2 (110) surface, which were in the SMSI state
after high temperature annealing in UHV. The encapsulation
layer with a striped zigzag structure on (111)-oriented Pt
clusters was proposed to be a slightly oxygen-rich, oxygen
terminated TiO1.1 (111) double layer [11] (see Fig. 19). More
often, STM is used for in situ morphological investigations of
metal overlayers during growth process, thermal treatment, or
gas adsorption. Surface processes, such as nucleation, growth,
coalescence, sintering, gas-induced island restructuring, etc.,
have been studied for various metals supported on SrTiO3 (100)
and TiO2 (110) surfaces [91,220,237242].
STM can also be applied to image wide band gap oxide
surfaces provided that a well-ordered ultrathin oxide film
is supported on a conductive support. Metal interactions
with oxide films have been extensively explored by STM,
for example, in metal/Al2 O3 /NiAl(110) systems by Freunds
group [26,243] and metal/SiO2 /Mo systems by Goodmans
group [100,161,244].
Scanning tunneling spectroscopy (STS) provides information on the local electronic structure of solid surfaces. Both the
occupied and unoccupied energy states are probed by ramping
the bias voltage in positive and negative directions. The magnitude of tunneling current (I ) at one bias voltage (V ) is related to
the density of states of the sample at that energy. The obtained
I V curves reflect the electronic character of the tunneling
sites. In supported model systems, STS spectra can be recorded
on an individual metal particle, which allows for the exploring

of the electronic structure of individual metal clusters. Valden
et al. acquired STS spectra from various Au clusters supported
on a TiO2 (110) surface. They observed a metallic to nonmetallic transition of the clusters with a decrease in size [245]. The
appearance of band gap on small clusters is caused by the depletion of the density of states near E F and probably due to quantum size effects [246]. In Pt/TiO2 (110) systems, I V curves
were acquired from clean Pt clusters and Pt clusters in the SMSI
state. The spectra showed that the electronic structure of clusters in the SMSI state is more semiconductor-like due to decoration with TiOx layers [11,247] (e.g. Fig. 19(f)).
Atomic force microscopy (AFM) or scanning force
microscope (SFM) is not limited by the sample conductivity,
but does not present a high spatial resolution as STM.
Recently, non-contact AFM and dynamic-mode SFM have
been developed to yield atomic-resolution images of some
oxide surfaces, such as -Al2 O3 (0001), TiO2 (110), and
MgO(100) [147,248,249].
3.2.3. Transmission electron microscopy
Transmission electron microscopy (TEM) is one of the most
powerful tools to study metal/oxide interfaces. Conventional
TEM (CTEM) is the most common technique to investigate the
nucleation, growth, and coalescence of metals on oxides [250].
The direct particle size and shape determination is possible
for particles larger than 3 nm. The orientation relationship of
metal islands with support can be acquired when employing
transmission electron diffraction (TED). Polli et al. [251]
have characterized Pt growth on SrO-terminated SrTiO3 (100)
surface via TEM. Fig. 20 gives plan-view and cross-sectional
452
acquired: one from a region containing the interface and

two from the nearby bulk materials on each side of the
interface. The interface-specific EELS components are obtained
using the spatial difference method. The difference spectrum
representing the EELS of interface atoms gives information
about oxidation state and chemical environment of these atoms.
Such a technique has been successfully applied to derive the
interface electronic structure at Cu/-Al2 O3 , Cr/SrTiO3 , and
Ni/SrTiO3 systems [257,258].
Fig. 20. Plan-view and cross-section bright field (BF) images recorded from
20 nm Pt films supported on SrTiO3 (100) surface covered by 2 ML SrO.
From [251].
images recorded from a 20 nm thick Pt film grown on

SrTiO3 (100). Both images clearly show the presence of two
kinds of Pt islands, as shown by the different morphologies
of Pt islands. The pyramids and truncated pyramids are from
(100)-oriented Pt grains while the flatter and hexagonal islands
present (111) orientation. The standard bright field (BF) images
essentially show 2D projections of the shapes of the particles.
Selected zone dark field (DF) and weak beam dark field
(WBDF) methods have been developed to obtain the 3D shapes
and internal structures of small particles [24,252].
The power of HRTEM in direct imaging of atomic
structures of solids is of great value in investigations of
metaloxide interactions. Interface reactions between metals
and oxides, e.g., redox reaction, encapsulation, alloy formation,
and interdiffusion, can be directly observed in HRTEM
(see Fig. 8). Bernal et al. [94,95,253,254] have shown that
electron microscopy can make an outstanding contribution
to understanding of the effects of SMSI. HRTEM revealed
the actual nature of the metalsupport interactions in several
metal/oxide systems, e.g. noble metal (NM)/CeO2 and
NM/TiO2 systems. More often, HRTEM is used to distinguish
between the reactive and non-reactive metal/oxide interfaces.
At reactive interfaces, the newly formed interface phases can
be identified. In the case of a non-reactive interface, which
is atomically sharp, experimental HRTEM results may be
compared with the results of image simulation and electronic
structure calculation. This can give information about interface
bonding, translation state, and defects (e.g. [29,255]).
Electrons can be used not only for imaging in TEM but
also for chemical analysis in the analytic electron microscope
(AEM). EELS and the related electron energy-loss nearedge structure (ELNES) analysis are based on the inelastic
scattering processes due to interactions of transmission electron
beams with samples. EELS can provide chemical analysis
at high spatial resolution. AEM offers a spatial resolution
and an energy resolution close to 0.1 eV [256].
close to 1 A
That means that electronic structures at atomic sites can
be probed. Experimentally, the spatial difference techniques
as well as EELS line scan across the interfaces are used
to study metal/oxide interfaces. Three spectra need to be
3.2.4. Other techniques

ISS, including low energy ion scattering (LEIS) and medium
energy ion scattering (MEIS) spectroscopy, may be used for
surface analysis. Among them, LEIS is extremely sensitive to
the composition of the topmost layer. Each surface element
gives a distinct peak in the ion scattering spectrum. The
area of one peak is proportional to the coverage of the
corresponding element in the top layer. Interface reactions,
in particular encapsulation, can be effectively studied by this
method (e.g., see Fig. 19(a) recorded from Pt/TiO2 (110) in the
SMSI state) [11,259,260].
X-ray scattering, in particular, the grazing incidence Xray scattering (GIXS) using synchrotron radiation sources,
is another powerful technique for characterization of single
crystal oxide surfaces and metal/oxide interfaces. Renaud has
given a detailed review of this topic [25]. Compared to the
electron-based surface techniques, this X-ray-based surface
analytical technique possesses a few unique advantages. First, it
is not subject to charging effects such that the surface analysis
will not be limited by the insulating character of most oxide
crystals. Secondly, this technique can be applied in situ, e.g., on
an oxide surface at high temperature, at high pressure, or
during the growth of metal overlayers (e.g. [261]). Furthermore,
X-ray interacts weakly with solids and, thus, quantitative
analysis based on a single scattering calculation is possible.
Therefore, the surface X-ray scattering technique allows for
the experimental determination of the atomic structure of oxide
interfaces with high accuracy (e.g. [173,262]).
Electron diffraction techniques including low electron
energy diffraction (LEED) and reflection high energy electron
diffraction (RHEED) are applied to investigate the surface
structure. On oxide surfaces, either LEED or RHEED delivers
quantitative information about surface reconstruction and
relaxation. Upon metal deposition, the interfacial structure,
such as orientation of metal overlayers with substrate, new
interface phases formed by reactions, and metal phase
transformation, can be effectively studied by the surface
electron diffraction methods [91,263,264].
Molecular beam experiments are frequently applied to study
the metaloxide interactions [243]. In the experiments, various
molecular probes were used to test the surface reactivity. For
example, metal catalysts in SMSI states present much lower
H2 and CO chemisorption capacity [14]. The charging state of
metal clusters supported on oxide surface can be detected by
frequency shifts of the intramolecular vibration of adsorbed CO
measured by infrared (IR) or EELS [265,266].
453
Fig. 21. (a) Structural model of the rutile TiO2 (110)-(1 1) surface. Large white balls represent oxygen, and small black balls represent titanium atoms. 6c-Ti: sixfold coordinated Ti atoms; 5c-Ti: five-fold coordinated Ti atoms; 2c-O: two-fold coordinated protruding O atoms (bridging O); 3c-O: fully three-fold coordinated O
showing surface defects. It is accepted that the contrast in empty-states STM images of TiO2 (110) is dominated by electronic
atoms. (b) STM image (200 A200
A)
effects. The bright rows correspond to the five-fold coordinated Ti atoms, while the dark troughs are from the bridging oxygen rows. Bright spots centered on dark
rows (defect A) are from missing single oxygen vacancies. The dark spots on bright rows (defect B) are attributed to subsurface oxygen vacancies. From [270].
4. Interaction of metals with mixed conducting oxides
4.1. Metals on TiO2
In electronic solids, atomic constituents exhibit negligible

mobility while electronic free carriers can be easily excited and
transferred. In ionic solids, e.g. solid electrolytes, at least one
ionic component is highly mobile whereas another (at least)
ionic component is virtually immobile. For mixed conducting
solids, both electrons (holes) and ion(s) are mobile such that
these solids exhibit electronic and ionic conductivities. These
materials have the ability to transform chemical energy or
information into electric energy or information (and vice versa),
and therefore possess many important applications [267].
Oxides, such as TiO2 , SrTiO3 , CeO2 , ZrO2 , SnO, and ZnO,
are typical mixed conductors. Inside these oxides, mobile
electronic and ionic defects can be generated according to
distinct defect reactions. The defect concentration is a function
of the corresponding mass-action laws, temperature, oxygen
partial pressure, and extrinsic dopants (e.g., see [268,269]).
Due to the mobility of both electronic and ionic defects
in a mixed conducting oxide, the interaction of a metal with
the oxide depends on both ionic and electronic conductivities.
Here, the electronic interaction is mainly determined by the
electronic character of the oxide. As discussed in Section 2.2.2,
the interfacial charge transfer between a metal and a mixed
conducting oxide can be treated similar to the case of
a metalsemiconductor junction. In this system, the oxide
behaves like a semiconductor. On the other hand, the chemical
interaction involves atom diffusion at interfaces such that the
transport of ionic defects in the oxide should be considered. In
such a case, the oxide is regarded as an ionic solid. Interactions
between metals and mixed conducting oxides (TiO2 and
SrTiO3 ) are reviewed here. We show how the interactions vary
with oxide surface properties, defect chemistry of oxides, and
metal overlayers.
There is a large body of literature concerning the surface and

interface studies of metalTiO2 systems, principal among them
an article by Diebold which gives an excellent overview of the
surface science of TiO2 [28]. Here, we focus on the interactions
of metals on well-defined rutile TiO2 (110) surfaces under UHV
conditions. TiO2 surface defects, TiO2 bulk defects, and metal
overlayers are found to be the most important factors when
determining the metalTiO2 interactions. These factors are now
detailed below.
4.1.1. TiO2 surfaces
The interactions of metals with TiO2 (110) surfaces at the
initial stage of metal growth (mainly at submonolayer coverage)
is discussed in this section. We show that surface defects of
a TiO2 (110) surface play a critical role in adsorption and
nucleation of metal adatoms. As expected, the metal interaction
on a defect-free surface is quite different than that on a defective
surface. It is therefore essential to discuss both cases separately.
4.1.1.1. Defect-free surfaces. Bulk truncated TiO2 (110) surfaces present two kinds of termination: the polar surface terminated with either Ti or O, and the non-polar surface containing
both under-coordinated Ti and O. Many experiments have confirmed that only the non-polar surface is stable [28]. On this surface there exist six-fold coordinated Ti atoms (6c-Ti), five-fold
coordinated Ti atoms (5c-Ti), two-fold coordinated protruding
O atoms (bridging O) (2c-O), and fully three-fold coordinated
O atoms (3c-O). These surface atoms are schematically shown
in Fig. 21.
In the case of metal adsorption on the TiO2 (110)-(1 1)
surface, the adatoms may register at different sites [271]: at
the monocoordinated site, atop of 5c-Ti or protruding 2c-O;
at the dicoordinated site, bridging two 2c-O; at the three-fold
454
coordinated site, between site formed by two 2c-O and one

3c-O or adjacent site by one 2c-O and two 3c-O; at the
tetracoordinated site, the hollow site surrounded by two 5c-Ti
and two 3c-O. Generally, one adsorption site is preferred over
the others. Such a preferable geometric arrangement results in
the strongest interface bonding between the metal and substrate
atoms. As discussed in Section 2.2.1.4, the bonding strength
is strongly correlated with electronegativity of two bonded
elements. As a rule of thumb, Pauling electronegativity (X M ) of
metals can be used to find the local interaction of metal adatoms
with the TiO2 (110)-(1 1) surfaces. As detailed below, metals
with X M < 1.9 form interface bonds with surface O whereas
metals with X M > 1.9 tend to register on surface Ti cations.
For metals with X M < 1.9 including alkali metals, early
transition metals, and Al, bonding of these metals with the
TiO2 surfaces mainly occurs via surface O. There is an indirect
charge transfer from the metal adatoms to surface Ti ions which
is mediated by the surface O. The adatoms try to maximize their
O coordination numbers and, thus, should sit on the three-fold
O coordination sites or sites bridging two 2c-O.
For example, in the case of Na-adsorption on TiO2 (110)
Onishi et al. [272] proposed a Na2 O-dimer model in which
a Na atom bonds to two bridging O atoms. Lagarde et al. [273]
applied extended X-ray absorption fine structure (EXAFS) to
determine the adsorption site of Na. They, however, found
that for Na coverage ranging between 0.25 and 0.5 ML the
metal is between a three-fold coordinated site where it is
bonded to two bridging O atoms and one in-plane O atom.
The same geometric configuration of Na adatoms on TiO2 (110)
has been experimentally observed by Murray et al. [274] and
Nerlov et al. [275]. In order to determine the adsorption site
of K on the TiO2 (110)-(1 1) surface, surface extended Xray absorption fine structure (SEXAFS) combined with STM
and non-contact AFM were used for the investigation of 0.15
ML K deposited on the TiO2 surface [276]. The experimental
data suggest that K also occupies a three-fold coordinated site
in which K bonds to two bridging O atoms and one in-plane O
atom. A similar three-fold hollow site seems plausible for Caadsorption on TiO2 (110) surface [277]. Ab initio calculations
and molecular dynamics (MD) simulations further confirmed
that the preferred adsorption positions for alkalies on TiO2 (110)
are the three-fold O coordination sites. Both between and
adjacent sites should be considered and the adsorption sites
may change with coverage [278,279].
Diebold et al. studied electronic structure of ultrathin Fe
films on TiO2 (110) with soft X-ray photoelectron spectroscopy
(SXPS) and resonance photoemission [18]. The lack of
hybridization of the Fe and Ti states suggests that the
bonding of Fe occurs predominantly via surface O and that atop
bonding above a Ti atom appears unlikely. Atomic-resolution
STM images recorded on a V/TiO2 (110) surface is proof that V
adatoms adsorb in the three-fold hollow sites. On these systems,
the V atom bonds to two bridging oxygen atoms and one basal
oxygen atom [280].
The formation of metalO interface bonds facilitates the
local charge transfer from metal to TiO2 . Since surface O atoms
are in their closed-shell configuration (O2 ) the additional
electrons transferred from the metal adatoms will populate

empty orbitals of Ti atoms. The transferred charge localizes at
the 3d orbital of the five-fold coordinated Ti ion resulting in
the appearance of occupied Ti3d-derived gap states [71]. This
electronic interaction can be observed by XPS measurements,
where shoulder peaks at low BE in XPS Ti2p spectra appear
due to the electron transfer from metal to surface Ti. The results
have been found in many metal/TiO2 interfaces [281286].
Metals with X M > 1.9, e.g. noble metals, try to bond with
surface Ti cations in most cases. There is a very weak, if any,
charge transfer from the substrate to the metal adatoms which
may be facilitated via formation of metalTi bonding.
Au-adsorption on the TiO2 (110)-(1 1) surface has been
extensively studied. Theoretical investigations show that Au
atom may bind to either a 5c-Ti atom on the basal plane
or atop a bridging O atom. The most favorable site depends
on Au coverage and calculation methods. Nevertheless, the
two sites have quite similar BEs and are comparable to each
other [271,287,288]. The electronic interaction between Au and
TiO2 is very weak. For example, MIGS and metal polarization
effects are found to contribute to the interaction [271,288].
Experimentally, STM images show that many of small Au
clusters are located on top of the bright rows that arise from
surface five-fold coordinated Ti atoms. Thus, it appears that Au
atoms nucleate on top of the Ti cations [240]. Another very
impressive STM result from Tong et al. also confirms that small
Au clusters, which are deposited on TiO2 (110)-(1 1) by sizeselected Au2 Au4 beams, are located atop of 5c-Ti [289].
XPS and LEIS studies in Ag deposition on stoichiometric
TiO2 (110) surface lead to a similar conclusion that Ag atoms
are preferentially bonded to Ti rather than O [290].
Horsley has conducted a molecular orbital study of Pt
interaction with TiO2 . The calculations favor a model in which
the nearest neighbor of Pt atom is the Ti ion and covalent
mixing between Ti3d and Pt5d orbitals occurs. There is some
ionic contribution to the PtTi bond and the Pt atom is
negatively charged by 0.11 e/atom [291]. A later ab initio
study by Xu et al. used the same geometric structure and
concluded that Pt is also negatively charged [292]. Adsorption
of Pt on a 5c-Ti atom on TiO2 (110) has been directly imaged
by STM [293]. XPS investigations of Pt-adsorption on an ideal
TiO2 surface indicate that the preferable Pt-adsorption site is
on top of 5c-Ti [294,295]. For Pd on TiO2 (110), STM imaging
shows that Pd atoms also adsorb on the 5c-Ti ions [236,240].
4.1.1.2. Defective surfaces. Various defects can be introduced
on the ideal TiO2 (110)-(11) surface and they are very critical
for metalTiO2 interactions. Oxygen vacancy and hydroxyl
group are the most important point defects on TiO2 (110)
surface. Surface oxygen vacancies can be created by UHV
heating, ion sputtering, and irradiation by electron beams or
UV light. The oxygen vacancies introduce Ti3d-derived gap
states 0.8 eV below the CBM and partially reduce the surface
Ti4+ ions. In STM images, these vacancies appear as bright
features centered on dark rows and connect two neighboring
bright rows, which suggest that they are from the bridging O
rows (see Fig. 21). The concentration of surface O vacancies
can be well-controlled by the surface treatments [28,296].

Dissociation of water on many defective oxide surfaces can be
observed [297]. On defective TiO2 (110) surfaces, it has been
found that water reacts very efficiently with oxygen vacancies
in a wide temperature range, which results in formation of
hydroxyl groups on the surface. The oxygen vacancy is imaged
as a faint bright spot in the dark rows while the feature from
the OH group appears brighter in the STM images than the
vacancy [298300]. Steps are always present on single crystal
surfaces due to a miscut of the crystal. On TiO2 (110) surfaces,
step edges run predominantly parallel to the h001i and h111itype directions. At the step edges, under-coordinated O and
Ti atoms can be found [270]. Finally, the (1 1) surface
may be subjected to reconstruction during heating the surface
in UHV or under oxidizing conditions. The most commonly
observed reconstruction is the (1 2) structure. Both missingrow and added-row models were proposed to explain the (12)
reconstructed surface structure [301304]. Presently, the addedrow reconstruction, in particular added Ti2 O3 rows model, is
generally considered to be a viable model [28].
Again, as discussed in Section 4.1.1.1, metal adsorption on
TiO2 (110) relies on X M of the metal. Reactive metals with
X M < 1.9 prefer to adsorb onto surface O via formation of
metalO bonds and the interface bonding strength is strong.
There is no tendency for the metals to decorate step edges
or oxygen vacancies because of an oxygen deficiency in
these sites. The metal atoms cover the surface evenly without
preferable decoration at any defective sites. It is also consistent
with the large adsorption energy and small diffusion length
which are expected for the highly reactive metals on the
stoichiometric TiO2 surface. Thornton and coworkers have
systematically studied the adsorption of Na [274], K [276],
and Ca [277] on TiO2 (110) surfaces with scanning probe
microscopy (SPM). The metal-induced surface structures are
observed to distribute evenly on the surfaces. Al and V interact
strongly with the TiO2 (110) surface. No obvious preferential
nucleation of Al [305] and V [280,306] clusters at the surface
defect sites has been observed.
For the noble metals with X M > 1.9, the surface defects
have a significant effect on the metal adsorption on TiO2 (110).
The role of steps, reconstructions, and oxygen vacancies in
the surface metal adsorption, which are all quite important, is
discussed.
Steps: Steps behave as trapping sites for noble metal
adsorption such that preferential nucleation of noble metal
clusters at the edge sites is often observed. Studies on Ag [220,
239,307], Cu [238], Au [245,308311], and Pd [236,312]
growth on the stoichiometric TiO2 (110) surfaces demonstrate
that most of metal clusters predominantly cover steps. The
weak interaction of the metals with the stoichiometric surface
sites enables that the adatoms have enough mobility on the
surfaces and they can diffuse to the step edges which contain
under-coordinated Ti atoms [248,270] where they act as sinks
for metal formation.
Reconstructions: On the TiO2 (110)-(1 2) surface, defect
sites, such as added rows and cross-link structures, act as traps
or diffusion barriers for metal atoms. This causes the mobility
455
of the noble metal atoms on the reconstructed surface to be

strongly reduced [311313]. The metal particles show high
dispersion on terraces and, therefore, have small size and high
island density. It has been found that the Ag cluster density
on the reconstructed surface is larger than that on the (1 1)
surface [220,313]. A similar phenomenon was observed for
Cu [314], Au [311,315], and Pd [312] on the (1 2) surface.
Berko et al. found that on the reconstructed surfaces Ir, Rh, and
Pt particles evenly covered both terraces and step edges and the
steps are not decorated preferentially [241,316,317].
Oxygen vacancies: On both (1 1) and (1 2)TiO2 (110)
surfaces, oxygen vacancies exert a strong influence on the
interaction of metal adatoms with the rutile surfaces. Tong et al.
confirmed that Ag clusters tend to nucleate at oxygen vacancies
due to the strong bonding of Ag atoms with the vacancies.
Their calculations show that the bonding of an Ag atom to
an oxygen vacancy site is stronger by about 0.5 eV than that
of any other site on the terrace [313]. High resolution STM
images reveal that Au clusters nucleate primarily at bridging
oxygen vacancies at 130 K. Calculations show that for a single
Au atom the most stable configuration is adsorption in an
oxygen vacancy site. The Au bond on the stoichiometric surface
originates from bond polarization while the Au vacancy bond
is covalent with very little charge transfer [309]. The result is
further confirmed by recent calculations [318]. They found that
Au remains neutral on regular sites of TiO2 (110) surface but
becomes negatively charged (Au ) when trapped at an oxygen
vacancy.
Surface hydroxyl groups: The recent research results suggest
that surface oxygen vacancies react quickly with residual
water even under UHV conditions such that the TiO2 (110)
surfaces are mostly hydroxylated [298]. Besenbacher and
coworkers found that the interaction between Au clusters and
the hydroxylated TiO2 (110) surface is quite weak while strong
AuTiO2 interaction occurs for the reduced TiO2 (110) surface
with oxygen vacancies and the oxidized TiO2 (110) surface with
adsorbed oxygen atoms [319].
4.1.2. TiO2 bulk defect chemistry
Chemical reactions of metal overlayers on TiO2 surfaces
involve the diffusion of ionic defects between the bulk and
surface of the oxide. Therefore, the metalTiO2 interface
reactions are closely related to TiO2 bulk defects. We show how
the metalTiO2 interface reactions including metal oxidation
and metal encapsulation depend on the bulk defect chemistry
of TiO2 crystals.
4.1.2.1. Defect chemistry of TiO2 . The defect chemistry and
electronic structure of TiO2 solids have been extensively
studied (see [28] and references therein). It is now wellaccepted that the main ionic defects in TiO2 are interstitial
titanium ions (Tii or Tii ) and oxygen vacancies (VO or
VO ) (KrogerVink notation). The diffusion mechanism of Ti
and O is that Ti diffuses in the solids as an interstitial atom and
O is transported via vacancy diffusion [320322].
456
Typical defect reactions in the solids occur according to the

following equations:
OOX VO + 2e0 + 1/2O2
X
TiTi
+ 2OOX
Tii
(34)
0
+ 4e + O2
(35)
for Nb donors on Ti sites.
(36)
and
X
NbTi
NbTi + e0 ,
X
Here, OOX and TiTi
are lattice O and Ti sites. Temperature and O2
partial pressure (PO2 ) determine the concentration of the ionic
and electronic defects in TiO2 . Furthermore, extrinsic dopants
such as Nb and Fe also affect the defect equilibria. Thus, the
ionic and electronic defects in TiO2 can be controlled by high
temperature heating, extrinsic doping, and ion sputtering [296,
323325]. We show how these treatments change the TiO2
defects and electronic structure.
High temperature heating: Heating TiO2 in an oxygendeficient atmosphere (e.g. UHV and H2 ) is known as high
temperature reduction (HTR). As shown in Eqs. (34) and (35),
this leads to the loss of O and contributes to the formation
of Ti and O defects, acting as intrinsic donors. Stoichiometric
rutile TiO2 is an insulator with band gap energy of 3.05 eV at
room temperature. In non-stoichiometric TiO2x (0 < x < 1),
defect states (essentially Ti3+ ), which originate from oxygen
vacancies, are present in the band gap at 2.3 eV above
VBM [296]. If the density of the defect states is sufficiently
high they can develop into a shallow CB, which produces free
electrons in the CB and thereby shifts E F towards the CB edge.
Extrinsic doping: Nb-doping in the parts-per-thousand range
gives rise to the formation of shallow donor states 0.02 0.03 eV
below the CBM. The donor states cause n-type semiconducting
behavior of TiO2 [323325].
Ion sputtering: Ar+ sputtering can preferentially remove O
atoms from the solid surfaces [296] by introducing O vacancies
in the topmost surface layer (e.g., Ar+ energy with 200 eV).
However, heavy ion sputtering (e.g., Ar+ energy with 3 keV)
produces O vacancies down to subsurface regions and may
change the stoichiometry of the bulk region near to surface [13,
92,326].
4.1.2.2. Effect of TiO2 bulk defect on metal oxidation on TiO2 .

Metal oxidation is the most common reaction at metalTiO2
interfaces. We show that the TiO2 bulk defect chemistry has a
strong effect on the metal oxidation reaction. Recently, Fu et al.
thickness) on three
studied the oxidation of Cr overlayers (6 A
differently prepared TiO2 (110) crystals [89]. The TiO2 defect
chemistry was controlled by HTR, doping, and sputtering.
In Case (A), TiO2 is doped with Nb (donor, 0.01 at.%) which
was Ar+ -sputtered (200 eV, 10 min) and UHV annealed at
800 C for 1 h. Case (B) is undoped TiO2 (110) which was Ar+ sputtered (200 eV, 10 min) and UHV annealed at 800 C for 1
h. Case C, which is also undoped TiO2 (110), was oxidized in
air (800 C, 6 h) and lightly Ar+ -sputtered (200 eV, 1 min). The
free electron concentration ([e0 ]) in Case A is higher than that
for Case B and C. The oxidized TiO2 (Case C) has the lowest
[e0 ] because annealing in an oxygen-rich atmosphere tends to
remove oxygen vacancies, i.e. the intrinsic donors, in the crystal

(Eq. (34)). Accordingly, we can conclude that [e0 ] in TiO2
decreases from the Case A to Case C. It should be mentioned
that introduction of donors, e.g. Nb, into TiO2 may decrease
[VO ] according to a compensation reaction [327]. Among the
three crystals, Case B possesses the largest [VO ].
To investigate the bulk defect effect on metal oxidation, Cr
overlayers were grown on the different TiO2 crystals and all
were stepwise heated in UHV to 600 C. In situ XPS was
performed to monitor the oxidation process after heating. The
following details the results. For Case A, Cr oxidation began
at 370 C, but metallic Cr was still observed on the surface at
550 C. With Case B, Cr was almost fully oxidized at 460 C,
strikingly different than Case A. Case C was already lightly
oxidized at 280 C and nearly converted to Cr oxide at 370 C.
These results indicate that the reaction rate of Cr oxidation
0
on TiO2 is inversely
proportional to [e0 ] in TiO2 rather than
correlated with VO in TiO2 . Lower [e ] in TiO2 favors faster
oxidation of Cr overlayers.
Domenichini et al. have systematically studied the role of
the TiO2 bulk stoichiometry in the interface reaction between
Mo and TiO2 (110) [92]. In that work, three different TiO2 (110)
crystals were prepared. The first (#1 TiO2 ) was annealed at
925 K for 72 h in air. This crystal then presents the bulk and
surface stoichiometry of TiO2 . From the first crystal, a second
one (#2 TiO2 ) was further prepared by Ar+ sputtering (20 min,
3 keV). This procedure produces defects in the upmost several
atomic layers, as discussed above. Thus, the #2 TiO2 crystal is
stoichiometric in the bulk but its surface is reduced. The third
one (#3 TiO2 ) was subjected to cycles of Ar+ sputtering (20
min, 3 keV) and UHV annealing (873 K, 30 min). The #3 TiO2
crystal is both bulk reduced and slightly surface reduced. Again,
one can conclude that [e0 ] in #3 TiO2 crystal is higher than that
in either #1 or #2 TiO2 crystals.
To compare these substrates, a varying amount of
Mo (13 eq ML) was deposited on surface #1 through #3,
which were then stepwise annealed in UHV up to about 750 C.
XPS investigations show that the deposited Mo becomes
oxidized for temperatures greater than 400 C on the #1 and
#2 TiO2 crystals, but Mo/#3 TiO2 was still metallic at 750 C
(see Fig. 22). Based on these results, three conclusions can
be made: (1) Mo oxidation is not influenced by the substrate
surface defects; (2) Mo oxidation relies much more on the bulk
stoichiometry; and (3) Faster Mo oxidation occurs on a TiO2
crystal with a lower concentration of electron defects, [e0 ].
The reaction results taken from both Cr/TiO2 (110) and
Mo/TiO2 (110) interfaces are quite consistent with each other,
which suggest that metal on a TiO2 crystal with a high [e0 ]
is more resistant against the oxidation reaction. These results
further support the concept suggested in Section 2.4 and
Fig. 13. For n-type doped TiO2 (Case A in [89]) or the TiO2
crystal with bulk non-stoichiometry (#3 TiO2 in [92]), E F is
close to the CBM. In case of contact between Cr (or Mo) and the
TiO2 crystals, E F (Cr or Mo) < E F (TiO2 ) results in electron
transfer from TiO2 to the metal. Positive space charges form
at the TiO2 surfaces which hinder O2 outward diffusion to
the interface. Therefore, metal oxidation is kinetically limited
457

Table 3
Treatments for the six TiO2 crystals
Crystals
Surface preparation
Color n
Encapsulation
#A, undoped
#B, undoped
#C, doped
#D, doped
#E, undoped
#F, undoped
Standard procedure
Standard procedure plus UHV heating at 800 C for 9 h
Standard procedure
Standard procedure plus UHV heating at 800 C for 9 h
Standard procedure plus light sputtering (200 eV, 1 min)
Standard procedure plus heavy sputtering (1000 eV, 10 min)
Pale grey Low

Blue High
Dark blue High
Dark blue High
Pale grey Low
Light blue High
No
Yes
Yes
Yes
No
Yes
A qualitative evaluation of the electron density (n) in CB of TiO2 was performed considering the defect chemistry of TiO2 and the crystals color after treatment.The
intensity of this blue color corresponds to different n in the CB of TiO2 : a darker color, a higher n [321,322]. Results on the Pd encapsulation are also included.
After [13].
Fig. 22. Evolution of the BE of the Mo3d5/2 peak main component during
annealing under UHV for 3.3 eqML of molybdenum deposited on bulk
and surface stoichiometric TiO2 , #1 TiO2 , (); 3.3 eqML of molybdenum
deposited on stoichiometric bulk and surface reduced TiO2 , #2 TiO2 , (); 3.05
eqML of molybdenum deposited on bulk and surface slightly reduced TiO2 , #3
TiO2 , (). From [92].
even at high temperature at Cr/TiO2 (Case A) and Mo/#3 TiO2

interfaces.
4.1.2.3. Effect of TiO2 bulk defect on metal encapsulation on
TiO2 . The encapsulation reaction is another important reaction
occurring at metalTiO2 systems. Like the metal oxidation on
TiO2 , metal encapsulation shows a strong dependence on the
bulk defect of the TiO2 substrate.
Fu et al. have systematically investigated Pd interaction
on six different TiO2 (110) crystals [13]. Table 3 presents the
data of these experiments. Generally speaking, electron-rich,
or n-type doped, crystals were obtained by HTR in UHV, Nbdoping, or by strongly sputtering the surface. Pd clusters with
nominal coverage of 1.5 nm were deposited on the differently
treated TiO2 crystals at 200 C and all were stepwise heated
in UHV to 720 C. Characterization by XPS, AES, AFM, and
high resolution Rutherford backscattering spectroscopy (RBS)
confirms that the encapsulation of Pd clusters on TiO2 mainly
depends on [e0 ] in the TiO2 crystals. Pd encapsulation only
occurs on the electron-rich or n-type doped TiO2 crystals.
Bourgeois et al. reached a similar conclusion, which was
derived from XPS studies of the Ni/TiO2 (100) system [328].
They found that the covering-up of nickel by a thin layer of
TiO2 occurs more easily in the case of bulk non-stoichiometric
TiO2x crystals. The TiO2x substrates were prepared by HTR
in ArH2 mixture at 1000 K for 10 h and they are n-type

doped due to the introduction of oxygen vacancies. The ntype conductivity of the TiO2x substrates contributes to the
covering-up reaction of Ni atop.
Furthermore, Berko et al. studied the thermal behaviors
of Rh supported on three TiO2 (110) surfaces [326]. No
encapsulation of Rh particles was observed on a well-ordered
TiO2 (110) surface and a slightly Ar+ -sputtered TiO2 (110)
surface. On the other hand, clear evidence is found for
encapsulation of Rh crystallites supported on a strongly
Ar+ -sputtered surface. The heavy ion sputtering produces
subsurface non-stoichiometry which leads to the presence of
an electron-rich near-surface region that contributes to the
encapsulation of Rh particles.
Experiments from other groups also show that encapsulation
of some noble metals, e.g. Pt, Pd, and Rh, takes place on
strongly reduced TiO2 crystals which are subjected to HTR
but does not occur on bulk stoichiometric crystals [8,11,231,
259,329,330]. In these cases, it known that the HTR treatment
leads to n-type doping in TiO2 crystals, which favors the
encapsulation reactions.
Based on these results, we come to a general conclusion that
metal encapsulation reactions are only possible on TiO2 crystals
that are electron-rich in the bulk. The n-type conductivity of
TiO2 can be obtained by HTR, extrinsic doping via donors,
or heavy ion sputtering. These results are consistent with
the concept described in Section 2.4 and Fig. 14. Since the
encapsulation process requires the amplified outward diffusion
of Tiin+ to TiO2 surfaces and an electronic configuration of
E F (TiO2 ) > E F (metal), it can be understood that n-type TiO2
crystals and noble metals with large work function (e.g. Pt, Pd,
and Rh) are the prerequisites for the encapsulation reaction.
Interface contact between the metals and the n-type TiO2 results
in formation of positive space charges at the TiO2 surfaces
which drives the outward diffusion of Ti cations and, thus, the
encapsulation reaction (Fig. 14).
4.1.3. MetalTiO2 interactions
In this section, we consider the effect of metal overlayers
on metalTiO2 interactions, in which both the thermodynamic
and kinetic factors is discussed. The interaction data of various
metals on TiO2 (110) surfaces are to be revisited. Four different
metal interactions with the TiO2 (110) surface can be observed.
The metal work function is the critical parameter which
influences the metalTiO2 interactions.
458
experience encapsulation reactions [13,111]. The dependence

of the encapsulation reactions on the surface energy of
metals was illustrated by Fig. 23 which was reproduced
from [13].
Fig. 23. Relationship between surface energy and work function of

different transition metals. If available, of fcc(111) and bcc(100) surfaces
are displayed because of the orientation of the metals on TiO2 (110). Otherwise
values are taken from polycrystalline samples [331]. The surface energy
data were taken from calculations of Mezey and Giber [332]; these data are
qualitatively consistent with experimental data [333,334], providing larger
values for Pt, Pd, Rh, and Ir and smaller values for Au, Ag, and Cu. In region
I, encapsulation is expected ( > 5.3 eV and > 2 J m2 ) while in region II
( < 4.7 eV) oxidation of metals on TiO2 is possible. After [13].
4.1.3.1. Thermodynamic aspect. Using thermodynamics, it is

possible to predict if a reaction between a metal and TiO2 is
feasible. A good example for a typical reaction at metalTiO2
interfaces is the redox reaction. The corresponding reaction
equation can be written as follows:
M + TiO2 MOx + TiO2x .
(37)
As discussed in Section 2.3.2, the heats of formation of oxides

(1H of ) can be simply used to find if the reaction occurs.
Diebold has made a general conclusion that redox reactions
are favored by the condition of 1H of < 250 kJ/mol O. This
simple rule has been confirmed by many redox reaction results
at metal/TiO2 interfaces [28].
The thermodynamic criteria for encapsulation at metalTiO2
interfaces mainly rely on the metal surface energy, . It has
been shown that the metals with > 2 J m2 are likely to
4.1.3.2. Kinetic aspect. Reaction kinetics at metalTiO2

interfaces is determined by mass transport processes at TiO2
surface regions. In TiO2 , the mass transport involves the
diffusion of defects of Tiin+ (n 4) and/or VOx+ (x =
1 or 2). Depending on the interface reaction, one has to
consider the different defect diffusion processes. For example,
the oxidation of metals on TiO2 is controlled by the outward
diffusion of oxygen anions (O2 ), i.e., the diffusion of oxygen
vacancies (Vx+
O ) in the reverse direction, in the vicinity
of the interfaces. The outward diffusion of O2 has been
observed at many metal/TiO2 interfaces, e.g., Al/TiO2 [335],
V/TiO2 [336], Cr/TiO2 [284], and Mo/TiO2 [337]. The
encapsulation of metals on TiO2 is limited by the outward
diffusion of titanium interstitial cations Tiin+ from TiO2 bulk
to the interface [13]. Many previous results have established
that Tiin+ ions possess a high diffusivity in TiO2 at elevated
temperatures (e.g. 500800 K) and the Ti ion diffusion is
critical to other TiO2 surface processes including oxygeninduced surface restructuring and bulk-assisted reoxidation on
TiO2 surfaces [301,329,338341].
Many experiments have shown that the metalTiO2
interface reactions are often thermally limited at relatively
low temperatures [18,20,29,90]. Therefore, it is important
to consider reaction kinetics. The generalized CabreraMott
theory suggests that the diffusion of ionic defects at TiO2
surfaces should be strongly coupled with the charge transfer at
metalTiO2 interfaces. We now discuss two different electronic
configurations at metalTiO2 interfaces: E F (metal) > E F
(TiO2 ) and E F (metal) < E F (TiO2 ). Different reaction kinetics
can be seen in the two cases.
In the case of E F (metal) > E F (TiO2 ), the interface contact
causes electron transfer from the metal to TiO2 , which results
in formation of negative space charges in TiO2 and a downward
bending of TiO2 bands. This case favors the outward diffusion
Fig. 24. Schematics showing the electronic interaction between metals and TiO2 . (a) E F (metal) > E F (TiO2 ) and downward bending after contact, favoring metal
oxidation; (b) E F (metal) < E F (TiO2 ) and upward bending after contact, favoring metal encapsulation.
Fig. 25. Dependence of the heat of formation of the metals oxide, 1Hof , on the
work function of the metals [331]. 1Hof was taken for the values of the most
stable oxide of the metals [331]. The solid horizontal depicts the borderline at
1H of = 250 kJ/(mol O). All metals can be classified into four regions I, II,
III, and IV (for details see text).
O2
of
in TiO2 and, thus, promotes the metal oxidation
(Figs. 13 and 24(a)).
The electronic configuration of E F (metal) < E F (TiO2 )
causes electron flow from TiO2 to the metal upon interface
formation. Positive space charges form at the TiO2 surface
and the TiO2 bands bend upwards, which drives the outward
diffusion of Tin+
i and favors metal encapsulation (Figs. 14 and
24(b)).
Therefore, it is expected that there is a strong dependence
of metalTiO2 reaction kinetics on the electronic structures
of the metals and TiO2 . In the following we present previous
experimental data to support the above working hypothesis.
4.1.3.3. MetalTiO2 interactions. Table 4 summarizes the
surface and interface studies of metals on TiO2 (110) which
have been investigated by PES, AES, TEM, and other
techniques. Metal oxidation is the reaction most commonly
observed at the metalTiO2 interfaces. The reaction strength
can be characterized experimentally by the thickness of the
metal layer which may undergo oxidation (see Table 4).
Electronic interactions, including the interface bonding and
space charge transfer, were studied primarily by PES. The
interface bonding results in reduction of surface Ti4+ or
oxidation of surface Ti3+ ions. The space charge transfer causes
band bending, 1(E F E V ) (eV), which was determined from
the BE shifts in PES spectra, such as Ti2p, O1s, O2p, etc. The
variation of surface work function, 1, reflects the relative
position of E F of the metal and TiO2 before contact between
the phases and was measured by UPS. All the electronic
interaction data including interface bonding, 1(E F E V ), and
1 are also listed in the table.
The metals can be classified into four groups according
to the different interactions. This is shown in Fig. 25, which
illustrates the category of the metals based on their work
functions. As shown below, each group will interact with
TiO2 (110) quite differently.
459
Fig. 26. Variation of , E bend , and h II (height of UPS peak around 1.0 eV
related with Na-adsorption), represented by full, half filled, and open circles,
respectively, with INa on TiO2 (110). The broken line indicates the completion
of the first layer. From [272].
(1) Alkali and alkaline earth metals (Cs, K, Na and Ca, Ba)
( < 3.0 eV, region I in Fig. 25).
These metals possess the highest reactivity to TiO2 (110)
surfaces concerning their relatively high oxygen affinities.
Within one monolayer coverage, fully ionized metals form
on TiO2 . Further exposure of the metals leads to formation
of multilayers of oxidized metals. This was observed by Lad
and Dake [342] where they reported that K2 O multilayers,
more than 9 monolayers thick, grew at 300 K by extracting
oxygen from the TiO2 bulk to the surface. The K2 O layers
are stable against annealing at 900 K. Similar K-adsorptioninduced O diffusion to the surface was reported by Heise
et al. [282]. For the Ca/TiO2 interface, XPS data showed that
3 nm thick Ca overlayers were oxidized during depositing Ca
on TiO2 surfaces [283]. For other alkali metals, oxidation was
also observed at multilayers coverage and the oxidation was
accompanied by a significant rearrangement of oxygen across
the interfaces [72,272,343].
Since the work function of these metals (<3 eV) is much
smaller than that of TiO2 (110) (ca. 5.2 eV [17]), electrons
flow from the metals to TiO2 after interface contact formation.
The transferred electrons may reside on surface Ti ions and
reduce them from Ti4+ to Ti3+ [272,282,342,343]. Some of
the electrons are delocalized into the CB of TiO2 within a
depth comparable to the Debye screening length L D [367]. The
accumulation of electrons in the TiO2 surface regions causes a
downward band bending as shown in Fig. 24(a) and leads to
an observed 1(E F E V ), ranging from 0.3 to 1 eV. In these
systems, the total decrease in the surface work function is in the
range of 2 to 4 eV, which is consistent with work function
differences between the metals and TiO2 . Fig. 26 displays
the variation of surface work function and band bending as
a function of Na coverage in the Na/TiO2 system [272]. It
can be seen that the band bending is as large as 1 eV with a
corresponding work function decrease of 3.4 eV. This effect has
also been seen in other alkali metalTiO2 systems [84,342,343].
The strong downward band bending and high density
negative space charges at the TiO2 surfaces accelerate O2
460
Table 4
Interaction of metal overlayers on TiO2 (110)
Band bending
1(E F E V )
(eV)
1max (eV)
Desorption
0.3 eV [84]
2.2 [84];
3.5 [72]
KOx is stable at 900 K
1 eV [342]
2.2 [342];
3.5 [282]
Na0+ & Na2 O: > 3 ML

Reduction of Ti4+ [272]
1 eV [272]
3.4 [272];
2.5 and
2.9 [275]
Ba [343]
Ba0+ & Bax+ : > 1 ML

Reduction of Ti4+
0.4 eV [343]
3.3 [343]
Ca [283]
CaO: > 3 nm
Reduction of Ti4+
Mg [344]
Formation of Mg+
Reduction of Ti4+
Al [335]
Al2 O3 : < 3 ML
Reduction of Ti4+
Oxidation of Al
Nb [29,90]
NbOx : < 2 ML
Reduction of Ti4+
Oxidation of Nb
Ti [345,346]
TiOx : < 2 ML
Reduction of Ti4+
Oxidation of Ti,
interdiffusion
V [29,347,348]
VOx at < 2 ML
Reduction of Ti4+
Oxidation of V
interdiffusion
(100)[010]V
k(110)[001]TiO2 [29]
Cr [88,284,349]
CrOx at submonolayer
Reduction of Ti4+
Oxidation of Cr,
interdiffusion
(100)[010]Cr
k(110)[001]TiO2 [88]; bcc
Cr(100) [349]
1.5 [284]
Fe [71,285,286,349,350]
FeOx at submonolayer
Reduction of Ti4+
Oxidation
bcc Fe(100) [349]
1.3 [284];
0.8 [350]
Mo [351,352]
Metallic Mo Reduction of
Ti4+
Oxidation
(100)[001]Mo
k(110)[001]TiO2 (reduced
crystal) [352]; bcc(110)
texture on oxidized
TiO2 (110) surface [351]
Ag [220,290,353]
Metallic Ag
No reduction of Ti4+
Sintering
Cu [29,238,350,354]
Metallic Cu
Sintering
(111) [110]Cu
k(110)[001]TiO2 [29]; fcc
Cu(111) [349]
Au [310,355,356]
Metallic Au
Sintering
(111) [110]Au
k(110)[001]TiO2 (RT);
(112) [110]Au
k(110)[001]TiO2 (775
K) [356]
Ni [183,227,328,357359]
Weak interaction
Ni+ [358]; Ni [227]
Sintering or
Encapsulation
(110)[110]Ni
k(110)[001]TiO2 [359]
Metals
Interaction@RT
Thermal stability
Cs [72,84]
CsOx & Cs:

1 ML < < 7 ML [84]
Reduction of Ti4+
K [282,342]
K2 O: > 9 ML [342]
Reduction of Ti4+
Na [272,275]
Epitaxy
Complete oxidation at
900 K
2.0 [344]
(100)[001]Nb
k(110)[001]TiO2
0.44 eV [345]
0.2 eV [347]
1.2 [347]
Downward
bending [290,
353]
0.5 [350]
0.10.15 eV [355];
0.2 [310]
0.5 [358]
461

Table 4 (continued)
1max (eV)
Metals
Interaction@RT
Thermal stability
Epitaxy
Band bending
1(E F E V )
(eV)
Rh [82,231,326]
Rh negatively charged
Oxidation of Ti3+ to Ti4+
Encapsulation
fcc Rh(111) [316]
0.8 [82]
Pd [13,330,360,361]
Metallic Pd
Encapsulation
(111)[121]Pd
k(110)[001]TiO2 [360]
0.3 eV [13]
0.5 [361]
Pt [11,83,259,294,
362365]
Metallic Pt
Encapsulation
fcc Pt(111) [11,259]
0.9 [83];
Upward bending
to form
Schottky
barrier [363,
365]
0.5 [294]
O2 [296,301,329,338,366]
Formation of TiO2 or
Tia Ob layers on
surfaces
0.8 eV [296]
1.1 [296]
: thickness of the metal layer which undergoes oxidation; 1(E F E V ), band bending: positive for the downward bending and negative for the upward bending;
1max shows the maximum change of surface work function during the adsorption of metals on TiO2 (110).
outward diffusion and promote oxidation at interfaces at room

temperature (see Figs. 13 and 24(a)). Therefore, adsorption
of the metals on TiO2 (110) surfaces results in growth of
multilayers (>3 ML) of metal oxides at room temperature.
(2) Early transition metals (Mo, Fe, Cr, V, Ti, Nb, and Hf)
and Al, 3.75 eV < < 5.0 eV (region II in Fig. 25).
These metals are thermodynamically favored to react with
TiO2 (1H of < 250 kJ/mol O). Redox reactions were often
observed right at the interfaces, i.e., reduction of surface Ti
and oxidation of the first metal layer. Further growth of metal
oxides necessitates the extraction of O from TiO2 . In many
cases the O diffusion and growth of oxidized metals is limited
at room temperature. The thickness of the oxidized metal layers
ranges from submonolayer coverage (Cr [284], Fe [71,285,
286,368], and Mo [351,352]), through 2 ML (Nb [29,90],
Ti [345,346], and V [29,347,348]), up to 3 ML (Hf [28]
and Al [335]). Mostefa-Sba et al. studied Fe deposition on
TiO2 (110) by XPS and AES. They found that the redox reaction
takes place only at the interface between the metal and the oxide
surface. i.e., during the completion of the first Fe layer or at
the periphery of the Fe islands [286]. The same conclusion was
reached by Nakajima et al. [368]. Marien et al. used HRTEM
to image the NbTiO2 (110) interface showing that only the
first two monolayers of Nb were oxidized at room temperature
(Fig. 8(a)) [90]. It should be mentioned that annealing these
interfaces at higher temperature in UHV may facilitate the
oxidation via bulk diffusion of oxygen in TiO2 [90,284,337]. At
some systems, interdiffusion occurs at elevated temperatures.
This further confirms that oxidation of the metals on TiO2
surfaces is thermally limited at room temperature due to the
low mobility of the diffusing atomic/ionic species.
The decrease in the surface work function upon adsorption
of these metals on TiO2 (110) surfaces was observed to be up
to 1.0 eV (Table 4). Such an electronic configuration with
E F (metal) > E F (TiO2 ) results in long-range charge flow from
metal bands to TiO2 bands and a downward band bending after
establishing interface contact (Fig. 24(a)). However, 1(E F

E V ) is observed to be below 0.5 eV, which is much weaker than
that at alkali metal/TiO2 interfaces. The local charge transfer is
facilitated by formation of interface bonding between metal and
surface oxygen. Fig. 27 shows XPS and UPS results recorded at
the V/TiO2 interface by Zhang and Henrich [347]. In the Ti2p
spectra (Fig. 27(a)), shoulder peaks at low BE appear after 2
ML V deposition. The change originated from a local electron
transfer from V atoms to surface Ti ions. On the other hand,
the long-range transfer of electrons from V to TiO2 bends the
TiO2 surface bands downwards such that the O2p VB spectrum
and O1s core level peak are shifted to higher BE by 0.2 eV
(Fig. 27(b)).
Table 4 shows that the charge transfer and downward band
bending at these metal/TiO2 interfaces is not as strong as that
in alkali metal/TiO2 systems. Therefore, it is well-expected that
the driving force behind the O2 diffusion is also smaller than
that at alkali metal/TiO2 interfaces such that oxidation of the
metals is generally limited within three ML.
(3) Mid-to-late transition and noble metals (Ag, Au, and Cu),
4.6 eV < < 5.4 eV (region III in Fig. 25).
The heats of oxide formation of these metals are above
the borderline of 250 kJ/mol O, and as a consequence,
neither metal oxidation nor TiO2 reduction was observed at
the metal/TiO2 interfaces over a wide range of temperatures.
In one example, the Ti2p spectra recorded from the TiO2 (110)
surface which was covered by Au [369] (Fig. 28) did not show
any changes in line shape of the spectra, which indicates that
no reactions occur at the Au/TiO2 interface. Due to the weak
interaction, the metal adatoms have quite high mobility on the
TiO2 surfaces, even at very low temperatures, and tend to form
3D clusters [220,238,239,354,369].
Work function of these metals is comparable to that of
TiO2 . The change in the surface work function during interface
formation is quite small, normally below 0.5 eV. There
is no significant band bending because of the similarities of
462
Fig. 27. (a) XPS spectra of Ti2p core levels for a stoichiometric TiO2 (110)
surface before (solid curve) and after (dashed curve) deposition of 2ML V. (b)
UPS spectra for a stoichiometric TiO2 (110) surface with different V coverage.
From [347].
Fig. 28. XPS spectra of Ti2p from TiO2 (110) surface covered by Au with a
(nominal thickness). From [369].
thickness from 1.1 to 52 A
E F (metal) and E F (TiO2 ). For example, upward bending of

0.2 eV was observed in case of 0.1 ML Au-adsorption on
TiO2 (110) [310] while 0.10.15 eV downward bending was
reported for Au on TiO2 with coverage less than 1 ML [355].
Due to the small charge transfer between the metal and TiO2 , it
was concluded that the interaction at the interfaces is very weak.
(4) VIII B metals (Pt, Pd, Rh, Ir, and Ni), > 5.4 eV (region
IV in Fig. 25).
There was no oxidation of the metals observed, even at
elevated temperatures, for these surfaces. Prolonged UHV
annealing at high temperatures, however, leads to the
encapsulation reaction. This was shown for Pt in [11,111,247,
259,362,363,370], Pd in [13,329,330,360], Rh in [82,231,316,

326,370], Ir in [184], and Ni in [183,328,357].
An increase in the surface work function during deposition
of the metals on TiO2 (110) was often observed, which indicates
an electronic configuration of E F (metal) < E F (TiO2 ) prior
to interfacial contact. For example, Negra et al. [361] and
Schierbaum et al. [83,294] reported a 0.5 eV increase in the
surface work function after Pd and Pt-adsorption on TiO2 (110)
or (100) surfaces. Because the metal Fermi level is lower than
that of TiO2 , contact between the metals and TiO2 causes
electron transfer from TiO2 to metals [13,82,231,294,326,361,
364] and may lead to the formation of a Schottky diode [83,
363,365]. This unique charge transfer process leads to upward
bending of TiO2 bands, formation of positive space charges at
TiO2 surfaces, oxidation of surface Ti, and negatively charged
metal clusters. For example, a detailed surface analysis of
Rh/TiO2 interface was performed by Sadeghi and Henrich [82].
They found that shoulder peaks originating from Tin+ (n <
4) on a sputtered TiO2 surface were largely weakened after
1.5 ML Rh deposition (Fig. 29(a)). Such a change originated
from a local electron transfer from Tin+ (n < 4) ions to Rh,
i.e., oxidation of surface Ti ions instead of surface Ti reduction
at V/TiO2 interfaces as shown in Fig. 27(a). The long-range
charge transfer from TiO2 to Rh resulted in an upward bending
of TiO2 bands, which was confirmed by the O2s shift to a lower
BE after 0.5 ML Rh deposition (Fig. 29(b)).
The unique electron transfer, which occurs from TiO2 to the
metals (Pt, Pd, Rh, etc.), results in formation of positive space
charges at TiO2 surfaces, which promotes the outward diffusion
of Tiin+ at the interfaces (see Figs. 14 and 24(b)). Therefore, it
can be concluded that encapsulation is uniquely observed for
metals with a large work functions (see Fig. 23) on TiO2 [13].
(5) Oxygen-induced restructuring and bulk-assisted reoxidation reactions on TiO2 surfaces.
One well-known TiO2 surface reaction is the so-called
oxygen-induced restructuring of TiO2 (110) surfaces. The
reaction describes the growth of surface phases when
heating TiO2 (110) surfaces in O2 environment. The dominant
mechanism of the reaction is the outward diffusion of Tiin+
ions from the bulk to surface, where they react with gaseous
oxygen and form phases like TiO2 or Tia Ob (e.g. Ti2 O3 ) [301,
329,338,366]. Due to the large electronegativity of oxygen it is
reasonable to suggest that the electronic energy level of O2p
is lower than E F of TiO2 (E O2p < E F (TiO2 )) when oxygen
is adsorbed on TiO2 surfaces. Thus, exposure of TiO2 to O2
leads to transfer of electrons from TiO2 to surface adsorbed
oxygen atoms. Such an electron transfer process causes upward
bending of the TiO2 surface energy bands and formation of
positive space charges at the TiO2 surface. Henrich et al. have
showed that exposure of reduced TiO2 surfaces to O2 at RT
results in 1(E F E V ) of 0.8 eV and 1 of 1.1 eV as shown
in Fig. 30 [296]. Therefore, the outward diffusion of Tiin+
towards the surface is driven by negatively charged oxygen
atoms on the TiO2 surfaces and the positive space charges at
the near-surface regions in TiO2 . The reaction process at the
O2 TiO2 interfaces can be well-compared to the encapsulation
at VIII B metals/TiO2 interfaces (see Figs. 14 and 24(b)). For
Fig. 29. (a) XPS spectra of Ti2p core levels for slightly reduced TiO2 (110)
before (solid curve) and after (dashed curve) deposition of 1.5 ML Rh. (b) UPS
spectra for slightly reduced TiO2 before (solid curve) and after (dashed curve)
deposition of about 0.5 ML Rh. Fom [82].
comparison, data concerning the O2 TiO2 interface reactions is

shown at the end of Table 4.
Another TiO2 surface reaction is the bulk-assisted reoxidation of reduced TiO2 surfaces in which the surface reduced
Ti ions are reoxidized upon heating a reduced TiO2 surface
in UHV [339,340]. The mechanism of the reaction is the outward diffusion of O2 or inward diffusion of Tiin+ . To drive
the O2 outward diffusion or Tiin+ inward diffusion, it is
clear that a downward band bending and formation of negative
space charges at TiO2 surfaces are necessary. Actually, electron accumulation at TiO2 surface and downward band bending of 0.20.3 eV have been observed on a reduced TiO2 surface, for example see Fig. 35 in Ref. [28]). The schematics
shown in Figs. 13 and 24(a) can also explain the bulk-assisted
reoxidation. Therefore, the oxygen-induced restructuring and
bulk-assisted reoxidation reactions are dominated by the same
concept.
In summary, we could show that the metalTiO2 interactions
depend strongly on of the metals. Redox reactions, observed
for metals with small work functions (e.g. Na, K, Al), show
a strong inverse correlation with work function: the smaller
the , the stronger the oxidation of the metal. On the other
hand, metals with very large work function (from Ni to Pt)
undergo encapsulation reactions on TiO2 (110). The interface
reactivity between metals and TiO2 can be categorized into the
four groups according to their work function values (Fig. 25).
463
Fig. 30. Work function change (1), Fermi level (E F E V ), position of

surface-state UPS peak (E S E V ), and normalized amplitude of the ELS
peak A versus normalized intensity of UPS surface-state peak ( S ) for Ar-ion
bombardment (solid points) and subsequent oxygen exposure (open points) of
the TiO2 (110) surface. Arrows indicate the sequence in which data were taken.
From [296].
Such a correlation cannot be drawn between the reactivity and

the thermodynamic data, e.g. 1H of (see Fig. 25). All the above
experimental results can be understood within the generalized
CabreraMott theory, as suggested in Section 2.4.
4.2. Metals on SrTiO3
SrTiO3 is a model material of the perovskite-type oxides
(ABO3 ). It has many technological applications, e.g., in
sensors, photocatalysts, and gate dielectrics. In addition,
SrTiO3 has been widely used as a substrate for epitaxial
growth of films of high-Tc superconductors, metals, and
oxides. SrTiO3 surfaces and metal interactions on these
surfaces were extensively studied. SrTiO3 surfaces, SrTiO3
bulk defect chemistry, and metal overlayers strongly affect the
metalSrTiO3 interactions.
4.2.1. SrTiO3 surfaces
The (100), (110), and (111) facets are the most commonly
studied SrTiO3 surfaces. Among them, the (110) and (111)
surfaces are polar, while the (100) surface is non-polar.
Most research focused on the more stable SrTiO3 (100)
surface. Important surface characteristics, including surface
terminations, surface defects, and surface reconstructions, are
464
addressed in the following, and the influence of these surface

properties on deposited metal overlayers are discussed in detail.
4.2.1.1. Surface terminations. A SrTiO3 crystal consists of
a stack of alternating TiO2 and SrO layers. Therefore, a
SrTiO3 (100) surface may have two distinct terminations: TiO2 termination and SrO termination. The surface termination
can be experimentally controlled. Kawasaki et al. [152] have
developed a wet chemical etching method to prepare the TiO2 terminated surface. It has been found that etching of prepolished SrTiO3 (100) substrates in NH4 F-buffered HF (BHF)
solutions selectively removes surface SrO groups as well as
calcium impurities leaving the surface terminated with Ti and
O [152,153,371]. The reproducibility of the surface etching
process has been improved through hydroxylation of the
topmost SrO layer, which enhances the etching-selectivity of
SrO relative to TiO2 in a BHF solution [91,372]. Atomically
flat surfaces with a TiO2 -termination can be obtained by the
chemical etching followed by heating in O2 and UHV. The
surface termination can be switched to the SrO-termination
by depositing a monolayer of SrO on the TiO2 -terminated
surface [251,373375].
EELS measurements indicate that the electronic structures
of the two differently terminated surfaces are different. TiO2 terminated surfaces show intrinsic surface states while SrOterminated surface does not exhibit such states [373]. It has
been seen that the photocatalytic activity on the SrO-terminated
surface is lower than that on the TiO2 -terminated surface [373,
375]. Furthermore, the surface termination has strong effects on
metal growth on the SrTiO3 (100) surface.
Polli et al. [251] have investigated Pt growth on the two
differently terminated SrTiO3 (100) surfaces. They showed that
the SrO-terminated surface has a mixture of Pt(100) and
Pt(111). Compare this to a TiO2 -terminated surface where
Pt islands form with a (100) orientation. The difference
between orientations was explained as being caused by the
weaker bonding between Pt and the SrO-terminated surface
as compared to Pt on the TiO2 -terminated surface. This
conclusion has been tested by a DFT calculation from Sholl
and coworker [376]. They found that Pt atoms prefer to
sit on top of O atoms on both surfaces, but the bonding
strength at Pt/TiO2 SrTiO3 interface is stronger than that at
Pt/SrOSrTiO3 interface in cases of a Pt coverage above 1
ML. A similar conclusion was made based on DFT studies
for Pd/SrTiO3 and Co/SrTiO3 interfaces. The TiO2 -terminated
substrate is energetically favorable for adhesion of Pd films
with Pd atoms bonded on top of O atoms [377,378]. At the
Co/SrTiO3 interface, the strongest cohesion between Co and
SrTiO3 is also found for the Co/TiO2 SrTiO3 interface where
the interfacial Co atoms sit on top of the interfacial O atoms.
Indirect electronic coupling was observed between interfacial
Ti and Co atoms mediated by the O atoms [379].
4.2.1.2. Surface point defects. Surface oxygen vacancies and
Ti3+ ions can be generated by ion bombardment, which
produces electronic states with predominant d electron
character in the band gap [380]. The gap states are located
several tenths of an eV below the CBM [380382]. Compared

to stoichiometric SrTiO3 (100) surfaces, defective SrTiO3 (100)
surfaces exhibit a higher reactivity to gases such as NO [382,
383], methanol, [384], and acetaldehyde [385]. It has been
demonstrated that the surface defects influence the metal
interaction as well. Conard et al. [386] studied Cu growth on
both stoichiometric and reduced SrTiO3 (100). They found that
the number of Cu islands is higher on the reduced surface than
on the stoichiometric surface. This was attributed to a stronger
interaction of Cu with the defective surface. We applied in situ
AES to monitor Cr growth on different SrTiO3 (100) surfaces.
The decay of substrate signals and increase of Cr signals for Cr
on Ar+ -sputtered SrTiO3 (100) are much faster than those on
stoichiometric SrTiO3 (100), which indicates a more layer-like
growth and, thus, stronger interaction of Cr on the reduced and
highly defective surface [387].
4.2.1.3. Surface reconstructions. Many reconstructions of the
SrTiO3 (100) surface have been experimentally observed. They
include 2 1 [145,388395], 2 2 [142,381,388,396
398], c(4 2) [142144,146,390,393395,399,400],
c(4
(
5
5)
4) [142,390,398], c(6 2) [143,144,389,395],
[141,142,401,402], c(13 13) [403], and (13
R
26.6
13) R 33.7 [142,389]. A summary of SrTiO3 (100) surface

reconstructions is given in Table 5. It can be seen that the
reconstruction depends sensitively on the annealing conditions,
e.g., atmosphere (oxidation or reduction), temperature, and
time, as well as surface conditions prior annealing. As one
example, Jiang and Zegenhagen [144] found that a c(4
2) surface is obtained by annealing a pristine surface with
assistance of hydrogen but a c(6 2) reconstruction is observed
upon heating O2 annealed surfaces. Castell [390,399] reported
that the creation of 2 1 and c(4 4) was achieved through
BHF etching and UHV annealing while a c(4 2) surface was
found by Ar+ ion sputtering and subsequent annealing in UHV.
Two kinds of surface models have been suggested to explain
the SrTiO3 surface reconstructions, Ti-rich surface models
and Sr-rich surface models. Oxygen vacancies on the topmost
TiO2 layer can be produced by vacuum annealing. Therefore,
the surface superstructure is attributed to an ordering of O
vacancies [144,381,401,402]. Formation of TiOx (x 2)-type
surface phases was suggested to explain some experimental
results, e.g., a Ti2 O3 surface phase for the (2 1) surface [390]
and a TiO2 overlayer atop a bulk-like TiO2 layer for (2 1) and
c(4 2) structures [145,146]. In contrast to the Ti-rich surface
models, Kubo et al. [141,142] found that a structural model
consisting of an ordered Sr adatom at the O four-fold site of
a TiO2 -terminated layer can explain previous experimental and
theoretical results. Liang and Bonnell [404] also explained the
intergrowth of a SrO-rich surface phase on SrTiO3 (100) with a
lamellae structure (Srn+1 Tin O3n+1 ).
In brief, the surface reconstruction is closely related
to the change in surface composition. Explanation of the
reconstructions can be complicated by a number of factors
which include surface non-stoichiometry, presence of TiO2
and SrO terminations, diffusion of defects between bulk and
surface, and surface segregation of impurities. In order to
465

Table 5
Surface reconstructions on SrTiO3 (100)
Reconstructions
Sample preparations
21
UHV, 800900 C, 20120 min
LEED, STM [388]
Vacuum, 750800 C, 1 h
RHEED [389]
UHV, 600800 C, 30 min9 h
STM [390392]
O2 , 10501100 C, 0.55 h, after ion milling
TEM [145,395]
UHV, 10001200 C, 220 min
LEED, STM [388]
Vacuum, T < 700 C

UHV, 650 & 730 C
LEED, AES [396]
UHV, 800 C, 5 h, after Ar+ sputtering
STM [397,398]
UHV, 10001250 C
STM [142]
H2 , 950 C, 2 h
LEED, STM, AES [143,144]
UHV, 9001150 C, only after sputtering
STM [390,393,394,398400]
22
c(4 2)
c(4 4)
c(6 2)
( 5 5)-R26.6
c( 13 13)
( 13 13)-R33.7
Techniques
RHEED [381]
UHV, 10001250 C
STM [142]
O2 , 830930 C, after ion milling
TEM [146,395]
UHV, 950 C, 2 h, only after O2 heating
LEED, STM, AES [143,144]
O2 , 8001000 C, 15 h
RHEED [389]
UHV, up to 1200 C, some minutes
STM [402]
UHV, 1200 C
STM [141,142]
O2 , 10501100 C, after ion milling
TEM [395]
O2 , 8001000 C, 15 h
RHEED [389]
UHV, 1200 C, several minutes
STM [142,401]
UHV, 950 C, 2 h
STM [403]
O2 , 8001000 C, 15 h
UHV, 1250 C
RHEED [389]
clarify the atomic structure of all the different SrTiO3 surface

reconstructions further detailed studies are necessary.
The reconstructed SrTiO3 surfaces are significant in
technological applications of SrTiO3 , e.g., in the fields of film
growth, catalysis, and other surface chemical reactions. For
example, Silly and Castell [393,394] performed an elegant
experiment of metal growth on SrTiO3 (100) surfaces where
they demonstrated that the shape and orientation of supported
metal nanocrystals can be selected by controlling the oxide
substrate reconstruction. On a SrTiO3 (100)-(2 1) surface Pd
forms hut-shaped nanocrystals; while on a SrTiO3 (100)-c(4
2) surface hexagonal and truncated pyramid nanocrystals are
observed (see Fig. 31). It was found that the two reconstructed
SrTiO3 (100) surfaces have different surface energies and the
interface energies between the metal and the substrates varies
with the substrate crystallography. As a consequence, the
supported nanocrystals can have different equilibrium shapes
and orientations.
4.2.2. SrTiO3 bulk defect chemistry
Like TiO2 , chemical reactions occurring on SrTiO3 surfaces
are closely related with SrTiO3 bulk defects. In this section, the
defect chemistry of SrTiO3 will be discussed first. Then, we
will show how the reactions on SrTiO3 surfaces are dependent
on the defect chemistry of SrTiO3 . The reactions include
STM [142]
SrTiO3 surface reactions upon high temperature annealing and

metalSrTiO3 interface reactions.
4.2.2.1. Defect chemistry of SrTiO3 . SrTiO3 is the prototype
of mixed conductors whose defect chemistry has been studied
in detail. Schottky-type defects are dominant in SrTiO3 solids,
which include oxygen vacancies, VO , and strontium vacancies,
VSr00 [405]. Assuming two-fold ionization of the O vacancies, the
defect reaction of O vacancies (VO ), free electrons (e0 ), and
lattice oxygen sites (OXO ) is described by
OOX VO + 2e0 + 1/2O2 .
(38)
The defect reaction of lattice Sr sites (SrXSr ) and OOX can be

written as
X
OOX + SrSr
VSr00 + VO + SrORP-phase .
(39)
The formation of VO and VSr00 has been suggested to be

accompanied by growth of RuddlesdenPopper (RP) phases
[SrO (SrTiO3 )n ] [406] in the solids which act as a sink for the
excess Sr. However, direct experimental proof of an RP phase
formation in the SrTiO3 solid has not been given as yet. Instead,
recent results indicate that Sr vacancies can be created only at
the crystal surfaces and all excess Sr migrates to the surfaces
where secondary SrOx phases grow on top of the surfaces of
466
Fig. 31. (a) STM image (140 140 nm2 ) of ca. two monolayers Pd deposited onto SrTiO3 (100)-(2 1) at room temperature followed by 650 C anneals showing
hut-shaped nanocrystals; (b) STM image (140 140 nm2 ) of Pd deposition of around one monolayer onto SrTiO3 (100)-c(4 2) at 175 C followed by 650 C
anneals showing two kinds of nanocrystal shapes, hexagons and pyramids. From [394].
the single crystals [407409]. The defect reaction of the cation

sub-lattice can be written as
X
SrSr
+ 2e0 + 1/2O2 VSr00 + SrOsecond-phase .
(40)
For doped SrTiO3 , the following equilibria may exist in the

solids:
X
LaSr
LaSr + e0 ,
X
NbTi
NbTi
+e,
0
for donor of La on Sr sites,

for donor of Nb on Ti sites,
and
X
FeTi
Fe0T i + h ,
for acceptor of Fe on Ti sites.
(41)
The above defect reactions suggest that the concentrations of

VO , VSr00 , e0 , and hole (h ) are a function of temperature, O2
partial pressure (pO2 ), and extrinsic dopant concentrations.
Qualitatively speaking, annealing in an oxygen-deficient
atmosphere (e.g. UHV) increases VO and e0 concentrations
[VO ] and [e0 ] while heating in an oxygen-rich environment

increases VSr00 and h concentrations [VSr00 ] and [h ] .
In SrTiO3 , all cations move via Sr vacancies and anions
diffuse via O vacancies. It is found that the diffusivity of Ti
is much lower compared to that of Sr and the Sr diffusivity
is orders of magnitude lower than the diffusivity of oxygen
vacancies [409,410].
In the following, we show that SrTiO3 defect chemistry
plays an important role in SrTiO3 surface reactions and in
metalSrTiO3 interface reactions.
4.2.2.2. Effect of SrTiO3 bulk defects on SrTiO3 surface
reactions. Surface phases often form on SrTiO3 surfaces
when heating the crystals at elevated temperatures. Generally,
annealing in an oxidizing atmosphere produces SrO-rich
surface phases (Eq. (40)) while heating in a reducing
environment results in growth of TiOx -rich surface phases (Eq.
(38)). The SrTiO3 surface reactions are similar to the surface
restructuring of TiO2 surfaces. As discussed in Section 4.1.3.3,
oxidation of reduced TiO2 crystals drives the outward diffusion

of titanium interstitial ions to the surfaces, where they react with
gaseous oxygen and additional TiO2 or Tia Ob structures form
on the surfaces [301,329,338,366]. The formation of surface
phases on SrTiO3 surfaces is also related to defect chemistry
in SrTiO3 . In particular, the production and diffusion of Sr
vacancies in SrTiO3 dominate the SrTiO3 surface reactions
occurring at high temperatures in both oxidizing and reducing
atmosphere.
Surface reactions in oxidizing atmosphere: According to
the defect reactions in SrTiO
3 , heating SrTiO3 in oxidizing
atmosphere increases VSr00 at the SrTiO3 surface (Eq. (40)).

The gradient in VSr00 drives the inward diffusion of VSr00 . This
means that Sr ions will diffuse outwards from the bulk to
the surface, where they can react with gaseous oxygen and
form SrO-rich surface phases. Szot et al. have observed a
continuous accumulation of SrOx on SrTiO3 (100) surfaces
in the case of oxidizing conditions [411,412]. Sr surface
segregation and growth of SrOx secondary phase on SrTiO3
surfaces under oxidizing conditions have been also confirmed
by many groups [407,413418]. The amount of the formed
surface SrOx layers increases with the doping level of La or
Nb in SrTiO3 since the extrinsic donors can create more Sr
vacancies and enhance Sr diffusion in the solids [413415].
Rahmati et al. also showed that the density of surface SrOrich islands depends on SrTiO3 surface orientations and island
density decreases in the sequence of (100), (110), and (111).
The result was attributed to that the Sr segregation rate is
related with SrTiO3 surface energy which shows the order of
(100) < (110) < (111) [416,417].
Surface reactions in reducing atmosphere: In reducing
conditions, oxygen may desorb from the surface (see Eq. (38)),
leading to an increase in [VO ] at the surface. According
to Schottky equilibrium, [VSr00 ] subsequently decreases. The
gradient in Sr drives inward diffusion of Sr ions from the
surface to the bulk. This results in Ti enrichment as well as
O- and Sr-deficiency at the surface. It has been demonstrated
467
that reducing SrTiO3 surfaces in vacuum above 900 C causes

formation of various Ti-rich surface phases, including Ti-rich
phases of TiO and Ti2 O [411], TiO2 and Magnelli-type Ti-rich
phases [419], Tix Oy Magneli phases [413], Ti2 O3 islands [420],
anatase TiO2 (100) islands [421], and TiO islands [422]. An
exception is the result reported by Liang and Bonnell [404].
They found that SrO-rich islands formed on SrTiO3 (100) after
annealing the surface in UHV at 1300 C.
4.2.2.3. Effect of SrTiO3 bulk defects on metal oxidation on
SrTiO3 . In order to study the influence of SrTiO3 defect
chemistry on metalSrTiO3 interface reactions, Cr/SrTiO3
interfaces were used as model systems and the reactivity of the
Cr/SrTiO3 interface was studied on various SrTiO3 crystals [85,
89]. Five SrTiO3 (100) crystals were prepared in different ways.
SrTiO3 #1: undoped and strongly reduced; SrTiO3 #2: Nb-doped
(n-type doping) and normally reduced; SrTiO3 #3: undoped and
normally reduced; SrTiO3 #4: Fe-doped (p-type doping) and
normally reduced; SrTiO3 #5: undoped and oxidized. According
to the SrTiO3 defect chemistry discussed in Section 4.2.2.1
and the defect model suggested

by Moos and Hardtl [405],
the defect concentrations VO , [VSr00 ], [e0 ], and [h ], can be
calculated. The data is shown in Fig. 32. It can be seen that e0
in the five crystals decreased in the order #1 to #5 whereas [VO ]
decreased in the sequence #1, #4, #3, #5, #2. The decreasing [e0 ]
in the crystals in the order #1 to #5 results in a shift of E F from
CB to VB, which transforms the n-type doping of #1 to p-type
doping of #5.
The thermal stability of Cr clusters (nominal film thickness
on the five SrTiO3 (100) crystals was studied under
6 A)
identical experimental conditions. XPS was used to monitor Cr
oxidation on the SrTiO3 surfaces and determine the reaction
onset temperature (TRO , the temperature point where oxidation
of Cr is first observed). We found that Cr on SrTiO3 #1 was not
oxidized below 810 C (TRO > 810 C) while Cr oxidation
occurred on SrTiO3 #5 as low as 280 C. The TRO of Cr
oxidation on SrTiO3 #2, SrTiO3 #3, and SrTiO3 #4 were 760 C,
640 C, and 460 C, respectively. The large difference in TRO
reflects significant variation in the Cr oxidation rate on SrTiO3 ,
which decreases in the order Cr/SrTiO3 #5 to Cr/SrTiO3 #1.
0
Plotting the data
increase
of
TRO with [e ] reveals a systematic
0
of TRO with e (Fig. 32(a)). A smaller [e0 ] in SrTiO3
corresponds to faster oxidation of Cr on SrTiO3 . In contrast,
we did not find such a systematic dependence between TRO and
[VO ] (Fig. 32(b)).
It has been shown that oxidation of metals on SrTiO3 (100)
necessitates outward diffusion of oxygen ions at the SrTiO3
surfaces [91,237]. If the oxygen diffusion is controlled by
Ficks law (Eq. (20)) the oxygen transport in SrTiO3 and
the rate of Cr oxidation on SrTiO3 (100) should be strongly
related with [VO ]. However, this work demonstrated that the
Cr oxidation rate was controlled by [e0 ] rather than [VO ] as
shown in Fig. 32. This behavior could be explained by assuming
that at relatively low temperatures (<900 C) the Cr oxidation
on SrTiO3 (100) is determined by the interface electric field
and [e0 ] in SrTiO3 . In the case of interface contact between Cr
and the electron-rich SrTiO3 , e.g. SrTiO3 #1 and SrTiO3 #2,
Fig. 32. TRO (the temperature point where oxidation of Cr is first observed
by XPS) for Cr oxidation on STO as a function of [e0 ] (a) and as a function
of [VO ] (b) in the STO crystals #1 to #5. Reproduced from [89]. The
defect concentrations were calculated according to the defect model suggested
in [405].
the interfacial electronic configuration is similar to that shown

in Fig. 14. In that case, the positive space charge hinders the
oxygen outward diffusion and Cr oxidation. On the other hand,
the contact of Cr with the p-type doped SrTiO3 , e.g. SrTiO3
#5, leads to the interfacial electronic configuration shown in
Fig. 13. Then, the oxygen outward diffusion and Cr oxidation
are promoted by a negative space charge (compare Fig. 9 in
Ref. [89]). Such a systematic investigation of Cr oxidation
on SrTiO3 (100) clearly shows that the metal/oxide interface
reactions are closely related to the defect chemistry in oxides,
and the reactions are tunable by variation in the electronic
structure of the oxides. These interfacial reactions can be
explained in the framework of the Generalized CabreraMott
theory (see Chapter 2.4.3).
4.2.3. MetalSrTiO3 interactions
Various metal overlayers have been grown on SrTiO3
surfaces. The main results of the metal interactions on
SrTiO3 (100) surfaces are listed in Table 6. Here, we focus a few
important characteristics of metalSrTiO3 interfaces, including
the interface electronic structure, growth and epitaxy of the
metal overlayers, and interface reactions.
468
Table 6
Summary of metal interactions on SrTiO3 (100) surfaces. OR I, OR II, OR III, OR IV, and OR V are different orientations of metals on SrTiO3 (100) (see the text)
Metal/SrTiO3 systems
Interaction
Y/SrTiO3 (100) [423]
Oxidation of Y
Y, Ba, Ti, /SrTiO3

(100) [424]
Oxidation of metals; reduction of Ti4+

to Ti3+ and Ti2+
Al/SrTiO3 (100) [89,108]
Reaction
Growth and epitaxy

Layer growth
Oxidation of metal overlayers at RT and

500 C; strong outward diffusion of O
Oxidation at RT
OR I
Cr/SrTiO3 (100) [89,91,177,

218,237,258]
Formation of interface bonding

between Cr and O; SrTiO3 band
bending after Cr deposition;
Oxidation at elevated temperatures; Reaction

temperature depends on SrTiO3 -doping and
surface orientation
Island growth; 100600 C: OR II;

>700 C: OR II & OR III
Mo/SrTiO3 (100) [89,263,

425,426]
Formation of MoO bonds at interfaces
Oxidation of Mo above 900 C
Metastable fcc Mo at low coverage;

OR II at low T ; OR III & OR IV at
higher T
Oxidation of Fe above 800 C
TiO2 -terminated: OR II; TiO2 - and

SrO-terminated surface: major OR II
& minor OR IV
Fe/SrTiO3 (100) [89,427]
Fe/SrTiO3 (100)c(4 2) [400]
Truncated pyramid islands with OR II
Co/SrTiO3 (100)(2 2) [397]
Truncated pyramid islands with fcc

OR I
Ni/SrTiO3 (100) [108,258,

428431]
Ni bonds with outmost surface O to

form 2D NiO; work function increases
by 1.3 eV after Ni deposition
Cu/SrTiO3 (100) [386,424,

432]
Cu bonds to O: strong interaction
Simultaneous Multilayer (SM)

growth [428]; OR I [108,258]
No reaction but sintering of Cu during

annealing at 500 C
Ag/SrTiO3 (100)(1 2) [392]

Au/SrTiO3 (100)(1 2) [391]
OR I at 100 C; Cu(100) and

Cu(111) with CuOx interface layer
Icosahedral nanocrystals
Wetting at submonolayer Au;

Dewetting as annealing
Truncated triangle island with small

amount of icosahedral islands; ORV
600 C: OR I T < 250 C: {111}
with ORV
Pd/SrTiO3 (100) [108,433,

434]
Pd/nanoline
structured SrTiO3 (100) [435]
Encapsulation of Pd with TiO1.36 or TiO1.37

layers
Pd/SrTiO3 (100)-(2 1) &

SrTiO3 (100)-c(4 2) [393,
394]
On SrTiO3 (100)-(2 1): hut islands

with OR III; On SrTiO3 (100)-(4 2)
@ RT: Hexagonal islands with OR V;
On SrTiO3 (100)-(4 2) @ 460 C:
truncated pyramid islands with OR I
Pt(621)/SrTiO3 (621) [436]
600 C: (621)[012]Pt
k(621)[012]SrTiO3
Pt(620)/SrTiO3 (620)
&(622) [437]
Pt adsorbs more at steps; interface

strength larger on low index surfaces
Pt/SrTiO3 (100) [86,221,

438]
Charge transfer from SrTiO3 to Pt:

0.6e/Pt atom; band bending of 0.6 eV
after Pt; SBH 0.6 eV
Pt/SrTiO3 (100) [439]
XPS, band bending of 0.6 eV
DFT: Possible step-flow growth from

DFT calculations
469

Table 6 (continued)
Metal/SrTiO3 systems
Interaction
Reaction
Growth and epitaxy
Pt/SrTiO3 (100) [251,440]
TiO2 -terminated surface: pyramidal

islands with OR I; SrO-terminated
surface: pyramidal islands with OR I
and hexagonal islands with OR V
Ir/SrTiO3 (100)
c( 13 13) [403]
Needle islands with OR I
Fig. 33. (a) Ti2p spectra from the ordered STO (100) surface as a function of Y coverage showing Ti reduction from 4+ to 3+ and 2+ nominal valences [424], and
(b) Ti2p spectra from the sputtered STO (100) surface with different coverages of Pt [221].
4.2.3.1. Electronic interaction of metals with SrTiO3 . The

adsorption of metals on SrTiO3 (100) surfaces may cause the
formation of interface bonding and charge transfer between
metal adatoms and SrTiO3 surface atoms. The local charge
transfer occurring via interfacial bonding depends on the
electronegativity of the metals. Reactive metals with small
electronegativity, such as Y, Ba, and Ti [424], K [441], and
Cr [91,218], could reduce the surface Ti4+ ions where charge
transfer occurs from the metal atoms to SrTiO3 . In contrast,
adsorption of noble metals with large electronegativity, e.g. Pt,
may result in electron transfer from SrTiO3 to the metal atoms
such that surface reduced Ti ions (e.g. Ti3+ or Ti2+ ) are
oxidized by the metals [86,221]. The two different charge
transfer processes were exemplified by the evolution of Ti2p
spectra from two SrTiO3 (100) surfaces which were covered by
Y and Pt, respectively (Fig. 33). The electron transfer from Y
to Ti resulted in an increase in the relative intensity of peaks
from reduced Ti while the intensity of the reduced Ti peak was
weakened by electron transfer from surface Ti to Pt upon Pt
deposition.
The space charge transfer between metals and SrTiO3 is

determined by the interface electronic configuration, i.e., E F
of the metal overlayers and E F of the SrTiO3 crystal before
contact. The work function of the clean and reduced SrTiO3
surfaces is around 4.2 eV [86,380,442]. Metals with surface
work function above 4.2 eV form Schottky-type contacts
with SrTiO3 single crystals. In case of interface contact,
electrons flow from the SrTiO3 CB to the metal VB which
produces positive space charges and upward band bending
at SrTiO3 surfaces. For example, Chung et al. [86] observed
a band bending of 0.6 eV on SrTiO3 (100) surface after
1 ML Pt deposition. Copel et al. have performed XPS and
UPS investigations of PtSrTiO3 (100) and AuSrTiO3 (100)
surfaces showing Pt-induced band bending of about 0.6 eV
and little band bending for Au metallizations [439]. Park
and Kim showed that metal/Nb-doped SrTiO3 (100) junction
with a low work function metal of Ti exhibited linear
currentvoltage (I V ) characteristics while for Ni and Pt
metals with large work function the junctions showed rectifying
I V characteristics and notable hysteresis due to the formation
470
Fig. 34. Plot of the lattice mismatch between the overlayer film and SrTiO3
substrates, | f | as a function of the oxygen affinity of the overlayer metal, pO
showing the range of epitaxy of metals. Region A: epitaxial growth of metals;
Region B: polycrystalline growth of metals; Region C: chemical reactions of
metals. From [108].
of Schottky barrier. The results clearly indicate that metal work

function influences the Schottky barrier height and the interface
resistance switching [443]. Shimuzu et al. have measured
the SBHs from various metal/SrTiO3 junctions and plotted
the SBHs as a function of Paulings electronegativity of the
metals. They show that the determined slope parameter S X is
about 1, which suggests that the interfacial electronic states
are not dominant in the interface charge transfer process at
metal/SrTiO3 interfaces (see discussion in Section 2.1.1) [438].
4.2.3.2. Growth and epitaxy of metals on SrTiO3 . SrTiO3 has
been often used as a substrate for epitaxial growth of metal
films. Here, we review the results of metal growth and epitaxy
on SrTiO3 . It is shown that the metal epitaxial growth is closely
dependent on the metalSrTiO3 interactions.
Growth: SrTiO3 (100) possesses a relatively low surface
energy about 0.9 J/m2 for the unreconstructed and TiO2 terminated surface ([444] and references therein). According to
expression (1), it is well-expected that most metals grow on the
SrTiO3 surfaces with the island growth mode (VolmerWeber
mode) [108]. Layer growth was observed only for a few cases,
Y growth on SrTiO3 (100) due to strong reaction between Y and
SrTiO3 [423] and the formation of wetted Au monolayer islands
on SrTiO3 (100)-(2 1) because of the commensurate epitaxy
at submonolayer coverage [391].
Metal islands on SrTiO3 surfaces tend to form in such a
manner that the total energy 1E is minimized
X
1E =
Me(hkl) AMe(hkl) + i Ai STO Ai .
(42)
(hkl)
Here, (hkl) Me(hkl) AMe(hkl) is the sum of the products of

surface energy of every metal facet, Me(hkl) and the metal
facet area, AMe(hkl) , i is the interface energy, STO is SrTiO3
surface energy, and Ai is interface area. Obviously, the shape
and orientation of metal islands are determined by Me(hkl) ,
STO , and particularly i . Wagner and coworkers [91,108,237,
251,263,425,433,445] have systematically studied the epitaxial
growth of various metals on SrTiO3 (100). They suggested that
P
i is closely related to lattice mismatch and interface reactivity.

The lattice mismatch is described by f = (aSub aMe )/aMe
(aMe : lattice parameter of metal; aSub : lattice parameter of
substrate) and the interface reactivity is characterized by the
oxygen affinity pO of the metals. i increases monotonically
with f 2 but decreases monotonically with pO. They found that
metal epitaxy can be achieved only in a certain range of | f |
and pO (see Fig. 34). For high | f | and low pO, polycrystalline
films may be observed. In case of a very high pO, a strong
chemical reaction between the metal and the substrate may
occur, which also prevents metal epitaxy [108]. Silly and
Castell have applied STM to image the growth of metal
nanocrystals on reconstructed SrTiO3 (100) surfaces [391394,
397,400]. They found that , which is the energy difference
between the interface energy and the substrate surface energy,
plays a critical role in the shape and orientation of nanocrystals.
The surface reconstruction leads to variation in SrTiO3 surface
energy and the interface energy and, thus, can be used to
manipulate the growth of metal nanocrystals [393,394].
Epitaxy: The most often observed epitaxial orientation
relationships for fcc and bcc metals on SrTiO3 (100) are OR I
and OR II, respectively (see Table 6):
OR I, cube-on-cube : (100)SrTiO3 k (100)Me ,
[001]SrTiO3 k [001]Me
OR II, cube-on-cube with 45 orientation :
(100)SrTiO3 k (100)Me , [001]SrTiO3 k [011]Me .
The OR I and OR II are mainly favored by their low interface
energies because the (100) surface of an fcc or bcc metal is not
the surface with the lowest surface energy. DFT calculations
have shown that Pt [376], Pd [377,378], and Co [379] atoms
tend to register on top of O atoms on the TiO2 -terminated
SrTiO3 (100) surface. Fig. 35 shows that OR I results in a
high density of near coincident sites for fcc metal atoms on
an O sub-lattice. Therefore, this orientation results in low
interface energies. The same argument applies for bcc metals on
SrTiO3 (100). Here, metals such as Cr and Mo [258,425,426],
try to bond with surface O. Formation of OR II is also promoted
by a high density of near coincident sites for bcc metal atoms
on O (Fig. 35). In addition, OR I and OR II are favored by the
small misfits.
As discussed above, the low interface energy in the cases
of OR I and OR II relies on the registry of metal adatoms on
the O sub-lattice on the TiO2 -terminated SrTiO3 (100) surface.
If SrTiO3 (100) is SrO-terminated, or there exists an interface
oxide layer between metal overlayers and the SrTiO3 surface,
the interface energy may be increased to the point that the (100)
orientations will not be dominant in the metal epitaxy. In such
a case, (110) orientations (OR III and OR IV) are observed for
bcc metals and (111) orientations (OR V) for fcc metals:
OR III : (100)SrTiO3 k (110)bcc-Me , [001]SrTiO3 k [001]bcc-Me ,
OR IV : (100)SrTiO3 k (110)bcc-Me , [110]SrTiO3 k [111]bcc-Me ,
OR V : (100)SrTiO3 k (111)fcc-Me , [001]SrTiO3 k [011]fcc-Me .
For example, growth of Cr on SrTiO3 (100) above 700 C
results in the formation of an interfacial CrOx layer, which
471
causes the appearance of OR II and (110) Cr with OR III [237].

A similar result was observed by Francis et al. [432]. They
found that purely epitaxial Cu(100) films were obtained on a
clean SrTiO3 (100) while a mixture of Cu(100) and Cu(111)
domains was observed in the presence of a small amount
of copper oxide at the interface. The presence of surface
terminated SrO layer could also lead to orientations other than
the (100) orientations. Ono et al. found that Fe grows on
the flat TiO2 -terminated SrTiO3 (100) with OR II. However,
both (100) Fe (OR II) and (110) Fe (OR IV) were observed
on the as-polished SrTiO3 surface, which consists of both
TiO2 - and SrO-terminations [427]. Polli et al. observed that Pt
presents OR I on purely TiO2 -terminated SrTiO3 (100) while
Pt(100) (OR I) and Pt(111) (OR V) have been recorded on the
SrO-terminated surface [251]. The epitaxial growth of metal
overlayers on SrTiO3 (100) can be controlled by modification
of the SrTiO3 (100) surface. This concept is applicable for most
oxide surfaces.
4.2.3.3. Chemical interaction of metals on SrTiO3 . The most
important interface reaction between metals and SrTiO3 is
metal oxidation. Metals of Y, Ba, Ti, Al, Si, Cr, Fe, and Mo,
can be oxidized on SrTiO3 surfaces [85,89,91,423,424]. A
simple thermodynamic criterion for metal oxidation on SrTiO3
surfaces is 1H of < 250 kJ/mol O. The heats of oxide
formation per mole of oxygen (1H of in kJ/mol O) were taken
from Ref. [23].
Hill et al. found that Y, Ba, and Ti reacted with SrTiO3 .
At 300 K, the reaction was diffusion limited. At elevated
temperatures, extended out-diffusion of O was activated and
the metal overlayers could be fully oxidized [424]. In order to
study the reaction kinetics in more detail, Fu et al. performed
a systematic study of the thermal stability of ultrathin films of
Al, Cr, Fe, and Mo on the SrTiO3 (100) surface [89]. Al, Cr,
were deposited onto
Fe, and Mo films of a thickness of 6 A
identical SrTiO3 (100) surfaces. The interface reactivity was
studied by stepwise UHV annealing combined with in situ XPS
measurements. The determined reaction onset temperatures
(TRO ) were plotted as a function of the metal work function M .
The figure is reproduced in Fig. 36 showing close correlation
with E F of metals (see Section 2.4.3.).
Encapsulation reactions were seldom observed for metals
on SrTiO3 surfaces. Silly and Castell recently reported the
encapsulation of Pd nanocrystals on SrTiO3 (100) [435] where
the SrTiO3 substrate has been subjected to extended UHV
annealing and the surface is covered by anatase TiO2 (100)
islands. On the strongly reduced SrTiO3 surface Ti and O
atoms are mobile enough at elevated temperatures such that Pd
clusters were encapsulated by TiO1.36 or TiO1.37 layers upon
UHV annealing. Therefore, the reaction can be regarded as Pd
encapsulation on TiO2 islands supported on the SrTiO3 (100)
surface.
5. Interaction of metals with insulating oxides
Al2 O3 , MgO, and SiO2 crystals are excellent insulators.
They have large band gap energies (E g ), i.e., E g (Al2 O3 ) =
Fig. 35. Schematics of the interface between a TiO2 -terminated SrTiO3 (100)
and a bcc metal overlayer (a) and the interface between a TiO2 -terminated
SrTiO3 (100) and an fcc metal overlayer (b) with OR I and OR II, respectively
(top view). From [108].
8.8 eV, E g (MgO) = 7.7 eV, and E g (SiO2 ) = 9 eV [17,

41]. Electrons in these oxides are strongly localized and
the production and diffusion of ionic defects in the oxides
are limited. Therefore, long-range charge transfer and ion
transport do not occur in these oxides at relatively low
temperature (e.g., <1000 C). Metal interactions with these
insulating oxides are often confined to the interfaces, which
only involve oxide surface atoms and metal adatoms in
contact with the substrate surface. The interactions are strongly
dependent on the surface properties of oxide substrates.
Surface stoichiometry, surface terminations, and surface defects
are the most important factors influencing the metaloxide
interactions.
In order to circumvent the charging problems encountered
on surfaces of single crystal insulating oxides, thin oxide
films have been grown on conductive supports which were
472
Fig. 36. Plot of TRO (the temperature point where oxidation of metal overlayer
is first observed by XPS) for Al, Cr, Fe, and Mo metal films on identical
SrTiO3 (100) and TiO2 (110) substrates as a function of M [89].
widely used as model systems [159]. For these systems of

metal overlayer/oxide film/conductive support, both the oxide
film surface and the support are critical in the metaloxide
interactions.
Interactions between metals and insulating oxides (Al2 O3 ,
MgO, and SiO2 ) are reviewed in this chapter. We show how
interactions vary with oxide surface properties and how the
supports of the thin oxide films influence the metal interactions
on the top of the oxide film surfaces.
5.1. Metals on Al2 O3
Alumina (Al2 O3 ) is widely used as a catalyst support,
for structural ceramics, and as a substrate for film growth.
Among the various polymorphs of alumina, corundum, Al2 O3 , is the most stable phase and has been subjected to
extensive studies [26,76]. In particular, the crystallographically
simple and energetically stable -Al2 O3 (0001) surface offers
a good playground for understanding of insulating oxide
surfaces and metalAl2 O3 interactions. In the following,
details of the surface properties of -Al2 O3 (0001), which
include surface terminations, surface reconstructions, and
surface hydroxylation, are given. The effects of these Al2 O3
surface properties and metal overlayers on the metalAl2 O3
interactions are discussed in detail. In addition to the single
crystal Al2 O3 surfaces, ultrathin alumina layers grown on
conductive substrates have been widely used as model systems
for supported metal overlayers. In the subsequent part, metal
interactions with alumina films are reviewed.
5.1.1. -Al2 O3 surfaces
Alumina surfaces are often prepared by mechanical
polishing, ion sputtering, and annealing. Naturally, different
surface preparation processes result in various surface
properties, which, in turn, may cause different behaviors
of metal growth on these surfaces. Here, we discuss the
surface terminations and reconstructions as well as the surface
hydroxylation of the -Al2 O3 surfaces.
5.1.1.1. Surface terminations and reconstructions. The unit

cell of bulk -Al2 O3 can be described as a hexagonal unit
cell containing six formula units of Al2 O3 [446]. This unit
cell consists of six close-packed hexagonal O layers. Al
layers, which are not coplanar but buckled, are placed between
these O layers. All of the ions are stacked along the c axis
of the unit cell in a sequence R-AlAlO3 -R (R: continuing
sequence in the bulk). For the bulk truncated and clean Al2 O3 (0001) surface there exist, from a geometrical standpoint,
three different terminations [25,156]: O layer termination
(O3 AlAl-R), single Al layer termination (AlO3 Al-R), and
double Al layer termination (AlAlO3 -R). Theoretical work
suggests that the 3 surfaces possess different thermodynamic
stabilities [156,447449]. The different surface terminations
have been experimentally observed by different groups
(Table 7).
O layer termination: The O layer terminated surface has
a large surface dipole moment and surface dangling bonds
which cause the surface to be energetically unstable under most
environmental conditions [156]. The O layer termination was
observed experimentally only by Toofan et al., who reported
a mixture of 2:1 O/Al-terminated surface domains [450].
Presence of H or adsorption of a reactive metal can stabilize
the O layer terminated surface.
Single Al layer termination: The single Al layer terminated
surface is generally accepted to be the most stable Al2 O3 (0001)-(1 1) surface [25,67,156,447453,468473]
because this surface is non-polar. The surface Al atoms can
relax inwards so that they are almost coplanar with respect to
the second O layer (see Fig. 37). The relaxation is accompanied
by a rehybridization of surface Al atoms to an sp2 orbital
configuration, which significantly stabilizes the surfaces via
charge autocompensation [468]. The surface relaxation was
calculated to be about 85% [67,156,448,469,471473]. The
experimentally observed relaxations were smaller, ranging
from 51% to 63% [25,451,452,474].
Double Al layer termination: The (1 1) surface may be
subjected to reconstructions in cases of surface O desorption
or Al deposition onto the surface [147150,466].
XRD and
LEED investigations have revealed a ( 31 31)R 9
reconstruction on -Al2 O3 (0001) surfaces which were heated
in UHV at T > 1200 C [148150] or covered by Al [150,
466]. The Al-rich reconstructed surface has been confirmed to
be terminated by a double Al layer, which contains surface
domains with a Al(111) structure separated by hexagonal
network of domain walls [147,148,448]. Such a structure has
been directly imaged using dynamic-mode SFM by Barth and
Reichling [147].
In addition, disordered Al-rich surface phases were obtained
through ion sputtering, which preferentially removes surface
oxygen [464,465,475].
Many research results have shown that the sapphire surface
terminations and reconstructions have a strong effect on the
nature of bonds at metal/Al2 O3 interfaces and the morphology
of the supported metal clusters.
An extensively investigated system is the Cu/-Al2 O3 (0001)
interface. Gautier et al. [476] found that the size of Cu clusters
473

Table 7
Selected experimental data showing the surface terminations of -Al2 O3 (0001)
Termination
Surface treatments
Surface structure
Techniques
O layer termination
Heating in O2 plasma
T < 750 C: (1 1); T < 1000 C :
( 3 3)-R30 ; T > 1100 C :
( 13 13)-R 9 ; Mixture of Oand Al-terminated domains
LEED [450]
Single Al layer termination
UHV annealing up to 1100 C;
(1 1); 63% inward relaxation of top

Al layer; H presence on surface even
after annealing at 1100 C
Al layer
Al layer
(1 1)
TOF-SARS, LEED [451]
Annealing in air at 1500 C, 3 h

Heating at 650 C using atomic
deuterium beam
Heating in O2 at 1570 K for 10 h
followed by UHV heating at 870 K
for 10 min
Double Al layer termination
UHV heating, O2 annealing, Al

deposition, or Si etching
UHV, 1350 C for 20 min

UHV, 1300 C
OH layer termination
LEED [452]
CAICISS, RHEED [453]
LEED [150]
GIXD [148,149]
SFM [147]
Exposure to 104 Pa H2 O
Formation of surface OH
HREELS, XPS [454,455]
Exposure to H2 O
(1 1); Chemisorption of H2 O
LITO, TPD, LEED [456,457]
Surface in ambient conditions
Fully hydrated surface with O

up disordered O
termination; 2.3 A
layer from adsorbed water
1 ML coverage of surface OH
CTR diffraction [458]
XPS [459]
0.5 0.1 ML coverage of surface OH
XPS [460]
Formation of surface OH
XPS [461463]
3 nm -Al2 O3 layer with high

density defects
Surface Al-rich phases; With
increasing T : change from (2 2),
(3 3 3 3)R30 , to
( 13 13)R 9
( 13 13)R 9 : Al-rich surface

phases
TEM [464]
Exposure of Al-terminated surface

to >1 Torr water followed by
oxygen plasma at RT
Exposure of clean surface to water
drops followed by oxygen plasma
at RT
Exposure to water vapor
Other surface phases
UHV, T > 1250 C : ( 13 13);

Excess Al deposition at 800 C:
( 13 13); O2 annealing at
10001200 C: (1 1); 13 13:

cubic layer with composition of Al2 O
or AlO
( 13 13)-R 9 : two Al planes

close to metallic Al(111)
( 13 13)-R 9 : hexagonal
(111) Al surface domains with
hexagonal domain walls
CTR diffraction [25]
1 keV Ar bombardment
UHV heating or ion sputtering
Al deposition
is larger on a reconstructed ( 31 31)R9 surface than that

on a (1 1) surface. The same conclusion was made by Gota
et al. [477] based on SEXAFS studies during the initial stage of
Cu growth on the two surfaces. It is concluded that the different
morphologies of Cu clusters on the two surfaces originate from
the different bonding strength of Cu with the Al2 O3 surface.
Stronger adhesion of Cu on the (1 1)Al2 O3 (0001) surface
favors the formation of higher density of Cu clusters atop.
LEED, XPS, EELS [465]
LEED, EELS, AES [466,467]
Scheu [475] has shown that bonding at a Cu/(0001)Al2 O3

interface depends on the substrate preparation and that the
bonding can be manipulated by different substrate cleaning
processes. As discussed above, the (11) Al2 O3 (0001) surface
is terminated by a single Al layer. On this surface strong inward
relaxation of the top Al layer results in that the second O
layer is almost coplanar with the topmost Al layer (Fig. 37).
With surface O available, Cu-adsorption onto the surface results
474
Fig. 37. Atomic layer sequence and layer spacings along the [0001] direction
for single Al-terminated -Al2 O3 surface model determined by Guenard
et al. [474]. Small grey balls represent Al, and big red balls represent O.
Note that Al and O are almost coplanar at the topmost surface. Reproduced
from [458].
in the formation of CuO bonds. A combined approach of

HRTEM and ELNES reveals the existence of CuO bonding at
the internal MBE-grown CuAl2 O3 interface. The bonding is
of mixed ioniccovalent character with substantial Cu3dO2p
hybridization similar to that in Cu2 O [257,475,478]. Cu-O bond
formation for copper growth on the (1 1) surfaces has also
been observed by Guo et al. [479] and Varma et al. [480]. In
contrast to the (1 1)Al2 O3 (0001) surface, the reconstructed
surfaces or Ar+ -sputtered surfaces are surface Al-rich. On these
surfaces, Cu atoms tend to bond with surface Al atoms, which
results in formation of interfacial metallic CuAl bonding.
Scheu has observed CuAl bonds at the Cu/(0001)Al2 O3
interface that had been Ar+ sputter cleaned [475] and Gota
et al. confirmed
that
are bonded to the Ar+ sputter
Cu clusters
cleaned ( 31 31)R 9 surface via Al atoms [477].

DFT calculations of Cu and Pd interactions with Al2 O3 (0001) demonstrate that the interaction mechanism
depends on the surface stoichiometry [481]. The interaction is
covalent-like for the metals on the Al-rich surface but ioniclike for the metals on the oxygen terminated surface. On the
stoichiometric surface the interaction is very weak and mainly
due to polarization effects. The first principle studies from
Zhang et al. [482,483] suggest that the interface energy of
metal/Al2 O3 is a function of the substrate surface stoichiometry
and oxygen chemical potential. The most stable structure of
Al/Al2 O3 and Nb/Al2 O3 interfaces is reached for metals on
the oxygen terminated surface as long as the oxygen chemical
potential is above a critical value. At the Ag/Al2 O3 interface, Al
termination results in the formation of a stable interface even at
relatively high oxygen potentials.
5.1.1.2. Surface hydroxylation. Many theoretical results show
that hydroxylation of the clean -Al2 O3 (0001) surfaces may
result in a further lowering of the energy of these surfaces [156,
448,449,484,485]. Thus, the above-mentioned O- and Alterminated surfaces are expected to be reactive to water. Ab
initio calculations revealed that molecularly adsorbed water
on Al-terminated surfaces is metastable and can dissociate
readily. The H2 O dissociative reactions produce two types
of surface OH groups: Oads H and Os H (Oads : water oxygen;
Os : surface oxygen) [471,472,486]. The OH-terminated Al2 O3 (0001) surfaces have been experimentally observed by
various techniques, e.g., SFM [147], XRD [458], thermal
desorption [456,457], EELS [454,487], and PES [455,459463,
488,489]. These OH-terminated surfaces were simply obtained
Fig. 38. O1s spectra measured at a 10 take-off angle for hydroxylated Al2 O3 surface (a), and the same after 0.8 ML Co deposition (b). From the
intensity ratio of OH to lattice oxygen O1s, the coverage of surface OH on the
hydroxylated surface was calculated to be 1 ML. Co deposition removed 0.4
ML surface OH groups. Reproduced from [459].
via exposure of the Al2 O3 surfaces to water or air, see Table 7.

The coverage of surface OH groups can be up to 1 ML.
For example, Fig. 38 shows that the OH coverage on a
hydroxylated Al2 O3 surface is about 1 ML [459]. The surface
OH groups are very sensitive to electron beam illumination,
Ar+ sputtering, UHV heating, and adsorption of reactive
metals. For example, exposure of a hydroxylated Al2 O3 surface
to a large electron beam dose or high energy ion sputtering
resulted in dehydroxylation of the surface [454,460,462,487].
High temperature annealing or Al deposition can transform a
hydroxylated surface to an Al-terminated surface [460,490].
Surface hydroxylation is of special importance for the Al2 O3 surfaces. Alumina surfaces are quite often covered by
water or exposed to air during handling. Therefore, OH groups
are always present on the surfaces in case of no further special
surface treatments. Furthermore, it is shown below that surface
OH groups exert a significant influence on the metal interaction.
Cu/-Al2 O3 (0001) interface is a well-studied system to
illustrate the effect of OH on the interface formation. Kelber
and coworkers applied XPS and DFT calculation to study Cu
interactions with hydroxylated -Al2 O3 (0001) surfaces [461,
462]. It was shown that the presence of surface OH leads to the
formation of a Cu (I) monolayer up to 1/3 ML coverage. In
agreement with this result, a recent DFT calculation indicates
that at Cu coverages below 1/3 ML Cu atom adsorbed onto a
OH-terminated Al2 O3 surface can remove the surface H and
bind the surface through the surface oxygen [491]. The strong
interaction between Cu and the hydroxylated surface enhances
the wetting of Cu which stabilizes Cu(I) adatoms in 2D islands.
The strong interaction may originate from an exothermic
reaction of surf-OH + Cu (gas) 12 H2 + surf-OCu [462,
491]. However, Wang et al. [492,493] and Lodziana et al. [481]
presented a quite different conclusion, suggesting weak

interaction between Cu and the surface OH. The interfacial
OH can be stable even in the presence of two ML Cu [492,
493]. Similar to the theoretical results, XPS experimental
data do not support any strong chemical reactions between
surface OH and Cu adatoms [460]. Lodziana et al. [481] have
suggested that surface defects which are introduced by the
hydroxylation, rather than surface OH groups, are responsible
for the experimentally observed enhancement of Cu interaction.
Experimental and theoretical work has shown that others
metals, such as Al [455,460,494], Ti [455,488], Rh [489], and
Co [459,460], can react with the surface OH to form oxidized
metal overlayers at submonolayer coverage. For example,
Chambers et al. [459] showed that Co deposition on fully
hydroxylated sapphire surfaces is accompanied by oxidation to
Co2+ and removal of OH (Fig. 38). The interface reactions lead,
instead of island growth, to technologically important 2D metal
films growth on oxides.
5.1.2. Metal interactions with bulk Al2 O3
Both electronic interactions and chemical reactions at
metalAl2 O3 interfaces are reviewed in detail here.
5.1.2.1. Electronic interaction of metals on Al2 O3 . The
interaction of metals with Al2 O3 surfaces presents a high
degree of complexity. Nevertheless, many theoretical and
experimental efforts have contributed to the understanding of
the nature of interactions between metals and Al2 O3 surfaces.
Different interaction mechanisms have been proposed, such as
van der Waals force, interfacial bonding (covalent or/and ionic),
and polarization. Both theoretical and experimental results are
discussed separately.
(1) Theoretical results: Johnson and Pepper [495] first
used a cluster model and molecular orbital theory to study
the metalsapphire interfacial strength. They concluded that a
direct chemical bond, which is mainly covalent, is established
between metal atoms (Fe, Ni, Cu and Ag) and oxygen anions.
The covalent interaction occurs through the hybridization of
metal d orbitals and non-bonding O2p orbitals which produce
bonding states and antibonding states. An ionic component
is associated with metal-to-oxygen charge transfer at the
interface. The increasing occupation of antibonding orbitals
and decreasing metal-to-oxygen charge transfer explain the
reduction in metalsapphire contact strength through the series
Fe, Ni, Cu and Ag. This picture was confirmed by Nath and
Anderson [496].
Alemany et al. [75] studied adhesion of 3d transition metals
on -Al2 O3 surfaces using extended Huckel tight-binding band
structure calculations. However, they found that two OM and
AlM interactions are responsible for the adhesion strength.
OM repulsive closed-shell interactions are a destabilizing
factor while AlM charge-transfer interactions favor interface
formation.
Verdozzi et al. [67] have investigated Pt- and Ag-adsorption
on Al-terminated -Al2 O3 (0001) using the local density
approximation (LDA) and thick slabs (up to 18 O layers) in
their calculations. The nature of the oxidemetal bond was
475
Fig. 39. Adsorbate-induced charge density difference plot for Nb, Ru, and
Pd. Solid lines indicate charge accumulation, and dashed lines depletion, with
logarithmic increments. The h100i cut goes through the center of the adsorbates
and oxygen ions. For metals to the right of Mo in the periodic table, e.g., Ru and
Pd, charge transfer occurs from d2Z orbitals to lateral d orbitals upon adsorption
on surface O. Metals to the left of Mo, e.g., Nb, present less than a halffull d shell, and the oxygen polarization goes from lateral d orbitals to d2Z
orbitals [66].
found to change drastically with the metal coverage. At 1

ML both Pt and Ag prefer atop-Al sites and no evidence of
significant charge transfer occurs between metal overlayers and
the oxide. The BE is dominated by the polarization effect. In
contrast to the case of 1 ML, isolated metal atoms at 1/3 ML
coverage bind the surface up to 5 times stronger through largely
ionic interactions. This result was generalized by Bogicevic
and Jennison in a study in adsorption of various metals (Li,
Al2 O3 K, Y, Nb, Ru, Pd, Pt, Cu, Ag, Au, Al) atop 5 A
films on Al(111) [66]. At 1/3 ML, the oxidemetal bond is
ionic, regardless of metal adsorbate. Metal atoms, attracted
to the O2 ions, prefer to bind in the three-fold hollow site.
The charge is transferred from metal atoms to the nearest
neighboring oxygen ions. The degree of ionicity depends on
the metal Pauling electronegativity and metal ion radius. At 1
ML coverage, the adhesion is almost purely electrostatic. The
metal overlayer is attracted to the O ions at the oxide surface by
lateral polarization as illustrated in Fig. 39.
It has been shown that the theoretical results of metal
interactions on alumina surfaces rely on the choice of
the surface termination (Al-terminated or O-terminated),
calculation models (cluster or slab), metal coverage (isolated
atom or metal film), and choice of computational method
and approximations used. Nevertheless, a qualitative image of
metalAl2 O3 interactions can be obtained based on the results
available. The interactions can be qualitatively interpreted on
the basis of the metals Pauling electronegativity.
The bonding between the metals with small Pauling
electronegativity and Al2 O3 is mainly ionic. As one example, at
Nb(111)/-Al2 O3 (0001) interface, strong ionic bonds form by
Nb4d O2p electron donation, which accounts for the high
adhesive strength of O-terminated interfaces [497]. Ab initio
calculations of bonding at Al(111)/-Al2 O3 (0001) interface
also indicate that AlO bonds constitute the primary interfacial
interaction. The bonds are very similar to the cationanion
bonds found in Al2 O3 bulk and are mainly ionic [473].
The interactions of the metals with large Pauling electronegativity with alumina surfaces are, however, dominated
by the polarization effect, in particular, on Al-terminated surfaces and at high coverage, e.g., 1 ML. For example, at Pd/-
476
Al2 O3 (0001) interfaces, the nature of the interaction has been

investigated by Gomes et al. [498,499]. They concluded that the
Pd-surface interaction comes from metal polarization, which
is caused by the surface electrostatic field. The same mechanism controls interactions of Pt and Au with -Al2 O3 (0001)
surfaces [66,67,500].
For metals with Pauling electronegativity in between e.g., Cu
and Ag, either charge transfer or metal polarization is claimed
to be responsible for the interactions [66,500,501].
(2) Experimental results: Adsorption of reactive metals onto
-Al2 O3 (0001) is often accompanied by oxidation of the metals
in the submonolayer coverage regime. Oxidation of Nb [502],
V [503], and Ti [504] on -Al2 O3 (0001) was observed using
PES. The formation of MO bonds is accompanied by charge
transfer from metal adatoms to the substrate surface. However,
reduction of surface Al3+ was not often observed. Biener
et al. [503] argued that a non-localized charge redistribution on
the substrate surface instead of the localized charge transfer to
the substrate atoms, as observed at the V/TiO2 interface [348]
seems to account for the oxidation of V atoms at the V/Al2 O3 (0001) interface.
In the case of non-reactive metals, such as Pd, on Al2 O3
surfaces charge transfer, if any, happens from the oxide to
the metals. For example, Ogawa and Ichikawa applied the
Kelvin probe technique to monitor changes in surface work
function during Pd clusters growth on -Al2 O3 (0001). Their
results confirmed that the charge transfer occurs from Al2 O3
to Pd at low Pd coverage [505]. Gillet and coworkers studied
Pd deposition on -Al2 O3 (1012). Analysis of the modified
Auger parameter of Pd indicates that formation of interfacial
PdAl bonding promotes an electronic transfer from the oxide
to Pd [506508].
In conclusion, both theoretical and experimental results
demonstrate that the electronic interactions between metals and
Al2 O3 surfaces are dominated by either interfacial bonding or
metal polarization effect. The local electronic interactions are
strongly dependent on metal electronegativity. However, longrange interaction, such as space charge transfer, which was
observed at metal/TiO2 and metal/SrTiO3 interfaces, has been
rarely observed.
5.1.2.2. Chemical interaction of metals on Al2 O3 crystals.
Various surface science studies have confirmed that oxidation
reactions can happen between metals and -Al2 O3 (0001)
surfaces near room temperature. The metals include Al [455,
460,466,467], Ti [232,455,488,504], Nb [502], V [503], and
Cu [462,480]. Almost exclusively, the oxidation reactions
are strictly limited to the interfaces. Subsequently, metallic
overlayers will develop with metal coverage above 1 ML.
HRTEM was applied to investigate the interfaces of Al/Al2 O3 [509], Ti/-Al2 O3 [510], Cr/-Al2 O3 [88,511], and
Cu/-Al2 O3 [257,510], which were all formed below 600 C.
The results clearly show that all the interfaces are atomically
sharp and no interface reaction phases thicker than a monolayer
have been observed.
The results indicate that only those metal adatoms right
at the interface, which are in contact with surface oxygen,
became oxidized. Alumina has a high thermodynamic stability

(1Hof 600 kJ mol1 O) and oxygen diffusion in the crystal
is highly limited. The oxidation of metal multilayers may be
either thermodynamically impossible or/and kinetically limited.
5.1.3. Metal interactions with alumina films
The main problem of Al2 O3 single crystals in surface studies
is the surface charging effect, which prevents the use of surface
techniques that involve the emission or adsorption of charged
particles. In order to circumvent this drawback ultrathin
alumina layers have been grown on conductive substrates. The
oxide film-based model systems provide a good playground for
surface analysis of alumina surfaces and studies in interactions
between metals and alumina films.
5.1.3.1. Growth of alumina films. Well-ordered alumina thin
films have been grown on different Ni/Al alloy surfaces.
For example, epitaxial Al2 O3 -films on NiAl(110) show a
high degree of crystallinity, very low surface roughness,
and good preparation reproducibility [173,512,513]. Thus,
the Al2 O3 /NiAl(110) system has been extensively used as
model oxide surfaces [26,27]. Ultrathin alumina films can be
obtained by oxidizing Ni3 Al(111) [514,515] and NiAl(111)
surfaces [516]. Oxidation of polycrystalline Al foils or single
crystal Al surfaces also produces alumina films [171,172,517,
518].
5.1.3.2. Electronic interaction of metals on alumina films. In
the systems of metal-overlayer/oxide-film/conductive-support,
both the oxide film surface and the support are critical in
the metaloxide interactions. Here, we demonstrate how the
surfaces of the grown alumina films and the supports of the
alumina films affect metalalumina interactions.
Metal interactions with alumina films are strongly influenced
by surface properties of the thin oxide layers. The ordered
Al2 O3 -films grown on NiAl(110) have been extensively studied
by STM and surface X-ray diffraction (SXRD). The surface is
found to be terminated by a layer of O2 -ions [173,513,519].
Therefore, adsorption of reactive metals, such as Al, V, and Cr,
tend to build interfacial MO bonds such that charge transfer
occurs from the metal adatoms to surface oxygen, which results
in the oxidation of the adsorbed metals [109,489,520]. Like
bulk Al2 O3 surfaces, surfaces of thin alumina layers can also be
hydroxylated by exposing the O-terminated surfaces to water.
The surface OH groups tend to increase the dispersion of metal
overlayers on the hydroxylated surface. For example, STM
experiments on Rh and Pd deposition on thin alumina layers
showed that island density of the metals on the hydroxylated
surfaces is higher than that on the clean surfaces [489,521,522].
In oxide film-based model systems the effect of the metal
supports has to be taken into consideration. If the thickness
of the supported oxide layer is very small, e.g. less than
1 nm, electron tunneling happens between the underlying metal
substrates and metal adsorbates [115,116,119]. In such a case,
the support can have a strong effect on metal interactions on
the oxide surfaces. For example, a recent STM study in Au grown on NiAl(110)
adsorption on a thin alumina film (5 A)
477
Rh-, Pd-, and Pt-adsorption on an alumina layer grown on

NiAl(110) [489]. The electronic configuration in the systems of
Ni/Al2 O3 /Al, Pd/Al2 O3 /Al, and Rh (Pd, Pt)/Al2 O3 /NiAl(110)
can now be described by schematic shown in Fig. 41(b). In
contrast to Fig. 41(a), charge transfer takes place from metal
supports to metal overlayers because of E F (metal overlayer) <
E F (support), which results in downward band bending and
negative BE shifts in the oxides.
Fig. 40. (a)(b) STM images of Au pentamers on an alumina film on NiAl

(110), which are adsorbed in A and B domains, respectively. The angles
between the chain (dotted lines) and the NiAl [001] direction (straight lines)
are indicated. (c)(d) Schemes showing the orientation of the hexagonal Als
lattice with respect to the NiAl substrate. From [523].
demonstrated that 1D Au chains exhibit a preferential

orientation close to [001] direction of the underneath NiAl(100)
(Fig. 40) [523]. The results unambiguously indicate the
participation of the NiAl substrate in the Au binding on the
alumina surface. Furthermore, many PES studies confirmed that
supports may change the electronic structure of metalalumina
interfaces. The support effects mainly originate from the charge
transfer between the metal overlayers and the underlying metal
supports. Depending on the E F of the metal overlayer and the
support, two different charge transfer processes can occur in the
metal-overlayer/oxide-film/conductive-supportsystem.
Deposition of Cs onto ultrathin alumina films supported on
Mo(110) induced large (0.91.1 eV) positive shifts in the BE
Al
of O KVV, O1s, Al2p feature of alumina film [524]. 1 A
deposition on an alumina layer grown on NiAl(110) resulted
in a shift of the oxide component in Al2p, O1s, and O2p to
higher BE, 0.47 0.03 eV [489]. A similar shift to higher BE
in the range of 0.5 eV was detected during depositing V onto
the same oxide surface [109]. Considering that Cs, Al, and V
have low work functions, a charge transfer is expected to take
place from Mo to Cs [524], from NiAl to Al [489], and from
NiAl to V [109]. Consequently, downward band bending of the
oxide layer may happen, which results in positive BE shifts
of the oxide components. The case of E F (metal overlayer) >
E F (support) is illustrated schematically in Fig. 41(a).
Deposition of large work function metals on the supported
alumina films can reverse the charge transfer direction. For
example, Sarapatka
observed parallel negative shifts of oxide
Al2p and O1s in the case of Ni or Pd deposition onto alumina
thin layers formed by surface oxidation of polycrystalline
Al foil [171,172]. Core level spectra and Auger parameters
recorded from Ni or Pd overlayers suggested negative charging
of the metal particles, which is consistent with a charge
transfer from the Al support. In another example, small
negative BE shifts (around 0.1 eV) are also observed for
5.1.3.3. Chemical interaction of metals on alumina films. In

oxide film-based model systems, the defect structure of thin
oxide layers provides some channels for metal migration into
metal supports such that alloy formation may take place beneath
the oxide layers. Interdiffusion, mainly metal diffusion through
oxide layers into the metallic substrate, can limit the thermal
stability and therefore is very critical at elevated temperatures.
The thermal stability of Co, Rh, and Pd particles on a thin
alumina layer supported on NiAl(110) was systematically
studied as a function of annealing temperature [26,522].
Annealing the surface up to 900 K almost completely removed
all metal particles (see Fig. 42). STM experiments proved that
the diffusion process is defect-mediated. Loss of material was
mainly observed in the vicinity of antiphase domain boundaries,
which act as an important diffusion channel. Chen et al. [518]
used EELS and AES to study the thermal behaviors of Ni
overlayers deposited on Al2 O3 -films, which were prepared
by oxidizing an atomically clean Al(111) surface. Annealing
the metal/oxide systems up to 700 K causes a strong inward
diffusion of Ni overlayers. It was concluded that the metallic
aluminum substrate provides a major driving force for the
inward Ni diffusion by NiAl alloy formation. A similar
result was observed in the Cu/Al2 O3 /Al(111) system [525].
In Pd/Al2 O3 /Al and Ni/Al2 O3 /Al systems, Sarapatka

found
inward diffusion of Pd and Ni through alumina oxide layers
and subsequent formation of PdAl and NiAl alloys under the
oxides [171,172]. The interdiffusion of Pd and Ni can be
retarded by increasing the thickness of the Al2 O3 layer.
5.2. Metals on MgO
MgO has high chemical, thermal, and mechanical stability
which allows it to be widely used as a substrate for epitaxial
growth of metal films and as a catalytic support. Moreover,
the MgO(100) surface is non-polar and experimentally wellcharacterized, making it a good model system for theoretical
studies of insulating oxide surfaces.
5.2.1. Metal interactions with bulk MgO
We start with the most stable surface, MgO(100), which is
often subjected to metal adsorption or deposition. Two critical
factors including MgO(100) surface properties and metal
overlayers in the metalMgO(100) interactions are discussed
in detail.
5.2.1.1. MgO(100) surfaces. The single crystal MgO can
be easily cleaved along the {100} planes, so well-formed
MgO(100) surfaces are often obtained by cleavage in UHV or
478
Fig. 41. Flat energy band models of metal overlayer/ultrathin oxide layer/metal substrate systems. (a) metal overlayers have lower work function () compared to
that of metal substrate; (b) metal overlayers have larger work function () than that of metal substrate.
Fig. 42. Particle density as a function of the annealing temperature of Rh, Pd,
and Co deposits on alumina layers grown on NiAl(110). The decrease in island
density is from metal migration into metal substrates. Co shows higher stability
due to its strong metal support interaction. Reproduced from [522].
air [526528]. The surface reconstruction and point defects are

strongly dependent on the surface preparation condition.
Most of the well-prepared MgO(100) surfaces are unreconstructed. However, surface defects may induce reconstructions.
For example, a ( 2 2) surface structure was observed

by helium atom scattering (HAS) measurements on cleaved
MgO(100) single crystals [529]. The reconstruction is due to
the presence of a large compressive surface stress and a certain
amount of surface defects. Ca segregation onto MgO(100) sur
faces leads to another ( 2 2)R45 reconstruction [528].

On the well-defined (1 1)MgO(100) surfaces, both surface
relaxation and surface rumpling are very small ([528] and references therein). For example, Robach et al. reported a relaxation of 0.56% 0.35% and a rumpling of 1.07% 0.5% on
MgO(100) which was determined by grazing incidence X-ray
scattering (GIXS) [528].
Typically, MgO(100) surfaces exhibit defects, including
extended defects and point defects. The extended defects are
steps, line defects attributable to missing rows of Mg2+ and
O2 ions, rectangular holes of nanometer size originating
from cleavage, complex adstructures due to adatoms, etc.
All these surface features have been imaged by dynamic-
mode SFM [249] and NC-AFM [530]. The point defects

can be categorized into four groups: low coordinated sites,
surface vacancies, divacancies, and impurity atoms. The low
coordinated sites are four-coordinated ions located at step and
edge sites and three-coordinated ions located at corners, kinks,
etc. A surface oxygen vacancy is usually called as a surface
F (Fs ) center. Divacancies are created by removing a neutral
MgO unit. MgO surface point defects have been reviewed
by Pacchioni [77]. Single point defects on flat terraces of
MgO(100) were imaged by dynamic-mode SFM at atomic
resolution (Fig. 43) [249]. The density of point defects is
estimated to be roughly 1012 1013 defects/cm2 . It has been
shown that surface defects strongly influence the nucleation
and growth of metals on MgO(100) surface. Didier and Jupille
found that the growth mode of Ag on MgO(100) strongly
depends on the chemical purity of the oxide surface. On a
carbon- and defect-free MgO(100) surface, Ag grows in a 2D
mode while the presence of any surface defects prevents such
a growth mode [531]. The nucleation of Au and Pd on cleaved
MgO(100) surfaces has been studied by AFM, which indicates
that nucleation is controlled by surface point defects [136,532].
5.2.1.2. MetalMgO interactions. Many experiments have
been performed to study the interface formation between metals
and MgO(100). Further theoretical work contributed much to
the understanding of these interactions. The following discusses
the interaction of various metals with MgO(100) surfaces,
including both experimental and theoretical results.
(1) Experimental results: Using GIXS Renaud and
coworkers have systematically investigated the structure and
morphology of Ag [533,534], Pd [535], and Ni [536] films
grown on MgO(100) at RT. In all three cases, the epitaxial
orientation of metals on MgO(100) is cube-on-cube and the
adsorption site of metals is above oxygen ions on MgO(100).
At submonolayer coverage, Ag overlayers have the bulk lattice
parameter but Ni overlayers are strained to register with the
surface oxygen lattice of MgO(100). Pd with coverage less than
1 ML has an average lattice parameter between those of bulk
Pd and MgO. At high coverage, e.g., >1 ML, Ag and Pd form
a network of interfacial misfit dislocations to relax the strain,
whereas the relaxation of Ni is facilitated by the growth of Ni
Fig. 43. Single point defects on a flat MgO(100) terrace imaged with atomic
resolution by dynamic-mode SFM in two consecutive measurements (a) and (b)
(9.1 9.1 nm2 ). From [249].
clusters with an additional orientation of Ni(110) k MgO(100).

These authors concluded that structure and morphology at
the metal/MgO interfaces are mostly influenced by bonding
strength at the interface rather than the lattice parameter misfit.
Trampert et al. [537] have investigated the atomistic structure
of the Ag/MgO interface with HRTEM. These authors showed
that Ag islands grow cube-on-cube on the (100) surface of the
MgO substrate. The interface was atomically abrupt, i.e. no
chemical reactions like interdiffusion or formation of interfacial
phases were observed. The misfit between Ag and MgO
is accommodated by partial edge dislocations, leading to a
situation where the Ag atoms between dislocations are sitting
either on Mg or O atoms/ions.
Other experiments suggest that there is a very weak
interaction between MgO surface and metals of Cu [140],
Ag [533], and Au [532]. Metals often grow in form of
islands on MgO(100) at high temperature (above RT) and/or
high coverages (above 1 ML). However, below a critical
submonolayer coverage, it was observed that Pd [138] and
Ag [538] grow in 2D islands on MgO(100) surfaces.
Chemical interaction of metals on MgO(100) was observed
for Ti- and Zr-adsorption on MgO. Ti and Zr adatoms diffuse
into the MgO bulk via Mg substitutional sites at RT. No
reactions were found at interfaces of Fe, Ni, Ge, and Ag with the
479
MgO surfaces. The trends of interface reaction at metalMgO

interfaces are interpreted in terms of the reactivity of the metal
adatoms to the surface oxygen atoms [135].
(2) Theoretical results: Many theoretical studies have been
conducted to study metalMgO interactions. Several key issues
on the interface formation have been addressed. They include
the adsorption sites of metal atoms on MgO, the nature of
electronic interaction between metals and MgO, and the effect
of surface defects on metal adsorption. All the three aspects are
reviewed below.
For adsorption sites, most of the results indicate that on
defect-free MgO(100) surfaces metal atoms prefer to adsorb on
top of the surface O ions, which is consistent with experimental
results [25]. Schonberger et al. calculated that the minimum
energy of the Ag/MgO interface occurs when the Ag atoms are
sitting on top of the O atoms/ions [539]. Rosch and coworkers
made a systematic DFT study in the adsorption of various
metal atoms on MgO(100). They show that metals of Cr,
Mo, W, Ni, Pd, Pt, Cu, Ag, and Au all register on the oxide
surface anions [540542]. The adsorption geometry of Pd [64,
543], Cu [544], Ag [74,545,546], Au [543], and Al [546] on
MgO(100) showed that the metal atoms sit on the surface O
sites.
The electronic interaction between metals and MgOsurfaces with low defect densities is generally very weak [65,
547]. The metalMgO electronic interaction can be attributed
to three mechanisms, metal polarization, chemical bonding,
and MIGS. Goniakowski [65] suggested that Pd polarization
caused by the MgO surface electrostatic field can be of great
importance. There is a redistribution of electrons between
different d components in the metal VB upon interface
formation, which is similar to the case observed at metal/Al2 O3
interface (Fig. 39). In the case of metal adatom above surface
O, electrons transfer from the d3z 2 r 2 orbital to the rest of the d
band. Such an effect is similar to the macroscopic image charge
force, which is observed at interfaces between a metal overlayer
and an ionic crystal [62]. The polarization effect was found to
dominate in Cu, Ag, and Au interactions with MgO(100) [540,
541]. Another example is the ab initio study in metal/MgO(100)
systems [547]. They found that conventional MIGS produce a
smooth DOS in the MgO band gap energy range. The interface
states together with the polarization effect and bonding states
determine the interface characteristics, such as charge transfer,
SBHs, etc. Finally, the interface bonding provides another
important contribution to the metalMgO interactions. In most
cases, the bonding is facilitated by the mixing of metal valence
electron orbitals with O2p orbitals, which forms covalent bonds
at the interface. Ni, Pd, Pt, and W [540,542] were found to form
strong bonding with the oxide anions and the bond has a polar
covalent nature with little charge transfer from the metal to the
oxide.
Finally, we come to the effect of surface defects on
metal interactions. Extensive theoretical work demonstrates
that the presence of surface defects enhances the metalMgO
interactions. Matveev et al. [548] calculated that the adsorption
of Cu, Ni, Ag, and Pd is stronger on Fs sites by 12.4 eV
compared with regular O2 sites. Zhukovskii et al. showed
480
that the nature of the interaction between Ag or Cu and MgO

substrates with low defect density is of physisorptive character.
Above the Fs sites, metal atoms are much more strongly bound.
Adsorption onto the defect sites is accompanied by a substantial
charge transfer towards each adatom (Cu = 0.41 eV and
Ag = 0.32 eV) as well as formation of partly covalent MeFs
bonds at the interface [549]. In the Au/MgO(100) system, the
adsorption of Au on the ideal MgO(100) surface is found to be
very weak with an adsorption energy of 0.13 eV/adatom. On
the defective MgO(100) surface, Au adatoms prefer the vacancy
sites where the adsorption energy is 1.93 eV/adatom [550].
The same role of oxygen vacancies in Pd-adsorption on
MgO(100) was identified based on periodic ab initio electronic
band structure investigation of the PdMgO interface [551].
5.2.2. Metal interactions with MgO films
Highly ordered and stoichiometric MgO thin films can be
epitaxially grown on metallic substrates. These well-defined
MgO-films are often used as model systems for supported metal
overlayers.
5.2.2.1. Growth of MgO films. MgO(100) thin films are
frequently prepared by evaporation of Mg onto metal substrates
in the presence of O2 . Occasionally, annealing in an oxygen
atmosphere is necessary in order to crystallize the films.
The supports include metal single crystals such as Mo(100)
[552554], Ag(100) [555], W(110) [556]. STM and electron
spectroscopic investigations demonstrate that the grown MgOfilms are smooth and possess the bulk properties of MgO [552,
554]. Furthermore, the presence of surface defects on MgOfilms has been identified and the density of the surface
defects can be regulated by post-annealing or electron
bombardment [557,558].
5.2.2.2. Metal interactions with MgO films. Like the Al2 O3 film-based model systems, metal interactions with MgO-films
depend on both the MgO film surfaces and the supports,
respectively.
(1) Surface effects: Similar to metal interactions on bulk
MgO crystals, surface defects play an important role in the
adsorption of metal atoms on MgO-films. Au-adsorption on
MgO(100) films has been systematically investigated through
ab initio calculations and laboratory experiments. Using massselected and softlanding techniques, Au8 clusters having the
smallest catalytically active size were deposited on MgO(100)
films grown on Mo(100) [557,559]. Yoon et al. found that Au8
clusters supported on defect-rich MgO(100) surfaces are active
whereas clusters deposited on virtually perfect MgO surface
remain chemically inert. As illustrated in Fig. 44, the stretch
vibration of CO adsorbed on mass-selected Au8 on MgO(100)
with coadsorbed O2 shows a red shift on an F-center-rich
surface with respect to the perfect surface. It was concluded
that there exists a larger degree of charge transfer from the Fcenter-rich MgO surface to the gold cluster than that on an Fcenter-free surface. Charging of the supported Au clusters by
the surface defects plays a key role in promoting their chemical
activity [265]. Sterrer et al. applied electron paramagnetic
resonance (EPR), infrared spectroscopy (IR), and STM to study

the interaction of Au clusters with F centers on MgO surfaces.
They also confirmed that Au particles adsorbed to color centers
are indeed negatively charged while Au particles on regular
terrace sites are neutral [558]. In addition to surface point
vacancies, hydroxyl groups on MgO surfaces also affect the
metal adsorption process strongly [560], which is similar to
the cases observed in metal adsorption on hydroxylated TiO2 ,
Al2 O3 , and SiO2 surfaces [319,461,462,489,491,561563].
(2) Support effects: The other important factor, which
influences the metalMgO-film interaction, is the metal support
effect. Pacchioni and coworkers demonstrate that a metal
support at the interface with a MgO-film results in charging of
adsorbed atoms with high electron affinity, like Ag and Au [543,
564]. They concluded that an adsorbed Au or Ag atom is
almost neutral on a single crystal MgO(100) surface, while it is
negatively charged on a MgO(100) film supported on Mo(100).
In Au/MgO(100) Au6s level is half filled and the configuration
is atom-like, 5d10 6s1 . However, in Au/MgO/Mo(100) the Au6s
level lies below the E F of Mo and both and components
of the Au6s level are filled. Therefore, Au carries a net negative
charge and becomes Au with an electronic configuration of
5d10 6s2 (as shown in Fig. 45). The charge transfer, which may
be facilitated by tunneling effects, can occur from a metal
support with low work function to adsorbed metal atoms with
high electron affinity.
5.3. Metals on SiO2
SiO2 plays an important role in many technological
applications, for example as dielectric layer in microelectronics
and as catalyst support in heterogeneous catalysis. Over a broad
range of temperatures and pressures -quartz is one of the most
stable structures of SiO2 , and the (0001) -quartz surface can
be considered a model surface. However, the band gap of bulk
SiO2 is very large, around 9.0 eV [565] such that it is difficult
to study bulk SiO2 surfaces with many surface techniques.
Studies of metalSiO2 interactions often involve SiO2 -films
grown on single crystal Si or refractory metal substrates, which
are then studied with various surface science techniques. Here,
we discuss the interaction of metals with bulk SiO2 and SiO2 films.
5.3.1. Metal interactions with bulk SiO2
There have been a few experimental studies in single crystal
quartz surfaces. The available results suggest two (0001) quartz surface structures. A LEED investigation of a (0001)
surface, which was etchedin HF solution,
indicated a (1
1)-type surface [566]. A ( 84 84)-R11 reconstruction

was observed when heating the -quartz surfaces in air above
600 C; the surface structure is related with the quartz
phase transition which occurs at 573 C [567]. Harte et al. have
studied the initial states of Cr and Ti growth on SiO2 (0001).
No large difference
in the metal growth was observed on both
(1 1) and ( 84 84)-R11 surfaces. Moreover, they found

that deposition of Ti results in the formation of a Ti oxide layer
on SiO2 , while Cr forms metal clusters on the surfaces [568].
481
Fig. 44. Mass spectrometric signals pertaining to the formation of CO2 on Au8 deposited on (a) F-center-rich and (b) F-center-free MgO(100) thin films. (c) and
(d) FTIR spectra measured for the same surfaces and with the same CO and O2 exposures as in (a) and (b), respectively, at various annealing temperatures. Mass
spectrometric data show that Au8 adsorbed on an F-center-free MgO(100) surface were essentially inactive for the combustion reaction (b); IR results indicate a
red shift of the CO stretching frequency for the molecule adsorbed on Au8 supported on the defect-rich MgO thin film. From [265].
Fig. 45. DOS (density of states) curves for a Au atom adsorbed on top of O. (a) unsupported MgO(100) (3 layers); (b) supported MgO(100) (3 layers) on Mo(100).
DOS of Au 2. From [543].
Fig. 46. STM images of a SiO2 /Mo (112) film. (a) 100 100 nm2 , Vs = 2 V, I = 0.2 nA, (b) 8 8 nm2 , Vs = 1.2 V, I = 0.35 nA, (c) 8 8 nm2 , Vs = 0.65 V,
I = 0.75 nA. The arrow indicates an antiphase domain boundary and insets in (b) and (c) show the close up of the atomically resolved STM image (left) and
simulated image (right). From [580].
The intrinsic defects on bulk SiO2 surfaces may be

divided into two families. Oxygen-related defects of SiO2
include the peroxyl bridge SiOOSi , the peroxyl
radical SiOO, the non-bridging oxygen SiO, and
+
overcoordinated oxygen O+
3 (Si3 ) or O3 (Si2 O). Si-related
defects include the two-fold coordinated silicon =Si, the
silicon dangling bond Si, and the oxygen vacancy, which

may be neutral VO (the weak Si Si ), the positively
charged VO corresponding to an E 0 center, or the wrong
bond SiSi . All these defects have been identified by
EELS [565] and ESR [569]. Experimental and theoretical data
showed that the interaction of metals, such as Au, Cu, Pd,
482
and Cs, on regular sites of SiO2 is weak, mainly through

dispersion interaction [570572]. The surface defects, however,
can stabilize the deposited metal atoms. Among the defect sites,
the non-bridging oxygen forms the strongest bonds with metal
atoms, which is followed by the silicon dangling bond, and the
neutral oxygen vacancies [570,572].
Like TiO2 , Al2 O3 , and MgO surfaces, hydroxylation may
occur on SiO2 surfaces [573]. Silanol groups (SiOH)
can be introduced by reaction of SiO2 surfaces with water.
The concentration, distribution, and nature of the silanol
groups clearly influence the properties for the technological
application [561563].
5.3.2. Metal interactions with silica films
SiO2 -films grown on conductive or semiconductive substrates possess the chemical and physical properties of bulk
SiO2 [574576]. Furthermore, the thin SiO2 -films can circumvent the surface charging effect encountered in bulk SiO2 systems. Therefore, the supported SiO2 -films are good model systems for studies in metalSiO2 interactions.
5.3.2.1. Growth of silica films. The simplest route to prepare
SiO2 -films is to oxidize single crystal Si surfaces, either
via thermal oxidation of Si in oxidizing atmospheres or
by wet etching in solutions [5,577]. Other techniques, such
as ALD [187,188] and thermal evaporation of SiO in O2
atmosphere [96,98], have been also used to grow thin SiO2 films. All the methods enable control over the thickness of the
SiO2 layers, while the structure of the as-deposited layers is
generally disordered.
From a structural point of view, it is highly desirable
that model systems consist of well-ordered films instead of
amorphous ones. Schroeder et al. [578,579] first reported
the preparation of a thin crystalline SiO2 -film on Mo(112).
The experimental procedure consists of repeated cycles of Si
deposition and subsequent oxidation, which is followed by a
final annealing step. The silica multilayer film is stoichiometric
and fully covers the support surface. Recently, Freund and
coworkers have modified this process and prepared a monolayer
crystalline SiO2 -film on Mo(112) (Fig. 46) [580,581]. The
atomic structure of the epilayer was resolved by STM,
infrared reflection-adsorption spectroscopy (IRAS), and DFT
calculation. The film consists of a 2D honeycomb-like network
of SiO4 tetrahedra with one oxygen of each tetrahedral unit
binding to the protruding Mo atoms of the metal surface,
while the other three form SiOSi bonds with the neighboring
tetrahedral unit. Goodman and coworkers [582,583] have also
grown highly crystalline, well-ordered thin SiO2 -films on
Mo(112) surfaces and characterized them by LEED, STM,
and HREELS. The derived structural model of the SiO2 -film
consists of a layer of isolated [SiO4 ] clusters arranged in a
c(2 2) structure on the Mo(112) surface with all oxygen
atoms bonding to the Mo substrate. The physical properties
of the SiO2 -films at 1 ML coverage are influenced by the Mo
substrate, while films with coverage greater than 2 ML show
properties comparable to bulk-like SiO2 samples [584]. Flat
SiO2 -films have been also deposited on Mo(110) [574,585,
586] and Mo(100) [575] surfaces. In addition to Mo surfaces,

Kundu and Murata [587] have prepared a single crystal SiO2
film with the structure of quartz on a Ni(111) surface and
Zhang et al. [588] deposited atomically flat silica films on a
Pd(100) surface.
5.3.2.2. Metal interactions with silica films. The effects of the
SiO2 film surfaces and the supports of the SiO2 -films on metal
interactions with SiO2 -films are given below.
(1) Surface effects: Surface defects are an important factor
in the interaction between metals and SiO2 . The presence of
steps, antiphase domain boundaries, and oxygen vacancies on
surfaces of SiO2 epilayers grown on Mo has been confirmed by
MIES, UPS, LEED, and STM [576,579,580,589]. Au clusters
were used to identify the role of the various defects in the
nucleation and growth of metals on the SiO2 surfaces. The
stability of Au nanoclusters on defect sites follows the tendency
of oxygen vacancy complexes > step edges > line defects >
single oxygen vacancies [589]. In addition, the defects of Ti
impurities introduced onto SiO2 surfaces can strengthen the
metal interaction with the surfaces. Metal clusters show a
marked increase in island density and are sinter-resistant on the
TiOx -modified SiO2 surface [590,591].
(2) Support effects: Depending on the thickness of the
grown SiO2 layer the supports may have a strong influence
on metalSiO2 interactions. Toyoshima and coworkers [169,
592] have studied the interaction of Ni with SiO2 -films
grown on Si(111) and CO-adsorption at the Ni/SiO2 /Si(111)
system. CO-adsorption was suppressed on the Ni clusters that
were deposited on SiOx /n-Si(111) but normal CO-adsorption
occurred on Ni supported on SiOx /p-Si(111). It was concluded
that in the Ni/SiOx /n-Si(111) system the charge transfer takes
place from the donor level of n-Si to the Ni d orbital via
Thus,
electron tunneling through the thin SiOx interlayer (3 A).
CO-adsorption at the Ni/SiO2 /n-Si(111) system is inhibited
due to retardation of -donation from CO to Ni. The charge
transfer between metal overlayers and Si(111) substrates has
been studied in metal/SiO2 (15 A)/Si(111)

systems by Ofner
et al. [170]. They observed that the charge transports from
Cs to the SiO2 /Si interface which built dipole field across
the SiO2 layer and induced Si band bending. However, In
deposition did not cause any changes in SiO2 /Si interface
electronic structure. The different electronic interactions in
the Cs/SiO2 /Si and In/SiO2 /Si systems are attributed to the
different electronegativities of Cs and In, which can be also
explained by the schematics shown in Fig. 41.
5.3.2.3. Chemical interaction of metals on silica films. Metal
adatoms on SiO2 surfaces can be subjected to various surface
and interface processes. Fig. 47 gives the schematic description
of the processes which could happen at metal/SiO2 /Si model
systems. The typical chemical interactions described in
Section 2.3.1, which include redox reaction, alloy formation,
encapsulation, and interdiffusion, have been observed at
metalSiO2 interfaces [100,593].
Redox reaction: Pretorious et al. [103] found that Ti, Zr, Hf,
V, and Nb react with SiO2 to produce oxides and silicides at
483
Fig. 47. Schematic of surface and interface processes in metal/SiO2 /Si model
systems. From [593].
the interfaces. They suggested a direct reaction between a silica

film and a metal overlayer as follows:
Mx + SiO2 M y Si + Mxy O2 .
The reaction can be predicted based on thermodynamic
consideration as discussed in Section 2.3.2. Metals with an
electronegativity of less than 1.5 on the Pauling scale should
react with a SiO2 substrate. Similar to the above result, reactive
metals such as Al, Mg, Ti, Si, and Ge were found to react
with SiO2 directly and cause a homogeneous decomposition
of SiO2 -films grown on Si(100) [594]. Annealing a CuMg
alloy deposited on 650 nm SiO2 layers results in the formation
of fcc MgO and the reduction of SiO2 [595]. In the
Ti/SiO2 (50A)/Si(111)
system, the metal oxidation and SiO2
reduction occur even at room temperature [596].
Encapsulation: Powell and Whittington [597] suggested a
mechanism of encapsulation to explain the deactivation of
SiO2 -supported Pt model catalysts. They observed that Pt
particles become partially immersed in the SiO2 surface with
a concurrent formation of a SiO2 ridge around the base of the
Pt particles when annealing the catalysts at 1200 and 1375 K.
The encapsulating process was driven by the minimization of
surface free energy in the Pt/SiO2 system. Van den Oetelaar
et al. studied the thermal stability of Cu particles supported on
a thick SiO2 layer (400500 nm) by LEIS, AFM, and RBS [593,
598]. The disappearance of Cu from the outermost atomic layer
of the UHV annealed Cu/SiO2 model catalysts was attributed
to encapsulation of the Cu particles by silicides. The UHVannealed Cu/SiO2 samples can be regenerated by exposure to
air at room temperature for several hours. The reversibility of
the surface process is very similar to the encapsulation reactions
observed in metal/TiO2 systems [8].
Interdiffusion: Interdiffusion of metals into SiO2 layers is
often observed upon annealing metal overlayers supported on
thin SiO2 -films (10 nm). In metal/SiO2 /Si systems, silicide
formation between the metal overlayers and the Si substrate
drives the metal diffusion through the SiO2 layer to the
SiO2 /Si interface [594]. It is believed that the interdiffusion is
facilitated by defects in the SiO2 layers, e.g., oxygen vacancies,
pinholes, microchannels, or microvoids. The interdiffusion of
Ni [599,600], Cu [593], Pd [601,602], and other transition
Fig. 48. HRTEM images of three particles after reduction showing formation
of (a) Pt3 Si with Cu3 Au structure, (b) monoclinic Pt3 Si, and (c) tetragonal
Pt12 Si5 . From [98].
metals [594] in SiO2 layers has been reported upon annealing

of the metal/SiO2 /Si model systems.
484
As we have discussed in Section 2.4, mass transport in an

oxide can be assisted by an electric field therein. For example,
Vogt and Drescher [603] demonstrated field-assisted Cu diffusion in SiO2 . When annealing a 100 nm Cu/150 nm SiO2 /Si
system at 450500 C for 1 h, no Cu diffusion in the SiO2 layer
was observed. However, the thermal treatment of the interface
with an external electric field results in strong enhancement of
Cu transport. The applied electric field accelerates the transport process by assisting the ionized copper atoms to migrate
through the oxide towards the Si substrate.
Alloy formation: In metal/SiO2 (thinlayer)/Si model systems,
many metals can diffuse to SiO2 /Si interface, where they react
with Si substrates to form metal silicides [593,594,596,598
602,604].
Other studies also show that metal silicide formation can
take place via direct reaction between metal and SiO2 . Van den
Oetelaar et al. [593] proposed a reaction mechanism to explain
the direct reaction between a noble metal and a thick SiO2 -film
under the condition of UHV annealing:
UHV,HT
xCu+ OSiO Cux Si + (x 1)SiO2

+ oxygen-compound .
The oxygen compound represents oxygen that is released
by the reduction of SiO2 . The strong interaction between Cu
and SiO2 supports results in the formation of Cu+ species,
which could alter the surface and interface energy in the system
and make the reaction thermodynamically possible. A similar
mechanism is found to be active for Rh3 Si formation in the case
of thermal treatment of a Rh/SiO2 /Mo model system above 850
K in UHV [605] and for Pd silicide formation in a Pd/SiO2 /Mo
model system upon UHV annealing at 1000 K [100,606].
In the presence of H2 , silicide formation takes place via a
different mechanism [99,607]:
Mx + SiO2 + 2H2 Mx Si + 2H2 O.
Thermodynamically, the driving force behind the process is
the formation of stable water molecules. For example, in a
Pt/SiO2 model system silicide formation was studied by various
TEM techniques [96,98,608]. Pt particles supported on freestanding amorphous SiO2 -films (25 nm thick) were reduced
in 1 bar H2 at 873 K. Pt3 Si with Cu3 Au structure, monoclinic
Pt3 Si, and tetragonal Pt12 Si5 were identified after the treatment
(see Fig. 48). The reaction products result from reduction
of SiO2 by atomic hydrogen, which involves dissociative
adsorption of H2 on Pt particles, reduction of SiO2 by atomic
hydrogen diffused from Pt, and migration of Si atoms into
Pt to form Pt silicides. An early investigation of Pt reaction
with SiO2 conducted by Lamber et al. [99] also confirmed
the formation of a cubic platinum-rich Pt3 Si phase, which is
intermediate to the monoclinic Pt3 Si phase. The hydrogeninduced metal silicide formation has been reported in many
other metal/SiO2 systems. For example, a Ni3 Si compound was
observed after prolonged heating of a Ni/SiO2 model system
in a hydrogen atmosphere [609]. A Pd2 Si phase was revealed
by TEM, TED, and Convergent Beam Electron Diffraction
(CBED) after heating a Pd/SiO2 system in H2 . They found that
metalsupport interaction is influenced by the pre-treatment

of SiO2 support, indicating that the presence of OH groups
on the silica facilitates chemical metalsupport interaction and
formation of a metal silicide [561]. Other Pd silicides, such as
Pd4 Si and Pd3 Si, have been observed after reduction of Pd/SiO2
catalysts in H2 up to 600 C [610,611].
6. Summary
In the present review, two fundamental questions at
metal/oxide interfaces were addressed. (1) What is the
electronic interaction during metal/oxide interface formation?
(2) What kind of chemical reaction occurs during metal/oxide
interface formation?
First, contact between a metal and an oxide results in charge
redistribution at the interface at the local range and/or the long
range. The electron redistribution is driven by principles of
energy minimization of the system and continuity of electric
potential in the solid.
Local charge redistribution occurs within a few atomic
layers close to the interface, which includes polarization
of metal electron orbitals, formation of image charges
in the metal, MIGS at interfaces, and interface bonding.
The local electronic interaction is particularly important
at metal/insulating oxide systems, e.g., metal/Al2 O3 (0001),
metal/MgO(100), and metal/SiO2 . Two aspects are critical in
the process. One of the important factors is the electronegativity
of the metal; the oxides surface property including surface
defects, surface stoichiometry, surface termination, and surface
hydroxylation is the other one.
The long-range electronic interaction at metal/oxide
interfaces is analogous to that at metalsemiconductor
junctions. The equilibration between E F of the metal and
oxide contacting phases induces charge transfer between the
metal and the space charge region in the oxide. This space
charge transfer relies on the electronic structures of the two
contacting phases, i.e. the surface work function (E F ) of the
solids. In cases of metals supported on mixed conducting
oxides, such as TiO2 (110) and SrTiO3 (100), the space
charge is very important. In oxide film-based model systems,
e.g., metal/Al2 O3 -film/support, metal/MgO-film/support, and
metal/SiO2 -film/support, electrons can tunnel through the thin
insulating oxide layers, which allow long-range charge transfer
to occur between the lower conductive support and the metal
overlayers. In this case, the charge transfer depends on the E F
of the support and the metal overlayers.
Secondly, when putting a metal onto an oxide surface, the
metal and substrate atoms may be involved in various mass
transport processes as shown in Fig. 17. One of the basic
processes is the surface diffusion of metal adatoms on oxides,
which enables the atoms to find the energetically preferable
surface adsorption sites. Metal adsorption on an oxide surface is
closely correlated with the local electronic interaction between
the metal and oxide, which depends on metal electronegativity
and oxide surface properties.
Metal and/or substrate atoms can diffuse across the interface,
which results in interfacial reactions. The reactions are driven
by thermodynamic forces but may be kinetically limited at

relatively low temperature (<1000 C). For mixed conducting
oxides, such as TiO2 and SrTiO3 , the interface reactions
involve ionic defect diffusion processes in the substrates, which
are closely related to defect chemistry of the oxides. For
example, the production and diffusion of oxygen vacancies
are critical in metal oxidation on TiO2 and SrTiO3 surfaces.
Diffusion of Ti interstitials Tiin+ (n 4) dominates in metal
encapsulation reactions, oxygen-induced restructuring, and
bulk-assisted reoxidation reactions on TiO2 surfaces. Transport
of Sr vacancies and oxygen vacancies between the surface
and bulk of SrTiO3 determines the surface reconstruction and
the formation of new surface phases at SrTiO3 surfaces. The
diffusion of ionic defects in oxides can be promoted or retarded
by an interfacial electric field. At the metal/oxide interfaces,
mass transport is coupled with the electronic interaction.
As has been shown in many cases, reactions at metal/TiO2
and metal/SrTiO3 interfaces show a strong dependence on
bulk electronic structure (free electron density or E F ) of
both contacting phases. The metaloxide interface reaction
mechanism is consistent with the observed oxidation reactions
of metal surfaces and etching reactions of semiconductor
surfaces. This behavior can be explained in the framework
of the generalized CabreraMott theory, i.e. mass transport
occurring in a bi-phase (solidsolid or solidgas) interface
reaction may be dependent on the bulk electronic structure
of the two contact phases. Thus, interface reactions can be
successfully tuned by controlling E F of the metal and/or oxide
phases, such as doping of oxides, alloying of metals, application
of external electric field, etc.
Metal/oxide interface is a fast-growing research field and
many systems have been investigated. In this paper metals on
a few typical oxides (TiO2 , SrTiO3 , Al2 O3 , MgO, and SiO2 )
have been reviewed, demonstrating the critical factors in determining the metaloxide interactions and the possible routes
to desirably control the formation of metal/oxide interfaces.
The discussed metal/oxide interfaces can be classified into
three systems: (1) metals on mixed conducting oxides including metal/TiO2 (110) and metal/SrTiO3 (100); (2) metals on insulating oxides, such as metal/Al2 O3 (0001), metal/MgO(100),
and metal/SiO2 (0001); (3) metals on supported thin oxide films,
e.g., metal/Al2 O3 -film/support, metal/MgO-film/support, and
metal/SiO2 -film/support. Most of the previous research efforts
performed in these systems have focused on the adsorption and
bonding of metals on the oxide surfaces, which show strong
dependence on the oxide surface properties. Some experiments
have shown that the nucleation, growth, and epitaxy of metals on oxides can be deliberately controlled through the surface modification of the oxide supports, e.g. the surface reconstruction, change in surface termination, and variation in surface point defects. Surface hydroxyl groups are often present
on many oxide surfaces and the surface hydroxylation is another important parameter to regulate the metaloxide interaction [319,459,460]. It is clear that more such studies should be
done in future.
Recent results indicate that the formation of metal/oxide interfaces is not only correlated with the oxide surface proper-
485
ties but also strongly influenced by the oxide bulk properties,

such as the bulk ionic defect and bulk electronic structure [13,
28,85,89]. This point is particularly important for systems of
metals on mixed conducting oxides, where space charge transfer and mass transport at the interfaces are feasible. We show
that the interactions and, in particular, chemical reactions at
the interfaces are determined by the defect chemistry and E F
in the oxide bulk and the experiments can be explained by the
concept of the generalized CabreraMott theory (see Figs. 13
15). Although these results were observed at metal/TiO2 and
metal/SrTiO3 interfaces it seems to be very likely that similar
results occur on the surfaces of other mixed conducting oxides,
such as CeO2 , ZrO2 , ZnO, SnO, etc.
For metals on supported thin oxide films, the film supports
are of great importance. Some example showed that electronic
interaction between the conductive support underneath the film
and the metal overlayer on top of the film occurs via electron
tunneling through the thin oxide films (see Fig. 41). Chemical
reactions, mainly alloy formation, between the metal overlayers
and the support may promote the interdiffusion through the
thin oxide films. Apparently, the charge transfer and mass
transport processes are sensitively dependent on the thickness
of the thin oxide films. Therefore, the choices of the conductive
support and the thickness of the overgrown oxide film are
critical parameters controlling the metaloxide interactions.
Nucleation, growth, electronic state, and surface chemistry of
the metal deposit on the thin oxide films have to be related
to properties and variations of the underlying support and film
thickness [159].
It is highly expected that the macroscopic properties of the
metal/oxide interfaces in both model and real systems, such
as catalytic performance, electrical properties, and mechanical
stability, can be tuned based on the above concepts developed
from the studies on the model systems.
Acknowledgements
The authors would like to thank Prof. Manfred Ruhle for his
support when this project was performed. We owe warm thanks
to Prof. Xinhe Bao, Prof. Robert Schlogl, Dr. Dangsheng Su
for their support during the writing of this manuscript; to Dr.
Jason White for critical reading of the manuscript; and to many
colleagues for providing us electronic versions of their figures.
We would also like to thank Prof. Ulrike Diebold for having
invited us to write this review and critically reading the article.
References
[1] M. Ruhle, A.G. Evans, M.F. Ashby, J.P. Hirth (Eds.), MetalCeramic
Interfaces, Pergamon, Oxford, 1990.
[2] F. Ernst, Metaloxide interfaces, Mater. Sci. Eng. R 14 (1995) 97.
[3] G.D. Wilk, R.M. Wallace, J.M. Anthony, High-k gate dielectrics: Current
status and materials properties considerations, J. Appl. Phys. 89 (2001)
5243.
[4] S.I. Association, International technology roadmap for semiconductors.
http://www.itrs.net/.
[5] R.M.C. de Almeida, I.J.R. Baumvol, Reactiondiffusion in high-k
dielectrics on Si, Surf. Sci. Rep. 49 (2003) 1.
[6] F. Solymosi, Importance of the electric properties of supports in the
carrier effect, Catal. Rev. Sci. Eng. 1 (1967) 233.
486
[7] G.M. Schwab, Electronics of supported catalysts, Adv. Catal. 27 (1978)

1.
[8] G.L. Haller, D.E. Resasco, Metalsupport interaction: Group VIII metals
and reducible oxides, Adv. Catal. 36 (1989) 173.
[9] G.L. Haller, New catalytic concepts from new materials: Understanding
catalysis from a fundamental perspective, past, present, and future,
J. Catal. 216 (2003) 12.
[10] C.G. Vayenas, S. Brosda, C. Pliangos, The double-layer approach to promotion, electrocatalysis, electrochemical promotion, and metalsupport
interactions, J. Catal. 216 (2003) 487.
[11] O. Dulub, W. Hebenstreit, U. Diebold, Imaging cluster surfaces with
atomic resolution: The strong metalsupport interaction state of Pt
supported on TiO2 (110), Phys. Rev. Lett. 84 (2000) 3646.
[12] D.W. Goodman, Catalytically active au on titania: Yet another example
of a strong metal support interaction (SMSI)? Catal. Lett. 99 (2005) 1.
[13] Q. Fu, T. Wagner, S. Olliges, H.D. Carstanjen, Metaloxide interfacial
reactions: Encapsulation of Pd on TiO2 (110), J. Phys. Chem. B 109
(2005) 944.
[14] S.J. Tauster, S.C. Fung, R.L. Garten, Strong metalsupport interactions.
Group 8 noble metals supported on TiO2 , J. Am. Chem. Soc. 100 (1978)
170.
[15] S.J. Tauster, S.C. Fung, R.T.K. Baker, J.A. Horsley, Strong interactions
in support-metal catalysts, Science 211 (1981) 1121.
[16] A.T. Bell, The impact of nanoscience on heterogeneous catalysis,
Science 299 (2003) 1688.
[17] V.E. Henrich, P.A. Cox, The Surface Science of Metal Oxides,
Cambridge University Press, Cambridge, 1994.
[18] U. Diebold, J.M. Pan, T.E. Madey, Ultrathin metal film growth on
TiO2 (110): An overview, Surf. Sci. 331333 (1995) 845.
[19] D.W. Goodman, Model catalysts: From extended single crystals to
supported particles, Surf. Rev. Lett. 2 (1995) 9.
[20] R.J. Lad, Interactions at metal/oxide and oxide/oxide interfaces studied
by ultrathin film growth on single-crystal oxide substrate, Surf. Rev. Lett.
2 (1995) 109.
[21] C. Noguera, Physics and Chemistry of Oxide Surfaces, Cambridge
University Press, Cambridge, 1996.
[22] M.W. Finnis, The theory of metalceramic interface, J. Phys.: Condens.
Matter 8 (1996) 5811.
[23] C.T. Campbell, Ultrathin metal films and particles on oxide surfaces:
Structural, electronic and chemisorptive properties, Surf. Sci. Rep. 27
(1997) 1.
[24] C.R. Henry, Surface studies of supported model catalysts, Surf. Sci. Rep.
31 (1998) 231.
[25] G. Renaud, Oxide surfaces and metal/oxide interfaces studied by grazing
incidence x-ray scattering, Surf. Sci. Rep. 32 (1998) 1.
[26] M. Baumer, H.-J. Freund, Metal deposits on well-ordered oxide films,
Prog. Surf. Sci. 61 (1999) 127.
[27] H.-J. Freund, Clusters and islands on oxides: From catalysis via
electronics and magnetism to optics, Surf. Sci. 500 (2002) 271.
[28] U. Diebold, The surface science of titanium dioxide, Surf. Sci. Rep. 48
(2003) 53.
[29] T. Wagner, J. Marien, G. Duscher, Cu, Nb and V on (110) TiO2 (rutile):
Epitaxy and chemical reactions, Thin Solid Films 398399 (2001) 419.
[30] L.J. Brillson, The structure and properties of metalsemiconductor
interfaces, Surf. Sci. Rep. 2 (1982) 123.
[31] W. Monch, On the physics of metalsemiconductor interfaces, Rep.
Prog. Phys. 53 (1990) 221.
[32] R.T. Tung, Recent advances in Schottky barrier concepts, Mater. Sci.
Eng. R 35 (2001) 1.
[33] W. Schottky, Halbleitertheorie der sperrschicht, Naturwissenschaften 26
(1938) 843.
[34] N.F. Mott, Note on the contact between a metal and an insulator or semiconductor, Proc. Cambridge Philos. Soc. 34 (1938) 568.
[35] S.M. Sze, Physics of Semiconductor Devices, John Wiley & Sons, New
York, 1981.
[36] H. Luth, Solid Surfaces, Interfaces and Thin Films, 4th ed., Springer,
2001.
[37] J. Bardeen, Surface states and rectification at a metal semi-conductor

contact, Phys. Rev. 71 (1947) 717.
[38] A.M. Cowley, S.M. Sze, Surface states and barrier height of
metalsemiconductor systems, J. Appl. Phys. 36 (1965) 3212.
[39] S. Kurtin, T.C. McGill, C.A. Mead, Fundamental transition in the
electronic nature of solids, Phys. Rev. Lett. 22 (1969) 1433.
[40] M. Schluter, Chemical trends in metalsemiconductor barrier height,
Phys. Rev. B 17 (1978) 5044.
[41] J. Robertson, Band offsets of wide-band-gap oxides and implications for
future electronic devices, J. Vac. Sci. Technol. B 18 (2000) 1785.
[42] V. Heine, Theory of surface states, Phys. Rev. 138 (1965) A1689.
[43] S.G. Louie, M.L. Cohen, Electronic structure of a metalsemiconductor
interface, Phys. Rev. B 13 (1976) 2461.
[44] J. Tersoff, Schottky barrier heights and the continuum of gap states, Phys.
Rev. Lett. 52 (1984) 465.
[45] R.T. Tung, Chemical bonding and Fermi level pinning at
metalsemiconductor interfaces, Phys. Rev. Lett. 84 (2000) 6078.
[46] R.T. Tung, Formation of an electric dipole at metalsemiconductor
interfaces, Phys. Rev. B 64 (2001) 205310.
[47] D.M. York, W. Yang, A chemical potential equalization method for
molecular simulations, J. Chem. Phys. 104 (1996) 159.
[48] A.K. Rappe, W.A. Goddard III, Charge equilibration for molecular
dynamics simulations, J. Phys. Chem. 95 (1991) 3358.
[49] J.M. Andrews, J.C. Phillips, Chemical bonding and structure of
metalsemiconductor interfaces, Phys. Rev. Lett. 35 (1975) 56.
[50] G. Ottaviani, K.N. Tu, J.W. Mayer, Interfacial reaction and Schottky
barrier in metal-silicon systems, Phys. Rev. Lett. 44 (1980) 284.
[51] L.J. Brillson, Transition in Schottky barrier formation with chemical
reactivity, Phys. Rev. Lett. 40 (1978) 260.
[52] L.J. Brillson, Chemical reaction and charge redistribution at
metalsemiconductor interfaces, J. Vac. Sci. Technol. 15 (1978)
1378.
[53] W. Monch, Metal-semiconductor contacts: Electronic properties, Surf.
Sci. 299/300 (1994) 928.
[54] L.J. Brillson, Chemical mechanisms of Schottky barrier formation, J.
Vac. Sci. Technol. 16 (1979) 1137.
[55] S.G. Louie, J.R. Chelikowsky, M.L. Cohen, Ionicity and the theory of
Schottky barriers, Phys. Rev. B 15 (1977) 2154.
[56] W. Monch, Mechanisms of Schottky barrier formation in
metalsemiconductor contacts, J. Vac. Sci. Technol. B 6 (1988)
1270.
[57] C. Noguera, G. Bordier, Theoretical approach to interfacial metaloxide
bonding, J. Physique III 4 (1994) 1851.
[58] J.V. Naidich, The wettability of solids by liquid metals, Prog. Surf.
Membrane Sci. 14 (1981) 353.
[59] F. Didier, J. Jupille, The van der Waals contribution to the adhesion
energy at metaloxide interfaces, Surf. Sci. 314 (1994) 378.
[60] J.G. Li, Chemical trends in the thermodynamic adhesion of
metal/ceramic systems, Mater. Lett. 22 (1995) 169.
[61] A.M. Stoneham, Systematics of metal-insulator interfacial energies: A
new rule for wetting and strong catalyst-support interactions, Appl. Surf.
Sci. 14 (1983) 249.
[62] A.M. Stoneham, P.W. Tasker, Metal-non-metal and other interfaces: The
role of image interactions, J. Phys. C: Solid State Phys. 18 (1985) L543.
[63] D.M. Duffy, J.H. Harding, A.M. Stoneham, Atomistic modeling of
metaloxide interfaces with image interactions, Philos. Mag. A 67
(1993) 865.
[64] J. Goniakowski, Electronic structure of MgO-supported palladium films:
Influence of the adsorption site, Phys. Rev. B 57 (1998) 1935.
[65] J. Goniakowski, Transition metals on the MgO(100) surface: Evolution
of adsorption characteristics along the 4d series, Phys. Rev. B 59 (1999)
11047.
[66] A. Bogicevic, D.R. Jennison, Variations in the nature of metal adsorption
on ultrathin Al2 O3 films, Phys. Rev. Lett. 82 (1999) 4050.
[67] C. Verdozzi, D.R. Jennison, P.A. Schultz, M.P. Sears, Sapphire(0001)
surface, clean and with d-metal overlayers, Phys. Rev. Lett. 82 (1999)
799.

[68] J. Goniakowski, C. Noguera, Electronic states and Schottky barrier
height at metal/MgO(100) interfaces, Interface Sci. 12 (2004) 93.
[69] L.N. Pauling, The nature of the chemical bond, Cornell University,
Ithaca, NY, 1960.
[70] N.B. Hannay, C.P. Smyth, The dipole moment of hydrogen fluoride and
the ionic character of bonds, J. Am. Chem. Soc. 68 (1946) 171.
[71] U. Diebold, H.S. Tao, N.D. Shinn, T.E. Madey, Electronic structure of
ultrathin Fe films on TiO2 (110) studied with soft-x-ray photoelectron
spectroscopy and resonant photoemission, Phys. Rev. B 50 (1994)
14474.
[72] M. Brause, S. Skordas, V. Kempter, Study of the electronic structure
of TiO2 (110) and Cs/TiO2 (110) with metastable impact electron
spectroscopy and ultraviolet photoemission spectroscopy (HeI), Surf.
Sci. 445 (2000) 224.
[73] A.M. Ferrari, G. Pacchioni, Metal deposition on oxide surfaces: A
quantum-chemical study of the interaction of Rb, Pd, and Ag atoms with
the surface vacancies of MgO, J. Phys. Chem. 100 (1996) 9032.
[74] Y.F. Zhukovskii, E.A. Kotomin, P.W.M. Jacobs, A.M. Stoneham, An
initio modeling of metal adhesion on oxide surfaces with defects, Phys.
Rev. Lett. 84 (2000) 1256.
[75] P. Alemany, R.S. Boorse, J.M. Burlitch, R. Hoffmann, Metalceramic
adhesion: Quantum mechanical modeling of transition metalAl2 O3
interfaces, J. Phys. Chem. 97 (1993) 8464.
[76] H.-J. Freund, H. Kuhlenbeck, V. Staemmler, Oxide surfaces, Rep. Prog.
Phys. 59 (1996) 283.
[77] G. Pacchioni, Quantum chemistry of oxide surfaces: From CO
chemisorption to the identification of the structure and nature of point
defects on MgO, Surf. Rev. Lett. 7 (2000) 277.
[78] H. Gronbeck, First principles studies of metaloxide surfaces, Top.
Catal. 28 (2004) 59.
[79] T. Ioannides, X.E. Verykios, Charge transfer in metal catalysts supported
on doped TiO2 : A theoretical approach based on metalsemiconductor
contact theory, J. Catal. 161 (1996) 560.
[80] W. Gopel, L.J. Brillson, C.F. Brucker, Surface point defects and Schottky
barrier formation on ZnO(1010), J. Vac. Sci. Technol. 17 (1980) 894.
[81] K.H. Ernst, A. Ludviksson, R. Zhang, J. Yoshihara, C.T. Campbell,
Growth model for metal films on oxide surfaces: Cu on ZnO(0001)-O,
Phys. Rev. B 47 (1993) 13782.
[82] H.R. Sadeghi, V.E. Henrich, Electronic interactions in the rhodium/TiO2
system, J. Catal. 109 (1988) 1.
[83] K.D. Schierbaum, S. Fischer, P. Wincott, P. Hardman, V. Dhanak, G.
Jones, G. Thornton, Electronic structure of Pt overlayers on (1 3)
reconstructed TiO2 (100) surfaces, Surf. Sci. 391 (1997) 196.
[84] A.W. Grant, C.T. Campbell, Cesium adsorption on TiO2 (110), Phys.
Rev. B 55 (1997) 1844.
[85] Q. Fu, T. Wagner, On the tunability of chemical reactions at metal/oxide
interfaces, Surf. Sci. 574 (2005) L29.
[86] Y.W. Chung, W.B. Weissbard, Surface spectroscopy studies of the
SrTiO3 (100) surface and the platinumSrTiO3 (100) interface, Phys.
Rev. B 20 (1979) 3456.
[87] A.M. Stoneham, J.H. Harding, Computer simulation of interfaces: What
do we need to know? Acta Mater. 46 (1998) 2255.
[88] T. Wagner, Q. Fu, C. Winde, S. Tsukimoto, F. Phillipp, A comparative
study of the growth of Cr on (110)TiO2 rutile, (0001) -Al2 O3 and (100)
SrTiO3 surfaces, Interface Sci. 12 (2004) 117.
[89] Q. Fu, T. Wagner, Metaloxide interfacial reactions: Oxidation of metals
on TiO2 (110) and SrTiO3 (100), J. Phys. Chem. B 109 (2005) 11697.
[90] J. Marien, T. Wagner, G. Duscher, A. Koch, M. Ruhle, Nb on (110) TiO2
(rutile): Growth, structure, and chemical composition of the interface,
Surf. Sci. 446 (2000) 219.
[91] Q. Fu, T. Wagner, Thermal stability of Cr clusters on SrTiO3 (100), Surf.
Sci. 505 (2002) 39.
[92] B. Domenichini, A.M. Flank, P. Lagarde, S. Bourgeois, Interfacial
reaction between deposited molybdenum and TiO2 (110) surface: Role
of the substrate bulk stoichiometry, Surf. Sci. 560 (2004) 63.
[93] G. Blanco, J.J. Calvino, M.A. Cauqui, P. Corchado, C. Lopez Cartes, C.
Colliex, J.A. Perez-Omil, O. Stephan, Nanostructured evolution under
reducing conditions of a Pt/CeTbOx catalyst: A new alternative system
as a TWC component, Chem. Mater. 11 (1999) 3610.
487
[94] S. Bernal, J.J. Calvino, M.A. Cauqui, J.M. Gatica, C. Lopez Cartes,
J.A. Perez-Omil, J.M. Pintado, Some contributions of electron
microscopy to the characterization of the strong metalsupport
interaction effect, Catal. Today 77 (2003) 385.
[95] S. Bernal, J.J. Calvino, M.A. Cauqui, J.M. Gatica, C. Larese, J.A. PerezOmil, J.M. Pintado, Some recent results on metal/support interaction
effects in NM/CeO2 (NM: Noble metal) catalysts, Catal. Today 50
(1999) 175.
[96] S. Penner, D. Wang, D.S. Su, G. Rupprechter, R. Podloucky, R. Schlogl,
K. Hayek, Platinum nanocrystals supported by silica, alumina and
ceria: Metalsupport interaction due to high-temperature reduction in
hydrogen, Surf. Sci. 532535 (2003) 276.
[97] S. Penner, D. Wang, R. Podloucky, R. Schlogl, K. Hayek, Rh and Pt
nanoparticles supported by CeO2 : Metalsupport interaction upon hightemperature reduction observed by electron microscopy, Phys. Chem.
Chem. Phys. 6 (2004) 5244.
[98] D. Wang, S. Penner, D.S. Su, G. Rupprechter, K. Hayek, R. Schlogl,
Silicide formation on a Pt/SiO2 model catalyst studied by TEM, EELS,
and EDXS, J. Catal. 219 (2003) 434.
[99] R. Lamber, N.I. Jaeger, On the reaction of Pt with SiO2 substrates:
Observation of the Pt3 Si phase with the Cu3 Au superstructure, J. Appl.
Phys. 70 (1991) 457.
[100] B.K. Min, A.K. Santra, D.W. Goodman, Understanding silica-supported
metal catalysts: Pd/silica as a case study, Catal. Today 85 (2003) 113.
[101] N. Tsud, V. Johanek, I. Stara, K. Veltruska, V. Matoln, XPS, ISS and
TPD study of PdSn interactions on PdSnOx systems, Thin Solid Films
391 (2001) 204.
[102] X.A. Zhao, E. Kolawa, M.A. Nicolet, Reaction of thin metal films with
crystalline and amorphous Al2 O3 , J. Vac. Sci. Technol. A 4 (1986) 3139.
[103] R. Pretorius, J.M. Harris, M.A. Nicolet, Reaction of thin metal films with
SiO2 substrates, Solid State Electron. 21 (1978).
[104] Q. Zhong, F.S. Ohuchi, Surface science studies on the Ni/Al2 O3
interface, J. Vac. Sci. Technol. A 8 (1990) 2107.
[105] R. Raj, A. Saha, L. An, D.P.H. Hasselman, F. Ernst, Ion exchange at a
metalceramic interface, Acta Mater. 50 (2002) 1165.
[106] Y. Yu, J. Mark, F. Ernst, T. Wagner, R. Raj, Diffusion reactions at
AlMgAl2 O4 interfaces and the effect of applied electric fields, J. Mater.
Sci. 41 (2006) 7785.
[107] T.B. Reed, Free Energy of Formation of Binary Compounds, MIT press,
Cambridge, 1971.
[108] T. Wagner, A.D. Polli, G. Richter, H. Stanzick, Epitaxial growth of
metals on (100) SrTiO3 : the influence of lattice mismatch and reactivity,
Z. Metallk. 92 (2001) 701.
[109] M. Baumer, J. Biener, R.J. Madix, Growth, electronic properties and
reactivity of vanadium deposited onto a thin alumina film, Surf. Sci. 432
(1999) 189.
[110] J. Zhou, Y.C. Kang, D.A. Chen, Oxygen-induced dissociation of Cu
islands supported on TiO2 (110), J. Phys. Chem. B 107 (2003) 6664.
[111] Y. Gao, Y. Liang, S.A. Chambers, Thermal stability and the role of
oxygen vacancy defects in strong metal support interaction-Pt on Nbdoped TiO2 (100), Surf. Sci. 365 (1996) 638.
[112] E. Taglauer, H. Knozinger, Spreading and wetting, in: G. Ertl,
H. Knozinger, J. Weitkamp (Eds.), Handbook of Heterogeneous
Catalysis, VCH, Wenheim, 1997, p. 216.
[113] M. Backhaus-Ricoult, Solid-state reactivity at heterophase interfaces,
Annu. Rev. Mater. Res. 33 (2003) 55.
[114] A.T. Fromhold Jr., E.L. Cook, Diffusion currents in large electric fields
for discrete lattices, J. Appl. Phys. 38 (1967) 1546.
[115] N.F. Mott, The theory of the formation of protective oxide films on
metals. 3, Trans. Faraday Soc. 43 (1947) 429.
[116] N. Cabrera, N.F. Mott, Theory of the oxidation of metals, Rep. Prog.
Phys. 12 (1949) 163.
[117] A. Atkinson, Transport processes during the growth of oxide films at
elevated temperature, Rev. Modern. Phys. 57 (1985) 437.
[118] A.T. Fromhold Jr., Theory of Metal Oxidation Volume 1 Fundamentals,
North-Holland, Amsterdam, 1976.
[119] A.T. Fromhold Jr., E.L. Cook, Kinetics of oxide film growth on metal
crystals: Electron tunneling and ionic diffusion, Phys. Rev. 158 (1967)
600.
488
[120] A.T. Fromhold Jr., E.L. Cook, Kinetics of oxide film growth on metal
crystals: Thermal electron emission and ionic diffusion, Phys. Rev. 163
(1967) 650.
[121] I. Popova, V. Zhukov, J.T. Yates Jr., Electrostatic field enhancement of
Al(111) oxidation, Phys. Rev. Lett. 89 (2002) 276101.
[122] V. Zhukov, I. Popova, J.T. Yates Jr., Electron-stimulated oxidation
of Al(11) by oxygen at low temperatures: Mechanism of enhanced
oxidation kinetics, Phys. Rev. B 65 (2002) 195409.
[123] H.D. Ebinger, J.T. Yates Jr., Electron-impact-induced oxidation of
Al(111) in water vapor: Relation to the CabreraMott mechanism, Phys.
Rev. B 57 (1998) 1976.
[124] Y.Z. Hu, R. Sharangpani, S.P. Tay, Kinetic investigation of copper film
oxidation by spectroscopic ellipsometry and reflectometry, J. Vac. Sci.
Technol. A 18 (2000) 2527.
[125] H.F. Winters, J.W. Coburn, T.J. Chuang, Surface processes in plasmaassisted etching environments, J. Vac. Sci. Technol. B 1 (1983) 469.
[126] H.F. Winters, D. Haarer, Influence of doping on the etching of Si(111),
Phys. Rev. B 36 (1987) 6613.
[127] F.A. Houle, Photoeffects on the fluorination of silicon. I. Influence of
doping on steady-state phenomena, J. Chem. Phys. 79 (1983) 4237.
[128] L. Baldi, D. Beardo, Effects of doping on polysilicon etch rate in a
fluorine-containing plasma, J. Appl. Phys. 57 (1985) 2221.
[129] Y.H. Lee, M.M. Chen, A.A. Bright, Doping effects in reactive plasma
etching of heavily doped silicon, Appl. Phys. Lett. 46 (1985) 260.
[130] J.A. Yarmoff, F.R. McFeely, Effect of sample doping level during
etching of silicon by fluorine atoms, Phys. Rev. B 38 (1988) 2057.
[131] C.G. Van de Walle, F.R. McFeely, S.T. Pantelides, Fluorinesilicon
reactions and the etching of crystalline silicon, Phys. Rev. Lett. 61 (1988)
1867.
[132] C.W. Lo, D.K. Shuh, J.A. Yarmoff, Influence of electronic structure on
XeF2 etching of silicon, J. Vac. Sci. Technol. A 11 (1993) 2054.
[133] S.R. Qiu, H.F. Lai, J.A. Yarmoff, Self-limiting growth of metal fluoride
thin films by oxidation reactions employing molecular precursors, Phys.
Rev. Lett. 85 (2000) 1492.
[134] R.H. Kingston, S.F. Neustadter, Calculation of the space charge, electric
field, and free carrier concentration at the surface of a semiconductor, J.
Appl. Phys. 26 (1955) 718.
[135] T. Suzuki, S. Hishita, K. Oyoshi, R. Souda, Initial stage growth
mechanisms of metal adsorbates Ti, Zr, Fe, Ni, Ge, and Ag on
MgO(001) surface, Surf. Sci. 442 (1999) 291.
[136] G. Hass, A. Menck, H. Brune, J.V. Barth, J.A. Venables,
K. Kern, Nucleation and growth of supported clusters at defect
sites: Pd/MgO(001), Phys. Rev. B 61 (2000) 11105.
[137] M. Meunier, C.R. Henry, Nucleation and growth of metallic clusters on
MgO(100) by helium diffraction, Surf. Sci. 307309 (1994) 514.
[138] C. Goyhenex, M. Meunier, C.R. Henry, Limitation of Auger electron
spectroscopy in the determination of the metal-on-oxide growth mode:
Pd on MgO(100), Surf. Sci. 350 (1996) 103.
[139] G. Fahsold, A. Pucci, K.H. Rieder, Growth of Fe on MgO(001) studied
by He-atom scattering, Phys. Rev. B 61 (2000) 8475.
[140] S. Colonna, F. Arciprete, A. Balzarotti, M. Fanfoni, M. De Crescenzi,
S. Mobilio, In situ X-ray absorption measurements of the Cu/MgO(001)
interface, Surf. Sci. 512 (2002) L341.
[141] T. Kubo, H. Nozoye, Surface Structure of SrTiO3 (100)-(root5 root5)R26.6 , Phys. Rev. Lett. 86 (2001) 1801.
[142] T. Kubo, H. Nozoye, Surface structure of SrTiO3 (100), Surf. Sci. 542
(2003) 177.
[143] Q.D. Jiang, J. Zegenhagen, SrTiO3 (001)-c(6 2): A long-range,
atomically ordered surface stable in oxygen and ambient air, Surf. Sci.
367 (1996) L42.
[144] Q.D. Jiang, J. Zegenhagen, c(6 2) and c(4 2) reconstruction of
SrTiO3 (001), Surf. Sci. 425 (1999) 343.
[145] N. Erdman, K.R. Poeppelmeier, M. Asta, O. Warschkow, D.E. Ellis,
L.D. Marks, The structure and chemistry of the TiO2 -rich surface of
SrTiO3 (001), Nature 419 (2002) 55.
[146] N. Erdman, O. Warschkow, M. Asta, K.R. Poeppelmeier, D.E. Ellis,
L.D. Marks, Surface structures of SrTiO3 (001): A TiO2 -rich reconstruction with a c(4 2) unit cell, J. Am. Chem. Soc. 125 (2003) 10050.
[147] C. Barth, M. Reichling, Imaging the atomic arrangements on the hightemperature reconstructed -Al2 O3 (0001) surface, Nature 414 (2001)
54.
[148] G. Renaud, B. Villette, I. Vilfan, A. Bourret, Atomic structure of the
Al2 O3 (0001) ( 31 31)R 9 reconstruction, Phys. Rev. Lett. 73

(1994) 1825.
[149] M. Gautier, G. Renaud, L. Pham Van, B. Villette, M. Pollak, N. Thromat,
F. Jollet, J.P. Duraud, -Al2 O3 (0001) surfaces: Atomic and electronic
structure, J. Am. Ceram. Soc. 77 (1994) 323.
[150] T.M. French, G.A. Somorjai, Composition and surface structure of the
(0001) face of -alumina by low-energy electron diffraction, J. Phys.
Chem. 74 (1970) 2489.
[151] H. Maki, N. Ichinose, N. Ohashi, H. Haneda, J. Tanaka, The lattice
relaxation of ZnO single crystal (0001) surface, Surf. Sci. 457 (2000)
377.
[152] M. Kawasaki, K. Takahashi, T. Maeda, R. Tsuchiya, M. Shinohara,
O. Ishiyama, T. Yonezawa, M. Yoshimoto, H. Koinuma, Atomic control
of the SrTiO3 crystal surface, Science 266 (1994) 1540.
[153] A.D. Polli, T. Wanger, M. Ruhle, Effect of Ca impurities and wet
chemical etching on the surface morphology of SrTiO3 substrates, Surf.
Sci. 429 (1999) 237.
[154] S.H. Overbury, P.V. Radulovic, S. Thevuthasan, G.S. Herman,
M.A. Henderson, C.H.F. Peden, Ion scattering study of the Zn and
oxygen-terminated basal plane surfaces of ZnO, Surf. Sci. 410 (1998)
106.
[155] O. Dulub, L.A. Boatner, U. Diebold, STM study of the geometric and
electronic structure of ZnO(0001)-Zn, (0001)-O, (1010), and (1120)
surfaces, Surf. Sci. 519 (2002) 201.
[156] X.G. Wang, A. Chaka, M. Scheffler, Effect of the environment on Al2 O3 (0001) surface structures, Phys. Rev. Lett. 84 (2000) 3650.
[157] J. Maier, Ionic conduction in space charge regions, Prog. Solid State
Chem. 23 (1995) 171.
[158] R. Merkle, J. Maier, Oxygen incorporation into Fe-doped SrTiO3 :
Mechanistic interpretation of the surface reaction, Phys. Chem. Chem.
Phys. 4 (2002) 4140.
[159] H.-J. Freund, Metalsupport ultrathin oxide film systems as designable
catalysts and catalyst supports, Surf. Sci. 601 (2007) 1438.
[160] S.A. Chambers, Epitaxial growth and properties of thin film oxides, Surf.
Sci. Rep. 39 (2000) 105.
[161] A.K. Santra, D.W. Goodman, Oxide-supported metal clusters: Models
for heterogeneous catalysts, J. Phys.: Condens. Matter 14 (2002) R31.
[162] S. Surnev, M.G. Ramsey, F.P. Netzer, Vanadium oxide surface studies,
Prog. Surf. Sci. 73 (2003) 117.
[163] G.S. Herman, M.C. Gallagher, S.A. Joyce, C.H.F. Peden, Structure of
epitaxial thin TiOx films on W(110) as studied by low energy electron
diffraction and scanning tunneling microscopy, J. Vac. Sci. Technol. B
14 (1996) 1126.
[164] T.V. Ashworth, G. Thornton, Thin film TiO2 on nickel(110): An STM
study, Thin Solid Films 400 (2001) 43.
[165] N.D. McCavish, R.A. Bennett, Ultra-thin film growth of titanium dioxide
on W(100), Surf. Sci. 546 (2003) 47.
[166] U. Berner, K.D. Schierbaum, Cerium oxides and cerium-platinum
surface alloys on Pt(111) single-crystal surfaces studied by scanning
tunneling microscopy, Phys. Rev. B 65 (2002) 235404.
[167] W. Weiss, W. Ranke, Surface chemistry and catalysis on well-defined
epitaxial iron-oxide layers, Prog. Surf. Sci. 70 (2002) 1.
[168] W. Weiss, R. Schlogl, An integrated surface science approach towards
metal oxide catalysis, Top. Catal. 13 (2000) 75.
[169] K. Tanaka, B. Viswanathan, I. Toyoshima, CO adsorption suppression
due to charge transfer in the NiSiOx -n-Si(111) system at low Ni
coverage, J. Chem. Soc. Chem. Commun. (1985) 481.
[170] H. Ofner,
R. Hofmann, J. Kraft, F.P. Netzer, J.J. Paggel, K. Horn, Metaloverlayer-induced charge-transfer effects in thin SiO2 -Si structures,
Phys. Rev. B 50 (1994) 15120.
[171] T.J. Sarapatka,

XPS-XAES study of charge transfers at Ni/A12 O3 /Al
systems, Chem. Phys. Lett. 212 (1993) 37.
[172] T.J. Sarapatka,

Pd-induced charge transports with Pd/Al2 O3 /Al interface
formation, J. Phys. Chem. 97 (1993) 11274.

[173] A. Stierle, F. Renner, R. Streitel, H. Dosch, W. Drube, B.C. Cowie, Xray diffraction study of the ultrathin Al2 O3 layer on NiAl(110), Science
303 (2004) 1652.
[174] M. Ohring, The Materials Science of Thin Films, Academic Press,
London, 1992.
[175] J.A. Venables, G.D.T. Spiller, M. Hanbucken, Nucleation and growth of
thin films, Rep. Prog. Phys. 47 (1984) 399.
[176] J.A. Venables, Atomic processes in crystal growth, Surf. Sci. 299300
(1994) 798.
[177] Q. Fu, T. Wagner, Diffusion-corrected simultaneous multilayer growth
model, Phys. Rev. Lett. 90 (2003) 106105.
[178] Q. Fu, T. Wagner, Distinguishing film growth modes via spectroscopy:
Simple analytic models, Appl. Surf. Sci. 240 (2005) 189.
[179] U. Heiz, F. Vanolli, L. Trento, W.D. Schneider, Chemical reactivity
of size-selected supported clusters: An experimental setup, Rev. Sci.
Instrum. 68 (1997) 1986.
[180] U. Heiz, W.D. Schneider, Nanoassembled model catalysts, J. Phys. D:
Appl. Phys. 33 (2000) R85.
[181] C. Binns, Nanoclusters deposited on surfaces, Surf. Sci. Rep. 44 (2001)
1.
[182] K. Judai, S. Abbet, A.S. Worz, U. Heiz, C.R. Henry, Low-temperature
cluster catalysis, J. Am. Chem. Soc. 126 (2004) 2732.
[183] M. Aizawa, S. Lee, S.L. Anderson, Sintering, oxidation, and chemical
properties of size-selected nickel clusters on TiO2 (110), J. Chem. Phys.
117 (2002) 5001.
[184] M. Aizawa, S. Lee, S.L. Anderson, Deposition dynamics and chemical
properties of size-selected Ir clusters on TiO2 , Surf. Sci. 542 (2003) 253.
[185] M. Leskela, M. Ritala, Atomic layer deposition chemistry: Recent
developments and future challenges, Angew. Chem. Int. Ed. 42 (2003)
5548.
[186] M. Ritala, K. Kukli, A. Rahtu, P.I. Raisanen, M. Leskela, T. Sajavaara,
J. Keinonen, Atomic layer deposition of oxide thin films with metal
alkoxides as oxygen sources, Science 288 (2000) 319.
[187] D. Hausmann, J. Becker, S.L. Wang, R.G. Gordon, Rapid vapor
deposition of highly conformal silica nanolaminates, Science 298 (2002)
402.
[188] J.W. Klaus, O. Sneh, S.M. George, Growth of SiO2 at room temperature
with the use of catalyzed sequential half-reactions, Science 278 (1997)
1934.
[189] A.M. Lemonds, J.M. White, J.G. Ekerdt, Surface chemistry of TaCl5 on
polycrystalline Ta, Surf. Sci. 527 (2003) 124.
[190] A.M. Lemonds, J.M. White, J.G. Ekerdt, Surface science investigations
of atomic layer deposition half-reactions using TaF5 and Si2 H6 , Surf.
Sci. 538 (2003) 191.
[191] T. Aaltonen, P. Alen, M. Ritala, M. Leskela, Ruthenium thin films grown
by atomic layer deposition, Chem. Vapor Depos. 9 (2003) 45.
[192] Q. Wang, J.G. Ekerdt, D. Gay, Y.M. Sun, J.M. White, Low-temperature
chemical vapor deposition and scaling limit of ultrathin Ru films, Appl.
Phys. Lett. 84 (2004) 1380.
[193] T. Aaltonen, M. Ritala, T. Sajavaara, J. Keinonen, M. Leskela, Atomic
layer deposition of platinum thin films, Chem. Mater. 15 (2003) 1924.
[194] B.S. Lim, A. Rahtu, R.G. Gordon, Atomic layer deposition of transition
metals, Nat. Mater. 2 (2003) 749.
[195] P.L.J. Gunter, J.W. Niemantsverdriet, Surface science approach to
modeling supported catalysts, Catal. Rev. Sci. Eng. 39 (1997) 77.
[196] F.P. Netzer, Interfacial oxide layers at the metaloxide phase boundary,
Surf. Rev. Lett. 9 (2002) 1553.
[197] K. Hayek, M. Fuchs, B. Klotzer, W. Reichl, G. Rupprechter, Studies of
metalsupport interactions with real and inverted model systems:
Reactions of CO and small hydrocarbons with hydrogen on noble metals
in contact with oxides, Top. Catal. 13 (2000) 55.
[198] F.P. Leisenberger, S. Surnev, G. Koller, M.G. Ramsey, F.P. Netzer,
Probing the metal sites of a vanadium oxidePd(111) inverse catalyst:
Adsorption of CO, Surf. Sci. 444 (2000) 211.
[199] J. Schoiswohl, S. Eck, M.G. Ramsey, J.N. Anderson, S. Surnev,
F.P. Netzer, Vanadium oxide nanostructures on Rh(111): Promotion
effect of CO adsorption and oxidation, Surf. Sci. 580 (2005) 122.
489
[200] B. Jenewein, M. Fuchs, K. Hayek, The CO methanation on Rh/CeO2

and CeO2 /Rh model catalysts: A comparative study, Surf. Sci. 532535
(2003) 364.
[201] S. Eck, C. Castellarin-Cudia, S. Surnev, M.G. Ramsey, F.P. Netzer,
Growth and thermal properties of ultrathin cerium oxide layers on
Rh(111), Surf. Sci. 520 (2002) 173.
[202] S. Eck, C. Castellarin-Cudia, S. Surnev, K.C. Prince, M.G. Ramsey,
F.P. Netzer, Adsorption and reaction of CO on a ceriaRh(111) inverse
model catalyst surface, Surf. Sci. 536 (2003) 166.
[203] S. Surnev, G. Kresse, M.G. Ramsey, F.P. Netzer, Novel interfacemediated metastable oxide phases: Vanadium oxides on Pd(111), Phys.
Rev. Lett. 87 (2001) 086102.
[204] S. Surnev, J. Schoiswohl, G. Kresse, M.G. Ramsey, F.P. Netzer,
Reversible dynamic behavior in catalyst systems: Oscillations of
structure and morphology, Phys. Rev. Lett. 89 (2002) 246101.
[205] S. Surnev, M. Sock, G. Kresse, J.N. Andersen, M.G. Ramsey,
F.P. Netzer, Unusual CO adsorption sites on vanadium oxidePd(111)
inverse model catalyst surfaces, J. Phys. Chem. B 107 (2003) 4777.
[206] J. Schoiswohl, G. Kresse, S. Surnev, M. Sock, M.G. Ramsey,
F.P. Netzer, Planar vanadium oxide clusters: Two-dimensional evaporation and diffusion on Rh(111), Phys. Rev. Lett. 92 (2004) 206103.
[207] M.G. Mason, Electronic structure of supported small metal clusters,
Phys. Rev. B 27 (1983) 748.
[208] P.H. Citrin, G.K. Wertheim, Y. Baer, Core level binding energy and
density of states from the surface atoms of gold, Phys. Rev. Lett. 41
(1978) 1425.
[209] P.H. Citrin, G.K. Wertheim, Photoemission from surface-atom core
levels, surface densities of states, and metalatom clusters: A unified
picture, Phys. Rev. B 27 (1983) 3176.
[210] P.S. Bagus, C.R. Brundle, G. Pacchioni, F. Parmigiani, Mechanisms
responsible for the shifts of core-level binding energies between surface
and bulk atoms of metals, Surf. Sci. Rep. 19 (1993) 265.
[211] P.S. Bagus, F. Illas, G. Pacchioni, F. Parmigiani, Mechanisms responsible
for chemical shifts of core-level binding energies and their relationship
to chemical bonding, J. Electron. Spectrosc. Relat. Phenom. 100 (1999)
215.
[212] W.F. Egelhoff Jr., Core-level binding-energy shifts at surfaces and in
solids, Surf. Sci. Rep. 6 (1987) 253.
[213] G.K. Wertheim, S.B. DiCenzo, S.E. Youngquist, Unit charge on
supported gold clusters in photoemission final state, Phys. Rev. Lett. 51
(1983) 2310.
[214] G.K. Wertheim, S.B. DiCenzo, D.N.E. Buchanan, Noble- and transitionmetal clusters: The d bands of silver and palladium, Phys. Rev. B 33
(1986) 5384.
[215] G.K. Wertheim, S.B. DiCenzo, Cluster growth and core-electron binding
energies in supported metal clusters, Phys. Rev. B 37 (1988) 844.
[216] G.K. Wertheim, Electronic structure of metal clusters, Z. Phys. D 12
(1989) 319.
[217] C. Kuhrt, M. Harsdorff, Photoemission and electron microscopy of small
supported palladium clusters, Surf. Sci. 245 (1991) 173.
[218] Q. Fu, T. Wagner, The interaction of ultrathin Cr layers with
SrTiO3 (100), Surf. Sci. 601 (2007) 1339.
[219] H.P. Steinruck, F. Pesty, L. Zhang, T.E. Madey, Ultrathin films of Pt on
TiO2 (110): Growth and chemisorption-induced surfactant effects, Phys.
Rev. B 51 (1995) 2427.
[220] K. Luo, T.P.S. Clair, X. Lai, D.W. Goodman, Silver growth on TiO2 (110)
(1 1) and (1 2), J. Phys. Chem. B 104 (2000) 3050.
[221] M.K. Bahl, S.C. Tsai, Y.W. Chung, Auger and photoemission
investigations of the platinumSrTiO3 (100) interface: Relaxation and
chemical-shift effects, Phys. Rev. B 21 (1980) 1344.
[222] C.D. Wagner, Chemical-shift of Auger lines, and Auger parameter,
Faraday Discuss. Chem. Soc. 60 (1975) 291.
[223] S.W. Gaarenstroom, N. Winograd, Initial and final state effects in the
ESCA spectra of cadmium and silver oxides, J. Chem. Phys. 67 (1977)
3500.
[224] D. Briggs, M.P. Seah, Practical Surface Analysis by Auger and X-ray
Photoelectron Spectroscopy, 2nd ed., John Wiley, Chichester, 1990.
490
[225] G. Moretti, Auger parameter and Wagner plot in the characterization

of chemical states by X-ray photoelectron spectroscopy: A review, J.
Electron. Spectrosc. Relat. Phenom. 95 (1998) 95.
[226] G.K. Wertheim, Auger shifts in metal clusters, Phys. Rev. B 36 (1987)
9559.
[227] C.C. Kao, S.C. Tsai, M.K. Bahl, Y.W. Chung, Electronic properties,
structure and temperature-dependent composition of nickel deposited on
rutile titanium dioxide (110) surfaces, Surf. Sci. 95 (1980) 1.
[228] P. Zurcher, R.S. Bauer, Photoemission determination of dipole layer
and VB-discontinuity formation during the MBE growth of GaAs on
Ge(110), J. Vac. Sci. Technol. A 1 (1983) 695.
[229] C. Argile, G.E. Rhead, Adsorbed layer and thin film growth modes
monitored by Auger electron spectroscopy, Surf. Sci. Rep. 10 (1989)
277.
[230] T. Wagner, J.Y. Wang, S. Hofmann, in: D. Briggs, J.T. Grant (Eds.),
Surface Analysis by Auger and X-ray Photoelectron Spectroscopy,
IMPublications, Chichester, 2003.
[231] H.R. Sadeghi, V.E. Henrich, Rh on TiO2 : Model catalyst studies of the
strong metalsupport interaction, Appl. Surf. Sci. 19 (1984) 330.
[232] S. Bernath, T. Wagner, S. Hofmann, M. Ruhle, Interface formation
between ultrathin films of titanium and (0001) sapphire substrates, Surf.
Sci. 400 (1998) 335.
[233] R. Fuchs, K.L. Kliewer, Optical modes of vibration in an ionic crystal
slab, Phys. Rev. 140 (1965) A2076.
[234] Z. Chang, S. Piligkos, P.J. Mller, High-resolution electron-energy-loss
spectroscopy of vanadium and vanadium oxide thin films on TiO2 (110)(1 1), Phys. Rev. B 64 (2001) 165410.
and
[235] W.T. Petrie, J.M. Vohs, Interaction of platinum films with the (0001)
(0001) surfaces of ZnO, J. Chem. Phys. 101 (1994) 8098.
[236] C. Xu, X. Lai, G.W. Zajac, D.W. Goodman, Scanning tunneling
microscopy studies of the TiO2 (110) surfaces: Structure and the
nucleation growth of Pd, Phys. Rev. B 56 (1997) 13464.
[237] Q. Fu, T. Wagner, Nucleation and growth of Cr clusters and films on
(100) SrTiO3 surfaces, Thin Solid Films 420421 (2002) 455.
[238] D.A. Chen, M.C. Bartelt, R.Q. Hwang, K.F. McCarty, Self-limiting
growth of copper islands on TiO2 (110)-(1 1), Surf. Sci. 450 (2000)
78.
[239] D.A. Chen, M.C. Bartelt, S.M. Seutter, K.F. McCarty, Small, uniform,
and thermally stable silver particles on TiO2 (110)-(1 1), Surf. Sci. 464
(2000) L708.
[240] X. Lai, T.P. St Clair, M. Valden, D.W. Goodman, Scanning tunneling
microscopy studies of metal clusters supported on TiO2 (110):
Morphology and electronic structure, Prog. Surf. Sci. 59 (1998) 25.
[241] A. Berko, F. Solymosi, Effects of different gases on the morphology of
IR nanoparticles supported on the TiO2 (110)-(1 2) surface, J. Phys.
Chem. B 104 (2000) 10215.
[242] A. Berko, J. Szoko, F. Solymosi, Effect of CO on the morphology of Pt
nanoparticles supported on TiO2 (110)-(1n), Surf. Sci. 566568 (2004)
337.
[243] J. Libuda, H.-J. Freund, Molecular beam experiments on model catalysts,
Surf. Sci. Rep. 57 (2005) 157.
[244] A.K. Santra, B.K. Min, D.W. Goodman, Ag clusters on ultra-thin,
ordered SiO2 films, Surf. Sci. 515 (2002) L475.
[245] M. Valden, X. Lai, D.W. Goodman, Onset of catalytic activity of gold
clusters on titania with the appearance of nonmetallic properties, Science
281 (1998) 1647.
[246] C. Xu, D.W. Goodman, Morphology and local electronic structure
of metal particles on metal oxide surfaces: A scanning tunneling
microscopic and scanning tunneling spectroscopic study, Chem. Phys.
Lett. 263 (1996) 13.
[247] J. Szoko, A. Berko, Tunneling spectroscopy of Pt nanoparticles
supported on TiO2 (110) surface, Vacuum 71 (2003) 193.
[248] K. Fukui, H. Onishi, Y. Iwasawa, Atom-resolved image of the TiO2 (110)
surface by noncontact atomic force microscopy, Phys. Rev. Lett. 79
(1997) 4202.
[249] C. Barth, C.R. Henry, Atomic resolution imaging of the (001) surface of
UHV cleaved MgO by dynamic scanning force microscopy, Phys. Rev.
Lett. 91 (2003) 196102.
[250] H. Poppa, Nucleation, growth, and TEM analysis of metal particles and
clusters deposited in UHV, Catal. Rev. Sci. Eng. 35 (1993) 359.
[251] A.D. Polli, T. Wagner, T. Gemming, M. Ruhle, Growth of platinum on
TiO2 - and SrO-terminated SrTiO3 (100), Surf. Sci. 448 (2000) 279.
[252] G. Rupprechter, H. Hayek, H. Hofmeister, Electron microscopy of
thin-film model catalysts: Activation of alumina-supported rhodium
nanoparticles, J. Catal. 173 (1998) 409.
[253] S. Bernal, F.J. Botana, J.J. Calvino, C. Lopez, J.A. Perez-Omil,
J.M. Rodrguez-Izquierdo, High-resolution electron microscopy investigation of metalsupport interactions in Rh/TiO2 , J. Chem. Soc. Faraday
Trans. 92 (1996) 2799.
[254] S. Bernal, J.J. Calvino, J.M. Gatica, C. Larese, C. Lopez-Cartes,
J.A. Perez-Omil, Nanostructural evolution of a Pt/CeO2 catalyst reduced
at increasing temperatures (4731223 K): A HREM study, J. Catal. 169
(1997) 510.
[255] R. Schweinfest, S. Kostlmeier, F. Ernst, C. Elsasser, T. Wagner,
Atomistic and electronic structure of Al/MgAl2 O4 and Ag/MgAl2 O4
interfaces, Philos. Mag. A 81 (2001) 927.
[256] W. Sigle, Analytic transmission electron microscopy, Annu. Rev. Mater.
Res. 35 (2005) 325.
[257] C. Scheu, G. Dehm, M. Ruhle, R. Brydson, Electron-energy-loss
spectroscopy studies of Cu--Al2 O3 interfaces grown by molecular
beam epitaxy, Philos. Mag. A 78 (1998) 439.
[258] C. van Benthem, C. Scheu, W. Sigle, M. Ruhle, Electronic structure
investigation of Ni and Cr films on (100) SrTiO3 substrate using electron
energy-loss spectroscopy, Z. Metallk. 93 (2002) 362.
[259] F. Pesty, H.P. Steinruck, T.E. Madey, Thermal stability of Pt films on
TiO2 (110): Evidence for encapsulation, Surf. Sci. 339 (1995) 83.
[260] S. Labich, E. Taglauer, H. Knozinger, Metalsupport interactions on
rhodium model catalysts, Top. Catal. 14 (2001) 153.
[261] N. Kasper, A. Stierle, P. Nolte, Y. Jin-Phillipp, T. Wanger,
D.G. de Oteyza, H. Dosch, In situ oxidation study of MgO(100)
supported Pd nanoparticles, Surf. Sci. 600 (2006) 2860.
[262] G. Renaud, R. Lazzari, C. Revenant, A. Barbier, M. Noblet, O. Ulrich,
F. Leroy, J.P. Jupille, Y. Borensztein, C.R. Henry, J.P. Deville, F.
Scheurer, J. Mane-Mane, O. Fruchart, Real-time monitoring of growing
nanoparticles, Science 300 (2003) 1416.
[263] Q. Fu, E. Tchernychova, T. Wagner, Texture study of molybdenum thin
films on SrTiO3 (100): A RHEED study, Surf. Sci. 538 (2003) L511.
[264] V. Oderno, C. Dufour, K. Dumesnil, A. Mougin, P. Mangin, G. Marchal,
Hexagonal surface structure during the first stages of niobium growth on
sapphire(1120), Phil. Mag. Lett. 78 (1998) 419.
[265] B. Yoon, H. Hakkinen, U. Landman, A.S. Woerz, J.M. Antonietti,
S. Abbet, K. Judai, U. Heiz, Charging effects on bonding and catalyzed
oxidation of CO on Au8 clusters on MgO, Science 307 (2005) 403.
[266] M. Frank, M. Baumer, R. Kuhnemuth, H.-J. Freund, Metal atoms and
particles on oxide supports: Probing structure and charge by infrared
spectroscopy, J. Phys. Chem. B 105 (2001) 8569.
[267] J. Maier, Nanoionics: Ion transport and electrochemical storage in
confined systems, Nat. Mater. 4 (2005) 805.
[268] S. Mrowec, Defects and Diffusion in Solids, Elsevier, 1980.
[269] J. Maier, Physical Chemistry of Ionic Materials-Ions and Electrons in
Solids, John Wiley & Sons Ltd., 2004.
[270] U. Diebold, J. Lehman, T. Mahmoud, M. Kuhn, G. Leonardelli,
W. Hebenstreit, M. Schmid, P. Varga, Intrinsic defects on a TiO2 (110)(1 1) surface and their reaction with oxygen: A scanning tunneling
microscopy study, Surf. Sci. 411 (1998) 137.
[271] N. Lopez, J.K. Nrskov, Theoretical study of the Au/TiO2 (110)
interface, Surf. Sci. 515 (2002) 175.
[272] H. Onishi, T. Aruga, C. Egawa, Y. Iwasawa, Modification of surface
electronic structure on TiO2 (110) and TiO2 (441) by Na deposition, Surf.
Sci. 199 (1988) 54.
[273] P. Lagarde, A.M. Flank, R.J. Prado, S. Bourgeois, J. Jupille, The defined
adsorption site of sodium on the TiO2 (110)-(1 1) surface, Surf. Sci.
553 (2004) 115.
[274] P.W. Murray, N.G. Condon, G. Thornton, Na adsorption sites on
TiO2 (110)-1 2 and its 2 2 superlattice, Surf. Sci. 323 (1995) L281.

[275] J. Nerlov, S.V. Christensen, S. Weichel, E.H. Pedersen, P.J. Mller,
A photoemission study of the coadsorption of CO2 and Na on
TiO2 (110)-(1 1) and -(1 2) surfaces: Adsorption geometry and
reactivity, Surf. Sci. 371 (1997) 321.
[276] C.L. Pang, C.A. Muryn, A.P. Woodhead, H. Raza, S.A. Haycock,
V. Dhanak, G. Thornton, Low-coverage condensation of K on TiO2 (110)
1 1, Surf. Sci. 583 (2005) L147.
[277] O. Bikondoa, C.L. Pang, C.A. Muryn, B.G. Daniels, S. Ferrero,
E. Michelangeli, G. Thornton, Ordered overlayers of Ca on TiO2 (110)-1
1, J. Phys. Chem. B 108 (2004) 16768.
[278] M.A. San Miguel, C.J. Calzado, J.F. Sanz, Modeling alkali atoms
deposition on TiO2 (110) surface, J. Phys. Chem. B 105 (2001) 1794.
[279] T. Bredow, E. Apr`a, M. Catti, G. Pacchioni, Cluster and periodic abinitio calculations on K/TiO2 (110), Surf. Sci. 418 (1998) 150.
[280] S. Agnoli, C. Castellarin-Cudia, M. Sambi, S. Surnev, M.G. Ramsey,
G. Granozzi, F.P. Netzer, Vanadium on TiO2 (110): Adsorption site and
sub-surface migration, Surf. Sci. 546 (2003) 117.
[281] B. Domenichini, S. Petigny, V. Blondeau-Patissier, A. Steinbrunn,
S. Bourgeois, Effect of the surface stoichiometry on the interaction of
Mo with TiO2 (110), Surf. Sci. 468 (2000) 192.
[282] R. Heise, R. Courths, A photoemission investigation of the adsorption
of potassium on perfect and defective TiO2 (110) surfaces, Surf. Sci.
331333 (1995) 1460.
[283] B. Demri, M. Hage-Ali, M. Moritz, J.L. Kahn, D. Muster, X-ray
photoemission study of the calcium/titanium dioxide interface, Appl.
Surf. Sci. 108 (1997) 245.
[284] J.M. Pan, U. Diebold, L.Z. Zhang, T.E. Madey, Ultrathin reactive
metal films on TiO2 (110): Growth, interfacial interaction and electronic
structure of chromium films, Surf. Sci. 295 (1993) 411.
[285] J.M. Pan, T.E. Madey, Ultrathin Fe films on TiO2 (110): Growth and
reactivity, J. Vac. Sci. Technol. A 11 (1993) 1667.
[286] H. Mostefa-Sba, B. Domenichini, S. Bourgeois, Iron deposition on
TiO2 (110): Effect of the surface stoichiometry and roughness, Surf. Sci.
437 (1999) 107.
[287] A. Vijay, G. Mills, H. Metiu, Adsorption of gold on stoichiometric and
reduced rutile TiO2 (110) surfaces, J. Chem. Phys. 118 (2003) 6536.
[288] Z. Yang, R.Q. Wu, D.W. Goodman, Structural and electronic properties
of Au on TiO2 (110), Phys. Rev. B 61 (2000) 14066.
[289] X. Tong, L. Benz, P. Kemper, H. Metiu, M.T. Bowers, S.K. Buratto,
Intact size-selected Aun clusters on a TiO2 (110)-(1 1) surface at room
temperature, J. Am. Chem. Soc. 127 (2005) 13516.
[290] C. Su, J.C. Yeh, J.L. Lin, J.C. Lin, The growth of Ag films on a
TiO2 (110)-(1 1) surface, Appl. Surf. Sci. 169170 (2001) 366.
[291] J.A. Horsley, A molecular orbital study of strong metalsupport
interaction between platinum and titanium dioxide, J. Am. Chem. Soc.
101 (1979) 2870.
[292] W.X. Xu, K.D. Schierbaum, W. Goepel, Ab initio study of the effect of
oxygen defect on the strong-metalsupport interaction between Pt and
TiO2 (Rutile)(110) surface, J. Solid State Chem. 119 (1995) 237.
[293] S. Takakusagi, K. Fukui, R. Tero, F. Nariyuki, Y. Iwasawa, Self-limiting
growth of Pt nanoparticles from MeCpPtMe3 adsorbed on TiO2 (110)
studied by scanning tunneling microscopy, Phys. Rev. Lett. 91 (2003)
066102.
[294] K.D. Schierbaum, S. Fischer, M.C. Torquemada, J.L. de Segovia,
E. Roman, J.A. Martn-Gago, The interaction of Pt with TiO2 (110)
surfaces: A comparative XPS, UPS, ISS, and ESD study, Surf. Sci. 345
(1996) 261.
[295] S. Fischer, K.D. Schierbaum, W. Gopel, Surface defects and platinum
overlayers on TiO2 (110) surfaces: STM and photoemission studies,
Vacuum 48 (1997) 601.
[296] V.E. Henrich, G. Dresselhaus, H.J. Zeiger, Observation of twodimensional phases associated with defect states on the surface of TiO2 ,
Phys. Rev. Lett. 36 (1976) 1335.
[297] M.A. Henderson, The interaction of water with solid surfaces:
Fundamental aspects revisited, Surf. Sci. Rep. 46 (2002) 1.
491
[298] S. Wendt, R. Schaub, J. Matthiesen, E.K. Vestergaard, E. Wahlstrom,

M.D. Rasmussen, P. thostrup, L.M. Molina, E. Lgsgaard, I. Stensgaard,
B. Hammer, F. Besenbacher, Oxygen vacancies on TiO2 (110) and their
interaction with H2 O and O2 : A combined high-resolution STM and
DFT study, Surf. Sci. 598 (2005) 226.
[299] S. Wendt, J. Matthiesen, R. Schaub, E.K. Vestergaard, E. Lgsgaard,
F. Besenbacher, B. Hammer, Formation and splitting of paired hydroxyl
groups on reduced TiO2 (110), Phys. Rev. Lett. 96 (2006) 066107.
[300] Z. Zhang, O. Bondarchuk, B.D. Kay, J.M. White, Z. Dohnalek, Imaging
water dissociation on TiO2 (110): Evidence for inequivalent geminate
OH groups, J. Phys. Chem. B 110 (2006) 21840.
[301] H. Onishi, Y. Iwasawa, Dynamic visualization of a metaloxidesurface/gas-phase reaction: Time-resolved observation by scanning
tunneling microscopy at 800 K, Phys. Rev. Lett. 76 (1996) 791.
[302] R.E. Tanner, M.R. Castell, G.A.D. Briggs, High resolution scanning
tunneling microscopy of the rutile TiO2 (110) surface, Surf. Sci. 412/413
(1998) 672.
[303] H. Onishi, H. Iwasaki, Reconstruction of TiO2 (110) surface: STM study
with atomic-scale resolution, Surf. Sci. 313 (1994) L783.
[304] C.L. Pang, S.A. Haycock, H. Raza, P.W. Murray, G. Thornton,
O. Gulseren, R. James, D.W. Bullett, Added row model of TiO2 (110)
1 2, Phys. Rev. B 58 (1998) 1586.
[305] X. Lai, C. Xu, D.W. Goodman, Synthesis and structure of Al clusters
supported on TiO2 (110): A scanning tunneling microscopy study, J. Vac.
Sci. Technol. A 16 (1998) 2562.
[306] J. Biener, J. Wang, R.J. Madix, Direct observation of the growth of
vanadium on TiO2 (110)-(1 2), Surf. Sci. 442 (1999) 47.
[307] L. Benz, X. Tong, P. Kemper, Y. Lilach, A. Kolmakov, H. Metiu,
M.T. Bowers, S.K. Buratto, Landing of size-selected Ag+
n clusters on
single crystal TiO2 (110)-(1 1) surfaces at room temperatures, J. Chem.
Phys. 122 (2005) 081102.
[308] A.K. Santra, F. Yang, D.W. Goodman, The growth of AgAu bimetallic
nanoparticles on TiO2 (110), Surf. Sci. 548 (2004) 324.
[309] E. Wahlstrom, N. Lopez, R. Schaub, P. Thostrup, A. Rnnau,
C. Africh, E. Lgsgaard, J.K. Nrskov, F. Besenbacher, Bonding of gold
nanoclusters to oxygen vacancies on rutile TiO2 (110), Phys. Rev. Lett.
90 (2003) 026101.
[310] T. Minato, T. Susaki, S. Shiraki, H.S. Kato, M. Kawai, K. Aika,
Investigation of the electronic interaction between TiO2 (110) surfaces
and Au clusters by PES and STM, Surf. Sci. 566568 (2004) 1012.
[311] Y. Maeda, T. Fujitani, S. Tsubota, M. Haruta, Size and density of Au
particles deposited on TiO2 (110)-(1 1) and cross-linked (1 2)
surfaces, Surf. Sci. 562 (2004) 1.
[312] M.J.J. Jak, C. Konstapel, A. van Kreuningen, J. Chrost, J. Verhoeven,
J.W.M. Frenken, The influence of substrate defects on the growth rate of
palladium nanoparticles on a TiO2 (110) surface, Surf. Sci. 474 (2001)
28.
[313] X. Tong, L. Benz, A. Kolmakov, S. Chretien, H. Metiu, S.K. Buratto,
The nucleation sites of Ag clusters grown by vapor deposition on a
TiO2 (110)-1 1 surface, Surf. Sci. 575 (2005) 60.
[314] J. Zhou, D.A. Chen, Controlling size distributions of copper islands
grown on TiO2 (110)-(1 2), Surf. Sci. 527 (2003) 183.
[315] J.R. Kitchin, M.A. Barteau, J.G. Chen, A comparison of gold and
molybdenum nanoparticles on TiO2 (110) 1 2 reconstructed single
crystal surfaces, Surf. Sci. 526 (2003) 323.
[316] A. Berko, G. Menesi, F. Solymosi, STM study of rhodium deposition on
the TiO2 (110)-(1 2) surface, Surf. Sci. 372 (1997) 202.
[317] A. Berko, A.M. Kiss, J. Szoko, Formation of vacancy islands tailored by
Pt nanocrystallites and Ar+ sputtering on TiO2 (110) surface, Appl. Surf.
Sci. 246 (2005) 174.
[318] A.S. Worz, U. Heiz, F. Cinquini, G. Pacchioni, Charging of Au atoms on
TiO2 thin films from CO vibrational spectroscopy and DFT calculations,
J. Phys. Chem. B 10 (2005) 18418.
[319] D. Matthey, J.G. Wang, S. Wendt, J. Matthiesen, R. Schaub,
E. Lgsgaard, B. Hammer, F. Besenbacher, Enhanced bonding of gold
nanoparticles on oxidized TiO2 (110), Science 315 (2007) 1692.
[320] J. Sasaki, N.L. Peterson, K. Hoshino, Tracer impurity diffusion in singlecrystal rutile (TiO2x ), J. Phys. Chem. Solids 46 (1985) 1267.
492
[321] M. Li, W. Hebenstreit, U. Diebold, A.M. Tyryshkin, M.K. Bowman,

G.G. Dunham, M.A. Henderson, The influence of the bulk reduction
state on the surface structure and morphology of rutile TiO2 (110) single
crystals, J. Phys. Chem. B 104 (2000) 4944.
[322] T. Sekiya, T. Yagisawa, N. Kamiya, D. Das Mulmi, S. Kurita,
Y. Murakami, T. Kodaira, Defects in anatase TiO2 single crystal
controlled by heat treatment, J. Phys. Soc. Jpn. 73 (2004) 703.
[323] M. Batzill, K. Katsiev, D.J. Gaspar, U. Diebold, Variations of the local
electronic surface properties of TiO2 (110) induced by intrinsic and
extrinsic defects, Phys. Rev. B 66 (2002) 235401.
[324] D. Morris, Y. Dou, J. Rebane, C.E.J. Mitchell, R.G. Egdell, D.S.L. Law,
Photoemission and STM study of the electronic structure of Nb-doped
TiO2 , Phys. Rev. B 61 (2000) 13445.
[325] S.A. Chambers, Y. Gao, Y.J. Kim, M.A. Henderson, S. Thevuthasan,
S. Wen, K.L. Merkle, Geometric and electronic structure of epitaxial
Nbx Ti1x O2 on TiO2 (110), Surf. Sci. 365 (1996) 625.
[326] A. Berko, I. Ulrych, K.C. Prince, Encapsulation of Rh nanoparticles
supported on TiO2 (110)-(1 1) surface: XPS and STM studies, J. Phys.
Chem. B 102 (1998) 3379.
[327] D.M. Smyth, The effects of dopants on the properties of metal oxides,
Solid State Ion. 129 (2000) 5.
[328] S. Bourgeois, P. le Seigneur, M. Perdereau, Study by XPS of ultra-thin
nickel deposits on TiO2 (100) supports with different stoichiometries,
Surf. Sci. 328 (1995) 105.
[329] R.A. Bennett, P. Stone, M. Bowker, Pd nanoparticle enhanced reoxidation of non-stoichiometric TiO2 : STM imaging of spillover and a
new form of SMSI, Catal. Lett. 59 (1999) 99.
[330] R.A. Bennett, C.L. Pang, N. Perkins, R.D. Smith, P. Morrall,
R.I. Kvon, M. Bowker, Surface structures in the SMSI state; Pd on (1
2) reconstructed TiO2 (110), J. Phys. Chem. B 106 (2002) 4688.
[331] R.C. Weast, M.J. Astle (Eds.), CRC Handbook of Chemistry and
Physics, CRC Press, 1982.
[332] L.Z. Mezey, J. Giber, The surface free energies of solid chemical
elements: Calculation from internal free enthalpies of atomization, Jpn.
J. Appl. Phys. 21 (1982) 1569.
[333] W.R. Tyson, W.A. Miller, Surface free energies of solid metals:
Estimation from liquid surface tension measurements, Surf. Sci. 62
(1977) 267.
[334] S.H. Overbury, P.A. Bertrand, G.A. Somorjai, The surface composition
of binary systems. Prediction of surface phase diagrams of solid
solutions, Chem. Rev. 75 (1975) 547.
[335] L.S. Dake, R.J. Lad, Electronic and chemical interactions at
aluminium/TiO2 (110) interfaces, Surf. Sci. 289 (1993) 297.
[336] J. Wang, J. Biener, R.J. Madix, Temperature effects on vanadium
overlayers on the TiO2 (110)-(1 2) surface, J. Phys. Chem. B 104
(2000) 3286.
[337] V. Blondeau-Patissier, B. Domenichini, A. Steinbrunn, S. Bourgeois,
MoOx (x < 2) ultrathin film growth from reactions between metallic
molybdenum and TiO2 surfaces, Appl. Surf. Sci. 175-176 (2001) 674.
[338] R.A. Bennett, P. Stone, N.J. Price, M. Bowker, Two (1 2)
reconstructions of TiO2 (110): Surface rearrangement and reactivity
studied using elevated temperature scanning tunneling microscopy,
Phys. Rev. Lett. 82 (1999) 3831.
[339] M.A. Henderson, Mechanism for the bulk-assisted reoxidation of ion
sputtered TiO2 surfaces: Diffusion of oxygen to the surface or titanium
to the bulk? Surf. Sci. 343 (1995) L1156.
[340] M.A. Henderson, A surface perspective on self-diffusion in rutile TiO2 ,
Surf. Sci. 419 (1999) 174.
[341] M. Li, W. Hebenstreit, U. Diebold, Oxygen-induced restructuring of the
rutile TiO2 (110) (1 1) surface, Surf. Sci. 414 (1998) L951.
[342] R.J. Lad, L.S. Dake, Electronic and structural properties of interfaces
created by potassium deposited on TiO2 (110) surfaces, Mat. Res. Soc.
Symp. Proc. 238 (1992) 823.
[343] Z.S. Li, J.H. Jrgensen, P.J. Mller, M. Sambi, G. Granozzi,
A photoemission and resonant photoemission study of Ba deposition at
the TiO2 (110) surface, Appl. Surf. Sci. 142 (1999) 135.
[344] M. Brause, V. Kempter, Mg interaction with TiO2 (100): MIES and UPS
(HeI) results, Surf. Sci. 490 (2001) 153.
[345] G. Rocker, W. Gopel, Titanium overlayers on TiO2 (110), Surf. Sci. 181
(1987) 530.
[346] J.T. Mayer, U. Diebold, T.E. Madey, E. Garfunkel, Titanium and reduced
titania overlayers on titanium dioxide(110), J. Electron. Spectrosc. Relat.
Phenom. 73 (1995) 1.
[347] Z.M. Zhang, V.E. Henrich, Electronic interactions in the
vanadium/TiO2 (110) and vanadia/TiO2 model catalyst systems,
Surf. Sci. 277 (1992) 263.
[348] J. Biener, M. Baumer, J. Wang, R.J. Madix, Electronic structure and
growth of vanadium on TiO2 (110), Surf. Sci. 450 (2000) 12.
[349] J.M. Pan, B.L. Maschhoff, U. Diebold, T.E. Madey, Structural study of
ultrathin metal films on TiO2 using LEED, ARXPS and MEED, Surf.
Sci. 291 (1993) 381.
[350] A.K. See, R.A. Bartynski, Electronic properties of ultrathin Cu
and Fe films on TiO2 (110) studied by photoemission and inverse
photoemission, Phys. Rev. B 50 (1994) 12064.
[351] V. Blondeau-Patissier, G.D. Lian, B. Domenichini, A. Steinbrunn,
S. Bourgeois, E.C. Dickey, Molybdenum thin-film growth on rutile
titanium dioxide (110), Surf. Sci. 506 (2002) 119.
[352] B. Domenichini, M. Petukhov, G.A. Rizzi, M. Sambi, S. Bourgeois,
G. Granozzi, Epitaxial growth of molybdenum on TiO2 (110), Surf. Sci.
544 (2003) 135.
[353] N. Nilius, N. Ernst, H.-J. Freund, On energy transfer processes at
clusteroxide interfaces: Silver on titania, Chem. Phys. Lett. 349 (2001).
[354] U. Diebold, J.M. Pan, T.E. Madey, Growth mode of ultrathin copper
overlayers on TiO2 (110), Phys. Rev. B 47 (1993) 3868.
[355] L. Zhang, F. Cosandey, R. Persaud, T.E. Madey, Initial growth and
morphology of thin Au films on TiO2 (110), Surf. Sci. 439 (1999) 73.
[356] F. Cosandey, L. Zhang, T.E. Madey, Effect of substrate temperature on
the epitaxial growth of Au on TiO2 (110), Surf. Sci. 474 (2001) 1.
[357] J. Zhou, S. Ma, Y.C. Kang, D.A. Chen, Dimethyl methylphosphonate
decomposition on titania-supported Ni clusters and films: A comparison
of chemical activity on different Ni surfaces, J. Phys. Chem. B 108
(2004) 11633.
[358] H. Onishi, T. Aruga, C. Egawa, Y. Iwasawa, Photoelectron spectroscopic
study of clean and CO adsorbed Ni/TiO2 (110) interfaces, Surf. Sci. 233
(1990) 261.
[359] R.E. Tanner, I. Goldfarb, M.R. Castell, G.A.D. Briggs, The evolution of
Ni nanoislands on the rutile TiO2 (110) surface with coverage, heating
and oxygen treatment, Surf. Sci. 486 (2001) 167.
[360] T. Suzuki, R. Souda, The encapsulation of Pd by the supporting
TiO2 (110) surface induced by strong metalsupport interactions, Surf.
Sci. 448 (2000) 33.
[361] M.D. Negra, N.M. Nicolaisen, Z.S. Li, P.J. Mller, Study of the
interactions between the overlayer and the substrate in the early stages
of palladium growth on TiO2 (110), Surf. Sci. 540 (2003) 117.
[362] Y.M. Sun, D.N. Belton, J.M. White, Characteristics of Pt Thin Films on
TiO2 (110), J. Phys. Chem. 90 (1986) 5178.
[363] A. Linsebigler, C. Rusu, J.T. Yates Jr., Absence of platinum enhancement
of a photoreaction on TiO2 CO photooxidation on Pt/TiO2 (110), J. Am.
Chem. Soc. 118 (1996) 5284.
[364] S. Gan, S. Liang, D.R. Baer, A.W. Grant, Effects of titania surface
structure on the nucleation and growth of Pt nanoclusters on rutile
TiO2 (110), Surf. Sci. 475 (2001) 159.
[365] X.Z. Ji, G.A. Somorjai, Continuous hot electron generation in Pt/TiO2 ,
Pd/TiO2 , and Pt/GaN catalytic nanodiodes from oxidation of carbon
monoxide, J. Phys. Chem. B 109 (2005) 22530.
[366] M. Li, W. Hebenstreit, L. Gross, U. Diebold, M.A. Henderson,
D.R. Jennison, P.A. Schultz, M.P. Sears, Oxygen-induced restructuring
of the TiO2 (110) surface: A comprehensive study, Surf. Sci. 437 (1999)
173.
[367] C. Kittel, Introduction to Solid State Physics, 7th ed., John Wiley & Sons,
1995.
[368] N. Nakajima, H. Kato, T. Okazaki, Y. Sakisaka, Photoemission study of
the modification of the electronic structure of transition-metal overlayers
on TiO2 surfaces I. Fe on TiO2 (110), Surf. Sci. 561 (2004) 79.
[369] L. Zhang, R. Persaud, T.E. Madey, Ultrathin metal films on a metal oxide
surface: Growth of Au on TiO2 (110), Phys. Rev. B 56 (1997) 10549.

[370] A. Berko, J. Szoko, F. Solymosi, High temperature postgrowing of Ptnanocrystallites supported and encapsulated on TiO2 (110) surface, Surf.
Sci. 532535 (2003) 390.
[371] M. Kawasaki, A. Ohtomo, T. Arakane, K. Takahashi, M. Yoshimoto,
H. Koinuma, Atomic control of SrTiO3 surface for perfect epitaxy of
perovskite oxides, Appl. Surf. Sci. 107 (1996) 102.
[372] G. Koster, B.L. Kropman, G.J.H.M. Rijnders, D.H.A. Blank, H. Rogalla,
Quasi-ideal strontium titanate crystal surfaces through formation of
strontium hydroxide, Appl. Phys. Lett. 73 (1998) 2920.
[373] A. Hirata, K. Saiki, A. Koma, A. Ando, Electronic structure of a SrOterminated SrTiO3 (100) surface, Surf. Sci. 319 (1994) 267.
[374] G. Koster, G. Rijnders, D.H.A. Blank, H. Rogalla, Surface morphology
determined by (001) single-crystal SrTiO3 termination, Physica C 339
(2000) 215.
[375] Y. Matsumoto, T. Ohsawa, R. Takahashi, H. Koinuma, Surface
termination effect on the photocatalysis on atomically controlled
SrTiO3 (001) surface, Thin Solid Films 486 (2005) 11.
[376] A. Asthagiri, D.S. Sholl, First principles study of Pt adhesion and growth
on SrO- and TiO2 -terminated SrTiO3 (100), J. Chem. Phys. 116 (2002)
9914.
[377] T. Ochs, S. Kostlmeier, C. Elsasser, Microscopic structure and bonding
at the Pd/SrTiO3 (001) interface. An ab initio local density functional
study, Integr. Ferroelectr. 32 (2001) 959.
[378] T. Ochs, C. Elsasser, Thin Pd films on SrTiO3 (001) substrates: Ab initio
local-density-functional theory, Z. Metallk. 93 (2002) 406.
[379] I.I. Oleinik, E.Y. Tsymbal, D.G. Pettifor, Atomic and electronic structure
of Co/SrTiO3 /Co magnetic tunnel junctions, Phys. Rev. B 65 (2002)
020401(R).
[380] V.E. Henrich, G. Dresselhaus, H.J. Zeiger, Surface defects and the
electronic structure of SrTiO3 surfaces, Phys. Rev. B 17 (1978) 4908.
[381] T. Nishimura, A. Ikeda, H. Namba, T. Morishita, Y. Kido, Structure
change of TiO2 -terminated SrTiO3 (001) surfaces by annealing in O2
atmosphere and ultrahigh vacuum, Surf. Sci. 421 (1999) 273.
[382] J.A. Rodriguez, S. Azad, L.-Q. Wang, Electric and chemical properties
of mixed-metal oxides: Adsorption and reaction of NO on SrTiO3 (100),
J. Chem. Phys. 118 (2003) 6562.
[383] S. Azad, J. Szanyi, C.H.F. Peden, L.-Q. Wang, Adsorption and reaction
of NO on oxidized and reduced SrTiO3 (100) surfaces, J. Vac. Sci.
Technol. A 21 (2003) 1307.
[384] L.-Q. Wang, K.F. Ferris, S. Azad, M.H. Engelhard, Adsorption and
reaction of methanol on stoichiometric and defective SrTiO3 (100)
surfaces, J. Phys. Chem. B 109 (2005) 4507.
[385] L.-Q. Wang, K.F. Ferris, S. Azad, M.H. Engelhard, C.H.F. Peden,
Adsorption and reaction of acetaldehyde on stoichiometric and defective
SrTiO3 (100) surfaces, J. Phys. Chem. B 108 (2004) 1646.
[386] T. Conard, A.C. Rousseau, L.M. Yu, J. Ghijsen, R. Sporken, R. Caudano,
R.L. Johnson, Electron spectroscopy study of the Cu/SrTiO3 (100)
interface, Surf. Sci. 359 (1996) 82.
[387] Q. Fu, T. Wagner, (unpublished results).
[388] Q.D. Jiang, J. Zegenhagen, SrTiO3 (001) surfaces and growth of ultrathin GdBa2 Cu3 O7x films studied by LEED/AES and UHV-STM, Surf.
Sci. 338 (1995) L882.
[389] M. Naito, H. Sato, Reflection high-energy electron diffraction study on
the SrTiO3 surface structure, Physica C 229 (1994) 1.
[390] M.R. Castell, Scanning tunneling microscopy of reconstructions on the
SrTiO3 (001) surface, Surf. Sci. 505 (2002) 1.
[391] F. Silly, M.R. Castell, Bimodal growth of Au on SrTiO3 (001), Phys. Rev.
Lett. 96 (2006) 086104.
[392] F. Silly, M.R. Castell, Growth of Ag icosahedral nanocrystals on a
SrTiO3 (001) support, Appl. Phys. Lett. 87 (2005) 213107.
[393] F. Silly, M.R. Castell, Selecting the shape of supported metal
nanocrystals: Pd huts, hexagons, or pyramids on SrTiO3 (001), Phys.
Rev. Lett. 94 (2005) 046103.
[394] F. Silly, A.C. Powell, M.G. Martin, M.R. Castell, Growth shapes of
supported Pd nanocrystals on SrTiO3 (001), Phys. Rev. B 72 (2005)
165403.
[395] N. Erdman, L.D. Marks, SrTiO3 (001) surface structures under oxidizing
conditions, Surf. Sci. 526 (2003) 107.
493
[396] P.J. Mller, S.A. Komolov, E.F. Lazneva, Selective growth of a MgO
(100)-c(2 2) superstructure on a SrTiO3 (100)-(2 2) substrate, Surf.
Sci. 425 (1999) 15.
[397] F. Silly, M.R. Castell, Self-assembled supported Co nanocrystals: The
adhesion energy of face-centered-cubic Co on SrTiO3 (001)-(2 2),
Appl. Phys. Lett. 87 (2005) 053106.
[398] F. Silly, D.T. Newell, M.R. Castell, SrTiO3 (001) reconstructions: The (2
2) to c(4 4) transition, Surf. Sci. 600 (2006) L219.
[399] M.R. Castell, Nanostructures on the SrTiO3 (001) surface studied by
STM, Surf. Sci. 516 (2002) 33.
[400] F. Silly, M.R. Castell, Fe nanocrystal growth on SrTiO3 (001), Appl.
Phys. Lett. 87 (2005) 063106.
[401] H. Tanaka, H. Matsumoto, T. Kawai, S. Kawai, Surface structure
and electronic property of reduced SrTiO3 (100) surface observed by
scanning tunneling microscopy /spectroscopy, Jpn. J. Appl. Phys. 32
(1993) 1405.
[402] H. Tanaka, T. Matsumoto, T. Kawai, S. Kawai, Interaction of oxygen
vacancies with O2 on a reduced SrTiO3 (100) root5 roo5-R26.6
surface observed by STM, Surf. Sci. 318 (1994) 29.
[403] B. Koslowski, R.N.P. Ziemann, Epitaxial growth of iridium on
strontium-titanate (001) studied by in situ scanning tunneling
microscopy, Surf. Sci. 496 (2002) 153.
[404] Y. Liang, D.A. Bonnell, Structures and chemistry of the annealed
SrTiO3 (001) surface, Surf. Sci. 310 (1994) 128.
[405] R. Moos, K.H. Hardtl, Defect chemistry of donor-doped and undoped
strontium titanate ceramics between 1000 C and 1400 C, J. Am.
Ceram. Soc. 80 (1997) 2549.
[406] S.N. Ruddlesden, P. Popper, New compounds of the K2 NiF4 type, Acta
Crystallogr. 10 (1957) 538.
[407] R. Meyer, R. Waser, J. Helmbold, G. Borchardt, Cationic surface
segregation in donor-doped SrTiO3 under oxidizing conditions, J.
Electroceram. 9 (2002) 87.
[408] R. Meyer, R. Waser, J. Helmbold, G. Borchardt, Observation of vacancy
defect migration in the cation sublattice of complex oxides by 18 O tracer
experiments, Phys. Rev. Lett. 90 (2003) 105901.
[409] K. Gomann, G. Borchardt, M. Schulz, A. Gomann, W. Maus-Friedrichs,
B. Lesage, O. Kaitasov, S. Hoffmann-Eifert, T. Schneller, Sr diffusion
in undoped and La-doped SrTiO3 single crystals under oxidizing
conditions, Phys. Chem. Chem. Phys. 7 (2005) 2053.
[410] K. Gomann, G. Borchardt, A. Gunhold, W. Maus-Friedrichs,
H. Baumann, Ti diffusion in La-doped SrTiO3 single crystals, Phys.
Chem. Chem. Phys. 6 (2005) 3639.
[411] K. Szot, W. Speier, Surfaces of reduced and oxidized SrTiO3 from
atomic force microscopy, Phys. Rev. B 60 (1999) 5909.
[412] K. Szot, W. Speier, U. Breuer, R. Meyer, J. Szade, R. Waser,
Formation of micro-crystals on the (100) surface of SrTiO3 at elevated
temperatures, Surf. Sci. 460 (2000) 112.
[413] A. Gunhold, K. Gomann, L. Beuermann, M. Frerichs, G. Borchardt,
V. Kempter, W. Maus-Friedrichs, Geometric structure and chemical
composition of SrTiO3 surfaces heated under oxidizing and reducing
conditions, Surf. Sci. 507510 (2002) 447.
[414] A. Gunhold, K. Gomann, L. Beuermann, V. Kempter, G. Borchardt,
W. Maus-Friedrichs, Changes in the surface topography and electronic
structure of SrTiO3 (110) single crystals heated under oxidizing and
reducing conditions, Surf. Sci. 566568 (2004) 105.
[415] H. Wei, L. Beuermann, J. Helmbold, G. Borchardt, V. Kempter,
G. Lilienkamp, W. Maus-Friedrichs, Study of SrO segregation on
SrTiO3 (100) surfaces, J. Eur. Ceram. Soc. 21 (2003) 1677.
[416] B. Rahmati, J. Fleig, W. Sigle, E. Bischoff, J. Maier, M. Ruhle, Oxidation
of reduced polycrystalline Nb-doped SrTiO3 : Characterization of surface
islands, Surf. Sci. 595 (2005) 115.
[417] B. Rahmati, J. Fleig, E. Bischoff, W. Sigle, J. Maier, M. Ruhle,
Microstructural studies on the reoxidation behavior of Nb-doped SrTiO3
ceramics, J. Eur. Ceram. Soc. 25 (2005) 2211.
[418] D. Kobayashi, R. Hashimoto, A. Chikamatsu, H. Kumigashira,
M. Oshima, T. Ohnishi, M. Lippmaa, K. Ono, M. Kawasaki,
H. Koinuma, Sr surface segregation and water cleaning for atomically controlled SrTiO3 (001) substrates studied by photoemission spectroscopy, J. Electron. Spectrosc. Relat. Phenom. 144147 (2005) 443.
494
[419] K. Szot, W. Speier, J. Herion, C. Freiburg, Restructuring of the surface

region in SrTiO3 , Appl. Phys. A 64 (1997) 55.
[420] A. Gunhold, L. Beuermann, M. Frerichs, V. Kempter, K. Gomann,
G. Borchardt, W. Maus-Friedrichs, Island formation on 0.1 at.% Ladoped SrTiO3 (1 0 0) at elevated temperatures under reducing conditions,
Surf. Sci. 523 (2003) 80.
[421] F. Silly, M.R. Castell, Formation of single-domain anatase TiO2 (001)-(1
4) islands on SrTiO3 (001) after thermal annealing, Appl. Phys. Lett.
85 (2004) 3223.
[422] S.B. Lee, F. Phillipp, W. Sigle, M. Ruhle, Nanoscale TiO island
formation on the SrTiO3 (001) surface studied by in situ high-resolution
transmission electron microscopy, Ultramicroscopy 105 (2005) 30.
[423] J.E.T. Andersen, P.J. Mller, Room-temperature interaction of ultrathin
film yttrium with SrTiO3 (100), LaAlO3 (100), and MgO(100) surfaces,
Phys. Rev. B 44 (1991) 13645.
[424] D.M. Hill, H.M. Meyer III, J.H. Weaver, Y, Ba, Cu, and Ti interface
reactions with SrTiO3 (100), J. Appl. Phys. 65 (1989) 4943.
[425] E. Tchernychova, C. Scheu, T. Wagner, Q. Fu, M. Ruhle, Electron
microscopy studies of thin Mo films grown by MBE on (100) SrTiO3
substrates, Surf. Sci. 542 (2003) 33.
[426] T. Classen, Diploma Thesis, University of Stuttgart, Stuttgart, 2001.
[427] T. Ono, T. Shinjo, Anisotropic structure and giant magnetoresistance in
Fe/Cr multilayers on SrTiO3 (100) substrates with step terraces, Surf. Sci.
438 (1999) 341.
[428] D. Vlachos, M. Kamaratos, S.D. Foulias, C. Argirusis, G. Borchardt, Ni
ultrathin film development on SrTiO3 (100) surface, Surf. Sci. 550 (2004)
213.
[429] D. Vlachos, M. Kamaratos, S.D. Foulias, C. Argirusis, G. Borchardt,
Adsorption of oxygen on a nickel covered SrTiO3 (100) surface
studied by means of Auger electron spectroscopy and work function
measurements, J. Phys.: Condens. Matter 17 (2005) 635.
[430] M. Kamaratos, D. Vlachos, S.D. Foulias, C. Argirusis, The development
of nickel ultra-thin films and the interaction with oxygen on the
SrTiO3 (100) surface studied by soft X-rays photoelectron spectroscopy,
Surf. Rev. Lett. 11 (2004) 419.
[431] Y. Kido, T. Nishimura, Y. Hoshino, H. Namba, Surface structures of
SrTiO3 and Ni/SrTiO3 (001) studied by medium-energy ion scattering
and SR-photoelectron spectroscopy, Nucl. Instrum. Methods B 161-163
(2000) 371.
[432] A.J. Francis, Y. Cao, P.A. Salvador, Epitaxial growth of Cu(100) and
Pt(100) thin films on perovskite substrates, Thin Solid Films 496 (2006)
317.
[433] T. Wagner, G. Richter, M. Ruhle, Epitaxy of Pd thin films on
(100)SrTiO3 : A three-step growth process, J. Appl. Phys. 89 (2001)
2606.
[434] G. Richter, T. Wagner, Nucleation and growth of Pd clusters on (001)
SrTiO3 : Determination of diffusion and adsorption energies from cluster
densities, J. Appl. Phys. 98 (2005) 094908.
[435] F. Silly, M.R. Castell, Encapsulated Pd nanocrystals supported by
nanoline-structured SrTiO3 (001), J. Phys. Chem. B 109 (2005) 12316.
[436] A.J. Francis, P.A. Salvador, Chirally oriented heteroepitaxial thin films
grown by pulsed laser deposition: Pt(621) on SrTiO3 (621), J. Appl.
Phys. 96 (2004) 2482.
[437] A. Asthagiri, D.S. Sholl, Pt thin films on stepped SrTiO3 surfaces:
SrTiO3 (620) and SrTiO3 (622), J. Mol. Catal. A 216 (2004) 233.
[438] T. Shimizu, N. Gotoh, N. Shinozaki, H. Okushi, The properties of
Schottky junctions on Nb-doped SrTiO3 (001), Appl. Surf. Sci. 117/118
(1997) 400.
[439] M. Copel, P.R. Duncombe, D.A. Neumayer, T.M. Shaw,
R.M. Tromp, Metallization induced band bending of SrTiO3 (100)
and Ba0.7 Sr0.3 TiO3 , Appl. Phys. Lett. 70 (1997) 3227.
[440] X. Chen, T. Garrent, S.W. Liu, Y. Lin, Q.Y. Zhang, C. Dong, C.L. Chen,
Scanning tunneling microscopy studies of growth morphology in highly
epitaxial c-axis oriented Pt thin film on (001) SrTiO3 , Surf. Sci. 542
(2003) L655.
[441] E.E. Mori, M. Kamaratos, Adsorption kinetics of potassium on
SrTiO3 (100), Surf. Rev. Lett. 13 (2007) 681.
[442] P.J. Mller, S.A. Komolov, E.F. Lazneva, A.S. Komolov, Unoccupied
states evolution with oxidation of ultrathin Mg, Zn and Cd layers on
SrTiO3 (100) surfaces, Appl. Surf. Sci. 175176 (2001) 663.
[443] C. Park, D.W. Kim, Interface resistance switching characteristics of
metal/Nb-doped SrTiO3 junctions, J. Korean Phys. Soc. 50 (2007) 1294.
[444] T. Sano, D.M. Saylor, G.S. Rohrer, Surface energy anisotropy of SrTiO3
at 1400 C in air, J. Am. Ceram. Soc. 86 (2003) 1933.
[445] A.E. Romanov, T. Wagner, On the universal misfit parameter at
mismatched interfaces, Scripta Mater. 45 (2001) 325.
[446] R.W.G. Wyckoff (Ed.), Crystal Structures, Krieger, Malabar, 1982.
[447] J. Guo, D.E. Ellis, D.J. Lam, Electronic structure and energetics of
sapphire (0001) and (1102)

surfaces, Phys. Rev. B 45 (1992) 13647.
[448] R. Di Felice, J.E. Northrup, Theory of the clean and hydrogenated
Al2 O3 (0001)-(1 1) surfaces, Phys. Rev. B 60 (1999) R16287.
[449] P.D. Tepesch, A.A. Quong, First-principles calculations of -alumina
(0001) surfaces energies with and without hydrogen, Phys. Status Solidi
b 217 (2000) 377.
[450] J. Toofan, P.R. Watson, The termination of the -Al2 O3 (0001) surface:
A LEED crystallography determination, Surf. Sci. 401 (1998) 162.
[451] J. Ahn, J.W. Rabalais, Composition and structure of the Al2 O3 {0001}-(1
1) surface, Surf. Sci. 388 (1997) 121.
[452] E.A. Soares, M.A. Van Hove, C.F. Walters, K.F. McCarty, Structure
of the -Al2 O3 (0001) surface from low-energy electron diffraction:
Al termination and evidence for anomalously large thermal vibrations,
Phys. Rev. B 65 (2002) 195405.
[453] T. Suzuki, S. Hishita, K. Oyoshi, R. Souda, Structure of -Al2 O3 (0001)
surface and Ti deposited on -Al2 O3 (0001) substrate; CAICISS and
RHEED study, Surf. Sci. 437 (1999) 289.
[454] V. Coustet, J. Jupille, High-resolution electron-energy-loss spectroscopy
of isolated hydroxyl groups on -Al2 O3 (0001), Surf. Sci. 307309
(1994) 1161.
[455] R. Lazzari, J. Jupille, Wetting and interfacial chemistry of metallic films
on the hydroxylated -Al2 O3 (0001) surface, Phys. Rev. B 71 (2005)
045409.
[456] C.E. Nelson, J.W. Elam, M.A. Cameron, M.A. Tolbert, S.M. George,
Desorption of H2 O from a hydroxylated single-crystal -Al2 O3 (0001)
surface, Surf. Sci. 416 (1998) 341.
[457] J.W. Elam, C.E. Nelson, M.A. Cameron, M.A. Tolbert, S.M. George,
Adsorption of H2 O on a single-crystal -Al2 O3 (0001) surface, J. Phys.
Chem. B 102 (1998) 7008.
[458] P.J. Eng, T.P. Trainor, G.E. Brown Jr., G.A. Waychunas, M. Newville,
S.R. Sutton, M.L. Rivers, Structure of the hydrated -Al2 O3 (0001)
surface, Science 288 (2000) 1029.
[459] S.A. Chambers, T. Droubay, D.R. Jennison, T.R. Mattsson, Laminar
growth of ultrathin metal films on metal oxides: Co on hydroxylated Al2 O3 (0001), Science 297 (2002) 827.
[460] Q. Fu, T. Wagner, M. Ruhle, Hydroxylated -Al2 O3 (0001) surfaces and
metal/-Al2 O3 (0001) interfaces, Surf. Sci. 600 (2006) 4870.
[461] J.A. Kelber, C. Niu, K. Shepherd, D.R. Jennison, A. Bogicevic, Copper
wetting of -Al2 O3 (0001): Theory and experiment, Surf. Sci. 446
(2000) 76.
[462] C. Niu, K. Shepherd, D. Martini, J. Tong, J.A. Kelber, D.R. Jennison,
A. Bogicevic, Cu interactions with -Al2 O3 (0001): Effects of surface
hydroxyl groups versus dehydroxylation by Ar-ion sputtering, Surf. Sci.
465 (2000) 163.
[463] P. Liu, T. Kendelewicz, G.E. Brown Jr., E.J. Nelson, S.A. Chambers,
Reaction of water vapor with -Al2 O3 (0001) and -Fe2 O3 (0001)
surfaces: Synchrotron X-ray photoemission studies and thermodynamic
calculations, Surf. Sci. 417 (1998) 53.
[464] T. Akatsu, C. Scheu, T. Wagner, T. Gemming, N. Hosoda, T. Suga,
M. Ruhle, Morphology and microstructure of the Ar-ion sputtered (0001)
-Al2 O3 surface, Appl. Surf. Sci. 165 (2000) 159.
[465] M. Gautier, J.P. Duraud, L. Pham Van, M.J. Guittet, Modifications
of -Al2 O3 (0001) surfaces induced by thermal treatments or ion
bombardment, Surf. Sci. 250 (1991) 71.
[466] M. Vermeersch, F. Malengreau, R. Sporken, R. Caudano, The
aluminum/sapphire interface formation at high temperature: An AES and
LEED study, Surf. Sci. 323 (1995) 175.

[467] M. Vermeersch, R. Sporken, P. Lambin, R. Caudano, The Al/Al2 O3
interface formation as studied by electron spectroscopies, Surf. Sci. 235
(1990) 5.
[468] T.J. Godin, J.P. LaFemina, Atomic and electronic structure of the
corundum (-alumina) (0001) surface, Phys. Rev. B 49 (1994) 7691.
[469] I. Manassidis, A. De Vita, M.J. Gillan, Structure of the (0001) surface of
-Al2 O3 from first principles calculations, Surf. Sci. 285 (1993) L517.
[470] V.E. Puchin, J.D. Gale, A.L. Shluger, E.A. Kotomin, J. Gunster,
M. Brause, V. Kempter, Atomic and electronic structure of the corundum
(0001) surface: Comparison with surface spectroscopies, Surf. Sci. 370
(1997) 190.
[471] K.C. Hass, W.D. Schneider, A. Curioni, W. Andreoni, The chemistry
of water on alumina surfaces: Reaction dynamics from first principles,
Science 282 (1998) 265.
[472] K.C. Hass, W.F. Schneider, A. Curioni, W. Andreoni, First-principles
molecular dynamics simulations of H2 O on -Al2 O3 (0001), J. Phys.
Chem. B 104 (2000) 5527.
[473] D.J. Siegel, L.G. Hector Jr., J.B. Adams, Adhesion, atomic structure,
and bonding at the Al(111)/-Al2 O3 (0001) interface: A first principles
study, Phys. Rev. B 65 (2002) 085415.
[474] P. Guenard, G. Renaud, A. Barbier, M. Gautier-Soyer, Determination of
the -Al2 O3 (0001) surface relaxation and termination by measurements
of crystal truncation rods, Surf. Rev. Lett. 5 (1998) 321.
[475] C. Scheu, Manipulating bonding at a Cu/(0001)Al2 O3 interface by
different substrate cleaning processes, Interface Sci. 12 (2004) 127.
[476] M. Gautier, J.P. Duraud, L. Pham Van, Influence of the Al2 O3 (0001)
surface reconstruction on the Cu/Al2 O3 interface, Surf. Sci. 249 (1991)
L327.
[477] S. Gota, M. Gautier-Soyer, L. Douillard, J.P. Duraud, P. Le F`evre, The
initial stages of the growth of copper on a (1 1) and a ( 31 31)R

9 -Al2 O3 (0001) surface, Surf. Sci. 352354 (1996) 1016.
[478] G. Dehm, C. Scheu, G. Mobus, R. Brydson, M. Ruhle, Synthesis
of analytical and high-resolution transmission electron microscopy to
determine the interface structure of Cu/Al2 O3 , Ultramicroscopy 67
(1997) 207.
[479] Q. Guo, P.J. Mller, On the thermal stability of copper deposits on a
(0001) sapphire surface, Surf. Sci. 244 (1991) 228.
[480] S. Varma, G.S. Chottiner, M. Arbab, Surface studies of (0001) Al2 O3
and the growth of thin films of Cu on Al2 O3 , J. Vac. Sci. Technol. A 10
(1992) 2857.
[481] Z. Lodziana, J.K. Nrskov, Adsorption of Cu and Pd on -Al2 O3 (0001)
surfaces with different stoichiometries, J. Chem. Phys. 115 (2001)
11261.
[482] W. Zhang, J.R. Smith, Nonstoichiometric interfaces and Al2 O3 adhesion
with Al and Ag, Phys. Rev. Lett. 85 (2000) 3225.
[483] W. Zhang, J.R. Smith, Stoichiometry and adhesion of Nb/Al2 O3 , Phys.
Rev. B 61 (2000) 16883.
[484] X.G. Wang, J.R. Smith, Hydrogen and carbon effects on Al2 O3 surface
phases and metal deposition, Phys. Rev. B 70 (2004) 081401(R).
[485] M.A. Nygren, D.H. Gay, C.R.A. Catlow, Hydroxylation of the surface
of the corundum basal plane, Surf. Sci. 380 (1997) 113.
[486] J.M. Wittbrodt, W.L. Hase, H.B. Schlegel, Ab initio study of the
interaction of water with cluster models of the aluminum terminated
(0001) -aluminum oxide surface, J. Phys. Chem. B 102 (1998) 6539.
[487] J.G. Chen, J.E. Crowell, J.T.J. Yates, Assignment of a surface vibrational
mode by chemical means: Modification of the lattice modes of Al2 O3 by
a surface reaction with H2 O, J. Chem. Phys. 84 (1986) 5906.
[488] R. Lazzari, J. Jupille, Chemical reaction via hydroxyl groups at the
titanium/-Al2 O3 (0001) interface, Surf. Sci. 507510 (2002) 683.
[489] J. Libuda, M. Frank, A. Sandell, S. Andersson, P.A. Bruhwiler,
M. Baumer, N. Martensson, H.-J. Freund, Interaction of rhodium with
hydroxylated alumina model substrates, Surf. Sci. 384 (1997) 106.
[490] Z. Lodziana, J.K. Nrskov, P. Stoltze, The stability of the hydroxylated
(0001) surface of -Al2 O3 , J. Chem. Phys. 118 (2003) 11179.
[491] J.F. Sanz, N.C. Hernandez, Mechanism of Cu deposition on the Al2 O3 (0001) surface, Phys. Rev. Lett. 94 (2005) 016104.
[492] X.G. Wang, J.R. Smith, M. Scheffler, Effect of hydrogen on Al2 O3 /Cu
interfacial structure and adhesion, Phys. Rev. B 66 (2002) 073411.
495
[493] X.G. Wang, J.R. Smith, Adhesion of copper and alumina from first
principles, J. Am. Ceram. Soc. 86 (2003) 696.
[494] D.R. Jennison, T.R. Mattsson, Atomic understanding of strong
nanometer-thin metal/alumina interfaces, Surf. Sci. 544 (2003) L689.
[495] K.H. Johnson, S.V. Pepper, Molecular-orbital model for metalsapphire
interfacial strength, J. Appl. Phys. 53 (1982) 6634.
[496] K. Nath, A.B. Anderson, Oxidative bonding of (0001) -Al2 O3 to closepacked surfaces of the first transition-metal series, Sc through Cu, Phys.
Rev. B 39 (1989) 1013.
[497] I.G. Bartirev, A. Alavi, M.W. Finnis, T. Deutsch, First-principle
calculations of the ideal cleavage energy of bulk niobium (111)/alumina (0001) interfaces, Phys. Rev. Lett. 82 (1999) 1510.
[498] J.R.B. Gomes, F. Illas, N.C. Hernandez, A. Marquez, J.F. Sanz,
Interaction of Pd with -Al2 O3 (0001): A case study of modeling the
metaloxide interface on complex substrates, Phys. Rev. B 65 (2002)
125414.
[499] J.R.B. Gomes, Z. Lodziana, F. Illas, Adsorption of small palladium
clusters on the relaxed -Al2 O3 (0001) Surface, J. Phys. Chem. B 107
(2003) 6411.
[500] N.C. Hernandez, J. Graciani, A. Marquez, J.F. Sanz, Cu, Ag and Au
atoms deposited on the -Al2 O3 (0001) surface: A comparative density
functional study, Surf. Sci. 575 (2005) 189.
[501] Y.F. Zhukovskii, E.A. Kotomin, B. Herschend, K. Hermansson,
P.W.M. Jacobs, The adhesion properties of the Ag/-Al2 O3 (0001)
interface: An ab initio study, Surf. Sci. 513 (2002) 343.
[502] F.S. Ohuchi, Surface science studies of Nb-(0001) Al2 O3 interfacial
reactions, J. Mater. Sci. Lett. 8 (1989) 1427.
[503] J. Biener, M. Baumer, R.J. Madix, P. Liu, E. Nelson, T. Kendelewisz,
G. Brown Jr., Growth and electronic structure of vanadium on Al2 O3 (0001), Surf. Sci. 449 (2000) 50.
[504] Y.S. Chaug, N.J. Chou, Y.H. Kim, Interaction of Ti with fused silica and
sapphire during metallization, J. Vac. Sci. Technol. A 5 (1987) 1288.
[505] S. Ogawa, S. Ichikawa, Observation of induced dipoles between small
palladium clusters and -(0001) Al2 O3 , Phys. Rev. B 51 (1995) 17231.
[506] B. Ealet, E. Gillet, Palladium alumina interface: Influence of the oxide
stoichiometry studied by EELS and XPS, Surf. Sci. 281 (1993) 91.
[507] E. Gillet, M.H.E. Yakhloufi, J.P. Disalvo, F. Ben Abdelouahab, In situ
characterization of ultra-thin palladium deposits on - and -alumina,
Surf. Sci. 419 (1999) 216.
[508] B. Ealet, E. Gillet, Metalalumina interface: Influence of the metal
electronegativity and of the substrates stoichiometry, Surf. Sci. 367
(1996) 221.
[509] G. Dehm, B.J. Inkson, T. Wagner, Growth and microstructural stability
of epitaxial Al films on (0001) -Al2 O3 substrates, Acta Mater. 50
(2002) 5021.
[510] G. Dehm, C. Scheu, M. Ruhle, R. Raj, Growth and structure of internal
Cu/Al2 O3 and Cu/Ti/Al2 O3 interfaces, Acta Mater. 46 (1998) 759.
[511] S. Tsukimoto, F. Phillipp, T. Wagner, Texture of MBE grown Cr films on
-Al2 O3 (0001): The occurrence of NishiyamaWassermann (NW) and
KurdjumovSachs (KS) related orientation relationships, J. Eur. Ceram.
Soc. 23 (2003) 2947.
[512] J. Libuda, F. Winkelmann, M. Baumer, H.-J. Freund, T. Bertrams,
Structure and defects of an ordered alumina film on NiAl(110), Surf.
Sci. 318 (1994) 61.
[513] G. Kresse, M. Schmid, E. Napetschnig, M. Shishkin, L. Kohler, P. Varga,
Structure of the ultrathin aluminum oxide film on NiAl(110), Science
308 (2005) 1440.
[514] A. Rosenhahn, J. Schneider, J. Kandler, C. Becker, K. Wandelt,
Interaction of oxygen with Ni3 Al(111) at 300 K and 1000 K, Surf. Sci.
433435 (1999) 705.
[515] A. Lehnert, A. Krupski, S. Degen, K. Franke, R. Decker, S. Rusponi,
M. Kralj, C. Becker, H. Brune, K. Wandelt, Nucleation of ordered Fe
islands on Al2 O3 /Ni3 Al(111), Surf. Sci. 600 (2006) 1804.
[516] R. Franchy, J. Masuch, P. Gassmann, The oxidation of the NiAl(111)
surface, Appl. Surf. Sci. 93 (1996) 317.
[517] N. Tsud, K. Veltruska, V. Matolin, Pd/Al2 O3 interaction: The influence
of ionicity character of different alumina surfaces, Surf. Sci. 507510
(2002) 808.
496
[518] J.G. Chen, J.E. Crowell, J.T.J. Yates, The metalmetal oxide interface:
A study of thermally-activated diffusion at the Ni/Al2 O3 interface using
electron spectroscopies, Surf. Sci. 185 (1987) 373.
[519] R.M. Jaeger, H. Kuhlenbeck, H.-J. Freund, M. Wuttig, W. Hoffmann,
R. Franchy, H. Ibach, Formation of well-ordered aluminum oxide
overlayer by oxidation of NiAl(110), Surf. Sci. 259 (1991) 235.
[520] M. Eriksson, J. Sainio, J. Lahtinen, Chromium deposition on ordered
alumina films: An x-ray photoelectron spectroscopy study of the
interaction with oxygen, J. Chem. Phys. 116 (2002) 3870.
[521] M. Heemeier, M. Frank, J. Libuda, K. Wolter, H. Kuhlenbeck,
M. Baumer, H.-J. Freund, The influence of OH groups on the growth
of rhodium on alumina: A model study, Catal. Lett. 68 (2000) 19.
[522] M. Heemeier, S. Stempel, S.K. Shaikhutdinov, J. Libuda, M. Baumer,
R.J. Oldman, S.D. Jackson, H.-J. Freund, On the thermal stability of
metal particles supported on a thin alumina film, Surf. Sci. 523 (2003)
103.
[523] M. Kulawik, N. Nilius, H.-J. Freund, Influence of the metal substrate
on the adsorption properties of thin oxide layers: Au atoms on a thin
alumina film on NiAl(110), Phys. Rev. Lett. 96 (2006) 036103.
[524] J.A. Rodriguez, M. Kuhn, J. Hrbek, Interaction of silver, cesium, and
zinc with alumina surfaces: Thermal desorption and photoemission
studies, J. Phys. Chem. 100 (1996) 18240.
[525] J.G. Chen, M.L. Colaianni, W.H. Weinberg, J.T.J. Yates, The
Cu/Al2 O3 /Al(111) interface: Initial film growth and thermally induced
diffusion of copper into the bulk, Surf. Sci. 279 (1992) 223.
[526] C. Duriez, C. Chapon, C.R. Henry, J.M. Rickard, Structural
characterization of MgO(100) surfaces, Surf. Sci. 230 (1990) 123.
[527] D. Abriou, F. Creuzet, J. Jupille, Characterization of cleaved MgO(100)
surfaces, Appl. Surf. Sci. 352354 (1996) 499.
[528] O. Robach, G. Renaud, A. Barbier, Very-high-quality MgO(001)
surfaces: Roughness, rumpling and relaxation, Surf. Sci. 401 (1998) 227.
[529] G. Benedek, G. Brusdeylins, V. Senz, J.G. Skofronick, J.P. Toennies,
F. Traeger, R. Vollmer, Helium atom scattering study of the surface
structure and dynamics of in situ cleaved MgO(001) single crystals,
Phys. Rev. B 64 (2001) 125421.
[530] T.V. Ashworth, C.L. Pang, P.L. Wincott, D.J. Vaughan, G. Thornton,
Imaging in situ cleaved MgO(100) with non-contact atomic force
microscopy, Appl. Surf. Sci. 210 (2003) 2.
[531] F. Didier, J. Jupille, Layer-by-layer growth mode of silver on magnesium
oxide (100), Surf. Sci. 307309 (1994) 587.
[532] K. Hjrup-Hansen, S. Ferrero, C.R. Henry, Nucleation and growth
kinetics of gold nanoparticles on MgO(100) studied by UHV-AFM,
Appl. Surf. Sci. 226 (2004) 167.
[533] O. Robach, G. Renaud, A. Barbier, Structure and morphology of the
Ag/MgO(001) interface during in situ growth at room temperature, Phys.
Rev. B 60 (1999) 5858.
[534] P. Guenard, G. Renaud, B. Villette, Structure, translational state and
morphology of the Ag/MgO(001) interface during its formation, Physica
B 221 (1996) 205.
[535] G. Renaud, A. Barbier, O. Robach, Growth, structure, and morphology
of the Pd/MgO(001) interface: Epitaxial site and interfacial distance,
Phys. Rev. B 60 (1999) 5872.
[536] A. Barbier, G. Renaud, O. Robach, Growth, annealing and oxidation of
the Ni/MgO(001) interface studied by grazing incidence x-ray scattering,
J. Appl. Phys. 84 (1998) 4259.
[537] A. Trampert, F. Ernst, C.P. Flynn, H.F. Fischmeister, M. Ruhle, High
resolution transmission electron microscopy studies of the Ag/MgO
interface, Acta Metall. Mater. 40 (1992) S227.
[538] A.M. Flank, R. Delaunay, P. Lagarde, M. Pompa, J. Jupille, Epitaxial
silver layer at the MgO(100) surface, Phys. Rev. B 53 (1996) R1737.
[539] U. Schonberger, O.K. Andersen, M. Methfessel, Bonding at
metalceramic interface; Ab Initio density-functional calculations
for Ti and Ag on MgO, Acta Metall. Mater. 40 (1992) S1.
[540] I. Yudanov, G. Pacchioni, K. Neyman, N. Rosch, Systematic Density
Functional Study of the Adsorption of Transition Metal Atoms on the
MgO(001) Surface, J. Phys. Chem. B 101 (1997) 2786.
[541] A.V. Matveev, K. Neyman, G. Pacchioni, N. Rosch, Density functional
study of M4 clusters (M = Cu, Ag, Ni, Pd) deposited on the regular MgO
(001) surface, Chem. Phys. Lett. 299 (1999) 603.
[542] G. Pacchioni, N. Rosch, Supported nickel and copper clusters on

MgO(100): A first-principles calculation on the metal/oxide interface,
J. Chem. Phys. 104 (1996) 7329.
[543] G. Pacchioni, L. Giordano, M. Baistrocchi, Charging of metal atoms on
ultrathin MgO/Mo(100) films, Phys. Rev. Lett. 94 (2005) 226204.
[544] V. Musolino, A. Selloni, R. Car, First principles study of adsorbed
Cun (n = 14) microclusters on MgO(100): Structural and electronic
properties, J. Chem. Phys. 108 (1998) 5044.
[545] C. Li, R. Wu, A.J. Freeman, C.L. Fu, Energetics, bonding mechanism,
and electronic structure of metalceramic interfaces: Ag/MgO(001),
Phys. Rev. B 48 (1993) 8317.
[546] T. Hong, J.R. Smith, D.J. Srolovitz, Theory of metalceramic adhesion,
Acta Metall. Mater. 43 (1995) 2721.
[547] C. Noguera, F. Finocchi, J. Goniakowski, First principles studies of
complex oxide surfaces and interfaces, J. Phys.: Condens. Matter 16
(2004) S2509.
[548] A.V. Matveev, K.M. Neyman, I.V. Yudanov, N. Rosch, Adsorption of
transition metal atoms on oxygen vacancies and regular sites of the
MgO(001) surface, Surf. Sci. 426 (1999) 123.
[549] Y.F. Zhukovskii, E.A. Kotomin, G. Borstel, Adsorption of single Ag and
Cu atoms on regular and defective MgO(001) substrates: An ab initio
study, Vacuum 74 (2004) 235.
[550] Z.X. Yang, R.Q. Wu, Q.M. Zhang, D.W. Goodman, Adsorption of Au on
an O-deficient MgO(001) surface, Phys. Rev. B 65 (2002) 155407.
[551] L. Giordano, J. Goniakowski, G. Pacchioni, Characteristics of Pd
adsorption on the MgO(100) surface: Role of oxygen vacancies, Phys.
Rev. B 64 (2001) 075417.
[552] M.C. Wu, J.S. Corneille, C.A. Estrada, J.W. He, D.W. Goodman,
Synthesis and characterization of ultra-thin MgO films on Mo(100),
Chem. Phys. Lett. 182 (1991) 472.
[553] M.C. Wu, J.S. Corneille, J.W. He, C.A. Estrada, D.W. Goodman,
Preparation, characterization, and chemical properties of ultrathin MgO
films on Mo(100), J. Vac. Sci. Technol. A 10 (1992) 1467.
[554] M.C. Gallagher, M.S. Fyfield, J.P. Cowin, S.A. Joyce, Imaging insulating
oxides: Scanning tunneling microscopy of ultrathin MgO films on
Mo(001), Surf. Sci. 339 (1995) L909.
[555] S. Altieri, L.H. Tjeng, G.A. Sawatzky, Electronic structure and chemical
reactivity of oxidemetal interfaces: MgO(100)/Ag(100), Phys. Rev. B
61 (2000) 16948.
[556] P. Stracke, S. Krischok, V. Kempter, Ag-adsorption on MgO:
Investigations with MIES and UPS, Surf. Sci. 473 (2001) 86.
[557] A. Sanchez, S. Abbet, U. Heiz, W.-D. Schneider, H. Haekkinen,
R.N. Barnett, U. Landman, When gold is not noble: Nanoscale gold
catalysts, J. Phys. Chem. A 103 (1999) 9573.
[558] M. Sterrer, M. Yulikov, E. Fishbach, M. Heyde, H.-P. Rust, G. Pacchioni,
T. Risse, H.-J. Freund, Interaction of gold clusters with color centers on
MgO(001) films, Angew. Chem. Int. Ed. 45 (2006) 2630.
[559] H. Hakkinen, S. Abbet, A. Sanchez, U. Heiz, U. Landman, Structural,
electronic, and impurity-doping effects in nanoscale chemistry:
Supported gold nanoclusters, Angew. Chem. Int. Ed. 42 (2003) 1297.
[560] D.E. Starr, S.F. Diaz, J.E. Musgrove, J.T. Ranneay, D.J. Bald, L. Nelen,
H. Ihm, C.T. Campbell, Heat of adsorption of Cu and Pb on hydroxylcovered MgO(100), Surf. Sci. 515 (2002) 13.
[561] R. Lamber, N. Jaeger, G. Schulz-Ekloff, Metalsupport interaction in
the Pd/SiO2 system: Influence of the support pretreatment, J. Catal. 123
(1990) 285.
[562] G.-M. Rignanese, J.-C. Charlier, X. Gonze, First-principles moleculardynamics investigation of the hydration mechanisms of the (0001) quartz surface, Phys. Chem. Chem. Phys. 6 (2004) 1920.
[563] J.J. Yang, E.G. Wang, Water adsorption on hydroxylated -quartz (0001)
surfaces: From monomer to flat bilayer, Phys. Rev. B 73 (2006) 035405.
[564] L. Giordano, M. Baistrocchi, G. Pacchioni, Bonding of Pd, Ag, Au atoms
on MgO(100) surfaces and MgO/Mo(100) ultrathin films: A comparative
DFT study, Phys. Rev. B 72 (2005) 115403.
[565] F. Bart, M. Gautier, F. Jollet, J.P. Duraud, Electronic structure of the
(0001) and (1010) quartz surfaces and of their defects as observed by
reflection electron energy loss spectroscopy (REELS), Surf. Sci. 306
(1994) 342.

[566] A.P.J. Jansen, R.A. van Santen, HartreeFockSlater calculations on
cation-induced changes in the adsorption of CO on Ir4 clusters, J. Phys.
Chem. 94 (1990) 6764.
[567] F. Bart, M. Gautier, A LEED study of the (0001) alpha-quartz surface
reconstruction, Surf. Sci. 311 (1994) L671.
[568] S.P. Harte, A.P. Woodhead, S. Vinton, T. Evans, S.A. Haycock,
C.A. Muryn, P.L. Wincott, V.R. Dhanak, C.E. Marsden, G. Thornton,
The initial stages of Cr and Ti growth on SiO2 (0001), Surf. Sci. 424
(1999) 179.
[569] E.P. OReilly, J. Robertson, Theory of defects in vitreous silicon dioxide,
Phys. Rev. B 27 (1983) 3780.
[570] N. Lopez, F. Illas, G. Pacchioni, Adsorption of Cu, Pd, and Cs atoms
on regular and defect sites of the SiO2 surface, J. Am. Chem. Soc. 121
(1999) 813.
[571] J.B. Zhou, H.C. Lu, T. Gustafsson, E. Garfunkel, Anomalously weak
adsorption of Cu on SiO2 and MgO surfaces, Surf. Sci. 293 (1993) L887.
[572] J.M. Antonietti, M. Michalski, U. Heiz, H. Jones, K.H. Lim, N. Rosch,
A.D. Vitto, G. Pacchioni, Optical absorption spectrum of gold atoms
deposited on SiO2 from cavity ringdown spectroscopy, Phys. Rev. Lett.
94 (2005) 213402.
[573] H.A. Al-Abadleh, V.H. Grassian, Oxide surfaces as environmental
interfaces, Surf. Sci. Rep. 52 (2003) 63.
[574] X. Xu, D.W. Goodman, New approach to the preparation of ultrathin
silicon dioxide films at low temperatures, Appl. Phys. Lett. 61 (1992)
774.
[575] J.W. He, X. Xu, J.S. Corneille, D.W. Goodman, X-ray photoelectron
spectroscopic characterization of ultrathin silicon oxide films on a
Mo(100) surface, Surf. Sci. 279 (1992) 119.
[576] Y.D. Kim, T. Wei, D.W. Goodman, Identification of defect sites on SiO2
thin films grown on Mo(112), Langmuir 19 (2003) 354.
[577] Asuha, S.S. Im, M. Tanaka, S. Imai, M. Takahashi, H. Kobayashi,
Formation of 1030 nm SiO2 /Si structure with a uniform thickness at
120 C by nitric acid oxidation method, Surf. Sci. 600 (2006) 2523.
[578] T. Schroeder, A. Hammoudeh, M. Pykavy, N. Magg, M. Adelt, M.
Baumer, H.-J. Freund, Single crystalline silicon dioxide films on
Mo(112), Solid State Electron. 45 (2001) 1471.
[579] T. Schroeder, J.B. Giorgi, M. Baumer, H.-J. Freund, Morphological and
electronic properties of ultrathin crystalline silica epilayers on a Mo(112)
substrate, Phys. Rev. B 66 (2002) 165422.
[580] J. Weissenrieder, S. Kaya, J.-L. Lu, H.-J. Gao, S. Shaikhutdinov,
H.-J. Freund, M. Sierka, T.K. Todorova, J. Sauer, Atomic structure of
a thin silica film on a Mo(112) substrate: A two-dimensional network of
SiO4 tetrahedra, Phys. Rev. Lett. 95 (2005) 076103.
[581] T.K. Todorova, M. Sierka, J. Sauer, S. Kaya, J. Weissenrieder,
J.-L. Lu, H.-J. Gao, S. Shaikhutdinov, H.-J. Freund, Atomic structure
of a thin silica film on a Mo(112) substrate: A combined experimental
and theoretical study, Phys. Rev. B 73 (2006) 165414.
[582] M.S. Chen, A.K. Santra, D.W. Goodman, Structure of thin SiO2 films
grown on Mo (112), Phys. Rev. B 69 (2004) 155404.
[583] M.S. Chen, D.W. Goodman, The structure of monolayer SiO2 on
Mo(112): A 2-D [SiOSi] network or isolated [SiO4 ] units? Surf. Sci.
600 (2006) L255.
[584] S. Wendt, E. Ozensoy, T. Wei, M. Frerichs, Y. Cai, M.S. Chen,
D.W. Goodman, Electronic and vibrational properties of ultrathin SiO2
films grown on Mo(112), Phys. Rev. B 72 (2005) 115409.
[585] X. Xu, J. Szanyi, Q. Xu, D.W. Goodman, Structural and catalytic
properties of model silica supported palladium catalysts: A comparison
to single crystal surfaces, Catal. Today 21 (1994) 57.
[586] K. Luo, T. Wei, C.-W. Yi, S. Axnanda, D.W. Goodman, Preparation
and characterization of silica supported AuPd model catalysts, J. Phys.
Chem. B 109 (2005) 23517.
[587] M. Kundu, Y. Murata, Growth of single-crystal SiO2 film on Ni(111)
surface, Appl. Phys. Lett. 80 (2002) 1921.
[588] Z. Zhang, Z.Q. Jiang, Y.X. Yao, D.L. Tan, Q. Fu, X.H. Bao, Preparation
and characterization of atomically flat and ordered silica films on a
Pd(100) surface, Thin Solid Films (in press).
[589] B.K. Min, W.T. Wallace, A.K. Santra, D.W. Goodman, Role of defects
in the nucleation and growth of Au nanoclusters on SiO2 thin films, J.
Phys. Chem. B 108 (2004) 16339.
497
[590] M.S. Chen, D.W. Goodman, An investigation of the TiOx SiO2 /

Mo(112) interface, Surf. Sci. 574 (2005) 259.
[591] B.K. Min, W.T. Wallace, D.W. Goodman, Synthesis of a sinter-resistant,
mixed-oxide support for Au nanoclusters, J. Phys. Chem. B 108 (2004)
14609.
[592] T. Asakawa, K. Tanaka, I. Toyoshima, Interaction of Ni with SiOx or
SiO2 formed on Si(111) and CO adsorption inhibition in Ni/SiOx /nSi(111) studied by XPS and AES, Langmuir 4 (1988) 521.
[593] L.C.A. van den Oetelaar, A. Partridge, S.L.G. Toussaint, C.F.J. Flipse,
H.H. Brongersma, A surface science study of model catalysts. 2.
metalsupport interactions in Cu/SiO2 model catalysts, J. Phys. Chem.
B 102 (1998) 9541.
[594] H. Dallaporta, M. Liehr, J.E. Lewis, Silicon dioxide defects induced by
metal impurities, Phys. Rev. B 41 (1990) 5075.
[595] M.J. Frederick, R. Goswami, G. Ramanath, Sequence of Mg segregation,
grain growth, and interfacial MgO formation in CuMg alloy films on
SiO2 during vacuum annealing, J. Appl. Phys. 93 (2003) 5966.
[596] M. Liehr, F.K. LeGoues, G.W. Rubloff, P.S. Ho, Chemical reactions at
Pt/oxide/Si and Ti/oxide/Si interfaces, J. Vac. Sci. Technol. A 3 (1985)
983.
[597] B.R. Powell, S.F. Whittington, Encapsulation: A new mechanism of
catalyst deactivation, J. Catal. 81 (1983) 382.
[598] L.C.A. van den Oetelaar, R.J.A. van den Oetelaar, A. Partridge,
C.F.J. Flipse, H.H. Brongersma, Reaction of nanometer-sized Cu
particles with a SiO2 substrate, Appl. Phys. Lett. 74 (1999) 2954.
[599] J.T. Mayer, R.F. Lin, E. Garfunkel, Surface and bulk diffusion of
adsorbed nickel on ultrathin thermally grown silicon dioxide, Surf. Sci.
265 (1992) 102.
[600] J.B. Zhou, T. Gustafsson, R.F. Lin, E. Garfunkel, Medium energy ion
scattering study of Ni on ultrathin films of SiO2 on Si (111), Surf. Sci.
284 (1993) 67.
[601] B. Schleich, D. Schmeisser, W. Goepel, Structure and reactivity of the
system Si/SiO2 /Pd: A combined XPS, UPS, and HREELS study, Surf.
Sci. 191 (1987) 367.
[602] R. Anton, U. Neukirch, M. Harsdorff, Auger-electron-spectroscopy
analysis of a plasmon loss in palladium silicide formed from Pd deposits
on silicon, Phys. Rev. B 36 (1987) 7422.
[603] M. Vogt, K. Drescher, Barrier behavior of plasma deposited silicon
oxide and nitride against Cu diffusion, Appl. Surf. Sci. 91 (1995)
303.
[604] J. Zhu, G.A. Somorjai, Formation of platinum silicide on a platinum
nanoparticle array model catalyst deposited on silica during chemical
reaction, Nano Lett. 1 (2001) 8.
[605] S. Labich, A. Kohl, E. Taglauer, H. Knozinger, Silicide formation by
high-temperature reaction of Rh with model SiO2 films, J. Chem. Phys.
109 (1998) 2052.
[606] B.K. Min, A.K. Santra, D.W. Goodman, Thermal stability of Pd
supported on single crystalline SiO2 thin films, J. Vac. Sci. Technol. B
21 (2003) 2319.
[607] F. Sadi, D. Duprez, F. Gerard, S. Rossignol, A. Miloudi, Morphological
and structural changes in reducing and steam atmospheres of SiO2 supported Rh catalysts, Catal. Lett. 44 (1997) 221.
[608] D. Wang, S. Penner, D.S. Su, G. Rupprechter, K. Hayek, R. Schlogl,
SiO2 -supported Pt particles studied by electron microscopy, Mater.
Chem. Phys. 81 (2003) 341.
[609] R. Lamber, N. Jaeger, G. Schulz-Ekloff, On the metalsupport
interaction in the NiSiO2 system, Surf. Sci. 227 (1990) 268.
[610] W. Juszczyk, D. Lomot, J. Pielaszek, Z. Karpinski, Transformation of
Pd/SiO2 catalysts during high temperature reduction, Catal. Lett. 78
(2002) 95.
[611] W. Juszczyk, Z. Karpinski, D. Lomot, J. Pielaszek, Transformation of
Pd/SiO2 into palladium silicide during reduction at 450 and 500 C, J.
Catal. 220 (2003) 299.
498
Glossary
AEM: analytic electron microscope
AES: Auger electron spectroscopy
AFM: atomic force microscopy
ALD: atomic layer deposition
ALE: atomic layer epitaxy
BE: binding energy
CB: conduction band
CBM: conduction band minimum
CNL: charge neutrality level
CSD: chemical solution deposition
CTEM: conventional transmission electron microscopy
CVD: chemical vapor deposition
DFT: density functional theory
DOS: density of states
EELS: electron energy-loss spectroscopy
E F : Fermi energy
ELNES: electron energy-loss near-edge structure
EPR: electron paramagnetic resonance
EXAFS: extended X-ray absorption fine structure
FTIR: Fourier transform infrared spectroscopy
FWHM: full width at half maximum
GIXS: grazing incidence X-ray scattering
HAS: helium atom scattering
HREELS: high resolution electron energy-loss spectroscopy
HRTEM: high resolution transmission electron microscopy
HTR: high temperature reduction
IR: infrared spectroscopy
IRAS: infrared reflection-adsorption spectroscopy
ISS: ion scattering spectroscopy

LDA: local density approximation
LEED: low energy electron diffraction
LEIS: low energy ion scattering
MBE: molecular beam epitaxy
MD: molecular dynamics
MEIS: medium energy ion scattering
MIGS: metal-induced gap states
ML: monolayer
MOS: metaloxidesemiconductor
MOVPE: metal-organic vapor phase epitaxy
MS: metalsemiconductor
PES: photoemission electron spectroscopy
OR: orientation
PVD: physical vapor deposition
RHEED: reflection high energy electron diffraction
SBH: Schottky barrier height
SEXAFS: surface extended X-ray absorption fine structure
SFM: scanning force microscopy
SMSI: strong metalsupport interaction
STM: scanning tunneling microscopy
STS: scanning tunneling spectroscopy
TEM: transmission electron microscopy
TED: transmission electron diffraction
UHV: ultrahigh vacuum
UPS: ultraviolet photoelectron spectroscopy
VB: valence band
VBM: valence band maximum
XRD: X-ray diffraction
XPS: X-ray photoelectron spectroscopy

Fu 2007

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Fu 2007

Caricato da

Copyright:

Formati disponibili

Surface Science Reports 62 (2007) 431498

Interaction of nanostructured metal overlayers with oxide surfaces

Accepted 29 July 2007

E-mail addresses: qfu@dicp.ac.cn (Q. Fu), t.wagner@fkf.mpg.de (T. Wagner).

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

For materials science, metal/oxide interfaces can be found in

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

Fig. 1. Schematic of a metal island on a flat oxide support in thermodynamic

this, consider Fig. 1 which shows a schematic of a metal

Here, oxide is the surface free energy of the oxide substrate,

Thus, the mechanical strength at the interfaces is closely

Fig. 2. Schematic of important regions of a MOS field-effect transistor gate

supported on TiO2 which are reduced at high temperatures [14,

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

The knowledge of metal/oxide interfaces has been discussed

Fig. 4. Energy band diagram of a metaln-type semiconductor interface at the

where (r ) is the space charge density, is the static

where E g is the band gap energy of the semiconductor. o

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

a metal comes into contact with a semiconductor, is due to

Fig. 5. Energy band diagram of a metaln-type semiconductor contact with an

Schottky limit and the Bardeen limit in order to analyze charge

with defined as the thickness of the interfacial layer, and it is

2.1.2. Chemistry at metalsemiconductor interfaces

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

the bond picture which considers interface chemical bonding

2.2.1.2. Image charge theory. Stoneham and Tasker [62]

where r is the distance parallel to the surface. At a metalionic

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

seems to allow some states originally in the valence and

In case of a compound solid, the electronegativity is given by

example of this is the oxidation of metal adatoms and reduction

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

Recall from 2.1.1 that ionic semiconductors, e.g., SrTiO3 ,

where N D is the donor concentration in the bulk semiconductor,

with the relevant boundary conditions, V (z = 0) = VBB =

The transferred charge per unit area from the semiconductor to

(2) The finite interface (metal clusters or discontinuous

As seen in the above equations, both d and Q are functions of

of 1.5 nm and only 0.01 for crystallites larger than 10 nm [79].

Of all the reactions occurring at metal/oxide interfaces,

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

been prepared by molecular beam epitaxy (MBE) at room

The former case, where MeII is reduced during alloy formation,

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

Main mass transport process

MeI O y k MeII Oxy

RMa on TiO2 & SrTiO3

NMb on CeO2 & SiO2

MeII Ox k MeI k MeII Ox

NMb on TiO2 & CeO2

Metal atoms and/

a RM: reactive metals.

In alloy formation processes, it is mostly expected that

This reaction results in blocking active catalytic sites on metal

the bulk thermodynamic data. For interface reactions, the

Q. Fu, T. Wagner / Surface Science Reports 62 (2007) 431498

case of a redox reaction at a metal/oxide interface, where an

2.3.3. Kinetic consideration

W is the activation energy; is the ionic attempt frequency;

which can be written as

with a diffusion coefficient D = 4a 2 exp(W/k B T ). This is