Chapter 7
Oxidation of NiAl(110) at medium
oxygen pressures
During thermally controlled oxidation of NiAl, aluminium oxide is formed and its struc-
tural properties, thickness and morphology are strongly dependent on the oxidation con-
ditions. Thin aluminium oxide layers can be grown upon exposure of low-index NiAl
surfaces. Their structure and the growth mode is strongly influenced by the substrate
orientation: a three-dimensional island growth was reported on the (100) surface [25],
whereas well-ordered ultra-thin oxide layers may be grown by oxidation of the (110) sur-
face [7, 8, 9, 25, 102].
We have investigated the ultra-thin aluminum oxide grown on the NiAl(110) and its
transition to bulk oxides as a function of the partial oxygen pressure and temperature.
The main experimental technique employed for this study is surface x-ray diffraction. In
addition, ex situ TEM measurements were performed in order to gain insight on the oxide
microstructure and the oxide/substrate interface.
This chapter is organized as follows: section 7.1 gives an overview of the previous
findings concerning the structural aspects of the ultra-thin aluminium layer, followed in
section 7.2 by the description of the sample preparation procedure and the experimental
set-up used for the diffraction experiments. A comparison between our experimental
diffraction data and the theoretical model of the ultra-thin aluminium model proposed by
Kresse et al. [8] is presented in section 7.3.1. The controversial interfacial structure for the
ultra-thin aluminium oxide on NiAl(110) is also discussed in section 7.3.1. The evolution of
the ultra-thin oxide layer at 350◦C has been followed in situ applying progressively higher
oxygen pressures, thus capturing the initial stages of bulk-like oxide growth. These results
are presented in section 7.3.3. The oxidation of the clean NiAl(110) surface at 800◦C is
discussed in the section 7.3.4. The chapter concludes with section 7.4 which comprises
the summary of the main experimental observations presented here.
106 Oxidation of NiAl(110) at medium oxygen pressures
7.1 Introduction
The ultra-thin aluminium oxide layer formed on the (110) surface of NiAl has a highly
complex structure and has attracted a lot of scientific interest during the last two decades.
The film exhibits a high degree of crystallinity and very good reproducibility in prepa-
ration. By dosing 1200 L oxygen at T=270◦C an amorphous oxide in formed, which
crystallizes upon annealing at T=800◦C [103, 104].
Sustained scientific efforts have been made over the past years in order to solve the
structure of the oxide layer. Various experiments, like low energy electron diffraction ,
electron energy loss spectroscopy (EELS), x-ray photoemission spectroscopy (XPS), angle-
resolved ultraviolet photoemission spectroscopy (ARUPS), auger electron spectroscopy,
scanning tunneling microscopy (STM) and ion scattering spectroscopy (ISS) have been
performed [103, 104]. The most important features proposed on the basis of these inves-
tigations are:
• The oxide film is atomically flat and has a layered structure with a thickness of 5 A˚.
This is compatible with two bilayers consisting each of an aluminium layer and a
distorted hexagonal oxygen layer;
• The oxide is commensurate with the substrate along the [11¯0]NiAl direction and
incommensurate in the [001]NiAl direction;
• There are two domains of the oxide overlayer which are rotated by 24◦ with respect
to the substrate unit cell;
• The oxide film is oxygen-terminated.
Although some initial studies were pointing to a structure derived from the bulk α- or
γ-alumina [103], a detailed atomistic model of the oxide layer was missing for more than
two decades. Extended SXRD investigations have been recently performed by Stierle et
al. [7] and a structural model has been proposed. It has the Al2O3 stoichiometry and
is derived from the κ-alumina polymorph. The oxide is composed of a double layer of
strongly distorted hexagonal arrangement of oxygen ions hosting aluminum ions on both
octahedral and tetrahedral sites with equal probability.
Fig. 7.1(a) shows the LEED pattern of the ultra-thin oxide on NiAl(110) recorded at an
electron energy of 67 eV. The reciprocal unit cell of the substrate is indicated in the figure.
The real space lattice constants were found to be a=18.01 A˚ and b=10.59 A˚, γ= 91.15◦
and the orientation of the oxide lattice relative to the substrate is shown schematically
7.1 Introduction 107
in Fig. 7.1(c). As previously mentioned, the oxide lattice can be rotated by ±24.01◦ with
respect to the substrate [11¯0] direction, corresponding to the formation of twin domains.
The in-plane diffraction map of the oxide layer corresponding to only one domain is
presented in Fig. 7.1(b). The full and the open half-circles indicate the experimental and
the best-fit structure factors, respectively. The red hexagon marks the reflections that
arise from the distorted hexagonal structure having a side length of 3 A˚. The out-of-plane
structure of the oxide was probed by measuring the intensity along the surface rods of
the oxide. Fig. 7.2 shows the surface rod scattering for different values of the in-plane
momentum transfer as a function of the reciprocal lattice coordinate, L. The red curves
represent best-fit structure factors for the proposed structural model, which is shown in
Fig. 7.3. The light and dark blue spheres represent the oxygen ions located at the surface
and interface, respectively, whereas the red and orange spheres indicate the aluminium
ions in the topmost and interfacial layer, respectively. The oxygen ion double layer was
found to be buckled along the [100] direction of the oxide lattice.
Another structural model has been proposed by Kresse et al. [8] based on STM mea-
surements and DFT calculations. The stoichiometry of this structure was reported to
be Al10O13, in contrast to the bulk Al2O3. The top and side views of the DFT-based
Figure 7.1: (a) LEED pattern of the ultra-thin oxide layer on NiAl(110). The bulk coordinates
of the substrate are indicated [7]. (b) In-plane diffraction data of the oxide layer for one twin
domain. The size of the half circles is proportional to the structure factor. Full half-circles,
experimental structure factors; open half circles, best-fit structure factors. The red hexagon
indicates the reflections that arise from a distorted hexagonal structure with a side length of
3 A˚. The dashed circle denotes the reciprocal space area displayed in the LEED pattern in (a) [7].
(c) Orientation relations between the NiAl lattice and the oxide film lattice. The oxide lattice
can be rotated by ±24.01◦ with respect to the [11¯0] direction, corresponding to the formation
of twin domains [7].
108 Oxidation of NiAl(110) at medium oxygen pressures
Figure 7.2: Experimental x-ray diffraction structure factors (black symbols) as a function of
the reciprocal lattice coordinate, L, for different in-plane momentum transfers. The red curves
represent best-fit structure factors for the proposed structural model. In the inset, the scattering
geometry is pictured [7].
model are shown in Fig. 7.4(a) and (b). The light and dark blue spheres represent the
interfacial (Ali) and surface (Als) aluminium ions, respectively, whereas the red spheres
represent the oxygen ions. For modeling the interface, the parallelogram-shaped commen-
surate super-cell indicated in the Fig. 7.4(a) has been adopted. It has twice the area of
the conventional oxide unit cell (white rectangle) and the lattice constants are a=20.67 A˚
Figure 7.3: Top and side view of the best-fit oxide structure. The diagram shows surface
oxygen ions (light blue), interfacial oxygen ions (dark blue), Al ions in between the two oxygen
ion layers (red), and interfacial Al ions (orange). The oxygen atoms that form rows along the
substrate [001] direction are shown in yellow [7].
7.1 Introduction 109
Figure 7.4: (a) Top and (b) side view of the DFT- and STM-based model for the ultra-
thin aluminum oxide on NiAl(110). (c) Experimental STM image of the oxide film at room
temperature. Figure taken from [8].
and b=21.85 A˚, γ=59.95◦. The topmost oxide layer has the bulk Al2O3 stoichiometry
and is characterized by an oxygen ion arrangement which can be described in terms of
squares and triangles. These are highlighted in Fig. 7.4(a) and (c) (green lines). The
aluminum ions are placed in between the oxygens, only slightly below the oxygen layer,
as can be observed in Fig. 7.4(f). Additional aluminum ions are present in the oxide layer
close to the interface, Ali which are bound strongly to the substrate, thus contributing
with only two electrons to the oxide film. The registry to the substrate is enforced by the
interfacial Ali atoms which are mostly sitting above the Ni rows. This model was found to
be energetically more favorable as compared to the previously described SXRD-model [8].
A comparison between the DFT-based model and our experimental diffraction data will
be shown in section 7.3.3.
110 Oxidation of NiAl(110) at medium oxygen pressures
7.2 Experimental details
Sample preparation
The experiments described in this chapter were performed using a nominally Ni50Al50
single crystal with the diameter 10 mm and the thickness of 2 mm. The specimen has
been cut parallel to the (110) planes and oriented by x-ray diffraction better than 0.1◦.
It has been polished on one side using alumina particles down to 0.05 µm.
Prior the diffraction experiments, the NiAl(110) sample surface was prepared in the
UHV preparation chamber at the MPI laboratory. The substrate has been fixed on a
molybdenum sample holder by spot-welding using thin tantalum foil, as can be seen in
the inset of Fig. 7.7. The cleanliness and the quality of the sample surface were checked
by AES and LEED. Three cycles of Ar+ sputtering (30 minutes at room temperature at
1 keV and pAr+=5×10−6 mbar) followed by 5 minutes annealing at 800◦C were sufficient
to remove the surface contaminants. Fig. 7.5 shows the AES spectrum recorded after the
above mentioned cleaning procedure. The specific Auger electron energies for the most
common surface contaminants were marked. No signal from either sulphur, carbon or
oxygen was detected. In the inset the AES spectrum in the energy region between 38 and
75 eV energy range is shown. The characteristic AES transitions of Ni◦ at 60 eV and Al◦
at 68 eV are clearly resolved.
The LEED pattern observed after the sputtering and annealing at 800◦C showed
a (1×1) rectangular net of sharp reflections, indicating a well-ordered surface. In the
Fig. 7.6(a) the LEED pattern measured at 91 eV is shown. The white arrows indicate
the unit vectors of the substrate reciprocal lattice.
0 100 200 300 400 500 600
-100
-50
0
50
40 50 60 70
-100
-50
0
50
OC
Energy (eV)
dN
/d
E
 
 
   Ni 
102 eV
S
   Al 
68 eV
  Ni 
61 eV
Figure 7.5: Auger electron spectra recorded for the NiAl(110) after the cleaning procedure.
7.2 Experimental details 111
(a) (b)
Figure 7.6: (a) The LEED pattern from the clean NiAl(110) surface. (b) The LEED pattern
of the ultra-thin aluminium oxide on NiAl(110).
An ultra-thin aluminium oxide layer was then grown using the well-established recipe:
oxidation at 270◦C with 5×10−6 mbar O2 for 15 minutes followed by 5 minutes annealing
at 800◦C [7]. In the Fig. 7.6 (b) the LEED pattern of the oxidized NiAl(110) is shown.
The energy of the incident electrons was 62 eV. Heating the oxidized sample in vacuum
at temperatures around 1000◦C leads to the desorption of the ultra-thin oxide film.
After preparation of the well-ordered ultra-thin aluminium oxide layer, the sample was
transferred into the portable UHV diffraction chamber, which was then shipped to the
synchrotron radiation facility.
Surface x-ray diffraction measurements
The SXRD measurements presented in this chapter have been performed at the beam-
line ID32 at the European Synchrotron Radiation Facility, Grenoble. The experimental
set-up used during the beamtime1 is illustrated in Fig. 7.7. In the inset of Fig. 7.7 the
NiAl(110) sample mounted on the inconel sample holder is shown. The oxidation experi-
ments were performed in the portable UHV diffraction chamber described in Chapter 4.
In addition, the chamber was equipped with an Ar+ ion sputtering gun which facilitated
the preparation of a clean substrate surface whenever needed during the measurements.
The temperature was monitored using a Chromel-Alumel thermocouple which was placed
close to the heater. A photon energy of 12.5 keV (λ=0.9918 A˚) was chosen to be far away
from any adsorption edge of Ni or Al, thus reducing the fluorescent background from the
substrate. The measurements were performed in a horizontal scattering geometry (with
1The diffraction data presented in this chapter were obtained during the ESRF1 beamtime.
112 Oxidation of NiAl(110) at medium oxygen pressures
Figure 7.7: Experimental set-up for the SXRD measurements.
the sample surface normal vertical). For the quantitative measurements the incident an-
gle was fixed to 0.4◦, which is higher than the critical angle for total reflection of NiAl.
The horizontal and vertical detector slits were set to 5×5 mm for both the in-plane and
out-of-plane measurements. The structure factor amplitudes have been obtained from
the integrated intensities using the software package ANA. In all cases, the appropriate
correction factors—background subtraction, active area, rod interception, Lorentz and
polarization factors—have been included [54].
Orientation matrix for NiAl(110)
A new unit cell defined in terms of surface orientation has been used to index the diffrac-
tion data from the NiAl(110) substrate. It offers a convenient representation of the re-
ciprocal lattice in which the scattering vector, Q, is expressed in terms of two in-plane
coordinates (H and K) which give the parallel component of the momentum transfer, Q‖,
and the third coordinate, L, defining the perpendicular momentum transfer, Q⊥.
The real space unit cell of NiAl(110) and its reciprocal lattice are shown in Fig. 3.4 in
Chapter 3. The lattice constants are related to the cubic one—ac=2.887 A˚—by: a=c=ac
and b=a0
√
2=4.082 A˚ and α=β=γ= 90◦. The magnitude of the reciprocal lattice vectors
is thus a∗=c∗=1.539 A˚−1 and b∗= 2.176 A˚−1. The correspondence between the real space
directions of the surface and the bulk unit cell of NiAl is: [100]surf ‖ [11¯0]bulk; [010]surf ‖
[001]bulk and [001]surf ‖ [110]bulk. The (2, 0, 0) and (0, 1, 0) substrate reflections have been
7.2 Experimental details 113
used for defining the NiAl orientation matrix. All the substrate diffraction data presented
in this chapter have been indexed according to this unit cell.
Orientation matrix for the ultra-thin aluminum oxide
A different coordinate system has been used to index the reflections arising from the
ultra-thin aluminium oxide layer. The orientation of the reciprocal oxide lattice with
respect to the underlying substrate is shown schematically in Fig. 7.8. Both aluminium
oxide twin domains are indicated (continuous and dotted red lines), together with the
substrate directions (black lines). The real space oxide unit cell lattice constants are
a=18.01 A˚, b=10.59 A˚ and α=β=90◦, γ=91.15◦. The reciprocal lattice is thus character-
ized by a∗=0.3489 A˚−1, b∗=0.5934 A˚−1, c∗=2.176 A˚−1 and α∗=β∗=90◦, γ∗=88.85◦.
Orientation matrix for γ-alumina
A hexagonal unit cell has been chosen to describe the symmetry of the (111) planes of
the γ-Al2O3 cubic unit cell. This unit cell is depicted in Fig. 7.9 and has the c axis
perpendicular to the (111) surface. The oxygen ions are represented by the blue spheres
and the octahedrally and tetrahedrally coordinated aluminium ions are marked by the red
and magenta spheres, respectively. Viewed along the [111] direction, γ-Al2O3 structure
is characterized by an ABCABC close-packed stacking of the oxygen layers with the
cations occupying the octahedrally and tetrahedrally coordinated interstitial sites, as can
be seen in Fig. 7.9. The intrinsic hexagonal lattice constants are related to the cubic one—
a0=7.911 A˚ [69]—by: a=b=a0/
√
2 , a=a0
√
3, with a=b=5.59 A˚, c=13.7 A˚ and α=β=90◦,
γ=120◦.
Figure 7.8: Reciprocal space representation of the relative orientation of the ultra-thin alu-
minium oxide twin domains (red) with respect to the NiAl lattice (black).
114 Oxidation of NiAl(110) at medium oxygen pressures
Figure 7.9: The hexagonal unit cell of γ-Al2O3.
7.3 Results and discussion
7.3.1 Ultra-thin aluminium oxide on NiAl(110)
A well-ordered ultra-thin aluminium oxide film was grown by oxidation of NiAl(110) using
the recipe described in section 7.2. Structural characterization of both the oxide layer and
the interface with the substrate has been performed by means of surface sensitive x-ray
diffraction. The experimental findings are discussed in the following sections.
A. Surface oxide structure
A set of 125 in-plane reflections of the ultra-thin oxide film was measured quantitatively
by rocking scans at L=0.03 and different H and K values ranging from H=0–12 and
K=8¯–8. The experimental structure factors have been obtained by integration using the
software package ANA, including the appropriate correction factors [54].
Fig. 7.10 shows the in-plane map of experimental structure factors (half filled circles)
together with those calculated (open half circles) for the DFT-based model proposed by
Kresse et al. [8]. The experimental in-plane structure factors are consistent with the
data published by Stierle et al. [7], as can be seen by comparison with the in-plane map
in Fig. 7.1(c). The in-plane structure factors are reproduced with less accuracy by the
DFT-based model, as compared to the SXRD-based model.
A convenient tool to test the in-plane projected structure is offered by the Patterson
function, P(r), which is defined as the electron density-density correlation function [52].
7.3 Results and discussion 115
Figure 7.10: In-plane diffraction data for one domain of the ultra-thin aluminium oxide layer.
The size of the half circles is proportional to the structure factor. Full half-circles, experimental
structure factors; open half circles, structure factors calculated for the DFT-model proposed in
Ref. [8].
For in-plane reflections (L ∼ 0) this reduces to
P (x, y) = 2
∑
HK
|FHK |2 · cos[2pi(Hx+Ky)],
where x, y are the atomic coordinates within the unit cell. The strength of the peaks
in a contour map of Patterson function is proportional to the product of the electron
densities of the atoms that produced that peak. In addition, the positions of the peaks in
a Patterson map give information on the sum and the difference between the interatomic
vectors corresponding to the investigated surface structure. Fig. 7.11(a) shows the contour
map of the Patterson function calculated from the experimental structure factors of the
in-plane oxide reflections. The real space unit cell is indicated by the red rectangle. For
comparison, the Patterson map of the DFT-based model is shown in Fig. 7.11(b). A
detailed visual inspection reveals the presence of some additional peaks in the Patterson
map of the calculated model. However, there is an overall satisfactory agreement between
the experimental and theoretical map.
Additional information on the out-of-plane structure of the oxide film was obtained
by measuring the intensity distribution along the oxide surface rods. The integrated
intensities of a total of 92 out-of-plane reflections have been recorded upon rocking the
116 Oxidation of NiAl(110) at medium oxygen pressures
Figure 7.11: (a) Patterson map from experimental data. (b) Patterson map calculated for the
model proposed by Kresse et al. [8].
sample around its normal. In Fig. 7.12 the experimental structure factor amplitudes
(black symbols) of five non-equivalent surface rods are shown. The red curves represent
the corresponding structure factors calculated for the DFT-based model.
Attempts to improve the agreement between the calculated and experimental data
have been made by performing a structural refinement using the software ROD [105]. The
aluminium and oxygen ions in the oxide were allowed to relax in x, y and z directions.
Both in-plane and out-of-plane diffraction data (in total, 214 structure factors) have been
included for the refinement, but no significant improvement of the fit could be achieved.
B. Interface structure
In general, the selective oxidation of an ordered alloy surface requires a rearrangement of
the atoms in substrate surface region, which leads eventually to the formation of point
defects. Two different mechanisms have been proposed in the literature for the formation
of the ultra-thin aluminium oxide layer on NiAl(110). It was suggested by Jaeger et
al. [103] that the 5 A˚ thick aluminium oxide layer is formed by internal oxidation, i.e.,
following dissociation, the oxygen penetrate the first two layers of the NiAl substrate
forming a disordered layer which crystallizes upon subsequent annealing. Simultaneously,
Ni dissolution in the bulk takes place. The second mechanism describes the oxide growth
in terms of external oxidation and suggests the segregation of Al atoms to the surface
as a key process during the oxide growth. Therefore, if this mechanism is valid, one
could eventually expect an enrichment in Al of the first substrate layer(s). It is however
not straightforward to discriminate between the different mechanisms based solely on the
interfacial structure, since other factors may also be involved (e.g., diffusion is expected
to play a role at those temperatures).
Although the issue of the interfacial structure has been already addressed in several
7.3 Results and discussion 117
0.0 0.5 1.0 1.5 2.0
102
(0, 4)
 
 
|F
H
KL
|
L(r.l.u.)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
101
102
(5, 1)
 
 
|F
H
KL
|
L(r.l.u.)
0.0 0.5 1.0 1.5 2.0
101
102
(6, 2)
 
 
|F
H
KL
|
L(r.l.u.)
0.0 0.5 1.0 1.5 2.0
101
102
(6, 2)
 
 
|F
H
KL
|
L(r.l.u.)
0.0 0.5 1.0 1.5 2.0
101
102
(1, 3)
 
 
|F
H
KL
|
L(r.l.u.)
Figure 7.12: Experimental structure factors (black symbols) as a function of the reciprocal
lattice coordinate, L, for different in-plane momentum transfers. For comparison, the structure
factors calculated (red curves) from the model proposed in Ref. [8] are shown.
studies [9, 102], there is still controversy regarding the correct model. In this context,
we have performed CTR measurements in order to probe the ultra-thin aluminium ox-
ide/NiAl(110) interface structure. Three independent substrate truncation rods—one su-
perstructure rod, (0, 1) and two fundamental rods, (2, 0) and (1, 1)—have been measured.
A total of 124 structure factors were obtained by integrating the intensity and apply-
ing the appropriate correction factors. The experimental structure factor amplitudes for
(0, 1), (2, 0) and (1, 1) crystal truncation rods are represented by the black symbols in
118 Oxidation of NiAl(110) at medium oxygen pressures
(a)
0.0 0.5 1.0 1.5 2.0
101
102
 
 
IF
I (
A
rb
. u
ni
ts
)
L (r.l.u.)
(0, 1) L 
(b)
0.0 0.5 1.0 1.5 2.0
101
102
 
 
IF
I (
A
rb
. u
ni
ts
)
L (r.l.u.)
(2, 0) L 
(c)
0.0 0.5 1.0 1.5 2.0
102
103
 
 
IF
I (
A
rb
. u
ni
ts
)
L (r.l.u.)
(1, 1) L 
Figure 7.13: Experimental structure factor amplitudes (black symbols) for (0, 1), (2, 0) and
(1, 1) NiAl crystal truncation rods. The dashed green, the solid red and the dotted blue lines
correspond to the structure factors calculated for the interface model I, II, and the DFT-based
model, respectively.
Fig. 7.13.
In the structural model proposed by Kresse et al. [8], the interface has been modeled
using the parallelogram-shaped super-cell—depicted in Fig. 7.4(a)—placed onto 33 NiAl
unit cells and the atoms in the first five substrate layers were allowed to relax in x, y
and z -directions. The average, minimum and maximum displacement of both Ni and Al
atoms in each layer is listed in the Table 7.1. The displacements are calculated as percents
from the NiAl interlayer spacing. The structure factors for (0, 1), (2, 0) and (1, 1) crystal
truncation rods have been calculated numerically using ROD and are represented by the
dotted blue lines in Fig. 7.13. The poor agreement with the experimental data shows that
the proposed model lacks some structural features and can not accurately describe the
oxide/alloy interface.
Detailed structure refinement of the interface structure has been performed by fitting
all the structure factor amplitudes simultaneously using the software ROD [105]. Two dif-
ferent structural models have been found to describe the observed shape of the truncation
rods and will be described below.
7.3 Results and discussion 119
Table 7.1: Displacements of the atoms in the first five substrate layers according to the DFT-
based model [8].
Layer Al Ni
∆z ∆y ∆x ∆z ∆y ∆x
1st Mean -0.88 -0.05 0.29 1.12 -0.06 0.32
Min. -6.90 -0.79 -1.01 -6.87 -1.39 -1.58
Max. 3.52 0.76 1.38 8.76 1.33 1.99
2nd Mean 0.21 -0.04 0.18 -0.37 -0.04 0.18
Min. -3.36 -0.65 -0.23 -4.22 -0.61 -0.35
Max. 2.60 0.60 0.59 2.30 0.54 0.61
3rd Mean -0.19 -0.01 0.07 -0.42 -0.01 0.05
Min. -1.87 -0.25 -0.11 -3.05 -0.42 -0.16
Max. 0.93 0.27 0.27 1.46 0.30 0.37
4th Mean -0.20 -0.01 0.03 -0.06 -0.00 0.03
Min. -1.56 -0.15 -0.09 -1.33 -0.17 -0.20
Max. 0.52 0.16 0.13 0.80 0.15 0.20
5th Mean -0.02 -0.00 0.02 -0.37 0.00 0.015
Min. -0.52 -0.08 -0.07 -0.99 -0.15 -0.16
Max. 0.30 0.13 0.09 0.01 0.16 0.21
I. Al anti-site model The characteristic structural feature of this model is the presence
of point defects in the first substrate layer. These are namely, Al anti-site atoms (Al
atoms residing on the Ni lattice) and a very low percent of Al vacancies. In addition
to these structural defects, the substrate atoms in the first two layers were allowed
to relax along the z -direction2. Nine free parameters were considered for the fit: one
overall scale factor, three occupancy parameters and five displacement parameters.
Fig. 7.13 shows the experimental structure factors (black symbols) together with
those calculated for the best-fit model (full red curves). The fit parameters are
listed in Tab. 7.2. There are 84% of Ni regular atoms in the first substrate layer, the
rest of Ni sites, namely 16%, being occupied by Al atoms. In turn, the occupancy
probability of Al sites in the same layer is 94%. There is a pronounced outward
2The interface has been modeled using the conventional unit cell of NiAl(110). No contribution from
the oxide layer was considered.
120 Oxidation of NiAl(110) at medium oxygen pressures
Table 7.2: Site occupancies and displacement parameters obtained for the best-fit model as-
suming the presence of Al anti-site atoms.
Layer Atom Occupancy ∆z
1st layer Al 0.94 - 2.78%
Ni 0.84 -3.16%
Al AS 0.16 +68%
2nd layer Al 1.00 +1.4%
Ni 1.00 -2.2%
displacement of the Al anti-site atoms (68%, if expressed in terms of NiAl interlayer
spacing), while both the regular Ni and Al atoms in the first layer are slightly
displaced inwards. This means effectively that the Al anti-site atoms are in fact
closer to the oxide layer than the substrate. The second substrate layer maintains
the bulk stoichiometry and small inward and outwards displacements were found
for the Ni and Al atoms, respectively. It is important to emphasize that this model
is consistent with the one previously proposed by Stierle et al. [9]: 18% of Al anti-
site atoms showing significant outward displacements were reported to be present
in the first substrate layer. In addition, there are 5 % of Ni anti-site atoms in
the second substrate layer. It was argued that the destructive interference close to
the anti-Bragg point (L=1) on the (0, 1) superstructure rod can be considered as a
signature for the presence of interfacial Al anti-site atoms, which are stabilized at the
interface by the presence of the aluminium oxide layer; their pronounced outward
displacement could be understood in terms of a strong interaction of Al anti-site
atoms with the oxide overlayer. However, according to the Ni-Al phase diagram,
these defects are forbidden in the bulk β-NiAl and Ni vacancies are expected to
form on the Al-rich side [see section 5.1 in Chapter 5].
II. Displacement model Recent theoretical studies [8] have suggested that the presence
of Al anti-site atoms at this interface is energetically very costly. A model which
assumes a perfect bulk stoichiometry and only displacements of the atoms in the first
few substrate layers is considered to be energetically more favorable. This would
imply that during the formation of the aluminium oxide, a vertical mass transport
of aluminium and nickel atoms sets in to cancel out the concentration gradient
which appears due to selective oxidation of the substrate. Therefore, an interfacial
model based solely on vertical displacements of the atoms residing in the first three
7.3 Results and discussion 121
Table 7.3: Parameters obtained for the best-fit model assuming only z-displacements.
Layer Atom Occupancy ∆z
1st layer Al 1.00 - 2.38%
Ni 1.00 - 2.46%
2nd layer Al 1.00 -1.5%
Ni 1.00 +1.36%
3rd layer Al 1.00 -0.74%
Ni 1.00 +1.16%
substrate layers was considered. Structural refinement3 was performed by fitting
simultaneously the three truncation rods shown in Fig. 7.13. We have used seven fit
parameters, namely the scale factor and six displacement parameters corresponding
to the Ni and Al atoms in the first three layers. The best-fit structure factors are
represented as dashed green lines in Fig. 7.13. The fit parameters corresponding
to this model are listed in Tab. 7.3. The displacements are expressed in terms of
NiAl(110) interlayer spacing, a/2=2.041 A˚. According to the proposed structural
model, all the atoms in the first substrate layer are displaced inwards, whereas in
the second and the third layers, the Al and Ni atoms are displaced inwards and
outwards, respectively.
It is evident from the Fig. 7.13 that a slightly better fit was obtained for the anti-site
model. However, this is predicted to be energetically less stable than the displacement-
based model [8]. As previously mentioned, the model proposed in Ref. [8] contains at the
interface Al ions in the oxide sitting directly above the Ni atoms in the substrate. An
attempt to reconcile the theoretical predictions with the experimental observations has
been made by considering a starting model in which six Ni atoms in the topmost substrate
located below Al ions have been removed. Structural refinement has been performed
allowing z -displacements of six Al atoms (previously sitting on top of the removed Ni
atoms). Also, vertical relaxations of the Ni and Al atoms in the first two substrate
layers were allowed. However, the resulting model did not reproduced the experimentally
observed surface rods of the oxide.
In this context, relying solely on the existing experimental evidence, the unambiguous
discrimination between these two models is still a critical issue. Anomalous scattering
3The interface was modeled using the parallelogram-shaped super-cell shown in Fig. 7.4(a); the oxide
layer was placed onto 33 NiAl unit cells.
122 Oxidation of NiAl(110) at medium oxygen pressures
measurements at the K adsorption edge of Ni are expected to provide valuable information
which will enable one to clearly distinguish between the two aforementioned models. (The
atoms having an absorption edge at or just above the incident energy absorb strongly the
radiation, which causes a change in the atomic form factors for these atoms. In this
situation, the dispersion corrections need to be considered).
7.3.2 Multiple oxidation
Motivated by the requirements for practical applications, attempts have been made to
increase the thickness of the well-ordered aluminium oxide layer, while preserving its
homogeneity and crystallinity. Various procedures have been suggested in the litera-
ture [106, 107, 108]. Among these, it was reported by Yoshitake et al. [106] that the
thickness of the ultra-thin aluminum oxide film on NiAl(110) can be increased by per-
forming multiple-oxidation cycles. The results were obtained by means of AES and LEED
investigations. The thickness of the oxide layer was estimated based on the O KLL (503
eV) / Ni LMM (848 eV) Auger peak intensity ratio, whose value as a function of oxidation
cycle is plotted in Fig. 7.14. The average oxide thickness after nine oxidation cycles was
reported to be 1.3 nm.
We have applied the above-mentioned procedure and verified the thickening of the
oxide layer by means of crystal truncation rod measurements. As discussed previously,
CTRs offer a very sensitive test for probing surface/interface structures,i.e., any change
in the structure or chemical order at the interface will lead to a change in intensity mea-
sured at the zone boundary (e.g. L=1)on the (0, 1) superstructure rod of the substrate.
Figure 7.14: Auger intensity ratio of O KLL (503 eV) to Ni LMM (848 eV) is plotted as a
function of oxidation cycles [106].
7.3 Results and discussion 123
0 200 400 600 800
500
1000
1500
 
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
Time (s)
 timescan on (0, 1, 0.8)
Figure 7.15: Time-dependent evolution of the intensity on the (0, 1, 0.8) position during second
oxidation cycle.
Therefore, an ultra-thin oxide layer has been prepared on NiAl(110) according to the
well-established recipe: oxidation at 5×10−6 mbar O2 and 270◦C for 15 minutes, followed
by 5 minutes annealing at 800◦C. The intensity at the (0, 1, 0.8) substrate peak was fol-
lowed in situ during the second oxidation cycle. No change in intensity was observed
after 15 minutes of oxygen exposure at 5×10−6 mbar O2 and 270◦C, as can be observed
in Fig. 7.15, where the intensity evolution as a function of time is plotted. An unaltered
interfacial structure points to the fact that no additional mass transport through the in-
terface took place during the second oxidation cycle. Since any further oxidation can not
proceed without mass transport, our results give clear evidence that no thickening of the
oxide layer takes place by repeating the above-mentioned oxidation procedure.
In the view of these findings, we suggest that the increase of the O KLL (503 eV) /
Ni LMM (848 eV) Auger peak intensity ratio reported in Ref. [106] could be attributed to
an increased oxide coverage as a function of oxidation cycles, rather than to oxide thick-
ening. This observation is supported also by Low Energy Electron Microscopy (LEEM)
studies [109], which show that even if a relatively uniform film is obtained after the initial
oxidation cycle, pinholes revealing the clean NiAl surface reappear after the annealing step
when crystallization of the amorphous oxide takes place. The prolonged annealing time
(1–2 hours at 800◦C in vacuum) used in the oxidation experiments reported in Ref. [106]
might cause this effect to be even more dramatic, thus leading to larger areas of oxide-free
regions, whose oxidation will then lead to a change in the O/Ni peak intensity ratio in
the Auger measurements.
124 Oxidation of NiAl(110) at medium oxygen pressures
7.3.3 Transition of the ultra-thin aluminium oxide to bulk Al2O3
The ultra-thin aluminium oxide film is a metastable state in which the system is held by
energy barriers. Thus, for a certain temperature, there will be a particular oxygen partial
pressure for which the mass transport through the oxide layer will be activated and the
oxide growth will proceed further. The question of how and under which conditions this
transition takes place has been addressed in this work by investigating in situ the stability
of the ultra-thin aluminium oxide layer as a function of the oxygen partial pressure at
350◦C.
In the first step, an ultra-thin aluminium oxide layer was prepared, then the sample
was heated to 350◦C and subsequently exposed at progressively higher oxygen pressures,
starting with 5×10−6 mbar O2. The evolution of the oxide layer and of the interfacial
structure was monitored in situ by performing rocking scans on the oxide (0, 4, 0.03) and
the substrate (0, 1, 0.8) reflections, respectively. If no change was observed, the oxygen
was first pumped off and a higher oxygen pressure was applied. The ultra-thin oxide layer
was found to be stable up to a partial oxygen pressure of 0.01 mbar. The sample was
then annealed for two hours at 800◦C in order to crystallize the newly formed oxide. The
evolution of the FWHM and the integrated intensity of both the (0, 4, 0.03) oxide and the
(0, 1, 0.8) substrate reflections as a function of pO2 is shown in Fig. 7.16(a). The filled
and open circles indicate the substrate and oxide reflections, respectively. In Fig. 7.16(b)
and (c) the rocking scans at the above mentioned positions measured at different partial
oxygen pressures are plotted. The same symbols and color code as in Fig. 7.16(a) were
used: the open and filled symbols represent the oxide and the substrate peak, respectively,
and each color indicates a different oxygen pressure which can be read on the x -axis of
the graph in Fig. 7.16(a). The values plotted in Fig. 7.16(a) were obtained by fitting both
the substrate and the oxide peaks shown in Fig. 7.16(b) and (c) by a Gaussian profile.
It can be noticed from Fig. 7.16(a) that a decrease in the integrated intensities of the
oxide and the substrate reflection took place at 10−2 mbar oxygen, while the FWHM of
both reflections remained unchanged. After increasing the oxygen pressure to 0.2 mbar
O2 followed by two hours annealing in vacuum at 800
◦C, a more significant decrease of
the intensity on the oxide peak took place concomitantly with a strong increase in the
integrated intensity of the (0, 1, 0.8) substrate peak. In the inset of Fig. 7.16(b) the rocking
scan on the (0, 1, 0.8) substrate peak after annealing at 800◦C is shown. In addition to the
increased intensity, the profile of the peak changes to a Lorentzian, indicative of structural
changes in the near surface region of the substrate. A detailed structural characterization
of the aluminium oxide/NiAl(110) system has been performed at this stage, with focus
both on the interface and the oxide layer structure.
7.3 Results and discussion 125
(a)
10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100
0.0
0.5
1.0
FW
H
M
 (o
)
p(mbar)
 
 
  I
nt
en
si
ty
 (A
rb
. u
ni
ts
) 0.0
0.5
 
 
 
 
(b)
101.0 101.5 102.0
500
1000
101.0 101.5 102.0 102.5
0
2000
4000
In
te
ns
ity
 (A
rb
. U
ni
ts
)
(deg)
In
te
ns
ity
 (A
rb
. U
ni
ts
)
(deg)
(0, 1, 0.8)
(c)
121 122 123 124 125
0
1000
2000
3000
121 122 123 124 125
0
50
100
 
 
(deg)
In
te
ns
ity
 (A
rb
. u
ni
ts
)
In
te
ns
ity
 (A
rb
. u
ni
ts
)
(deg)
 (0,  4,  0.03)
Figure 7.16: (a) Evolution of the normalized integrated intensity and the FWHM of the
(0, 4, 0.03) oxide peak (open circles) and the (0, 1, 0.8) substrate peak (filled circles) as a func-
tion of the partial oxygen pressure during oxidation at 350◦C. (b) Rocking scans on (0, 1, 0.8)
substrate reflection during oxidation at 350◦C and different oxygen pressures. (c) Rocking scans
on (0, 4, 0.03) oxide reflection during oxidation at 350◦C and different oxygen pressures. [Each
color indicates a different oxygen pressure, which can be read on the x -axis of the graph in (a)].
126 Oxidation of NiAl(110) at medium oxygen pressures
0.0 0.5 1.0 1.5 2.0
102
(0, 4)
 
 
|F
H
KL
|
L(r.l.u.)
0.0 0.5 1.0 1.5 2.0
101
102
(6, 2)
 
 
|F
H
KL
|
L(r.l.u.)
Figure 7.17: Experimental structure factors of the (0, 4) and (6, 2¯) oxide surface rods (green
symbols) after exposure at 0.2 mbar oxygen and annealing at 800◦C. The corresponding structure
factors for the ultra-thin oxide layer (black symbols) are shown for comparison. The red curves
represent the structure factors calculated from the model proposed in Ref. [8].
A. Ultra-thin aluminium oxide
Fig. 7.17 shows the structure factor amplitude along the (0, 4) and (6, 2¯) surface rods
(green symbols) measured after oxidation with 0.2 mbar O2 followed by two hours an-
nealing at 800◦C. The corresponding surface rods (black symbols) of the initial ultra-thin
oxide layer are also shown for comparison, together with the structure factors calculated
for the DFT-based model (red lines).The experimental profiles of the surface rods are
very similar, indicating that both the structure and the thickness of the surface oxide
are essentially unchanged. However, a decrease of the structure factor amplitude is to
be noticed due to a reduced coverage of the substrate with ultra-thin aluminium oxide.
Indeed, in addition to the latter, bulk γ-Al2O3 islands started to form, as will be discussed
further below. Thus, in this experiment we have captured precisely the transition from
the ultra-thin oxide layer to the bulk oxide growth. Based on the unchanged profiles
of the surface rods, it can be concluded that the ultra-thin oxide doesn’t grow thicker.
No intermediate structures were observed and further oxidation proceeds by bulk oxide
formation.
B. Epitaxial γ-Al2O3
Under the aforementioned oxidation conditions, γ-alumina formed epitaxially with the
(111) plane parallel to the NiAl(110) surface. Both in-plane and out-of-plane diffraction
measurements were performed in order to obtain structural information on the newly
formed γ-Al2O3 oxide. To index the bulk oxide reflections the hexagonal unit cell de-
scribed in section 7.2 has been used. The in-plane oxide lattice constant was determined
experimentally to be a=b=5.5 A˚, that is 1.6% smaller than the reported value [69]. The H
7.3 Results and discussion 127
Figure 7.18: Possible in-plane orientation relationships at the fcc(111) / bcc(110) interface.
andK indices describe the in-plane momentum transfer and the L index the perpendicular
one, expressed in units of the reciprocal oxide lattice (r.l.u.).
In-plane structure Before proceeding to present the in-plane structure of the oxide
layer with respect to the substrate, the reader is reminded that the NiAl has the CsCl-
structure on a bcc-type of lattice, while the bulk γ-Al2O3 has a cubic spinel structure with
the oxygen ions forming a fcc sublattice. A brief description of the possible structural
arrangements at this particular type of interface is given below.
The main in-plane orientation relationships predicted for the fcc(111)/bcc(110) inter-
face are: Nishiyama-Wassermann (NW) [110, 111], Kurdjumov-Sachs (KS) [112] and the
R30◦ [113] orientation. The schematic real space representation for each of these orien-
tations is depicted in Fig. 7.18. The blue spheres represent the oxygen ions in γ-Al2O3,
whereas red and green spheres indicate Al and Ni atoms in the substrate, respectively.
Nishiyama-Wassermann For the NW relationship, the [11¯0] direction of the fcc lattice
is parallel to the [001]bcc direction. One can define an angle θ such that θ=0 when
[11¯0]fcc ‖ [001]bcc.
Kurdjumov-Sachs For the KS orientation relationship there are two domains having
θ=±5.26◦ and [11¯0]fcc ‖ [11¯1]bcc.
R30◦ The third possible orientation relationship is R30◦ in which [21¯1¯]fcc ‖ [001]bcc and
θ=30◦.
128 Oxidation of NiAl(110) at medium oxygen pressures
(a)
160 180 200 220 240
102
103
104
105
(4
20
) K
S 
II(4
20
) K
S 
I
(2
20
) K
S 
II
 
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
 (deg.)
(2
20
) K
S 
I
(b)
Figure 7.19: (a) Diffracted intensity as a function of the in-plane rotation angle, θ, with the
scattering angle fixed on the (2, 2, 0) reflection of γ-Al2O3. (b) In-plane view of the reciprocal
space of the NiAl(110)/γ-Al2O3(111) interface having a KS orientation relationship. The recip-
rocal unit cells of NiAl(110) and γ-Al2O3(111) are represented: black open circles indicate the
substrate reflections; the dark and light blue filled circles mark the γ-alumina reflections.
Fig. 7.19(a) shows the diffracted intensity as a function of the azimuthal angle, θ
(rotation of the sample around its surface normal) for the in-plane momentum transfer
Q=4.56 A˚−1, which corresponds to the (2, 2, 0)-type reflections of γ-Al2O3(111). The
observed in-plane reciprocal map of the NiAl(110)/γ-Al2O3(111) interface is schemati-
cally shown in Fig. 7.19(b). The reciprocal unit cells of NiAl(110) and γ-Al2O3(111) are
represented: black open circles indicate the substrate reflections; the dark and light blue
filled circles mark the γ-alumina reflections. The characteristic pattern of the Kurdjumov-
Sachs orientation relationship could be identified: there are two γ-Al2O3 domains which
are rotated by ±5◦ with respect to the (100) direction of the substrate. It is interesting to
note that the reciprocal space orientation of the oxygen sublattice in γ-alumina showing
the aforementioned orientation [blue dotted line hexagons in Fig. 7.19(b)] is similar to
that exhibited by the (12, 0, 0)-type of planes of the ultra-thin aluminium oxide. The
corresponding reflections are indicated by the green full circles in Fig. 7.19(b). They
appear at slightly smaller momentum transfer, which implies a contraction of the corre-
sponding lattice planes upon transformation to bulk γ-Al2O3. In conclusion, the relative
orientation relationship at the γ-Al2O3(111)/NiAl(110) interface was determined to be
the Kurdjumov-Sachs (KS) orientation [112].
Out-of-plane structure The structure of γ-Al2O3(111) layer was further investigated
by performing out-of-plane measurements along the oxide rods, for which an additional
7.3 Results and discussion 129
Figure 7.20: Out-of-plane reciprocal (H0L) plane of γ-Al2O3.
momentum transfer is provided perpendicular to the surface. From bulk structure factor
considerations only reflections at L=4 and 10 are expected for both (0, 2, L) and (2, 0, L)
directions, as can be observed in the left panel of Fig. 7.20. The presence of reflections
at L=8 indicate the formation of twin domains, which occur where an ABC stacking
sequence of the oxygen layers is reversed to CBA, resulting in ABCABCBACBA stacking
sequence. A reversed stacking of the oxygen ion layers from ABC to CBA corresponds to a
180◦ rotation of the reciprocal (H0L) plane of γ-Al2O3 (middle panel in Fig. 7.20. By this
rotation, another reciprocal lattice is obtained: along the (0, 0, L) and (3, 0, L) directions
there is an overlap of the reflections arising from the above mentioned twin domains (grey
circles), while along the (1, 0, L), (2, 0, L), (4, 0, L) directions the two domains contribute
at different positions. The reflections arising from the ABC and CBA twin domains are
indicated by the black and light grey circles, respectively, whereas those for which both
domains are contributing are marked by the dark grey circles. Whenever both the ABC
and CBA stacking of the oxygen ion layers are present, in addition to the (2, 0, 4) and
(2, 0, 10) reflections corresponding to the ABC stacking there are also reflections at L=2
and 8 due to the reversed CBA stacking. This observations make this plane sensitive to
the stacking and twin faults in the subsequent ABCABC layers of oxygen sublattice.
In Fig. 7.21(a) three out-of-plane scans as a function of the γ-Al2O3 reciprocal lattice
coordinate, L, are shown. L=1 corresponds to the reciprocal lattice unit of the oxide
|c∗|=2pi/c=0.4663 A˚−1. The presence of both ABC and CBA oxide domains could be
identified. Due to the KS orientation relationship, there is a superposition of substrate
Bragg reflections along certain out-of-plane reciprocal space direction, as can be seen in
Fig. 7.21(b). Therefore, additional peaks at non-integer L values can be observed on
the oxide rods. In Fig. 7.21(a), the peaks were marked by arrows and labeled with the
corresponding index of the substrate reflections. The FWHM of the (2, 0, 4) and (2, 0, 8)
oxide peaks were found to be ∆L= 0.66 and 0.84 r.l.u., respectively. The average film
130 Oxidation of NiAl(110) at medium oxygen pressures
(a)
0 2 4 6 8
103
104
(2
02
)
(2
22
)
(1
11
)
(2, 0) 
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
L (r.l.u.)
(2
02
)
(2
04
)
(2
08
)
(0
42
) (0
44
)
(0
48
)
(0
22
)
(0
24
)
(0
28
)
103
(2
20
)
(0, 4) 
  
 
103
104
  
 
 
(0, 2) 
(b)
Figure 7.21: (a) L-scans along different reciprocal lattice directions normal to the surface. (b)
Out-of-plane view of the NiAl(110)/γ-Al2O3(111)(KS) reciprocal space.
thickness was thus be estimated to be 20 and 16 A˚, respectively4. No finite film thickness
oscillations were observed indicating a broad distribution of oxide island thicknesses.
C. Interface structure
Two independent in-plane momentum transfers have been probed in order to obtain quan-
titative information on the interfacial structure. The integrated intensity along the (0, 1)
and (2, 0) NiAl crystal truncation rods was measured and the structure factors were
obtained by integration and applying the suitable correction factors using the program
package ANA. In Fig. 7.22 the structure factors along the (0, 1) and (2, 0) NiAl crystal
truncation rods are shown (black symbols).
A simple visual inspection indicates a drastic change of the (0, 1) superstructure rod
profile as compared to the profile measured after the ultra-thin oxide layer formation
(Fig. 7.13, black symbols). However, no significant change was observed for the profile of
the (2, 0) fundamental rod. Recalling the discussion on the different origin of fundamental
and superstructure rods [see Chapter 2 for details], it is clear that the fundamental rods
are not sensitive to the chemical ordering, only to the mean value of the atomic form
4The (2, 0, 4) and (2, 0, 8) oxide peaks are sensitive only to the ABC and CBA stacking, respectively.
7.3 Results and discussion 131
(a)
0.0 0.5 1.0 1.5 2.0
102
103
 
 
IF
I (
A
rb
. u
ni
ts
)
L (r.l.u.)
(0 1) L
(b)
0.0 0.5 1.0 1.5 2.0
102
103
(2 0) L
L (r.l.u.)
IF
I (
A
rb
. u
ni
ts
)
 
 
Figure 7.22: Experimental structure factors (black symbols) of the (a) (0, 1) and (b) (2, 0)
NiAl CTRs after oxidation at 350◦C and annealing at 800◦C. The structure factors calculated
for the best-fit interface model are indicated by the full red curves [see text for details].
factor per layer, f¯ . In the particular case of NiAl, one can express this quantity as:
f¯Al = θ
Al
AlfAl + θ
Al
NifNi
f¯Ni = θ
Ni
NifNi + θ
Ni
AlfNi,
where θAlAl and θ
Ni
Ni represent the occupancies of regular Al and Ni sites, respectively, and
θNiAl and θ
Al
Ni indicate the occupancies of anti-site Al and Ni atoms, respectively. Thus,
the shape of the (2, 0) fundamental rod would remain essentially unchanged as long as
the quantity f¯Ni + f¯Al is similar to the fNi + fAl. On the other hand, an increase of the
intensity measured at the anti-Bragg point of a superstructure rod is to be observed when
f¯Ni - f¯Al 6= fNi - fAl. In the following it will be shown that the observed modulation on
the (0, 1) superstructure rod of oxidized NiAl is related to the presence of point defects
in the topmost substrate layers.
A simultaneous fit of the (0, 1) and (2, 0) rods was performed using the program
package ROD [105]. Several free parameters were included, namely, an overall scaling
factor, displacement parameters corresponding to relaxations in z -direction of the atoms
in the first three substrate layers and variable site occupancies for the sites in the first two
substrate layers. The structure factors given by the best-fit model are indicated by the
full red lines in Fig. 7.22. The model considers the presence of Ni anti-site atoms which
occur with 69% and 58% probability in the first two substrate layers. There are also Ni
and the Al vacancies in both substrate layers. The occupancy probability for different
sites and the average atomic form factor per layer are listed in the Tab. 7.4. In addition,
z -displacements of the atoms in the first three substrate layers were considered for the
structural refinement.
132 Oxidation of NiAl(110) at medium oxygen pressures
θNiNi θ
Al
Al θ
Al
Ni θ
Ni
Al f¯Ni + f¯Al f¯Ni - f¯Al
1st layer 0.889 0.105 0.695 0 45.715 4.085
2nd layer 0.7587 0.218 0.576 0 40.21 2.36
Table 7.4: The occupancy of different sites for the best-fit structural model of the aluminium
oxide/NiAl(110) interface [see text for details].
The best-fit structural model for the aluminium oxide/NiAl(110) interface is shown
schematically in Fig. 7.23. The displacements for each type of atom are indicated in the
figure. They are calculated as percents from the interlayer spacing, 2.0414 A˚. The outward
relaxation of the Al atoms in the topmost substrate layer amounts 11%, while the Ni and
the Ni anti-site atoms show only 3% and 4% displacements, respectively.
To separate the effect of the occupancies and displacements on the rods profile, the
structure factor has been calculated numerically by considering in turn only the displace-
ments and the occupancy parameters listed above. The rods profiles are indicated in
Fig. 7.22 by the green dotted curve (only z -relaxations) and the blue dotted curve (only
occupancies). It can be noticed that the observed modulation is reproduced qualitatively
even if one does not include any z -displacement of the atoms. In addition, it should be
Figure 7.23: Structural model of the aluminium oxide/NiAl(110) interface [oxide layer not
shown].
7.3 Results and discussion 133
mentioned that no reasonable fit could be obtained by including only z -displacements
as free fit parameters. This increases our confidence that the model gives a realistic
description of the interface structure.
The Ni anti-site atoms are reported as the common structural defect which appear
on the Ni-rich side of NiAl phase diagram [see Chapter 4]. The Ni enrichment at the
interface appears as a consequence of Al consumption by selective oxidation. A recent
ab initio thermodynamics study [10] on the structural stability of the Al2O3/ NiAl(110)
interface predicts the formation of two types of stable structures depending on the alloy
composition. For aluminium oxide being in contact with a Ni-rich alloy, the formation of
Ni anti-site atoms at the interface has been predicted. Qualitatively, this is in agreement
with our experimental observations. However, Feng et al. [10] have indicated the formation
of Ni anti-site atoms on only one substrate layer, whereas our model shows Ni enrichment
in the first two NiAl layers. The structure factor has been calculated numerically by
assuming bulk occupancy for the second substrate layer. The result is shown by the dot-
dashed magenta line in Fig. 7.22 and it clearly indicates that the observed modulation
on the (0, 1) superstructure rod can not be reproduced by a model assuming only one
Ni-enriched substrate layer.
7.3.4 Oxidation at 800◦C
The increase of both the oxidation temperature and the partial oxygen pressure induces
considerable changes in the structure of the oxide layer. Exposing the clean NiAl(110)
surface at 10 mbar oxygen at 800◦C for 90 minutes leads to formation of an oxide layer
which consists solely of bulk oxides; the coexistence of epitaxial γ-Al2O3 oxide and poly-
crystalline δ-Al2O3 was observed.
A. Epitaxial γ-Al2O3
In-plane structure Fig. 7.24(a) shows the diffracted intensity as a function of the az-
imuthal angle, θ for the in-plane momentum transfer Q=4.53 A˚−1 measured after the
oxidation at 800◦C (red symbols). The same scan as in the Fig. 7.19(a) is shown for
comparison (blue symbols). There are now three peaks corresponding to (2, 2, 0)- type
reflections of γ-Al2O3 which appear at different azimuthal angles compared with the previ-
ous scan. This can be understood in terms of a coexistence of oxide domains showing two
different relative orientations with respect to the substrate: the Nishiyama-Wassermann
(NW) [111, 110] and the R30◦ orientation [113]. The in-plane view of the reciprocal space
of the NiAl(110)/γ-Al2O3(111) interface having the above mentioned orientation relation-
ships is shown in Fig. 7.24(c). The represented reciprocal unit cells are NiAl(110) (black
134 Oxidation of NiAl(110) at medium oxygen pressures
(a)
160 180 200 220 240
101
102
103
104
105
106
107
108
109
(4
20
) N
W
(2
20
) R
30
o
(2
20
) N
W
(4
20
) K
S
 I
I
(4
20
) K
S
 I
(2
20
) K
S
 I
I  
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
 (deg.)
(2
20
) K
S
 I
(b)
220 225 230 235 240
0
100
200
KS
 I
I
 (deg.)
 
 
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
KS
 I
(4
20
) N
W
(c)
Figure 7.24: (a) Diffracted intensity as a function of the in-plane rotation angle, θ, with
the scattering angle fixed on the (2, 2, 0) reflection of γ-Al2O3. (b) Rocking scan on the (42¯0)
reflection of γ-Al2O3 (θ = 230◦) after oxidation at 800◦C. (c) In-plane view of the reciprocal space
of the NiAl(110)/γ-Al2O3(111) interface having the NW (blue filled circles) and the R30◦(red
filled circles) orientation relationship.
open circles) and γ-Al2O3(111) having the NW (blue filled circles) and the R30
◦(red filled
circles) orientation relationship. According to this map, the peaks which appear at θ =
170◦ and 230◦ are due to the oxide domains showing a NW orientation, while the peak at
θ = 200◦ is attributed to R30◦ orientated domains. A rather large FWHM characterizes
the oxide peaks in the rocking scan shown in Fig. 7.24(a), indicative of a large degree
of in-plane misorientation. Fig. 7.24(b) shows the diffracted intensity as a function of
the azimuthal angle, θ, with the scattering angle fixed on the (42¯0) reflection of γ-Al2O3
measured after the oxidation at 800◦C. It is basically the oxide peak corresponding to θ
∼ 230◦ shown in the Fig. 7.24(a) (red line). Three reflections were identified by fitting
this peak with a Gaussian lineshape [Fig. 7.24(b)]: the central peak arises from the oxide
domains having NW orientation, while the two side peaks correspond to the KS domains.
This is indicative of an incomplete transition from KS to NW orientation, whose origin is
7.3 Results and discussion 135
explained further below. Based on the rigid lattice model, the KS orientation is expected
at the γ-Al2O3 (111)/NiAl(110) interfaces as it minimizes the interfacial energy [113].
However, as the film grow thicker the KS domains tend to coalesce in order to decrease
the dislocation and defect energy at the domain boundaries. This implies a rotation of
the two KS domains by ±5.26◦ towards each other, a process that was not complete
under the investigated oxidation conditions, i. e., 800◦C and 10 mbar O2. The above-
mentioned transition was previously observed for the Pdfcc(111)/Crbcc(110) system and
the coalescence of the KS domains was investigated as a function of Pd thickness and
growth temperature [114].
Out-of-plane structure The out-of-plane structure of the epitaxial oxide was checked
by performing L-scans in several reciprocal space directions for both the NW and the R30◦
oxide domains. These are shown in Fig. 7.25 and correspond to the γ-Al2O3 structure [see
Fig. 7.20for comparison]. The presence of (2, 0, 4) and (0, 2, 4) reflections is a signature
for twin domains formation. Additional peaks at integer L values appear along (2, 0, L)
and (0, 2, L) reciprocal space directions for both NW and R30◦ domains. They represent
contribution from the neighboring oxide rods and are labeled accordingly in the Fig. 7.25.
B. Polycrystalline δ-Al2O3
A radial scan performed in a non-high symmetry direction is presented in Fig. 7.26. It re-
veals the presence of a polycrystalline component which was identified to be δ-Al2O3.
At oxidation temperatures as high as 800◦C and prolonged oxidation times, various
metastable alumina phases could form on the NiAl surface. Most often, γ-Al2O3 [101, 115]
or θ-Al2O3 [11] were reported to form on NiAl(110) surface. However, the reflections ob-
served in Fig. 7.26 correspond to those of the tetragonal δ-Al2O3 structure reported by
Lippens et al. [84].
Careful examination of the diffraction data is required when it comes to discriminate
between various transition alumina reflections, since there is a considerable number of
common d -spacings. This is illustrated in Fig. 7.26(b) where different d -spacings and
momentum transfers characteristic for γ-, θ- and δ-Al2O3 reflections are shown. It can
be observed that there are a number of δ-Al2O3 reflections which might help for an
unambiguous interpretation of the experimental data.
136 Oxidation of NiAl(110) at medium oxygen pressures
(a)
0 2 4 6 8
103
104
105
106
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
L (r.l.u.)
(2
02
) (2
04
)
(2
08
)
(0
42
)
(0
44
)
(0
48
)
(0
22
) (0
24
)
(0
28
)
102
103
104
105
(1
23
)R
30
(1
1.
14
)
(1
1.
14
)
(1
13
)R
30
(2
22
)
(2
22
)
  
 
104
105
106
(4
0.
12
)
 (2, 0) NW
 (0, 4) NW
  
 
 
 (0, 2) NW
(b)
0 2 4 6 8
103
104
105
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
L (r.l.u.)
(2
02
) (2
04
)
(2
08
)
(0
42
) (0
44
)
(0
48
)
(0
22
) (0
24
)
(0
28
)
102
103
  
 
103
104
105
(2
13
)N
W
(1
13
)N
W
(2
16
)N
W
(1
16
)N
W
(2, 0) R30o
(0, 4) R30o
  
 
 
(0, 2) R30o
(2
22
)
(4
0.
12
)
(2
22
)
(1
1.
14
)
(1
1.
14
)
Figure 7.25: L-scans in different reciprocal lattice directions normal to the surface. The
reflections were indexed using the coordinates of the hexagonal γ-Al2O3 unit cell. Both the
(a) NW and the (b) R30◦ domains were probed.
C. Interface structure
Fig. 7.27 shows the structure factor amplitude for the (0, 1) and the (2, 0) substrate
truncation rods. While no significant changes in the (2, 0) rod profile, a peak was observed
on the (0, 1) superstructure rod. A closer inspection of Fig. 7.24(c) reveals the fact that the
(1¯, 2, L) reciprocal space direction of the γ-Al2O3 having the NW orientation relationship
overlaps with the (0, 1) superstructure rod of the substrate. Therefore, the peaks at L ∼
0.8 and 1.6 are identified as a contribution from the (1¯, 2, 3) and (1¯, 2, 6) reflections of
γ-alumina domains showing the NW orientation, which in turn hampers an atomic level
investigation of the interface structure.
D. Ex situ TEM characterization
After the diffraction experiment, cross-sectional TEM specimens were prepared [116] and
examined5 using an in-house JEM-ARM 1250 microscope operated at an accelerating
voltage of 1250 kV.
Fig. 8.10(a) shows a conventional transmission electron microscopy image of the alu-
5The TEM measurements were performed by Yun Jin-Phillipp, Max-Planck-Institut fu¨r Metall-
forschung, Stuttgart.
7.3 Results and discussion 137
(a)
1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
102
103
 
(3
39
/1
1.
15
)
(4
40
)
(4
0.
12
)
(4
26
/3
1.
11
)
Q(Å-1)
(5
23
/5
16
/2
0.
15
)
(3
33
)
(3
18
)(2
26
)(3
14
/3
05
)
(3
12
)
(2
22
)
(1
16
)
(1
14
/1
05
)
 
 
In
te
ns
ity
 (A
rb
.u
ni
ts
)
(b)
1.5 2.0 2.5 3.0 3.5 4.0 4.5
1
2
3
4
 
 
 -alumina
 -alumina
 -alumina
d-
sp
ac
in
g 
(Å
)
Q (Å-1)
Figure 7.26: (a) Radial in-plane scan in a non-high symmetry direction after oxidation at
800◦C. (b) Comparison between different d -spacings characteristic for γ-, θ- and δ-Al2O3.
(a)
0.0 0.5 1.0 1.5 2.0
102
103
(0 1) L
(1
26
) N
W
L (r.l.u.)
In
te
ns
ity
 (A
rb
. u
ni
ts
)
 
 
(1
23
) N
W
(b)
0.0 0.5 1.0 1.5 2.0
103
104
(2
04
) N
W
(2 0) L
L (r.l.u.)
In
te
ns
ity
 (A
rb
. u
ni
ts
)
 
 
(2
02
) N
W
Figure 7.27: (a) The structure factor amplitude on the (0, 1) superstructure rod of NiAl. (b)
The structure factor amplitude on the (2, 0) fundamental rod of NiAl.
minium oxide/NiAl(110) interface. The overall thickness of the oxide layer is estimated
to be ∼700 A˚. In Fig. 8.10(b) a 150 nm wide cavity can be identified at the oxide/alloy
interface. The formation of cavities is related to the condensation of vacancies created at
the interface as a result of selective oxidation, as perviously explain for the CoGa(100)
oxidation [see Chapter 6]. It was reported in the literature [117] that when both transient
aluminas and α-Al2O3 are present on the surface, the cavities form only underneath the
transient phases. This is due to the fact that transient oxides grow mainly by outward
Al diffusion through the lattice, which results in an inward diffusion of vacancies to the
interface where they reach critical supersaturation level, agglomerate and form voids. On
the other hand, α-Al2O3 is supposed to grow mainly by inward oxygen diffusion. There-
fore, in the regions where the nucleation of the α-Al2O3 took place in the early oxidation
stages, the amount of injected vacancies is reduced.
The epitaxial γ-Al2O3 observed in our study does not form a continuous layer on top of
138 Oxidation of NiAl(110) at medium oxygen pressures
Figure 7.28: Cross sectional TEM micrographs showing the Al2O3/NiAl(110) interface.
7.4 Summary 139
the substrate. Relatively small oriented domains of γ-Al2O3 are found to be embedded in
a polycrystalline δ-Al2O3 matrix, as can be observed in Fig. 8.10(c). The formation of the
polycrystalline oxide is favored by the prolonged oxidation time employed for this experi-
ment (90 minutes). Much shorter exposure times under similar conditions of temperature
and pressure (10 mbar O2 at 800
◦C) are expected to lead to an oxide layer consisting
solely of γ-Al2O3. This idea is supported by the fact that 10 minutes exposure at 870
◦C
and 1 bar O2 leads solely to the growth of an epitaxial γ-Al2O3, as will be discussed in
the Chapter 8.
7.4 Summary
This chapter presented a study of NiAl(110) oxidation at various temperatures and oxygen
pressures ranging from 5 ·10−6 mbar up to 10 mbar O2. Surface x-ray diffraction measure-
ments performed at the synchrotron radiation source ESRF in Grenoble were employed
for structural characterization and in situ evolution of the oxide/alloy system during ox-
idation. Additional information were obtained by means of ex situ TEM investigations.
The main results are summarized in the following:
Ultra-thin aluminium oxide/NiAl(110) interface model
A set of in-plane reflections and several surface rods corresponding to the ultra-thin
aluminium oxide layer have been measured. The DFT-based model proposed by Kresse et
al. [8] was found to reproduce reasonably good the surface oxide rods, however structural
features were found to be missing in this model for it does not reproduce the observed
shape of the crystal truncation rods.
Structural refinement of the interface has been performed and two different models
were found to reproduce the experimental data. A better fit was obtained for the Al
anti-site model, but this is predicted to be energetically less favorable than the model
based solely on relaxation of the atoms in the first three substrate layers. The unam-
biguous discrimination between these two models is still a critical issue and anomalous
scattering measurements at the K adsorption edge of Ni are expected to enable one to
clearly distinguish between the two aforementioned models.
Transition of the ultra-thin aluminium oxide to bulk Al2O3
We demonstrated that the thickness of the ultrathin aluminum oxide can not be further
increased by multiple oxidation cycles and it only exists as a 5 A˚ thick layer. At 350◦C, the
surface oxide was found to be stable up to a 0.01 mbar O2. The sample was subsequently
annealed at 800◦C in order for the newly formed oxide to crystallize. Further oxidation
proceeds by the formation of epitaxial bulk γ-Al2O3 islands having Kurdjumov-Sachs
140 Oxidation of NiAl(110) at medium oxygen pressures
orientation that coexist with the initial oxide layer.
Oxidation at 800◦C
γ-Al2O3 oxide islands having both the Nishiyama-Wassermann and R30
◦ in-plane ori-
entation with respect to the substrate formed by oxidation of the NiAl(110) substrate
at 800◦C and 10 mbar O2. In addition, the presence of polycrystalline δ-Al2O3 was
identified. The transition from the KS to NW orientation was previously observed for
the Pd(111)/Cr(110) system [114]. As it will be shown in the following chapter, at
slightly higher growth temperatures (870◦C) and shorter oxidation times (10 minutes),
only epitaxial γ-alumina showing a R30◦ orientation relationship was formed. This is
an indication that as the oxide grows thicker, the R30◦ orientation becomes more favor-
able. This experimental observation is supported by a recent ab initio thermodynamics
study [10], which showed that the R30◦ orientation relationship is the most stable one at
the Al2O3/NiAl(110) interface.
Chapter 8
Oxidation of NiAl(110) at
atmospheric pressure
8.1 Introduction
NiAl is an ordered intermetallic material which possesses low density, good oxidation
resistance, and metal-like electrical and thermal conductivity. As previously mentioned,
owing to their high temperature oxidation resistance, the Ni-Al intermetallics may be
used as bond coats in thermal barrier coating (TBC) systems. Therefore, the oxidation
of NiAl at high temperature was intensively studied [11, 118, 119, 101, 120, 115, 121, 122,
123, 124] using different techniques. However, to the best of the our knowledge, no in
situ atmospheric pressure oxidation studies exist on NiAl single crystals, which allow to
follow the transition from metastable oxide phases to α-Al2O3, offering at the same time
the possibility to perform a structural characterization of the oxide and oxide/substrate
interface.
We have studied in situ the transition from epitaxial γ-Al2O3 to α-alumina during
atmospheric oxidation of a NiAl(110) single crystal. The method of choice is Grazing In-
cidence X-ray Diffraction (GIXRD) which has proved to be a very useful non-destructive
technique that can be used to investigate oxidation processes in situ. Working at grazing
angles ensures the required surface sensitivity, since it is the oxide and the oxide/substrate
interface we are interested in. In contrast to other surface science techniques, GIXRD
does not have limitations concerning the insulating character of the material under in-
vestigation and does not require ultra-high vacuum environment. Additional structural
information were obtained from cross-section Transmission Electron Microscopy (TEM)
images. The cross-sectional TEM specimens were prepared [116] after the in situ diffrac-
tion experiments. The TEM measurements were performed1 using an in-house JEM-ARM
1The TEM measurements have been performed by Amalia Catanoiu-Soare and Gunther Richter.
142 Oxidation of NiAl(110) at atmospheric pressure
1250 microscope operated at an accelerating voltage of 1250 kV.
This chapter is organized as follows: the experimental set-up for the GIXRD measure-
ments is presented in Section 8.2, followed in Section 8.3.1 by the ex situ x-ray diffraction
characterization of the preoxidized γ-Al2O3/NiAl(110) system. Section 8.3.2 comprises
the in situ experimental observation on the transformation of epitaxial γ-Al2O3 layer to
the stable α-Al2O3 during high temperature atmospheric pressure oxidation. The ex situ
x-ray diffraction and TEM observations on the α-Al2O3/NiAl(110) system are presented
in Section 8.4. The chapter concludes with a summary of the experimental findings.
8.2 Experimental details
The experiments were performed using a nominally Ni50Al50 single crystal grown by the
floating-zone technique [125]. The sample was cut and polished in (110) orientation better
than 0.1◦. The surface was cleaned by sputter-annealing cycles in UHV until the surface
contaminants were removed. Prior to the in situ experiments, the sample was oxidized
at 870◦C and at a pure oxygen pressure of 1 bar for 10 minutes. The oxide layer was
first characterized by performing ex situ GIXRD measurements. The first set of data
was taken using a six circle diffractometer installed at a laboratory rotating anode x-ray
source. The measurements were done in a horizontal sample geometry (with its surface
normal vertical) using Mo Kα radiation with a wavelength of 0.70926 A˚. A graphite
monochromator was used allowing a separation of the Kα and Kβ lines. The angular
acceptance of the detector was 0.4◦ normal to the surface and 0.4◦ parallel to the surface.
Structural information of the oxide layer were obtained from the out-of-plane diffraction
data. In addition, some in-plane data were taken in order to obtain additional information
regarding the orientation relationship between the oxide and the underlying substrate.
For the examination of the oxide phase transformations during further oxidation, in
situ in-plane measurements2 were carried out using synchrotron radiation at the MPI-MF
surface diffraction beamline at ANKA, Karlsruhe [61]. The wavelength was chosen to
be 1.1808 A˚ (10.5 keV). The sample was mounted in vertical scattering geometry inside
the Anton Paar HTK 1200N high temperature furnace. The temperature was measured
using a Pt 10% RhPt thermocouple. The experimental set-up used during the beamtime
is illustrated in Fig. 8.1. The oxidation experiments were performed in air at temperatures
ranging from 800 to 1200◦C. Due to the limited width of the furnace’s window it was not
possible to perform in situ out-of-plane measurements. Therefore, after the oxidation at
1200◦C, the out-of-plane structure of the oxide layer was investigated performing ex situ
2The diffraction data presented in this chapter were obtained during the ANKA3 beamtime.
8.3 Results and discussions 143
Figure 8.1: Experimental set-up for the GIXRD measurements at ANKA.
measurements using the six circle diffractometer described above.
8.3 Results and discussions
8.3.1 Characterization of the preoxidized NiAl(110)
Prior to the in situ GIXRD measurements, the NiAl(110) crystal has been oxidized for
10 minutes at 870◦C and 1 bar oxygen and a layer of epitaxial γ-Al2O3 oxide. Ex situ
x-ray diffraction investigations were performed to probe the crystallographic structure of
the oxide using the laboratory set-up described above. The in-plane and the out-of-plane
structure of the oxide layer will be described in the following.
A. In-plane structure of γ-Al2O3
In order to establish the orientation of the γ-Al2O3 layer with respect to the substrate,
in-plane rocking scans over a wide angular range (in which the value of the in-plane
momentum transfer, Q, is kept constant) and in-plane radial scans are necessary. The
hexagonal unit cell—shown in Fig. 7.9—was chosen to describe the symmetry of the
(111) planes of the γ-Al2O3 bulk unit cell; the intrinsic hexagonal lattice constants are
a=b=5.531 A˚, c=13.55 A˚ and α=β=90◦, γ=120◦. It should be noted that all the γ-Al2O3
reflections were indexed using the hexagonal unit cell, whereas the substrate reflections
are labeled according to the bulk unit cell.
144 Oxidation of NiAl(110) at atmospheric pressure
(a)
-60 -30 0 30 60 90 120
100
101
102
103
104
105
(2
40
)
(4
20
)
(2
20
)
(0
02
)
(0
02
)
In
te
ns
ity
 (c
ou
nt
s/
20
se
c.
)
 
  (deg) (b)
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
100
101
102
103
104
105
106
107
(220)
 
 
In
te
ns
ity
 (A
rb
. u
ni
ts
) 
Q (Å-1)
(110)
(110)
 
 
(c)
Figure 8.2: (a) Diffracted intensity as a function of the in-plane rotation angle, θ, with the
scattering angle fixed on the (0, 0, 2) reflection of NiAl. (b) Radial in-plane scan through NiAl
(1, 1¯, 0) reflection, (1, 1, 0) and (2, 2, 0) reflections of γ-Al2O3. (c) In-plane view of the reciprocal
in-plane unit cell of the substrate (open black symbols) and the relaxed γ-Al2O3(111) (filled red
symbols) having the R30◦ orientation relationship.
Fig. 8.2(a) shows the diffracted intensity as a function of the azimuthal sample angle,
θ (rotation of the sample around the surface normal). The scan was performed with the
incident angle equal to the critical angle for total external reflection of the oxide layer, i. e.,
αi=0.26
◦. The in-plane momentum transfer was kept at Q=4.3764 A˚−1, corresponding to
(0, 0, 2) reflection of NiAl. In between the (0, 0, 2) and (0, 0, 2¯) reflections of NiAl, three
additional peaks can be observed. These are compatible with the six-fold symmetry of
the (111) planes in γ-Al2O3 and were identified to be the γ (2, 2, 0)-type reflections (or
the (4, 4¯, 0) reflection using the bulk γ-Al2O3 coordinates).
A radial scan through the oxide peaks [not shown here] revealed that their position is
slightly shifted to a higher Q-value, i. e., 4.544 A˚−1, as compared to that of the (0, 0, 2)
reflection of NiAl, indicating that the oxide lattice is relaxed with respect to the underlying
substrate. However, the oxide peaks were found to be rather broad in the radial direction
8.3 Results and discussions 145
(a)
0 2 4 6 8 10 12
101
102
103
I
II
III
 
02l
In
te
ns
ity
(c
ou
nt
s/
20
se
c.
)
L (r.l.u)
101
102
103
II
II
I
I20l
  
 
101
102
103
I+II
I+II
I+II
22l
  
 
 
(b)
Figure 8.3: (a) L-scans in different reciprocal lattice directions normal to the surface. (b) Re-
ciprocal (H, 0, L) plane of γ-Al2O3.
which explains why there is a contribution even in the rocking scan performed with
Q=4.3764 A˚−1. Along the [11¯0] direction of NiAl, in addition to the substrate (1, 1¯, 0)
peak, the γ (2, 2, 0) and γ (1, 1, 0) reflections of the γ-Al2O3 were found. This is shown
in Fig. 8.2(b). The scan was performed at an incident angle equal with the critical angle
of the oxide in order to enhance the signal from the γ-Al2O3 layer. The presence of the γ
(1, 1, 0) reflection is an additional proof that the oxide layer consists of γ-Al2O3, since for
α-Al2O3 this reflection does not appear. Fig. 8.2(c) shows a sketch of the reciprocal in-
plane unit cell of the substrate (open black symbols) and the relaxed γ-Al2O3(111) (filled
red symbols) having the R30◦ orientation relationship. The aforementioned experimental
observations give direct evidence for the R30◦ orientation relationship3 which can be
summarized as follows: NiAl(110)bcc ‖ γ-Al2O3 (111)fcc and [21¯1¯]fcc γ-Al2O3 ‖ [001]bcc
NiAl.
B. Out-of-plane structure of γ-Al2O3
The measurements described in the previous section probed only the in-plane structure
of the oxide layer with respect to the substrate. In order to complete the structural
characterization, additional information are needed. These are obtained performing out-
of-plane measurements along the reciprocal lattice directions, providing an additional
momentum transfer perpendicular to the surface,Q⊥ [52].
Fig. 8.3(a) shows three out-of-plane scans as a function of the γ-Al2O3 reciprocal lattice
coordinate, L, where L=1 corresponds to the reciprocal lattice unit of the oxide |c∗| =
2pi/c=0.4638 A˚−1. The reflections were indexed using the coordinates of the hexagonal
3A description of the possible orientation relationships at fcc(111)/bcc(110) interface was offered in
Chapter 7.
146 Oxidation of NiAl(110) at atmospheric pressure
0 5 10 15 20 25 30
700
800
900
1000
1100
1200
 
 
Te
m
pe
ra
tu
re
(o
C
)
Time (h)
RT
Figure 8.4: Temperature program during the in situ GIXRD measurements.
γ-Al2O3 unit cell. The presence of (2, 0, 4) and (0, 2, 4) reflections indicate the formation
of twin domains, which occur whenever the ABC sequence is reversed to CBA, resulting
in ABCABCBACBA stacking [see Section 7.3.3 for details]. The reflections arising from
the two 60◦ twin domains I and II are indicated by black and white circles, respectively,
whereas those for which both domains are contributing (I + II) are marked by the grey
circles. The missing reflections could be an indication for a poor cation ordering. Along
the (2, 0, L) direction, in addition to the (2, 0, 2) and (2, 0, 8) reflections corresponding to
the ABC stacking [open circles in Fig. 8.3(b)], one can observe also reflections at L=2
and 10, due to the CBA stacking of the oxygen layers [black circles in Fig. 8.3(b)].
8.3.2 In situ oxidation at atmospheric pressures
After the ex situ characterization, the sample was introduced into the Anton Paar furnace
described previously and In situ measurements were performed at the MPI-MF beamline
at ANKA, Karlsruhe using a photon energy of 10.5 keV to study the The phase transfor-
mation and oxidation behavior was investigated during oxidation in air at temperatures
ranging from 800–1200◦C. The sample was first heated at 800◦C with a heating rate of
20◦C per minute. As can be seen in Fig. 8.4, the temperature was then raised in five steps
to 1025◦C. During each step, rocking scans were performed on the γ(2, 2, 0) reflection to
study the evolution of the mosaic distribution in the oxide layer. A continuous increase in
the FWHM (Full Width of Half Maximum) of the γ(2, 2, 0) peak was observed, which is
a precursor for the shearing of the oxygen ion planes during the fcc-hcp–like martensitic
phase transformation from γ-Al2O3 to α-Al2O3. In addition, a radial scan in the (11¯0)
direction of the substrate is shown in Fig. 8.5(a) for T=1025◦C, which indicates the for-
mation of a new metastable phase, δ-Al2O3. One can observe the apparition of a new
peak close to the γ(2, 2, 0) reflection: in addition to the NiAl (1, 1¯, 0) and the γ(2, 2, 0)
8.3 Results and discussions 147
(a)
3.0 3.5 4.0 4.5
101
102
103
(220)
Q (Å-1)
In
te
ns
ity
 (A
rb
. u
ni
ts
) 
(110)
(516)
 
 
(b)
4.40 4.45 4.50 4.55
0
1000
2000
3000
4000
Q (Å-1)
3030)
(220)  
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
(40.12)
(c)
Figure 8.5: (a) Radial in-plane scan in the (1, 1¯, 0) direction of the substrate during oxidation
at 1025◦C. (b) Radial scan through the (2, 2, 0) and (3, 0, 3¯, 0) reflection of γ-Al2O3 and α-Al2O3,
respectively. (c) Diffracted intensity as a function of the in-plane rotation angle, θ, with the
scattering angle fixed on the (3, 0, 3¯, 0) reflection of α-Al2O3 during oxidation at 1025◦C (after
the sample was cooled down to room temperature and heated up again).
reflections, the (5, 1, 6) reflection of δ-Al2O3 is present.
After eight hours at 1025◦C, the sample was cooled down to room temperature and
heated up again at 1025◦C. After another hour at 1025◦C, a high resolution radial scan
through the γ(2, 2, 0) reflection revealed the presence of the δ(4, 0, 12) and α(3, 0, 3¯, 0)
peaks, as seen in Fig. 8.5(b). The fitting was done using a Gaussian lineshape for each
of the peaks: red peak corresponds to the γ (2, 2, 0) reflection, the magenta and blue
peaks correspond to the δ(4, 0, 12) and α(3, 0, 3¯, 0) reflections, respectively. A rocking
scan performed with the scattering angle fixed on the α(3, 0, 3¯, 0) reflection is plotted
in the Fig. 8.5(c). The sharp peak superimposed on a strong background indicates the
coexistence of polycrystalline and epitaxial α-Al2O3, as will be evidenced further below.
The fitting was done using a Lorentzian lineshape with a constant background. The
sharp component is due to the epitaxial α-Al2O3, while the broad component is due to
the contribution of the polycrystalline alumina (Debye-Scherrer ring).
The relative orientation relationship between the epitaxial α-alumina and the un-
148 Oxidation of NiAl(110) at atmospheric pressure
Figure 8.6: [left] In-plane reciprocal lattice of NiAl(110) (black open circles), γ-Al2O3(111) (red
circles) and α-Al2O3(0001) (blue circles). [right] Real space orientation of the oxygen sublattice
in epitaxial γ- and α-alumina with respect to the underlying substrate.
derlying substrate was determined to be NiAl(110) ‖ α-Al2O3 (0001) and [11¯0] NiAl ‖
[213¯0] α-Al2O3. It should be noted that during the transformation of the epitaxial γ- to
epitaxial α-Al2O3 the orientation of the oxygen sublattice with respect to the substrate
remains unchanged. It is the originally fcc oxygen arrangement in γ-alumina (ABCABC
layer sequence) that is converted into the hcp stacking ABAB by shearing of the O2−
ion planes and cation rearrangement. The (111) γ-Al2O3 plane becomes the (0001) plane
of corundum. A top view of the real space orientation of the oxygen sublattice (blue
spheres) in the epitaxial γ and α-alumina with respect to the substrate unit cell is shown
in Fig. 8.6[right]. The green and the orange spheres represent the Ni and the Al inside
the substrate unit cell, respectively. The sketch in Fig. 8.6[left] represents the correspond-
ing in-plane reciprocal space of NiAl(110) (black open circles), γ-Al2O3 (111) (red filled
circles) and α-Al2O3 (0001) (blue circles).
Fig. 8.7 shows rocking scans with the scattering angle fixed on the (3, 0, 3¯, 0) reflection
of α-Al2O3 performed during the oxidation at 1100
◦C. The incident angle was fixed to
αi=0.4
◦ and the exit angle equal to αf=0.56◦, αf=2◦ and αf=4◦, respectively. The
presence of a peak in the rocking scan even at large exit angles—the CTR signal—is an
indication of a smooth oxide layer.
At 1100◦C the radial scans revealed the presence of polycrystalline δ- and α-Al2O3,
as can be seen in Fig. 8.8(a). After two hours at this temperature, the transformation to
α-Al2O3 was complete: no δ-Al2O3 reflections were present. The first scan (magenta line)
was performed immediately after heating at 1100◦C, whereas the other scan (blue line)
was recorded two hours later. At 1200◦C the oxide scale was found to consist of solely
8.3 Results and discussions 149
22 23 24 25 26 27 28
40
i = 0.4º
f = 0.56º
FWHM = 0.727o
 
In
te
ns
ity
 (A
rb
. u
ni
ts
)
 (deg.)
4
i = 0.4º
f = 2º
FWHM = 1.209o
 
 
 
1
2 i = 0.4º
f = 4º
FWHM = 2.423o
 
 
 
 
Figure 8.7: Diffracted intensity as a function of the in-plane rotation angle, θ, with the scat-
tering angle fixed on the (3, 0, 3¯, 0) reflection of α-Al2O3 during oxidation at 1100◦C.
α-Al2O3. Fig. 8.8(b) shows two radial in-plane scans performed through the α(3, 0, 3¯, 0)
reflection (red line) and in an arbitrary direction (black line), respectively. For a better
comparison the intensity of the black scan was multiplied by a factor of two. These scans
were recorded during oxidation at 1200◦C with the incident angle fixed to αi=0.4◦. All the
peaks were identified to belong to α-Al2O3. It can be easily observed, that the α(3, 0, 3¯, 0)
reflection is much weaker when the scan is performed in an arbitrary direction, in which
all the diffracted intensity is due to polycrystalline alumina. Since the epitaxial α-Al2O3
contributes to the diffracted intensity only when the scan is performed in the [11¯0]NiAl
direction, this is a clear indication that the epitaxial α-Al2O3 is preserved even at 1200
◦C.
After five hours of oxidation at 1200◦C, the sample was cooled down to room temper-
ature with 20◦C per minute and the same in-plane measurements were repeated. From
the room temperature in-plane XRD measurements, d -spacings of several α-Al2O3 planes
(a)
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
102
103
104
(1
10
) N
iA
l
Q (Å-1)
In
te
ns
ity
 (A
rb
. u
ni
ts
) 
(5
23
/5
16
/2
0.
15
)
(3
33
)
(3
18
)(2
26
)
(3
14
/3
05
)
(3
12
)
(2
22
)
(1
17
)
(1
16
)
(1
14
/1
05
)
(1
13
)
(1
11
)
 
 
 1100oC
 after 2h at 1100oC;  arbitrary direction 
 
 
(b)
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
102
103
104
105
 
In
te
ns
ity
 (A
rb
. u
ni
ts
) 
Q (Å-1)
 
(2
13
7)
(1
01
.1
0)
(2
02
8)
(3
03
0)
(2
13
4)
(1
23
2)
(2
13
1)
(1
12
6)
(0
22
4)
(2
02
2)
(1
12
3)
(0
00
6)
 1200oC 
 1200oC;  arbitrary direction
(1
12
0)(
10
14
)
(0
11
2)
(1
10
) N
iA
l
Figure 8.8: (a) Radial in-plane scans during oxidation at 1100◦C. (b) Radial in-plane scans
through α-Al2O3 (3, 0, 3¯, 0) reflection (black line) and in a non high symmetry direction (red
line) during oxidation at 1200◦C.
150 Oxidation of NiAl(110) at atmospheric pressure
Table 8.1: The d-spacing from the diffraction data (d), strain free d-spacing (d0) and the
calculated strain values for α-Al2O3. The strain free lattice constants were considered to be
a0=4.7589 A˚ and c0=12.991 A˚ [87]. The thermal expansion coefficient of α-Al2O3 at 1200◦C is
α=10.2·10−6 K−1 [126].
Room temperature 1200◦
Reflection dRT (A˚) d0 (A˚) Strain a (A˚) d
1200 (A˚) d0(1+α∆T ) (A˚) Strain
(3, 0, 3¯, 0)epi 1.366 1.373 -0.518% 4.733 1.389 1.389 0%
(1, 1, 2¯, 0)poly 2.378 2.379 -0.038% 4.756 2.378 2.379 -0.038%
were evaluated in order to get information about the in-plane strain in the oxide scale.
The epitaxial α-Al2O3 is under compressive stress, due to the difference in the thermal
expansion coefficients of the oxide and the substrate. However, there is a negligible value
of the in-plane strain in the polycrystalline oxide [see Table 8.1]. The low value of strain
in the polycrystalline alumina can be attributed to the similar thermal expansion coeffi-
cient of the polycrystalline and epitaxial alumina. In order to estimate the growth strains
in the oxide scale, the d-spacings for several alumina lattice planes were calculated from
the data measured at 1200◦C. The thermal expansion coefficient of alumina at the above
mentioned temperature is α=10.2×10−6 K−1 [126]. The growth strain was found to be
negligible, both in the epitaxial and in the polycrystalline α-Al2O3 [see Table 8.1]. This is
in agreement with previous studies, reporting an initial tensile stress state which declined
to zero as soon as the transformation of metastable to α-Al2O3 was complete [11, 12].
8.3.3 Ex situ measurements after oxidation at 1200◦C
A. Grazing incidence x-ray diffraction measurements
Out-of-plane structure of α-Al2O3 The out-of-plane structure of α-Al2O3 was probed
by measuring the diffracted intensity along nine different rods with the laboratory set-up
described before. Fig. 8.9(a) shows the diffracted intensity along the (1¯1¯2L), (1¯01L) and
(2¯02L) reciprocal lattice directions as a function of the reciprocal lattice coordinate, L.
The reflections were indexed using the α-Al2O3 coordinates. The lattice parameters of α-
Al2O3 unit cell are a=b=4.7142 A˚ and c=12.93 A˚, so that L corresponds to the reciprocal
lattice unit |c∗| = 2pi/c=0.4859 A˚−1.
In Fig. 8.9(b) the fcc-like (011¯0)/(0001) and hcp-like (112¯0)/(0001) reciprocal lattice
planes are shown. The fcc-like reciprocal plane is due to the stacking of the missing Al3+
ions in the corundum structure. Therefore, similar considerations as for the (H0L) plane
of γ-Al2O3 hold [see Fig. 8.3(b) in the section 8.3.1.B]. As for the of γ-Al2O3, twin domains
8.3 Results and discussions 151
(a)
0 2 4 6 8 10
102
104
(3
03
0)
I+II I+II
I+III+II(112L)
  
L (r.l.u.)
In
te
ns
ity
 (c
ou
nt
s 
/ 2
0 
s.
)
(0
22
4)
102
104
(2
13
4)
(0
22
4)
(1
12
3)
II
II
I
I(101L)
  
 
102
104
(2
13
4)
(1
12
6)
 
I
II
(202L)
 
 
 
II
I
(b)
Figure 8.9: (a) L-scans as a function of the relative coordinate L in different reciprocal lattice
directions normal to the surface. (b) The (011¯0)/(0001) (fcc-like) and (112¯0)/(0001) (hcp-like)
reciprocal planes of α-Al2O3.
are also present, which can be deduced by the presence of reflections with L=4, 10 and
L=2, 8 in the (01¯1L) and (02¯2L) directions, respectively. Additional peaks appear at
non-integer L values, which are due to the intersection of the L-scan with Debye-Scherrer
rings of polycrystalline alumina.
B. Transmission electron microscopy characterization
To confirm and complete the diffraction data concerning the epitaxial α-Al2O3 layer,
cross-section transmission electron microscopy (TEM) images were recorded for the ox-
ide/substrate interface. Both high resolution (HR-) and conventional (CC-) TEM was
performed. Fig. 8.10(a) shows a conventional cross-section transmission microscopy im-
age of α-Al2O3(0001) / NiAl(110) system, which directly reveals a well-defined interface
between the substrate and the epitaxial oxide layer, whereas small voids could be seen
at the epitaxial/polycrystalline alumina interface. In all the TEM specimens which were
analyzed, the epitaxial oxide formed a continuous layer all along the interface, having an
average thickness of about 150 nm. An atomic view on the epitaxial alumina/substrate
interface can be seen in the HR-TEM image in Fig. 8.10(b). The interface is smooth,
in agreement with the observed CTR signal [see Fig. 8.7]. The epitaxial relationship de-
termined from the Fourier transformation of the HR-TEM images agrres well with the
GIXRD findings. In Fig. 8.11 the atomistic model of the α-Al2O3(0001)/NiAl(110) in-
terface viewed along the [101¯0] direction of α-Al2O3 is shown. The small red spheres
represent Al ions in the oxide, while the oxygen ions are colored in dark blue. The green
and the red spheres represent Ni and Al atoms inside the substrate, respectively.
152 Oxidation of NiAl(110) at atmospheric pressure
Figure 8.10: (a) Conventional cross-section transmission microscopy image of α-Al2O3(0001) /
NiAl(110)interface. (b) High-resolution cross-section transmission microscopy image of epitaxial
α-Al2O3(0001) / NiAl(110)interface.
8.4 Summary
In conclusion,we have performed grazing incidence x-ray diffraction measurements to
study the oxidation of NiAl(110). The oxidation at 870◦C and 1 bar O2 leads to the
formation of a well-ordered γ-Al2O3 epitaxial layer. The orientation relationship of the
oxide layer with respect to the underlying substrate has been determined from the in-
plane diffraction measurements and it corresponds to the R30◦ orientation relationship:
NiAl(110)bcc ‖ γ-Al2O3 (111)fcc and [21¯1¯]fcc γ-Al2O3 ‖ [001]bcc NiAl. Out-of-plane struc-
tural characterization of the γ-Al2O3 layer indicate the formation of twin domains, which
occur where the ABC sequence of the oxygen layers is reversed to CBA.
The stability of the γ-Al2O3 epitaxial layer and its transition to α-Al2O3 phase was
8.4 Summary 153
Figure 8.11: Atomistic model of the α-Al2O3(001)/ NiAl(110)interface viewed along the [101¯0]
direction of α-Al2O3.
further investigated by performing in situ GIXRD measurements during oxidation in air
at different temperatures up to 1200◦C. The oxide transformation sequence was found
to be γ- → δ- → α-Al2O3. No θ-Al2O3 was formed at any stage of the oxidation, in
contrast to other experimental observations: Heuer et al. [11] have reported θ-Al2O3 to
be the first oxide phase to form by oxidation of NiAl(110) at 1100◦C, which gradually
transformed to the thermodynamically stable α-Al2O3. We argue that the formation of
different alumina polymorphs on the same orientation of NiAl single crystals is due to
the different oxidation conditions during the two experiments: in the study by Heuer et
al. [11], the NiAl crystal was heated directly to 1100◦C, whereas in the present study
the temperature was raised in six steps to 1100◦C. It is known that δ-Al2O3 is less stable
then θ-Al2O3, therefore the former is expected to form at lower oxidation temperatures.
In addition, the formation of δ-Al2O3 might be promoted by the preexisting γ-Al2O3 due
to the structural similarities between the two phases.
At temperatures around 1000◦C there is a coexistence of epitaxial γ- and polycrys-
talline δ-Al2O3. The observed reflections of δ-Al2O3 correspond to those of the tetragonal
structure reported by Lippens et al. [84]. α-alumina starts to form at 1025◦C and only
at 1100◦C the δ- → α-Al2O3 transformation was complete.
At 1200◦C the oxide scale was observed to consist of both polycrystalline and epitaxial
α-Al2O3. The relative orientation relationship between the epitaxial α-alumina and the
underlying substrate was determined to be NiAl(110) ‖ α-Al2O3 (0001) and [11¯0] NiAl ‖
[213¯0] α-Al2O3. A recent ab initio thermodynamics study [10] on the α-Al2O3(0001) /
154 Oxidation of NiAl(110) at atmospheric pressure
NiAl(110) interface stability reports the experimentally observed orientation relationship
as the most stable one for this interface. Several other configurations have been tested
which lead to much higher interfacial mismatch, and therefore much larger interfacial
energies as compared to the aforementioned orientation.
The cross-section TEM observations confirmed the presence of a continuous layer of
epitaxial α-Al2O3 having a thickness of about 150 nm. The TEM images revealed also a
well-defined interface between the substrate and the epitaxial layer. Therefore, it can be
concluded that the initial γ-Al2O3 layer transforms directly into the epitaxial α-Al2O3.
This corresponds to a fcc–hcp transformation of the oxygen ion sublattice. It is important
to underline that during this transformation the orientation of the oxygen sublattice with
respect to the substrate remains unchanged. Its crystallographic nature resembles that
of a martensitic transformation and it proceeds by shearing of the oxygen ion planes (the
stacking is changed from ABCABC in γ-Al2O3 to ABAB in α-Al2O3), as well as by the
rearrangement of the cations in the lattice.
From the room temperature in-plane XRD measurements, the residual in-plane strain
in the oxide scale was estimated. The epitaxial α-Al2O3 was found to be under compressive
stress, due to the difference in the thermal expansion coefficients of the oxide and the
substrate. A negligible value of the residual in-plane strain in the polycrystalline alumina
was found which is explained similar thermal expansion coefficients of the polycrystalline
and epitaxial alumina. The growth strain was found to be negligible, both in the epitaxial
and in the polycrystalline α-Al2O3, in agreement with previous studies, reporting an initial
tensile stress state which declined to zero as soon as the transformation of metastable
aluminas to α-Al2O3 was complete [11, 12].
Chapter 9
Summary
In this work, the transition from ultra-thin oxide layers to bulk oxides during thermally
controlled oxidation of CoGa(100) and NiAl(110) has been studied as a function of tem-
perature and the oxygen pressure. In particular, the structure, morphology and thickness
of the oxide layer, epitaxial orientation relationship, domain structure, and changes in the
substrate due to selective oxidation, were thoroughly investigated. The main experimental
methods employed in this work were surface x-ray diffraction (SXRD) and high resolution
core level spectroscopy (HRCLS). Prior to the synchrotron experiments, the cleanliness
and structure of the samples surface were monitored by low energy electron diffraction
(LEED) and Auger electron spectroscopy (AES). The SXRD and HRCLS measurements
have been performed making use of the brilliant synchrotron radiation at various facili-
ties, i.e., Angstro¨mquelle Karlsruhe (Germany), European Synchrotron Radiation Facil-
ity (France) and Max-LAB (Sweden). Complementary information has been obtained by
means of transmission electron microscopy (TEM)1 and atomic force microscopy (AFM)2
measurements. Moreover, the results of first-principles thermodynamics3 on the stability
of surface oxides were used to complement the experimental data.
Oxidation of CoGa(100)
The formation of a well-ordered ultra-thin surface gallium oxide layer was previously ob-
served after exposing a CoGa(100) surface at relatively low temperatures and oxygen
pressures. LEED measurements show a (2×1) reconstruction in two domains perpen-
dicular on each other. In this work, an atomic level understanding of the structure of
this surface gallium oxide has been achieved based on the combined use of experimen-
1The TEM measurements have been performed by A. Catanoiu, Y. Jin-Phillipp and G. Richter.
2The AFM measurements have been performed by E. Barrena and X. Zhang.
3The DFT calculations have been performed by G. Kresse and M. Marsman.
156 Summary
tal (SXRD, HRCLS) and theoretical methods (DFT-based calculations). The resulting
structural model contains the basic building block of bulk β-Ga2O3, although deviations
from the bulk phase stoichiometry are observed. The oxide film consists of an oxygen ion
double layer with two Ga ions located in truncated octahedral and tetrahedral sites of
the fcc oxygen ion sublattice. At the interface, Ga ions occupy sites of the corresponding
stacking sequence of the bulk β-Ga2O3. The two oxygen interfacial atoms share half of
their bonds with the strongly buckled Ga substrate atoms, recovering the formal oxida-
tion state of Ga3+ in the oxide film. The thickness of the surface gallium oxide film was
determined to be 4.5 A˚. Additional information have been obtained by means of HRCLS
measurements. The decomposition of the Ga 3d spectrum revealed three oxygen-induced
components, whereas two components split by 0.6 eV could be identified in the O 1s
spectrum. The calculated Ga 3d and O 1s core-level shifts (including final state effects)
agree well with the experimental core-level spectra.
The stability diagram of gallium oxide on the CoGa(100) surface was mapped out by
means of in situ SXRD measurements. Both a surface oxide and the bulk β-Ga2O3 phase
were identified, even though, according to thermodynamical considerations, only the bulk
β-Ga2O3 is predicted to form throughout the investigated (p-T ) range. The surface oxide
forms under metastable equilibrium conditions. The surface oxide structure was preserved
at room temperature even at pressures as high as 1 bar O2, maintaining its high degree of
crystallinity. The bulk oxide growth was found to be limited by kinetics, which hinder its
formation for temperatures lower than 350◦C (within the experimentally accessed time
scale). At higher temperatures, three-dimensional bulk Ga2O3 islands grows epitaxially
on CoGa(100), provided that enough oxygen is supplied to the system.
HRCLS measurements have been performed after oxidation at temperatures up to
750◦C and pressures between 10−8 and 5×10−5 mbar O2. An additional higher binding
energy component has been identified for both the Ga 3d and O 1s core-level spectra after
oxidation at 750◦C. This could be an indication of a CoGa2O4 spinel formation.
SXRD measurements showed that at temperatures higher than 750◦C, the formation
of the bulk gallium oxide is accompanied by substrate faceting. The facets and the
substrate join at a 26◦ angle, as determined from the SXRD measurements, corresponding
to (1, 0, 2)-type of planes. Ex situ AFM measurements indicated a similar, i.e., faceted,
morphology of the oxide surface.
In conclusion, the multi-technique approach adopted in this work is very well suited for
investigating the structure, morphology and stability of oxide layers grown on different
metallic alloys. This study shows that the thickness and structural perfection of the
oxide layers on alloys can be tailored by the appropriate choice of oxygen pressure and
157
temperature, which is of crucial importance for all applications involving ultrathin oxide
films.
Oxidation of NiAl(110)
Medium pressure oxidation of NiAl(110)
Within this project, SXRD measurements have been employed for structural characteri-
zation and in situ evolution of the aluminium oxide/NiAl(110) system during oxidation
at different temperatures and oxygen pressures ranging from 5×10−6 up to 10 mbar O2.
An ultra-thin aluminum oxide layer was grown by thermal oxidation using the well-
established recipe: oxidation at 270◦C with 5×10−6 mbar O2 for 15 minutes followed by
annealing at 800◦C for 5 minutes [7]. We have performed SXRD measurements in order
to probe the ultra-thin aluminium oxide/NiAl(110) structure. The experimental in-plane
structure factors are consistent with the data published by Stierle et al. [7]. A comparison
between our experimental data and the model proposed by Kresse et al. [8] was made.
The experimental in-plane structure factors were reproduced with less accuracy by this
model, as compared to the model proposed by Stierle et al. [7]. Although reasonable
agreement was found on the surface rods (sensitive to the oxide structure), the poor
agreement on the crystal truncation rods proved that the DFT-based model lacks some
structural features and can not accurately describe the oxide/alloy interface.
Structural refinement of the oxide/alloy interface was performed and two different
models were found to reproduce the experimental data. The characteristic structural
feature of the first model is the presence of point defects (Al anti-site atoms) in the
topmost substrate layer. It is consistent with the model previously proposed by Stierle et
al. [9]. The second model assumes a perfect bulk stoichiometry and z-displacements of the
atoms in the first three substrate layers. The unambiguous discrimination between these
two models is still a critical issue. We expect that anomalous scattering measurements at
the K adsorption edge of Ni will provide the information needed to single out the correct
model.
The transition of the ultra-thin aluminium oxide to bulk Al2O3 was followed in situ at
350◦C. The ultra-thin oxide was found to be stable up to 0.01 mbar O2. The sample was
subsequently annealed at 800◦C in order for the newly formed oxide to crystallize. The
coexistence of surface oxide and epitaxial γ-Al2O3 oxide islands showing the Kurdjumov-
Sachs orientation was observed. Structural refinement of the oxide/alloy interface was
performed based on the CTR data. The main structural feature of the resulting interface
model is the presence of Ni anti-site atoms which occur in the first two substrate layers
158 Summary
with 69% and 58% probability, respectively. The enrichment of Ni at the interface appears
as a consequence of Al consumption by selective oxidation. For an aluminium oxide layer
being in contact with a NiAl(110) alloy having a composition in the Ni-rich region, the
formation of Ni anti-site atoms at the interface has been predicted in a recent ab initio
thermodynamics study [10], in agreement with our experimental observations.
After oxidation of the NiAl(110) substrate at 800◦C and 10 mbar O2, γ-Al2O3 oxide
islands formed showing the Kurdjumov-Sachs, Nishiyama-Wassermann and R30◦ in-plane
orientation. In addition, the presence of polycrystalline δ-Al2O3 was identified. Cross-
section transmission electron microscopy images of the aluminium oxide/NiAl(110) inter-
face revealed an oxide layer having an overall thickness of about ∼70 nm. The epitaxial
γ-Al2O3 did not form a continuous layer on top of the substrate. The formation of cavities
was identified at the oxide/alloy interface.
In conclusion, our work provides detailed insight into the stability of the ultra-thin
aluminium oxide layer on NiAl(110) at different conditions of oxygen pressures and tem-
peratures. We demonstrated that the thickness of the ultrathin aluminum oxide can
not be further increased by multiple oxidation cycles and it only exists as a 5 A˚ thick
layer. Further oxidation induces a drastic change in the oxide morphology, which in turn
can have important implications on the behavior of model catalysts deposited on the
aluminium oxide layer/NiAl(110), since changes in the oxide structure can influence the
catalytic behavior of metal particles.
Atmospheric pressure oxidation of NiAl(110)
Grazing incidence x-ray diffraction (GIXRD) and TEM measurements were employed
to study the atmospheric pressure oxidation of NiAl(110). The oxidation at 870◦C and
1 bar O2 leads to the formation of a well-ordered γ-Al2O3 epitaxial layer. The orientation
relationship at the oxide/substrate interface has been determined to be NiAl(110)bcc ‖ γ-
Al2O3 (111)fcc and [21¯1¯]fcc γ-Al2O3 ‖ [001]bcc NiAl, which corresponds to R30◦ orientation
relationship.
The transition from γ- to α-Al2O3 was further investigated during oxidation in air
at different temperatures up to 1200◦C by means in situ GIXRD measurements. The
oxide transformation sequence was found to be γ- → δ- → α-Al2O3. α-alumina starts to
grow at 1025◦C. At 1200◦C no metastable aluminas were present and the oxide scale was
observed to consist of both polycrystalline and epitaxial α-Al2O3. The relative orientation
relationship between the epitaxial α-alumina and the underlying substrate was determined
to be NiAl(110) ‖ α-Al2O3 (0001) and [11¯0] NiAl ‖ [213¯0] α-Al2O3.
From the room temperature in-plane XRD measurements, the residual in-plane strain
159
in the oxide scale was estimated. The epitaxial α-Al2O3 was found to be under compressive
stress, due to the difference in the thermal expansion coefficients of the oxide and the
substrate. A negligible value of the residual in-plane strain in the polycrystalline alumina
was found, which is understood in terms of similar thermal expansion coefficients of the
polycrystalline and epitaxial alumina. The growth strain was found to be negligible, both
in the epitaxial and in the polycrystalline α-Al2O3, in agreement with previous studies
reporting an initial tensile stress state which declined to zero as soon as the transformation
of metastable to α-Al2O3 was complete [11, 12].
Ex situ TEM investigations gave evidence of a continuous layer of epitaxial α-alumina
having a thickness of about 150 nm. The initial γ-Al2O3 layer transforms directly into
the epitaxial α-Al2O3 by a fcc–hcp martensitic-like transformation of the oxygen ion
sublattice. However, the orientation of the oxygen sublattice with respect to the substrate
remains unchanged during the transformation, that means it is the same as in the γ-Al2O3
having the R30◦ orientation. A recent theoretical study [10] has put forward the observed
orientation as being the most stable one at the α-Al2O3(0001) / NiAl(110) interface.
We conclude that the orientation of the initial γ-alumina layer plays a key role in
the further growth of epitaxial α-Al2O3: it was shown that depending on the oxidation
conditions, γ-Al2O3 having different orientation relationships can be grown on NiAl(110)
surface. As described earlier, oxidation of NiAl(110) at 800◦C and 10 mbar O2 results
in the growth of γ-Al2O3 showing all three orientation relationships embedded in a poly-
crystalline δ-Al2O3 oxide. In contrast, a well-ordered γ-Al2O3 layer was found to form at
a slightly higher temperature, 870◦C and 1 bar O2.
We argue that the induced growth of an epitaxial α-Al2O3 layer could be significant
for the practical applications of Ni-Al intermetallics as bond coats in the thermal barrier
coating systems. Since the bond coat is polycrystalline, it will be interesting to investigate
the growth of epitaxial α-alumina on NiAl single crystals having other orientations, i. e.,
(100) and (111). Pre-oxidizing the bond coat to form large epitaxial oxide grains is
expected to lead to a decreased growth rate of the thermally grown oxide (α-alumina),
since less grain boundaries would be available for ion diffusion. This, in turn, would
be beneficial for the lifetime of the aforementioned systems, since it was shown that
the failure of thermal barrier coating can be correlated with a critical thickness of the
thermally grown oxide [13].

Appendix A
Atomic coordinates of surface gallium oxide model
Table 9.1: Atomic coordinates for the (a) experimental and (b) DFT model of the surface
gallium oxide on CoGa(100).
Atom x y z
Co 0.5 0 -0.5041
Co 0 0 -0.5041
Ga 0.2423 0.4838 -0.0019
Ga 0.7358 0.4856 0.0352
Co 0.4880 -1.0540 0.4917
Co 0.0049 -1.0554 0.5008
Ga 0.2190 -0.5355 0.9890
Ga -0.2787 -1.0299 1.1605
O -0.2928 -0.5315 1.5439
O 0.1995 -0.5338 1.6585
Ga 0.2318 -0.0408 2.0398
Ga -0.2711 -0.5410 2.2225
O -0.4656 -0.0499 2.2959
O -0.0721 -1.0711 2.3820
(a) Experimental model
Atom x y z
Co -0.50094 -1.00116 -0.50226
Co -0.00636 -1.0012 -0.50211
Ga 0.24433 -0.50326 -0.01268
Ga -0.25621 -0.50402 0.0266
Co -0.00018 -0.006 0.48726
Co -0.51478 -0.0061 0.48783
Ga 0.2414 -0.50795 0.97835
Ga -0.25871 -0.01996 1.1414
O -0.25803 -0.5193 1.57555
O 0.24023 -0.51727 1.64646
Ga 0.23171 -0.01758 2.05517
Ga -0.2684 -0.51714 2.24527
O -0.4889 -1.0169 2.37792
O -0.04684 -1.01747 2.37816
(b) DFT model
162 Appendix B
Appendix B
HRCLS fit parameters
Table 9.2: The fit parameters for the Ga 3d spectrum shown in Fig. 6.13(a).
Peak BE Γ α σ FWHM
(eV) (meV) (meV) (meV)
1 18.398 155 0.1 190.5 316
2 18.838 153.4 0.09 192.5 316.3
3 18.538 157 0.1 180 307.2
4 18.978 155.4 0.09 178 303.5
5 19.079 164 0.06 400 519.6
6 19.539 162.3 0.06 395.9 514.1
7 20.060 256 0.06 555.3 740.0
8 20.500 254.3 0.06 550.4 733.5
Table 9.3: The fit parameters for the O 1s spectrum shown in Fig. 6.13(b).
Peak BE Γ α σ FWHM
(eV) (meV) (meV) (meV)
1 529.6085 280 0.09 600.3 834.1
2 530.1985 283 0.09 604.8 841.6
164 Appendix C
Appendix C
Acronyms
AES Auger Electron Spectroscopy
ANKA A˚ngstrømquelle Karlsruhe
AFM Atomic Force Microscopy
CLS Core Level Spectra
CTR Crystal Truncation Rod
DFT Density Functional Theory
ESRF European Synchrotron Radiation Facility
EELS Electron Energy Loss Spectroscopy
FWHM Full Width at Half Maximum
GFWHM Gaussian Full Width at Half Maximum
GIXRD Grazing Incidence X-ray Diffraction
HRCLS High Resolution Core Level Spectroscopy
IMPF Inelastic Mean Free Path
LEED Low Energy Electron Diffraction
LFWHM Lorentzian Full Width at Half Maximum
MRAM Magnetic Random Access Memories
RFA Retarding Field Analyzer
RT Room temperature
SCLS Surface Core Level Shift
SXRD Surface X-ray Diffraction
STM Scanning Tunneling Microscopy
TBC Thermal Barrier Coating
166 Appendix C
TGO Thermally Grown Oxide
TEM Transmission Electron Microscopy
UHV Ultra High Vacuum
UPS Ultraviolet Photoelectron Spectroscopy
Bibliography
[1] N. P. Padture, M. Gell, and E. H. Jordan, Science 296, 280 (2002).
[2] R. Franchy, J. Masuch, and P. Gassmann, Appl. Surf. Sci. 93, 31 (1996).
[3] R.-P. Blum, D. Ahlbehrendt, and H. Niehus, Surf. Sci. 396, 176 (1998).
[4] R.-P. Blum, D. Ahlbehrendt, and H. Niehus, Surf. Sci. 366, 107 (1996).
[5] H. Graupner, L. Hammer, K. Heinz, and D. M. Zehner, Surf. Sci. 380, 335 (1997).
[6] G. Schmitz, P.Gassmann, and R. Franchy, Surf. Sci. 397, 339 (1998).
[7] A. Stierle, F. Renner, R. Streitel, H. Dosch, W. Drube, and B. C. Cowie, Science
303, 1652 (2004).
[8] G. Kresse, M. Schmid, E. Napetschnig, M. Shishkin, L. Ko¨hler, and P. Varga,
Science 308, 1440 (2005).
[9] A. Stierle, F. Renner, R. Streitel, and H. Dosch, Phys. Rev. B 64, 165413 (2001).
[10] J. Feng, W. Zhang, W. Jiang, and H. Gu, Phys. Rev. Lett. 97, 246102 (2006).
[11] A. Heuer, A. Reddy, D. Hovis, B. Veal, A. Paulikas, A. Vlad, and M. Ru¨hle,
Scripta Metallurgica et Materialia 54, 1907 (2006).
[12] P. Y. Hou, A. P. Paulikas, and B. W. Veal, Mat. Sci. Forum 461, 671 (2004).
[13] A. G. Evans, D. R. Mumm, J. W. Hutchinson, G. H. Meier, and F. S. Pettit,
Progr. Mat. Sci. 46, 505 (2001).
[14] Corrosion and Metal Artifacts — A Dialogue Between Conservators and Archae-
ologists and Corrosion Scientists, edited by B. F. Brown, H. C. Burnett, W. T.
Chase, M. Goodwa, J. Kruger, and M. Pourbaix (National Association of Corrosion
Engineers, Huston, Texas, 1991).
168 Bibliography
[15] G. della Porta, Magiae naturalis, 1558 (quotation from English Translation, Natural
Magic, London, 1658).
[16] M. Atalla, E. Tannenbaum, and E. J. Scheibner, Bell Syst. Tech. J. 38, 749 (1959).
[17] J. S. Moodera, L. R. Kinder, T. M. Wong, and R. Meservey, Phys. Rev. Lett. 74,
3273 (1995).
[18] T. Miyazaki and N. Tezuka, J. Magn. Magn. Mater. 139, L231 (1995).
[19] J. N. M. Covington and D. Song, Appl. Phys. Lett. 76, 3965 (2000).
[20] Z. Li, C. de Groot, and J. H. Moodera, Appl. Phys. Lett. 77, 22 (2000).
[21] STEINEL Solutions AG, Einsiedeln. www.steinel.ch.
[22] K. R. Lawless, Rep. Prog. Phys. 37, 231 (1974).
[23] J. Gustafson, Ph.D. thesis, Lund University, 2006.
[24] E. Lundgren, A. Mikkelsen, J. N. Andersen, G. Kresse, M. Schmid, and P. Varga,
J. Phys.: Condens. Matter. 18, 481 (2006).
[25] A. Stierle, V. Formoso, F. Comin, and R. Franchy, Surf. Sci. 85, 467 (2000).
[26] R. Franchy, Surf. Sci. Rep. 38, 195 (2000).
[27] R. Franchy, G. Schmitz, P. Gassmann, and F. Bartolucci, Appl. Phys. A 65, 551
(1997).
[28] G. Zhou and J. C. Yang, J. Mater. Res. 20, 1684 (2005).
[29] J. A. T. Fromhold, Theory of Metal Oxidation, Vol. I - Fundamentals of Defects in
Crystalline Solids (North-Holland Publishing Company, Amsterdam · New York ·
Oxford, 1976).
[30] C. Wagner, Z. Physik. Chem. B 21, 25 (1933).
[31] N. Cabrera and N. F. Mott, Rep. Prog. Phys. 12, 163 (1949).
[32] A. T. Fromhold and E. L. Cook, Phys. Rev. Lett. 158, 600 (1967).
[33] A. T. Fromhold and E. L. Cook, Phys. Rev. Lett. 17, 1212 (1966).
[34] N. F. Mott, Rep. Prog. Phys. 35, 1175 (1939).
[35] A. Stierle and H. Zabel, Europhys. Lett. 37, 365 (1997).
[36] O. Kubaschewski and B. E. Hopkins, Oxidation of Metals and Alloys (Butterworths,
London, 1962).
[37] A. Atkinson, Rev. Mod. Phys. 57, 437 (1985).
Bibliography 169
[38] G. R. Wallwork, Rep. Prog. Phys. 39, 401 (1976).
[39] K. Reuter and M. Scheffler, Phys. Rev. B 65, 035406 (2001).
[40] M. Todorova, Ph.D. thesis, Technical University Berlin, 2004.
[41] E. Lundgren, J. Gustafson, A. Mikkelsen, J. N. Andersen, A. Stierle, M. T. H. Dosch,
J. Rogal, K. Reuter, and M. Scheffler, Phys. Rev. Lett. 92, 046101 (2004).
[42] L. I. Ponomarev, The Quantum Dice (Mir Publishers, Moscow, 1988).
[43] J. Als-Nielsen and D. McMorrow, Elements of Modern X-ray Physics (John Wiley
and Sons Ltd., West Sussex, England, 2001).
[44] R. Feidenhans’l, Surf. Sci. Rep. 10, 105 (1989).
[45] F. Kirchner and F. Lassen, Adv. Phys. 24, 113 (1935).
[46] M. v. Laue, Adv. Phys. 26, 55 (1936).
[47] D. H. Bilderback, P. Elleaume, and E. Weckert, J. Phys. B: At. Mol. Opt. Phys.
38, 773 (2005).
[48] B. E. Warren, X-ray Diffraction (Dover Publications, Inc., New York, 1990).
[49] J. M. Cowley, Diffraction Physics, 2nd ed. (North-Holland Publishing Company,
Amsterdam, 1981).
[50] B. D. Cullity, Elements of X-ray Diffraction, 2nd ed. (Addison-Wesley Publishing
Company, Inc., Phillipines, 1978).
[51] A. Authier, Dynamical Theory of X-ray Diffraction (Oxford University Press Inc.,
New York, 2001).
[52] I. K. Robinson and D. J. Tweet, Rep. Prog. Phys. 55, 599 (1992).
[53] I. K. Robinson, Phys. Rev. B 33, 3830 (1986).
[54] E. Vlieg, J. Appl. Cryst. 30, 523 (1997).
[55] C. Berg, S. Raaen, A. Borg, C. N. Andersen, E. Lundgren, and R. Nyholm,
Phys. Rev. B 47, 13063 (1993).
[56] A. Stierle, C. Tieg, H. Dosch, V. Formoso, E. Lundgren, J. Andersen, L. Ko¨hler,
and G. Kresse, Surf. Sci. 569, 263 (2003).
[57] S. Doniach and M. Sˇunjic´, J. Phys. C: Solid State Phys. 3, 285 (1970).
[58] D. L. Adams, FitXPS Version 2.12.
[59] J. K. G. Ertl, Low Energy Electrons and Surface Chemistry, 2nd ed. (VCH Verlags-
gesellschaft mbH, ADDRESS, 1985).
170 Bibliography
[60] ANKA Instrumentation Book, edited by T. Baumbach, J. Go¨ttlicher, and M. Hagel-
stein (ANKA Angstroemquelle Karlsruhe, ISS Institute for Synchrotron Radiation,
Karlsruhe, 2006).
[61] A. Stierle, A. Steinha¨user, A. Ru¨hm, F. U. Renner, R. Weigel, N. Kasper, and H.
Dosch, Rev. Sci. Inst. 75, 5302 (2004).
[62] www.certif.com.
[63] F. Comin, Rev. Sci. Inst. 66, 2082 (1995).
[64] Homepage of ID32 beamline at www.esrf.fr.
[65] R. Nyholm, J. Andersen, U. Johansson, B. Jensen, and I. Lindau, Nuclear Instru-
ments and Methods in Physics Research A 467-468, 520 (2001).
[66] N. Ma˚rtensson, P. Baltzer, P. Bru¨hwiler, J.-O. Forsell, A. Nilsson, A. Stenborg, and
B. Wannberg, J. Electron Spectrosc. Relat. Phenom. 70, 117 (1994).
[67] T. B. Massalski, Binary alloy phase diagrams, 2 ed. (William W. Scott jr., ASM
International, 1992).
[68] S. M. Kim, Phys. Rev. B 70, 054113 (1986).
[69] W. Pies and A. Weiss, Numerical data and functional relationships in science and
technology, New series (Springer Verlag, Berlin, 1997).
[70] R. Roy, V. Hill, and E. Osborn, J. Am. Chem. Soc. 74, 719 (1952).
[71] J. A˚hman, G. Svensson, and J. Albertsson, Acta Crystallogr. C52, 1336 (1996).
[72] S. Geller, J. Chem. Phys. 33, 676 (1960).
[73] H. He, M. A. Blanco, and R. Pandey, Appl. Phys. Lett. 88, 261904 (2006).
[74] Y. Tomm, J. Ko, A. Yoshikawa, and T. Fukuda, Solar Energy Materials and So-
lar Cells 66, 369 (2001).
[75] L. Binet, D. Gourier, and C. Minot, J. Sol. State Chem 113, 420 (1994).
[76] A. Taylor and N. J. Doyle, J. Appl. Cryst. 5, 201 (1972).
[77] A. Taylor and N. J. Doyle, Proc. Roy. Soc. London A159, 56 (1937).
[78] D. B. Miracle, Acta Metall. Mater. 41, 649 (1993).
[79] R. Krachler and H. Ipser, Phys. Rev. B 70, 054113 (2004).
[80] M. Kogachi, Y. Takeda, and T. Tanahashi, Intermetallics 3, 129 (1995).
[81] P. Korzhavyi, I. Abrikosov, and B. Johansson, Mat. Res. Soc. Symp. Proc 552,
5351 (1999).
Bibliography 171
[82] D. Farkas, B. Mutasa, C. Vailhe, and K. Ternes, Modelling Simul. Mater. Sci. Eng.
3, 201 (1995).
[83] S. J. Wilson, Proc. Br. Ceram. Soc. 28, 281 (1979).
[84] B. C. Lippens and J. H. D. Boer, Acta Crystallogr. 17, 1312 (1964).
[85] I. Levin and D. Brandon, J. Am. Ceram. Soc. 81, 1995 (1998).
[86] L. Pauling and S. B. Hendricks, J. Am. Chem. Soc. 47, 781 (1925).
[87] W. E. Lee and K. P. D. Lagerlof, J. Electron Micr. Tech.. 2, 247 (1985).
[88] V. Jayaram and C. G. Levi, Acta Metallurgica 37, 569 (1989).
[89] B. Ealet, M. H. Elyakhloufi, E. Gillet, and M. Ricci, Thin Solid Films 250, 92
(1994).
[90] S.-D. Mo, Y.-N. Xu, and W.-Y. Ching, J. Am. Ceram. Soc. 80, 1193 (1997).
[91] G. Paglia, A. L. Rohl, C. E. Buckley, and J. D. Gale, Phys. Rev. B 71, 224115
(2005).
[92] G. Paglia, C. E. Buckley, A. L. Rohl, B. A. Hunter, R. D. Hart, J. V. Hanna, and
L. T. Byrne, Phys. Rev. B 68, 144110 (2003).
[93] E. Eumann, G. Schmitz, and R. Franchy, Surf. Sci. 72, 3440 (1998).
[94] G. Schmitz, P. Gassmann, and R. Franchy, Surf. Sci. 397, 339 (1998).
[95] R. Franchy, M. Eumann, and G. Schmitz, Surf. Sci. 470, 337 (2001).
[96] R. Streitel, Ph.D. thesis, .
[97] A. Stierle, R. Streitel, P. Nolte, A. Vlad, I. Costina, M. Marsman, G. Kresse, E.
Lundgren, J. N. Andersen, R. Franchy, and H. Dosch, New J. Phys. 9, 331 (2007).
[98] G. Schmitz, P. Gassmann, and R. Franchy, J. Appl. Phys. 83, 2533 (1998).
[99] D. Stull and H. Prophet, JANAF Thermochemical Tables, 2 ed. (US National Bu-
reau of Standards, Washington DC, ADDRESS, 1971).
[100] U. . Koops and M. . Martin, J. Phys. C: Solid State Phys. 136-137, 971 (2000).
[101] J. C. Yang, K. Nadarzinski, E. Schumann, and M. Ru¨hle, Scripta Metallur-
gica et Materialia 33, 1043 (1995).
[102] M. Finnis, A. Lozovoi, and A. Alavi, Annu. Rev. Mater. Res. 35, 167 (2005).
[103] R. M. Jaeger, H. Kuhlenbeck, H. J. Freund, M. Wuttig, W. Hoffmann, R. Franchy,
and H. Ibach, Surf. Sci. 259, 235 (1991).
172 Acknowledgments
[104] J. Libuda, F. Winkelmann, M. Ba¨umer, H. J. Freund, T. Bertams, H. Neddermeyer,
and K. Mu¨ller, Surf. Sci. 318, 61 (1994).
[105] E. Vlieg, J. Appl. Cryst. 33, 401 (2000).
[106] M. Yoshitake, B. Mebarki, and T. T. Lay, Surf. Sci. 511, 313 (2002).
[107] T. T. Lay, M. Yoshitake, and W. Song, Appl. Surf. Sci. 239, 451 (2005).
[108] Y. Lykhach, V. Moroz, and M. Yoshitake, Appl. Surf. Sci. 241, 250 (2005).
[109] K. F. McCarty, Surf. Sci. 474, 165 (2001).
[110] G. Wassermann, Arch. Eisenhu¨ttenwes 126, 647 (1933).
[111] Z. Nishiyama, Sci. Rep. Tohoku Univ. 23, 638 (1934).
[112] G. Kurdjumov and G. Sachs, Z. Phys. 64, 325 (1930).
[113] Y. Gotoh and H. Fukuda, Surf. Sci. 223, 315 (1989).
[114] O. Hellwig, K. Theis-Bro¨hl, G. Wilhelmi, A. Stierle, and H. Zabel, Surf. Sci. 398,
379 (1998).
[115] H. J. Grabke, Intermetallics 7, 1153 (1999).
[116] A. Strecker, U. Salzberger, and J. Mayer, Prakt. Mettalogr. 30, 482 (1993).
[117] H. Svensson, K. Stiller, and J. Angenete, Mat. Sci. Forum 461, 231 (2004).
[118] J. C. Yang, E. Schumann, H. Mu¨llejans, and M. Ru¨hle, J. Phys. D: Appl. Phys. 29,
1716 (1996).
[119] J. C. Yang, E. Schumann, I. Levin, and M. Ru¨hle, Acta Materialia 46, 2195 (1998).
[120] M. Bobeth, E. Bischoff, E. Schumann, M. Rockstroh, and M. Ru¨hle, Corrosion Sci.
37, 657 (1995).
[121] M. W. Brumm and H. J. Grabke, Corrosion Sci. 33, 1677 (1992).
[122] M. W. Brumm and H. J. Grabke, Corrosion Sci. 34, 547 (1993).
[123] M. W. Brumm, H. J. Grabke, and B. Wagemann, Corrosion Sci. 36, 37 (1994).
[124] E. Schumann, C. Sarioglu, J. R. Blachere, F. S. Pettit, and G. H. Meier, Oxida-
tion of Metals 53, 259 (2000).
[125] U. Essmann, R. Henes, U. Holzwarth, F. Kloppfer, and E. Bu¨chler,
Phys. Stat. Sol. A 160, 487 (1997).
[126] Y. S. Touloukian, R. K. Kirby, R. E. Taylor, and T. Y. R. Lee, Thermal Expansion:
Nonmetallic Solids (IFI, Plenum, 1977), Vol. 13.
Acknowledgements
A number of people offered me an important support, not only in scientific matters but
also in making these years spent in Stuttgart truly exceptional. I would like to take this
opportunity to express my sincere gratitude to:
• Prof. Helmut Dosch for giving me the opportunity to carry out this work in
a scientifically exciting environment and for his constant support throughout the
course of this thesis. The friendly and supportive atmosphere inherent to his group
contributed essentially to the final outcome of my work.
• My supervisor, Dr. Andreas Stierle, for introducing me to the surface x-ray
diffraction field, as well as for his patient and continuous guidance at every stage
of my research, for important inputs to the present thesis, and for reading and
correcting the thesis manuscript.
• Prof. Manfred Ru¨hle for his help and support and for having accepted me
as member of the Research Training Group ”Innere Grenzfla¨chen in kristallinen
Materialien”, which gave me the opportunity to widen my knowledge in the field of
materials science.
• Prof. Horst Strunk for his efforts as the main referee.
• Prof. Helmut Bertagnolli for being part of the examination committee.
• My colleagues, for participating at the beamtimes and making those days and
nights really enjoyable: Ioan Costina, Nikolai Kasper, Melissa Delheusy and
Philipp Nolte. I would also like to acknowledge the help of Annette Weißhardt
and Frank Adams, as well as the technical support offered during the beamtimes
by Ralf Weigel (ANKA) and Frank Renner (ESRF).
• Edvin Lundgren for helpful discussions concerning the interpretation of the core-
level spectra.
174 Acknowledgments
• Esther Barrena and Xuena Zhang for the performance of the AFM measure-
ments.
• Amalia Ca˘ta˘noiu-Soare, Gunther Richter, Yun Jin-Phillipp for the perfor-
mance of the TEM measurements.
• Georg Kresse and Martijn Marsman for performing the DFT calculations.
• Sebastian Scho¨der for translating the thesis summary into German.
• Ja´nos Major for his helpfulness and very useful LaTex tips.
• All the members of the Department Dosch at the Max-Planck-Institut fu¨r Metall-
forschung in Stuttgart for all the nice moments I experienced next to them in the
last years, for creating a really pleasant and stimulating working environment, for
the help and advice they offered whenever needed. I would like to thank partic-
ularly to Amalia Ca˘ta˘noiu-Soare, Ioan Costina, Melissa Delheusy, Claus
Ellinger, Nikolai Kasper, Ruslan Kurta, Cristian Mocuta, Philipp Nolte,
Max Nu¨lle, Ingo Ramsteiner,Alexander Reicho, Sebastian Scho¨der, Sorin
Soare, Alexander Udyansky, Vedran Vonk.
Finally, I want to especially thank my family. No matter how far, they have always
been here and have given me affection and support all along the way, instilling in me
confidence and a drive for pursuing my PhD. Mult¸umesc!
This project has been financially supported by the Max Planck Society, Deutsche
Forschungsgemeinschaft (DFG) within the Graduiertenkolleg ”Innere Grenzfla¨chen in
kristallinen Materialien” (GRK 285) and by the EC/NSF Project HIPERCOAT (GRD2-
200-30 211).