Initial oxidation of zirconium: oxide-film growth kinetics ...
Transcript of Initial oxidation of zirconium: oxide-film growth kinetics ...
Max-Planck-Institut für Intelligente Systeme (ehemals Max-Planck-Institut für Metallforschung) Stuttgart
Initial oxidation of zirconium: oxide-film growth kinetics and mechanisms
Georgijs Bakradze
Dissertation an der Universität Stuttgart Bericht Nr. 238 November 2011
Initial oxidation of zirconium:
oxide-film growth kinetics and mechanisms
von der Fakultät Chemie der Universität Stuttgart
zur Erlangung der Würde eines Doktors der Naturwissenschaften (Dr. rer. nat.)
genehmigte Abhandlung
vorgelegt von
Georgijs Bakradze
aus Riga/Lettland
Hauptberichter: Prof. Dr. Ir. E. J. Mittemeijer
Mitberichter: Prof. Dr. J. Bill
Prüfungsvorsitzender: Prof. Dr. E. Roduner
Tag der Einreichung: 12.09.2011
Tag der mündlichen Prüfung: 17.11.2011
MAX-PLANCK-INSTITUT FÜR INTELLIGENTE SYSTEME
(ehemals MAX-PLANCK-INSTITUT FÜR METALLFORSCHUNG)
INSTITUT FÜR MATERIALWISSENSCHAFT DER UNIVERSITÄT STUTTGART
Stuttgart 2011
Wovon man nicht sprechen kann,
darüber muss man schweigen.1
L. J. J. Wit tgenstein (1889-1951)
1 (Germ.) What we cannot speak about we must pass over in silence.
Contents
Contents .......................................... .......................................................... 5
1. General introduction ........................... ................................................. 9
1.1 Initial oxidation of metals ....................... ...................................................... 9
1.2 Focus of the thesis ............................... ...................................................... 12
1.3 Zirconium and zirconium oxide (zirconia) .......... ...................................... 12
1.3.1 Zirconium ............................................................................................. 12
1.3.2 Zirconium oxide (zirconia) .................................................................... 14
1.4 Short overview on zirconium oxidation studies ..... ................................. 15
1.5 Methods of characterization ....................... ............................................... 17
1.5.1 Angle-resolved X-ray photoelectron spectroscopy ............................... 18
1.5.2 Real-time in-situ spectroscopic ellipsometry ........................................ 20
1.5.3 Scanning tunneling microscopy ........................................................... 21
1.5.4 Time-of-flight secondary ion mass-spectrometry ................................. 23
References ........................................ ................................................................. 25
2. The different initial oxidation kinetics of Zr(0 001) and Zr(10-10) surfaces .......................................... .........................................................29
2.1 Introduction ...................................... ........................................................... 29
2.2 Experimental ...................................... ......................................................... 31
2.3 Data evaluation ................................... ........................................................ 33
2.3.1 AR-XPS data ....................................................................................... 33
2.3.2 RISE data ............................................................................................ 36
2.4 Results and discussion ............................ .................................................. 38
2.4.1 Oxide-film constitution.......................................................................... 38
2.4.2 Oxide film thickness: comparison of AR-XPS and RISE analyses ....... 38
2.4.3 Oxide-film growth kinetics .................................................................... 40
2.5 Conclusions ....................................... ......................................................... 43
References ........................................ ................................................................. 44
3. Valence-band and chemical-state analyses of Zr a nd O in thermally-grown thin zirconium-oxide films: an XPS study ..............47
3.1 Introduction ...................................... ........................................................... 47
3.2 Experimental procedure and spectra evaluation ..... ................................ 49
3.3 Results and discussion ............................ .................................................. 53
3.3.1 The oxide-film valence band spectra ................................................... 53
3.3.2 The local chemical states of O and Zr in the oxide films ...................... 56
3.4 Conclusions ....................................... ......................................................... 62
References ........................................ ................................................................. 62
4. An STM study of the initial oxidation of single- crystalline Zr surfaces .......................................... .........................................................65
4.1 Introduction ...................................... ........................................................... 65
4.2 Experimental ...................................... ......................................................... 65
4.3 Results and discussion ............................ .................................................. 67
4.3.1 Oxide-film microstructure at T = 300-450 K ......................................... 69
4.3.2 Evolution of the oxide microstructure at 450 K ..................................... 74
4.4 Conclusions ....................................... ......................................................... 78
References ........................................ ................................................................. 78
5. Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation, as-derived from 18O-tracer experiments ....................................... .....................................................83
5.1 Introduction ...................................... ........................................................... 83
5.2 Experimental ...................................... ......................................................... 86
5.2.1 Specimen preparation .......................................................................... 86
5.2.2 Thermal oxidation ................................................................................ 87
5.2.3 In-situ deposition of an Al capping layer .............................................. 89
5.2.4 XPS, ToF-SIMS and HR-TEM analysis................................................ 90
5.3 The oxide-film microstructure ..................... .............................................. 91
5.4 18O-tracer depth distributions: identification of gov erning transport mechanisms ........................................ ............................................................... 93
5.5 Proposed oxidation mechanism ...................... .......................................... 97
5.6 Conclusions ....................................... ......................................................... 99
References ........................................ ............................................................... 100
6. Summary ........................................ ...................................................103
7. Zusammenfassung................................. ..........................................107
List of used abbreviations ........................ ...........................................113
List of publications .............................. ................................................115
Acknowledgements................................... ...........................................117
Chapter 1
General introduction
1.1 Initial oxidation of metals
The chemical interaction of a metal surface with oxygen gas is a typical example of a
heterogeneous chemical reaction, which can be described by the following chemical
equation:
x·Me(s) + y/2·O2(g) ⇄ MexOy(s). (1.1)
The oxidation reaction typically has a large thermodynamic driving force, owing to a
negative standard free-energy change, ∆G. Consequently, most oxidation reactions run
spontaneously in the forward direction under ambient conditions: i.e. they do not require any
activation. As a result, a native oxide film inevitably forms on most metal surfaces under
ambient conditions. However, the sign and magnitude of ∆G only states the feasibility (i.e.
the thermodynamic driving force) for the oxidation reaction. It does not bear any information
on the rate of oxide formation, as governed by the reaction kinetics at the phase boundaries
(i.e. the gas/oxide and oxide/metal interface), as well as by the transport rate of the reactants
(metal, oxygen and electron species) through the initial oxide film [1]. The native oxide layer
on a metal surface typically acts as a diffusion barrier between the reactants, thereby
inhibiting further oxidation of the underlying metallic compound. In practice, this knowledge
is utilized to enhance the corrosion resistance of metallic coating systems under operation at
higher temperatures by the deliberate (pre-)formation of a protective (i.e. adherent and stable)
α-Al2O3 or α-Cr2O3 overlayer [2-3].
The thin surficial oxide on a metal has a direct influence on the chemical and physical
properties of metallic components (e.g. corrosion resistance, adhesion activity, thermal
stability, catalytic properties, tribologic properties, wear resistance, electrical properties etc.),
as applied in state-of-the art technologies, such as optical, adhesive and corrosion-resistance
coating systems, functionalized surfaces, microelectronics, heterogeneous catalysis and
sensor devices [4]. Clearly, most of the above-mentioned properties are determined by the
microstructure (e.g. thickness, morphology, chemical constitution, crystallographic and
defect structure) of the surficial oxide. In order to control the oxide-film microstructure and,
thereby, the material's properties during fabrication and subsequent operation, a profound
10 Chapter 1
knowledge on the relationships between the oxide microstructure and the oxidation
conditions (e.g. temperature, oxygen partial pressure, surface purity) is required. In addition,
fundamental understanding of the transport phenomena in thin growing oxide layers is
needed to improve e.g. the corrosion resistance of metal and alloy surfaces.
Up to date, the growth kinetics, the microstructural evolution, as well as the
mechanical and transport properties, of oxide layers grown on metals and alloys by thermal
oxidation have been investigated mainly at elevated oxidation temperatures (say at T > 500
K) [5]. At such elevated temperatures, solid-state transport through the growing oxide film
(under influence of chemical potential gradients) becomes pronounced and, consequently,
relatively thick (typically up to several hundreds of micrometers) oxide scales develop. The
microstructure of these oxide scales can be well-characterized by conventional analysis
techniques, such as gravimetry, interferometry, light or electron microscopy and XRD
analysis. Contrarily, our knowledge on the thermodynamics and kinetics of the oxidation
process at low temperatures (of, say, below 500 K) is far from complete and still suffers from
the lack of reliable quantitative experimental data. This can be mainly attributed to the fact
that the oxide films developing on metallic surfaces at low temperatures are typically only
very thin (thickness < 10 nm). Delicate and expensive ultra-high vacuum (UHV) systems for
surface preparation, controlled oxidation and surface analysis are mandatory to process and
characterize the thin oxide-film microstructure in-situ. Additional experimental difficulties
arise due to elaborate in-vacuo surface-preparation procedures. Indeed, first studies on the
controlled oxidation of metals at low temperatures (under well-defined conditions) were
reported in parallel with the fast development of UHV technologies, starting from the late
1960s. Already long before it was recognized that the electrochemical and mechanical
properties of metal and alloy surfaces depend, to a great extent, on the crystallographic,
defect and electronic structure of the native oxide.
Wagner was among the first to develop a theoretical description for the high-
temperature oxidation of metals and alloys on the basis of the thermally-activated diffusion of
reactant species through the developing oxide scale under influence of the acting chemical
potential gradients [6]. His theoretical framework in turn promoted the postulation of theories
on the low-temperature oxidation behaviour of metal surfaces (i.e. in the absence of thermally
activated diffusion), as governed by the surface field setup by chemisorbed oxygen species on
the oxidizing surface (cf. Mott-Cabrera [7], Fromhold [2] and Fromm [3]). Despite these
important advances in the field of low-temperature oxidation, the processes that occur and the
changes that take place at the metal surface, in the developing oxide film, and at the
General introduction 11
metal/oxide and oxide/gas interfaces during the initial and subsequent stages of oxide-film
growth are still only partially understood.
Clearly, the initial oxidation of a bare metal surface is a complex physicochemical
process, which involves a series of competing and overlapping processes (see Fig. 1.1), such
as [2]: (i) oxygen molecule impingement on the metal surface and (ii ) subsequent
phys isorpt ion (i.e. O2(g) ⇄ O2(phys) and/or O2(g) ⇄ 2O(phys)), (iii ) chemisorpt ion of oxygen
species (i.e. O(phys) + e ⇄ O–(chem) and/or O2(phys) + e ⇄ O2
–(chem)); (iv) place exchange between
atoms in the metal subsurface and chemisorbed oxygen species, incorporat ion of anions
and cations into the growing oxide film and (v) oxide nucleat ion and continued oxide-film
growth.
Fig. 1.1. Schematic illustration of some competing physical and chemical processes for the initial
stages of dry thermal oxidation of a bare metallic substrate as described by Eq. 1.1: (i) impingement
and (ii ) physisorpt ion of oxygen molecules on the surface, (iii ) dissociative oxygen
chemisorpt ion, (iv) incorporat ion of anions and cations into the growing oxide film. Note: only
transport processes in the direction parallel to the surface normal are shown.
Oxide nucleation, lateral growth and coalescence of oxide nuclei, followed by
continued oxide-film growth (thickening), generally involve both volume and short-circuit
transport of the reactants (metal cations, oxygen anions and their vacancies, as well as
electrons and their holes) through the developing oxide film. The aforementioned theoretical
descriptions of the oxidation process typically assume that the initial oxide layer covers the
metal surface uniformly and that further oxide-film growth proceeds in a uniform way (i.e. by
a layer-by-layer thickening). Evidently, these oversimplified assumptions are serious
deficiencies in the theoretical treatments [2-3, 7] of, particularly, the initial oxidation stages
[8]. The situation becomes even more complicated by the fact that the coupled (charged)
currents of ions/vacancies and electrons typically act in the presence of steep gradients in the
defect concentrations, intrinsic stress-level and/or the electric-field across the developing
oxide film. Finally, it is noted that the oxidation behavior not only depends on the developing
12 Chapter 1
oxide-film microstructure, but also on e.g. the crystallographic orientation of the parent metal
substrate, the presence of any (segregated) impurities, and in some cases on the rate of
dissolution of oxygen into the metal (especially for elevated oxidation temperatures and low
oxygen partial pressures).
1.2 Focus of the thesis
The present PhD work addresses the growth kinetics (Chapters 2), chemical composition
(Chapters 2, 3), morphology (Chapter 4) and transport properties (Chapter 5) of zirconium-
oxide films, as grown by the dry, thermal oxidation of single-crystalline Zr surfaces at low
oxidation temperatures (for details see Section 1.4). To this end, oxide films with thicknesses
in the range of 1-10 nm were grown on bare (i.e. without a native oxide) Zr(0001) and Zr(10
1 0) surfaces by controlled exposure to O2(g) in the temperature range of T = 300-450 K at an
oxygen partial pressure of pO2 = 1×10-4 Pa in an especially-designed UHV system.
This study reveals, for the first time, the effect of the metal substrate orientation on
the low-temperature growth kinetics and microstructural evolution of the initial oxide
overgrowths on Zr single-crystalline surfaces with basal and prism orientations. Furthermore,
two-stage tracer oxidation experiments using 16O and 18O isotopes were performed to reveal
the governing atomic transport mechanism(s) in the thin oxide films developing on both Zr
single-crystalline surfaces at 450 K.
1.3 Zirconium and zirconium oxide (zirconia)
1.3.1 Zirconium
Zirconium (Z = 40, [Kr]4d25s2) belongs to the IV group of the periodic table and, together
with titanium and hafnium, forms a small IVb subgroup of metals, which all to a large extent
exhibit very similar physical and chemical properties.
Metallic Zr is extracted from minerals, such as zircon (ZrSiO4) and the natural form
of zirconium oxide – baddeleyite (ZrO2). Typical natural impurities in zirconium are oxygen,
iron and hafnium. Zirconium has a silver-grayish colour with a characteristic metallic blister.
Under atmospheric pressure zirconium has two allotropic modifications: the hexagonal α-
phase with a Mg-type crystal structure (c/a = 1.59, a = 3.2312 Å, stable at T < 1139 K) and
the BCC cubic β-phase with a α-Fe-type crystal structure (stable at T > 1139 K) [9]. The
relatively low α-to-β transition temperature in combination with the high chemical reactivity
General introduction 13
of metallic Zr make it very difficult to prepare pure Zr single crystals. Pure Zr is ductile and
can be readily processed by conventional metal forming techniques like forging, cold-rolling
or drawing and welding under inert atmosphere [10]. However, the presence of oxygen,
nitrogen, carbon and hydrogen strongly affects the mechanical and chemical properties of Zr
[5].
Fig. 1.2. The Zr-O phase-diagram [11]. Note an extensive region of solid solutions of O in α-Zr even
at low temperatures.
Zirconium and its alloys are stable in water, air and acids (except hydrofluoric and
concentrated sulphuric acid) at room temperature. Due to its good corrosion resistance at low
temperatures, Zr and its alloys are increasingly being used in applications for chemical
processing equipment, oceanic instruments and marine hardware [12-13]. Some Zr alloys are
used in biomedical applications (e.g. for hip joint implants) due to their wear resistance and
compatibility with biological tissues [4].
Zr has a high affinity for oxygen (∆G = –1037 kJ/mol at standard conditions [14]) and
actively absorbs gases (like O2 and H2) [5] at elevated temperatures T > 500 K. In addition,
the O solubility in α-zirconium is as high as 30 at.% even at moderate temperatures (see Fig.
1.2). Consequently, Zr is very reactive to the residual gases in vacuum (especially to CO
14 Chapter 1
which dissolves readily). These properties insure the use of Zr-alloys as a getter material in a
new generation of getter pumps for UHV applications [15].
For many decades, zirconium finds his primary use in the nuclear industry1 for in-
reactor components, especially in the cladding of the fuel rods, due to zirconium's low
neutron scattering cross-section and passivating oxidation behavior at low operating
temperatures < 500-600 K [12-13]. The integrity of the passivating oxide film on the cladding
elements is crucial in order to assure the safe reactor operation. Hence thorough information
is required concerning the structure, composition, kinetics and mechanism of the growth of
oxide films on Zr and its alloys.
1.3.2 Zirconium oxide (zirconia)
Zirconium-oxide consists in three crystalline forms: the monoclinic α-phase at low
temperatures, the tetragonal β-phase above 1400 K and the cubic γ-phase above 2600 K [14].
Crystalline zirconium-oxide generally exists in the monoclinic phase at room temperature.
However, the cubic phase can be stabilized at lower temperatures by the enhanced formation
of vacancies in the anion sublattice, as induced by the addition of e.g. ZrN, CaO, Y2O3, MgO
[17]. The existence of a metastable, amorphous Zr-oxide phase has also been reported for the
thermal oxidation of Zr substrates for short oxidation times at low temperatures (up to about
573 K) (see Ref. [18]) and can be rationalized on a thermodynamic basis [19]. In
nanomaterials, the tetragonal ZrO2 phase can be stabilized due to its lower interfacial energy
(as compared to monoclinic phase) [20-21].
Zirconia finds diverse applications in jewellery, microelectronics, fuel cells and
oxygen sensors [22-23]. Due to its high melting point (2953 K), chemical durability and high
hardness, zirconium dioxide has long been used for refractory containers and as an abrasive
medium. Zirconia-based materials have similar thermal expansion coefficients as some high-
temperature-resistant metallic alloys and are therefore widely applied in thermal barrier
coating systems for jet-engines [24]. In recent years, zirconia has also been considered as a
promising candidate for tunnel barrier applications in next-generation metal-oxide-
semiconductor field-effect transistor (MOS-FET) devices owing to its high dielectric constant
(κ ≈ 25), its large conduction band offset with Si, as well as its predicted stability with respect
to solid-state reactions with Si [25-26].
1 Unlike other metal alloys, zirconium alloys applied in nuclear applications are rather pure (>97-98% Zr) and can almost be considered as single-component systems [4]. However, a strong influence of foreign elements on the diffusion in Zr has been reported in [16].
General introduction 15
1.4 Short overview on zirconium oxidation studies
Up to date, the oxidation behaviour of zirconium and its alloys has been extensively
investigated under various oxidation conditions (e.g. oxidation temperature and time, partial
oxygen pressure, oxidizing atmospheres; cf. Refs. [27-32]) and by using a broad range of
(surface-)analytical techniques. Particularly, the effect of the oxidation temperature and
partial oxygen pressure on the oxide-film growth kinetics and the developing oxide-film
microstructure has been studied thoroughly, but unfortunately unsystematically. Many of the
earlier oxidation studies on Zr pertain to the high-temperature oxidation regime and often
suffer from the fact that only a single analytical technique has been applied to characterize
the developing oxide microstructure.
Several theoretical and experimental studies have focused on the very initial stages of
oxygen interaction with the bare Zr(0001) surface (i.e. for oxygen exposures < 50 L) [33-40].
It was found that, at 90 K, 293 K and 473 K the adsorbed oxygen atoms preferentially occupy
octahedral subsurface sites in the Zr(0001) substrate for low oxygen coverages (< 0.5 ML)
[33, 40]. The oxygen atoms may penetrate into the subsurface octahedral sites, thereby
leaving free sites on the top surface layer and thus enabling further adsorption and
incorporation of oxygen. Initial exposure of the bare Zr(0001) surface up to about 1
Langmuir (see footnote 1) of O2(g) and subsequent flash-annealing to about 473 K yields a
well-ordered (2×2) O-adsorbate overlayer structure by low-energy electron diffraction
(LEED) [38, 41-43]. On the prism plane the stable surface phase is Zr(10 10)-O(2×4) [44].
The oxidation kinetics of the Zr(0001) surface for more prolonged oxygen exposures
have been studied by Auger electron spectroscopy [40]. It was found that oxide-layer growth
proceeds by a layer-by-layer growth mechanism at 90 K, whereas at T ≥ 293 K oxidation
occurs by initial formation of oxide islands, which predominantly grow inwardly into the
metal substrate. Oxidation of the Zr(0001) surface at room temperature results in the
formation of a ultra-thin disordered (amorphous), overall non-stoichiometric oxide film [38,
41]. Investigations by X-ray Photoelectron Spectroscopy (XPS) have revealed that the oxide
film is constituted of a non-stoichiometric suboxidic layer (ZrOx with x < 2, which contains
several Zrδ+ valence states with δ < 4) at the interface with the parent Zr metal and a (near-
)stoichiometric ZrO2 oxide adjacent to the surface [4, 30-31, 45].
1 1 Langmuir = 1×10-6 Torr·s.
16 Chapter 1
Only a few studies have been reported on the oxidation of single-crystalline Zr
surfaces other than Zr(0001). Noteworthy, the oxidation behaviour of the Zr(10 10) prism
plane is of particular industrial interest, because the cold-rolled Zr samples are textured with
{10 1 0}-planes parallel to the surface [34, 44]. Already in the early 50s, it was found that Zr
(10 11) and Zr(1120) surfaces have a lower oxidation rates than Zr(0001) and Zr(10 10)
surfaces [34]. For the oxidation of Zr at 773 K in steam [46], the oxidation rate of Zr surfaces
of different crystallographic orientation increased in the order: (10 12)<(1120)<(10 10)<
(10 11)<(0001). However, these findings are in contradiction with the results reported in Ref
[47], stating that the oxidation rate of polycrystalline Zr reached its minimum when the c-axis
was normal to the surface plane, whereas the maximum oxidation rate is attained when the c-
axis is inclined to the surface plane of the sample by 20°. The kinetics of oxidation of
Zr(0001) and Zr(10 10)(1×4) surfaces were investigated by the Norton group [38, 43-44] in
the 90-473 K range, however the oxygen exposures in all experiments were very low and the
oxidation kinetics of both substrates has not been done.
Current scientific and technological interest concerns primarily the successive stages
of the oxidation process associated with: (i) formation of a closed oxide film on the Zr metal
surface and (ii ) subsequent thickening of the thin (thickness < 100 nm) oxide-film by
transport of reactant species (i.e. cations, anions and their vacancies, as well as electrons and
electron holes). Surprisingly, the initial oxidation of Zr at intermediate temperatures (i.e. T =
300-600 K), where oxide films can be grown with controllable thicknesses in the nanometer
range, have been largely unaddressed up to date [34, 44]. Also comprehensive knowledge on
the developing oxide-film microstructure and its atomic transport mechanisms during the
oxidation process at intermediate temperatures is still lacking. This is also evidenced by the
contradictory statements in the scientific literature on the oxidation mechanisms of Zr at low
and intermediate temperatures. For example, it was postulated in Ref. [48] that the growth of
an amorphous Zr-oxide film on Zr at low temperatures proceeds by the coupled currents of
cations and electrons across the oxide-film under influence of the surface-charge field (as
setup by chemisorbed oxygen species at the oxidizing surface). Also in Ref. [31], the
transport of Zr4+ ions to the film surface was proposed as the rate-limiting step for the
oxidation process. However, according to Ref. [49], the rate-limiting step in the oxidation
process is either the O2 dissociation rate or the O transport rate through the growing ZrO2
film. There are also contradictions regarding the rate-controlling step for oxygen dissolution
into the Zr metal at elevated temperatures. As evidenced from O18-isotope tracer oxidation
General introduction 17
experiments, the inward migration of oxygen along oxide grain boundaries (GBs) plays a
dominant role in the growth of thick, polycrystalline oxide scales on Zr [50-52] and its alloys
[53] at elevated temperatures. According to Ref. [54], the O dissolution rate is governed by
the transport rate of O through the interfacial suboxide layer and not by the rate of oxygen
dissolution in the metal at the suboxide/metal interface (as postulated in Ref. [33]). Upon
oxidation of zirconium alloys at higher oxidation temperatures (T > 600 K) oxygen short-
circuit transport along the crystallite GBs is commonly considered as the predominating
transport mechanism [29, 51].
Recently, an investigation on the initial oxidation of weakly textured, polycrystalline
Zr surfaces in the temperature range of T = 304-573 K was carried out at Max Planck
Institute for Metals Research (Stuttgart). In that study the native oxide on the Zr surface was
removed by Ar+ sputter-cleaning (SC) under UHV conditions [55-56], but without
performing a final in-vacuo annealing step prior to oxidation (to restore the distorted
crystallography at the ion-bombarded surface). Consequently, the surfaces were not in the
crystallographically well-defined state and hence the effect of the substrate orientation on the
oxidation process could not be established.1
The present study, for the first time, presents a direct comparison of the initial
oxidation of single-crystalline Zr surfaces with basal and prism orientations (i.e. Zr(0001) and
Zr(101 0), respectively) performed under the same, well-controlled experimental conditions
in the temperature range of 300-450 K and at pO2 = 1×10-4 Pa. To this end, well-defined,
single-crystalline Zr surfaces were prepared by an elaborate cyclic treatment of alternating
Ar+ SC and in-vacuo annealing steps under UHV conditions. The relationships between the
oxidation kinetics, the developing microstructure and the crystallographic orientation of the
parent metal substrate were established by application of various (surface-)analytical
techniques (see Section 1.5). Furthermore, two-stage tracer oxidation experiments using 16O
and 18O isotopes were successfully employed to reveal the atomic transport mechanisms in
the very thin (thickness < 10 nm) oxide films, as grown on the single-crystalline Zr surfaces
by thermal oxidation at 450 K and pO2 = 1×10-4 Pa.
1.5 Methods of characterization
Thorough characterization of the growth kinetics and microstructural evolution of thin (< 10
nm) oxide overgrowths on bare Zr substrates requires a complex experimental approach by
1 To see the effect of SC on the surface morphology see Fig.4.1.
18 Chapter 1
various in-situ, preferably non-destructive, surface-sensitive analytical techniques. In the
present thesis, a combined experimental approach by in-situ angle-resolved X-ray
photoelectron spectroscopy (AR-XPS), real-time in-situ spectroscopic ellipsometry (RISE),
in-situ scanning tunnelling microscopy (STM), ex-situ High-resolution Transmission
Electron Microscopy (HR-TEM) and ex-situ time-of-flight secondary mass-spectrometry
(ToF-SIMS) has been applied to study the microstructural evolution and growth kinetics, as
well as transport mechanism, during growth of thin oxide overgrowth on Zr single crystals.
1.5.1 Angle-resolved X-ray photoelectron spectroscopy
XPS belongs to the vast family of electron spectroscopic techniques and is widely used for
chemical characterization of near-surface regions in solid compounds. The working principle
of XPS is based on the measurement of the energy spectrum of electrons emitted due to the
outer photoeffect, i.e. due to the high-energy photon-induced photoemission of electrons from
characteristic energy levels of the atoms in the sample [57].
Fig. 1.3. Schematic view of the AR-XPS principle. The sample is exposed to a flux of monochromatic
photon radiation with energy hν. The kinetic energies, Ek, of the ejected electrons are recorded at
different detection angles θ, thus obtaining depth-resolved information of the investigated chemical
species. Adopted from Refs. [57-58].
In an AR-XPS set-up the kinetic energy of the emitted photoelectrons, Ek, can be
recorded at various detection angles θ (with respect to the sample surface normal), as
illustrated in Fig. 1.3. The sensitivity of XPS to the surface composition of the sample
originates from the fact that emitted photoelectrons with kinetic energies below 500 eV are
General introduction 19
easily inelastically scattered in the solid. For example, for the electrons with energies of 20-
200 eV in inorganic solids, the inelastic means free paths (IMFP) is generally less than 10 Å.
Only the photoelectrons, which are emitted in the near-surface region of the solid, have a
finite probability to escape from the solid into vacuum and reach the detector without kinetic
energy (KE) loss [59]. The measured KE of these unscattered and elastically scattered
electrons can be easily be converted into the respective binding energies (BE), Eb, using the
following equation:
Eb = hν – Ek – Φ, (1.2)
where Φ denotes the work function of the spectrometer (for conductive samples it equals the
work function of the sample). If the effects of elastic scattering of the traversing
photoelectrons are neglected, 95% of the unscattered and elastically scattered photoelectrons
(for θ = 0) will originate from depths up to 3 times the IMFP below the sample surface. The
effect of elastic scattering can be accounted for by using the so-called effective attenuation
length (EAL, symbol λeff) instead of the IMFP. The information depth then varies with the
photoelectron detection angle according to: 3λeff×cos θ. Angle-resolved XPS measurements
thus principally allow the investigation of the depth distribution of various chemical species
in very thin (thickness < 6 nm) oxide films. In Chapters 2 and 3 of this thesis, AR-XPS
analysis of the bare and oxidized metal substrates has been applied to determine the
thickness, chemical composition and constitution of the oxide overgrowths on Zr.
One of the most challenging problems in the quantitative processing of the recorded
XPS data is the deconvolution of a measured spectrum into its individual spectral
contributions as originating from different the chemical states of the elements in the solid
[59]. The accuracy of quantitative AR-XPS analysis strongly depends on the methods used to
reconstruct and subtract the superimposed background intensity of inelastically scattered
electrons to the measured spectrum [58]. For a measured binding energy region of a core-
level photoelectron line with a single chemical-state contribution, the inelastic background
can be easily removed by subtracting some arbitrary background, such as a linear or a
Shirley-type background [58]. However, if the recorded XPS core-level spectrum contains
several (overlapping) peaks (as is typically the case for a recorded XPS spectrum of an
oxidized metal, which consists of at least one metallic and one or more oxidic main peaks;
see Fig. 2.3), each main peak provides its own background of inelastically scattered electrons
and, consequently, the simplified methods for the background subtraction can cause
significant errors in the determination of e.g. the oxide-film thickness and composition. In
20 Chapter 1
this work an advanced procedure for the evaluation of the Zr 3d XPS spectra of zirconium-
oxide has been applied, which is based on the reconstructions of the metallic and oxidic Zr 3d
spectral contributions by convolution of physically realistic functions for the X-ray energy
distribution, the core-level main peak, the cross-sections for intrinsic and extrinsic excitations
and instrumental broadening (see Chapter 2) [60].
The local chemical state of an element in a solid can also be accessed by XPS on the
basis of the so-called Auger parameter (AP) [61-63], which is defined as the sum of the
kinetic energy of the most prominent and sharp core-level-like Auger transition and the
binding energy of the most prominent and sharp core-level photoelectron line of an element
in the solid. The AP value provide a unique direct measure of the electronic polarizability of
the local chemical environment around the photoemitting atom and is, thus, sensitive to
structural changes in the nearest coordination spheres of the constituent atoms [64-66]. Also
the structure of the valence band (VB), as measured by XPS, is very sensitive to
microstructural changes in elemental solids and compounds, because valence electrons are
directly involved in chemical bonding, which is not the case for the core-level photoelectrons.
Therefore, the evolution of the oxide-film microstructure was also derived from the measured
changes in shape (i.e. fine-structure) of the resolved oxide-film VB spectra, as well as of the
accompanied AP shifts of the Zr and O ions in the oxide film with increasing oxidation
temperature (see Chapter 3).
1.5.2 Real-time in-situ spectroscopic ellipsometry
Ellipsometry offers a unique opportunity to perform in-situ dynamic (real-time) and non-
destructive optical analysis of a thin oxide film developing on a metal surface upon oxidation.
To this end, linearly polarized light is irradiated onto a sample at the Brewster angle (or close
to that), and the optical constants and film thickness of the sample can be extracted from the
measured change in the polarization state of light upon reflection or transmission [67]. Figure
1.4 illustrates the basic measurement principle of ellipsometry. Although all optical
techniques are typically inherently diffraction limited, ellipsometry exploits both phase
information and the polarization state of light, and thus can achieve Ångström resolution and
high precision in measurement [67-68].
In ellipsometry set-up, the polarization states of incident and reflected light are
described by the coordinates of two orthogonal p- and s-polarizations (i.e. in the incidence
plane and perpendicular to that, cf. Fig. 1.4). After reflection of a linearly polarized light from
the sample surface the light wave is generally elliptically polarized, i.e. the tip of the electric
General introduction 21
field vector Er = Erp + Ers describes an ellipse in any plane normal to the direction of
propagation. The ellipsometric parameters (ψ, ∆) are defined as the ratio of the amplitudes of
reflection coefficients, rp and rs, for p- and s-polarizations [67-68]:
==×=
is
rs
ip
rp
s
pi)exp()tan(E
E
E
E
r
r∆ψρ . (1.3)
Therefore, ψ represents the angle determined from the amplitude ratio between reflected p-
and s-polarizations, while ∆ expresses the phase difference between reflected p- and s-
polarizations (cf. Fig. 1.4).
Fig. 1.4. Illustration of the measurement principle of ellipsometry. The sample surface is illuminated
with a linearly polarized light. The difference in the polarization state between the incident (linearly
polarized) and reflected light (typically elliptically polarized) beam is determined. Adopted from Ref.
[67].
For the current ellipsometric investigations on the oxidation of Zr, the courses of ψ
and ∆ as a function of oxidation time, t, were recorded simultaneously over a wavelength
range λ = 250-900 nm (as generated by a Xe light source). Conclusive information on
particularly the growth kinetics of the developing oxide films was obtained by fitting the
calculated data of ψ(λ,t) and ∆(λ,t) to the measured data ψ(λ,t) and ∆(λ,t) by adopting a
physically realistic model for the evolving substrate/oxide-film system [67-68] and
employing realistic values for the optical constants of the substrate and the thin film. The
applied optical model description for the evolving substrate/oxide-film system was
constructed on the basis of the pre-knowledge on the chemical constitution of the oxide films,
as obtained by AR-XPS.
1.5.3 Scanning tunnelling microscopy
The STM was first introduced in 1981 and ever since has become an invaluable tool in many
surface science studies [69-70]. STM enables the mapping of the topography of conducting
22 Chapter 1
surfaces with a resolution down to the atomic scale. The working principle of STM is based
on the quantum-mechanical tunnelling of electrons through the vacuum barrier separating the
tip and sample: e.g. electrons in states within energy e·Vt above the Fermi level on the
negative side tunnel into empty states within energy e·Vt above the Fermi level on the
positive side, as determined by the sign of the applied sample bias voltage, Vt [69].
Fig. 1.5. Scheme showing the measurement principle of STM: The bias voltage Vt is applied between
tip and sample (separated by the tunneling gap d) and the resulting tunneling current It is measured. A
feedback loop controls the contraction/extension of the z-piezo in order to maintain a constant It. The
lateral movement of the tip and acquisition of the regulated variation of z (to keep It constant) gives a
topographic image of the surface. Adopted from Ref. [71].
The basic operational principle of an STM is schematically sketched in Fig. 1.5. If a
bias voltage Vt is applied between the sample and the tip, and if the tip is sufficiently close to
the sample (within several tens of Ångströms), electrons can tunnel between the tip and the
sample, thus producing a measurable tunnel current, I t, of the order of 1 nA. Since the
tunnelling current I t is a monotonous function of the tip-to-sample distance d, namely, I t ∝
exp(-d), it will have a well-defined value for a given reference current, I t [69, 71].
General introduction 23
In the so-called constant-current or topographic mode, a constant tunnelling current is
maintained while scanning the sample surface by continuously recording and adjusting the
distance between the tip and the sample surface by a feedback loop [71]. The feedback loop
is adjusted in the vertical z-direction by regulating the applied voltage on a z-piezoelectric
tube to keep the tunnel current constant. Similarly, lateral scanning of the STM tip (along the
x- and y-directions) is realized by applying voltages to the side electrodes of the piezoelectric
tube [70]. The resolutions of the STM are mainly determined by: (i) the mechanical stability
of the instrument and (ii ) the electronic structure of the tip.
One of the main disadvantages of STM is its inability to image surfaces of insulating
(bulk) materials, since the STM operation relies on the electron tunnelling between the
surface and the tip. However, imaging of very thin insulator layers on conducting substrates
is still possible [72]. Another disadvantage of STM is that the image quality typically
depends not only on the quality of the tip, but also on its electronic structure. Moreover,
recorded images typically do not represent merely topography of the scanned area, but also
contain information on electron density distribution over the scanned area. Nevertheless,
sufficiently large scanned areas can be regarded as purely topographic images.
In this project in-situ STM has been applied to investigate the topography of very thin
(thickness < 10 nm) oxide films, as grown on the Zr surfaces by thermal oxidation (see
Chapter 4 for details).
1.5.4 Time-of-flight secondary ion mass-spectrometry
Time-of-flight secondary ion mass-spectrometry (ToF-SIMS) is a mass-sensitive surface-
analysis technique for surface studies of solids, which dates back to the late 1960s and 1970s.
The method is based on the bombardment of the sample surface with high energy primary
ions to desorb and ionize species (i.e. secondary ions) from a sample surface [73]. The
resulting secondary ions are accelerated into a mass spectrometer and are mass-analyzed by
measuring their time-of-flight from the sample surface to the detector.
Figure 1.6 illustrates the basic measurement principle of ToF-SIMS. The sample
surface is bombarded by a pulsing primary ion beam with a typical energy of a several
kiloelectron-volts. The primary ions interact with the surface of the investigated material,
resulting in the emission of neutral fragments, positively and negatively charged secondary
ions from the analyzed sample surface. To filter particles of only one polarity they are
directed to the mass analyzer, where they are accelerated to a given kinetic energy and mass-
analyzed by time-of-flight measurement. Finally they are counted in an ion detector (e.g.
24 Chapter 1
electron multiplier, a Faraday cup or a channel plate). ToF-SIMS principally allows a
simultaneous recording of emitted secondary ions (including possibility of differentiation of
particular isotopes) over a wide mass range with very high mass sensitivity (m/∆m > 9000 at
m/q = 29), ion transmission efficiencies and good lateral resolution (down to 50 nm).
Fig. 1.6. Schematic diagram showing the principle of ToF-SIMS: the surface of the sample is
sputtered with a primary beam of low-energy ions. The generated secondary ions are directed to the
detector and mass-analyzed by time-of-flight. Adopted from Ref. [73].
The main advantages of ToF-SIMS is the possibility to detect practically all the
elements of the periodic system (including hydrogen; the noble gases are difficult to detect as
they have high ionization energies) and very high sensitivities (in ppb range). The main
disadvantage of ToF-SIMS is related to the matrix effect, which makes a qualitative analysis
of the measurements complicated [73]. The matrix effect is a general term used to describe
the effects of physical and/or chemical nature which cause changes in the Auger electron,
photoelectron, secondary ion yield, scattered ion intensity, the energy or shape of the signal
of an element in a certain chemical environment as compared to these quantities in a pure
element [74]. These matrix effects can be accounted for by using standards for each element.
However, the extreme sensitivity of the sputtered ion yields to variations in the matrix
composition of the sample imposes severe restrictions on the nature of acceptable standards
[73]. If standards are not available (as it was the case it this project) or/and one is only
General introduction 25
interested in the isotopes distribution of the same element, this problem can be overcome by
analyzing the isotope fraction profiles (see Chapter 5 for details).
In the present study, ToF-SIMS has been used to resolve the depth-distribution of 18O-
tracer atoms in very thin (thickness < 10 nm) oxide films as grown on the Zr(0001) and Zr(10
1 0) surfaces by two-stage oxidation experiments at 450 K and pO2 = 1×10-4 Pa. As such,
detailed knowledge is obtained on the governing transport processes in the developing oxide
films upon oxidation (see Chapter 5).
References
[1] K. R. Lawless, Rep. Prog. Phys. 37 (1974) 231.
[2] A. T. Fromhold, Theory of Metal Oxidation, North-Holland, Amsterdam (1976).
[3] E. Fromm, Kinetics of Metal-Gas Interactions at Low-Temperatures, Springer, Berlin
(1998).
[4] N. Stojilovic, E. T. Bender, and R. D. Ramsier, Prog. Surf. Sci. 78 (2005) 101.
[5] P. Kofstad, High-Temperature Oxidation of Metals, John Wiley & Sons, New York,
London, Sydney (1966).
[6] C. Wagner, Z. Phys. Chem. B 21 (1933) 25.
[7] N. Cabrera and N. F. Mott, Rep. Prog. Phys. 12 (1948) 163.
[8] G. W. Zhou, Appl. Phys. Lett. 94 (2009) 201905.
[9] Landolt-Börnstein, Group III Condensed Matter, Structure Data of Elements and
Intermetallic Phases, Springer-Verlag, (1971).
[10] B. Lustman and F. Kerze, ed. The Metallurgy of Zirconium (1955) McGraw-Hill
Book Comp.: New York.
[11] J. P. Abriata, J. Garcés, and R. Versaci, Bull. Alloy Phase Diag. 7 (1986) 116.
[12] B. Kammenzind and M. Limbäck, Zirconium in the nuclear industry: 15th
international symposium, ASTM International, West Conshohocken, PA (2009).
[13] A. L. Lowe and G. W. Parry, Zirconium in the nuclear industry: proceedings of the
3rd International Conference, ASTM, Philadelphia (1977).
[14] CRC Handbook of Chemistry and Physics, Internet Version 2005,
http://www.hbcpnetbase.com, ed. by D. R. Lide, CRC Press, Taylor and Francis
Group, Boca Raton, FL (2005).
[15] W. J. Lange, J. Vac. Sci. Technol. 14 (1977) 582.
[16] Y. A. Smirnov, Fiz. Met. Metalloved. 86 (1998) 15.
[17] Science and technology of zirconia, Technomic Pub. Co., (1993).
26 Chapter 1
[18] D. L. Douglass and J. Vanlandu, Acta Metall. Mater. 13 (1965) 1069.
[19] F. Reichel, L. P. H. Jeurgens, and E. J. Mittemeijer, Acta Mater. 56 (2008) 5894.
[20] W. Qin, C. Nam, H. L. Li, and J. A. Szpunar, Acta Mater. 55 (2007) 1695.
[21] R. C. Garvie, J. Phys. Chem.-US 82 (1978) 218.
[22] C. I. Howe, B. Mcenaney, V. D. Scott, and M. G. C. Cox, J. Phys. E 14 (1981) 1308.
[23] R. M. Ormerod, Chem. Soc. Rev. 32 (2003) 17.
[24] C. G. Levi, Curr. Opin. Solid State & Mater. Sci. 8 (2004) 77.
[25] T. Kurniawan, K. Cheong, K. Razak, Z. Lockman, and N. Ahmad, J. Mater. Sci.
(2010) 1.
[26] G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89 (2001) 5243.
[27] J. M. Sanz, A. R. Gonzalezelipe, A. Fernandez, D. Leinen, L. Galan, A. Stampfl, and
A. M. Bradshaw, Surf. Sci. 309 (1994) 848.
[28] J. S. Foord, P. J. Goddard, and R. M. Lambert, Surf. Sci. 94 (1980) 339.
[29] B. Cox, J. Nucl. Mater. 218 (1995) 261.
[30] C. Morant, J. M. Sanz, L. Galan, L. Soriano, and F. Rueda, Surf. Sci. 218 (1989) 331.
[31] P. Sen, D. D. Sarma, R. C. Budhani, K. L. Chopra, and C. N. R. Rao, J. Phys. F 14
(1984) 565.
[32] C. O. d. Gonzalez and E. A. Garcia, Appl. Surf. Sci. 44 (1990) 211.
[33] B. Li, A. R. Allnatt, C. S. Zhang, and P. R. Norton, Surf. Sci. 330 (1995) 276.
[34] H. G. Kim, T. H. Kim, and Y. H. Jeong, J. Nucl. Mater. 306 (2002) 44.
[35] R. A. Ploc, J. Nucl. Mater. 110 (1982) 59.
[36] Y. M. Wang, Y. S. Li, and K. A. R. Mitchell, Surf. Sci. 343 (1995) L1167.
[37] Y. M. Wang, Y. S. Li, and K. A. R. Mitchell, Surf. Sci. 380 (1997) 540.
[38] C. S. Zhang, B. J. Flinn, I. V. Mitchell, and P. R. Norton, Surf. Sci. 245 (1991) 373.
[39] C. S. Zhang, B. J. Flinn, and P. R. Norton, J. Nucl. Mater. 199 (1993) 231.
[40] C. S. Zhang, B. J. Flinn, and P. R. Norton, Surf. Sci. 264 (1992) 1.
[41] K. C. Hui, R. H. Milne, K. A. R. Mitchell, W. T. Moore, and M. Y. Zhou, Solid. State
Commun. 56 (1985) 83.
[42] P. C. Wong and K. A. R. Mitchell, Can. J. Phys. 65 (1987) 464.
[43] Y. M. Wang, Y. S. Li, and K. A. R. Mitchell, Surf. Sci. 342 (1995) 272.
[44] C. S. Zhang, B. Li, and P. R. Norton, Surf. Sci. 313 (1994) 308.
[45] L. Kumar, D. D. Sarma, and S. Krummacher, Appl. Surf. Sci. 32 (1988) 309.
General introduction 27
[46] J. N. Wanklyn, Recent Studies of the Growth and Breakdown of Oxide Films on
Zirconium and Zirconium Alloys, in Corrosion of Zirconium Alloys, W. Anderson,
Editor^Editors. 1964, ASTM. p. 58.
[47] J. P. Pemsler, J. Electrochem. Soc. 105 (1958) 315.
[48] L. P. H. Jeurgens, A. Lyapin, and E. J. Mittemeijer, Acta Mater. 53 (2005) 4871.
[49] K. O. Axelsson, K. E. Keck, and B. Kasemo, Surf. Sci. 164 (1985) 109.
[50] J. A. Davies, B. Domeij, J. P. S. Pringle, and F. Brown, J. Electrochem. Soc. 112
(1965) 675.
[51] S. Chevalier, G. Strehl, J. Favergeon, F. Desserrey, S. Weber, O. Heintz, G.
Borchardt, and J. P. Larpin, Mater. High Temp. 20 (2003) 253.
[52] J. B. Holt and L. Himmel, J. Electrochem. Soc. 116 (1969) 1569.
[53] A. Grandjean and Y. Serruys, J. Nucl. Mater. 273 (1999) 111.
[54] P. E. West and P. M. George, J. Vac. Sci. Technol. A 5 (1987) 1124.
[55] A. Lyapin, L. P. H. Jeurgens, P. C. J. Graat, and E. J. Mittemeijer, J. Appl. Phys. 96
(2004) 7126.
[56] A. Lyapin, L. P. H. Jeurgens, and E. J. Mittemeijer, Acta Mater. 53 (2005) 2925.
[57] S. Hüfner, S. Schmidt, and F. Reinert, Nucl Instrum Meth A 547 (2005) 8.
[58] S. Hüfner, Photoelectron Spectroscopy: Principles and Applications, Springer, Berlin,
Heidelberg, New York (2003).
[59] D. Briggs and J. T. Grant, ed. Surface Analysis by Auger and X-Ray Photoelectron
Spectroscopy (2003) IM Publications and Surface Spectra Limited: Chichester and
Manchester.
[60] A. Lyapin and P. C. J. Graat, Surf. Sci. 552 (2004) 160.
[61] C. D. Wagner, L. H. Gale, and R. H. Raymond, Anal Chem 51 (1979) 466.
[62] C. D. Wagner, Anal. Chem. 44 (1972) 967.
[63] C. D. Wagner and A. Joshi, J. Electron Spectrosc. 47 (1988) 283.
[64] G. Moretti, J. Electron Spectrosc. 95 (1998) 95.
[65] P. C. Snijders, L. P. H. Jeurgens, and W. G. Sloof, Surf. Sci. 589 (2005) 98.
[66] L. P. H. Jeurgens, F. Reichel, S. Frank, G. Richter, and E. J. Mittemeijer, Surf.
Interface Anal. 40 (2008) 259.
[67] H. Fujiwara, Spectroscopic Ellipsometry, Wiley, New York (2007).
[68] H. G. Tompkins and E. A. Irene, Handbook of ellipsometry, Springer, New York
(2005).
28 Chapter 1
[69] J. A. Stroscio and W. J. Kaiser, ed. Scanning Tunneling Microscopy (1993) Academic
Press, Inc.: San Diego.
[70] D. Bonnell, ed. Scanning Probe Microscopy and Spectroscopy (2001) Wiley-VCH:
New York.
[71] P. Wahl, Local Spectroscopy of Correlated Electron Systems at Metal Surfaces, in
Fachbereich Physik. 2005, Universität Konstanz: Konstanz. p. 127.
[72] D. A. Bonnell, Prog. Surf. Sci. 57 (1998) 187.
[73] D. Briggs and M. P. Seah, ed. Practical Surface Analysis: Ion and Neutral
Spectroscopy (1992) Wiley: New York.
[74] IUPAC Compendium of Chemical Terminology: The Gold Book, ed. by International
Union of Pure and Applied Chemistry, (2002).
Chapter 2
The different initial oxidation kinetics of
Zr(0001) and Zr(10 10) surfaces
Georgijs Bakradze, Lars P.H. Jeurgens and Eric J. Mittemeijer
Abstract
The growth kinetics of thin (thickness < 10 nm) oxide films on Zr(0001) and Zr(101�0) single-crystal
surfaces were investigated by RISE and AR-XPS. To this end, clean crystalline Zr(0001) and
Zr(101�0) surfaces were prepared under UHV conditions by a cyclic treatment of alternating SC and
in-vacuo annealing steps. The thus obtained bare Zr surfaces were then exposed to dry O2(g) in the
temperature range of 300-450 K (at a partial oxygen pressure of 1×10-4 Pa), while monitoring the
growth kinetics by RISE. It was found that the less-densely packed Zr(101�0) surface oxidizes more
readily than the densely packed Zr(0001) surface. A near-limiting thickness of the oxide film on both
surfaces is attained only at oxidation temperatures T < 375 K. At T ≥ 375 K, the oxidation rate
becomes controlled by the thermally-activated dissolution and diffusion of oxygen in the α-Zr
substrate. The higher oxidation rate of the Zr(101�0) surface for T ≥ 375 K is attributed mainly to the
higher oxygen diffusivity in α-Zr along the Zr[101�0] direction than along the Zr[0001] direction.
2.1 Introduction
The possibility of controlled growth of thin, insulating zirconium-oxide films upon thermal
oxidation of zirconium or zirconium-based alloy surfaces is of crucial importance for many
state-of-the-art technologies and applications, such as heterogeneous catalysis [1-4],
microelectronics [5-7] and gas-sensors [1, 7], as well as in the production of corrosion-
resistant coating systems for fuel cladding materials in nuclear reactors [8-10]. In such
applications, the microstructure (e.g. thickness, morphology, chemical constitution,
crystallographic structure and texture) of the thin (thickness < 10 nm) oxide overgrowth (co-)
determines the performance and durability of the material component in operation.
Previous studies on the initial oxidation of various bare (i.e. without a native oxide on
the surface prior to oxidation) metal and alloy surfaces (see Refs. [11-16] and references
therein) have indicated that a precise tailoring of the growth kinetics and developing
30 Chapter 2
microstructure of the oxide overgrowth can in principal be achieved by controlled variation
of, in particular, the oxidation temperature (T), oxygen partial pressure (pO2), the
crystallographic orientation and the surface condition (e.g. grain size and texture, cleanliness,
roughness, defect structure) of the parent substrate prior to oxidation. As recently
demonstrated for the controlled oxidation of bare Al(111), Al(100) and Al(110) surfaces, the
crystallographic orientation of the metal substrate has a strong effect on the oxidation kinetics
and the developing oxide-film microstructure [15-16]. Yet, systematic investigations on the
dependencies of the growth kinetics and the developing microstructure of the initial oxide
overgrowth on the metal substrate orientation are very scarce. Comprehensive substrate-
orientation-dependent oxidation studies have only been performed for several common
metals like Al [15-16], Ni [17-19], Cr [18] and Cu [20]. To the best of our knowledge, such
systematic experimental investigation for other less common pure single-crystalline metals of
the titanium group (e.g. Ti, Zr and Hf) has not been reported until now.
Previous theoretical and experimental work on the oxidation of bare single-crystalline
Zr surfaces [21-27] was mainly focused on the initial stages of oxygen interaction with the
bare Zr(0001) metal surface at low oxygen exposures (< 50 L). Oxidation of the Zr(101 0)
prism plane, as well as the substrate-orientation-dependent oxidation of Zr after prolonged
oxygen exposures (i.e. during oxide-film growth), have been virtually unaddressed up to date
[22, 28]. Current scientific and technological interest concerns primarily the successive stages
of the oxidation process associated with (i) formation of a closed oxide film on the Zr metal
surface and (ii ) subsequent oxide-film growth by transport of reactant species (i.e. cations,
anions and their vacancies, as well as electrons and electron holes) through the thickening Zr-
oxide film. The oxidation behaviour of the prism plane is particularly of significant industrial
interest, because the cold-rolled Zr samples are textured with {101 0} parallel to the surface
[28].
A RISE and AR-XPS study on the initial oxidation of Zr metal has recently been
carried out by our group for weakly textured, polycrystalline Zr surfaces, where the native
oxide was removed before oxidation by Ar+ SC under UHV conditions [29]; a subsequent in-
vacuo annealing step of the SC surfaces was not performed, and thus a distorted crystallinity
and a large-scale roughness of the ion-bombarded Zr surfaces prevailed prior to oxidation.
Against this background the present study addresses the initial oxidation of high-
purity Zr(0001) and Zr(101 0) single crystals with crystallographically well-defined bare,
pure surfaces, which were obtained by an elaborate cyclic treatment of alternating Ar+ SC
The different oxidation kinetics of Zr(0001) and Zr(1010) surfaces 31
and in-vacuo annealing steps (see Section 2.2). The differences in the initial oxidation
kinetics for two orientations of the single-crystal Zr substrates could thus be established by
in-situ RISE for the dry, thermal oxidation of the bare Zr(0001) and Zr(101 0) surfaces in the
temperature range of 300-450 K (at a pO2 = 1×10-4 Pa).
2.2 Experimental
Disc-shaped Zr(0001) and Zr(101 0) single crystals were cut (diameter: 6 mm; thickness 1
mm; orientation alignment within ±0.5º of the nominal surface plane) from a single-
crystalline unalloyed α-Zr rod, and mechanically polished (last step 0.05 µm). Main
contaminations in the prepared samples were identified by Inductively Coupled Plasma
Optical Emission Spectroscopy analysis: (in mass parts) Hf (60 ppm); Fe (25 ppm); Ti (1
ppm); Cu, Zn, Mn, Ca, Na (< 2 ppm).
Samples prepared as described above were introduced into a combined UHV system
for sample processing and in-situ analysis (base pressure < 3×10-8 Pa). The (native) oxide and
other adventitious contaminants on the surface were removed by SC at room temperature
with a focussed 1 kV Ar+ beam (rastering the entire sample surface and employing sample
rotation at a speed of about 2 rpm) until no other element than Zr was detected in a measured
XPS survey spectrum recorded over the BE range from 0 to 1200 eV. Then the sample and
sample holder were outgassed by a cycling treatment of alternating SC and in-vacuo
annealing steps (see above), while gradually increasing the sample temperature during each
successive in-vacuo annealing step up to 1000 K. For the in-vacuo annealing steps at T > 750
K, Fe was found (by means of in-situ AR-XPS) to segregate at the SCed Zr surfaces. To
obtain segregant-free Zr surfaces, the clean single-crystalline surfaces were extensively SCed
(with Ar+ at 1 kV, total sputter time > 120 hrs), while keeping the sample at a constant
temperature in the range 823-873 K (according to Ref. [30], the Fe segregation reaches its
maximum surface coverage on the Zr(101 0) plane at 823 K) and employing sample rotation.
After these elaborate in-vacuo cleaning procedures, no segregated Fe or (Fe-rich) precipitates
(such as reported in Refs. [30-31]) were detected at the surface.
As a final surface-preparation step, prior to each oxidation experiment, the SCed
surfaces were in-vacuo annealed at 1000 K for 300-600 s to restore their crystallinity in the
ion-bombarded near-surface region, which was verified by in-situ LEED (employing a Specs
4-grid ER-LEED system using primary electron energies in the range of 30-200 eV). Typical
32 Chapter 2
LEED patterns of the thus-obtained bare Zr(0001) and Zr(101 0) surfaces prior to oxidation
are shown in Figs.2.1a and b, respectively.
Fig. 2.1. LEED patterns as recorded from (a) the bare Zr(0001) with a primary electron energy of 66
eV and (b) the bare Zr(1010) single-crystal surfaces with a primary electron energy of 85 eV (i.e.
prior to oxidation). See Section 2.2.
Next, oxide films were grown at 300, 325, 350, 375, 400, 425 and 450 K by in-situ
exposure of the bare Zr(0001) and Zr(101 0) surfaces for a period of t = 7200 s to pure
oxygen gas (purity ≥ 99.9999 vol.% with a specified residual gas content of H2O ≤ 0.5 vpm,
N2 + Ar ≤ 2.0 vpm, CnHm ≤ 0.1 vpm and CO2 ≤ 0.1 vpm) at pO2 = 1×10-4 Pa. The oxidation
temperature was measured with a type K thermocouple, which was put in direct mechanical
contact with the sample surface.
For each conducted oxidation experiment, the changes in the spectra of the
ellipsometric parameters ψ(λ) and ∆(λ) were recorded as function of the oxidation time (t) in
the wavelength range λ = 250-1000 nm using a Woollam M-2000L spectroscopic
ellipsometer equipped with a Xe light source and mounted directly to the flanges of the UHV
reaction chamber (with fixed angles of incidence and reflection of 70° relative to the sample
surface normal).
In-situ AR-XPS analysis of the sample surface before and after each oxidation was
conducted with a Thermo VG Thetaprobe system employing monochromatic Al Kα radiation
(hν = 1486.68 eV). XPS survey spectra, covering a BE range from 0 to 1200 eV, were
recorded with a step size of 0.2 eV and constant pass energy of 200 eV. For the bare and
oxidized sample surfaces, AR-XPS spectra of the Zr 3d region were recorded over the BE
range from 176 eV to 206 eV with a step size of 0.1 eV at a constant pass energy of 50 eV.
The different oxidation kinetics of Zr(0001) and Zr(1010) surfaces 33
The AR-XPS measurements were performed according to the so-called parallel data
acquisition mode, by detecting the photoelectrons simulteneously over the angular detection
range from θ = 26.75 to θ = 79.25° in six ranges of 7.5° degrees each (θ is the center of the
corresponding angular detection angle range with respect to the sample-surface normal) [32].
To maximize the surface-analysis area, the AR-XPS spectra were measured at four defined
locations on the surface (incident X-ray spot size about 20 µm × 400 µm), equally distributed
over an entire analysis area of about 2 mm × 5 mm.
2.3 Data evaluation
2.3.1 AR-XPS data
For quantification, the recorded Zr 3d spectra of the bare and oxidized metal were first
averaged over the measured positions on the sample surface for each photoelectron detection
range (see Section 2.2). Next, the thus obtained spectra were corrected for the electron-
kinetic-energy dependent transmission of the hemispherical analyzer by adopting the
corresponding correction factor as provided by the manufacturer. The intrinsic metallic and
oxidic contributions to the corrected Zr 3d spectra, as well as their associated inelastic
backgrounds, were reconstructed according to the procedure, as described in detail in Ref.
[33]. The spectral reconstruction method is based on the convolution of physically realistic
functions for the intrinsic Zr 3d core-level main peaks, for the cross-sections of the intrinsic
and extrinsic electron energy losses and for the instrumental and natural spectral (X-ray)
broadening. To this end, the intrinsic spectra and extrinsic loss functions were determined
separately for the Zr(0001) and Zr(101 0) bare substrates, as well as for thick (i.e. thickness
>> 10 nm) oxidized Zr substrates, following the procedures described in Ref. [34].
For the oxidations for t = 7200 s at T ≤ 375 K, the Zr 3d spectra of the oxidized
Zr(0001) and Zr(101 0) substrates could be accurately described with (see Fig. 2.2): (i) one
metallic main peak due to the Zr substrate, (ii) one predominant Zr4+ oxidic main peak (due
to the presence of stochiometric ZrO2 oxide in the grown oxide film adjacent to the oxide-
film surface; further designated as ZrO2 main peak), and (iii ) (a minimum of) two weaker
suboxidic main peaks due to Zr cations with a lower valence state than Zr4+ (due to the
presence of non-stochiometric Zr-oxide in the grown oxide film adjacent to the metal/oxide
interface), in accordance with previous XPS studies of oxidized polycrystalline Zr metal
surfaces [29, 35].
34 Chapter 2
Fig. 2.2. Exemplary reconstruction of the measured Zr 3d XPS spectrum, as recorded from the
oxidized Zr(1010) single crystal (oxidized for 7200 s at 375 K and at pO2 = 1×10-4 Pa), at a single
detection angle of θ = 34.25o. The calculated spectral contributions (i.e. the various intrinsic Zr 3d
spectral contributions plus their individual inelastic backgrounds), as originating from the metal
substrate, interfacial suboxide or stoichiometric ZrO2, have been indicated. For details, see Section
2.3.1.
Table 2.1: BE positions of the metallic and oxidic Zr 3d5/2 peaks, as resolved by reconstruction of the
measured AR-XPS spectra of the oxidized single-crystalline Zr surfaces (see Section 2.3.1 and Fig.
2.2).
peak designation BE (eV)
metallic Zr 3d Zr0 178.6
suboxidic Zr 3d Zrδ+
Zrγ+
179.7
181.2
main oxidic Zr 3d Zr4+ 183.1
The total thicknesses of the oxide films grown at T ≤ 375 K were calculated by
iteration from the sum of the primary zero loss (PZL) intensity of the predominant ZrO2 main
peak and the PZL intensities of the suboxidic Zr 3d main peaks, following the procedure as
The different oxidation kinetics of Zr(0001) and Zr(1010) surfaces 35
outlined in Ref. [29]. In the calculations, the average molar densities of Zr in the ZrO2 top
layer and in the suboxidic bottom layer (i.e. in the oxide film adjacent to metal/oxide
interface) were taken equal to 46.10 mol/dm3 and 58.71 mol/dm3, respectively [34]. Values
for the effective attenuation lengths (EALs; symbol λ) of the detected Zr 3d photoelectrons
(see Table 2.2) and the anisotropy of the Zr 3d photo-ionization cross-section, as required in
the thickness calculations, were determined according to Refs. [32, 36] while using the
physical constants as listed in Tables 2.1 and 2.2.
Table 2.2: Effective attenuation lengths (EALs; in nm) of the detected Zr 3d photoelectrons in Zr
metal (symbol: Zrλ ) and ZrO2 (symbol: 2ZrOλ ) for the various detection angles (symbol θ; as defined
with respect to the sample surface normal; see Section 2.2), as employed in the AR-XPS
quantification. For the calculation of the EALs [32, 36], the density of Zr metal and ZrO2 were taken
equal to 6.52 and 5.68 g/cm3, respectively [37]. The asymmetry parameter, describing the angular
distribution of the ejected Zr 3d photoelectrons, was taken from Ref. [38]: i.e. β = 1.16. A value of
5.42 eV was adopted for the energy band gap of ZrO2 [39].
θ (°) Zrλ (nm) 2ZrOλ (nm)
26.75 2.664 2.360
34.25 2.591 2.311
41.75 2.529 2.268
49.25 2.477 2.232
56.75 2.433 2.203
64.25 2.395 2.177
71.75 2.362 2.155
79.25 2.328 2.132
For the oxidations at elevated temperatures T > 375 K, the oxide-film thickness after t
= 7200 s approaches or exceeds the average information depth of the XPS analysis of
3λ×cosθ < 7.5 nm (see Section 2.4). Consequently, the metallic contribution from the
underlying Zr substrate can hardly be discerned in the measured AR-XPS spectra of the
oxidized Zr metal. Furthermore, as discussed in Section 2.4.2, the oxide-film thickness
becomes non-uniform. Consequently, for the thicker oxide films of non-uniform thickness
grown after t = 7200 s at T > 375 K, the above described fitting procedure of the oxidized Zr
3d spectra fails and the applied XPS quantification procedures can no longer be applied.
36 Chapter 2
2.3.2 RISE data
The oxide-film growth kinetics (i.e. oxide-film thickness, L, versus oxidation time, t) were
deduced from the recorded RISE raw data by application of an optical model of the evolving
substrate/film system, which accurately describes the measured courses of the ellipsometric
parameters ∆(λ, t) and ψ(λ, t) over the concerned wavelength range as function of oxidation
time. The applied model is constituted of a ZrO2 surface (top) layer and a suboxidic (i.e. non-
stoichiometric) interfacial (bottom) layer, with respective (variable) uniform thicknesses,
ox ( )L t and EMA ( )L t , respectively, on top of an infinitely thick Zr metal substrate. The
suboxidic interfacial layer was introduced to account for the characteristic absorption due to
the development of suboxidic species at the metal/oxide interface, analogously to the RISE
data evaluation procedures adopted in Refs. [34, 40]. The interfacial sublayer, as introduced
in the spectroscopic modeling, accounts for the presence of the non-stoichiometric suboxide
layer, as determined by AR-XPS (see Ref. [29] and a related discussion in Section 2.4.2).
The optical properties (i.e. the complex index of refraction) of the bare substrate at the
oxidation temperature were determined from the measured ψ(λ) and ∆(λ) spectra of the bare
substrate prior to each oxidation (i.e. with the bare substrate held at the oxidation temperature
under UHV conditions) using the pseudodielectric approximation [41]. The wavelength
dependence of the real part of the refractive index (n) of the transparent ZrO2 top layer (i.e.
the extinction coefficient (i.e. imaginary part of the refraction index, k) was taken equal to
zero over the concerned wavelength range) was described by a Cauchy-type relation, i.e.
( ) 2 4n A B Cλ λ λ= + + (with the constants A, B and C as defined below). The optical
constants of the interfacial suboxide were estimated using an effective medium
approximation (EMA) based on the Maxwell-Garnett formulation and defining the
stoichiometric ZrO2 top layer and the parent Zr metal substrate as the “matrix” and the
“inclusion”, respectively [34]. The average EMA-fraction of “metal inclusion” was taken
equal to 0.5 (as determined in a separate fit-parameter study). The Cauchy constants (A, B, C)
of the ZrO2 top layer were determined by fitting of the calculated to the measured spectra of
∆(λ, t=7200s) and ψ(λ, t=7200s) (i.e. at the end of the oxidation after t = 7200 s) at all
oxidation temperatures simultaneously, adopting A, B, C, ox ( 7200s)L t = and
EMA ( 7200s)L t = as the only fit parameters. This resulted in values for the Cauchy constants
for the ZrO2 top layer of: A = 2.393, B = 0 µm2, C = 1.315×10-3 µm4 for the oxidized
Zr(0001) face and of A = 2.367, B = 0 µm2, C = 1.155×10-3 µm4 for the oxidized Zr(101 0)
The different oxidation kinetics of Zr(0001) and Zr(1010) surfaces 37
face. Typical values of the refractive index at λ = 600 nm are n = 2.18 [42] for bulk ZrO2 and
1.88 ≤ n ≤ 2.08 for micrometer-thick ZrO2 films deposited by thermal evaporation [43].
These literature values are significantly lower than the corresponding n-value of 2.38±0.02,
as determined for the much thinner (< 10 nm) stoichiometric ZrO2 top layer in the present
study: thin films typically have n-values higher than thick layers (or bulk materials) due to
e.g. growth stresses and relatively high defect densities [44].
Fig. 2.3. Exemplary as-measured and fitted courses of the ellipsometric parameters ∆(t) (phase
difference) and ψ(t) (amplitude ratio) as function of oxidation time at a central wavelength of λ = 550
nm, for the oxidation of the bare Zr(1010) surface at 375 K (at pO2 = 1×10-4 Pa). The measured and
calculated (fitted) data are indicated by markers and solid lines, respectively.
Using the optical constants for the top and bottom sublayers indicated above, the total
oxide-film growth curves, )t(LSE
tot = ox ( )L t + EMA ( )L t , were obtained by fitting the calculated
to the measured spectra of ∆(λ, t) and ψ(λ, t) (over the wavelength range from 350-800 nm
using the WVASE 32 software package), while adopting ox ( )L t and EMA ( )L t as the only
time-dependent fit parameters. Exemplary results of the fitting of ∆(λ, t) and ψ(λ, t), at a
typical wavelength of λ = 550 nm (i.e. in the centre of the simultaneously-fitted whole
wavelength range) for the oxidation of the Zr(101 0) surface at 375 K, are shown in Fig. 2.3.
38 Chapter 2
2.4 Results and discussion
2.4.1 Oxide-film constitution
For oxidation temperatures T ≤ 375 K, the measured Zr 3d photoelectron spectra of the
oxidized Zr sample can be accurately described with one metallic, one predominant oxidic
and two weaker suboxidic spectral components (designated as ZrO2 and suboxidic
components, respectively, see Fig. 2.2, Table 2.1 and Section 2.3.1). The predominant Zr4+
oxidic contribution is associated with stoichiometric ZrO2 in the region of the oxide film
adjacent to the surface, whereas the suboxidic contributions arise from an interfacial suboxide
layer (beneath the stoichiometric ZrO2 top layer): see Ref. [34]. Furthermore, as also
demonstrated in Ref. [34], the interfacial suboxide is enriched in Zr (as compared to ZrO2);
the degree of Zr enrichment in the oxide film increases towards the metal/oxide interface,
resulting in an overall decrease of the average valence state of Zr from +4 to 0 (from the ZrO2
top layer to the parent metal substrate).
As discussed in Sections 2.3.1 and 2.4.2, for the thicker oxide films of non-uniform
thickness grown at T > 375 K after t = 7200 s, the AR-XPS fitting procedure of the Zr 3d
spectra fails and the oxide-film thickness can no longer be quantified by AR-XPS.
2.4.2 Oxide film thickness: comparison of AR-XPS and RISE
analyses
The total oxide-film thicknesses, as determined by RISE ( SEtotL ) and AR-XPS ( XPS
totL ), agree
fairly well for the oxidation of the Zr(101 0) and, in particular, the Zr(0001) surface for 7200
s of oxidation in the temperature range 300-375 K: see Fig. 2.4. However, the RISE thickness
values are systematically larger than the corresponding AR-XPS thickness values. In
particular, the RISE thickness value becomes increasingly larger than the corresponding AR-
XPS thickness value with increasing oxidation temperature for both substrates.
As revealed by in-situ STM investigations (see Ref. [45] or Chapter 4), the oxidized
Zr(0001) and Zr(101 0) surfaces are significantly rougher than the respective bare (i.e. SCed
and annealed) Zr surfaces (i.e. prior to oxidation), particularly at the higher oxidation
temperatures. It should be noted here that the EMA sublayer, as determined by RISE, not
only accounts for a deviation of the oxide stoichiometry adjacent to the interface, but also
effectively describes combined effects of roughness at the metal/oxide interface and/or oxide
surface [41]. Therefore, the RISE quantification procedure (see Section 2.3.2) tends to
The different oxidation kinetics of Zr(0001) and Zr(1010) surfaces 39
overestimate the total oxide-film thickness, particularly at the higher oxidation temperatures:
see Fig. 2.4.
The oxidic Zr 3d photoelectron intensity, as detected by AR-XPS, is proportional to
the number Zr cations in the probed oxide volume. Consequently, the presence of
substrate/interface roughness and/or non-uniformity of the oxide-film thickness, to a first
approximation, do not affect the determination of the total oxide film thickness by AR-XPS.
However, as discussed in Section 2.3.1, the XPS quantification procedure becomes unreliable
as soon as the film thickness approaches the information depth of the XPS analysis (i.e.
3λ×cosθ < 7.5 nm). Further, the AR-XPS analysis is performed over an area of about 20 µm
×400 µm, which is much smaller than the area of 2 mm × 5 mm analyzed by RISE (Section
2.2): i.e. the RISE quantification procedure averages over a much larger area of the oxidized
surfaces.
Fig. 2.4. Comparison of the total oxide-film thickness, SEtotL , after t = 7200s of oxidation, as obtained
by RISE, with the corresponding total oxide-film thickness, XPStotL , as independently determined by
AR-XPS, for the oxidation of the bare (a) Zr(0001) and (b) Zr(101 0) substrates at various
temperatures in the range of 300 to 375 K (at pO2 = 1×10-4 Pa). The dashed lines represent the ideal
relationship SEtotL = XPS
totL . Note that a comparison of the thicknesses by AR-XPS and RISE was not
possible for oxidation temperatures T > 375 K, as discussed in Section 2.4.1.
The average thicknesses of the interfacial suboxide layers, as determined by AR-XPS,
are 0.43±0.05 nm and 0.20±0.01 nm for the Zr(0001) and Zr(101 0) surfaces, respectively,
40 Chapter 2
independent of the oxidation temperature in the range of T = 300-350 K. These values are
compatible with those determined for the near-limiting EMA thicknesses, as determined by
RISE (see Fig. 2.5).
2.4.3 Oxide-film growth kinetics
Total oxide-film growth curves (i.e. total oxide-film thickness, SEtotL , versus oxidation time, t)
for the thermal oxidation of the bare Zr(0001) and Zr(101 0) surfaces in the temperature
range 300-450 K (at pO2 = 1×10-4 Pa) are shown in Figs. 2.5a and b, respectively. The total
oxide-film thickness, as determined by RISE, corresponds to the sum of the thickness, Lox(t),
of the ‘stoichiometric’ ZrO2 top layer (see Figs. 2.5a and b) and the thickness, LEMA(t), of the
suboxide (‘non-stoichiometric’) interface layer (see Section 2.3.2 and Figs. 2.5c and d).
Fig. 2.5. (a, b) Total oxide-film thickness, SEtotL , and (c, d) corresponding EMA thickness, LEMA, of the
suboxide interfacial layer, as function of oxidation time for the oxidation of the bare Zr(0001) and
Zr(1010) substrates at various oxidation temperatures in the range of 300 to 450 K at pO2 = 1×10-4 Pa
(as determined by RISE; cf. Section 2.3.2 and Fig. 2.3).
The different oxidation kinetics of Zr(0001) and Zr(1010) surfaces 41
It follows that, for T ≤ 350 K, the oxide-film growth rate is initially very fast, but
drastically decreases already within the first 500 to 1900 s of oxidation. As a result, near-
limiting oxide-film thicknesses of 1.26±0.11 nm and 1.44±0.15 nm are attained on the
Zr(0001) and Zr(101 0) surfaces after 7200 s of oxidation at 300 K, respectively. This
passivation behavior is typical for the oxidation of metals and alloys at low temperatures,
when the rate of diffusion of cations and/or anions through the developing oxide film under
influence of the (electro)chemical potential (i.e. concentration) gradients is negligibly small
[14, 46-48]. The here-determined near-limiting oxide-film thickness values of 1.26±0.11 nm
and 1.44±0.15 nm at 300 K are comparable to the near-limiting oxide-film thickness value of
≈1.45 nm [49], as reported for the oxidation of SCed polycrystalline Zr surfaces at 304 K
(and pO2 ≈ 10-4 Pa). Note that the polycrystalline Zr substrate surfaces, as employed in Ref.
[49], have a distorted crystallinity at their surface (as induced by a 3 keV Ar+ SC step prior to
oxidation), which complicates a direct comparison with the present results, obtained for the
oxidation of well-defined single-crystalline Zr surfaces.
The decrease of the oxide-film growth rate after the initial fast oxidation regime
becomes less pronounced with increasing oxidation temperature (Fig. 2.5); for the oxidation
of the Zr(0001) and Zr(101 0) surfaces at T > 350 K, a passivation behavior of the oxide-film
growth kinetics (i.e. with the occurrence of a near-limiting oxide-film thickness) is no longer
observed. Moreover, at these elevated oxidation temperatures the oxide-film growth kinetics
on the Zr(0001) and Zr(101 0) surface become distinctly different: i.e. for oxidation times t <
500 s, the oxide-film growth rate is considerably lower for the densely packed Zr(0001)
surface than for the less densely packed Zr(101 0) surface (at the same oxidation temperature
T ≥ 350 K); compare Figs. 2.5a and b. Two hours of oxidation at 400 K results in a total
oxide-film thickness of 2.98±0.30 nm and 4.59±0.37 nm, respectively. The here observed
non-passivating oxidation kinetics of the Zr single-crystal surfaces at T > 350 K is in line
with previous results on the oxidation of polycrystalline Zr [26-27, 29].
The difference in the oxide-film growth kinetics for the Zr(0001) and Zr(101 0)
surfaces becomes only distinct above a certain oxidation temperature (say, T ≥ 350 K); the
deviation between the growth curves increases with increasing temperature (see Figs. 2.5a
and b). This behaviour hints at a thermally activated nature of the rate-limiting step(s) in the
oxidation process, in dependence on the orientation of the parent metal surface. Recognizing
the enhanced solubility of O in α-Zr at elevated T [29, 50], the substrate-orientation
dependence of the oxidation kinetics at T > 375 K may be attributed to the anisotropy of the
42 Chapter 2
oxygen diffusion in α-Zr [9]1. Several studies have reported a distinct anisotropy of the O
diffusion coefficient in α-Zr with ||D D⊥ > , where ⊥D and ||D denote the O diffusion
coefficients in α-Zr in directions perpendicular and parallel to the c-axis (i.e. along the [101
0] and [0001] directions, respectively) [21, 52]. As will be argued below, a realistic value for
the anisotropy of the oxygen diffusion in α-Zr suffices to induce the substrate-orientation-
dependent oxide-film growth kinetics, as observed in the present study.
Provided that the oxygen-incorporation rate (i.e. dissolution; see Chapter 5 or Ref.
[51] and footnote 1) into the substrate is the rate-limiting step2 in the α-Zr oxidation process
at 375 ≤ T ≤ 450 K, a rough estimate of the anisotropy ratio, ⊥= DD||κ , for the diffusion of
oxygen in α-Zr can be obtained recognizing that the amount of diffusant taken up in a semi-
infinite solid is proportional with tD ⋅ [53], with D as the diffusion coefficient of O in α-Zr
and t as the diffusion (oxidation) time. For substitutional diffusion of oxygen in the α-Zr
substrate, the amount of incorporated oxygen is proportional with the metal volume
consumed in the oxidation process and thus with the total oxide-film thickness, totL . It then
follows that ( )2tottot|| (prism)(basal)κ LLDD ≈= ⊥ , where (basal)totL and (prism)totL are the
total oxide-film thicknesses on the basal and prism Zr planes for the same oxidation
conditions. The total oxide-film thicknesses on the basal and prism Zr planes, as determined
by RISE after 7200s of oxidation at T = 450 K (see Fig. 2.5), amount to (basal)SEtotL = 6.5 nm
and (prism)SEtotL = 9.2 nm, respectively, which results in κ ≈ 0.5 (at T = 450 K). Not only the
sign (i.e. ||D D⊥ > ), but also the magnitude of the anisotropy coefficient κ, is in line with data
reported in Refs. [21, 52] for O diffusion in α-Zr at T > 523 K.
An estimate for the separate diffusion coefficients ⊥D and ||D may be obtained by
proceeding one step further as follows: (i) the thickness of the Zr metal layer converted into
oxide may be obtained from PBtot rL , where PBr is the Pilling-Bedworth ratio (i.e. the ratio of
the molar volumes per metal atom of metal oxide and metal); (ii ) the thickness of the
converted Zr metal layer may be equated with tD ⋅ . From (i) and (ii ) it follows that D can
1 As postulated in Chapter 5 and Ref. [51], the O vacancies, generated in the oxide layer upon the dissolution of O atoms/ions into the Zr substrate, sustain an outward flux of O vacancies, which diffuse through the oxide film to the surface and are annihilated there (i.e. filled-up with oxygen ions from the oxygen atmosphere). 2 Note: at even higher oxidation temperatures (say, T > 573 K) the oxygen dissolution rate into α-Zr becomes much faster than the formation of the oxide film (e.g. as observed in Ref. [29]). Then oxygen incorporation into the substrate is no longer rate-determining for the oxide-film growth.
The different oxidation kinetics of Zr(0001) and Zr(1010) surfaces 43
be estimated by ( )2
tot PB1D t L r≈ ⋅ . Then taking 561PB .r = [54] and the above data for
(basal)totL and (prism)totL , it is obtained ||D ≈2.4×10-21 m2/s and D⊥ ≈4.8×10-21 m2/s for the
oxygen diffusivities in α-Zr at T = 450 K. These crude estimates are in a very good agreement
with the overall value for the diffusion coefficient reported in Ref. [55] for the dissolution
rate of chemisorbed oxygen into polycrystalline Zr surfaces upon in-situ annealing 1.0×10-21
m2/s at T = 450.
The oxide films, as grown on the Zr(0001) and Zr(101 0) surfaces in the range of T =
300-450 K, have also been investigated by in-situ XPS (see Ref. [56] or Chapter 3) and in-
situ STM (see Chapter 4). The in-situ XPS VB analysis (Chapter 3 or [56]) shows that the
oxide films formed at T < 400 K are predominantly amorphous, whereas those formed at T >
400 K are predominantly crystalline. The development of long-range order in the thickening
oxide films at T > 400 K, as indicated by XPS, runs parallel with a coarsening of the oxide-
film microstructure, as observed by in-situ STM. From these data it follows that the oxide-
film growth kinetics on the Zr(0001) and Zr(101 0) surfaces become distinctly different at
temperatures where a crystalline oxide phase develops. The in-situ STM image analysis
further indicates that the grain size of the predominantly crystalline oxide films formed at T >
400 K is, on average, smaller (and, thus, the GB density is on average higher) for the oxide
overgrowths on the Zr(101 0) substrate, which substrate exhibits the higher oxidation rate at
T ≥ 350 K. It might therefore not be fully ruled out that the difference in GB density of the
crystalline oxide films on the basal and prism Zr surfaces contributes to the observed
substrate-orientation dependence of the oxidation kinetics as well (i.e. in addition to the
above-discussed contribution originating from the anisotropy of the oxygen diffusion in α-
Zr).
2.5 Conclusions
The growth kinetics of the oxide films formed on bare Zr(0001) and Zr(101 0) single-
crystalline surfaces upon thermal oxidation in pure oxygen gas in the temperature range of
300-450 K can be successfully determined by in-situ RISE through application of a three-
layer optical model, which is constituted of a thickening ZrO2 layer and a thinner EMA
suboxide interfacial layer adjacent to the bare Zr substrate. AR-XPS analysis of the oxidized
Zr surfaces confirms the existence of this suboxide layer.
44 Chapter 2
After a short initial stage of fast oxide-film growth, a near-limiting thickness of the
oxide film is attained at T < 375 K on both Zr surfaces. The non-passivating oxidation
kinetics of the single-crystal Zr surfaces at T ≥ 350 K are in accordance with previous reports
on the thermal oxidation of polycrystalline Zr surfaces.
Distinct differences in the oxidation kinetics for the two Zr substrate orientations
become apparent at T > 375 K: the Zr(101 0) prism plane oxidizes more readily than the
more densely-packed Zr(0001) basal plane under the same experimental conditions. At T >
375 K, the oxidation rate of both Zr faces becomes governed by thermally-activated
dissolution and diffusion of oxygen into the α-Zr substrate. On this basis quantitative
estimates for the diffusion coefficients of oxygen in α-Zr parallel and perpendicular to the
crystallographic c-axis can be obtained, which indicate that oxygen diffusion along the Zr[10
1 0] direction is faster (by about a factor of two at 450 K) than along the Zr[0001] direction.
References
[1] N. Stojilovic, E. T. Bender, and R. D. Ramsier, Prog. Surf. Sci. 78 (2005) 101.
[2] C. Stampfl, M. V. Ganduglia-Pirovano, K. Reuter, and M. Scheffler, Surf. Sci. 500
(2002) 368.
[3] B. M. Reddy, M. K. Patil, K. N. Rao, and G. K. Reddy, J. Mol. Catal. A 258 (2006)
302.
[4] J. A. Anderson and M. F. Garcia, Supported Metals in Catalysis, Imperial College
Press, London (2005).
[5] C. L. Chang and S. Ramanathan, J. Electrochem. Soc. 154 (2007) G160.
[6] M. Gutowski, J. E. Jaffe, C. L. Liu, M. Stoker, R. I. Hegde, R. S. Rai, and P. J. Tobin,
Appl. Phys. Lett. 80 (2002) 1897.
[7] G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89 (2001) 5243.
[8] C. S. Zhang, B. J. Flinn, and P. R. Norton, Surf. Sci. 264 (1992) 1.
[9] J. P. Pemsler, J. Electrochem. Soc. 105 (1958) 315.
[10] C. O. Degonzalez and E. A. Garcia, Appl. Surf. Sci. 44 (1990) 211.
[11] F. P. Fehlner and N. F. Mott, Oxid. Met. 2 (1970) 59.
[12] A. Atkinson, Rev. Mod. Phys. 57 (1985) 437.
[13] K. R. Lawless, Rep. Prog. Phys. 37 (1974) 231.
[14] M. Martin and E. Fromm, J Alloy Compd 258 (1997) 7.
The different oxidation kinetics of Zr(0001) and Zr(1010) surfaces 45
[15] F. Reichel, L. P. H. Jeurgens, G. Richter, and E. J. Mittemeijer, J. Appl. Phys. 103
(2008) 093515.
[16] F. Reichel, L. P. H. Jeurgens, and E. J. Mittemeijer, Acta Mater. 56 (2008) 2897.
[17] D. F. Mitchell, P. B. Sewell, and M. Cohen, Surf. Sci. 61 (1976) 355.
[18] L. P. Bonfrisco and M. Frary, J. Mater. Sci. 45 (2010) 1663.
[19] D. F. Mitchell, P. B. Sewell, and M. Cohen, Surf. Sci. 69 (1977) 310.
[20] F. W. Young, J. V. Cathcart, and A. T. Gwathmey, Acta Metall. Mater. 4 (1956) 145.
[21] B. Li, A. R. Allnatt, C. S. Zhang, and P. R. Norton, Surf. Sci. 330 (1995) 276.
[22] H. G. Kim, T. H. Kim, and Y. H. Jeong, J. Nucl. Mater. 306 (2002) 44.
[23] R. A. Ploc, J. Nucl. Mater. 110 (1982) 59.
[24] Y. M. Wang, Y. S. Li, and K. A. R. Mitchell, Surf. Sci. 343 (1995) L1167.
[25] M. Yamamoto, C. T. Chan, K. M. Ho, and S. Naito, Phys. Rev. 54 (1996) 14111.
[26] M. Yamamoto, S. Naito, M. Mabuchi, and T. Hashino, J. Chem. Soc. Faraday Trans.
86 (1990) 3797.
[27] M. Yamamoto, S. Naito, M. Mabuchi, and T. Hashino, J. Chem. Soc. Faraday Trans.
87 (1991) 1591.
[28] C. S. Zhang, B. Li, and P. R. Norton, Surf. Sci. 313 (1994) 308.
[29] A. Lyapin, L. P. H. Jeurgens, and E. J. Mittemeijer, Acta Mater. 53 (2005) 2925.
[30] C. S. Zhang, B. Li, and P. R. Norton, Surf. Sci. 338 (1995) 157.
[31] C. S. Zhang, B. Li, and P. R. Norton, J. Nucl. Mater. 223 (1995) 238.
[32] M. S. Vinodh and L. P. H. Jeurgens, Surf. Interface Anal. 36 (2004) 1629.
[33] A. Lyapin and P. C. J. Graat, Surf. Sci. 552 (2004) 160.
[34] A. Lyapin, L. P. H. Jeurgens, P. C. J. Graat, and E. J. Mittemeijer, J. Appl. Phys. 96
(2004) 7126.
[35] C. Morant, J. M. Sanz, L. Galan, L. Soriano, and F. Rueda, Surf. Sci. 218 (1989) 331.
[36] A. Jablonski and C. J. Powell, Surf. Sci. Rep. 47 (2002) 35.
[37] CRC Handbook of Chemistry and Physics, Internet Version 2005, ed. by D. R. Lide,
Taylor and Francis Group, LLC, (2005).
[38] R. F. Reilman, A. Msezane, and S. T. Manson, J. Electron Spectrosc. 8 (1976) 389.
[39] B. Kralik, E. K. Chang, and S. G. Louie, Phys. Rev. B 57 (1998) 7027.
[40] M. S. Vinodh, L. P. H. Jeurgens, and E. J. Mittemeijer, J. Appl. Phys. 100 (2006) 9.
[41] H. Fujiwara, Spectroscopic Ellipsometry, Wiley, New York (2007).
[42] V. I. Aleksandrov, V. F. Kalabukhova, E. E. Lomonova, V. V. Osiko, and V. I.
Tatarintsev, Inorg. Mater. 13 (1977) 1747.
46 Chapter 2
[43] I. Ohlidal, D. Necas, D. Franta, and V. Bursikova, Diam. Relat. Mater. 18 (2009) 364.
[44] H. G. Tompkins and E. A. Irene, Handbook of ellipsometry, Springer, New York
(2005).
[45] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, Surf. Sci. submitted (2011).
[46] N. Cabrera and N. F. Mott, Rep. Prog. Phys. 12 (1948) 163.
[47] L. P. H. Jeurgens, A. Lyapin, and E. J. Mittemeijer, Acta Mater. 53 (2005) 4871.
[48] A. T. Fromhold and E. L. Cook, Phys. Rev. 158 (1967) 600.
[49] A. Lyapin, L. P. H. Jeurgens, P. C. J. Graat, and E. J. Mittemeijer, Surf. Interface
Anal. 36 (2004) 989.
[50] J. P. Abriata, J. Garcés, and R. Versaci, Bull. Alloy Phase Diag. 7 (1986) 116.
[51] G. Bakradze, L. P. H. Jeurgens, U. Starke, T. Acartürk, and E. J. Mittemeijer, Acta
Mater. 59 (2011) 7498.
[52] G. M. Hood, H. Zou, S. Herbert, R. J. Schultz, H. Nakajima, and J. A. Jackman, J.
Nucl. Mater. 210 (1994) 1.
[53] J. Crank, The mathematics of diffusion, Clarendon Press, Bristol (1975).
[54] C. J. Rosa, J. Less-Common Met. 15 (1968) 35.
[55] J. S. Foord, P. J. Goddard, and R. M. Lambert, Surf. Sci. 94 (1980) 339.
[56] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, J. Phys. Chem. C 115 (2011)
19841.
Chapter 3
Valence-band and chemical-state analyses of Zr and
O in thermally-grown thin zirconium-oxide films:
an XPS study
Georgijs Bakradze, Lars P.H. Jeurgens and Eric J. Mittemeijer
Abstract
In-situ XPS was applied to investigate the VB spectrum and to analyse the local chemical states of Zr
and O in thin (thickness < 10 nm) oxide films, grown on bare single-crystalline Zr surfaces by dry
thermal oxidation in the temperature range of T = 300-450 K. The oxide films grown at T ≤ 400 K are
predominantly amorphous. The measured upper VB region of the grown oxide films shows
pronounced changes in shape with increasing oxidation temperature, which can be attributed to the
gradual formation of a tetragonal ZrO2-like phase at T > 400 K. The resolved Zr 3d5/2 and O 1s
photoelectron lines and the Zr M45N1N23, Zr M45N23N23, Zr M45N23V and O KL23L23 Auger transitions
were combined to construct so-called Wagner plots for Zr and O in the oxide films. The observed
decreases of the Zr and O Auger-parameter values with increasing oxidation temperature evidence a
lowering of the electronic polarizability around core-ionized Zr and O atoms. It was concluded that
the amorphous-to-crystalline transition of the oxide films with increasing oxidation temperature is
accompanied with an increase of the Zr-O bond ionicity and changes in the first coordination spheres
of both Zr and O. The results obtained for the amorphous-to-crystalline transition of zirconium-oxide
films were compared with those for Al2O3 films.
3.1 Introduction
For many years zirconium and its alloys have been applied in the production of cladding
materials in nuclear reactors due to the low neutron scattering cross-section of Zr [1-3]. More
recently, zirconium and zirconia-based materials have also been applied in various
nanotechnologies: for example, as catalyst in heterogeneous catalysis [4-7] and as gate
material in MOS-FET devices (due to the high dielectric constant of ZrO2) [8-10]. The
microstructure of the zirconia layers to a large extent determines the performance and
durability of the Zr-based component in operation. Therefore, profound knowledge of the
48 Chapter 3
relationships between the oxidation conditions (e.g. temperature, oxygen partial pressure) and
the developing oxide microstructure is required.
XPS is a powerful surface analytical technique to reveal and characterize the
(different) chemical bonding states of elements in near-surface regions (i.e. up to depths of
about 10 nm) of a solid compound. For example, different oxidation states of an element in
an oxide can often be distinguished on the basis of the chemical shifts of the respective core-
level photoelectron lines with respect to some well-defined reference state (e.g. the metallic
state or free atom in a gas phase). The local chemical state of an element in a solid can
particularly efficaciously (see below) be assessed by XPS on the basis of the so-called
modified Auger parameter (AP), as first introduced by Wagner [11-13]. The modified AP of
an element in a compound is defined as the sum of the KE of the most prominent and sharp
core-level-like Auger transition (line) and the BE of the most prominent and sharp core-level
photoelectron (line) (see Section 2.53.3.2 and [11, 14]). The AP provides a direct measure of
the electronic polarizability of the chemical environment around the core-ionized atom and is
therefore sensitive to structural changes in the nearest coordination sphere of the element
considered [15-17].
The value of the AP is independent of the selected energy reference level (i.e. the
position of the Fermi level in the band gap) and thus is unaffected by aberrational energy
shifts of the measured photoelectron and Auger-electron lines due to charging (as
encountered in the XPS analysis of insulating compounds) or due to band-bending effects (as
commonly observed in thin films). Against the above background and recognizing the high
surface sensitivity offered by XPS, the AP provides a unique tool for the direct experimental
assessment of the changes in the local chemical state of elements in thin films as function of
e.g. the growth conditions.
Since valence electrons are directly involved in the chemical bonding, the shape of the
VB is, next to the AP (see above), sensitive to structural changes in elemental solids and
compounds. For example, various XPS VB studies [18-23] have demonstrated that
polymorphic modifications of an oxide can be distinguished on the basis of differences in the
fine structure of the upper VB (UVB) spectrum. Such structural modifications of a compound
(i.e. without any associated change in the chemical composition) are generally not revealed
by energy shifts of the XPS core-level lines and Auger lines. Although the fine-structure of
the VB spectrum, as measured by XPS or ultraviolet photoelectron spectroscopy, is highly
sensitive to structural changes, the obtained information on the associated changes in the
Valence-band and chemical state analyses of Zr and O in thermally-grown thin zirconium oxide films 49
local chemical environments of the constituent atoms remains of only qualitative nature if
theoretical descriptions of the corresponding densities of states are not available.
In the current XPS investigation of the initial thermal oxidation of Zr surfaces, the
measured fine-structure of the oxide-film VB, as well as the APs of Zr and O, were employed
to reveal the microstructural evolution of oxide overgrowths as function of the oxidation
temperature in the range of T = 300-450 K. To this end, bare (i.e. without a native oxide)
crystalline Zr(0001) and Zr(101 0) surfaces were prepared under UHV conditions and
subsequently exposed for 7200 s to dry O2(g) at a partial oxygen pressure of pO2 = 1×10-4 Pa
and at a temperature in the range of 300-450 K (see Section 3.2). The XPS measurements of
the oxidized Zr surfaces were performed in-situ at room temperature immediately after each
oxidation experiment. To the best of our knowledge, for the first time, Wagner (chemical
state) plots have been constructed for Zr and O in thin (thickness < 10 nm) zirconium oxide
films. On the basis of the analysis of the observed changes in fine-structure of the resolved
oxide-film VB spectra and the accompanying changes in the local chemical states of Zr and
O (see Section 3.3), the microstructural evolution of the oxide overgrowths with increasing
oxidation temperature has been revealed.
3.2 Experimental procedure and spectra evaluation
Disc-shaped Zr(0001) and Zr(101 0) single crystals were cut (diameter 6 mm; 1 mm thick;
orientation alignment within ±0.5º of the nominal surface plane) from a single-crystalline
unalloyed α-Zr rod and single-side polished (last polishing step 0.05 µm). Main bulk
impurities in the as-prepared samples, as identified by Inductively Coupled Plasma Optical
Emission Spectroscopy analysis, are (in mass parts): Hf (60 ppm); Fe (25 ppm); Ti (1 ppm);
Cu, Zn, Mn, Ca, Na (< 2 ppm).
The samples were introduced in a multi-chamber UHV system (base pressure < 2×10-
8 Pa) for in-vacuo sample processing (e.g. SC, annealing, oxidation) and in-situ analysis by
AR-XPS, RISE, LEED, Reflection High Energy electron Diffraction (RHEED) and STM.
First, the (native) oxide and other contaminants on the Zr surface were removed by
SC at room temperature with a focussed 1 kV Ar+ beam (rastering the entire sample surface
and employing sample rotation at a speed of about 2 rpm) until no other element than Zr was
detected in a measured XPS survey spectrum recorded over the BE range from 0 to 1200 eV.
Next, the sample and sample holder were outgassed by a cyclic treatment of alternating SC
50 Chapter 3
(as above) and in-vacuo annealing steps, while gradually increasing the sample temperature
during each successive in-vacuo annealing step up to 1000 K. For the in-vacuo annealing
steps at T > 750 K, Fe was found (by means of in-situ AR-XPS) to segregate at the SCed Zr
surfaces. To obtain segregant-free Zr surfaces, the clean single-crystalline surfaces were
extensively SCed (with Ar+ at 1 kV, total sputter time > 120 h; employing sample rotation),
while keeping the sample at a constant temperature of 823 K. As a result, no segregated Fe or
other (Fe-rich) precipitates (such as reported in [24-25]) were detected at the surface (as
verified by AR-XPS). As a final surface-preparation step prior to each oxidation experiment,
the SCed surfaces were in-vacuo annealed at 1000 K for 300-600 s to restore the crystallinity
at the ion-bombarded surface, as verified by in-situ LEED and STM; corresponding LEED
and STM analyses of the bare Zr(0001) and Zr(10 0) surfaces have been presented in Ref.
[26].
Fig. 3.1. (a) Exemplary reconstruction of the measured Zr 3d XPS spectrum, as recorded from the
oxidized Zr(10 0) single crystal (oxidized for 7200 s at 375 K and at pO2 = 1×10-4 Pa), at a detection
angle of θ = 34.25o. The calculated spectral contributions (plus their individual inelastic
backgrounds), as originating from the metal substrate, interfacial suboxides and stoichiometric ZrO2
have been indicated, see Ref. [26] for details. (b) Exemplary reconstruction of the measured O 1s XPS
spectrum, as recorded at a detection angle of θ = 34.25o from the same oxidized Zr surface as in (a).
The resolved predominant high-BE (HBE) and surface-adjacent low-BE (LBE) O 1s main peaks have
been indicated.
Next, oxide films were grown at substrate temperatures in the range of 300-450 K by
in-situ exposure of the bare Zr(0001) and Zr(10 0) surfaces for a period of t = 7200 s to pure
oxygen gas (purity ≥ 99.9999 vol.% with a specified residual gas content of H2O ≤ 0.5 vpm,
1
1
1
Valence-band and chemical state analyses of Zr and O in thermally-grown thin zirconium oxide films 51
N2+Ar ≤ 2.0 vpm, CnHm ≤ 0.1 vpm and CO2 ≤ 0.1 vpm) at a partial oxygen pressure of pO2 =
1×10-4 Pa. The oxidation temperature was measured with a calibrated type K thermocouple,
which was put in direct mechanical contact with the sample surface.
Fig. 3.2. The as-measured (a) Zr MNN and (b) O KLL Auger spectra, as recorded (at θ = 53o) from
the same oxidized Zr surface as in Fig. 3.1. The positions of the Zr M45N1N23, Zr M45N23N23, Zr
M45N1V, Zr M45N23V and O KL23L23 Auger peaks have been indicated.
In-situ XPS analyses of the Zr surfaces before and after each oxidation were
conducted with a Thermo VG Thetaprobe system employing monochromatic Al Kα radiation
(hν = 1486.68 eV). The Thetaprobe system is equipped with a special radian lens (i.e. a 180°-
spherical-sector-analyzer with an exceptionally large angular acceptance of 60°), which
allows simultaneous detection of all photoelectrons over the angular range (θ) between 23°
and 83° (with respect to the sample-surface normal; i.e. the central detection angle equals
53°). For the bare and oxidized surfaces, XPS spectra of the Zr 3d region were recorded over
the BE range from 176 to 206 eV with a step size of 0.05 eV at a constant pass energy of 50
eV; the Zr MNN Auger lines were recorded over the BE range from 1320 to 1410 eV with a
step size 0.1 eV at a constant pass energy of 100 eV; the O 1s region was measured over the
BE range from 520 to 540 eV with a step size of 0.05 eV and a constant pass energy of 50
eV; the O KLL region was measured over the BE range from 955 to 1035 eV with a step size
of 0.1 eV and a constant pass energy of 50 eV; finally, the VB spectra were recorded over the
BE range from -5 to 28 eV with a step size 0.1 eV at a constant pass energy of 100 eV.
The measured Zr 3d, Zr MNN, O 1s, O KLL and VB spectra of the bare and oxidised
Zr single crystal surfaces were all corrected for the electron KE dependent transmission of the
52 Chapter 3
analyser by adopting the corresponding correction factor as provided by the manufacturer.
Furthermore, the lower BE side of the thus corrected XPS spectra was set to zero
(background) intensity by subtraction of a constant background, the value of which was taken
equal to the averaged minimum intensity at the lower BE side of the corresponding peak
envelop.
Next the BE position of the predominant O 1s peak (at the lower BE side of the O 1s
peak envelop; see Fig. 3.1b) were resolved by the dedicated spectrum-reconstruction
procedures, as described in detail in Refs. [27-28]. The peak positions of the Zr 3d5/2, peak
(corresponding to the Zr4+ oxidation state; see Fig. 3.1a and footnote 1), Zr M45N1N23, Zr
M45N23N23, Zr M45N23V and O KL23L23 Auger lines (see Fig. 3.2) were straightforwardly
determined from the zero value of the first derivative of a third order polynomial function,
which was fitted to the top region of the concerned main peak after subtraction of a linear
background.
Fig. 3.3. As-measured XPS VB spectra, as recorded from the thermally oxidized (at temperatures as
indicated) (a) Zr(0001) and (b) Zr(10 0) single-crystalline surfaces. The oxide films were grown by
in-situ exposure of the bare Zr surface for 7200 s to pure O2(g) at a constant substrate temperature in
the range of 300-450 K and at pO2 = 1×10-4 Pa. The positions of the lower (LVB) and UVB have been
indicated. The arrow indicates the direction of increasing oxidation temperature.
1 The two predominant XPS peaks around 183 eV and 186 eV in Fig. 3.1a constitute the Zr 3d spin-orbit doublet, as associated with the Zr4+ valence state in the oxide film. The much weaker suboxidic Zr 3d contributions due to non-stoichiometric oxide phases at the Zr/oxide interface (see Fig. 3.1a; as resolved by the spectrum-reconstruction procedure described in Refs. [17, 24]), were not considered in the present chemical state analyses.
1
Valence-band and chemical state analyses of Zr and O in thermally-grown thin zirconium oxide films 53
The oxidic VB rest spectrum (Fig. 3.5), which corresponds to the contribution of the
oxide-film to the measured VB spectrum, was obtained by subtraction of the scaled VB
spectrum of the bare metal (after correcting for a zero-background offset) from the measured
VB spectrum of the oxidized metal (Fig. 3.3): see Fig. 3.4. The scaling of the bare metal VB
spectrum was performed within the BE range of -5 to 3 eV. It is thus implicitly assumed that
the intensity within this BE range of the measured VB spectrum of the oxidized metal
originates only from the Zr metal substrate (i.e. any potential contribution arising from
localized electron states in the oxide film due to O defects, which can trap electrons in the BE
range of 1.5-2.0 eV [29] is neglected). To account for possible small energy shifts (i.e.
typically smaller than 0.02 eV) of the reference spectrum of the bare metal, with respect to
that of the VB spectrum of the oxidized metal, the BE range of the bare metal VB spectrum
was allowed to shift during the iterative minimization procedure.
Fig. 3.4. Exemplary spectral reconstruction of an oxide-film VB spectra using the measured VB
spectra of the bare and the oxidized Zr(10 0) single-crystalline surface (oxidised for 7200 s at 325 K
and at pO2 = 1×10-4 Pa). Linear least squares minimization of the oxidic rest spectrum, as obtained
after subtraction of the scaled metallic contribution, was performed in the BE range from -5 to 3 eV.
See Section 3.2 for details.
3.3 Results and discussion
3.3.1 The oxide-film valence band spectra
The as-measured VB spectra of the oxidized Zr(0001) and Zr(10 0) single-crystal surfaces
and the resolved oxide-film VB spectra (cf. Fig. 3.4), are shown in Figs. 3.3 and 3.5,
respectively. The positions of the LVB and the UVB, separated by the intraband gap, have
1
1
54 Chapter 3
been indicated. Evidently, only the UVB of the (resolved) oxide-film VB spectra shows
pronounced differences in shape with increasing oxidation temperature (Fig. 3.5). It is known
that the UVB of zirconia is predominantly constituted of O 2p states with some admixing of
Zr valence states (in particular, the Zr 4d and Zr 5s states) [30-31]. At 300 K the oxide-film
UVB shows only one (broad) intensity maximum, whereas towards higher oxidation
temperatures (T ≥ 400 K), two peak maxima, around 6.2 eV and 8.5 eV, emerge in the oxide-
film UVB (see Fig. 3.5). The higher BE side of the UVB, around the developing (with
increasing oxidation temperature) peak maximum at 8.5 eV, has a more bonding character,
whereas the lower BE side of the UVB, around the developing (with increasing oxidation
temperature) peak maximum at 6.2 eV, is more non-bonding in nature [31].
Fig. 3.5. Series of reconstructed (cf. Fig. 3.4) oxide-film VB spectra of the bare and oxidized (a)
Zr(0001) and (b) Zr(10 0) single-crystalline surfaces. The positions of the lower (LVB) and UVB
have been indicated. The arrow indicates the direction of increasing oxidation temperature. See
Section 3.2 for details.
A similar emergence of a bonding/non-bonding UVB fine structure in the resolved
oxide-film UVB with increasing oxidation temperature was reported in a previous XPS VB
study on the thermal oxidation of Al metal [18]: at 373 K, an amorphous Al2O3 film with a
featureless oxide-film UVB spectrum develops, whereas a crystalline γ-Al2O3 film with a
pronounced bonding/non-bonding UVB fine structure forms at 773 K; the bonding/non-
bonding fine structure in the oxide-film UVB gradually emerges with increasing oxidation
temperature in the range of 373-773 K and is attributed to the gradual development of long-
range order in the Al2O3 films. Further, several studies [18-20, 23, 32] have demonstrated that
the fine structure of the XPS VB spectra can indeed be used as a sensitive fingerprint to
1
Valence-band and chemical state analyses of Zr and O in thermally-grown thin zirconium oxide films 55
distinguish crystalline and amorphous modifications, as well as distorted (e.g. by ion
bombardment [19, 32]) crystalline states, of oxides and semiconductors (e.g. Si and Ge). The
wide variation of bond configurations in a disordered phase gives rise to a broad and rather
featureless shape of the UVB spectrum (cf. Fig. 3.5 for T ≤ 400 K). Subsequent development
of long-range order results in more specific bond configurations, as reflected by the
emergence of distinct (sub-)bands of more bonding-like and more non-bonding-like character
in the UVB [18] (cf. Fig. 3.5 for T > 400 K).
Indeed, the Zr-oxide films grown at T ≤ 400 K (with rather featureless UVB spectra:
see Fig. 3.5) are found to be amorphous by in-situ LEED and RHEED analyses performed in
this work. At 425 K and 450 K, only two very diffuse spots can be distinguished in the
RHEED patterns, while still no diffraction spots can be discerned in the corresponding LEED
patterns (presumably due to roughness and/or charging of the oxidized surfaces [26]).
Although these RHEED patterns are much too diffuse to attribute these diffuse spots to any
of the known zirconia modifications (i.e. tetragonal, monoclinic and cubic), their appearance
suggests the precursor stages of an ordered (poly-)crystalline oxide phase. Indeed, the
appearance of these weak, diffuse diffraction spots in the RHEED patterns coincides with the
emergence of the bonding/non-bonding fine structure (i.e. two maxima) in the resolved UVB
spectra (see Fig. 3.5). The polycrystalline nature of the tetragonal ZrO2 oxide films formed
after t = 7200 s of oxidation at T = 450 K was also confirmed by HR-TEM [[33]].
As supported by first-principles molecular orbital calculations [31] and XPS VB
studies [21-22], the bonding/non-bonding fine structure of the UVB, that evolves with
increasing oxidation temperature, is characteristic for the development of a crystalline ZrO2
phase, particularly the tetragonal ZrO2 phase. Although monoclinic ZrO2 is the stable bulk
phase below 1300 K, the tetragonal ZrO2 phase can be thermodynamically preferred in nano-
sized systems due to its relatively low surface energy [34-36]. The cubic and monoclinic
ZrO2 phases have a less distinct UVB fine structure (as also holds for the amorphous oxide
films grown in this study at low temperatures; see above) owing to a coordination of the Zr
ions different from that for the tetragonal modification (in the cubic and monoclinic ZrO2
phases, the Zr ion has seven-fold coordination with O ions, whereas in the tetragonal form the
Zr ion has eight-fold coordination with O ions) [21-22, 31]. As evidenced by RISE
investigations performed in the same project [26, 37], the formation of (poly)crystalline oxide
films at oxidation temperatures of about 400 K and higher, as revealed by the present UVB
analysis, coincides with an increase of the oxide-film growth rate: at oxidation temperatures T
56 Chapter 3
< 375 K, the developing oxide-film attains a near-limiting thickness, whereas at T > 375 K
the oxide-film continuously thickens upon prolonged oxidation.
3.3.2 The local chemical states of O and Zr in the oxide films
The Zr MNN Auger line extends over the KE range from 85 eV to 150 eV and is constituted
of various peaks due to the different MNN transitions, as identified in Ref. [38]: see Fig. 3.2a.
Fig. 3.6. Wagner plots for the Zr cations in the
oxide layers grown on the Zr(0001) and Zr(10
0) surfaces by thermal oxidation for 7200 s
and at pO2 = 1×10-4 Pa in the temperature
range of 300-450 K. The family of dashed
diagonal lines with a slope of -1 represents
lines of constant according to Eq. 3.1.
The corresponding range of the AP values, as
reported in the NIST database, have also been
indicated [14]. The arrow indicates the
direction of increasing oxidation temperature.
See text for details.
1
Zrα
Valence-band and chemical state analyses of Zr and O in thermally-grown thin zirconium oxide films 57
In the present study, the three predominant Zr MNN Auger transitions (i.e. M45N1N23
at KE ≈ 93 eV, M45N23N23 at KE ≈ 118 eV and M45N23V at KE ≈ 149 eV; see Fig. 3.2a) each
in combination with the Zr 3d5/2 photoelectron line of the oxidic main peak (at BE ≈ 183 eV;
see Fig. 3.1a) were employed to investigate the (local) chemical state of the Zr and O species
in the oxide films grown for 7200 s at pO2 = 1×10-4 Pa at different oxidation temperatures
(ranging for 300 K to 450 K)1. To this end, Wagner plots for Zr were constructed by plotting
the measured KE of the Zr MNN Auger peak (either Ek(M45N1N23) or Ek(M45N23N23) or
Ek(M45N23V)) along the ordinate versus the corresponding BE of the Zr 3d5/2 oxidic peak
(Eb(Zr 3d5/2)) along the abscissa in reverse direction: see Figs. 3.6a, b and c.
A specific chemical state of Zr in the oxide film corresponds with a unique position in
the Wagner plot. The family of dashed diagonal straight lines with slope of -1 in Figs. 3.6a, b
and c represents the values of the modified AP of Zr, according to [13, 15]:
)3d(ZrMNN)(Zr 25bkZr EEα += . (3.1)
Analogously, such a Wagner plot can also be constructed for O in the oxide film by
plotting the measured KE of the O KL23L23 Auger peak, Ek(O KLL), versus the BE of the
predominant LBE O 1s peak, Eb(O 1s); see Fig. 3.7. The family of dashed diagonal straight
lines with slope of -1 in Fig. 3.7 then represents the values of the modified AP of O,
according to:
1s)(OKLL)(O bkO EEα += . (3.2)
As revealed by Figs. 3.6 and 3.7, the shifts of the core-level BEs with increasing oxidation
temperature are smaller than the shifts of the KEs of the respective Auger lines. This is a
direct consequence of the smaller final state relaxation energy for the single core-hole final
state in the photoemission process as compared to the double core-hole final state in the
Auger process [15]. Following the convention (cf. Ref. [15] and references therein), the total
final state relaxation energy, R, involved in the creation of a core-hole state in the
photoemission process can be written as R = Ra + Rea, where Ra and Rea represent the
contributions due to the atomic relaxation and extra-atomic relaxation (or polarization),
respectively [15].
The magnitude of Rea is determined by the electronic polarizability of the
neighbouring atoms (ligands) around the central (i.e. core-ionized) atom upon core-hole
formation [13, 15]. If two different chemical states of the same element are examined (and
1 For the bare Zr surfaces, the M45N1N23, M45N23N23 and M45N23V Auger lines are positioned at a KE of 94.2 eV, 117.5 eV and 149.3 eV, respectively. The Zr 3d5/2 main peak of the bare Zr surfaces is positioned at a BE of 179.2 eV.
58 Chapter 3
provided that the number of valence electrons of the core-ionized atom remains constant in
the final state of the photoemission process), it holds |∆Rea| >> |∆Ra|1. Then the corresponding
difference in the measured value of α for the two chemical states of the same element can be
related to the difference in Rea for the two chemical states by [13, 15]:
ea∆2∆ Rα ⋅= . (3.3)
Equation 3.3 explicitly expresses that any change in the local chemical state (i.e. the
nearest coordination sphere) of the core-ionized atom, resulting in a change of Rea, leads to a
change of the AP.
Fig. 3.7. Wagner plot for the O anions in the oxide layers grown on the Zr(0001) and Zr(10 0)
surfaces by thermal oxidation for 7200 s at pO2 = 1×10-4 Pa in the temperature range of 300-450 K.
The family of dashed diagonal lines with a slope of -1 represents lines of constant Oα according to
Eq. 3.2. The arrow indicates the direction of increasing oxidation temperature. See text for details.
The APs of Zr0 (i.e. unoxidized metallic state) as determined in this study are: 0Zrα
(M45N1N23) = 273.3 eV, (M45N23N23) = 296.7 eV and (M45N23V) = 328.5 eV; these
values are considerably higher than the APs, Zrα , for the oxide films (see Fig. 3.6). The
values of and Oα for Zr and O in the oxide films both decrease with increasing oxidation
temperature. For example, the value of for Ek(M45N1N23) decreases from about 273.2 eV
1 If the valence-electron charge around the core-ionized atom is the same in the initial (ground) state and in the final state [11], the relaxation process does not involve electron transfer from the surrounding atoms to the core-ionized atom (so-called non-local screening mechanism, for details see Ref. [15]).
1
0Zrα 0Zr
α
Zrα
Zrα
Valence-band and chemical state analyses of Zr and O in thermally-grown thin zirconium oxide films 59
at 300 K (with a corresponding total oxide-film thickness Ltot ≈ 1.4 nm [26, 37]) to 272.8 eV
at 450 K (with Ltot > 6.0 nm); the corresponding value of Oα decreases from about 1041.9 eV
at 300 K to 1041.2 eV at 450 K. Since the Zr M45N23V Auger process involves the Auger
emission of valence electrons of Zr, the number of valence electrons available for extra-
atomic relaxation is less. Consequently [15, 39], the change of upon increasing oxidation
temperature determined on the basis of the core-level M45N23V Auger transition is less
pronounced than the corresponding change of determined on the basis of the M45N1N23
and M45N23N23 Auger transitions (cf. Figs. 3.6c with a and b). Thus, the Zr M45N23V Auger
transition appears to be less suitable for chemical-state analysis and it is has been excluded
from the further discussion. Surprisingly, up to date, in the literature only the M45N23V Auger
transition has been used to determine an AP of Zr (cf. NIST database [14]).
Fig. 3.8. The relative shift of the AP of the Zr
cations, , (defined with respect to Zr0 in
the metal) versus the relative shift of the AP of
the O anions, , (defined with respect to
O2- in the H2O molecule) for the oxide films
grown on the Zr(0001) and Zr(10 0) surfaces
by thermal oxidation for 7200 s at pO2 = 1×10-
4 Pa in the temperature range of 300-450 K.
See text for details.
Zrα
Zrα
Zrα∆
Oα∆
1
60 Chapter 3
The Zr and O APs for the grown oxide films can be taken with respect to the AP
values for Zr in the metal (i.e. 0Zr Zr Zrα α α ∆ = − ) and for O2- in the H2O molecule (i.e.
2O O O in H Oα α α ∆ = − with 2O in H Oα = 1038.5 eV [40]), respectively. The thus-determined
values of Zrα∆ have been plotted versus those of Oα∆ in Fig. 3.8. It follows that Zrα∆ < 0
and that decreases with increasing oxidation temperature from -0.06 eV to -0.86 eV
(with similar shifts for the Zr M45N23N23 and Zr M45N1N23 peaks), because a core-hole in Zr
is more effectively screened in the metal than in the oxide. On the other hand, > 0 and
decreases with increasing oxidation temperature from 3.4 eV to 2.6 eV, which are
typical values for transition oxides [41]. Most strikingly, the changes of and with
increasing oxidation temperature are approximately equal (see straight lines with slope +1 in
Fig. 3.8), in contrast to results obtained for the oxidation of Al, where the metal and oxygen
AP shifts of the grown Al-oxide films (with increasing oxidation temperature) are much more
pronounced for the O anions than for the Al cations [16-17].
In ionic compounds the cations must be coordinated by anions (and vice versa), else
the structure is not stable. Hence, upon structural change in ionic compounds the type (anion
or cation) of the atoms surrounding the core-ionized atom is not expected to change, but the
coordination (sphere) of the core-ionized atom can change (not necessary the coordination
number). Thus, the simultaneous occurrence of (in this case comparable) shifts of and
with increasing oxidation temperature thus indicates concurrent changes in the first
coordination spheres of both Zr and O upon the development of long-range order in the oxide
phase (cf. Section 3.3.1). Structural changes in (initially amorphous) thermally grown Al2O3
films, on the other hand, proceed by the development of long-range order in the network of
edge-sharing [AlO4] and [AlO6] polyhedra, which results in a pronounced shift of only the O
AP (i.e. the value of the Al AP remains approximately constant, because the local chemical
environment around the Al cations remains largely unaffected) [16-17]. This suggests that
"building blocks" of Zr cations with well-defined first coordination spheres of O anions
(similar to the [AlO4] and [AlO6] polyhedra in the amorphous Al2O3 films) are absent in the
initially amorphous (see Section 3.3.1) Zr-oxide films grown on Zr. The (concurrent)
decreases of the Zr and O APs with increasing oxidation temperature imply a decrease of the
electronic polarizabilities (i.e. a decrease in Rea) of the first coordination spheres around the
Zrα∆
Oα∆
Oα∆
Zrα∆ Oα∆
Zrα∆
Oα∆
Valence-band and chemical state analyses of Zr and O in thermally-grown thin zirconium oxide films 61
Zr and O atoms, which is compatible with a gradual oxide-film densification in combination
with changes in the first coordination spheres (see above) [16-17].
The amorphous Zr-oxide films exhibit a pronounced deviation from the stoichiometric
composition adjacent to the Zr/oxide interface (see the suboxidic components, revealed by
the AR-XPS analysis in Fig. 3.1a and see Refs. [28, 42]): the films exhibit an O-deficiency,
which increases towards the Zr/oxide interface [26, 28, 42]. This suggests that the structural
disorder in the Zr-oxide films grown at low temperatures (i.e. T < 400 K, see Section 3.1) is
accompanied with compositional disorder. This is compatible with the above discussion.
Indeed, recognizing the presence of [AlO4] and [AlO6] polyhedra in amorphous Al2O3 films,
as studied in Refs. [16-17], the amorphous Al2O3 films are overall stoichiometric1.
The increase of Eb(Zr 3d5/2) and the simultaneous decrease of Ek(Zr MNN) with
increasing oxidation temperature (with respect to their corresponding values for Zr metal, cf.
Fig. 6) reveal an increase in ionicity of the chemical bonds in the grown oxide films [15],
which runs in parallel with the development of a crystalline oxide phase (see Section 3.3.1).
The experimental data points of Ek versus Eb for both Zr and O approximately fall on straight
lines with a slope of -3 (cf. Figs. 3.6 and 3.7), which is compatible with the theoretical
relationship between Ek and Eb as derived in Refs. [15]:
bMk 3-)](2[const EqkVE ⋅⋅+⋅+= , (3.4)
where VM denotes the Madelung potential (site or electrostatic self-potential) in the oxide, k
represents the change in core potential resulting from removal of a valence electron and q is
the valence charge of the considered atom in its initial (ground) state (see Refs. [15, 44] for
details). Since the shifts in Ek and Eb for Zr and O obey the linear relationship of Eq. 3.4 (cf.
Fig. 3.7), it follows that, although the value of the Madelung potential changes with
increasing oxidation temperature due to the increase of the ionicity of the chemical bond (see
above) and the development of long-range order in the oxide phase (see discussion in Section
3.3.1), the initial-state term in Eq. 3.4 remains approximately constant upon
increasing oxidation temperature, i.e. upon increasing structural (long-range) order. It could
be argued that the strength of the Madelung field is too weak to realize the formation of a
periodic arrangement of ions (i.e. the development of long-range order) in the oxide at low
oxidation temperatures. Only if the Madelung-field strength is increased by a local
redistribution of electronic charge at higher temperatures (as indicated by the increase of the
1 The relatively small suboxidic 'interface' contribution for the grown Al2O3 films (as evidenced by AR-XPS) arises from the deficient coordination of Al cations by nearest-neighbour O anions at the Al/Al2O3 interface [43].
[ ]MV k q+ ⋅
62 Chapter 3
ionicity of the Zr-O bonds, possibly accompanied with changes in local composition), does
its magnitude becomes high enough to induce the formation of a crystalline structure.
3.4 Conclusions
Initially amorphous oxide films thermally grown on Zr(0001) and Zr(10 0) surfaces exhibit
a gradual development of long-range order with increasing oxidation temperature in the range
of T = 300-450 K. At T ≤ 400 K, the oxide films remain predominantly amorphous, resulting
in a single-peak shape of the resolved oxide-film UVB spectra. At T > 400 K a bonding/non-
bonding UVB fine structure emerges, which is characteristic for the tetragonal ZrO2 phase.
Wagner plots constructed for the first time for Zr and O in zirconium oxide show
decreases of the Zr and O APs with increasing oxidation temperature, thereby exhibiting
concurrent changes in the first coordination spheres of both Zr and O in the oxide films. The
structural changes are due to a lowering of the electronic polarizability of the first
coordination sphere around the core-ionized Zr and O atoms in the oxide-films.
The Zr 3d5/2 BE increases and the Zr MNN KE decreases with increasing oxidation
temperature, thereby demonstrating an increase in ionicity of the Zr-O bonds in the oxide
with increasing oxidation temperature.
The present results reveal a remarkable difference in the development of the long-
range order in initially amorphous thin oxide films between Al-oxide films and Zr-oxide
films: whereas in Al2O3 films long-range order develops by the ordering of edge-sharing
[AlO 4] and [AlO6] polyhedral structural blocks already present in the amorphous oxide phase,
such "building blocks" do not occur in the amorphous Zr-oxide films, as demonstrated by the
concurrent changes in the coordination spheres of both Zr and O upon the development of
long-range order.
References
[1] C. S. Zhang, B. J. Flinn, and P. R. Norton, Surf. Sci. 264 (1992) 1.
[2] J. P. Pemsler, J. Electrochem. Soc. 105 (1958) 315.
[3] C. O. Degonzalez and E. A. Garcia, Appl. Surf. Sci. 44 (1990) 211.
[4] C. Stampfl, M. V. Ganduglia-Pirovano, K. Reuter, and M. Scheffler, Surf. Sci. 500
(2002) 368.
[5] S. Furuta, H. Matsuhashi, and K. Arata, Biomass & Bioenergy 30 (2006) 870.
1
Valence-band and chemical state analyses of Zr and O in thermally-grown thin zirconium oxide films 63
[6] B. M. Reddy, M. K. Patil, K. N. Rao, and G. K. Reddy, J. Mol. Catal. A 258 (2006)
302.
[7] N. Stojilovic, E. T. Bender, and R. D. Ramsier, Prog. Surf. Sci. 78 (2005) 101.
[8] C. L. Chang and S. Ramanathan, J. Electrochem. Soc. 154 (2007) G160.
[9] M. Gutowski, J. E. Jaffe, C. L. Liu, M. Stoker, R. I. Hegde, R. S. Rai, and P. J. Tobin,
Appl. Phys. Lett. 80 (2002) 1897.
[10] G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89 (2001) 5243.
[11] C. D. Wagner, L. H. Gale, and R. H. Raymond, Anal. Chem. 51 (1979) 466.
[12] C. D. Wagner, Anal. Chem. 44 (1972) 967.
[13] C. D. Wagner and A. Joshi, J. Electron Spectrosc. 47 (1988) 283.
[14] C. D. Wagner, A. V. Naumkin, A. Kraut-Vass, J. W. Allison, C. J. Powell, and J. R.
Rumble, NIST X-ray Photoelectron Spectroscopy Database. 2007, Measurement
Services Division of the National Institute of Standards and Technology (NIST)
Technology Services.
[15] G. Moretti, J. Electron Spectrosc. 95 (1998) 95.
[16] P. C. Snijders, L. P. H. Jeurgens, and W. G. Sloof, Surf. Sci. 589 (2005) 98.
[17] L. P. H. Jeurgens, F. Reichel, S. Frank, G. Richter, and E. J. Mittemeijer, Surf.
Interface Anal. 40 (2008) 259.
[18] P. C. Snijders, L. P. H. Jeurgens, and W. G. Sloof, Surf. Sci. 496 (2002) 97.
[19] J. M. Sanz, A. R. Gonzalezelipe, A. Fernandez, D. Leinen, L. Galan, A. Stampfl, and
A. M. Bradshaw, Surf. Sci. 309 (1994) 848.
[20] M. Gautier, J. P. Duraud, L. P. Van, and M. J. Guittet, Surf. Sci. 250 (1991) 71.
[21] D. Majumdar and D. Chatterjee, Thin Solid Films 206 (1991) 349.
[22] D. Majumdar and D. Chatterjee, Thin Solid Films 236 (1993) 164.
[23] R. P. Gupta, Phys. Rev. B 32 (1985) 8278.
[24] C. S. Zhang, B. Li, and P. R. Norton, Surf. Sci. 338 (1995) 157.
[25] C. S. Zhang, B. Li, and P. R. Norton, J. Nucl. Mater. 223 (1995) 238.
[26] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, J. Appl. Phys. 110 (2011)
024904.
[27] A. Lyapin and P. C. J. Graat, Surf. Interface Anal. 36 (2004) 812.
[28] A. Lyapin, L. P. H. Jeurgens, P. C. J. Graat, and E. J. Mittemeijer, J. Appl. Phys. 96
(2004) 7126.
[29] A. B. Sobolev, A. N. Varaksin, O. A. Keda, and A. P. Khaimenov, Phys. Stat. Solidi B
162 (1990) 165.
64 Chapter 3
[30] F. Zandiehnadem, R. A. Murray, and W. Y. Ching, Physica B & C 150 (1988) 19.
[31] M. Morinaga, H. Adachi, and M. Tsukada, J. Phys. Chem. Solids 44 (1983) 301.
[32] M. H. Brodsky and M. Cardona, J. Non-Cryst. Solids 31 (1978) 81.
[33] G. Bakradze, L. P. H. Jeurgens, U. Starke, T. Acartürk, and E. J. Mittemeijer, Acta
Mater. 59 (2011) 7498.
[34] R. C. Garvie, J. Phys. Chem.-US 82 (1978) 218.
[35] L. P. H. Jeurgens, Z. M. Wang, and E. J. Mittemeijer, Int. J. Mater. Res. 100 (2009)
1281.
[36] W. Qin, C. Nam, H. L. Li, and J. A. Szpunar, Acta Mater. 55 (2007) 1695.
[37] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, Surf. Interface Anal. 42 (2010)
588.
[38] M. Yamamoto, S. Naito, M. Mabuchi, and T. Hashino, J. Chem. Soc. Faraday Trans.
87 (1991) 1591.
[39] T. D. Thomas, J. Electron Spectrosc. 20 (1980) 117.
[40] G. Moretti, J. Electron Spectrosc. 58 (1992) 105.
[41] J. A. D. Matthew and S. Parker, J. Electron Spectrosc. 85 (1997) 175.
[42] L. P. H. Jeurgens, A. Lyapin, and E. J. Mittemeijer, Surf. Interface Anal. 38 (2006)
727.
[43] F. Reichel, L. P. H. Jeurgens, G. Richter, and E. J. Mittemeijer, J. Appl. Phys. 103
(2008) 093515.
[44] W. F. Egelhoff, Surf. Sci. Rep. 6 (1987) 253.
Chapter 4
An STM study of the initial oxidation of single-
crystalline zirconium surfaces
Georgijs Bakradze, Lars P.H. Jeurgens and Eric J. Mittemeijer
Abstract
The microstructural development of very thin (thickness < 10 nm) oxide layers grown on Zr surfaces
by thermal oxidation was investigated by in-vacuo STM and XPS. To this end, single-crystalline
Zr(0001) and Zr(101 0) surfaces were prepared under UHV conditions by a cyclic treatment of ion-
sputtering and in-vacuo annealing steps and then exposed to dry O2(g) in the temperature range of
300-450 K (at pO2 = 1×10-4 Pa). Oxidation proceeds by the very fast formation of a dense
arrangement of tiny oxide nuclei, which cover the entire Zr surface. The initial oxide cluster size is
about 2.0±0.5 nm. The transport processes of adsorbed O species and/or Zr species on the oxidizing
surface become promoted with increasing temperature and thereby the oxide clusters rearrange into
bigger agglomerates with increasing oxidation time. At the same time, a long-range atomic order
develops in the oxide overgrowths, as evidenced from the emergence of a bonding/non-bonding fine
structure in the resolved oxide-film upper VB, as measured in-situ by XPS.
4.1 Introduction
The dry oxidation of metals by molecular oxygen is still among the most extensively (both
experimentally and theoretically) studied heterogeneous gas-solid reactions [1-5]. Thin oxide
films, as formed on metallic (or semi-conductor) substrate surfaces in an oxidizing
environment, influence important properties of such components, such as the corrosion
resistance, the thermal stability, the catalytic activity and the electrical, adhesive and
tribological properties.
The majority of initial stage oxidation studies of metallic surfaces have hitherto
focused on: (i) the determination of oxidation-rate laws [3, 6] (i.e. the oxidation kinetics) as a
function of the oxidation conditions, and on (ii ) the characterization of the developing oxide
(micro)structure and the chemical constitution [7-10]. The information on the
amorphous/crystal structure of very thin (< 10 nm) oxide films, as formed on metal and alloy
surfaces under oxidizing conditions at low temperatures (up to, say, 500 K), is usually
66 Chapter 4
derived from observations in reciprocal space using diffraction techniques: e.g. LEED [11-
12], surface X-ray diffraction [13], transmission electron diffraction [14-15]. On the other
hand, the chemical constitution of such thin oxide films is typically determined by
spectroscopic techniques, such as XPS or Auger-electron spectroscopy (see e.g. [11-13]).
Although the microstructural characteristics of the developing oxide film play a crucial role
for many important material properties (see above), the above mentioned diffraction and
spectroscopic techniques do not provide direct information on the developing oxide
morphology.
Direct microscopic imaging, in real space and down to the atomic scale, of the surface
morphology of very thin oxide films can typically only be performed by TEM [10, 16] or
STM. Hence, since the advent of STM in 1981, numerous STM studies on the initial stage of
thermal oxidation of metallic surfaces have been published, e.g. for Al(111) [17], Ni3Al(110)
[18], Cr(110) [19-20], Cu(111) [21-23], Mg(0001) [24], W(110) [25], Pt(110) [26], Fe(111)
[27], Fe(110) [28] and Rh(111) [29]. These studies have shown that the initial stages of
oxidation of metallic surfaces typically proceed by a series of consecutive, but overlapping
steps, such as (i) oxygen chemisorption, (ii ) oxide nucleation and (iii ) continued growth of
the oxide nuclei until the entire metal surface has been covered with oxide (i.e. a "closed"
oxide film has been formed). At low oxidation temperatures, after an initial, very fast oxide-
film growth regime, the transport rate of the migrating species through the closed oxide film
becomes negligible upon continued exposure to oxygen, which results in a so-called
passivating behaviour (see Chapter 2, Refs. [30-31] and Section 4.3.1 below). At higher
oxidation temperatures, on the other hand, the initial, fast oxide-film growth regime is
generally succeeded by a stage of slower oxidation corresponding with a continuous
thickening of the insulating oxide film [31-33]. STM investigations at such later stages of the
oxidation process are often complicated by experimental problems, such as unstable
tunnelling contacts (in particular for thick insulating oxide films) [34].
To our knowledge, no STM studies on the thermal surface oxidation of Zr or its alloys
have been presented so far. Comprehensive knowledge on the oxidation behaviour of
zirconium surfaces is required in many technological areas, such as corrosion protection [35],
heterogeneous catalysis/electrochemistry [36-37] and microelectronics [38-39]. Against this
background the present in-vacuo STM study addresses the initial stages of the dry, thermal
oxidation of bare single-crystalline Zr(0001) and Zr(101 0) surfaces in the temperature range
of 300-450 K and at a partial oxygen gas pressure of pO2 = 1×10-4 Pa. As demonstrated in
this work, the STM observations on the evolving surface topography of the oxide
An STM study of the initial oxidation of single-crystalline zirconium surfaces 67
overgrowths on the Zr(0001) and Zr(101 0) surfaces are complementary to earlier reported
results on: (i) the oxide-film growth kinetics, as obtained by RISE [30, 32], and (ii ) the
evolution of the oxide-film microstructure, as derived from oxide-film VB studies by XPS
[40-43].
4.2 Experimental
Disc-shaped Zr(0001) and Zr(101 0) single crystals were cut (diameter 6 mm; 1 mm thick;
orientation alignment within ±0.5° of the nominal surface plane) from a single-crystalline
unalloyed α-Zr rod and subsequently single-side mechanically polished (last step 0.05 µm).
Main impurities in the polished specimens, as identified by inductively coupled plasma
optical emission spectroscopy analysis, are (in mass parts): Hf (60 ppm); Fe (25 ppm); Ti (1
ppm); Cu, Zn, Mn, Ca, Na (< 2 ppm).
The polished specimens were introduced into a combined UHV system, possessing
facilities for SC and XPS and STM analyses (base pressure < 5×10-8 Pa). Next, the (native)
oxide and other adventitious contaminants (mainly C) on the surface were removed by SC,
first with 3 keV Ar+ ions and subsequently with 1 keV Ar+ ions (see what follows), rastering
the entire surface area, until no other element than Zr was detected in a measured XPS survey
spectrum (see below). Roughening of the ion-bombarded single-crystalline surfaces due to
local differences in the sputter yield by ion channelling and shadowing effects [44] was
suppressed by employing continuous specimen rotation at a speed of about 2 rpm during the
SC treatments. The SC was performed with 3 keV Ar+ ions until all C surface contamination
was removed (as verified by XPS); all subsequent SC treatments (e.g. to remove remaining O
contamination) were performed with 1 keV Ar+ ions. Next, the specimen and specimen
holder were outgassed by a cycling treatment of alternating 1 kV Ar+ SC and in-vacuo
annealing steps, while gradually increasing the specimen temperature during each successive
in-vacuo annealing step up to 1000 K. As a final surface preparation step prior to each
oxidation experiment, the SCed surfaces were in-vacuo annealed at 1000 K for 300-600 s.
The SC treatment destroys the crystallinity at the specimen surface and also induces
sputter-induced surface roughness (see Section 4.3.1). However, the surface crystallinity in
the ion-bombarded surface region is fully restored (as verified by in-situ LEED in Ref. [31])
and the surface becomes atomically flat (as evidenced by in-situ STM in this work; see
Section 4.3.1) during the final in-vacuo annealing step at 1000 K for 300-600 s. The single-
68 Chapter 4
crystalline Zr(0001) and Zr(101 0) surfaces, as obtained after the final in-vacuo annealing
step, are further designated as bare Zr surfaces.
Next, oxide films were grown at 300, 375 and 450 K by exposure of the bare Zr(0001)
and Zr(101 0) surfaces for different times to pure oxygen gas (purity ≥ 99.9999 vol.% with a
specified residual gas content of H2O ≤ 0.5 vpm, N2 + Ar ≤ 2.0 vpm, CnHm ≤ 0.1 vpm and
CO2 ≤ 0.1 vpm) at pO2 = 1×10-4 Pa. The oxidation temperature was measured with a type K
thermocouple, which was put in direct mechanical contact with the single-crystal surface.
XPS was applied to determine the chemical constitution of the SCed, annealed and
oxidized surfaces. The XPS analysis was performed with a Thermo VG Thetaprobe system
employing monochromatic Al Kα radiation (hν = 1486.68 eV). XPS survey spectra, covering
a BE range from 0 to 1200 eV, were recorded with a step size and pass energy of 0.2 eV and
of 200 eV, respectively.
The STM studies of the bare and oxidized Zr surfaces (after cooling down to room
temperature) were performed in an UHV side-chamber (base pressure < 8×10-9 Pa), which is
interconnected to the UHV chambers for in-situ oxidation and in-situ AR-XPS analysis. The
STM investigations were performed with a Specs Aarhus 150 Scanning Tunnelling
Microscope (Createc edition), operated in constant tunnelling current (topographic) mode, as
controlled by the Specs SPC-260 electronics. "Imaging" (i.e. mapping) was performed at
room temperature by scanning the surface employing a steady state positive specimen bias
voltage (Vt) in the range of 300-2700 mV and a constant tunnelling current (I t) in the range of
0.2-1.5 nA (see figure captions for the values of It, Vt and the maximum height difference,
∆Z; in the images shown the scanning lines are parallel to the vertical directions of the
images). Before the STM investigation, the shape of the electrochemically etched tungsten tip
of the STM was cleaned in-situ by sputtering for 900 s with a parallel (defocused) 3 keV Ar+
beam (total sputter current of about 1.5 µA). Also during the subsequent STM investigation
(i.e. while scanning the specimen surface), the tip was repeatedly cleaned and sharpened by
shortly pulsing (for about 0.1 s) of the specimen bias voltage up to about 10 V (on specimen
areas different from those investigated). The electronic properties of the oxidized surfaces
were investigated by measuring I t-Vt curves at selected positions on the specimen surface (see
results in Section 4.3.1). The obtained STM images were post-processed (using standard
scanning probe microscopy filters, i.e. background subtraction, flatten etc.) with the WSxM
4.0 software [45]
An STM study of the initial oxidation of single-crystalline zirconium surfaces 69
4.3 Results and discussion
4.3.1 Oxide-film microstructure at T = 300-450 K
Exemplary STM images, as recorded from the bare (i.e. SCed and annealed) Zr substrates
after different stages of the in-vacuo specimen preparation procedure (see Section 4.2) are
shown in Figs. 4.1a-c. Evidently, the rough Zr surface, as obtained after the initial SC
treatment (see Fig. 4.1a), becomes atomically flat and single-crystalline by the final in-vacuo
annealing step at 1000 K for 300-600 s: see Fig. 4.1b and see corresponding LEED patterns
of the bare Zr surfaces, as presented in Ref. [31]. The measured individual step heights of
about 2.5 Å between the atomically flat terraces on the bare Zr(101 0) surface (see Fig. 4.1b)
closely match the (ideal) interplane distance of 2.8 Å along the α-Zr[101 0] direction and thus
represent mono-atomic steps. The striped appearance of the bare Zr(101 0) terraces (see Fig.
4.1b) is due to a 1×4 reconstruction of the (101 0) crystal surface of α-Zr, as reported in Ref.
[12].
Fig. 4.1. STM images (see Section 4.2) as recorded in-situ from: (a) the Zr(101 0) surface after SC
with 3 keV Ar+ for 3 hrs; (b) the SCed Zr(101 0) surface after annealing at 1000 K for 900 s; (c) the
Zr(0001) surface (i.e. SCed and annealed as in (b)) after residing for 1800 s under UHV conditions.
The adsorbed contaminant atoms (predominantly oxygen) form triangled structures (see inset). The
images were recorded with the following constant specimen tunneling bias (Vt) and tunneling current
(It); (a) Vt = 1806 mV, It = 0.240 nA, ∆Z = 19.2 nm; (b) Vt = 362 mV, It = 0.330 nA, ∆Z = 0.5 nm; (c)
Vt = 2087 mV, It = 0.250 nA, ∆Z = 0.7 nm; insert in (c) Vt = 670 mV, It = 1.320 nA, ∆Z = 0.2 nm.
The bare Zr surfaces are highly reactive and therefore easily contaminate (typically
within 1800 s) with rest-gas species in the UHV chamber (e.g. CO and H2O), even at a base
pressure as low as 1×10-8 Pa (as shown by STM and XPS). An STM image of an adsorbate-
70 Chapter 4
contaminated bare Zr(0001) surface (in-situ XPS analysis indicated that O is the main surface
contaminant) is shown in Fig. 4.1c.
As demonstrated by RISE [30-31], exposure of the bare single-crystalline Zr surfaces
to pure O2(g) at pO2 = 1×10-4 Pa at a constant substrate temperature in the range of 300-450
K results in the very fast (i.e. within 500 to 1900 s) formation of a closed oxide film, which
attains a near-limiting thickness at T < 375 K. The total oxide-film thickness, as attained after
7200 s of oxidation (relevant to the STM images shown in Figs. 4.2 and 4.3 discussed below)
increases with increasing oxidation temperature from about 1.3 nm at T = 300 K to 9.2 nm at
T = 450 K: see Table 4.1. For T > 375 K, the more open Zr(101 0) prism plane oxidizes more
readily than the densely packed Zr(0001) basal plane, in particular at the higher temperatures
[30-31].
Representative STM images of the Zr(0001) and Zr(101 0) surfaces at 300 K and 375
K after 7200 s of oxidation (i.e. after reaching near-limiting oxide-film thicknesses) are
shown in Figs. 4.2a to d. STM images of the Zr(0001) and Zr(101 0) surfaces at 450 K for t =
7200 s (corresponding to the highest oxidation temperature applied in the present study) are
shown in Figs. 4.3e and j, respectively.
Table 4.1: Average lateral oxide cluster size (d) as determined from the recorded STM images of the
oxidized Zr(0001) (basal) and Zr(101 0) (prism) surfaces oxidized for different times, t, at various
temperatures, T (see Figs. 4.2 and 4.3). The total oxide layer thickness Ltot was obtained by RISE [30-
31].
T = 300 K 375 K 450 K
t (s) 7200 s 7200 s 300 s 600 s 1200 s 2400 s 7200 s
basa
l d (nm) 1.6 2.0 1.8 2.2 2.8 3.9 3.2
Ltot (nm) 1.3 2.6 1.5 1.9 2.5 3.5 6.5
pris
m d (nm) 1.6 2.0 1.9 2.3 2.8 3.6 4.9
Ltot (nm) 1.4 2.9 2.5 3.0 4.3 6.0 9.2
Evidently, the STM images of the bare and oxidized surfaces (cf. Fig. 4.1b and 4.2c
for the Zr(101 0) surface) are strikingly different. The atomically-flat terraces, characteristic
for the bare Zr surfaces (Fig. 4.1b), are covered with irregular protrusions after oxidation. The
An STM study of the initial oxidation of single-crystalline zirconium surfaces 71
protrusions on both oxidized surfaces are constituted of very small oxide clusters, as
indicated in Fig. 4.2a. The values for the average lateral sizes of the oxide clusters at t = 7200
s for various oxidation temperatures in the range of T = 300-450 K have been gathered in
Table 4.1.
Fig. 4.2. STM images (see Section 4.2) as recorded in-situ from the Zr(0001) (left panels a and b) and
Zr(101 0) (right panels c and d) surfaces after oxidation for 7200 s at pO2 = 1×10-4 Pa at T = 300 K
(upper panels a and c) and 375 K (lower panels b and d). The images were recorded with the
following constant specimen tunnelling bias (Vt) and tunnelling current (It): (a) Vt = 2837 mV, It =
0.640 nA, ∆Z = 1.0 nm; (b) Vt = 593 mV, I t = 0.400 nA, ∆Z = 1.5 nm; (c) Vt = 2152 mV, It = 0.210
nA, ∆Z = 0.8 nm; (d) Vt = 2894 mV, It = 0.790 nA, ∆Z = 1.8 nm.
72 Chapter 4
Similar irregular oxide morphologies have been observed by STM for oxidized
Cr(110) surfaces (after 0.75 L O2-exposure at 300 K [20]), oxidized Cu(111) surfaces (after
970 L O2-exposure at 300 K [21]), oxidized Fe(111) surfaces (after 400 L O2-exposure at 300
K [27]) and oxidized Cu0.7Zn0.3(111) surfaces (after O2-exposures up to 1280 L at 300 K
[46]). Such small oxide clusters have been previously designated as "grains of the oxide film"
[20], "oxide islands" [27, 46] or "adsorbed oxygen (clusters)" [46]. Although the average of
size of these initial oxide clusters (i.e. with an average diameter of about 2.0±0.5 nm, see
Table 1 and Section 4.3.2) by far exceeds the lattice parameter of zirconium or zirconia, the
atomic structure of the clusters could not be resolved by STM (as performed at RT), as also
holds for the STM analysis of similar oxide structures, as formed during RT oxidation of
Cr(110), Fe(110) and Ni surfaces [17, 27].
The designation "oxide grains" (cf. Ref. [20]) implicitly assumes a crystalline
structure of the oxide overgrowth. However, the oxide layers, as grown on Zr(0001) and
Zr(101 0) surfaces at T < 400 K (see Fig. 4.2), are amorphous as demonstrated by LEED,
RHEED and XPS VB studies (see Section 4.3.2 and Ref. [47]). The term "oxide islands" (cf.
Refs. [27, 46]), suggests the existence of bare patches of metal surface in-between the "oxide
islands". However, I t-Vt-measurements performed in this work after 7200 s of oxidation in the
range of 300-450 K (see Section 4.2) of the oxide clusters and of areas in-between the oxide
clusters both displayed insulating behavior. This indicates that the Zr(0001) and Zr(101 0)
surfaces at this stage of oxidation have been fully covered by oxide, in accordance with the
drop of the oxidation rate associated with the occurrence of a near-limiting oxide-film
thickness at T < 375 K [30].
The in-situ XPS analysis1, performed on the surfaces of the oxidized specimens, only
evidences the presence of oxidic states of Zr and O, indicating that the observed small
protrusions do not pertain to chemisorbed O species (as referred to in Ref. [46]). Therefore,
the small protrusions are denoted as "oxide clusters" (or "oxide nuclei", see Section 4.3.2).
The density of the oxide clusters is, on average, slightly lower on the oxidized
Zr(0001) surface than on the oxidized Zr(101 0) surface for the same oxidation conditions at
T = 300 K, 375 K and 450 K (see Figs. 4.2a,c and Fig. 4.3). For T ≤ 375 K, the average oxide
cluster size (as indicated by the average cluster diameter d) after t = 7200 s is very similar for
both surfaces; the average oxide cluster diameter for t = 7200 s increases from about d =
1.6±0.5 nm at T = 300 K to d = 2.0±0.5 nm at T = 375 K (independent of the substrate
1 The in-situ STM and XPS analyses of the oxidized Zr surfaces were performed under UHV conditions (i.e. after evacuating the O2 gas; see Section 4.2).
An STM study of the initial oxidation of single-crystalline zirconium surfaces 73
orientation): see Table 4.1 and Fig. 4.2. At T = 450 K, on the other hand, the oxide cluster
size for t = 7200 s is, on average, larger for the oxide overgrowths on the Zr(101 0) surface:
i.e. d = 3.2±0.5 nm for Zr(0001) and d = 4.9±0.5 nm for Zr(101 0) (see Table 4.1).
An average oxide cluster size in the range of 2.0-5.0 nm, as determined for the
oxidized Zr surfaces at t = 7200 s (corresponding to an O2-exposure of 5400 L) for T = 300-
450 K, is comparable to reported oxide cluster sizes of 4.0±0.5 nm, for a Cr(110) surface
after 80 L O2-exposure at 300 K [20], and of 3.0-4.0 nm [28], for a Fe(110) surface after 80 L
O2-exposure at 300 K.
The thermally activated transport processes of adsorbed O species and/or Zr species
on the oxidizing surface become promoted with increasing temperature and, consequently,
upon continued oxidation the oxide clusters coalesce into larger agglomerates, as exhibited
for increasing oxidation temperature at constant oxidation time of 7200 s (see Figs. 4.2a,b,d
and 4.3). Smaller oxide particles, encompassed by a surface of generally smaller radius of
curvature, have, according to the Gibbs-Thompson relationship, a higher Gibbs energy than
larger ones; thereby, a driving force for oxide-cluster coarsening, development of
"agglomerates" by restructuring/reorientation, exists. The agglomerates do not show any
specific regularities in shape and have different sizes; their average size is typically larger for
the oxidized Zr(0001) surface than for the oxidized Zr(101 0) surface (cf. Figs. 4.2b, d and
Table 4.1).
The oxide agglomerates eventually constitute the oxide grains of the polycrystalline
oxide layer that evolves upon prolonged oxidation at 450 K. The reduction of the total oxide-
surface/interface area upon oxide-cluster agglomeration coarsens the oxide microstructure,
which can cause a change of governing atomic transport mechanism (i.e. bulk vs. GB
transport) upon oxide growth, as indeed reported for the oxidation of Zr surfaces at 450 K in
Ref. [31].
The fast occurrence of a limiting oxide-film thickness at T < 375 K (see above
discussion and Ref. [30]) evidently hinders a meaningful STM investigation of the successive
stages of development of the oxide-layer microstructure with increasing time. At T ≥ 450 K,
the retardation of the oxide-film growth rate, after the initial, fast oxidation regime, is much
less pronounced and, instead, the oxide film grows continuously (i.e. a near-limiting oxide-
film thickness is no longer established) [30]. A comparative STM study of the successive
stages of development of the oxide-layer microstructure at T = 450 K is thus possible, which
is presented in Section 4.3.2.
74 Chapter 4
4.3.2 Evolution of the oxide microstructure at 450 K
The evolution of the oxide-layer microstructure, as monitored by STM, for successive
oxidation times, t = 300, 600, 1200, 2400 and 7200 s, at T = 450 K is depicted for the
Zr(0001) surface in Figs. 4.3a-e (i.e. left column of Fig. 4.3) and for the Zr(101 0) surface in
Figs. 4.3f-j (i.e. right column of Fig. 4.3).
Fig. 4.3. STM images (see also the facing page) as recorded in-situ from the (a-e) Zr(0001) and (f-j)
Zr(101 0) surfaces after oxidation at 450 K for 300 s, 600 s, 1200 s, 2400 s and 7200 s (pO2 = 1×10-4
Pa): (a) Vt = 2219 mV, It = 0.330 nA, ∆Z = 1.4 nm; (b) Vt = 1770 mV, It = 0.110 nA, ∆Z = 1.5 nm; (c)
Vt = 1317 mV, It = 0.180 nA, ∆Z = 2.8 nm; (d) Vt = 2219 mV, It = 0.610 nA, ∆Z = 2.2 nm; (e) Vt =
2837 mV, It = 0.190 nA, ∆Z = 4.1 nm; (f) Vt = 2532 mV, I t = 0.750 nA, ∆Z = 1.2 nm; (g) Vt = 2668
mV, It = 0.270 nA, ∆Z = 1.0 nm; (h) Vt = 2587 mV, It = 0.200 nA, ∆Z = 1.4 nm; →→→→
An STM study of the initial oxidation of single-crystalline zirconium surfaces 75
→→→→ (i) Vt = 2588 mV, It = 0.730 nA, ∆Z = 4.4 nm; (j) Vt = 2152 mV, It = 0.990 nA, ∆Z = 3.3 nm.
76 Chapter 4
Note that these oxidation experiments have not been performed in a cumulative
manner: i.e. for each oxidation time a freshly prepared, bare Zr substrate was utilized (see
Section 4.2 for details). Consequently, the recorded STM images for one crystal plane after
different oxidation times do not represent the same location of an oxidized surface.
After the first 300 s of oxidation at T = 450 K (equivalent to 0.75 L O2-exposure at
pO2 = 1×10-4 Pa), the bare Zr(0001) and Zr(101 0) surfaces are (already, cf. results for 7200 s
at oxidation temperatures as low as 300 K, discussed in Section 4.3.1) densely covered with
small oxide clusters (see Figs. 4.3a, f and Table 4.1). The lateral size of the oxide clusters
gradually increases with increasing oxidation time: see Table 4.1 and Fig. 4.3. The initial
lateral size of the oxide cluster is comparable with the overall oxide-film thickness, Ltot, as
determined by RISE [30-31] and reported in Table 4.1. At these low O2-exposures (e.g. of
0.75 L; see Figs. 4.3a and 4.3f), the distribution of the lateral size of the oxide clusters is very
narrow, which hints at either a high oxygen sticking coefficient (a low activation-energy
barrier for oxide nucleation on the bare Zr surface) and/or a high activation-energy barrier for
their further growth. (Indeed an oxygen sticking coefficient of unity has been reported for the
bare Zr(0001) surface [48-49]). The oxide clusters thus are interpreted as oxide nuclei that
have formed on the bare Zr surfaces at the onset of the oxidation process. The oxide
nucleation on the terraces appears more homogeneous (random) on the bare Zr(0001) surface
than on the bare 1×4-reconstructed Zr(101 0) surface (compare Fig. 4.3a and 4.3f).
As evidenced by the recorded STM images after 0.75 L O2-exposure (see Figs. 4.3a
and 4.3f), the consecutive processes of oxide nucleation, growth and coalescence to form a
laterally-closed oxide layer are completed within 300 s of O2-exposure at T = 450 K and pO2
= 1×10-4 Pa.1
Upon continued oxidation (i.e. t > 600 s), the oxide clusters gradually
restructure/reorient into bigger agglomerates: see Figs. 4.3a-e and Figs. 4.3f-j. The evolving
oxide agglomerates have a characteristic lateral size of about 8 nm on the basal plane for t =
600 s at T = 450 K (see Fig. 4.3b), which by far exceeds the corresponding oxide-film
thickness of 1.9 nm (as determined by RISE: see Table 4.1). For example, only three
agglomerates are visible in the scanned STM area of 50 nm × 50 nm, as recorded from the
oxidized basal plane after 2400 s of oxidation (see Fig. 4.3d). The agglomeration process
represents a gradual coarsening of the oxide structure, accompanied with an increase of the
1 This observation is compatible with the preceding drop (at Ltot = 1.5 and 2.5 nm for the Zr(0001) and Zr(10-10) surfaces, respectively) of the oxide-film growth rate [5, 48], as measured by RISE [28, 42].
An STM study of the initial oxidation of single-crystalline zirconium surfaces 77
oxide-film roughness (cf. an increase of the ∆Z values with increasing t; see caption of Fig.
4.3).
Fig. 4.4. UVB spectrum of the oxide overgrowths, as resolved from the measured XPS spectra of the
Zr(101 0) surface, after oxidation at T = 450 K and pO2 = 1×10-4 Pa for oxidation times of t = 300 s, t
= 600 s, t = 1800 s and t = 3600 s. The arrow points in the direction of increasing oxidation time. For
details on the spectral evaluation procedure, see Refs. [43, 47].
The UVB region of the grown oxide films, as resolved from the measured XPS
spectra of the Zr(101 0) surface (for details, see Refs. [43, 47]) after oxidation for successive
oxidation times (i.e. 300, 600, 1800 and 7200 s) at T = 450 K and pO2 = 1×10-4 Pa, is shown
in Fig. 4.4. The UVB region of zirconia is predominantly constituted of O2p states with some
admixing by Zr (4d and 5s) states [50-51]. Evidently (see Fig. 4.4), the UVB region of the
oxidized Zr(101 0) surface exhibits pronounced differences in shape with increasing
oxidation time at T = 450 K: i.e. for t = 300 s the oxide-film UVB is rather flat and
featureless, whereas for t = 7200 s two distinct peak maxima at BE ≈ 6.2 eV and BE ≈ 8.5 eV
have emerged. The higher BE side of the UVB, around the developing peak maximum of 8.5
eV, has a more bonding character, whereas the lower BE side of the UVB, around the
evolving peak maximum of 6.2 eV, has a more non-bonding character [51]. As discussed in
e.g. Refs. [42-43, 47, 52-54], a rather broad, featureless structure of the UVB is characteristic
for a disordered (amorphous) phase with a relatively broad distribution of chemical bonding
78 Chapter 4
configurations. Indeed at T < 400 K for t = 7200 s [47], as well as for short oxidation times at
T = 450 K (see Fig. 4.4), the oxide films are still predominantly amorphous (as demonstrated
by LEED, RHEED and XPS VB studies [47]). The gradual development of long-range order
(and thus of specific bond configurations) in the oxide overgrowth, with increasing oxide
time at 450 K, results in the appearance of distinct bonding and non-bonding features in the
UVB region: see Fig. 4.4. Hence a polycrystalline oxide film develops upon prolonged
oxidation at 450 K, in accordance with HR-TEM (see Chapter 5 or Ref. [55]) and RHEED
(see Chapter 3 or Ref. [47]) analyses.
As follows from the above discussion, the development of long-range order in the
oxide overgrowth with increasing oxidation time at T = 450 K runs parallel with the
rearrangement of the oxide clusters into larger oxide agglomerates. Hence, the development
of long-range order in the oxide overgrowths is accompanied with the coarsening of the
oxide-film microstructure, as driven by the Gibbs-Thomson effect (see Section 4.3.1). The
initial oxide clusters, i.e. before agglomeration, have a lateral size in the range of 1.6-4.9 nm
and a height of about the thickness of the film; thus, a single cluster comprises about 2500
atoms. Apparently, such a small oxide-cluster volume, characterized by a high surface-to-
volume ratio, obstructs the development of long-range order (but short-range order is
possible). It could be argued that the strength of the Madelung field in such small oxide
clusters is too weak to realize the formation of a periodic arrangement of ions (i.e. the
development of long-range order) (for details see Chapter 3 or Ref. [47]). Although the
microstructure of the oxide overgrowth, as investigated by STM after t = 7200 s at T = 450 K
(cf. Figs. 4.3i and 4.3j), does not reveal distinct facets, indicative of the formation of a well-
defined crystalline oxide phase, the boundaries between the oxide agglomerates at e.g. t =
2400 s can be conceived as the GBs in the evolving polycrystalline oxide layer, as identified
by cross-sectional HR-TEM in Fig. 5.3 (see Chapter 5 or Ref. [55]); the densities of the
boundaries between the oxide agglomerates (STM) and of the GBs (HR-TEM) are
comparable.
4.4 Conclusions
Exposure of the bare Zr surfaces to pure O2(g) at substrate temperatures in the range of 300-
450 K and at pO2 = 1×10-4 Pa leads to the initial, very fast formation of a dense arrangement
of small oxide clusters/protrusions; in between the protrusions also oxide is present: the
whole surface is covered with oxide. The consecutive processes of oxide nucleation, growth
and coalescence, leading to a "laterally-closed" oxide layer, have completed within t = 300 s
An STM study of the initial oxidation of single-crystalline zirconium surfaces 79
of O2-exposure (at T = 300-450 K and at pO2 = 1×10-4 Pa). The average lateral size of the
oxide clusters increases gradually with increasing oxidation time at constant oxidation
temperature and with increasing oxidation temperature at constant oxidation time. The
average lateral size of the oxide clusters after 7200 s of oxidation at T = 300-450 K is in the
range 2.0-5.0 nm (dependent on t and T).
The transport processes of adsorbed O species and/or Zr species on the oxidizing
surface become promoted with increasing temperature, thereby promoting the
restructuring/reorientation of the oxide clusters into bigger agglomerates, e.g. with increasing
oxidation time at constant temperature, as driven by the Gibbs-Thomson effect.
At T < 400 K for t = 7200 s, as well as for shorter oxidation times at T = 450 K, the
oxide films are predominantly amorphous, because no long-range order can develop in the
oxide clusters having confined, small oxide-cluster volumes. Long-range order in the oxide
overgrowths can only develop in the larger oxide agglomerates, leading to the emergence of a
characteristic fine structure in the resolved oxide-film UVB spectrum as measured by XPS.
The boundaries between the evolving oxide agglomerates are the GBs in the evolving
polycrystalline oxide layer.
References
[1] D. A. King and D. P. Woodruff, ed. The Chemical of Solid Surface and
Heterogeneous Catalysis (1990) Elsevier: Amsterdam.
[2] A. T. Fromhold, Theory of Metal Oxidation, North-Holland, Amsterdam (1976).
[3] E. Fromm, Kinetics of Metal-Gas Interactions at Low-Temperatures, Springer, Berlin
(1998).
[4] L. P. Bonfrisco and M. Frary, J. Mater. Sci. 45 (2010) 1663.
[5] G. W. Zhou, Appl. Phys. Lett. 94 (2009) 201905.
[6] F. P. Fehlner and N. F. Mott, Oxid. Met. 2 (1970) 59.
[7] F. Reichel, L. P. H. Jeurgens, and E. J. Mittemeijer, Acta Mater. 56 (2008) 2897.
[8] F. Reichel, L. P. H. Jeurgens, and E. J. Mittemeijer, Surf. Interface Anal. 40 (2008)
281.
[9] F. Reichel, L. P. H. Jeurgens, G. Richter, P. A. v. Aken, and E. J. Mittemeijer, Acta
Mater. 55 (2007) 6027.
[10] F. Reichel, L. P. H. Jeurgens, G. Richter, and E. J. Mittemeijer, J. Appl. Phys. 103
(2008) 093515.
[11] Y. M. Wang, Y. S. Li, and K. A. R. Mitchell, Surf. Sci. 343 (1995) L1167.
80 Chapter 4
[12] C. S. Zhang, B. Li, and P. R. Norton, Surf. Sci. 313 (1994) 308.
[13] A. Stierle, V. Formoso, F. Comin, G. Schmitz, and R. Franchy, Physica B 283 (2000)
208.
[14] R. A. Ploc, J. Nucl. Mater. 110 (1982) 59.
[15] R. A. Ploc, J. Nucl. Mater. 115 (1983) 110.
[16] E. Panda, L. P. H. Jeurgens, and E. J. Mittemeijer, J. Appl. Phys. 106 (2009) 114913.
[17] H. Brune, J. Wintterlin, J. Trost, G. Ertl, J. Wiechers, and R. J. Behm, J. Chem. Phys.
99 (1993) 2128.
[18] G. F. Cotterill, H. Niehus, and D. J. Oconnor, Surf. Rev. Lett. 3 (1996) 1355.
[19] M. Muller and H. Oechsner, Surf. Sci. 387 (1997) 269.
[20] V. Maurice, S. Cadot, and P. Marcus, Surf. Sci. 458 (2000) 195.
[21] F. Wiame, V. Maurice, and P. Marcus, Surf. Sci. 601 (2007) 1193.
[22] F. Jensen, F. Besenbacher, E. Laegsgaard, and I. Stensgaard, Surf. Sci. 259 (1991)
L774.
[23] T. Matsumoto, R. A. Bennett, P. Stone, T. Yamada, K. Domen, and M. Bowker, Surf.
Sci. 471 (2001) 225.
[24] A. U. Goonewardene, J. Karunamuni, R. L. Kurtz, and R. L. Stockbauer, Surf. Sci.
501 (2002) 102.
[25] K. Radican, S. I. Bozhko, S. R. Vadapoo, S. Ulucan, H. C. Wu, A. McCoy, and I. V.
Shvets, Surf. Sci. 604 1548.
[26] W. X. Li, L. Österlund, E. K. Vestergaard, R. T. Vang, J. Matthiesen, T. M. Pedersen,
E. Lægsgaard, B. Hammer, and F. Besenbacher, Phys. Rev. Lett. 93 (2004) 146104.
[27] F. Qin, N. P. Magtoto, M. Garza, and J. A. Kelber, Thin Solid Films 444 (2003) 179.
[28] A. Wight, N. G. Condon, F. M. Leibsle, G. Worthy, and A. Hodgson, Surf. Sci. 333
(1995) 133.
[29] J. Klikovits, M. Schmid, L. R. Merte, P. Varga, R. Westerström, A. Resta, J. N.
Andersen, J. Gustafson, A. Mikkelsen, E. Lundgren, F. Mittendorfer, and G. Kresse,
Phys. Rev. Lett. 101 (2008) 266104.
[30] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, Surf. Interface Anal. 42 (2010)
588.
[31] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, J. Appl. Phys. 110 (2011)
024904.
[32] A. Lyapin, L. P. H. Jeurgens, and E. J. Mittemeijer, Acta Mater. 53 (2005) 2925.
An STM study of the initial oxidation of single-crystalline zirconium surfaces 81
[33] M. C. Gallagher, M. S. Fyfield, J. P. Cowin, and S. A. Joyce, Surf. Sci. 339 (1995)
L909.
[34] D. A. Bonnell, Prog. Surf. Sci. 57 (1998) 187.
[35] J. M. West, Basic Corrosion and Oxidation, Elsevier, Amsterdam (1986).
[36] N. Stojilovic, E. T. Bender, and R. D. Ramsier, Prog. Surf. Sci. 78 (2005) 101.
[37] C. Stampfl, M. V. Ganduglia-Pirovano, K. Reuter, and M. Scheffler, Surf. Sci. 500
(2002) 368.
[38] M. Gutowski, J. E. Jaffe, C. L. Liu, M. Stoker, R. I. Hegde, R. S. Rai, and P. J. Tobin,
Appl. Phys. Lett. 80 (2002) 1897.
[39] G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89 (2001) 5243.
[40] D. Majumdar and D. Chatterjee, Thin Solid Films 206 (1991) 349.
[41] D. Majumdar and D. Chatterjee, Thin Solid Films 236 (1993) 164.
[42] J. M. Sanz, A. R. Gonzalezelipe, A. Fernandez, D. Leinen, L. Galan, A. Stampfl, and
A. M. Bradshaw, Surf. Sci. 309 (1994) 848.
[43] P. C. Snijders, L. P. H. Jeurgens, and W. G. Sloof, Surf. Sci. 496 (2002) 97.
[44] A. Zalar, Thin Solid Films 124 (1985) 223.
[45] I. Horcas, R. Fernandez, J. M. Gomez-Rodriguez, J. Colchero, J. Gomez-Herrero, and
A. M. Baro, Rev. Sci. Instrum. 78 (2007)
[46] F. Wiame, V. Maurice, and P. Marcus, Surf. Sci. 601 (2007) 4402.
[47] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, J. Phys. Chem. C 115 (2011)
19841.
[48] C. S. Zhang, B. J. Flinn, and P. R. Norton, Surf. Sci. 264 (1992) 1.
[49] E. Fromm and O. Mayer, Surf. Sci. 74 (1978) 259.
[50] F. Zandiehnadem, R. A. Murray, and W. Y. Ching, Physica B & C 150 (1988) 19.
[51] M. Morinaga, H. Adachi, and M. Tsukada, J. Phys. Chem. Solids 44 (1983) 301.
[52] R. P. Gupta, Phys. Rev. B 32 (1985) 8278.
[53] M. Gautier, J. P. Duraud, L. P. Van, and M. J. Guittet, Surf. Sci. 250 (1991) 71.
[54] M. H. Brodsky and M. Cardona, J. Non-Cryst. Solids 31 (1978) 81.
[55] G. Bakradze, L. P. H. Jeurgens, U. Starke, T. Acartürk, and E. J. Mittemeijer, Acta
Mater. 59 (2011) 7498.
Chapter 5
Atomic transport mechanisms in thin oxide
films grown on zirconium by thermal oxidation,
as-derived from 18O-tracer experiments
Georgijs Bakradze, Lars P.H. Jeurgens, Ulrich Starke,
Tolga Acartürk and Eric J. Mittemeijer
Abstract
Two-stage oxidation experiments using 16O and 18O isotopes were performed to reveal the governing
atomic transport mechanism(s) in thin (thickness < 10 nm) oxide films grown during the initial stages
of dry thermal oxidation of pure Zr at 450 K. To this end, bare (i.e. without a native oxide) Zr(0001)
and Zr(101 0) single-crystalline surfaces were prepared under UHV conditions by a cyclic treatment
of alternating SC and in-vacuo annealing steps. Next, the bare Zr surfaces were oxidized at 450 K and
at pO2 = 1×10-4 Pa, first in 16O2(g) and subsequently in 18O2(g). The 18O-tracer depth distributions in
the oxide films were recorded by ToF-SIMS. It was concluded that the early stage of the oxidation
process is governed by oxygen transport to the metal/oxide interface through the lattice and along the
GBs of the nano-sized oxide grains, whereas upon continuing oxidation only oxygen lattice transport
controls the oxidation process. An oxide-film growth mechanism is proposed.
5.1 Introduction
In many application areas, such as powder metallurgy, microelectronics, gas sensors, surface
coatings and catalysis, comprehensive knowledge of the atomic transport phenomena in thin
oxide layers is a prerequisite in order to control and optimize the functional properties of
metallic- and/or semiconductor-based components under varying operation conditions [1-6].
Fundamental investigations on the transport mechanisms in oxide films, as developing on
metal and alloy surfaces by e.g. thermal oxidation, should particularly address: (i) the type of
the (rate-determining) migrating species (anions, cations and the corresponding vacancies, as
well as electrons and electron holes) and the associated reaction fronts (e.g. the oxide/gas
and/or metal/oxide interface), as well as (ii ) the transport paths (e.g. lattice versus GB
transport) and transport mechanisms (e.g. vacancy, interstitial, substitutional) of the migrating
84 Chapter 5
species. The quantitative, direct experimental assessment of (iii ) the rate of the migrating
species (e.g. diffusivities) [7] and (iv) the (steady-state) defect concentrations at the
metal/oxide and oxide/gas interfaces presents a considerable experimental challenge,
particularly for very thin (thickness < 10 nm) oxide films, as dealt with in the present study,
and is beyond the scope of this work.
Fig. 5.1. Schematic O-tracer depth distribution profiles in an oxide layer, as grown under control of
different (combinations of) transport processes (after Refs. [8-10]): (a) outward metal transport (either
through the lattice or along short-circuits); (b) inward oxygen transport by a vacancy (or interstitial)
mechanism; (c) inward oxygen short-circuit transport (without isotope exchange); (d) inward oxygen
short-circuit transport with isotope exchange. See Section 5.1 for details.
In the past, several types of marker and tracer experiments were designed to identify
the predominant transport species during surface oxidation, in particular, by determining the
reaction front(s) of the oxidation reaction [8-9, 11-12]. For example, in a so-called ‘two-stage
oxidation experiment’, when the substrate is first oxidized in a ‘natural’ 16O2(g) atmosphere
and subsequently in a ‘labelled’ (but chemically identical) 18O2(g) atmosphere; the following
determination of the specific distribution of the 18O tracer in the grown oxide layer can
disclose the more mobile species in the oxidation process. Determination of such isotope
depth distributions in thin films requires a high depth resolution, which can be offered with a
mass-sensitive surface-analytical technique as ToF-SIMS or nuclear reaction analysis. From
the established depth-profiles of the fraction of the 18O-tracer component, ( )c τ (see Section
5.2.4 for details), the predominant transport mechanisms during the oxidation process can be
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 85
deduced by considering the various possible cases for the governing transport mechanism, as
follows (see Fig. 5.1) [8, 10].
If outward metal transport is the governing mechanism during the second oxidation
step performed in a pure 18O2(g) atmosphere, an 18O-rich oxide will be formed preferentially
at the gas/oxide interface and the ( )c τ profile will resemble Fig. 5.1a. If inward oxygen
transport predominates during the second oxidation step, the shape of the ( )c τ profile will
depend on the particular oxygen transport path: (i) if O is transported through the oxide
lattice by a vacancy mechanism or by an interstitial mechanism (which is unlikely), the ( )c τ
profile will look like Fig. 5.1b, whereas (ii ) for inward O transport through short-circuits (e.g.
through GBs or cracks), 18O-rich oxide will mainly form near the oxide/metal interface (see
Fig. 5.1c). In general, isotopic exchange at oxide GBs cannot be neglected and, consequently,
the ( )c τ profile for case (ii ) (see Fig. 5.1c) will rather resemble that of Fig. 5.1d. A
combination of the above-described predominant transport mechanisms typically results in
more complicated shapes of the measured ( )c τ profiles [8, 10, 13-14], as encountered in the
present study (see Section 5.4).
To date, two-stage oxidation experiments, as sketched above, have been mainly
conducted for the case of growth of micrometer thick (crystalline) oxide scales, e.g. as grown
by thermal oxidation at elevated temperatures (T > 700 K) on pure Ni [10, 15], pure Cr and
Al [16], NiAl intermetallics and Ni-Cr-Al-Y alloys [17], Fe-Cr alloy and zircaloy-4 [10],
zirconium [18] and zirconium-tin alloys [19]. The present work aims to apply two-stage
oxidation experiments to the case of growth of very thin (thickness < 10 nm) oxide
overgrowths, as developing at low temperatures up to about 500 K [20-21]. Further, in this
way the present 18O-tracer oxidation study is meant to clarify an existing controversy in the
literature regarding the governing mechanism for the low-temperature oxidation of Zr: as
proposed in Ref. [22], the low-temperature oxidation rate of Zr would be limited by either
inward O transport or the dissociative chemisorption of O2(g) at the oxide surface, whereas
according to Ref. [23] outward cation transport would be the rate-limiting step in the
oxidation process. Understanding the oxidation mechanism(s) of Zr in the low-temperature
regime is not only of great importance for applications of very thin Zr-oxide layers in state-
of-the-art nanotechnologies and microelectronics, but also to complement existing knowledge
on the corrosion resistance and embrittlement of Zr (and its alloys), in conventional nuclear
cladding applications [24-26].
86 Chapter 5
In the present study, single-crystalline Zr(0001) and Zr(101 0) surfaces were oxidized
by successive exposures (for different times) to 16O2(g) and 18O2(g) at 450 K and at a partial
oxygen pressure of pO2 = 1×10-4 Pa. ToF-SIMS was applied to establish the 18O-tracer-depth
profiles in the grown oxide films (Section 5.3). On the basis of the identified governing
transport mechanisms at different stages of the oxidation process (Section 5.4), a
corresponding, overall oxide-film growth mechanism has been proposed for the oxidation at
450 K (Section 5.5). Details on the growth kinetics and the microstructural evolution of the
developing oxide films have been addressed in the previous Chapters 2 and 4, respectively.
5.2 Experimental
5.2.1 Specimen preparation
Disc-shaped Zr(0001) and Zr(101 0) single crystals were cut (diameter 6 mm; 1 mm thick;
orientation alignment within ±0.5º of the nominal surface plane) from a single-crystalline
unalloyed α-Zr rod and subsequently mechanically (single-side) polished (last step to 0.05
µm diamond paste). Main impurities in the polished specimens, as identified by inductively
coupled plasma optical emission spectroscopy analysis, are (in mass parts): Hf (60 ppm); Fe
(25 ppm); Ti (1 ppm); Cu, Zn, Mn, Ca, Na (< 2 ppm).
The polished specimens were introduced into a combined UHV system for specimen
processing and in-situ analysis (base pressure < 3×10-8 Pa). The (native) oxide and other
adventitious contaminants (mainly C) on the surface were removed by SC, rastering the entire
surface area, until no other element than Zr was detected in a XPS survey spectrum.
Roughening of the ion-bombarded single-crystalline surfaces due to local differences in the
sputter yield by ion channelling and shadowing effects [27] was prevented by employing
continuous sample rotation at a speed of about 2 rpm during SC. The SC was performed with
3 kV Ar+ ions until all C surface contamination was removed (as verified by XPS); all
subsequent SC treatments (e.g. to remove remaining O contamination) were performed with 1
kV Ar+ ions.
Next, the specimen and the specimen holder were outgassed by a cycling treatment of
alternating 1 kV Ar+ SC and in-vacuo annealing steps, while gradually increasing the sample
temperature during each successive in-vacuo annealing step up to 1000 K. As a final surface
preparation step, prior to each oxidation experiment, the SCed surfaces were in-vacuo
annealed at 1000 K for 300-600 s to fully restore the surface crystallinity in the ion-
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 87
bombarded region (as verified by in-situ LEED, see Chapter 2 or Ref. [28]) and, at the same
time, reduce the sputter-induced surface roughness (by thermally activated surface diffusion).
5.2.2 Thermal oxidation
Exposure of bare (i.e. without a native oxide) Zr metal surfaces to O2(g) at T ≤ 400 K, results
in the very fast formation (i.e. within less than a minute) of an ultrathin (approximately 1 to 3
nm thick), amorphous oxide film of self-limiting thickness [29-30]. At such low
temperatures, it is practically impossible to conduct two-stage oxidation experiments at
controllable, variable oxide-film thicknesses, as well as to record meaningful 18O-tracer-depth
profiles by ToF-SIMS with its depth resolution of about 1 nm.
Fig. 5.2. Total oxide-film thickness as a function of oxidation time for the oxidation of the bare (a)
Zr(0001) and (b) Zr(101 0) substrates at 450 K and at pO2 = 1×10-4 Pa (as determined by RISE); see
Chapter 2 or Ref. [29] for experimental details). The interruption moments, t16→18, for the ∆L16:∆L18 ≈
1:1 and ∆L16:∆L18 ≈ 2:1 tracer oxidation experiments (see Section 5.2.2) have been indicated by the
dashed lines.
For temperatures T ≥ 400 K, the oxidation kinetics of Zr metal surfaces are no longer self-
limiting (see Fig. 5.2) and, consequently, the oxide-film thicknesses after the initial (first
88 Chapter 5
stage) 16O2(g) oxidation and the subsequent (second stage) 18O2(g) oxidation can be
controlled over a sufficiently large range (i.e. from about 3 to 9 nm; see Fig. 5.2) to allow
reliable determination of the 18O-tracer-depth profiles by ToF-SIMS (provided that a closed,
uniform oxide layer has formed after the first oxidation step in 16O2(g), as is the case here: see
Section 5.3).
Thus bare Zr(0001) or bare Zr(101 0) surfaces were in-situ heated to a constant
temperature of 450 K (as measured with a type K thermocouple put in direct mechanical
contact with the single-crystal surface) and subsequently exposed for a defined period of time
(see Fig. 5.2) to ultrapure O2 gas (hereafter referred to as "16O2 gas", with purity ≥ 99.9999
vol.% with a specified residual gas content of H2O ≤ 0.5 vpm, N2+Ar ≤ 2.0 vpm, CnHm ≤ 0.1
vpm and CO2 ≤ 0.1 vpm; supplied by Westfalen AG) at pO2 = 1×10-4 Pa. The partial oxygen
pressure during oxidation was kept constant by a feedback of the (oxygen) pressure in the
UHV chamber, as measured with a Bayerd-Alpert nude pressure gauge (in Volts), to the
setpoint of the oxygen leakage valve control unit (Balzers, RVG 050B). Next, the 16O2-
preoxidized surfaces were exposed for a defined time period to a high purity isotopically-
enriched 18O2 gas (hereafter referred to as "18O2 gas", with an 16O2-isotope content of only 1
at.%; supplied by ICON Isotope Services Inc.) The switching from the 16O2 gas atmosphere
to the 18O2 gas atmosphere was performed in-situ such that after the 16O2 oxidation step, as
ended by closing the 16O2 valve and simultaneously opening the 18O2 valve, the O2 partial
pressure in the UHV chamber was restored by 18O2 gas, i.e. reached 1×10-4 Pa, in less than
120 s, while keeping the specimen at the temperature of 450 K. In the following, the 16O2-18O2 interruption moment, t16→18, will be taken as the oxidation time at which the oxidizing
medium was switched from 16O2(g) to 18O2(g).
For all two-stage oxidation experiments, the total oxidation time (i.e. the sum of the 16O2 and 18O2 exposure times) equalled 7200 s, which resulted in total thicknesses of Lt=7200s ≈
6.5 nm and Lt=7200s ≈ 9.2 nm for the oxide films grown on the Zr(0001) and Zr(101 0)
surfaces, respectively (as established by RISE) under identical conditions in the same
experimental set-up [29]). The values of t16→18 were selected on the basis of the oxide-film
growth kinetics, as determined by RISE [29], as follows. For two-stage oxidation
experiments it is generally advised to make the second oxidation stage considerably shorter
than the first one [14], in order to assure significant 18O-concentration gradients. This is
easier to realize at later oxidation stages, for which the oxidation rate levels off (see below
and Fig. 5.2). Moreover, a shorter second oxidation stage will reduce unwanted isotopic
intermixing. The depth resolution of the ToF-SIMS analysis is limited to about 1 nm (see
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 89
Section 5.2.4) and this requires that the minimal oxide-layer thickness, as grown in each of
both oxidation stages, exceeds 1 nm. On the basis of the above considerations, while
accounting for the different oxide-film growth kinetics on the Zr(0001) and Zr(101 0)
surfaces (see Figs. 5.2a and b), in a first set of two-stage oxidation experiments, the 16O2-18O2
interruption moments (i.e. t16→18 = 2400 s and t16→18 = 1500 s for the Zr(0001) and Zr(101 0)
surface, respectively; see Fig. 5.2) were chosen such that roughly the same nominal
thicknesses of 'normal' 16O-oxide (designated as ‘normal’ in Fig. 5.2), ∆L16, and 'labelled' 18O-oxide (designated as ‘labelled’ in Fig. 5.2), ∆L18, formed during the 16O2- and 18O2-
oxidation stages: i.e. ∆L16:∆L18 ≈ 1:1. The nominal thickness of the oxide, as formed during
the second oxidation step, is given by the oxide-film thickness increase during the second
oxidation stage, i.e. during the 18O2 exposure step, as determined from RISE data [28-29]: i.e.
∆L18 = Lt=7200s – ∆L16. In a second set of two-stage oxidation experiments, the 16O2-18O2
interruption moments were chosen such (t16→18 = 4200 s and t16→18 = 2400 s for the Zr(0001)
and Zr(101 0) surface, respectively; see Fig. 5.2), that roughly twice as much oxide is formed
during the first 16O-oxidation stage than during the second 18O-oxidation stage; i.e. ∆L16:∆L18
≈ 2:1.
5.2.3 In-situ deposition of an Al capping layer
After each two-stage oxidation experiment and cooling to room temperature, the oxide film
was sealed in-situ by depositing a laterally-closed, about 7 nm thick Al capping layer using a
thermal effusion source in a UHV molecular beam epitaxy (MBE) side-chamber, while
cooling the specimen holder with liquid nitrogen to prevent a local heating up of the oxidized
specimen surface due to heat irradiation from the effusion source, which could have led to
microstructural changes and isotopic intermixing (as driven by entropy effects) within the
oxide film and chemical interaction of the Al capping layer with the oxide-film surface (note:
for the two-stage oxidation of the Zr(101 0) surface with ∆L16:∆L18 ≈ 1:1, the deposited Al
capping layer was thinner than in other experiments, only about ≈ 5 nm, see Fig. 5.4a).
Next, the oxidized Zr single-crystal substrate with Al capping layer was removed
from the UHV system and immediately transferred (under atmospheric conditions) to the
vacuum chamber for ToF-SIMS analysis (see Section 5.2.4). Thereby, inevitably a 2 to 3 nm
thick (amorphous) Al2O3 film formed on the surface of the Al capping layer, which served as
a further protective layer to prevent intermixing of natural oxygen from the ambient
atmosphere with the two-stage oxidized Zr surface. Moreover, the Al2O3/Al double seal layer
90 Chapter 5
serves as a sacrificial layer for achieving a stationary state for the sputter conditions at the
specimen surface at the onset of the ToF-SIMS analysis (see Section 5.2.4), which is crucial
for a successful ToF-SIMS analysis of such very thin oxide films.
5.2.4 XPS, ToF-SIMS and HR-TEM analysis
XPS was applied to determine the chemical constitution of the SCed, the annealed and the
oxidized surfaces (see Sections 5.2.1 and 5.2.2). The XPS measurements were performed in-
situ with a Thermo VG Thetaprobe system employing monochromatic Al Kα radiation (hν =
1486.68 eV). The measured XPS survey spectra, which cover a binding energy range from 0
to 1200 eV, were recorded with a step size and constant pass energy of 0.2 eV and of 200 eV,
respectively.
The distribution of the 18O-tracer in the grown oxide films (with total thicknesses < 10
nm; see Fig. 5.2) was measured ex-situ by ToF-SIMS using a ToF-SIMS IV instrument (ION-
TOF GmbH, Münster, Germany). Secondary ions for analysis were generated with a focused
primary Ga+ beam of 25 keV and 0.07 pA operated in burst mode and scanned over an area
of 30.3 µm × 30.3 µm. The secondary ions were detected employing a negative detection
mode. Cs+ ions (as generated by electrostatic field ionization of liquefied Cs) were used for
surface ablation (sputtered area of 500 µm × 500 µm; acceleration voltage of 500 eV; beam
current 35 nA) to acquire a depth profile of the secondary ions. It is noted that for the
oxidized Zr(0001) surface with ∆L16:∆L18 ≈ 1:1, an Cs+ acceleration voltage of 250 eV
(instead of 500 eV) was employed, which affects only the sputter rate for the depth profiles.
For further instrumental details, see Ref. [31].
The strong dependencies of the recorded SIMS signal intensities on the selected
measurement parameters (e.g. the type of primary ions and their energy) and on matrix
effects (e.g. due to abrupt changes in the chemical composition at interfaces) were
circumvented in the data analysis by considering the relative change in the measured amount
of 18O isotope: i.e. the value of [ ]181618)τ( IIIc += was determined, as function of the
sputter time, where )τ(18I and )τ(16I denote the intensities of the recorded 18O and 16O
isotope signals, respectively. The sputter-time scale was converted into a sputter-depth scale
by assuming equal relative sputter-rates of Al2O3, Al metal, Zr-oxide and Zr metal and
employing the known oxide-film thickness values after t = 7200 s, as obtained from RISE
(see Fig. 5.2).
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 91
Storage of the oxidized specimens under laboratory conditions might result in gradual
changes in the 18O-tracer depth distribution by isotopic intermixing (as driven by entropy
effects). To minimize these ageing effects, the ToF-SIMS measurements were performed
within several hours after oxidation and subsequent capping. Only for one two-stage
oxidation experiment (the two-stage oxidation of the Zr(101 0) surface with ∆L16:∆L18 ≈ 2:1),
the specimen with Al-capping layer unintentionally had to reside for two weeks in UHV
before the ToF-SIMS analysis could be performed.
After ToF-SIMS analysis of the oxide film, as grown on the Zr(101 0) by the
∆L16:∆L18 ≈ 1:1 two-stage oxidation experiment at 450 K (see Section 5.2.2), a cross-
sectional lamella for high-resolution transmission electron microscopical (HR-TEM) analysis
was cut from the sealed oxidized specimen using a dual Focused Ion Beam (FIB; Nova
Nanolab 600, FEI Co.) Subsequent HR-TEM analysis was performed using a JEOL JEM-
ARM1250 electron microscope with a very high acceleration voltage of 1250 kV and a point-
to-point resolution with the side entry lens of 0.12 nm, while cooling the lamellae with liquid
nitrogen during the analysis to retard any microstructural changes in the irradiated area of the
TEM lamella, as inflicted by the high-energy electron beam. The negatives of the recorded
micrographs were digitized for further quantitative evaluation. Internal calibration of the
length scale of the recorded micrographs was performed using the known lattice constant of
the Zr metal (0.32312 nm [32]). For further details on the FIB preparation procedure and the
HR-TEM analysis, see Ref. [33].
5.3 The oxide-film microstructure
As shown by RISE and angle-resolved XPS [28-29], the initial thermal oxidation of bare Zr
surfaces exposed to pure O2(g) at pO2 = 1×10-4 Pa in temperature range of 300-450 K results
in the formation of an oxide film, which exhibits a gradient of O-deficiency (with respect to
ZrO2) that increases from the interior of the oxide film towards the metal/oxide interface (see
discussion in Section 5.5 and Refs. [30, 34-35]). A closed oxide film is already formed within
the first 300 s of oxidation and, subsequently, the oxide-film approaches a near-limiting
thickness at T < 375 K [29]. The fast occurrence of a limiting oxide-film thickness at T < 375
K obviously hinders the performance of meaningful two-stage tracer oxidation experiments.
At T ≥ 450 K, the retardation of the oxide-film growth rate after the initial fast oxidation
regime is much less pronounced and instead the oxide film grows continuously (i.e. a near-
92 Chapter 5
limiting oxide-film thickness is no longer established) by the formation of a stoichiometric
ZrO2 outer layer [29-30, 34]: see Fig. 5.2.
Fig. 5.3. Cross-sectional HR-TEM micrograph of the Zr-oxide overgrowth on the Zr(101 0) surface
after the ∆L16:∆L18 ≈ 1:1 two-stage oxidation experiment, i.e. oxidation at 450 K and at pO2 = 1×10-4
Pa for 7200 s (and subsequent in-situ deposition of a MBE-grown Al seal). The direction of the
incident electron beam was along the [0001] zone axis of the Zr(101 0) substrate. Insets (i) and (ii ):
fast Fourier transformations (FFT, i.e. "diffraction patterns") of the areas (i) a single oxide grain and
(ii ) the substrate as indicated in the micrograph.
The temperature of T = 450 K is thus considered a suitable temperature for the conduction of
two-stage oxidation experiments on the bare Zr surfaces in order to study mechanisms of
initial oxidation. Oxidation of the Zr(0001) and Zr(101 0) surfaces for 7200 s at 450 K and at
pO2 = 1×10-4 Pa results in total oxide-film thicknesses of Lt=7200s ≈ 6.5 nm and Lt=7200s ≈ 9.2
nm, respectively (see Fig. 5.2).
Oxide-film VB studies by in-situ XPS (in combination with in-situ investigations by
low energy electron diffraction and HR-TEM) performed in this project (see Ref. [36] and
Fig. 5.3) indicate that the ZrO2 films on Zr(0001) and Zr(101 0) are predominantly
amorphous for oxidation temperatures T < 400 K, whereas a crystalline tetragonal-ZrO2-like
phase develops at T ≥ 400 K (note: the present ToF-SIMS study pertains to oxidation at T =
450 K; see Section 5.2.2). Additionally, as evidenced by HR-TEM (see Fig. 5.3) and in-situ
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 93
STM (see Chapter 4 or [28]), the crystalline ZrO2 films grown at 450 K develop a
polycrystalline (granular) structure with an average grain size in the range of 10-15 nm. A
cross-sectional HR-TEM micrograph of the oxide-film as grown on the Zr(101 0) surface
after the ∆L16:∆L18 ≈ 1:1 two-stage tracer oxidation experiment at 450 K (see Section 5.2.2
and Fig. 5.2) is shown in Fig. 5.3.
Some nanograins within this polycrystalline oxide film could indeed be identified as
tetragonal zirconia by fast Fourier transformation (FFT) of selected regions of the digitized
HR-TEM micrographs, which each enclosed a single oxide grain (a corresponding, indexed
"diffraction pattern", as obtained by the FFT, is given in the inset in Fig. 5.3). The granular
structure coarsens with increasing oxidation time, accompanied with an increase of the
average surface roughness is less than 1 nm after 7200 s of oxidation, which is still much less
than the corresponding oxide-film thicknesses (see above and Fig. 5.2). The in-situ STM
studies show that the oxide layers formed on the Zr surfaces at 450 K are fully closed within
the first 300-600 s of oxidation. Furthermore the cross-sectional TEM analysis (see Fig. 5.3)
demonstrates that the film thickness of the oxide overgrowths on the Zr surface at 450 K is
more or less uniform. Thereby two necessary conditions for meaningful two-stage, tracer
oxidation experiments in the very thin oxide-film regime are fulfilled [8, 16].
5.4 18O-tracer depth distributions: identification of
governing transport mechanisms
As discussed in Section 5.2.4, the depth distribution of the 18O-isotope in the grown oxide
films, as measured by ToF-SIMS, is best visualized by plotting the measured 18O-isotope
fraction, [ ]181618 IIIc += , as function of the sputter depth or time, τ. The measured profiles
of ( )c τ for the first set of tracer oxidation experiments at 450 K and at pO2 = 1×10-4 Pa with
∆L16:∆L18 ≈ 1:1 are shown in Figs. 5.4a and c for the Zr(0001) substrate and the Zr(101 0)
substrate, respectively (cf. Fig. 5.2). The measured profiles of ( )c τ for the second set of
tracer oxidation experiments at 450 K and at pO2 = 1×10-4 Pa with ∆L16:∆L18 ≈ 2:1 are
presented in Figs. 5.4b and d. The first nanometers of the sputter-depth scale in the profiles of
Fig. 5.4 concern the removal of the thin Al capping layer including its native oxide. The ( )c τ
profiles in the underlying oxide films with ∆L16:∆L18 ≈ 1:1 show two clear intensity maxima,
94 Chapter 5
one near the gas/oxide interface and one close to the oxide/metal interface (see Figs. 5.4a and
c).
Fig. 5.4. 18O-tracer depth distributions, [ ]181618 IIIc += , (as recorded by ToF-SIMS; see Section
5.2.4) in Zr-oxide layers, as grown by two-stage thermal oxidation (i.e. first in 16O2(g) then in 18O2(g);
see Section 5.2.2) of bare Zr surfaces at 450 K (and at pO2 = 1×10-4 Pa) for a total oxidation time of
7200 s. The Al seal was deposited in-situ after the oxidation (see Section 5.2.3). Two-stage oxidation
of the Zr(0001) surface: (a) for 2400 s under 16O2 followed by 4800 s under 18O2, with "normal" (i.e. 16O) and "labelled" (i.e. 18O) nominal thicknesses corresponding to ∆L16:∆L18 ≈ 1:1; (b) for 4200 s
under 16O2(g) followed by 3000 s under 18O2(g), with "normal" and "labelled" nominal thicknesses
corresponding to ∆L16:∆L18 ≈ 2:1. Two-stage oxidation of the Zr(101 0) surface: (c) for 1500 s under 16O2 followed by 5700 s under 18O2, with "normal" and "labelled" nominal thicknesses corresponding
to ∆L16:∆L18 ≈ 1:1; (d) for 2400 s under 16O2(g) followed by 4800 s under 18O2(g), with "normal" and
"labelled" nominal thicknesses corresponding to ∆L16:∆L18 ≈ 2:1.
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 95
For the oxidized Zr(0001) plane, the two maxima are of similar (normalized) height
(see Fig. 5.4a), whereas, for the oxidized Zr(101 0) plane, the intensity maximum near the
oxide/metal interface is higher than that close to the gas/oxide interface (see Fig. 5.4c). The
occurrence of these two intensity maxima in the measured ( )c τ profiles suggests that the
oxidation rate is controlled by a combination of lattice and short-circuit transports of O and/or
Zr (see Fig. 5.1 and its discussion in Section 5.1).
The shape of the measured ( )c τ profiles for ∆L16:∆L18 ≈ 2:1 (see Figs. 5.4b and d) is
clearly different from those observed for ∆L16:∆L18 ≈ 1:1 (see Figs. 5.4a and c): only one
intensity maximum, close to the gas/oxide-interface, is observed, suggesting outward Zr
transport (either through the lattice or short-circuits; see Fig. 5.1a) and/or inward O lattice
transport (see Fig. 5.1b) as the governing transport mechanisms. Hence, as revealed by the
introduction of 18O-tracer at a later stage of the oxidation process, a change in the oxide-film
growth mechanism occurs upon continued oxidation.
At elevated temperatures, O lattice transport in crystalline ZrO2 is considerably faster
than Zr lattice transport in ZrO2 (about one order of magnitude faster at 973 K in undoped1
m-ZrO2 [38-40]), which is attributed to the higher migration enthalpy of cation vacancies
[41]. Further, as discussed in Section 5.3, the evolving Zr-oxide film is overall O-deficient
(particularly near the oxide/metal interface) and defective (i.e. contains oxygen vacancies)
[29-30, 35, 42-43]. Therefore (see also footnote 1), it is unlikely that outward transport of Zr
cations, via an interstitial mechanism, occurs at a (much) faster rate than inward O transport
via the vacancy mechanism. Hence, for the interpretation of the observed differences in the
measured ( )c τ profiles, only lattice transport of O ions through the bulk oxide phase (see
Fig. 5.1b) and short-circuit transport of O (see Figs. 5.1c and d) along GBs in the developing
oxide film (see Fig. 5.3) are considered as possible governing transport mechanisms. Note,
that a polycrystalline oxide film develops on the Zr surfaces at 450 K (cf. Fig. 5.3), with an
average oxide-grain size that increases with increasing oxidation time up to about 15 nm (i.e.
somewhat larger than the corresponding oxide-film thickness).
As follows from comparison of the ( )c τ profiles obtained for the case ∆L16:∆L18 ≈
1:1 (Figs. 5.4a and c) with Figs. 5.1b and Fig. 5.1c (or Fig. 5.1d), the measured 18O-tracer
depth profiles can then be understood as governed by lattice transport of O through the 1 Most studies [38-39] on diffusion in ZrO2 deal with the high-temperature cubic phase of ZrO2 (c-ZrO2), which is typically doped with yttria, calcia or magnesia, to stabilize the c-ZrO2 phase at low temperatures. The presence of such an aliovalent stabilizer on Zr sites is accompanied by the formation of structural vacancies on the O sublattice, which for doped zirconia leads to a further enhancement of the oxygen lattice diffusivity as compared to the cation diffusivity [37].
96 Chapter 5
interior of the nanosized oxide grains in combination with much faster short-circuit transport
of O along the GBs between the nanosized oxide grains. Hence the intensity maximum near
the oxide/metal interface for ∆L16:∆L18 ≈ 1:1 (see Figs. 5.4a and c) originates from fast, short-
circuit O diffusion.
If the 18O-isotope is introduced at a later stage of the oxidation process (i.e. for
∆L16:∆L18 ≈ 2:1), only the intensity maximum near the gas/oxide interface is observed (see
Figs. 5.4b and d). This then indicates (cf. Fig. 5.1b) that only inward lattice transport of O
prevails at a later oxidation stage (i.e. the contribution of short-circuit diffusion of O has
become marginal). The in-situ STM investigations [28] have revealed a coarsening of the
nanosized grains of the oxide films on the Zr(0001) and Zr(101 0) surfaces with increasing
oxidation time (at 450 K). Consequently, the GB density and thereby the density of O short-
circuit transport paths decreases with increasing oxidation time at 450 K. It may further be
suggested that the boundaries between the nanosized oxide grains, as formed at the onset of
oxidation (i.e. during the very fast oxidation regime; see Fig. 5.2), possess a non-equilibrium
structure facilitating short-circuit transport. The coarsening of the oxide grains occurring on
continued oxidation (cf. Section 5.3) can be accompanied by an equilibration and
densification of the GB structure, which also reduces the possible contribution of short-circuit
transport. Moreover, the transport along GBs in ionic materials, such as zirconia, can be
suppressed by the accumulation of space charge in the vicinity of GBs [44-46]. It is therefore
suggested that, not only the reduction of the GB density, but also an equilibration of the GB
structure and/or the accumulation of space charge at the GBs can contribute to the retardation
of the O short-circuit transport rate with increasing oxidation time at 450 K.
The occurrence of a maximum near the oxide/metal interface, which is higher than
that close to the gas/oxide interface, as observed for the ( )c τ profiles of the oxidized Zr(101
0) prism plane for ∆L16:∆L18 ≈ 1:1 (see Fig. 5.4c), suggests that the initial inward O short-
circuit transport is more pronounced for oxide films grown on the Zr(101 0) plane (as
compared to oxide films grown on the Zr(0001) plane under similar conditions for ∆L16:∆L18
≈ 1:1, cf. Fig. 5.4a), which is well-compatible with the observed higher initial growth rate
(see Fig. 5.2 and Ref. [29]) and an, on average, smaller grain size [36] for the oxide
overgrowth on the Zr(101 0) surface.
It is thus concluded that the initial, fast oxidation stage of the Zr(0001) and Zr(101 0)
surfaces at 450 K (and pO2 = 1×10-4 Pa) proceeds by inward O lattice transport (through the
interior of the nano-sized oxide grains) in combination with fast inward O transport along
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 97
short-circuits (along GBs of the nano-sized Zr-oxide grains). However, the relative
contribution of O short-circuit diffusion decreases with increasing oxidation time (at 450 K)
due to an overall decrease of the oxide GB density (by coarsening of the nanosized grain
structure) and a concurrent equilibration of the GB structure, in possible combination with the
accumulation of space charge at the GBs. Consequently, the relatively slow O lattice
transport through the bulk of the oxide grains governs the oxidation at a later stage of the
oxidation process.
Here it is emphasized that the above discussed change in oxidation mechanism upon
continued oxide-film growth would not have been revealed solely on the basis of one
measured ( )c τ profile. At least two two-stage, tracer oxidation experiments with different
exposure times to 16O2(g) and 18O2(g) are required to conclude upon the governing transport
mechanism(s) during the oxidation process.
5.5 Proposed oxidation mechanism
Oxygen lattice transport requires coupled fluxes of inwardly migrating O anions and
outwardly migrating O vacancies. Recognizing that the rate of dissolution of O in α-Zr is
(still) significant at temperatures as low as 450 K [47-48], the O vacancies could be injected
into the thickening oxide film by continuous O dissolution in the metal substrate, i.e. (using
the Kröger-Vink notation):
••− +→ (oxide) O2(metal) i
xO vOO . (5.1)
If sufficiently high concentrations of oxygen have interstitially dissolved in α-Zr (i.e. in the
range of 10 at.% up to the solubility limit of about 28 at.% at 450 K in the bulk [49-50]), the
closed-packed hexagonal structure of α-Zr becomes distorted, thereby forming a variety of
partially-ordered solid-solution phases of O in α-Zr (depending on the local composition; cf.
Refs. [49-50]). These solid-solution phases of interstitially dissolved O in α-Zr are revealed
by the in-situ AR-XPS and RISE analysis as an interfacial ZrOx suboxide layer (see Section
5.3 and Refs. [29-30, 34-35, 42]). The existence of the ZrOx interfacial layer is not revealed
by the cross-sectional HR-TEM analysis (see Fig. 5.3; but it is noted that the oxide/metal
interface is not at all atomically sharp).
It is thus suggested that O vacancies, as generated in the ZrOx interfacial layer by the
slow, but continuous, dissolution of O into the α-Zr substrate, diffuse outwardly through the
98 Chapter 5
O sublattice of the crystalline ZrO2 overlayer towards the oxide/gas interface (to be filled by
chemisorbed O surface species at the gas/oxide interface) according to:
(ads)xO
O
(phys)O vOO2v
(chem)
+→+⋅+••
4434421
''
'e . (5.2)
The thus-established outward flux of O vacancies is accompanied by a net inward flux of O
through the oxide film. The gradient of the O-defect concentration in the oxide layer can vary
with depth below the oxide surface, because of compositional and structural differences, in
particular between the ZrOx interfacial and ZrO2 surficial sublayers (see above). This
suggests that the self-diffusion coefficient of O in the growing oxide layers is not only time-
dependent, but also position- (and thus thickness-) dependent (the self-diffusion coefficient is
proportional to the defect concentration [51]) and, consequently, it is not possible to extract a
single value for the O self-diffusion coefficient in the oxide film from the measured data. As
demonstrated by RISE [29, 34], the later oxidation stages at elevated temperatures (i.e. T ≥
450 K) proceeds by thickening of the ZrO2 overlayer, while the thickness of the resolved
ZrOx interfacial layer remains about constant. This implies that the inward migration of the
ZrOx/α-Zr interface due to continuous O dissolution into α-Zr proceeds at a similar rate as
the inward migration of the supposedly "atomically-sharp" ZrO2/ZrOx interface (as modelled
by RISE [29, 34]) due to the conversion of ZrOx into ZrO2 (i.e. steady-state equilibria of the
O defect concentrations are established at the ZrOx/α-Zr and ZrO2/ZrOx interfaces).
The following processes1 could be rate-limiting at later stages of the Zr oxidation
process at 450 K (i.e. after the contribution of O short-circuit transport has become
negligible): (i) O-vacancy diffusion in the oxide lattice, (ii ) O dissolution into the metal
substrate (i.e. generation of O-vacancies at the ZrOx/α-Zr interface), (iii ) a surface (sub-
)reaction at the O2(g)/ZrO2 interface (e.g. dissociative chemisorption, filling of O-vacancies
at the gas/oxide interface), as argued in Ref. [22]. Ad (i) and ad (ii ): If O-vacancy transport
through the oxide lattice would be much slower than the rate of O dissolution into α-Zr, the
oxide-film thickness would have to decrease upon prolonged oxidation (as is the case for T >
573 K [34]), which is not the case here (see Fig. 5.2). Ad (iii ): If the oxide-film growth rate
would be limited by a reaction rate at the oxide surface, the growth kinetics is likely
independent of the crystallographic orientation of the metal substrate, which is also not the
1 Here electron transport is not considered as a rate-limiting process, because for the growth of such thin and overall non-stoichiometric oxide films (see Section 5.3) by thermal oxidation, electron transport by quantum mechanical tunneling (for film-thicknesses up to about 3 nm) and thermionic (or Schottky) emission is intrinsically fast (see Refs. [2, 53] for details).
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 99
case here (cf. Fig. 5.2). It is therefore concluded that, after the contribution of O short-circuit
transport has become negligible, the overall oxide-film growth rate at 450 K is limited by O
dissolution (i.e. supply of O vacancies into the growing oxide film) at the ZrOx/α-Zr
interface. In view of the results shown in Fig. 5.2, for later stages of oxidation, this implies a
higher O dissolution rate for the more open (i.e. less-densely packed) Zr(101 0) surface, as
compared to the more densely-packed Zr(0001) surface.
The growth mechanism, as proposed here for the thermal oxidation of Zr, pertains to
oxidation at T = 450 K. At even lower oxidation temperatures T < 350 K [20-21], oxygen
molecule dissociation can become slow and the processes of O dissolution in α-Zr and
oxygen anion/vacancy diffusion are no longer thermally activated: in this case ultra-thin
(thickness < 3 nm) [29-30] amorphous oxide films of limiting thickness are formed under
influence of the surface-charge field induced by chemisorbed O species on the oxide surface
[20, 52-54]. In the high-temperature oxidation regime, which commences at T > 573 K for
the oxidation of Zr [34], micrometer thick ZrO2 layers develop for which inward O short-
circuit transport along GBs and cracks (induced by the development of compressive growth
stresses; the Pilling-Bedworth ratio for Zr/ZrO2 is 1.55) becomes predominant [18-19, 55-57].
5.6 Conclusions
Two-stage, tracer oxidation experiments, i.e. successive 16O2(g) and 18O2(g) oxidation steps,
provide a powerful means to reveal the atomic transport mechanisms operating in the growth
of very thin (thickness < 10 nm) oxide films, as developing upon low-temperature oxidation
of metallic surfaces. A change of the rate-governing atomic transport mechanism upon
progress of oxide-film growth can be deduced on the basis of several two-stage oxidation
experiments with different, successive exposure times in 16O2(g) and 18O2(g) atmospheres.
For the first time, two-stage, tracer oxidation experiments have been applied to study
the atomic transport mechanisms during the initial stages of the low-temperature oxidation of
Zr surfaces. It follows that the initial thermal oxidation of bare, single-crystalline Zr(0001)
and Zr(101 0) surfaces at 450 K and at pO2 = 1×10-4 Pa comprises a change of the
predominant transport mechanism from inward O transport by a combination of vacancy and
short-circuit mechanisms to inward O transport by only the vacancy mechanism. The
contribution of O short-circuit transport becomes negligible with increasing oxidation time
owing to a reduction of the oxide-film GB density and the associated equilibration of the GB
100 Chapter 5
structure, in possible combination with the accumulation of space charge in the vicinity of the
oxide-grain boundaries. Oxygen short-circuit transport is more pronounced for the oxidation
of the Zr(101 0) plane (as compared to the Zr(0001) plane), as revealed by a higher initial
growth rate and as in accordance with an on average, smaller grain size in the oxide
overgrowths on the Zr(101 0) surface.
The oxide-film growth rate at 450 K (after the contribution of O short-circuit transport
has become negligible) is governed by the rate of O dissolution into the parent α-Zr substrate
(i.e. the injection of O-vacancies into the oxide at the ZrOx/α-Zr interface), which is higher
for the more "open" Zr(101 0) plane (as compared to the Zr(0001) plane).
Acknowledgements
We are grateful to U. Eigenthaler for the FIB preparation of the TEM lamellae, G. Richter for
the HR-TEM analysis of the oxide overgrowths and P. A. van Aken for provision of HR-
TEM facilities. Further, we are indebted to R. Merkle from the Max Planck Institute for Solid
State Research for helpful discussions on diffusion in oxide grain boundaries.
References
[1] L. P. H. Jeurgens, Z. M. Wang, and E. J. Mittemeijer, Int. J. Mater. Res. 100 (2009)
1281.
[2] A. T. Fromhold, Theory of Metal Oxidation, North-Holland, Amsterdam (1976).
[3] E. Fromm, Kinetics of Metal-Gas Interactions at Low-Temperatures, Springer, Berlin
(1998).
[4] P. Viklund and R. Pettersson, Oxid. Met. 76 (2011) 111.
[5] E. Panda, L. P. H. Jeurgens, and E. J. Mittemeijer, Surf. Sci. 604 (2010) 588.
[6] P. C. J. Graat, M. A. J. Somers, and E. J. Mittemeijer, Z. Metallkd. 93 (2002) 532.
[7] H. Schmidt, M. Gupta, T. Gutberlet, J. Stahn, and A. Bruns, Acta Mater. 56 (2008)
464.
[8] S. N. Basu and J. W. Halloran, Oxid. Met. 27 (1987) 143.
[9] A. Atkinson, Rev. Mod. Phys. 57 (1985) 437.
[10] S. Chevalier, G. Strehl, J. Favergeon, F. Desserrey, S. Weber, O. Heintz, G.
Borchardt, and J. P. Larpin, Mater. High Temp. 20 (2003) 253.
[11] R. W. Cahn and P. Haasen, Physical Metallurgy, North-Holland, Amsterdam (1996).
Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation 101
[12] J. A. Davies, B. Domeij, J. P. S. Pringle, and F. Brown, J. Electrochem. Soc. 112
(1965) 675.
[13] Y. M. Mishin and G. Borchardt, J. Phys. III 3 (1993) 945.
[14] Y. M. Mishin and G. Borchardt, J. Phys. III 3 (1993) 863.
[15] S. Chevalier, F. Desserrey, and J. Larpin, Oxid. Met. 64 (2005) 219.
[16] M. J. Graham, J. I. Eldrige, D. F. Mitchell, and R. J. Hussey, Mater. Sci. Forum 43
(1989) 207.
[17] E. W. A. Young, H. E. Bishop, and J. H. W. Dewit, Surf. Interface Anal. 9 (1986)
163.
[18] J. B. Holt and L. Himmel, J. Electrochem. Soc. 116 (1969) 1569.
[19] M. W. Mallett and W. M. Albrecht, J. Electrochem. Soc. 102 (1955) 407.
[20] N. Cabrera and N. F. Mott, Rep. Prog. Phys. 12 (1948) 163.
[21] F. P. Fehlner and N. F. Mott, Oxid. Met. 2 (1970) 59.
[22] K. O. Axelsson, K. E. Keck, and B. Kasemo, Surf. Sci. 164 (1985) 109.
[23] P. Sen, D. D. Sarma, R. C. Budhani, K. L. Chopra, and C. N. R. Rao, J. Phys. F 14
(1984) 565.
[24] N. Stojilovic, E. T. Bender, and R. D. Ramsier, Prog. Surf. Sci. 78 (2005) 101.
[25] A. L. Lowe and G. W. Parry, Zirconium in the nuclear industry: proceedings of the
3rd International Conference, ASTM, Philadelphia (1977).
[26] B. Kammenzind and M. Limbäck, Zirconium in the nuclear industry: 15th
international symposium, ASTM International, West Conshohocken, PA (2009).
[27] A. Zalar, Thin Solid Films 124 (1985) 223.
[28] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, J. Appl. Phys. 110 (2011)
024904.
[29] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, Surf. Interface Anal. 42 (2010)
588.
[30] A. Lyapin, L. P. H. Jeurgens, P. C. J. Graat, and E. J. Mittemeijer, J. Appl. Phys. 96
(2004) 7126.
[31] R. A. D. Souza, J. Zehnpfenning, M. Martin, and J. Maier, Solid State Ionics 176
(2005) 1465.
[32] Crystallography, Structure and Morphology, ed. by G. Chiarotti, Springer-Verlag,
Berlin (1995).
[33] F. Reichel, L. P. H. Jeurgens, G. Richter, P. A. v. Aken, and E. J. Mittemeijer, Acta
Mater. 55 (2007) 6027.
102 Chapter 5
[34] A. Lyapin, L. P. H. Jeurgens, and E. J. Mittemeijer, Acta Mater. 53 (2005) 2925.
[35] C. Morant, J. M. Sanz, L. Galan, L. Soriano, and F. Rueda, Surf. Sci. 218 (1989) 331.
[36] G. Bakradze, L. P. H. Jeurgens, and E. J. Mittemeijer, J. Phys. Chem. C 115 (2011)
19841.
[37] M. Kilo, G. Borchardt, B. Lesage, O. Kaïtasov, S. Weber, and S. Scherrer, J. Eur.
Ceram. Soc. 20 (2000) 2069.
[38] J. P. Pemsler, J. Electrochem. Soc. 113 (1966) C208.
[39] J. Levitan, J. E. Draley, and C. J. Vandrune, J. Electrochem. Soc. 114 (1967) 1086.
[40] S. K. Sankaranarayanan and S. Ramanathan, J. Phys. Chem. C 112 (2008) 17877.
[41] H. Drings, U. Brossmann, H. D. Carstanjen, A. Szokefalvi-Nagy, C. Noll, and H. E.
Schaefer, Physica Status Solidi A 206 (2009) 54.
[42] L. P. H. Jeurgens, A. Lyapin, and E. J. Mittemeijer, Surf. Interface Anal. 38 (2006)
727.
[43] W. W. Smeltzer, R. R. Haering, and J. S. Kirkaldy, Acta Metall. Mater. 9 (1961) 880.
[44] D. Gryaznov, J. Fleig, and J. Maier, Solid State Ionics 177 (2006) 1583.
[45] P. J. Gellings and H. J. M. Bouwmeester, The CRC handbook of solid state
electrochemistry, CRC Press, Boca Raton, Fla. (1997).
[46] X. Guo and J. Maier, J. Electrochem. Soc. 148 (2001) E121.
[47] C. O. Degonzalez and E. A. Garcia, Appl. Surf. Sci. 44 (1990) 211.
[48] P. Prieto, L. Galan, J. M. Sanz, and F. Rueda, Surf. Interface Anal. 16 (1990) 535.
[49] J. P. Abriata, J. Garcés, and R. Versaci, Bull. Alloy Phase Diag. 7 (1986) 116.
[50] T. B. Massalski, Binary alloy phase diagrams (2nd ed.), ASM International, Materials
Park, Ohio (1990).
[51] P. Heitjans and J. Kärger, Diffusion in condensed matter: methods, materials, models,
Springer, Berlin (2005).
[52] L. P. H. Jeurgens, A. Lyapin, and E. J. Mittemeijer, Acta Mater. 53 (2005) 4871.
[53] F. Reichel, L. P. H. Jeurgens, and E. J. Mittemeijer, Acta Mater. 56 (2008) 2897.
[54] A. T. Fromhold and E. L. Cook, Phys. Rev. 158 (1967) 600.
[55] B. Cox and J. P. Pemsler, J. Nucl. Mater. 28 (1968) 73.
[56] B. Cox and C. Roy, Electrochem. Technol. 4 (1966) 121.
[57] B. Cox and C. Roy, J. Electrochem. Soc. 112 (1965) C188.
Chapter 6
Summary
Up to date, the growth kinetics, the microstructural evolution, as well as the mechanical and
transport properties of oxide layers grown on metals and alloys by thermal oxidation have
been investigated mainly at elevated oxidation temperatures (say, at T > 500 K) and elevated
oxygen partial pressures (0.1 < pO2 < 105 Pa). Contrarily, our knowledge on the low-
temperature (of, say, below 500 K) oxidation process is far from complete and still suffers
from the lack of reliable quantitative experimental data. This is mainly due to the fact that the
oxide films developing on metallic surfaces at low temperatures are typically only very thin
(thickness < 10 nm) and thus delicate and expensive UHV systems for surface preparation
and analysis are mandatory to process the specimens and characterize the oxide film
microstructure. Although important advances in the field of low-temperature oxidation have
been made, the processes that occur and the changes that take place on the metal surface, in
the growing oxide film and at the metal/oxide and oxide/gas interfaces during the initial and
subsequent stages of oxide-film growth are still only partially understood.
The present PhD project addresses the growth kinetics (Chapter 2), chemical
constitution (Chapter 3), morphology (Chapter 4) and atomic transport mechanisms (Chapter
5) of zirconium-oxide films, as grown by the dry, thermal oxidation of single-crystalline Zr
surfaces at low oxidation temperatures. To this end, bare (i.e. without a native oxide) well-
defined single-crystalline Zr(0001) and Zr(101 0) surfaces were exposed to O2(g) in the
temperature range of T = 300-450 K at an oxygen partial pressure of pO2 = 1×10-4 Pa in an
especially-designed UHV system. The current study, for the first time, presents a direct
comparison of the initial oxidation of single-crystalline Zr surfaces with basal and prism
orientations. The relationships between the oxidation kinetics, the developing microstructure
and the crystallographic orientation of the parent metal substrate have been established by
application of various (surface-)analytical techniques (see Section 1.5). On the basis of the
thus-obtained results, a Zr oxidation mechanism in the low-temperature regime (< 500 K) has
been proposed (see below and Section 5.5 for details).
As evidenced by RISE (Chapter 2), after a short initial stage of fast oxide-film
growth, a near-limiting thickness of the oxide film is attained at T < 375 K on both Zr
surfaces. The non-passivating oxidation kinetics of the single-crystalline Zr surfaces at T ≥
350 K are in accordance with previous reports on the thermal oxidation of polycrystalline Zr
104 Chapter 6
surfaces. Distinct differences in the oxidation kinetics for the two Zr substrate orientations
become apparent at T > 375 K: the Zr(101 0) prism plane oxidizes more readily than the
more densely-packed Zr(0001) basal plane under the same experimental conditions (cf. Fig.
2.5). At T > 375 K, the oxidation rate of Zr becomes governed by thermally-activated
dissolution and diffusion of oxygen into the α-Zr substrate. On this basis quantitative
estimates for the anisotropy ratio of the oxygen diffusion coefficient in α-Zr parallel and
perpendicular to the crystallographic c-axis have been made (see Section 2.4.3). It follows
that oxygen diffusion along the Zr[101 0] direction is faster (by about a factor of two at 450
K) than along the Zr[0001] direction.
In Chapter 3 the changes in the local chemical states of Zr and O in the thin
zirconium-oxide films have been studied with increasing oxidation temperature. To this end,
the oxide-film VBs and the APs of Zr and O in the grown oxide films were resolved from
measured XPS spectra of the oxidized Zr surfaces in the oxidation-temperature range of T =
300-450 K. The changes in the shape of the oxide-film VB spectra and the local chemical
states of Zr and O in the oxide films evidence that the oxide films grown at T ≤ 400 K are
predominantly amorphous, whereas a tetragonal ZrO2-like phase develops at T > 400 K. The
formation of the tetragonal zirconia modification in the oxide films developing at 450 K is
supported by ex-situ (post-mortem) HR-TEM analysis (see Fig. 5.3 and text for details). The
resolved Zr 3d5/2 and O 1s photoelectron lines and the Zr MNN and O KL23L23 Auger lines
were combined to construct Wagner plots for Zr and O in the oxide films (Figs. 3.6 and 3.7).
The observed decreases of the Zr and O AP values with increasing oxidation temperature (see
Fig. 3.8) evidence a lowering of the electronic polarizability around core-ionized Zr and O
atoms upon increasing oxidation temperature. It was concluded that the amorphous-to-
crystalline transition of the zirconium-oxide films is accompanied with an increase of the Zr-
O bond ionicity and changes in the first coordination spheres of both Zr and O with
increasing oxidation temperature. The results obtained for the amorphous-to-crystalline
transition of zirconium-oxide films were compared to those for the amorphous-to-crystalline
transition of aluminium-oxide films. It follows that, whereas in Al2O3 films long-range order
develops by the ordering of edge-sharing [AlO4] and [AlO6] polyhedral structural blocks
already present in the amorphous oxide phase, such "building blocks" do not occur in the
amorphous Zr-oxide films.
As shown in Chapter 4 exposure of the bare Zr surfaces to pure oxygen gas at low
temperatures in the range of T = 300-450 K (at pO2 = 1×10-4 Pa) leads to the initial, very fast
Summary 105
formation of a dense arrangement of small oxide nuclei clusters, which completely cover the
bare Zr surface after 300 s of exposure (see Figs.4.2-4.3). The concurrent processes of oxide
nucleation, growth and coalescence leading to a "laterally-closed" oxide layer have completed
within t = 300 s of O2-exposure. The average lateral size of the oxide clusters increases
gradually with increasing oxidation time and with increasing oxidation temperature. The
average lateral size of the oxide clusters after 7200 s of oxidation at T = 300-450 K is in the
range of 2.0-5.0 nm (dependent on t and T).
The mobility of adsorbed O species and/or Zr species on the oxidizing surface and in
the developing oxide becomes promoted with increasing temperature, thereby promoting the
restructuring/reorientation of the oxide clusters into bigger agglomerates, e.g. with increasing
oxidation time at constant temperature, as driven by the Gibbs-Thomson effect.
At T < 400 K for t = 7200 s, as well as for shorter oxidation times at T = 450 K, the
oxide films maintain predominantly amorphous, because no long-range order can develop in
the oxide clusters having confined, small oxide-cluster volumes. Long-range order in the
oxide overgrowths can only develop in the larger oxide agglomerates, leading to the
emergence of a characteristic bonding/non-bonding fine structure in the resolved oxide-film
UVB spectrum as measured by XPS. The boundaries between the evolving oxide
agglomerates are the GBs in the evolving polycrystalline oxide layer.
Fig. 6.1. Schematic illustration showing elementary physical and chemical processes for the initial
stages of dry thermal oxidation of Zr substrate. Note: only transport processes in the direction parallel
to the surface normal have been shown.
106 Chapter 6
In Chapter 5 of this PhD thesis for the first time, two-stage tracer oxidation
experiments have been applied to study the atomic transport mechanisms in very thin (< 10
nm) oxide films, as formed during the initial stages of the low-temperature oxidation of the
bare, single-crystalline Zr(0001) and Zr(101 0) surfaces at 450 K and at pO2 = 1×10-4 Pa. The
observed differences in shape of the measured tracer profiles for different stages of oxidation
(see Fig. 5.4 and Section 5.4 for details) indicate a change in the oxide-film growth
mechanism during oxidation: i.e. a change of the predominant transport mechanism from
inward oxygen transport by a combination of lattice and short-circuit mechanisms during the
initial oxidation stage to inward oxygen transport by only the lattice mechanism during later
oxidation stages.
On the basis of the experimental findings, as obtained by 18O-tracer ToF-SIMS depth
profiling, HR-TEM, AR-XPS and STM (Chapters 2-5), a low-temperature oxidation
mechanism for zirconium at T = 450 K was proposed. Oxygen transport through the
developing oxide film requires coupled fluxes of inwardly migrating O anions and outwardly
migrating O vacancies, as supplied by the slow continuous O dissolution into α-Zr. The
resulting Zr(O) solid-solution phases formed at the metal/oxide interface are evidenced by an
interfacial ZrOx suboxide layer in the in-situ AR-XPS and RISE analysis (see Fig. 2.2 and
Chapter 2 for details). O vacancies, as generated in the ZrOx interfacial layer by the slow, but
continuous, dissolution of O into the α-Zr substrate, diffuse outwardly through the O
sublattice of the crystalline ZrO2 overlayer towards the oxide/gas interface to be filled by
chemisorbed O surface species at the gas/oxide interface (see Fig. 6.1). The thus-established
outward flux of O vacancies is accompanied by a net inward lattice flux of oxygen anions. At
the early oxidation stage, oxygen is transported inwardly via both the lattice and short-circuit
transport mechanism. At later oxidation stages, the contribution of O short-circuit transport
becomes negligible, as attributed to (see Chapter 4 for details) a reduction of GB density in
the oxide-film in association with an equilibration of the GB structure, in possible
combination with the accumulation of space charge in the vicinity of the GB in the oxide-
film. The overall oxide-film growth rate at 450 K is limited by the O dissolution rate (i.e.
supply of oxygen vacancies into the growing oxide film) at the ZrOx/α-Zr interface.
Chapter 7
Zusammenfassung
Bis heute sind die Wachstumskinetik, die mikrostrukturelle Entwicklung, sowie die
mechanischen Eigenschaften und Transporteigenschaften von Oxidschichten auf metallischen
Oberflächen vor allem bei relativ hohen Oxidationstemperaturen (d.h. T > 500 K) und hohen
Sauerstoffdrücken (d.h. 0.1 < pO2 < 105 Pa) untersucht worden. Jedoch ist unser Wissen über
Oxidationsprozesse bei niedrigeren Temperaturen (d.h. unter 500 K) immer noch
unvollständig und leidet unter dem Mangel an zuverlässligen, quantitativen experimentellen
Daten. Dies ist vor allem darauf zurückzuführen, dass die Oxidschichten, die sich auf
metallischen Oberflächen bei niedrigen Temperaturen bilden, in der Regel sehr dünn (Dicke
< 10 nm) sind. Somit sind empfindliche und teure UHV-Anlagen für die
Oberflächenverarbeitung und -analyse erforderlich, um die Proben vorzubereiten und die
Mikrostruktur der Oxidschicht zu charakterisieren. Trotz des enormen Fortschritts im Bereich
der Niedertemperatur-Oxidation in den letzten Jahrzehnten werden die Prozesse und die
Veränderungen, die in der wachsenden Oxidschicht während den anfänglichen und späteren
Oxidschichtswachstumsstadien auftreten, nach wie vor nur teilweise verstanden.
Diese Doktorarbeit widmet sich den anfänglichen Stadien der trockenen, thermischen
Oxidation von reinen, einkristallinen Zr-Oberflächen bei niedrigen Oxidationstemperaturen.
Im Einzelnen befasst sich die Doktorabreit mit der Wachstumskinetik (Kapitel 2), der
chemischen Zusammensetzung (Kapitel 3), der Morphologie (Kapitel 4) und den atomaren
Transportmechanismen in den wachsenden Zirkoniumoxid-Schichten (Kapitel 5). Zu diesem
Zweck wurden blanke (d.h. ohne eine natürliche Oxidschicht), klar definierte, einkristalline
Zr(0001)- und Zr(101 0)-Oberflächen reinem O2-Gas (pO2 = 1×10-4 Pa) bei verschiedenen
Temperaturen von T = 300-450 K in einer speziell angefertigten UHV-Anlage ausgesetzt. In
dieser Doktorarbeit wurde zum ersten Mal ein direkter Vergleich zwischen der anfänglichen
Oxidation von einkristallinen Zr(0001)- und von einkristallinen Zr(101 0)-Oberflächen
durchgeführt. Die Beziehungen zwischen der Oxidationskinetik, der mikrostrukturellen
Entwicklung und der kristallographischen Orientierung des Metallsubstrats wurden durch die
Anwendung verschiedener oberflächenanalytischer Techniken etabliert (siehe Abschnitt 1.5).
Aufgrund der erhaltenen Ergebnisse wurde ein Oxidationsmechanismus für die Oxidation
von Zr im Niedertemperaturbereich (< 500 K) vorgeschlagen (siehe unten und Abschnitt 5.4
für Details).
108 Chapter 7
Die mit spektroskopischer in-situ Echtzeitellipsometrie (RISE) ermittelte
Wachstumskinetik der Oxidschichten für die thermische Oxidation bei T < 375 K ist auf
beiden Zr-Oberflächen durch ein kurzes Anfangsstadium sehr schnellen Schichtwachstums
gekennzeichnet, dem ein zweites Oxidationsstadium folgt in dem die Wachstumsrate der
Schicht sehr klein wird, d.h. die limitierende Oxidschichtdicke wird erreicht (Kapitel 2). Die
nicht-passivierende Oxidationskinetik der einkristallinen Zr-Oberflächen bei T ≥ 350 K ist in
Übereinstimmung mit früheren Berichten über die thermische Oxidation von polykristallinen
Zr-Oberflächen. Erhebliche Unterschiede in der Oxidationskinetik werden beim Vergleich
der beiden Orientierungen des Zr-Substrats bei T > 375 K deutlich: Unter den gleichen
experimentellen Bedingungen oxidiert die offenere Zr(101 0)-Oberfläche schneller als die
dichtgepackte Zr(0001)-Oberfläche (vgl. Abb. 2.5). Bei höheren Temperaturen von T > 375
K wird die Oxidation von Zr durch thermisch aktivierte Lösung und Diffusion von Sauerstoff
in α-Zr Substrat bestimmt. Anhand dieser Grundlage wurden quantitative Einschätzungen des
Anisotropieverhältnisses des Sauerstoff-Diffusionskoeffizienten in α-Zr, parallel und
senkrecht zur kristallographischen c-Achse, vorgenommen (siehe Abschnitt 2.4.3). Daraus
resultiert, dass die Sauerstoffdiffusion entlang der Zr[101 0]-Richtung schneller erfolgt (etwa
um den Faktor zwei bei 450 K) als entlang der Zr[0001]-Richtung.
Kapitel 3 befasst sich mit Veränderungen in der lokalen chemischen Umgebung der
Zr- und O-Spezies in dünnen Zirkoniumoxid-Schichten bei verschiedenen
Oxidationstemperaturen. Zu diesem Zweck wurden die Valenzband-Spektren (VB) und die
Augerparameter (AP) der Zr- und O-Spezies in Oxidschichten aus gemessenen Röntgen-
Photoelektronen-Spektren (XPS) ermittelt. Die Veränderungen in der Form der VB-Spektren
und der lokalen chemischen Zustände von Zr- und O-Spezies in den Oxidschichten weisen
darauf hin, dass Oxidschichten, die bei T ≤ 400 K gewachsen sind, überwiegend amorph sind.
Im Vergleich dazu entwickeln Oxidschichten, die bei T > 400 K gewachsen sind, eine
tetragonale ZrO2-Phase. Die Bildung des tetragonalen Zirkoniumoxids in den Oxidschichten
bei 450 K wurde auch durch eine ex-situ (post-mortem) hochauflösende
transmissionselektronenmikroskopische (HR-TEM) Untersuchung (siehe Abb. 5.3 und Text
für Details) nachgewiesen. Die aus den XPS-Spektren ermittelten Zr3d5/2 und O1s
Photoelektronen-Linien, Zr MNN und O KL23L23 Auger-Linien wurden kombiniert, um die
sogenannten Wagner Diagramme für Zr- und O-Spezies in Oxidschichten zu konstruieren
(Abb. 3.6 und 3.7). Der beobachtete Verlauf der AP-Werte der Zr- und O-Spezies mit
steigender Oxidationstemperatur (siehe Abb. 3.8) weist auf eine Senkung der elektronischen
Polarisierbarkeit in der Umgebung der photoionisierten Zr- und O-Atome bei steigender
Zusammenfassung 109
Oxidationstemperatur hin. Es wurde festgestellt, dass die Umwandlung von amorphem zu
kristallinem Oxid in den Zirkoniumoxid-Schichten mit einem Anstieg der Ionizität der Zr-O-
Bindung und mit Veränderungen in der ersten Koordinationssphäre der beiden Zr- und O-
Spezies bei steigender Oxidationstemperatur begleitet wird. Die erhaltenen Ergebnisse für die
Umwandlung von amorphem zu kristallinem Oxid in den Zirkoniumoxid-Schichten wurden
mit den Ergebnissen für die Umwandlung von amorphem zu kristallinem Oxid in
Aluminiumoxid-Schichten verglichen. Daraus folgt, dass, während sich eine Fernordnung in
den Al2O3-Schichten durch die kantenverknüpfte Anordnung von polyedrischen [AlO4]- und
[AlO6]-Blöcken bereits in der amorphen Oxidphase entwickelt, solche "Bausteine" in
amorphen Zirkoniumoxid-Schichten nicht auftreten.
Wie im Kapitel 4 gezeigt wurde, führt eine Oxidation der blanken Zr-Oberflächen mit
reinem Sauerstoff in einem Temperaturbereich von T = 300-450 K und bei pO2 = 1×10-4 Pa
zu einer sehr schnellen Bildung dicht-angeordneter kleiner Oxidcluster, die die blanken Zr-
Oberflächen schon nach den ersten 300 s der Exposition vollständig bedecken (siehe Abb.
4.2-4.3). Die gleichzeitigen Prozesse der Oxidkeimbildung, des Wachstums und der
Koaleszenz führen zur Bildung einer "seitlich geschlossenen" Oxidschicht. Die
durchschnittliche laterale Größe der Oxidcluster nimmt mit zunehmender Oxidationszeit
allmählich zu, wobei dies bei höheren Oxidationstemperaturen deutlich ausgeprägter ist. Die
durchschnittliche laterale Größe der Oxidcluster nach 7200 s Oxidation bei T = 300-450 K
liegt im Bereich von 2.0-5.0 nm (abhängig von t und T).
Die Beweglichkeit der adsorbierten O- und/oder Zr-Spezies auf den oxidierenden
Oberflächen und in den wachsenden Oxidschichten wird mit steigender Temperatur
gefördert. Das Ergebnis ist eine Umstrukturierung/Umorientierung der Oxid-Cluster zu
größeren Agglomeraten, die vor allem bei höheren Oxidationstemperaturen begünstigt und
vom Gibbs-Thomson-Effekt bedingt wird.
Bei T < 400 K und t = 7200 s, sowie auch für kürzere Oxidationszeiten bei T = 450 K,
sind die Oxidschichten überwiegend amorph, da sich keine Fernordnung im begrenzten
Volumen der Oxidcluster (die jeweils nur etwa 2500 Atome enthalten) entwickeln kann. Eine
Fernordnung in den Oxidcluster kann sich nur in größeren Oxidagglomeraten entwickeln und
ist durch die Bildung einer bindenden/nicht-bindenden Feinstruktur in den ermittelten XPS
VB-Spektren gekennzeichnet. Die Grenzen zwischen den sich entwickelnden
Oxidagglomeraten werden schließlich zu Korngrenzen in der wachsenden polykristallinen
Oxidschicht.
110 Chapter 7
Im Kapitel 5 dieser Doktorarbeit wurden zum ersten Mal zweistufige
Oxidationsexperimente eingesetzt, um die atomaren Transportmechanismen in sehr dünnen
Oxidschichten (Dicke < 10 nm) während den Anfangsstadien der Niedertemperaturoxidation
von blanken, einkristallinen Zr(0001)- und Zr(101 0)-Oberflächen bei 450 K und bei pO2 =
1×10-4 Pa festzustellen. Die beobachteten Unterschiede in der Form der gemessenen Tracer-
Profile für unterschiedliche Stadien der Oxidation (siehe Abb. 5.4 und Abschnitt 5.4 für
Details) weisen auf eine Veränderung des Wachstumsmechanismus während des
Oxidationsprozesses, bzw. auf eine Änderung des vorherrschenden Transportmechanismus
hin: In früheren Oxidationsstadien findet der nach innen gerichtete Sauerstofftransport
hauptsächlilch über eine Kombination von Gitter- und Korngrenzendiffusion statt, während
in späteren Oxidationsstadien der Sauerstofftransport ausschließlich über Gitterdiffusion
erfolgt.
Abb. 7.1. Schematische Darstellung von physikalischen und chemischen Elementarprozessen
während den anfänglichen Stadien der trockenen, thermischen Oxidation von Zr-Substrat.
Anmerkung: Es werden nur senkrecht zur Oberfläche verlaufende Transportprozesse aufgezeigt.
Auf Grundlage der experimentellen Ergebnisse von ToF-SIMS 18O-
Tracertiefenprofilmessungen, HR-TEM, AR-XPS und STM (Kapitel 2-5) wurde ein
Oxidationsmechanismus für Zirkonium bei T = 450 K vorgeschlagen. Der nach innen
gerichtete Sauerstofftransport durch die wachsende Oxidschicht hindurch erfordert
gekoppelte Ströme von nach innen wandernden O-Anionen und von nach außen wandernden
O-Leerstellen, die durch die langsame kontinuierliche Sauerstofflösung in α-Zr geliefert
werden. Die daraus resultierende Zr(O)-Mischkristallphasenschicht an der Metall/Oxid-
Grenzfläche wurde als eine ZrOx-Suboxidschicht mithilfe AR-XPS und RISE-Analysen
Zusammenfassung 111
nachgewiesen (siehe Abb. 2.2 und Kapitel 2 für Details). Die O-Leerstellen, die sich in der
ZrOx Grenzschicht durch eine langsame kontinuierliche Lösung des Sauerstoffs in α-Zr-
Substrat bilden, diffundieren auswärts durch die kristalline ZrO2-Schicht an die Oxid/Gas-
Grenzfläche, wo sie durch chemisorbierte O-Spezies gefüllt werden (siehe Abb. 6.1). Der so
etablierte Strom von O-Leerstellen wird von einem inwärts gerichteten Strom von
Sauerstoffanionen begleitet. In frühen Oxidationsstadien erfolgt der nach innen gerichtete
Sauerstofftransport sowohl über das Gitter als auch über die Korngrenzen. In späteren
Oxidationsstadien wird dagegen ein relativer Beitrag des O-Korngrenzentransports
vernachlässigbar. Dies resultiert aus einer Reduktion der Korngrenzendichte in der
Oxidschicht in Verbindung mit einem Ausgleich der Korngrenzenstruktur in möglicher
Kombination mit der Entstehung einer Raumladung nahe Korngrenzen in der Oxidschicht
(siehe Kapitel 4 für Details). Die gesamte Oxidschichtwachstumsrate wird bei 450 K durch
die O-Lösegeschwindigkeit (d.h. Lieferung von O-Leerstellen in die wachsende Oxidschicht)
an der ZrOx/α-Zr-Grenzfläche beschränkt.
List of used abbreviations
AP Auger parameter
AR-XPS angle-resolved X-ray photoelectron spectroscopy
BE binding energy
EAL effective attenuation length
EMA Effective medium approximation
GB grain boundary
FIB focussed ion beam
FFT fast Fourier transformation
IMFP inelastic mean free path
HBE high binding energy
HR-TEM high-resolution transmission electron microscopy
KE kinetic energy
LBE low binding energy
LEED low energy electron diffraction
LVB lower valence band
MBE molecular beam epitaxy
MOS-FET metal-oxide-semiconductor field-emission transistor
PZL primary zero loss
RHEED reflection high energy electron diffraction
RISE real-time in-situ spectroscopic ellipsometry
SC sputter cleaning, sputter-cleaned
STM scanning tunneling microscopy
ToF-SIMS time-of-flight secondary mass-spectrometry
UHV ultra-high vacuum
UVB upper valence band
VB valence band
List of publications
1. G. Bakradze, L.P.H. Jeurgens, E.J. Mittemeijer, An STM study of the initial oxidation of sinlge-crystalline zirconium surfaces, in preparation (Chapter 4 of this thesis).
2. G. Bakradze, L.P.H. Jeurgens, U. Starke, T. Acartürk, E.J. Mittemeijer, Atomic transport mechanisms in thin oxide films grown on zirconium by thermal oxidation, as-derived from 18O-tracer experiments, Acta Materialia 59(20): 7498-7507 (2011) (Chapter 5 of this thesis).
3. G. Bakradze, L.P.H. Jeurgens, E.J. Mittemeijer, Valence-band and chemical-state analyses of Zr and O in thermally-grown thin zirconium-oxide films: an XPS study, Journal of Physical Chemistry C 115(40): 19841-19848 (2011) (Chapter 3 of this thesis).
4. G. Bakradze, L.P.H. Jeurgens, E.J. Mittemeijer, The different initial oxidation kinetics of Zr(0001) and Zr(10-10) surfaces, Journal of Applied Physics 110(2): 024904 (2011) (Chapter 2 of this thesis).
5. G. Bakradze, L.P.H. Jeurgens, E.J. Mittemeijer, Oxide-film growth kinetics on Zr(0001) and Zr(10-10) single-crystal surfaces, Surface and Interface Analysis 42(6-7): 588-591 (2010).
6. J. Kajaks, G. Bakradze, A. Viksne, S. Reihmane, M. Kalnins, R. Krutohvostov, The use of polyolefins-based hot melts for wood bonding, Mechanics of Composite Materials 45(6): 643-650 (2009).
7. F. Muktepavela, G. Bakradze, L. Grigorjeva, R. Zabels, E. Tamanis, Properties of ZnO coatings obtained by mechanoactivated oxidation, Thin Solid Films, 518(4): 1263-1266 (2009).
8. F. Muktepavela, G. Bakradze, V. Sursaeva, Micromechanical properties of grain boundaries and triple junctions in polycrystalline metal exhibiting grain-boundary sliding at 293 K, Journal of Materials Science, 43(11): 3848-3854 (2008).
9. G. Bakradze, J. Kajaks, S. Reihmane, R. Krutohvostovs, V. Bulmanis, The influence of water sorption-desorption cycles on the mechanical properties of composites based on recycled polyolefine and linen yarn production waste, Mechanics of Composite Materials, 43(6): 573-578 (2007).
10. G. Bakradze, J. Kajaks, S. Reihmane, J. Lejnieks, Correlation between the mechanical properties and the amount of desorbed water for composites based on a recycled low-density polyethylene and linen yarn production waste, Mechanics of Composite Materials, 43(5): 427-432 (2007).
11. F. Muktepavela, G. Bakradze, S. Stolyarova, Nanostructured metal/oxide coatings, Physica Status Solidi C 4(3): 740-743 (2007).
12. F. Muktepavela, G. Bakradze, E. Tamanis, S. Stolyarova, N. Zaporina, Influence of mechanoactivation on the adhesion and mechanical properties of metal/oxide interfaces, Physica Status Solidi C 2(1): 339-342 (2005).
Acknowledgements
This thesis has been performed at the Institute for Materials Science in the University of
Stuttgart and at the Max Planck Institute for Intelligent Systems (formerly Max Planck
Institute for Metals Research), Stuttgart.
First of all, I would like to use this chance to express my deep sense of gratitude to
my supervisor Prof. Dr. Ir. E.J. Mittemeijer for giving me the opportunity to perform the
exciting research project in his department. I have immensely benefited from his expert
guidance, numerous fruitful scientific discussions and continuous encouragement.
It is really hard to express how greatly I am indebted to my daily supervisor Dr.
L.P.H. Jeurgens, who never spared his time for sharing his knowledge by numerous helpful
scientific discussions and corrections of the raw papers. His inexhaustible enthusiasm and
encouragement have supported me during the course of this PhD project and have made a
deep impression on me.
I would also like to express my sincere gratitude to Prof. Dr. E. Roduner and Prof. Dr.
J. Bill for readily accepting to be my thesis examiners.
I thank Katharina Weller for correcting the German version of the summary.
I am grateful to International Max Planck Research School (IMPRS) for providing
financial support for my doctoral work.
I would also like to thank many present and former colleagues at both the Max Planck
Institutes in Stuttgart for their constant technical assistance during my research.
Upon my stay in Stuttgart I have encountered many nice people from all over the
world and I would like to thank them all for making my stay in Stuttgart so pleasant and
enjoyable.
Finally, I have no words to experess my thanks to my family in Riga, who have been
a constant support to me during my stay abroad.
Curriculum Vitae
Personal data
Name Georgijs Bakradze
Date and Place of Birth 22.08.1984, Riga (Latvia)
Marital Status single
Nationality Latvian
Schooling
1991 – 2002 Rigas Daugavgrivas Secondary School, Riga, Latvia
Higher education
Sept. 2002 – June 2005 Bachelor of Engineering (Materials Science)
Riga Technical University, Riga, Latvia
Sept. 2005 – July 2007 Master of Engineering (Materials Science)
Riga Technical University, Riga, Latvia
Research experience
Nov. 2003 – May 2007 Junior Research Fellow
Institute of Polymer Materials, Riga Technical University, Riga,
Latvia
Mar. 2004 – Aug. 2007 Junior Research Fellow
Institute of Solid State Physics, Riga, Latvia
Dissertation
Aug. 2007 – Nov. 2011 PhD at the Max Planck Institute for Intelligent Systems (formerly
MPI for Metals Research) and Universität Stuttgart
Title: Initial oxidation of zirconium: oxide-films growth kinetics and
mechanisms
Erklärung
Hiermit erkläre ich, Georgijs Bakradze, dass ich die vorliegende Doktorarbeit selbstständig
verfasst habe. Es wurden nur die in der Arbeit angegebenen Quellen und Hilfsmittel benutzt.
(Georgijs Bakradze)