Growth of NiO on Ag(001): Atomic environment, strain, and interface relaxations studied by polarization dependent extended X-ray absorption fine structure

Article abstract of DOI:10.1021/jp027788n

This paper reports a structural study of 3 and 10 monolayer (ML) thick NiO films grown on Ag(001). Polarization-dependent X-ray absorption spectroscopy at the Ni K edge allowed us to obtain an accurate description of the local atomic environment of the Ni atoms up to the seventh coordination shell, including the determination of the in-plane and out-of-plane strains for the NiO films (determined with an accuracy better than 1percent, resulting from the determination of Ni-O bond distances with a statistical error lower than 0.01 A?). Multishell multiple scattering calculations have been used to fit the data. Within the experimental errors, the 3 ML film has both in and out-of-plane Ni-O distances compatible with those expected in the case of perfect pseudomorphism on Ag(001). A rough film morphology, possibly including also the presence of NiO islands, is also suggested by the optimized coordination numbers. Conversely the 10 ML film is partially relaxed without significant deviation from ideal bidimensional film. The present EXAFS study allows us to rule out any significant atomic interdiffusion process between the substrate and the NiO film. Comparison with previous literature data allows us to make the hypothesis of a two step growth mechanism for NiO films on Ag(001). The NiO-Ag(001) interface distance determined in this EXAFS study (d_interface = 2.36 ± 0.05 A?) is in remarkable agreement with theoretical calculations (d_interface = 2.38 A?).

Full text of DOI:10.1021/jp027788n

J. Phys. Chem. B 2003, 107, 4597-4606
4597
Growth of NiO on Ag(001): Atomic Environment, Strain, and Interface Relaxations
Studied by Polarization Dependent Extended X-ray Absorption Fine Structure
E. Groppo, C. Prestipino, and C. Lamberti*
Department of Inorganic, Physical and Materials Chemistry, UniVersity of Torino,
Via P. Giuria 7, I-10125 Torino Italy; also with INFM U.d.R. di Torino
P. Luches and C. Giovanardi
INFM- National Center on nanoStructures and bioSystem at Surfaces (S3) and Dipartimento di Fisica,
UniVersita` di Modena e Reggio Emilia, Via G. Campi 213/a, 41100 Modena, Italy
F. Boscherini
INFM and Dipartimento di Fisica, UniVersita` di Bologna, Viale C. Berti Pichat 6/2, 40127 Bologna, Italy
ReceiVed: December 20, 2002; In Final Form: March 7, 2003
This paper reports a structural study of 3 and 10 monolayer (ML) thick NiO films grown on Ag(001).
Polarization-dependent X-ray absorption spectroscopy at the Ni K edge allowed us to obtain an accurate
description of the local atomic environment of the Ni atoms up to the seventh coordination shell, including
the determination of the in-plane and out-of-plane strains for the NiO films (determined with an accuracy
better than 1%, resulting from the determination of Ni-O bond distances with a statistical error lower than
0.01 Å). Multishell multiple scattering calculations have been used to fit the data. Within the experimental
errors, the 3 ML film has both in and out-of-plane Ni-O distances compatible with those expected in the
case of perfect pseudomorphism on Ag(001). A rough film morphology, possibly including also the presence
of NiO islands, is also suggested by the optimized coordination numbers. Conversely the 10 ML film is
partially relaxed without significant deviation from ideal bidimensional film. The present EXAFS study allows
us to rule out any significant atomic interdiffusion process between the substrate and the NiO film. Comparison
with previous literature data allows us to make the hypothesis of a two step growth mechanism for NiO films
on Ag(001). The NiO-Ag(001) interface distance determined in this EXAFS study (dinterface ) 2.36 ( 0.05
Å) is in remarkable agreement with theoretical calculations (dinterface ) 2.38 Å).
1. Introduction
efficient chemisorption processes performed on single crystals,
thin films, and powdered systems.2-7
Few monolayer (ML) thick NiO and MgO films on Ag(001)
exhibit a surprising chemical reactivity with a high degree of
chemical specificity and the sticking coefficient for H2O
chemisorption, which is close to unity.1 This has been suggested
to be due to the presence of hybridized states with strong O 2p
character, which are pushed close to and above the Fermi level
due to the mixing with the substrate Ag 5sp states, in an energy
region within the bulk oxide band gap. The confirmation of
this scenario would imply that, because of hybridization effects,
oxide-metal interfaces have unprecedented chemical properties.
However, the high hydroxylation rate observed for MgO and
NiO films on Ag(100) might also be caused by the intrinsic
crystal structure of the oxide thin film. In fact, it is by now
generally recognized that the chemical activity of oxide surfaces
is strongly determined by the coordination number and the
oxygen-cation bond length, so that, for example, low coordi-
nated step edges and corners are preferential sites for very
Of particular relevance in the understanding of several
fundamental phenomena has been the study of the interaction
of H2O with NiO and parent MgO oxide surfaces.2,8-12 Reissner
et al. reported that the H2O uptake on thin NiO films grown on
Ag(001) increases with increasing surface defectivity.11 H2O
undergoes heterolytic dissociation, through a nucleophilic attack
by the basic O2- anions; hence, the MgO surface becomes
immediately covered by a full layer of hydroxyl groups, as
demonstrated by some IR investigations.13 In this regard also,
the contributions of Echterhoff et al.14 and Kno¨zinger et al.15
together with some theoretical contribution16-21 should be noted.
Among them, Scamehorn et al.18 report two stable configurations
for H2O molecules physisorbed on MgO(001): H2O either forms
a bifurcate hydrogen bond with the O2- ions of the surface or
an oxygen down configuration on top of the Mg2+ cation. Xu
and Goodman22 reported a combined TPD, IRAS, and LEED
study on D2O adsorbed on thin MgO(001) films grown on a
Mo(001) substrate. As far as experiments on MgO(001) single
crystals are concerned, elastic helium scattering and LEED
investigations were carried out by Ferry et al.23,24 and a
* To whom correspondence should be addressed. E-mail:
carlo.lamberti@unito.it. Phone: +390116707841. Fax: +390116707858.
10.1021/jp027788n CCC: $25.00 © 2003 American Chemical Society
Published on Web 04/16/2003

4598 J. Phys. Chem. B, Vol. 107, No. 19, 2003
Groppo et al.
combined LEED and IR study (with polarized IR source) was
reported by Heidberg et al.25 The hydroxyl groups formed on
low coordinated sites of MgO(001) after dissociative adsorption
of water have been characterized with high-resolution EELS
by Xu et al.26 and by Coustet and Jupille in a combined high-
resolution EELS and XPS study.27-29 Of particular relevance
for the present work is the thickness-dependent investigation
of the strain in MgO films (1-20 ML) pseudomorphically
grown on Ag(001).30 The final picture emerging from all these
studies is that, on both NiO and MgO surfaces, regular five
coordinated sites are not able to dissociate H2O and that the
process occurs on four and three coordinated sites only. The
reactivity of NiO films on Ag(001) toward potassium has been
studied using ultraviolet photoelectron spectroscopy (UPS),
X-ray photoelectron spectroscopy (XPS), and low energy
electron diffraction (LEED).31
dicular to the electric vector of the impinging radiation, it is
possible to preferentially probe out-of-plane or in-plane atomic
correlations. For example, the technique has been recently used
to directionally probe distortions of the local structure induced
by heteroepitaxial growth in cubic semiconductor alloys45 and
in hexagonal nitrides.46 The sensitivity to small strain-induced
variations in bond lengths has been demonstrated for InGaAs
on InP.47,48 The ability of EXAFS to detect atomic interdiffusion
processes at interfaces is illustrated by work on Ge dots on Si.49
In the field of oxides, EXAFS has been used to probe the local
structure in magnetron-sputtered NiO films50 and to determine
the adsorption site of Ag on MgO.51
2. Experimental Section
2.1. Sample Preparation. The samples used for the present
study are two ex situ grown NiO films on Ag(001) single
crystals 3 and 10 ML thick. The films were protected by a
capping layer of 5 ML of MgO to prevent hydroxylation of the
NiO film surface. Given the small lattice parameter mismatch
between NiO and MgO (0.8%), the possible modifications of
the structure induced by the capping layer are expected to be
negligible. It is so assumed that the local structure of ex situ,
MgO-capped, NiO/Ag(001) films is virtually equivalent to that
of in situ NiO/Ag(001) films. The Ag(001) substrate was cleaned
by repeated cycles of sputtering (0.6 keV, 1µA/cm2) and
annealing (700 K) in UHV. The evaporation rates were
measured by means of a quartz crystal microbalance. Both NiO
films and MgO capping layers were grown using Knudsen cells
and a directional O2 flux on the sample resulting in a background
oxygen pressure of 7 × 10-8 Torr in the growth chamber. The
rate of oxide formation was approximately 1 ML/min, as
evaluated by the deposition rates and the relative Ni (Mg)
density in metallic Ni (Mg) and in NiO (MgO). The Ag substrate
was kept at 460 K during film growth. In the following, we
shall express the film thickness in monolayers (ML), where 1
ML corresponds to a number of atoms/cm2 equal to the surface
density of atoms in bulk NiO (1.15 1015 Ni and O atoms/cm2).
A sintered bulk NiO sample was measured as a reference.
Nickel hydroxide powder, sintered at 900 °C, was used as a
precursor of NiO. The resulting NiO is highly sintered and is
characterized by many cubic or prismatic microcrystals preva-
lently exposing (100) and (111) planes and with dimension in
the range of 500-2000 Å.5
From this picture, it is clear that any modification of the bond
length and of the coordination number due to lattice deforma-
tions associated with the lattice misfit of the oxide with the
substrate can significantly influence the chemical properties of
the oxide thin film. For this reason, the choice of a substrate
with a lattice parameter very similar to that of the film and with
the same crystallographic structure is of fundamental importance.
The lattice parameter of Ag is sufficiently close to that of NiO
and MgO, yielding misfits of 2.2% and 3%, respectively;
furthermore, the speed of Ag oxidation is lower than that of Ni
or Mg and this prevents the substrate oxidation and consequent
structural changes.
Ultrathin NiO films grown on Ag(001) present also interesting
magnetic properties32,33 different from the bulk.34 Their structure
has been studied with different techniques, such as Auger
electron diffraction,35 photoelectron diffraction,36 primary beam
diffraction modulated electron emission (PDMEE),37 LEED,37-39
electron energy loss,39 and direct31,40 and inverse40 electron
photoemission spectroscopies. The films are found to be
pseudomorphic to Ag up to a thickness of ∼5 ML. For higher
coverages, the strain is suggested to be released by the
introduction of misfit dislocations.38 XPS has been used by
Luches et al.36 and by Reissner et al.,10,11 the latter group having
also reported an ultraviolet photoelectron spectroscopy study.
The film morphology has been investigated by means of
STM,41,42 which suggested a rather complex interface with NiO
island formation and the presence of vacancy islands in the Ag
substrate after NiO deposition.
2.2. Data Collection. The EXAFS measurements were
performed at the GILDA beamline, at the European Synchrotron
Radiation Facility (Grenoble, France). The photon energy in
the Ni K edge range was obtained by means of a double crystal
monochromator used in dynamical focusing mode.52 The
EXAFS spectra of the films were recorded in the fluores-
cence mode using a 13 elements hyper-pure Ge detector,
whereas the sintered NiO sample was measured in the transmis-
sion mode.
Being the synchrotron radiation linearly polarized in the plane
of the electron orbit, the substrate Ag(001) single crystal can
be oriented in the electric field (E) of the photon beam in order
to modulate the angle θ between the growth axis and E ideally
in the 0-90° range. To discuss the expected polarization
dependence of the spectra, we will use the approximation in
which each two-body atomic correlation contributes to the total
signal with a weight equal to cos2(R) where R is the angle
between the interatomic vector r and the electric vector E. This
approximation is valid within the plane wave and small atom
limits and, in fact, has not been adopted in the data analysis;
In this work, we have used extended X-ray absorption fine
structure (EXAFS) to determine the local atomic environment
of Ni within thin oxide films pseudomorphically grown on
Ag(001) substrate. The good quality of the EXAFS data allows
us to perform the data analysis by including up to the 7th
coordination shell around Ni. This result represents a strong
improvement with respect to a preliminary analysis performed
up to the third coordination shell only.43 In this work, we have
determined the variation of interatomic distances induced by
strain and the epilayer-substrate atomic correlations with
remarkably low statistical errors. EXAFS is a powerful tool for
local structural determination, which has been used to study
surface and interface structure.44 The use of third generation
synchrotron radiation sources, coupled to state of the art
beamlines and experimental setups allows us to obtain data of
high quality even on few ML-thin epilayers. The polarization
dependence of the EXAFS cross section allows structural
determination with directional sensitivity; in fact, by exploiting
the linear polarization of the synchrotron beam and by orienting
the sample with the surface normal either parallel or perpen-

Growth of NiO on Ag (001)
J. Phys. Chem. B, Vol. 107, No. 19, 2003 4599
the integration, in k space, of the amplitude function of each
path (A). The value of the A integral of the first-shell single
scattering contribution is arbitrarily set to 100 and all remaining
A values scale in a proportional way. All paths having an A
value higher than 10 or lower than 5 have been included or
ignored in a straightforward manner. For all paths having 5 e
A e 10, a manual choice has been done depending on the k
and R regions where they give their main contribution. It is
worth noticing that, in the particular case of oxygen and nickel
atoms, which are separated in the periodic table by 20 elements,
the maxima of the respective scattering amplitudes are shifted
in k space by ∼4 Å-1, so that the oxygen and nickel atoms
contribute mainly in the 1-6 Å-1 and 5-15 Å-1 ranges,
respectively.
Figure 1. Representation of the portion of the NiO cluster adopted
for the FEFF calculation lying in the positive region of the x, y, and z
axes. Small spheres represent the Ni, whereas the big ones refer to O.
Number 0 identifies the absorbing Ni atom, whereas the progressive
1-7 numbers refer to the 1st-7th coordination shell around Ni(0). All
atoms belonging to the same shell are degenerate in the cubic symmetry
of the bulk case and so do the scattering paths. The letters following
the numbers are used to distinguish atoms of the same shell once the
degeneration is removed by tetragonal distortion.
The data processing was performed through five subsequent
steps. (i) The ø(k) function has been extracted from the raw
absorption data using AutoBK58 program. (ii) A structural model
of the reference compound (NiO bulk) was constructed using
the ATOMS program.59 (iii) Theoretical phase shifts and
amplitudes, for the paths satisfying the Rpath e Rcut-off relation-
ship, were generated by means of FEFF 8.10 code57 on the basis
of the NiO rock-salt structure; the polarization was explicitly
taken into account (vide infra) when generating the theoretical
signal. (iv) For the reference sample (sintered NiO), the
experimental EXAFS signal was fitted with the theoretical one,
using the FEFFIT 2.55 program.60 This allowed us to determine
the most significant multiple scattering paths and the S02 factor
(superposition integral of the passive electrons wave functions)
and to verify the ability of the adopted phases and amplitudes
to reproduce the structure known by diffraction data. (v) On
the basis of the results obtained for the reference sample, data
fitting of the EXAFS signal was performed for the films also.
however, it does provide the major polarization dependence and
is thus useful for the sake of a qualitative discussion. As a
consequence, a modulation in θ implies a modulation in cos2(R),
i.e., in the relative weight between in-plane and out-of-plane
distances of the pseudomorphic film. In the ideal cases of θ )
0 and 90°, only one contribution is present. For practical reasons
the θ ) 90° and 0° limit cases cannot be reached. For the films,
we used two different scattering geometries at θ ) 75° and
15°, to be selectively sensitive to in-plane and out-of-plane
structures respectively, hereafter labeled as normal and grazing
incidence. In such cases, the mixing with the undesired
contribution is, within the plane wave approximation, equal to
cos2(15°), i.e., less than 7%, which can be considered as
negligible in the EXAFS data analysis. To allow the simulta-
neous determination of both in-plane and out-of-plane distances
also a geometry with θ ) 45° has been used.
2.3. EXAFS Data Analysis. Nickel oxide, which has the
rock-salt structure with the Ni ions forming a face-centered-
cubic (rocksalt, see Figure 1) lattice, is an example of compound
where multiple scattering is expected to be strong because of
the presence of forward-scattering paths consisting of collinear
arrangements of neighboring atoms (also called focusing ef-
fects).53,54 The importance of the MS contributions in the
EXAFS data of rock-salt oxides has clearly been evidenced in
previous papers reporting EXAFS studies on both NiO55 and
MgO56 polycrystalline powders and on NiO films on polymide
film substrate.50 For this reason, data analysis was carried out
using the FEFF 8.10 code57 where the effect of multiple
scattering can be included by calculating all single and multiple
scattering paths, starting and ending on the absorbing atom,
within a total path length shorter than a cutoff threshold
(2Rcut-off). For single scattering paths, the total path length is
simply twice the distance between the absorbing and the
scattering atoms, i.e., twice the distance of the ith coordination
shell (hereafter Ri). To allow a direct comparison between
scattering path lengths and atomic distances, FEFF 8.10 orders
the different paths as a function of Rpath, defined as the half of
the total path length. In such a way, for all single scattering
paths, Rpath will coincide with the corresponding Ri.
The model ø(k) used to compare to data is evaluated as a
sum over all selected paths and depends on the scattering
amplitudes and phase-shifts and on the standard EXAFS
parameters (coordination number, change in distance, Debye-
Waller Factors) of each path and on a unique shift in energy
origin, common to all paths. The EXAFS contribution of each
selected scattering path is adjusted by optimizing the above-
mentioned parameters until the best fit to the data is found.
Coming to the analysis of the reference sample (sintered NiO),
an Rcut-off ) 6 Å was adopted. Within this value, 14 different
scattering paths (see the Appendix) were considered to simulate
the theoretical EXAFS signal, each one weighted by the
corresponding degeneration factor (6 for the 1st shell SS, 12
for the 2nd shell SS, and so on), automatically computed by
the FEFF code. As described above, the standard EXAFS
parameters are a single S02 factor and a single energy (∆E) shift
(common to all paths), the Debye-Waller factors (σi), the
coordination numbers (Ni), and the change in distances (∆Ri
with respect to an initial Ri guess). This results in a huge number
of parameters for 14 different paths. To minimize the number
of optimized fitting parameters we have fixed the coordination
numbers to the values expected for a rocksalt structure and all
14 path length variations ∆Ri have been expressed as a function
of a single parameter â, representing the isotropic volume
expansion (contraction if â < 1) coefficient. It is so assumed
that the NiO cluster undergoes an isotropic expansion; i.e., the
distances along all three axes expand or contract by the same
constant of proportionality. In such a way, for all paths the
Because the number of paths Npath(Rcut-off) rapidly diverges
with Rcut-off, [Npath(3 Å) ) 2, Npath(4 Å) ) 5, Npath(5 Å) ) 21,
Npath(6 Å) ) 75 for bulk NiO], a discrimination method must
be applied to limit the number of paths included in the
simulations. The adopted method is based on the evaluation of
optimized path length can be expressed as the product R Rp0ath
,
being R0path the initial guess given by XRD. In such approach,
the total number of optimized parameters is reduced to 13: (i)
2
one S0 ; (ii) one ∆E; (iii) one â; and (iv) 10 σi factors. Note

4600 J. Phys. Chem. B, Vol. 107, No. 19, 2003
Groppo et al.
that σi factors have been independently optimized for the seven
SS paths, whereas for the MS paths, we have fitted independ-
ently only those related to the two collinear paths contributing
in the 4th shell region. A single Debye-Waller factor was
considered for the four MS paths contributing in the 7th shell
region.
Coming to the films analysis, the polarization of the
synchrotron beam (i.e., the direction of the electric field E of
the incident beam) was explicitly taken into account. Assuming
the z axis as the film growth axis and θ as the angle between
z and E, hereafter, we will use the following correspondence
between polarization direction and incidence geometry: polar-
ization along [100] S normal beam incidence (θ ) 75°),
polarization along [001] S grazing beam incidence (θ ) 15°),
polarization along [101] S 45° incidence (θ ) 45°). The
introduction of the polarization removes the degeneration of
several scattering paths. As a consequence, the number of paths
within Rcut-off ) 6 Å increases from 75 to, in the case of [100]
polarization as an example, 167. In the Appendix, the case of
polarization along the [100] direction is reported in detail as an
example. The paths considered to simulate the theoretical
EXAFS signal of the films were almost the same used for the
reference sample. However, in this case, we must distinguish
between the in-plane and out-of-plane contributions because of
the different interatomic distances in the two directions (see
Figure 1 and the Appendix for the [100] polarization case).
Regarding the fitting parameters, a separate discussion has
to be done for the signal amplitude and the interatomic distances.
Figure 2. Vertically shifted for clarity, from top to bottom the k3-
weighted, phase uncorrected, FT of the ø(k) of experimental (points)
superimposed with the best fit up to the 7th coordination shell (solid
line); 1st shell SS; 2nd shell SS; 3rd shell SS; 4th shell SS; Ni(0)-
O(1a)-Ni(4a)-Ni(0) MS path; Ni(0)-O(1a)-Ni(4a)-O(1a)-Ni(0)
MS path; 5th shell SS; Ni(0)-O(1a)-O(5a)-Ni(0); 6th shell SS; 7th
shell SS; Ni(0)-Ni(2a)-Ni(-2a)-Ni(0) MS path; Ni(0)-Ni(2a)-
Ni(7a)-Ni(0) MS path; Ni(0)-Ni(2a)-Ni(0)-Ni(-2a)-Ni(0) MS
path; Ni(0)-Ni(2a)-Ni(7a)-Ni(2a)-Ni(0) MS path. See Figure 1 for
the path definition. White and gray-dashed parts differentiate the R
region where the experimental data have been fitted to those excluded
from the fitting procedure.
2
The EXAFS signal amplitude depends essentially on the S0 ,
the Debye-Waller factors, and the coordination numbers, for
which the following choices were done. (i) Being S02 exclusively
a function of the absorbing atom, it was fixed at the value
obtained for the reference sample. (ii) The Debye-Waller
factors of all of the selected scattering paths were fitted starting
from the values obtained for the reference sample; for the MS
paths, we have adopted the same strategy as for the reference
sample. (iii) Only the coordination numbers of the 1st and 2nd
shells were fitted, because of the necessity of minimizing the
fitting parameters; the remaining coordination numbers were
fixed at the values expected for a rocksalt structure in the case
of the 10 ML film and were multiplied by a scaling factor x
constant for all of the shells in the case of the 3 ML film
measured at normal incidence.
Coming to the distances, the fits of the films data were
performed on the hypothesis of an ideal tetragonal distortion;
that is, all of the in-plane and out-of-plane distances are
contracted/extended in the same proportion. On these bases, we
have expressed all of the changes in interatomic distances with
respect to the bulk case in terms of variations of the in-plane
(∆R|) and out-of-plane (∆R ) Ni-O bond distance (half of the
lattice parameter, see the Appendix for the [100] polarization
case). With the NiO lattice parameter being less than the Ag
one, we expect a compression of the in-plane distances (∆R|<
0) and an extension of the out-of-plane ones (∆R > 0).
weight to emphasize the high wavenumber part of the spectrum.
In the bottom part of Figure 2, the single contributions of all of
the SS and MS paths taken into account during the fit are also
shown, vertically shifted for clarity. The experimental curve is
characterized by five dominant peaks. By comparing the
experimental curve with the single path contributions, the
following considerations can be drawn (quoted distances are
phase uncorrected). The peak around 1.7 Å is due to the first
shell Ni-O atomic correlation, whereas the peak at ∼2.6 Å is
mainly due to the second shell Ni-Ni one. The higher distances
peaks around 3.8, 4.7, and 5.5 Å result from the superimposition
of several contributes but are basically centered where the 4th,
6th, and 7th coordination shells (Ni scatterers in all three cases)
occur. The peaks relative to the 3rd and 5th coordination shells
are not resolved due to the low backscattering amplitude of O
compared to Ni (vide supra): corresponding contributions
appear in the minimum of the experimental FT between the
2nd and 3rd and between 3rd and 4th peaks. The obtained â
(1.002 ( 0.002) value results in a Ni-O first shell distance of
R ) 2.092 ( 0.004 Å, to be compared with a value of 2.088
obtained from XRD. This datum, together with the obtained
Rfactor (0.037), confirms the validity of the choice of the
scattering paths previously discussed and the accuracy of the
corresponding phases and amplitudes functions computed by
FEFF8.10 code. On the basis of these results, we have extended
the fitting model to the films.
3. Results
3.1. EXAFS Data Analysis of the NiO Bulk Reference
Material. Figure 2 (top) shows the magnitude of the phase
uncorrected Fourier transform (FT) of the k3ø(k) of the reference
sample, both for experimental data (points) and for the best fit
(solid line). The fit was performed in R space in the 1.0-6.0 Å
range (evidenced by the white region in the figure), to take into
account all the contributions until the 7th coordination shell.
The FT was calculated in the range 2.5-15 Å-1, with a k3
3.2. EXAFS Data Analysis of the NiO/Ag(001) Films. The
raw EXAFS data are reported in Figure 3a. The six spectra
collected in the fluorescence mode are of remarkable quality

Growth of NiO on Ag (001)
J. Phys. Chem. B, Vol. 107, No. 19, 2003 4601
Figure 3. Part a: Raw kø(k) data of, from top to bottom: NiO sintered,
10 ML (θ ) 75°); 10 ML (θ ) 45°); 10 ML (θ ) 15°); 3 ML (θ )
75°); 3 ML (θ ) 45°); 3 ML (θ ) 15°). Part b: First shell contributions
as reported by the modulus of the k3-weighted, phase uncorrected, FTs
of the data shown in part a of 10 ML (θ ) 75°) and sintered NiO
(dotted and full lines, respectively). The amplitude of the FT of the
film has been multiplied by an arbitrary factor to result in a first shell
signal of the same magnitude as the reference sample.
Figure 5. Comparison between experimental (points) and best fit (solid
line) for the modulus of the k3-weighted, phase uncorrected, FTs of
the experimental ø(k) of 3 ML (θ ) 15°). The upper curves refer to a
fit including the signals from the tetragonally distorted NiO film only
(i.e., with the same approach used in the fits reported in Figure 4).
Bottom curves refer to the fit performed by including the 2nd shell
Ni-Mg signal from MgO cap (dotted line) and the 2nd shell Ni-Ag
signal from the Ag substrate (dashed line). White and gray-dashed parts
as in Figure 2.
of-plane SS paths. It is so the best candidate to observe, on a
qualitative ground, the presence of tetragonal strain. In the case
of the film measured at normal incidence, a shift toward lower
R is evident for the first shell contribution, reflecting a
contraction of the in-plane Ni-O bond length (R|), as expected
from pseudomorphism on a substrate with smaller lattice
parameter (Figure 3b). Conversely, the R region where the first
shell peak of the sample measured at grazing incidence appears
is almost similar to that of the reference sample. We are so
unable to estimate, on a qualitative ground (i.e., from a simple
inspection of FT’s), if the expected increase of the out-of-plane
Ni-O bond length (R ) occurs or not. To quantify the R|
contraction and to establish or confute the R elongation, a
careful EXAFS data analysis must be carried out (see section
2.3).
Figure 4. Comparison between experimental (points) and best fit (solid
line) for the modulus of the k3-weighted, phase uncorrected, FTs of
the experimental ø(k) of 10 ML (θ ) 75°), 10 ML (θ ) 15°), and 3
ML (θ ) 75°), vertically shifted for clarity. Only signals from the
tetragonally distorted NiO film contribute to the fit. See Tables 1 and
2 for the quantitative results extracted from these fits. White and gray-
dashed parts as in Figure 2.
The modulus of the k3-weighted, FTs of the experimental ø(k)
for the 10 ML film, at normal and grazing incidence, and for
the 3 ML film at normal incidence is shown in Figure 4 (points),
superimposed to the best fit (solid line, vide infra for the
optimization strategies). The fits were performed in R space in
the 1.0-5.8 Å range for the 10 ML film (for both geometries)
and in the 1.0-5.1 Å range for the 3 ML film (fit ranges are
emphasized by the white region in the figure), to take into
account all the contributions until the 7th and the 6th coordina-
tion shell, respectively. The FTs were calculated in the ranges
2.0-11.0 and 1.5-10.0 Å-1 for the 10 and 3 ML films,
respectively, with a k3 weight. A qualitative comparison of the
films FTs (Figure 4) with that of the reference sample (Figure
2, top) confirms that no major changes occur in the local
environment of Ni in the epilayers; the only apparent change is
a variation of the relative intensities of the different peaks, being
that related to the second shell signal the most relevant one.
by considering that they have been collected on 3 and 10 ML
thick films, corresponding to ∼3.45 × 1015 and ∼1.15 × 1016
Ni atoms cm-2. In all fluorescence spectra, the oscillations and
the main beats of the kø(k) of the sintered NiO (measured in
transmission mode, top spectrum) are visible up to 10 Å-1 and
in some cases up to 12 Å-1. Such closeness with the fine
structure of the model spectrum indicates that the local
environment of Ni atoms in the films reflects the structure of
NiO within several coordination shells (vide infra Figures 4 and
5 for the corresponding FTs).
The first coordination shell around Ni is the only one
unaffected by MS contributions showing only in-plane and out-

4602 J. Phys. Chem. B, Vol. 107, No. 19, 2003
Groppo et al.
TABLE 1: In-Plane (R|) and Out-of-Plane (R ) Ni-O First
Shell Bond Lengths and Corresponding Strains E| )
(R| - R0)/R0 and E ) (R - R0)/R0 for the 3 and 10 ML
Thick NiO/Ag(001) Films as Obtained from the EXAFS Data
Analysis at Different Polarization Geometriesa
N1 was optimized exactly to the expected single-crystal value).
Conversely, a strong improvement has been obtained for the 3
ML (θ ) 75°) and 3 ML (θ ) 45°), for which x was optimized
to 0.7 ( 0.1. Notwithstanding the use of variable N1 and N2
parameters, the optimization of the 3 ML (θ ) 15°) signal still
results incompatible with the experimental data.
It is finally worth noticing that the present EXAFS study
allows to rule out any significant atomic interdiffusion process
between the substrate and the NiO film, a kind of defectivity
easily detected by EXAFS.49
sample
θ
R| (Å)
2.088
2.092 ( 0.004 +0.2 ( 0.2 as R|
ꢀ| (%)
R
(Å)
ꢀ
(%)
XRD
sintered
0
as R|
as ꢀ|
as ꢀ|
10 ML 75 2.052 ( 0.006 -1.7 ( 0.3
10 ML 15
10 ML 45 2.062 ( 0.015 -1.2 ( 0.7 2.091 ( 0.024 +0 ( 1
3 ML
3 ML
3 ML
2.101 ( 0.006 +0.6 ( 0.3
75 2.046 ( 0.009 -2.0 ( 0.4
15
45 2.044 ( 0.009 -2.1 ( 0.4 2.133 ( 0.018 +2.1 ( 0.9
3.3. EXAFS Data Analysis of the 3 ML Thick NiO/Ag(001)
Films with Beam Polarization in the [001] Direction. A
different strategy must be used for the 3 ML film measured at
grazing incidence. Notwithstanding the good quality of the
experimental spectrum (bottom spectrum in Figure 3a), we were
not able to perform a fit using the NiO contributions only (see
Figure 5, upper curves). In fact, in grazing incidence geometry,
we are sensitive to the out-of-plane distances so that we cannot
ignore the contributions of the Ag substrate and of the MgO
capping layer. Note that for the 10 ML thick sample such
contributions have a much lower weight (1/20 instead of 1/6) and
can be ignored.
According to the experimental results of Giovanardi et al.,68
to the theoretical calculation of Duffy et al.62 and of Casassa et
al.33 for the NiO/Ag(001) system and in analogy with the MgO/
Ag(001) case,63 we assumed that the oxygen atoms accom-
modate themselves on the substrate silver atoms. In the
hypothesis of an ideal planar interface without interdiffusion,
Ag atoms give rise to a 2nd shell contribution. This contribution
has to be weighted for a 1/6 factor, because it represents half of
the 2nd shell neighbors (the inferior ones) seen by 1/3 of the Ni
atoms (those of the 1st ML). The SS Ni-Ag contribution to
the EXAFS signal was simulated with a path generated by
FEFF8.10 in the case of bulk NiO after substitution of the Ni
2nd shell atoms with Ag atoms. In such a way, the path is
characterized by an initial Ni-Ag distance value equal to that
of the Ni-Ni distance in bulk NiO which must be necessarily
optimized during the fit.
2.12 ( 0.02
+1.4 ( 0.9
a Experiments performed in normal (grazing) incidence are sensitive
to the parallel (perpendicular) values only, whereas measures collected
with θ ) 45° give information on both (although with higher
incertitude). For comparison also, the unstrained Ni-O values obtained
from XRD (R0 ) 2.088 Å) and from the EXAFS data analysis on the
sintered NiO have been reported. See Figures 2, 4, and 5 for a view on
the quality of the fits. The perfect pseudomorphism predicts R| )
1/2aAg(001), i.e., 2.045 Å.
TABLE 2: Optimized First and Second Shell Coordination
Numbers for the 3 and 10 ML Thick NiO/Ag(001) Films
Measured at Normal and Grazing Geometriesa
N1[100]
N1[001]
N2[100]
N2[001]
bulk theory
10 ML theory
10 ML
3ML theory
3 ML
2
2
2
2
2
1.8
2.0 ( 0.6
1.33
8
8
8
8
8
7.2
7 ( 1
5.3
2.8 ( 1.2
1.0 ( 0.2
6.1 ( 1.6
3.2 ( 0.9
a Nonoptimized parameters can be evidenced by the absence of the
relative error bars. For comparison, the “theory” lines report the values
expected in the case of a perfect NiO tridimensional single crystal (bulk
theory) and in the case of perfect bidimensional 10 and 3 ML thick
NiO films. See Figures 4 and 5 for a view on the quality of the fits.
The data collected on the 10 ML film (at θ ) 75°, 15°, and
45°, the latter not shown for brevity) have been satisfactorily
optimized using R| and R as the unique structural parameters
of the fit, being the coordination numbers fixed to those expected
from a perfect NiO single crystal. This approach did not result
in a satisfactory fit for the data collected on the 3 ML film for
any polarization, reflecting the deviation from a perfect NiO
single crystal, even if tetragonally distorted. For θ ) 15° and
45°, this deviation is expected to be due to the limited thickness
of the film. The inability to fit the 3 ML (θ ) 75°) data with
fixed coordination numbers means that the film cannot be
considered as a perfect bidimensional film and that the problem
of surface roughness and of island formation must be considered.
The direct observation of islands on the NiO layer surface has
been indeed evidenced in the STM study of Sebastian et al.41
All data have so been subjected to new optimization cycles
where the coordination numbers of the 1st and 2nd shell were
fitted (hereafter N1 and N2). To minimize the number of fitting
parameters, the coordination numbers of remaining shells and
of MS paths were fixed at the values expected for a rocksalt
structure multiplied by the scaling factor x constant for all of
the shells. The results of these new optimization cycles are
summarized for all cases in Tables 1 and 2 for bond distances
and coordination numbers, respectively. In the 10 ML (θ ) 75°)
and 10 ML (θ ) 45°) cases the introduction of N1 and N2 and
x as fitting parameters does not improve the fit; for this reason,
Tables 1 and 2 report the values obtained from the previous
optimization cycles (N1 and N2 fixed, x ) 1). For 10 ML (θ )
15°), a slightly improved fit was obtained resulting in a little
decrement of N2 and x values, which are however compatible
with the single-crystal values within the error bars (note that
Regarding the MgO capping layer, its influence could be in
principle negligible, because of the small lattice mismatch
between NiO and MgO (about 0.8%); nevertheless, we expect
a deformation in the EXAFS signal because of the smaller
scattering amplitude of Mg with respect to Ni. For this reason,
we have simulated also the Mg contribution (again 2nd shell
contribution) using the 2nd shell path generated by FEFF8.10
in the case of MgO with Ni as absorbing atom.
These two new contributions imply the following additional
parameters to be optimized: RNi-Ag; σNi-Ag; σMg-Ag, being the
1
coordination numbers fixed to those expected by /6 of second
shell neighbors and being the RNi-Mg distance optimized together
with the second shell RNi-Ni. To limit the total number of fitting
parameters, the following restrictions have been done on the
basis of what learnt from the previous optimizations: (i) The
shift in energy origin (∆E) was fixed to the value obtained from
the fit of the 3 ML (θ ) 75°) and (ii) the variation of the in-
plane lattice parameter ∆R| was fixed to the value expected in
the case of a perfect pseudomorphism (R| ) 1/2aAg(001) ) 2.045
Å), a value compatible with that optimized in the 3 ML (θ )
75°) case, see Table 1.
In Figure 5 (lower curves), a comparison between experi-
mental (points) and best fit (including substrate and cap
contributions, solid line) for the modulus, FT of the experimental
k3ø(k) of the 3 ML film measured at grazing incidence is shown.
The FT was calculated in the 1.5-10 Å-1 range. The fit was

Growth of NiO on Ag (001)
J. Phys. Chem. B, Vol. 107, No. 19, 2003 4603
performed in R space in the 1.0-3.5 Å range (evidenced by
the white region of the Figure) to take into account all of the
contributions until the 3rd coordination shell. In the picture also,
the 2nd shell Ni-Mg signal from MgO cap (dotted line) and
the 2nd shell Ni-Ag signal from the Ag substrate (dashed line)
are specifically shown. The presence of the Ni-Ag contri-
bution, occurring in a region where that of the NiO is absent,
is mandatory to obtain a good fit (see upper curves in Fig-
ure 5). Also the introduction of the Ni-Mg contribution
improves the fit, in the 2-3 Å region. We can so, a posteriori,
understand the failure of the fits previously attempted. The
structural parameters extracted from this fit are summarized in
Tables 1 and 2, and the optimized Ni-Ag distance is of 3.13
( 0.04 Å.
4. Discussion
4.1. Coordination Numbers. As previously discussed, a
qualitative inspection of the experimental FTs reveals that the
films spectra are very similar to that of NiO bulk (Figure 3a)
and that the major change is a reduction in the amplitude of
second shell signal with respect to the first shell one. This
reflects a decrease of the 2nd shell neighbors with respect to
the 1st shell ones, which is quantified in the data summarized
in Table 2. It is clear that for the 10 ML film N1 and N2 values
measured in both polarizations are compatible with a model of
a perfectly bidimensionally flat film without roughness effects.
Conversely, for the 3 ML film, only N1[100] fits the theoretical
value within the experimental errors, being N2[100], N1[001],
and N2[001] lower than the expected values. Comparing
experimental and theoretical values, it is clear that the 10 ML
film can be considered, within the EXAFS sensitivity, as
structurally perfect. Instead, in the case of the 3 ML film, a
great discrepancy with the expected values can be noted. This
experimental evidence is in agreement with the hypothesis of
an increasing roughness in surface morphology with decreasing
film thickness, because of island formation at the first stages
of the growth.41,42
4.2. Interatomic Distances. Figure 6 summarizes the out-
of-plane Ni-O distance as a function of the in-plane one,
according with the values reported in Table 1. The open circle
represents the unstrained bulk NiO (R| ) R ) R0 ) 2.088 Å)
as determined by XRD, whereas the full circle represents the
EXAFS value obtained from sintered NiO. The straight line is
obtained following the elastic theory approximation, according
to the equation R ) -γR| + (1 + γ)R0, where the ratio γ is
defined as γ ) 2C12/C11 and C11 and C12 are the elastic stiffness
Figure 6. Out-of-plane Ni-O bond distance vs. in-plane one of thin
NiO films: full square 3 ML (θ ) 15° and 75°), open square 3 ML (θ
) 45°), full triangle 10 ML (θ ) 15° and 75°), open triangle 10 ML
(θ ) 45°) and full circle sintered NiO determined in this EXAFS study.
Straight line represents the theoretical R vs R| relationship predicted
by the elastic theory, adopting the Poisson ratio of the NiO bulk (γ )
0.818).72 Big dot represents the unstrained NiO bulk (XRD), and the
star represents the ideal perfect pseudomorphism of NiO on Ag(001).
to those expected for a perfect pseudomorphism, whereas the
10 ML film appears partially relaxed along the growth direction.
This suggests that the critical thickness for strain release lies
between 3 and 10 ML. The in-plane contraction of the 10 ML
film is in disagreement with the results of other works which,
on the basis of electron scattering experiments, find a critical
thickness for strain release of 5 ML.36,38 This difference can be
understood on the basis of the different sampling depth of the
different techniques. In the case of EXAFS spectra detected in
fluorescence mode, the information comes from the whole
thickness of the sample under study and allows to affirm that
the most part of the atoms in the 10 ML film are still in an
in-plane contracted local environment. Electron scattering
techniques, instead, sample only the outermost film layers, which
can be in a more relaxed environment. With the R values being
obtained from EXAFS an average of the situation seen by Ni
atoms in all layers, the combination of the present EXAFS
results and the electron diffraction ones36,38 suggests a model
where strain relaxation starts around 5-6 ML and affects only
the subsequent layers, being the first 4-5 still under perfect
pseudomorphism.
constants of bulk NiO; values of C11 ) 344.6 GPa and C12
)
141 GPa64 give γ ) 0.818. The star represents the values
expected for an ideal NiO layer in perfect pseudo-
1
morphism on Ag(001): R| ) /2aAg(001) ) 2.045 Å and R )
2.125 Å. The 10 ML (full triangle) and 3 ML (full square)
positions have been represented using the R| and R values
obtained from analysis of the normal and grazing incidence data,
respectively. The R| and R values obtained using the data
collected at θ ) 45° are represented with open triangle and
square for the 10 ML and the 3 ML film, respectively. On the
basis of this picture three main considerations can be drawn.
(i) The values obtained fitting the data collected at incidence
angle of 45° are perfectly consistent, within the experimental
error, with those obtained from the combined analysis of the
data collected at normal and grazing incidence, confirming the
goodness of the model.
(iii) Marcus65 has suggested that the Poisson ratio for thin
films is different from the bulk one. An attempt to perform a
linear fit on the experimental data reported in Figure 6 seems
to support this thesis for the NiO system; however, the too
limited number of experimental points and the relative large
error bars prevent any definitive conclusion. Moreover, par-
ticularly for the 3 ML case, it must also be considered that the
morphology of the system is possibly not two-dimensional but
rather in the form of islands: in this case, the relation between
in-plane and out-of-plane strain is not expected to be simply
via the bulk Poisson ratio.
(ii) Both 10 and 3 ML films seem to be compressed in the
growth plane, to fit the smaller Ag(001) substrate lattice
constant. The 3 ML film presents R| and R values very similar
4.3. NiO-Ag(001) Interfacial Distance. It has been recently
proposed that ultrathin oxide films and oxide-metal interfaces

4604 J. Phys. Chem. B, Vol. 107, No. 19, 2003
Groppo et al.
may actually have unprecedented intrinsic chemical-physical
properties, because of image potential screening of charge
fluctuations and interfacial hybridizational effects.1,66,67 The
presence of a metal substrate, where electrons can move freely,
implies a consistent reduction in the energy cost to transfer one
electron from the valence band (with mainly O2p character) to
the conduction band (mainly centered on the cation), because
of the formation of image charges on the metal at the metal/
oxide interface, resulting in a net reduction of the band gap. In
the case of ultrathin layers (1 ML), even the insulating-metal
transition has been inferred. Altieri et al.1,67 have demonstrated
from X-ray photoemission spectroscopy and Auger electron
spectroscopy experiments that the on-site Coulomb interaction
(defined as the energy necessary to transfer an electron from a
cation to another) and the charge-transfer energy (defined as
the energy cost to transfer an electron from the anion to the
cation) in thin MgO films becomes significantly smaller than
in the bulk. The influence of the metal substrate on these
physical quantities, strictly related to the band-gap width, can
still be felt as far as 10 ML away from the interface. The
demonstrated ability to control the film thickness represents
therefore a new technological opportunity for band-gap engi-
neering. For a given oxide/metal system, the structural parameter
that establishes the Eg vs the film thickness relationship is the
oxide/metal interface distance (dinterface). It is so evident that this
parameter must be carefully determined in order to design a
film with desired electronic characteristic. In principle, once
dinterface has been measured, the Eg vs the film thickness
relationship is fixed, and so the latter can be ad hoc selected to
yield the desired Eg.
Regarding the Ag substrate contribution, the fit performed
on the 3 ML film measured at grazing incidence (see bottom
of Figure 5) gives a Ni-Ag distance of 3.13 ( 0.04 Å. In accord
with the hypothesis of an interface in which the O atoms are
located on top of Ag atoms and assuming a perfect pseudo-
morphic growth (i.e., a Ni-O distance at the interface equal to
1/2aAg(001), a value in agreement with the measured R| ) 2.044
( 0.009 Å, see Table 1), we have calculated the interfacial
distance between the Ag surface and the NiO layer as dinterface
) 2.37 ( 0.05 Å. This value is expanded if compared to both
NiO and Ag half lattice parameters (2.088 and 2.045 Å,
respectively) and is perfectly in agreement with that obtained
with periodic ab initio calculation of Casassa et al.33 performed
with CRYSTAL98 code on one monolayer of NiO on Ag
(dinterface ) 2.38 Å) and with that measured by Giovanardi et
al.68 with PDMEE technique (dinterface ) 2.3 ( 0.1 Å).
5. Conclusions
Polarization dependent EXAFS measurements and their
multiple scattering multishell analysis, up to the 7th coordination
shell, have allowed us a quantitative determination of the
structure of ultrathin NiO films on Ag(001). We found that the
local structure is the rock-salt one and that the intermixing with
the Ag substrate is negligible at the atomic scale. The 3 ML
film has both R| and R values compatible with those expected
in the case of perfect pseudomorphism on Ag(001), within the
experimental errors (0.009 and 0.02 Å respectively). Conversely,
the 10 ML film is partially relaxed: in this case Ni-O bond
distances have been determined with an accuracy of 0.006 Å
for both polarizations. For the 3 ML film, a rough morphology,
possibly including also the presence of NiO islands, is also
suggested by the optimized coordination numbers. From a merge
of all of the literature data and of the present EXAFS results, a
two steps growth mechanism for NiO films on an Ag(001)
substrate can be hypothesized. The first step is characterized
by the nucleation of NiO crystallites perfectly matched with
the substrate, starting from islands: this phase is characterized
by a great surface roughness so that the N2[100], N1[001], and
N2[001] values are lower than the expected ones. In the second
step, occurring around 4-6 ML, the elastic energy stored during
the distortion increases until it is partially released through the
creation of dislocations. At the same time, the NiO islands grow
and begin to connect together until the film grows in a
completely relaxed manner showing, within the experimental
errors, coordination numbers in agreement with the hypothesis
of a perfect bidimensional film. The EXAFS data collected in
grazing geometry on the 3 ML sample detected a significant
Ni-Ag contribution which has allowed us to determine the
NiO-Ag(001) interface distance: dinterface ) 2.37 ( 0.06 Å, a
value in remarkable agreement with both theoretical33 (dinterface
) 2.38 Å) and independent PDMEE experiments68 (dinterface
2.3 ( 0.1 Å).
)
Acknowledgment. This work has been funded by INFM
under PRA-ISADORA and Commissione Luce di Sincrotrone.
We are indebted to the staff of GILDA BM8 beamline for the
important support during EXAFS measurements. We thank D.
Scarano and S. Bertarione for the preparation of the NiO sintered
sample. We are indebted to E. Giamello, A.M. Ferrari, S.
Casassa, C. Pisani, S. D’Addato, and S. Valeri for fruitful
discussion. C. Pisani is also acknowledged for a critical reading
of the manuscript.
TABLE 3: Type, Degeneration, and Length Variation, Due to Tetragonal Distortion of SS Paths Involved in the EXAFS Data
Analysis of the Films in the Case of [100] Polarizationa
paths
degeneration
path length variation
Ni(0)-O(1a)-Ni(0)
Ni(0)-Ni(2a)-Ni(0)
Ni(0)-Ni(2b)-Ni(0)
2
4
4
∆R1 ) ∆R|
x
∆R2in
)
2∆R|
2
1
2
|
2
x
∆R2out
)
2R + 2R (∆R + ∆R ) + ∆R + ∆R - 2R1
x
1
|
2
2
|
2
Ni(0)-O(3)-Ni(0)
8
x
∆R3 ) 3R + 2R (2∆R + ∆R ) + 2∆R + ∆R - 3R1
x
x
1
|
Ni(0)-Ni(4a)-Ni(0)
Ni(0)-O(5a)-Ni(0) and Ni(0)-O(5b)-Ni(0)
2
4-4
∆R4 ) 2∆R|1
2
2
x
∆R5in
)
5R + 10R ∆R + 5∆R - 5R1
1 1 | |
2
1
2
1
2
|
2
Ni(0)-O(5c)-Ni(0)
Ni(0)-O(5e)-Ni(0)
Ni(0)-Ni(6a)-Ni(0)
Ni(0)-Ni(6c)-Ni(0)
4
4
x
x
∆R5out
∆R5out
)
5R + 2R (4∆R + ∆R ) + 4∆R + ∆R - 5R1
x
1
|
2
|
2
)
5R + 2R (∆R + 4∆R ) + ∆R + 4∆R - 5R1
x
1
|
2
1
2
|
2
8
x
x
∆R6 ) 6R + 2R (5∆R + ∆R ) + 5∆R + ∆R - 6R1
x
1
|
2
2
|
2
16
∆R6 ) 6R + 4R (∆R + 2∆R ) + 2∆R + 4∆R - 6R1
x
x1
1
|
Ni(0)-Ni(7a)-Ni(0)
Ni(0)-Ni(7b)-Ni(0)
4
4
∆R7in ) 2 2∆R|
2
1
2
|
2
x
∆R7out ) 2 2R + 2R (∆R + ∆R ) + ∆R + ∆R - 2 2R1
x
1
|
a The atomic labels are referred to Figure 1.

Growth of NiO on Ag (001)
J. Phys. Chem. B, Vol. 107, No. 19, 2003 4605
TABLE 4: As Table 3 for the MS Paths Involved in the EXAFS Data Analysis of the Films in the Case of Polarization [100]
paths
degeneration
2
path length variation
Ni(0)-O(1a)-O(-1a)-Ni(0) and
Ni(0)-O(1a)-Ni(0)-O(-1a)-Ni(0)
Ni(0)-O(1a)-Ni(4a)-Ni(0) and
Ni(0)-O(1a)-Ni(4a)-O(1a)-Ni(0)
Ni(0)-Ni(2a)-Ni(-2a)-Ni(0) and
Ni(0)-Ni(2a)-Ni(0)-Ni(-2a)-Ni(0)
Ni(0)-Ni(2b)-Ni(-2b)-Ni(0) and
Ni(0)-Ni(2b)-Ni(0)-Ni(-2b)-Ni(0)
Ni(0)-Ni(2a)-Ni(7a)-Ni(0) and
Ni(0)-Ni(2a)-Ni(7a)-Ni(2a)-Ni(0)
Ni(0)-Ni(2b)-Ni(7b)-Ni(0) and
Ni(0)-Ni(2b)-Ni(7b)-Ni(2b)-Ni(0)
Ni(0)-O(1a)-O(5a)-Ni(0)
∆R ) 2∆R|
∆R ) 2∆R|
4-2
4-2
4-2
8-4
8-4
x
∆R ) 2 2∆R|
2
1
2
2
x
∆R ) 2 2R + 2R (∆R + ∆R ) + ∆R + ∆R - 2 2R1
x
1
|
|
x
∆R ) 2 2∆R|
2
1
2
2
x
∆R ) 2 2R + 2R (∆R + ∆R ) + ∆R + ∆R - 2 2R1
x
1
|
|
8
8
x
x
∆R ) 1/2[(1 + 2 + 5)∆R|]
∆R ) 1/2[∆R| + 2R2 + 2R (∆R + ∆R ) + ∆R2 + ∆R2 +
Ni(0)-O(1a)-O(5c)-Ni(0)
x
1
1
|
|
2
1
2
2
x
x
5R + 2R (4∆R + ∆R ) + 4∆R + ∆R - ( 2 + 5)R1]
x
1
|
|
Ni(0)-O(1a)-Ni(2a)-Ni(0)
Ni(0)-O(1a)-Ni(2b)-Ni(0)
8
8
x
∆R ) 1/2(2 + 2)∆R|
2
1
2
|
2
x
∆R ) 1/2[∆R| + ∆R + 2R + 2R (∆R + ∆R ) + ∆R + ∆R - 2R1]
x
1
|
Appendix
References and Notes
(1) Altieri, S.; Tjeng, L. H.; Sawatzky, G. A. Phys. ReV. B 2000, 61,
16948.
(2) Zecchina, A.; Scarano, D.; Bordiga, S.; Spoto, G.; Lamberti, C.
AdV. Catal. 2001, 46, 265.
In section 2.3, we have discussed the number of scattering
paths considered to simulate the theoretical EXAFS signal. Here,
all of these paths are described in detail, for the [100]
polarization, referring to Figure 1. Note that a negative shell
number (e.g., Ni(-2a)) refers to the atom obtained from that
reported in Figure 1 with positive shell number (e.g., Ni(2a))
by axis reflection. The scattering paths involved in the analysis
of the reference sample were (i) all of the single scattering paths
(SS) from the 1st to the 7th coordination shell; (ii) the collinear
multiple scattering paths (MS) involving Ni(0)-O(1a)-Ni(4a)-
O(1a)-Ni(0), Ni(0)-O(1a)-Ni(4a)-Ni(0), Ni(0)-Ni(2a)-
Ni(-2a)-Ni(0), Ni(0)-Ni(2a)-Ni(0)-Ni(-2a)-Ni(0), Ni(0)-
Ni(2a)-Ni(7a)-Ni(0), Ni(0)-Ni(2a)-Ni(7a)-Ni(2a)-Ni(0);
and (iii) triangle path involving Ni(0)-O(1a)-O(5a)-Ni(0).
(3) (a) Zecchina, A.; Scarano, D.; Bordiga, S. In Handbook of
Heterogeneous Catalysis; Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; Wiley-
VCH: Weinheim, Germany, 1997; Vol. 2, p 728. (b) Zecchina, A.; Scarano,
D.; Bordiga, S.; Ricchiardi, G.; Spoto, G.; Geobaldo, F. Catal. Today 1996,
27, 403. (c) Zecchina, A.; Scarano, D.; Galletto, P.; Lamberti, C. Il NuoVo
Cimento D 1997, 19, 1773.
(4) (a) Freund, H. J.; Ba¨umer, M.; Kuhlenbeck, H. AdV. Catal. 2000,
42, 334. (b) Freund, H. J.; Dillman, B.; Seiferth, O.; Klivenyi, G.; Bender,
M.; Ehrlich, D.; Hemmerich, I.; Cappus, D. Catal. Today 1996, 32, 1. (c)
Freund, H.-J. Angew. Chem Int. Ed. 1997, 36, 452.
(5) (a) Escalona Platero, E.; Scarano, D.; Spoto, G.; Zecchina, A.
Faraday Discuss. Chem. Soc. 1985, 80, 183. (b) Escalona Platero, E.;
Scarano, D.; Zecchina, A.; Meneghini, G.; De Franceschi, R. Surf. Sci. 1996,
350, 113. (c) Escalona Platero, E.; Coluccia, S.; Zecchina, A. Surf. Sci.
1986, 171, 465.
Regarding the film analysis, we have said that the introduction
of the polarization removes the degeneration of several scattering
paths. As an example, the SS contribution from the 1st
coordination shell is split in two different contributions when a
polarization along the [100] direction is taken into account: the
first path has a component along the polarization direction and
involves two first neighboring atoms (O(1a) and O(-1a), see
Figure 1), whereas the second one has a null component along
the polarization direction and involves four first neighboring
atoms (O(1b), O(-1b), O(1c), and O(-1c)). The intensity of
this last contribution is negligible with respect to the pre-
vious one. The same happens for the successive SS and MS
paths, so that 29 different paths have to be considered in the
data analysis, as reported in detail in the first column of Tables
3 and 4 for SS and MS paths, respectively. All of these 29
different paths are characterized by different interatomic
distances. In Tables 3 and 4 (third columns), we report the
analytical expressions for the variation in scattering path lengths,
caused by pseudomorphism, defined as the difference between
the scattering path lengths in the bulk and in the film cases
respectively, in terms of variations of the in-plane (∆R|) and
out-of-plane (∆R ) Ni-O interatomic distances. All of these
29 paths have been weighted by the corresponding degeneration
factors (second columns in Tables 3 and 4). Because of the
symmetry of the system, the number and the type of the paths
obtained in the case of polarization along the [001] direction
are the same.
(6) Street, S. C.; Xu, C.; Goodman, D. W. Annu. ReV. Phys. Chem.
1997, 48, 43.
(7) Pacchioni, G. Surf. ReV. Lett. 2000, 7, 277.
(8) Thiel, A.; Madey, T. E. Surf. Sci. Rep. 1987, 7, 211.
(9) Brown, G. E., Jr.; Henrich, V. E.; Casey, W. H.; Clark, D. L.;
Eggleston, C.; Felmy, A.; Goodman, D. W.; Gra¨tzel, M. G. M.; McCarthy,
M. I.; Nealson, K. H.; Sverjensky, D. A.; Toney, M. F.; Zachara, J. M.
Chem. ReV. 1999, 99, 77.
(10) Reissner, R.; Schulze, M. Surf. Sci. 2000, 454, 183.
(11) Reissner, R.; Radke, U.; Schulze, M.; Umbach, E. Surf. Sci. 1998,
402, 71.
(12) Schulze, M.; Reissner, R. Surf. Sci. 2001, 482, 285.
(13) Coluccia, S.; Marchese, L.; Lavagnino, S.; Ampo, M. Spectrochim.
Acta A 1987, 43, 1573.
(14) Echterhoff, R.; Kno¨zinger, E. J. Mol. Struct. TEOCHEM 1988, 174,
343.
(15) Kno¨zinger, E.; Jacob, K.-H.; Singh, S.; Hofmann, P. Surf. Sci. 1993,
290, 288.
(16) Goniakowski, J.; Russo, S.; Nougera, C. Surf. Sci. 1993, 284,
315.
(17) Thiel, A.; Madey, T. E. Chem. Phys. Lett. 1993, 209, 340.
(18) (a) Scamehorn, C. A.; Hess, A. C.; McCarthy, M. I. J. Chem. Phys.
1993, 99, 2786. (b) Scamehorn, C. A.; Harrison, N. M.; McCarthy, M. I. J.
Chem. Phys. 1994, 101, 1547.
(19) Anchell, J. L.; Hess, A. C. J. Phys. Chem. 1996, 100, 18317.
(20) Refson, K.; Wogelius, R. A.; Fraser, D. G.; Payne, M. C.; Lee, M.
H.; Milman, V. Phys. ReV. B 1995, 52, 10823.
(21) Langel, W.; Parinello, M. J. Chem. Phys. 1995, 103, 3240. Langel,
W.; Parrinello, M. Phys. ReV. Lett. 1994, 73, 504.
(22) Xu, C.; Goodman, D. W. Chem. Phys. Lett. 1997, 265, 341.
(23) Ferry, D.; Glebov, A.; Senz, V.; Suzanne, J.; Toennies, J. P.; Weiss,
H. Surf. Sci. 1997, 377-379, 377.
(24) Ferry, D.; Glebov, A.; Senz, V.; Suzanne, J.; Toennies, J. P.; Weiss,
H. J. Chem. Phys. 1995, 105, 1697.

4606 J. Phys. Chem. B, Vol. 107, No. 19, 2003
Groppo et al.
(25) Heidberg, J.; Redlich, B.; Wetter, D. Ber. Bunsen-Ges. Phys. Chem.
1995, 99, 1333.
(26) Xu, C.; Estrada, C. A.; Goodman, D. W. Phys. ReV. Lett. 1991,
67, 2910.
(27) Coustet, V.; Jupille, J. Surf. Interface Anal. 1994, 22, 280.
(28) Coustet, V.; Jupille, J. Il NuoVo Cimento D 1997, 19, 1657.
(29) Abriou, D.; Jupille, J. Surf. Sci. Lett. 1999, 430, L527.
(30) Valeri, S.; Altieri, S.; di Bona, A.; Luches, P.; Giovanardi, C.; Moia,
T. S. Surf. Sci. 2002, 507-510, 311.
(31) Schulze, M.; Reissner, R. Surf. Sci. 2002, 507, 851.
(32) Spanke, D.; Solinus, V.; Knabben, D.; Hillebrecht, F. U.; Ciccacci,
F.; Gregoratti, L.; Marsi, M. Phys. ReV. B 1998, 5201.
(33) Casassa, S.; Ferrari, A. M.; Busso, M.; Pisani, C. J. Phys. Chem.
B 2002, 106, 12978.
(47) Romanato, F.; DeSalvador, D.; Berti, M.; Drigo, A. V.; Natali, M.;
Tormen, M.; Rossetto, G.; Pascarelli, S.; Boscherini, F.; Lamberti, C.;
Mobilio, S. Phys. ReV. B 1998, 57, 14619.
(48) Woicik, J. C.; Gupta, J. A.; Watkins, S. P.; Crozier, E. D. Appl.
Phys. Lett. 1998, 73, 2219.
(49) Boscherini, F.; Capellini, G.; Di Gaspare, L.; Rosei, F.; Motta, N.;
Mobilio, S. Appl. Phys. Lett. 2000, 76, 682.
(50) Kuzmin, A.; Purans, J.; Rodionov, A. J. Phys.-Condes. Matter 1997,
9, 6979.
(51) Flank, A. M.; Delaunay, R.; Lagarde, P.; Pompa, M.; Jupille, J.
Phys. ReV. B 1996, 53, R1737.
(52) Pascarelli, S.; Boscherini, F.; D’Acaptito, F.; Hrdy, J.; Meneghini,
C.; Mobilio, S. J. Synchrotron Radiat. 1996, 3, 147.
(53) Teo, B. K. J. Am. Chem. Soc. 1981, 103, 3990.
(54) Filipponi, A.; Di Cicco, A.; Natoli, C. R. Phys. ReV. B 1995, 52,
15122.
(34) Dovesi, R.; Orlando, R.; Roetti, C.; Pisani, C.; Saunders, V. R.
Phys. Status Solidi B 2000, 217, 63.
(35) Marre, K.; Neddermeyer, H.; Chasse, A.; Rennert, P. Surf. Sci. 1996,
358, 233.
(36) Luches, P.; Altieri, S.; Giovanardi, C.; Moia, T. S.; Valeri, S.;
Bruno, F.; Floreano, L.; Morgante, A.; Santaniello, A.; Verdini, A.; Gotter,
R.; Hibma, T. Thin Solid Films 2001, 400, 139.
(55) Corrias, A.; Mountjoy, G.; Piccaluga, G.; Solinas, S. J. Phys. Chem.
B 1999, 103, 10081.
(56) Aritani, H.; Yamada, H.; Nishio, T.; Shiono, T.; Imamura, S.; Kudo,
M.; Hasegawa, S.; Tanaka, T.; Yoshida, S. J. Phys. Chem. B 2000, 104,
10133.
(37) Giovanardi, C.; Di Bona, A.; Altieri, S.; Luches, P.; Liberati, M.;
Rossi, F.; Valeri, S. Thin Solid Films 2003, 428, 195.
(38) Wollschlager, J.; Erdos, D.; Goldbach, H.; Hopken, R.; Schroder,
K. M. Thin Solid Films 2001, 400, 1.
(39) Muller, F.; de Masi, R.; Steiner, P.; Reinicke, D.; Stadtfeld, M.;
Hufner, S. Surf. Sci. 2000, 459, 161.
(40) Portalupi, M.; Duo, L.; Isella, G.; Bertacco, R.; Marcon, M.;
Ciccacci, F. Phys. ReV. B 2001, 64, art. no. 165402.
(41) Sebastian, I.; Bertrams, T.; Meinel, K.; Neddermeyer, H. Faraday
Discuss. 1999, 114, 129.
(42) Bertrams, T.; Neddermeyer, H. J. Vac. Sci. Technol. B 1996, 14,
1141.
(43) Luches, P.; Groppo, E.; Prestipino, C.; Lamberti, C.; Giovanardi,
C.; Boscherini, F. Nucl. Instr. Methods B 2003, 200, 371.
(44) Boscherini, F.; Lamberti, C.; Pascarelli, S.; Rigo, C.; Mobilio, S.
Phys. ReV. B 1998, 58, 10745.
(45) Tormen, M.; De Salvador, D.; Drigo, A. V.; Romanato, F.;
Boscherini, F.; Mobilio, S. Phys. ReV. B 2001, 63, art. no. 115326.
(46) D’Acapito, F.; Boscherini, F.; Mobilio, S.; Rizzi, A.; Lantier, R.
Phys. ReV. B 2002, 66, 205411.
(57) Ankudinov, A. L.; Rave, L. B.; Rehr, J. J.; Conradson, S. D. Phys.
ReV. B 1998, 58, 7565.
(58) Newville, M.; Livins, P.; Yacoby, Y.; Rehr, J. J.; Stern, E. A. Phys.
ReV. B 1993, 47, 14126.
(59) Ravel, B. J. Synchrotron Radiat. 2001, 8, 314.
(60) Newville, M.; Ravel, B.; Haskel, D.; Rehr, J. J.; Stern, E. A.;
Yacoby, Y. Phys. B 1995, 209, 154.
(61) Boscherini, F.; de Panfilis, S.; Weissmuller, J. Phys. ReV. B 1998,
57, 3365.
(62) Duffy, D. M.; Harding, J. H.; Stoneham, A. M. Acta Metall. Mater.
1995, 43, 1559.
(63) Giovanardi, C.; Di Bona, A.; Moia, T. S.; Valeri, S.; Pisani, C.;
Sgroi, M.; Busso, M. Surf. Sci. 2002, 505, L209.
(64) Wang, J. F.; Fisher, E. S.; Manghnzmi, M. H. Chin. Phys. Lett.
1991, 8, 153.
(65) Marcus, P. M. Surf. Sci. 1996, 366, 219.
(66) Duffy, D. M.; Stoneham, A. M. J. Phys. C 1983, 16, 4087.
(67) Altieri, S.; Tjeng, L. H.; Sawatzky, G. A. Thin Solid Films 2001,
400, 9.
(68) Giovanardi, C. Ph.D. Thesis in Physics, Modena, 2002.