You are on page 1of 26

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/222644735

On the high-temperature oxidation of nickel

Article  in  Corrosion Science · January 2003


DOI: 10.1016/S0010-938X(02)00085-9

CITATIONS READS

94 529

1 author:

Reidar Haugsrud
University of Oslo
155 PUBLICATIONS   2,771 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Hydration Thermodynamics of High Temperature Proton Conductors (HydraThermPro) View project

Nitrogen in Oxides (NITROX) View project

All content following this page was uploaded by Reidar Haugsrud on 13 June 2018.

The user has requested enhancement of the downloaded file.


Corrosion Science 45 (2003) 211–235
www.elsevier.com/locate/corsci

On the high-temperature oxidation of nickel


R. Haugsrud *

Department of Chemistry, Centre for Materials Science, University of Oslo, Gaustadall


een 21,
Oslo N-0349, Norway
Received 20 March 2002; accepted 17 April 2002

Abstract

This paper summarizes on some of the extensive experimental data and corresponding
models suggested to account for the oxidation mechanism of Ni in the temperature range 500–
1400 °C. In addition it reports on in-house experimental data from investigations related to the
oxidation of high-purity Ni from 500 to 1300 °C in the oxygen pressure range 1  104 –1 atm
based on TG, measurements of surface kinetics, two-stage oxidation, scanning electron mi-
croscopy, atomic force microscopy, secondary ion mass spectroscopy etc. The main part of
this paper focuses on the more complex models suggested to account for experimental ob-
servations of the oxidation kinetics and the oxide morphology below 1000 °C.
Ó 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Nickel A; High-temperature oxidation C; Oxidation kinetics C; Oxide morphology; Oxidation


mechanism

1. Introduction

Ni is one of the base metals in several of today’s superalloys [1]. In addition to


being mechanically stable, these alloys must withstand severe service conditions with
respect to temperature and oxidizing environments. Although Ni is the major
constituent, NiO does usually not provide the corrosion barrier for these alloys;
superalloys are, generally, chromia or alumina formers [2]. However, since Ni oxi-
dation shows many parallels to systems of industrial relevance, the oxidation
mechanism of Ni has been investigated during almost a century. Regardless of this,
there are still discrepancies and important questions with reference to the oxidation

*
Tel.: +47-22-840659; fax: +47-22-840651.
E-mail address: reidar.haugsrud@fys.uio.no (R. Haugsrud).

0010-938X/03/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 0 - 9 3 8 X ( 0 2 ) 0 0 0 8 5 - 9
212 R. Haugsrud / Corrosion Science 45 (2003) 211–235

mechanism. In addition to reporting in-house experimental data, this paper aims at


summarizing the important findings in the literature on high-temperature oxidation
of Ni, and, moreover, to discuss different aspects related to the oxidation mecha-
nisms of Ni.

2. Literature

After Carl Wagner derived the theory for high-temperature oxidation of metals,
some of the transition metal–oxide systems––including Ni–O––became subject of
intensive research. Through these studies, the relation between parabolic oxidation
kinetics and the predominating point defect in the oxide was verified. To discuss the
high-temperature oxidation mechanism of Ni it is, therefore, appropriate to start
with a brief survey of some of the literature on point defect dependent properties of
NiO.

2.1. Defect dependent properties of NiO

The point defect structure of NiO has been investigated with a number of different
experimental techniques that include electrical conductivity [3–11], thermogravime-
try [12–14], coulometric titration [15,16], tracer diffusion measurements [17–22] and
high-temperature oxidation of nickel [23–26]. From these investigations, NiO has
been concluded to be a metal deficient p-type semiconductor, with nickel vacancies
as the predominating defects. Oxygen pressure dependences of the different point
defect governed processes have been reported and interpreted to reflect that the
vacancies may be either singly or doubly charged.
The non-stoichiometry in NiO is rather low and, by way of example, the con-
centration of nickel vacancies at 1000 °C in 1 atm oxygen is of the order 103 –104
(atomic fraction). Owing to the low concentration of native defects, aliovalent
doping easily influences the defect structure of NiO [6,7,9–11]. Hence, sensitivity
toward impurities is probably one important reason for scatter among the values of
non-stoichiometry reported in the literature. To get reliable data on the native defect
structure of NiO, caution with respect to selection of materials and specimen
treatment must therefore be shown.
Tracer studies have been important in determining transport mechanisms in NiO,
both at high and lower temperatures [17–22]. One of the most comprehensive in-
vestigations was carried out by Atkinson and Taylor [20,21]. Diffusion profiling of
63
Ni by secondary ion mass spectroscopy (SIMS) was conducted as a function of
annealing temperature from 500 to 1400 °C. Through these measurements, Atkinson
and Taylor distinguished between bulk, dislocation and grain boundary diffusion,
and deduced empirical expressions for the diffusion coefficient of the respective
transport mechanism. They also determined the approximate thickness, d, of the
grain boundary for NiO to be 70 A . Measurements of the oxygen pressure depen-
dence of grain boundary diffusion of nickel indicate that the transport occurs ac-
cording to a vacancy mechanism and that singly charged vacancies predominate [21].
R. Haugsrud / Corrosion Science 45 (2003) 211–235 213

Molecular dynamic simulations of diffusion of Ni ions along grain boundaries


suggest, on the other hand, that doubly charged Ni vacancies are responsible for the
cation grain boundary diffusion [27].
Dubois et al. [28] determined the self-diffusion of oxygen in NiO between 1100
and 1600 °C by tracer diffusion measurements with 18;18 O2 . Compared to the self-
diffusion of nickel, the lattice diffusivity of oxygen is lower by more than five orders
of magnitude in the temperature region to which most investigations of the oxida-
tion mechanisms refer. The activation energy for oxygen diffusion in NiO was de-
termined to be as high as 530 kJ/mol. There is some uncertainty as to the nature of
the oxygen point defect but modeling indicates that interstitials is the preferred
defect [29]. Measurement of creep rates in NiO indicates that the mechanism of
oxygen transport is consistent with respect to the high activation energy, but it has
been suggested that the oxygen diffusion rather occurs via vacancies [30]. Atkinson
et al. [31] measured the diffusion of oxygen along grain boundaries and determined
the rate to be much slower than Ni grain boundary diffusion. In fact, it was com-
parable with the diffusivity of Ni in the NiO lattice, and with activation energy of 240
kJ/mol.

2.2. High-temperature oxidation of Ni

The high-temperature oxidation mechanism for unalloyed Ni may, in gross terms,


be divided in two different categories [2]. At the higher temperatures bulk diffusion
determines the oxidation rate, while short-circuit transport mechanisms come into
play at lower temperatures [23,32–38]. The physical distinction may be given by the
temperature as a consequence of the difference in activation energy for the transport
mechanisms. At which range in temperature this change in growth mechanism occurs
depends on factors that affect the grain size of the oxide, e.g. the purity of Ni and
surface preparation. One should also note, since the non-stoichiometry is rather low,
that aliovalent impurities that dissolve in NiO change the native defect concentra-
tion. According to point defect chemistry, ions with a higher valence than Ni(II)
increase the concentration of metal vacancies and, in turn, the oxidation rate,
whereas ions with a lower valence than Ni(II) should, ideally, decrease the oxidation
rate.
Oxidation of nickel in the temperature range above 1100 °C is concluded to be
parabolic. The parabolic rate constant, kp , is proportional to the surrounding oxygen
1=6 1=4
pressure [32–34], with dependences reported in the range kp / pO2 –kp / pO2 . Some
even find that the pressure dependence changes over large experimental ranges in
oxygen pressure [23]. As for the other defect dependent properties, this is attributed
to growth governed by lattice diffusion via singly or doubly charged nickel vacancies
[3–22]. The activation energies reported for the oxidation at these temperatures are
rather consistent and in the order of 220–250 kJ/mol. Oxide scales grown above 1100
°C consist of columnar grains which extend through the entire scale. These oxides are
essentially dense except for some microporosity at the oxide–metal interface.
As the temperature of oxidation decreases (temp < 1000 °C), the kinetics grad-
ually change into a sub-parabolic behavior [35–37]. This means that the oxidation
214 R. Haugsrud / Corrosion Science 45 (2003) 211–235

rate decreases faster as a function of time than as for high-temperature parabolic


kinetics, i.e. oxidation rate / t1=>2 . The discrepancies among the reported rate
constants and, consequently, in the physicochemical properties of NiO deduced from
the oxidation kinetics increase dramatically in this temperature range. It has been
demonstrated that the oxidation behavior of Ni in this temperature range is very
much depending on the pretreatment of the metal prior to oxidation [25,39–44]. In a
rather extensive investigation of nickel oxidation from 600 to 1200 °C, Graham et al.
[25,42–44] focused on the sub-parabolic kinetics, and compared variations in the
instantaneous parabolic rate constant and the oxide morphology with time for
several different surface pretreatments, as well as for different annealing procedures
prior to the oxidation. The oxidation rate was markedly influenced by the surface
treatments of the specimen. At 600 °C, differences in values of the oxidation rate
constant were observed as high as 4 orders of magnitude. This was attributed to
varying amounts of short circuit paths, such as grain boundaries and dislocations.
Different concentrations of rapid diffusion paths with different preparation proce-
dures, also gave rise to variations in the apparent activation energy for the oxidation
process.
The surface morphology of oxidized Ni changes as a function of temperature. In
the low temperature regime a fine-grained surface structure, sometimes with platelets
and whiskers is formed. Increasing the temperature (700 °C < temperature < 900
°C), ridges are often observed at the surface. Above 1000–1100 °C facetted NiO
grains dominates the surface morphology.
The important morphological feature of thermally grown NiO cross-sections,
both for oxidation of the unalloyed metal and particularly when oxidizing Ni-rich
alloys, is the so-called duplex scale structure: The oxide scales can be divided into an
outer region consisting of columnar grains and an inner region with equiaxed grains.
The inner layer has, by different techniques (combined two-stage oxidation and
SIMS profiling, marker studies a.o.) been concluded to grow by inward transport of
oxygen. Since the oxygen diffusivity in NiO is much lower than self-diffusion of
nickel, lattice diffusion cannot explain the growth of the inner fraction of the scale
[20,21,28,31]. Several mechanisms have been suggested to account for the formation
of the inner layer, including inward diffusion of oxygen along grain boundaries,
ingress by oxygen dissociation across cavities during outward nickel diffusion,
opening of microfissures due to anomalous diffusivity along grain boundaries and in
the bulk phase, transport though micropores etc [45–56]. The thickness of the inner
layer can, among other factors, be related to the purity of the material [57].
It has been shown that oxidation of Ni is affected by the presence of oxides of so-
called reactive elements, including CeO2 [58–61], La2 O3 [62,63] and Y2 O3 [64,65]. A
number of papers deal with these effects. It has, generally, been believed that the
reactive elements segregate to grain boundaries either as oxide particles or as ions,
and that these decrease the outward flux of cations along grain boundaries. The
overall oxidation rate will, consequently, decrease with values typically of one to two
orders of magnitude. Effects of reactive element on the high-temperature oxidation
of Ni will not be addressed any further herein, but beneficial effects of surface
coatings of reactive element oxides will be the focus in a forth-coming paper [66].
R. Haugsrud / Corrosion Science 45 (2003) 211–235 215

3. Experimental

3.1. Preparation procedure

The same batch of high-purity Ni (99.999% Goodfellow LTD) has been used
throughout the investigation with specimens of dimensions 10  15  0:5 and 8 
15  0:5 mm. To test parameters that may affect the oxidation resistance of Ni,
different preparation procedures have been applied prior to oxidation. Furthermore,
two different methods have been used to determine the oxidation kinetics.

3.2. Measurements of the oxidation

The oxidation rate has been measured by means of continuous thermogravimetry


with two types of commercial microbalances. In addition oxidation kinetics have
been determined from manometric registration of the pressure decrease in a closed
reaction chamber as a function of time. The manometric equipment was attached to
a mass-spectrometer, which enabled control of the composition of the oxidizing gas.
This system provided a vacuum high enough to avoid oxidation during high-tem-
perature annealing, and it was possible to remove residual water and ethanol ad-
sorbed both to the walls of the reaction chamber and to the specimen surface.
Moreover, gases (most importantly H2 ) dissolved in the metallic matrix could be
removed. The actual amount of H2 present in the specimens was 101 ppm. Pro-
cedures for out-gassing and quantitative determination of the amount of H2 in the
specimens have been given elsewhere [67].
Prior to oxidation the specimens were polished with SiC-paper with different
grades of roughness in the range 500–2400, followed by cleaning in deionized water.
Ultrasound in ethanol was, furthermore, used to degrease the specimen surface.
After drying at room temperature, the specimens were introduced into the reaction
chambers. The rest of the procedures for measurements of the oxidation kinetics
differ in the manometric and thermogravimetric systems.
In the thermogravimetric systems, the specimens were introduced into the reac-
tion zone with a magnetic device from the cold zone of the setup. In this way, the
desired temperature and oxygen pressure could be established before the specimens
were exposed to any high-temperature oxidizing environment. Approximately 20 s
elapsed from the specimen entered the reaction zone to the weight change registra-
tion could be started. Oxygen partial pressures were varied by two different tech-
niques; (i) oxygen was diluted with argon at 1 atm total pressure, and (ii) oxygen was
continuously pumped through the reaction chamber at different total pressures. The
oxidation tests ended by retracting the specimens out of the reaction zone, again with
the magnetic device, and cooled to room temperature.
The manometric reaction chamber consisted of high-purity SiO2 tubes with closed
volumes of 73 and 97 ml. Precise pressure gauges permit an accurate determination
of the pressure decrease as a consequence of the reaction between Ni and the at-
mosphere and, as such, the oxidation rate. The system has a movable furnace so that
the specimen was introduced into the reaction chamber at room temperature. High
216 R. Haugsrud / Corrosion Science 45 (2003) 211–235

vacuum was established at room temperature and the out-gassing during this process
was monitored with the mass spectrometer to ensure that the gas level, especially of
water, was low enough to increase the temperature without any oxidation of the
specimen. A small amount of oxygen was then introduced. The oxidation was
conducted at pressures in the range 22–17 mbar O2 , meaning that oxygen was refilled
to 22 mbar once the pressure reached 17 mbar as a consequence of the oxidation
reaction. This was done to ensure that the same defect situation prevails in the oxide
during the experiments.

3.3. Measurements of surface oxide kinetics

The rate of oxygen dissociation at the NiO surface was determined by mixing
16;16
O2 and 18;18 O2 and measuring the development of the combined gaseous oxygen
molecule 16;18 O2 as a function of time. Additionally, the rate of exchange of oxygen
bonded to the oxide surface with the surrounding gas was estimated from the two-
stage oxidation experiments. In both these cases the dissociation of oxygen on the
empty SiO2 -tube (background) was measured separately and taken into account.

3.4. Characterization

Scanning electron microscopy (SEM) and atomic force microscopy (AFM) have
been applied to characterize surfaces and undersides of the oxide scales. To examine
the fractured cross-sections and the grain structure of the underside of NiO scales,
the Ni core was removed electrochemically. Cross-sections have, additionally, been
characterized by a standard metallographic route.
Two-stage oxidation experiments were followed by transverse SIMS to evaluate
the distribution of the two oxygen isotopes and determine the location of the major
oxide growth.

4. Results

4.1. Oxidation and surface kinetics

High-purity Ni has been oxidized from 500 to 1300 °C in oxygen pressures


ranging from 1  104 to 1 atm. The differences observed in the oxidation kinetics
over the respective temperature range are demonstrated by Fig. 1, which shows the
weight gain as a function of the square root of time. At temperatures above 1100 °C
the oxidation follows the parabolic rate law, whereas, in the range 700–1000 °C, two
different behaviors are encountered: (i) The oxidation rates are initially rapid and
then decrease to become parabolic or (ii) the oxidation rate never conforms to a
parabolic rate law, but decreases more rapidly with time (sub-parabolic). At 600 and
500 °C the oxidation kinetics again conforms essentially to a parabolic rate law.
The high-temperature kp depends on the oxygen pressure yielding an essentially
linear relationship in a logarithmic representation of kp versus (pO2 ). This is illus-
R. Haugsrud / Corrosion Science 45 (2003) 211–235 217

Fig. 1. The weight gain as a function of the square root of time for oxidation of Ni at 800 and 1300 °C in 1
atm O2 . The 800 °C data belongs to the right y-axis and the upper x-axis, whereas the 1300 °C data belongs
to the left y-axis and the lower x-axis.

Fig. 2. The parabolic rate constant as a function of the ambient oxygen pressure at 700, 1000, 1100, 1200
and 1300 °C.

trated in Fig. 2, from which it is also evident that the oxidation rate is essentially
independent of (pO2 ) at 700 °C. One should, however, bear in mind that the kinetics
may be sub-parabolic in the lower temperature range and that the rate constants at
700 °C are more of a qualitative matter. The independence of the ambient oxygen
pressure on the oxidation rate at 700 °C is, furthermore, illustrated in Fig. 3 where
the weight gain is plotted as a function of the square root of time in 1  103 and 1
atm O2 .
On basis of the changes in the oxidation kinetics one may expect a non-linear
temperature dependence, not conforming to one single Arrhenius relation. This is
also the case as seen from Fig. 4. Essentially linear behaviors are observed for the
low (temperature < 800 °C) and the high (temperature > 1000 °C) temperatures
218 R. Haugsrud / Corrosion Science 45 (2003) 211–235

Fig. 3. The weight gain as a function of the square root of time for oxidation at 700 °C in 1  103 and 1
atm O2 .

Fig. 4. The parabolic rate constant as a function of the inverse absolute temperature from 500 to 1300 °C
in 1 atm O2 .

with apparent activation energies of 150 and 200 kJ/mol, respectively. Between these
regions the oxidation rate goes through a transition with lower temperature de-
pendence.
Fig. 5 illustrates the difference in the initial oxidation rate (1 h oxidation) for a
specimen annealed 20 h and a specimen subjected directly to the oxidizing envi-
ronments. The kp of the specimen subjected to the annealing is lower by a factor of
3. Compared to other investigations this difference is small [25,42–44]. This is
probably due to the fact that the grain size in the starting material already was rather
coarse and that the annealing, therefore, did not lead to any significant grain growth
in the Ni metal. In other words, comparable grain sizes is formed in the initial oxide
and, consequently, similar oxidation rates.
R. Haugsrud / Corrosion Science 45 (2003) 211–235 219

Fig. 5. The weight gain as a function of the square root of time during the first hours of oxidation at 900
°C in 0.02 atm O2 for a specimen subjected to vacuum annealing and a specimen oxidized directly after
grinding.

Fig. 6. The amount of 16;18 O2 accumulated in the reaction chamber of the MS-assembly as a function of
time from a 1:1 mixture of 16;16 O2 and 18;18 O2 at 800 °C.

The oxygen dissociation rate at the NiO surface can be estimated from Fig. 6,
which shows the evolution of 16;18 O2 from a 1:1 mixture of 16;16 O2 and 18;18 O2 . The
linear rate constant for the oxygen dissociation is furthermore shown as a function of
the inverse absolute temperature from 500 to 800 °C in Fig 7.

4.2. Atomic force microscopy

AFM was applied to characterize the surface oxide and the underside of the oxide
scales for reaction conditions where fine-grained oxide structures were formed. Fig. 8
shows a topographic view of the surface, (A), and the underside, (B), of an oxide
scale formed under oxidation at 600 °C in air for 27 h (yielding a thickness of 3.7 lm).
220 R. Haugsrud / Corrosion Science 45 (2003) 211–235

Fig. 7. The rate of oxygen dissociation on NiO as a function of the inverse absolute temperature from 500
to 900 °C.

Fig. 8. Topographic AFM micrograph of the surface (A) and the underside (B) of the oxide scale formed
during oxidation at 600 °C in air for 27 h.

There is no significant difference in the appearance of the oxide at the two interfaces
and the grain diameter is in the range 200–400 nm. The oxide grain size is generally
uniform as illustrated by Fig. 9.

4.3. Scanning electron microscopy

Fig. 10A–D show the cross-section of the oxide scales formed at 700, 800, 900 and
1200 °C in air. At 700 °C, in Fig. 10A, a rather fine-grained oxide is formed both
toward the oxide–metal and the oxide–gas interface meanwhile the intermediate
layer has a coarser and more columnar appearance. When increasing the tempera-
ture of oxidation the familiar duplex structure appears with a coarse outer region
and an inner fine-grained, rather porous region. This is demonstrated for the spec-
R. Haugsrud / Corrosion Science 45 (2003) 211–235 221

Fig. 9. Topographic AFM micrograph of the surface formed during oxidation of Ni at 600 °C in air.

Fig. 10. SEM-micrographs showing cross-sections of oxide scales formed during oxidation in air at 700 °C
(A), 800 °C (B), 900 °C (C) and 1200 °C (D) in 1 atm O2 .

imen oxidized at 800 and 900 °C (Fig. 10B and C). With a further increase in the
temperature coarse columnar grains are formed that extend through the scales, from
222 R. Haugsrud / Corrosion Science 45 (2003) 211–235

Fig. 11. SEM-micrograph of the surface morphology formed during oxidation at 900 °C in air. The
micrograph illustrates the rather extensive formation of oxide ridges formed under intermediate temper-
atures.

the oxide–gas interface to the oxide–metal interface. This is illustrated here in Fig.
10D with a cross-section taken from a specimen oxidized at 1200 °C. The grain
diameter is in the order of 30–50 lm.
At the intermediate temperature of oxidation the surface of the oxide exhibits a
ridge like morphology. This is evident from Fig. 10C further illustrated in Fig. 11.

4.4. Secondary ion mass spectroscopy

Two-stage oxidation experiments with subsequent transverse SIMS profiling have


been performed for different oxidation conditions, and Fig. 12 presents the distri-
bution for 16 O and 18 O at 500, 700 and 800 °C for 1 lm thick oxide scales. The
reaction with oxygen corresponding to the first 500 nm of the scales was done in
16;16
O2 and the last 500 nm in 18;18 O2 . One should note that it is more difficult to
locate the oxide–metal interface from the profile of the lower temperature (500 °C).
This behavior reflects that the oxide–metal interface is non-uniform.
SIMS profiles of thicker scales, 17 and 10 lm at 700 and 800 °C, respectively,
are shown in Fig. 13. In the experiment at 700 °C, only the last 28% of the overall
reaction with oxygen was done in 18;18 O2 .

5. Discussion

The results from this investigation agree with observations from the literature on
high-temperature oxidation of Ni. The oxidation kinetics are parabolic at temper-
ature P 1100 °C, whereas a sub-parabolic behavior is observed at the lower tem-
peratures, temperature 6 1000 °C (cf. Fig. 1). There is a strong relation between the
R. Haugsrud / Corrosion Science 45 (2003) 211–235 223

Fig. 12. Transverse SIMS profiles across an oxide scale formed after two-stage oxidation in 16;16 O2 fol-
lowed by oxidation in 18;18 O2 at 500 °C (A), 700 °C (B) and 800 °C (C). The overall oxide thickness is 1
lm. Normalized profiles of the 18 O signal is shown in D.

Fig. 13. SIMS profiles across oxide scales formed during two-stage oxidation at 700 °C (A) and 800 °C
(B). The overall oxide thickness is 17 and 10 lm at 700 and 800 °C, respectively.

oxidation kinetics and the oxide morphology, as the morphology also changes with
temperature (cf. Figs. 8–11). From the following sections it becomes clear that the
origin of these relations may be quite complex and depend on several processes. To
determine the oxidation mechanism of Ni, one must identify and account for how
the different processes that affect the oxidation respond to changes in the reaction
224 R. Haugsrud / Corrosion Science 45 (2003) 211–235

conditions. On this basis it is important to analyze the changes in the oxidation rate
and the oxide morphology with changes in the temperature and the oxygen pressure.

5.1. Oxygen pressure dependence

The oxidation mechanism in the high-temperature region is relatively straight-


forward, and it is well established that the oxidation rate is determined by outward
lattice diffusion of Ni via vacancies. In this investigation the oxygen pressure de-
pendence changes from essentially kp / (pO2 )1=6 at 1000 °C to close to kp / (pO2 )1=4
at 1300 °C (cf. Fig. 2). By assuming that the diffusion proceeds through native nickel
vacancies, the dependence indicates that the relative ratio between doubly and singly
charged vacancies decreases with increasing temperature. This is the opposite be-
havior of what has been observed from electrical conductivity measurements where
the number of doubly charged vacancies increases with increasing temperature [3].
An alternative interpretation of the pO2 -dependence is to assume that doping affects
the oxidation at the lower temperatures and that its influence gradually disappears as
the native defect concentration increases with increasing temperature. Furthermore,
it may also be a question of whether the overall oxidation rate still is affected by
grain boundary transport at 1000–1100 °C, and that another point defect predom-
inate in the grain boundaries than in the oxide lattice. If so, this corresponds to the
pressure dependence expected on basis of simulations of the grain boundary defect
structure [27]. However, the fact that the rate of oxidation is essentially independent
of oxygen pressure at 700 °C may also point in direction of effects of doping (cf. Figs.
2 and 3).

5.2. Temperature dependence

The temperature dependence of the oxidation in Fig. 4 shows three different re-
gimes. At the lower (temp < 800 °C) and higher temperatures (temp > 1000 °C), the
oxidation rate follows an Arrhenius behavior, although with different activation
energies. The values of these activation energies are expected to reflect transport
along grain boundaries and/or dislocations at the lower temperature, and in the
oxide lattice at the higher temperatures. The apparent activation energies obtained
from Fig. 4 are somewhat lower (150 vs 170 kJ/mol and 200 vs 240 kJ/mol) than
those determined by Atkinson and Taylor [20,21], for tracer Ni diffusion along grain
boundaries and in the bulk.
The intriguing with the data presented in Fig. 4 is, actually, not the linear end
regions of the plot, but rather the intermediate transition region where the oxidation
rate shows a weak temperature dependence. This behavior indicates that at least
three processes are important in describing the overall oxidation and, moreover, that
the influence of the temperature on one of these are stronger in a certain temperature
interval. One alternative combination of processes in the scale that could explain this
behavior is: lattice diffusion––grain boundary diffusion––grain growth: Grain
growth is slow at the lower temperatures and the oxidation should be essentially
parabolic, although governed by transport along grain-boundaries. Increasing the
R. Haugsrud / Corrosion Science 45 (2003) 211–235 225

temperature, the rate of grain growth increases accordingly. If the rate controlling
grain size increases more rapidly with temperature than the overall transport across
the scale, the dependence of the oxidation rate on the temperature will decrease. At
higher temperatures, lattice diffusion predominates and a ‘‘normal’’ Arrhenius be-
havior will be retained even if grain growth is more pronounced under these con-
ditions. Another similar possibility is that the influence from inward short circuit
transport of oxygen decreases faster than the diffusional processes contributing to
the overall oxidation. The fact that the flat region appears between 800 and 1000 °C
supports the mechanism based on grain growth, since SIMS studies show that in-
gress of oxygen does not significantly affect the oxidation rate above 700 °C [24].
Peraldi et al. [34] observed a similar temperature dependence of Ni oxidation, but
did not comment any further the flat region of the temperature dependence. In
principle, this behavior should be encountered also in other systems where the oxi-
dation rate is governed or partly governed by short circuit transport mechanisms in
certain temperature regions.

5.3. Oxygen transport and inward growth

Formation of duplex oxide scales during high-temperature oxidation has been


subject of considerable experimental studies and discussions in the literature [24,45–
55]. Such scale structures are observed for NiO formed at temperatures below 1000
°C as shown in Fig. 10. Transverse SIMS profiling of scales formed during two-stage
oxidation verifies that growth of the inner region of the scales is at least partly as-
sociated with inward, probably gaseous, transport of oxygen (cf. Figs. 12 and 13 and
Refs. [24,68]). For scales formed on high-purity Ni, the inner layer is generally
thinner than the outer columnar part of the oxide, although its relative thickness
increases with decreasing temperature. In general, the relative thickness of the inner
layer depends on the purity of the starting material, decreasing with increasing pu-
rity. Peraldi et al. [69] analyzed the ratio between the thickness of the two layers for
different overall scale thickness after oxidation at 700 °C in flowing oxygen. The ratio
was constant as a function of time and equal to 2  0:2.
Different mechanisms have been proposed to account for the inward oxygen
transport through the oxide scale. Modeling of oxygen diffusion in NiO suggests that
motion of interstitial ions is responsible for any oxygen transport, both in the bulk
and along the grain boundaries [29]. The inward oxide growth can, indeed, not be
explained by oxygen lattice diffusion, since the diffusion coefficient of oxygen is far
too low ðDO ½lat=DNi ½latÞ < 105 [20,28]. Grain boundary diffusion coefficients of
oxygen estimated from theoretical calculations and experimental studies are in the
same order as lattice diffusivity of Ni ðDO ½gb  DNi ½latÞ [20,31]. From the relative
ratio of the diffusivity of oxygen and Ni along grain boundaries (103 –105 ), At-
kinson et al. [24,31] concluded that inward grain boundary diffusion of oxygen does
not contribute significantly in the overall oxidation of Ni.
In conclusion, since the oxygen diffusion is slow in NiO, oxygen must penetrate the
scale as gaseous species in agreement with isotope/SIMS studies. Hence, there must
exist a continuous, or partly continuous network of microchannels or interconnected
226 R. Haugsrud / Corrosion Science 45 (2003) 211–235

porosity across the oxide scale. Such microfissures have been observed by in situ
SEM during oxidation of copper forming Cu2 O and by TEM characterization of
scales on Ni–0.1%Cr alloys [36,38,70]. Robertson and Manning [49] envisage a per-
manently open network of fissures. This is not so likely, since the oxidation kinetics
either is parabolic or sub-parabolic and one should expect the kinetics to become
linear like with permanently open fissures. Atkinson and Smart [48] suggested that the
fissures have a transient existence, thus being continually generated and resealed.
Consequently, the number of fissures open at any one instant in time is only a small
fraction of the total number of fissures that have contributed to the oxide growth,
and, hence, the scale keeps its structural integrity. Traces from healed fissures have
been detected from SIMS depth profiles by analyses of the polyatomic oxygen signals
[68].
Three main routes have been suggested to account for the presence of microfis-
sures and inward growth of NiO: (i) dissociation of the scale into and along defects
(porosity and line defects) [45,47,52,54], (ii) stress-induced fissuring in the oxide
[24,48,49,55] and (iii) opening of microfissures as a consequence of differences in the
rate of diffusional deformation across the growing oxide scale [53].
(i) Several authors have accounted for inward oxygen transport and opening of
microfissures by mechanisms based on dissociation of the scale: Cations and cation
vacancies diffuse in opposite directions for an oxide growing by outward metal
diffusion, and the incoming vacancies must be annihilated at or close to the oxide–
metal interface. Several mechanisms are proposed to accommodate these vacancies
at the internal interface [56,71–73]. If the processes responsible for vacancy annihi-
lation are slow, vacancies coalescence, and pores develop at the interface. This leads
to discontinuities for outward metal diffusion. Nevertheless, metal ions continue to
diffuse outward from the outer surface of the pore as a consequence of the gradient
in the chemical potential across the scale. Correspondingly, the metal activity de-
creases, whereas the oxygen activity increases at the outer surface of the pore.
Through this process, oxygen dissociates into the pore resulting in inward oxygen
transport. By extending this mechanism fissures have been suggested to open up
along grain boundaries due to preferential outward metal transport along these rapid
diffusion paths [52,56]. Such fissures cause an even more direct ingress of oxygen,
compared to the mechanism based merely on dissociation [45–47,54]. These two
mechanisms are schematically illustrated in Fig. 14A and B.
(ii) In general, oxide scales growing by outward diffusion will, as a consequence of
the adhesive forces preventing the scale from detaching the metallic core, be in a state
of compressive stresses parallel to the oxide–metal interface and tensile stresses
normal to the oxide–metal interface. Formation of new oxide within a growing scale
normally produces large compressive stress. For an ideal oxide with a uniform stress
distribution, compressive stresses should prevent formation of microfissures. How-
ever, since oxides, commonly, have non-uniform grain structures, growth along
grain boundaries within the scale may well yield localized tensile stress [74–76]. Since
both compressive and tensile forces may be generated as a consequence of growth
within the scale, healing of some grain boundary fissures may create stress fields,
which again open others in a self-perpetuating way.
R. Haugsrud / Corrosion Science 45 (2003) 211–235 227

Fig. 14. Schematic illustration of the dissociation mechanism (A) and the formation of microfissure based
on anomalous rates of diffusion and accordingly dissociation along grain boundaries and in the bulk (B).

(iii) Strains are generally accommodated by dislocation glide, but since too few
slip systems are likely to be active during oxidation at intermediate and low tem-
peratures, creep is the only mechanism to relieve strain [77]. The main mechanisms
responsible for deformation of growing oxide scales will, therefore, be diffusional
flow and grain boundary sliding [78,79]. Combinations of tensile and compressive
forces give rise to an elongation of the oxide grains for scales growing by outward
diffusion. If these oxide scales are to remain dense and coherent, grain boundary
sliding must occur [78]. On this basis, Kofstad [53] advanced a model describing
formation of fissures based on the differences in rates of deformation across growing
oxides as a consequence of the gradient in oxygen activity. If creep, for instance, is
more oxygen pressure dependent than grain boundary sliding, fissures will have a
tendency to form in the interior of the scale [53]. It has been argued that a combi-
nation of these deformation mechanisms is more difficult for scales with columnar
microstructures than for equiaxed grain structures [80]. However once an inner
equiaxed layer is formed, grain boundary sliding may occur in the inner region, and
provided there is an electrochemical potential gradient across this region, the pre-
suppositions made by Kofstad will be valid.
Microfissures clearly are important in describing the overall oxidation mecha-
nisms of Ni. With respect to the above models, it is interesting to analyze whether the
influence of inward oxygen transport changes as a function of time during oxidation.
228 R. Haugsrud / Corrosion Science 45 (2003) 211–235

Moreover, to speculate on if and why the number of open fissures changes as a


function of time during the oxidation. Experimentally, isotopic labeling of the oxide
with oxidation in 16;16 O2 and 18;18 O2 and subsequent SIMS profiling may give a clue.
Harris and Atkinson [68] have shown that the relative amount of Ni18 O close to the
oxide–metal interface increases from scales with 2.5–4 lm thickness for oxidation at
600, 700 and 800 °C. SIMS profiles shown in this investigation (cf. Figs. 12 and 13),
on the contrary, indicate that the relative influence of inward growth decrease as a
function of time. As previously mentioned, Peraldi et al. [69] have observed that the
thickness of the two layers is constant with time at 700 °C. In other words, these
three investigations points in different directions with respect to variations in the
number of open fissures during the course of the oxidation. One should recognize,
when comparing results from different investigations, that the inward growth is
probably sensitive to differences in the surface finish of the specimen and purity of
the Ni–NiO interface.
In considering different aspects of the mechanisms suggested to explain the for-
mation of fissures, (i)–(iii), they all rely on available space at the metal–oxide in-
terface. This space may form as a consequence of pileup of the incoming vacancies.
Accordingly, the relative ratio of the reactions removing vacancies at the metal–
oxide interface and the rate of incoming vacancies is essential and, as such, the
processes keeping the adherence between the scale and the metallic core. As long as
the oxide–metal interface are relatively intact, vacancies can either be removed by
annihilation at interfacial sinks by dislocation climb or through injection of the
vacancies into the metal [56,71–73]. In the oxide, the plastic deformation is also
important since this will partly determine whether a pore or a fissure forms when the
oxide detaches the metal. If the deformation processes in the oxide are fast enough
the cavity initiated at the metal–oxide interface will be taken into the oxide and either
eliminated by sintering or moved outward by dissociation and deformation.

5.4. Evolution of grain structure

Graham et al. [25,42–44] demonstrated that the oxidation rate is strongly de-
pendent on the surface morphology and texture of Ni. At 600 °C the initial oxidation
rate was observed to vary with as much as 4 orders of magnitude for different
pretreatments of the Ni specimen. Moreover, the very initial oxide is essential in
determining the overall, long-term oxidation rate. This is particularly important at
the lower temperatures, since the grain growth is slow and the grain size remains
relatively constant with time. It is therefore of interest to analyze the evolution of the
oxide grain structure during the oxidation and how different factors may influence
the grain structure of the scale.
During the initial period of the oxidation when a continuous layer has been
formed, the ratio between the rate of grain growth and nucleation of new oxide
grains are important. As long as the scale is very thin, the grain boundary area may
actually not be important to the overall oxidation rate, even though the grain
boundary diffusivity is much higher than bulk diffusion. However, once a quasi
steady state has developed, the grain size of the scale becomes important. Mea-
R. Haugsrud / Corrosion Science 45 (2003) 211–235 229

Fig. 15. Instantaneous consume of oxygen as a function of time for oxidation of Ni at 800 °C in 0.02 atm
O2 . The figure includes also the oxygen dissociation rate under similar conditions.

surements of the rate of oxygen dissociation (cf. Figs. 5 and 6) during the initial
oxidation show that after the first 5–10 nm of oxide scale has been formed, the rate
of oxygen dissociation is faster than the instantaneous consume of oxygen by the
oxidation reaction (see Fig. 15). This means that after the first 150 s of the oxidation,
the Wagnerian parabolic conditions are satisfied with respect to equilibrium at the
oxide–gas interface.
It has been shown that the oxidation rate varies for different orientations between
the oxide and the metal grains [81,82]. This may explain why thin scales, generally,
are more non-uniform than thicker scales (high temperatures vs low temperatures
and short exposures versus longer exposures). The faster growing orientations
eventually will overgrow the slower one, but at low temperatures this may take some
time. As previously argued, this presumably is the reason for the non-uniform oxide–
metal interface at 500 °C evident from the SIMS data (cf. Fig. 12A).
The tendency to form columnar grains is small in the beginning of the oxidation,
since the rate of nucleation is high compared to the processes that promote a col-
umnar structure. New oxide nuclei form on the highly defective surface without
much change to organize. However, as the scale grows, columnar grains become
more favorable and new oxide is formed by extension of already existing grains. The
columnar structure is also eventually stabilized as a result of lateral compressive and
longitudinal tensile stresses in the scale. Consequently, the grain structure shows
increasing columnar tendency with time and temperature.
In further growth of the scale, the processes occurring at the oxide–metal interface
will, as we have seen, be just as important as growth at the surface (oxide–gas
230 R. Haugsrud / Corrosion Science 45 (2003) 211–235

interface). The relative ratio between the oxidation rate––i.e. the rate at which Ni
vacancies reach the internal interface––and the rate of deformation and removal of
vacancies at the interface is essential. Once the scale detaches the metal, inward
oxygen transport may be initiated, and the duplex structure will start to form. Since
inward growth is most pronounced below 700 °C (cf. Fig. 12) and also since the grain
growth is slow at the lower temperatures, one should expect the low temperature
scales to become rather equiaxed. This is verified from the micrographs of the oxide
morphology (cf. Figs. 8–11). At 700 °C, Fig. 10A, one should note that there is a
tendency to a columnar structure in the middle of the scale, whereas the grains in the
surface region and at the oxide–metal interface are fairly equiaxed. The fine-grained
surface is probably a consequence of the large difference between grain boundary
and bulk diffusivities and that these actually are formed by the same mechanism as
the ridges in Fig. 11 (further discussed below). One may also speculate on whether
the columnar intermediate structure at 700 °C is formed by recrystallization of the
top layer, since the diffusional transport is slow at this temperature.

5.5. Grain growth during oxidation

On the basis of simple models and empirical expressions for the diffusivities given
in the literature, it is possible to estimate the rate governing grain size of the NiO.
Atkinson et al. [83] applied Smeltzer et al.’s [84] model, which expresses the effective
diffusion coefficient as a combination of lattice diffusivity and grain boundary diffu-
sivity, and showed that kp for oxidation of Ni in 1 atm O2 may be described by
kp ðoverallÞ ¼ 6:4fDNi ½lat þ ð2dDNi ½gbÞ=gg, where d is the grain boundary width and
DNi [lat] and DNi [gb] are, respectively, the tracer diffusion coefficients of Ni in the
lattice and the grain boundaries. Now, by using Atkinson and Taylor’s [20,21] data
based on 63 Ni diffusion in NiO, one may apply the kp s from Fig. 4 to estimate the rate
controlling grain size as a function of temperature. This is presented in Fig. 16, and

Fig. 16. Estimated values for the rate controlling grain size in NiO as a function of the inverse absolute
temperature.
R. Haugsrud / Corrosion Science 45 (2003) 211–235 231

the agreement with the experimental observations from the microscopy above 700 °C
is rather good (Figs. 8–11). At the lower temperatures, however, the grain size ob-
served by microscopy is a factor 2–3 lower than the estimated grain size in Fig. 16.
This is in agreement with Atkinson [85], who also observed from his data on oxidation
kinetics and microstructure of NiO that the rate controlling grain size predicted from
the oxidation rate constants always was larger than the coarsest grains observed by
microscopy. It was suggested that the reason for this discrepancy was due to the
presence of low angle grain boundaries with relatively slow transport rates.
The data from Fig. 16 do not follow one simple Arrhenius behavior, but one may
still use them to estimate an apparent activation energy for grain growth in a growing
NiO scale. Based on the previous considerations one should expect the flatter region
of the temperature dependence of kp to be most representative for the influence of
grain growth on the overall rate of oxidation. On this basis an activation energy for
grain growth of 90 kJ/mol can be calculated from the temperature range 800–1000 °C.
One should recognize that the activation energy calculated based on instantaneous
parabolic rate constants represents an average for the given situation at the time t.
There have been a few investigations where the activation energy for the grain
growth in NiO has been determined. Atkinson et al. [86] and Rhines and Connell [87]
measured activation energies from microscopy observations of the grain size in
thermally growing NiO scales under the assumption that the grain growth follows an
essentially parabolic behavior. The activation energy of 90  40 kJ/mol is in agree-
ment with the one obtained here (cf. Fig. 16). These are, however, much lower than
the value obtained by Iida [88] of 230 kJ/mol from examination of the sintering of
NiO powders between 1500 and 1700 °C. Since grain growth in a ceramic body is
closely related to the diffusivity of the slower moving species, its activation energy
should reflect the one for oxygen diffusivity, i.e. 250 kJ/mol for NiO. The activation
energy reported by Iida [88] is therefore more consistent with respect to this prin-
ciple.
Differences in the activation energies may often be accounted for by different
purity of the materials. However, here the difference between 90 and 250 kJ/mol is
presumably too large to be explained by the presence of secondary phases in the
ceramic body. The value of 90 kJ/mol is remarkably low, comparing it with the
activation energies of self-diffusional processes in NiO. It has been speculated
whether the assumption that grain growth conforms to a parabolic rate law made by
Atkinson et al. [86] and Rhines and Connel [87] is wrong [37]. Monceau et al. [89]
demonstrated that the grain size may change considerably with time below 1000 °C,
but did not give any indications on the kinetics of grain growth. It has been observed
during the present investigation that grain growth is most pronounced in the first
period of the oxidation. This is consistent with results from the extensive work of
Atkinson [85], which addresses the development of the oxide grain structure in NiO.
He observes that the oxide grain size for a given distance from the oxide–metal in-
terface is essentially constant with time after 5 h of oxidation.
Grain growth requires transport of both oxygen and Ni ions. Since NiO scales
grow both by inward short circuit transport of oxygen and outward short circuit
transport of Ni, one may speculate whether grain growth may be affected by the
232 R. Haugsrud / Corrosion Science 45 (2003) 211–235

counter transport in the scale. Immediately, one might expect that counter transport
lead to growth inside the scale, resulting in a strained scale where grain growth is
slow. It has been observed that the activation energy for the chemical diffusion co-
efficient in the near-to-surface layer is 90 kJ/mol [90]. If the oxygen transport oc-
curs by gas transport or as surface diffusion of dissociated oxygen species with
similarly low or even lower activation energy, one may speculate if surface diffusion
is rate determining in grain growth during thermal NiO growth.

5.6. Surface morphologies

The surface of thermally grown NiO scales exhibits different morphologies with
different reaction conditions, and may also change within the course of oxidation for
one given reaction condition. During oxidation at low temperatures, the oxide scales
may be rather non-uniform as a consequence of different growth rates for different
grain orientations. At the low temperatures, disc-shaped NiO platelets have also
been observed [91,92]. In general, for different metal–oxide systems, such platelets
are usually encountered during oxidation at low temperatures and especially in
contact with water containing atmospheres. Analyses of the texture of this mor-
phological feature have revealed that they are usually generated in connection with
oriented defects like microtwins or stacking faults [91,92]. Their growth rates depend
on the relative growth by ledge displacement and surface diffusion [34]. NiO platelets
have not been observed during these experiments except for when the oxidation has
been performed in wet air (2% water vapor).
Increasing the temperature platelets disappear and become replaced by a ridge
like surface morphology. These ridges seemingly follow patterns of the grain
boundaries as evident from Figs. 10 and 11. Two different mechanisms have been
suggested to account for their existence. An increased growth above grain boundary
regions may be a consequence of the rapid outward grain boundary diffusion. It has
also been suggested that ridges form on grain boundaries in the surface region due to
stress generations as a consequence of oxide formation inside the scales [34]. In this
mechanism the scale is assumed to relax by plastic deformation squeezing the oxide
up into ridges. The amount of ridges decrease with increasing exposure and this may
be in favor of the former explanation. This may reflect that the overall outward
transport decreases with increasing scale thickness meanwhile the rate of surface
diffusion dispersing the material on the surface remains essentially constant as a
function of thickness. In addition, generation of stresses is commonly believed to
increase during scale growth and from this the amount of surface ridges should
increase rather than decrease as a function of time.

6. Summary and concluding remarks

In this paper the high-temperature oxidation of Ni has been revisited. The ex-
periments have been performed as a function of temperature in the range 500–1300 °C
and oxygen pressure from 1  104 to 1 atm. The surface kinetics and oxidation
R. Haugsrud / Corrosion Science 45 (2003) 211–235 233

kinetics have been investigated by mass spectroscopy and thermogravimetry and the
oxide scales have been characterized by AFM and SEM. Two-stage oxidation has
been followed by transverse SIMS profiling to elucidate the location of oxide growth
in the scale. In addition to the in-house experimental data, the discussion of this paper
also builds on some of the extensive literature on high-temperature oxidation of Ni.
The high-temperature oxidation mechanism of Ni above 1100 °C is well estab-
lished. The oxidation behaves according to a parabolic rate law, and the oxidation
rate is governed by outward lattice diffusion of Ni via either singly or doubly charged
Ni vacancies. The oxide morphology is consistent with this interpretation, as the scale
consists of coarse columnar grains that penetrate the entire scale. The dominating
outward growth is verified through two-stage oxidation combined with transverse
SIMS, as well as with marker studies. Different physicochemical constants deduced
under these conditions correspond to data from other experimental techniques. Re-
cent investigations indicate that there may be a certain influence from short circuit
transport mechanisms even at high temperatures (temp J 1000 °C). However, these
effects are usually of no or minor importance to the overall oxidation behavior.
The oxidation mechanism of Ni becomes more complex and the interpretations of
the oxidation behavior given in the literature correspondingly more contradictory as
the oxidation temperature decreases. Short circuit mechanisms of both Ni and oxy-
gen are of importance, where the influence of oxygen transport increases with de-
creasing temperature. The mechanism of oxygen ingress is not fully clear, but since
the oxygen diffusivity in NiO is too slow to account for the inward growth, oxygen is
assumed to penetrate the scale as gaseous species.
As evident from this paper, certain aspects of the oxidation mechanism still re-
main unclear. Extensive studies of the development of the grain structure and
characterization of the grain boundary texture as a function of the reaction condi-
tions is essential. This would be of interest not only to resolve the discrepancies
reported in the Ni system, but also to clarify the influence of grain boundary texture
on high-temperature oxidation mechanisms in general. As part of this, one should
address the mechanisms of grain growth during oxidation and the activation energy
corresponding to these processes.
Different aspects of the growth of the inner region also need further attention.
Measurements of the relative thickness of the inner region as a function of the oxy-
gen pressure for different temperatures and a function of time for one given reaction
would be interesting to explore the mechanism for the formation of the inner layer
and possibly the formation of microfissure.

Acknowledgements

The author would like to acknowledge the following coworkers: Dr. G. Hultquist
at The Royal Institute of Technology in Stockholm for the introduction to the Gas
Phase Analysis equipment, Dr. U. S€ odervall at Chalmers Technical University in
Gothenburg for performing the SIMS analysis, O. Dyrlie for running the AFM,
234 R. Haugsrud / Corrosion Science 45 (2003) 211–235

S. Jørgensen for doing the XPS analysis and Prof. T. Norby for discussion during
preparation of the manuscript, the last three persons at Centre for Materials Science,
University of Oslo. Finally this research has been carried out with financial support
from The Research Council of Norway, projects 110899/410 and 134957/432.

References

[1] A.M. Beltran, E.E. Brown, W.L. Chambers, D.S. Chang, W.H. Coust Jr., in: C.T. Sims, N.S. Stoloff,
W.C. Hagel (Eds.), Super Alloys II, John Wiley & Sons, New York, 1987.
[2] P. Kofstad, High Temperature Corrosion, Elsevier Applied Science, London, New York, 1988.
[3] S. Pizzini, R. Morlotti, J. Electrochem. Soc. 114 (1967) 1179.
[4] R. Uno, Phys. Soc. Jpn. 22 (1967) 1502.
[5] I. Branski, N.M. Tallan, J. Phys. Chem. 49 (1968) 1243.
[6] C.M. Osburn, R.W. Vest, J. Phys. Chem. Solids 32 (1971) 1343.
[7] C.M. Osburn, R.W. Vest, J. Phys. Chem. Solids 32 (1971) 1331.
[8] J. Deren, S. Mrowec, J. Mater. Sci. 8 (1973) 545.
[9] R. Fahri, G. Petot-Ervas, J. Phys. Chem. Solids 39 (1978) 1169.
[10] R. Fahri, G. Petot-Ervas, J. Phys. Chem. Solids 39 (1978) 1175.
[11] N.L. Peterson, Solid State Ionics 12 (1984) 201.
[12] S.P. Mitoff, J. Phys. Chem. 35 (1961) 882.
[13] W.C. Tripp, N.M. Tallan, J. Am. Ceram. Soc. 53 (1970) 531.
[14] R. Haugsrud, T. Norby, Solid State Ionics 111 (1998) 323.
[15] H.-G. Sockel, H. Schmalzried, Ber. Bunsenges Physik. Chem. 72 (1968) 745.
[16] Y.D. Tretyakov, R.A. Rapp, Trans. AIME 245 (1969) 1235.
[17] R. Lindner, A . A
kerstr€
om, Disc. Farad. Soc. 23 (1957) 133.
[18] M.T. Shim, W.J. Moore, J. Chem. Phys. 26 (1957) 802.
[19] M.L. Volpe, J. Reddy, J. Chem. Phys. 53 (1970) 1117.
[20] A. Atkinson, R.I. Taylor, Phil. Mag. A 39 (1979) 581.
[21] A. Atkinson, R.I. Taylor, Phil. Mag. A 43 (1981) 979.
[22] E.G. Moya, G. Deyme, F. Moya, Scr. Metall. Mater. 24 (1990) 2447.
[23] K. Fuecki, J.B. Wagner Jr., J. Electrochem. Soc. 112 (1965) 384.
[24] A. Atkinson, R.I. Taylor, P.D. Goode, Oxid. Met. 13 (1979) 519.
[25] M.J. Graham, G.I. Sproule, D. Capland, M. Cohen, J. Electrochem. Soc. 119 (1972) 883.
[26] N.N. Koi, W.W. Smeltzer, J.D. Embury, J. Electrochem. Soc. 122 (1975) 1495.
[27] T. Karakasidis, M. Meyer, Phys. Rev. B, Condens. Matter 55 (1997) 13853.
[28] C. Dubois, C. Monty, J. Philibert, Phil. Mag. A 46 (1982) 419.
[29] D.M. Duffy, P.W. Tasker, Phil. Mag. 54 (1986) 759.
[30] J. Cano-Cabrera, J. Casaing, J. Phys. Lett. 42 (1980) 119.
[31] A. Atkinson, F.C.W. Pummery, C. Monty, in: G. Simkovich, V.S. Stubican (Eds.), Transport in Non-
stoichiometric Compounds, Plenum, NY, 1985.
[32] C. Wagner, K. Gr€ unewald, Z. Physik Chem. B 24 (1938) 455.
[33] L. Berry, J. Paidassi, Compt. Rend. 255 (1962) 2253.
[34] R. Peraldi, D. Monceau, B. Pieraggi, in: M. McNallan, E. Opila, T. Maruyama, T. Narita (Eds.), High
Temperature Corrosion and Materials Chemistry (Electrochem. Soc. Proc., vol. 99–38) , 1999, p. 204.
[35] J.M. Perrow, W.W. Smeltzer, J.D. Embury, Acta Metall. 16 (1968) 1209.
[36] R.A. Rapp, Metall. Trans. 15A (1984) 65.
[37] A. Atkinson, R.I. Taylor, A.E. Hughes, Phil. Mag. 45 (1982) 823.
[38] C.K. Kim, L.W. Hobbs, Oxid. Met. 45 (1996) 247.
[39] G.M. Raynaud, L. Brossard, Proc. Int. Congr. Met. Corros. 1 (1984) 27.
[40] F. Barbier, J. Bernardini, F. Moya, M. Dechamps, Mater. Sci. Res. 21 (1987) 549.
[41] F.H. Stott, Z. Peide, W.A. Grant, R.P.M. Procter, Corros. Sci. 22 (1992) 305.
R. Haugsrud / Corrosion Science 45 (2003) 211–235 235

[42] D. Caplan, M.J. Graham, M. Cohen, J. Electrochem. Soc. 119 (1972) 1205.
[43] M.J. Graham, D. Caplan, M. Cohen, J. Electrochem. Soc. 120 (1972) 1265.
[44] M.J. Graham, R.J. Hussey, M. Cohen, J. Electrochem. Soc. 119 (1973) 1523.
[45] A. Br€uckman, R. Emmerich, S. Mrowec, Oxid. Met. 5 (1972) 137.
[46] F. Maak, C. Wagner, Werkstoffe Korros. 12 (1961) 273.
[47] A. Br€uckman, Corros. Sci. 7 (1967) 51.
[48] A. Atkinson, D.W. Smart, J. Electrochem. Soc. 135 (1988) 2886.
[49] J. Robertson, M.I. Manning, Mater. Sci. Technol. 4 (1988) 1064.
[50] D.P. Moon, Oxid. Met. 31 (1989) 71.
[51] D.P. Moon, A.W. Harris, P.R. Chalker, S. Mountfort, Mater. Sci. Technol. 4 (1988) 1101.
[52] S. Mrowec, Corros. Sci. 7 (1967) 563.
[53] P. Kofstad, Oxid. Met. 24 (1985) 265.
[54] G.J. Yurek, H. Schmaltzried, Ber. Bunsenges. Phys. Chem. 79 (1975) 255.
[55] A.G. Evans, D. Radjev, D.L. Douglass, Oxid. Met. 4 (1972) 151.
[56] G.B. Gibbs, R. Hales, Corros. Sci. 17 (1977) 487.
[57] J.S. Sheasby, D.S. Cox, Oxid. Met. 37 (1992) 373.
[58] A.T. Chadwick, R.I. Taylor, Solid State Ionics 12 (1984) 343.
[59] D.P. Moon, Oxid. Met. 32 (1989) 47.
[60] F. Czerwinski, J.A. Szpunar, W.W. Smeltzer, J. Electrochem. Soc. 143 (1996) 3000.
[61] R. Haugsrud, Corros. Sci. 47 (2002) 1569.
[62] X. Peng, D. Ping, T. Li, W. Wu, J. Electrochem. Soc. 145 (1998) 389.
[63] R. Haugsrud, A.E. Gunnaes, O. Nilsen, Oxid. Met., submitted for publication.
[64] A.A. Moosa, S.J. Rothman, L.J. Nowicki, Oxid. Met. 24 (1985) 115.
[65] A.A. Moosa, S.J. Rothman, Oxid. Met. 24 (1985) 133.
[66] R. Haugsrud, in preparation.
[67] kermark, G. Hultquist, L. Gr
T. A asj€
o, J. Trace, Microprobe Technol. 14 (1996) 377.
[68] A.W. Harris, A. Atkinson, Oxid. Met. 34 (1990) 229.
[69] R. Peraldi, D. Monceau, B. Pieraggi, Mater. Sci. Forum 369–372 (2001) 189.
[70] S.K. Verma, G.M. Raynaud, R.A. Rapp, Chemical Metallurgy––A Tribute to Carl Wagner, in:
N. Gokcen (Ed.), Proceedings of AIME Conference, 1981, p. 419.
[71] J.E. Harris, Acta Metall. 26 (1978) 1033.
[72] H.E. Evans, Mater. Sci. Technol. 4 (1989) 1089.
[73] B. Pieraggi, R.A. Rapp, Acta Metall. 36 (1988) 1281.
[74] A.M. Huntz, C. Liu, M. Kornmeier, J.L. Lebrun, Corros. Sci. 35 (1993) 989.
[75] P.Y. Hou, R.M. Cannon, Mater. Sci. Forum 251–254 (1997) 325.
[76] J.G. Goedjen, D.A. Shores, J.A. Stout, Mater. Sci. Eng. A 222 (1997) 58.
[77] J. Stringer, Corros. Sci. 10 (1970) 513.
[78] A.G. Evans, T.G. Langdon, Progr. Mater. Sci. 21 (1976) 177.
[79] D.L. Douglass, Oxidation of Metals and Alloys, ASM, Metal Park, Ohio, 1971.
[80] H. Kyung, C.K. Kim, Mater. Sci. Eng. B 76 (2000) 173.
[81] R. Herchl, N.N. Khoi, T. Homma, W.W. Smeltzer, Oxid. Met. 4 (1972) 35.
[82] F. Czerwinski, J.A. Szpunar, Corros. Sci. 41 (1999) 729.
[83] A. Atkinson, R.I. Taylor, A.E. Hughes, in: R.A. Rapp (Ed.), High Temperature Corrosion, NACE,
Houston TX, 1983.
[84] W.W. Smeltzer, R.R. Haering, J.S. Kirkaldy, Acta Metall. 9 (1961) 880.
[85] H.V. Atkinson, Oxid. Met. 28 (1987) 353.
[86] A. Atkinson, R.I. Taylor, A.E. Hughes, Phil. Mag. A 45 (1982) 823.
[87] F.N. Rhines, R.G. Connel, J. Electrochem. Soc. 124 (1977) 1122.
[88] Y.J. Iida, J. Am. Ceram. Soc. 41 (1958) 397.
[89] D. Monceau, R. Peraldi, B. Pieraggi, Defect Diffus. Forum 194–199 (2001) 1675.
[90] Z. Adamczyk, J. Nowotny, J. Electrochem. Soc. 127 (1980) 1117.
[91] L.C. Dufour, F. Morin, Oxid. Met. 37 (1993) 137.
[92] F. Morin, L.C. Dufour, G. Trudel, Oxid. Met. 37 (1993) 39.

View publication stats

You might also like