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

net/publication/286545911

Thermodynamics and Theory of External and Internal Oxidation of Alloys

Article · December 2010

DOI: 10.1016/B978-044452787-5.00012-3

CITATIONS READS

12 5,877

1 author:

Brian M Gleeson
University of Pittsburgh
186 PUBLICATIONS   4,453 CITATIONS   

SEE PROFILE

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

Thermal barrier coatings View project

All content following this page was uploaded by Brian M Gleeson on 12 January 2016.

The user has requested enhancement of the downloaded file.

 This article was originally published in Shreir’s Corrosion, published by
Elsevier. The attached copy is provided by Elsevier for the author’s benefit and
for the benefit of the author’s institution. It may be used for non-commercial
research and educational use, including (without limitation) use in instruction
at your institution, distribution to specific colleagues who you know, and
providing a copy to your institution’s administrator.

All other uses, reproduction and distribution, including (without limitation)

commercial reprints, selling or licensing of copies or access, or posting on
open internet sites, personal or institution websites or repositories, are
prohibited. For exception, permission may be sought for such use through
Elsevier’s permissions site at:

http://www.elsevier.com/locate/permissionusematerial

Gleeson B 2010 Thermodynamics and Theory of External and Internal

Oxidation of Alloys. In: Richardson J A et al. (eds.) Shreir’s Corrosion, volume
1, pp. 180-194 Amsterdam: Elsevier.
 Author’s personal copy

1.09 Thermodynamics and Theory of External and Internal

Oxidation of Alloys
B. Gleeson
Department of Mechanical Engineering and Materials Science, The University of Pittsburgh, 647 Benedum Hall, Pittsburgh,
PA 15261, USA

ß 2010 Elsevier B.V. All rights reserved.

1.09.1 Introduction 180

1.09.2 Pure Metal Reactions 181
1.09.2.1 Thermodynamics of a Single-Oxidant Reaction 181
1.09.2.2 Thermodynamics of Dual Oxidant Reactions 184
1.09.2.3 Kinetics of Scale Formation 186
1.09.2.3.1 Parabolic rate law 186
1.09.2.3.2 Linear rate law 187
1.09.2.3.3 Logarithmic rate law 187
1.09.2.4 Transport Properties of Metal Oxides 187
1.09.2.5 Wagner’s Theory of Metal Oxidation 188
1.09.3 Alloy Reactions 190
1.09.3.1 Thermodynamics of Alloy Oxidation 190
1.09.3.2 Criterion for the Sustained Exclusive Growth of a Protective Scale 191
1.09.3.3 Internal Oxidation 192
1.09.3.4 Transition from Internal Oxidation to External Scale Formation 193
1.09.4 Epilogue 193
References 193

1.09.1 Introduction
Symbols
ai Chemical activity of phase or component i
High temperature corrosion plays an important role
Di Diffusion coefficient of component i (m2 s
1)
in the selection of materials in modern indus-
K Equilibrium constant for a given reaction
try. Numerous commercial processes such as electric
kl Linear rate constant (m s
1) or (kg m
2 s
1)
power generating plants, aerospace, gas turbines,
kp Parabolic rate constant (m2 s
1) or (kg2 m
4 s
1)
heat-treating, and mineral and metallurgical proces-
Ni Mole fraction of component i
sing operate at temperatures exceeding 500 C.1 Oxi-
Pi Partial pressure of gaseous species i
dation is often the most important high temperature
R Universal gas constant (8.314 J mol
1 K
1)
corrosion reaction in these commercial processes.
T Temperature (K)
Indeed, most high temperature alloys are designed to
t Time (s)
react with the oxidizing environment in such a way
Vm Molar volume of metal or alloy (m3 mol
1)
that a protective oxide scale forms.2 The degradation
X Thickness of metal consumed due to scaling (m)
resistance of a high temperature alloy depends on
x Scale thickness (m)
sustaining the formation of this protective scale.
DG Gibbs free enthalpy in the standard state
The properties of the scale determine the extent to
(J mol
1)
which protection can be provided. Ideally, the scale
DW Weight change (kg m
2)
should exhibit a slow growth rate, good adherence to
gi Chemical activity coefficient of solute i
the alloy substrate, a high stability, and be continuous
n Stoichiometric factor for the oxide BOn
and free of defects such as microcracks or large

180
Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194
 Author’s personal copy

Thermodynamics and Theory of External and Internal Oxidation of Alloys 181

voids. In general, a chromia, alumina, or silica scale can 1.09.2.1 Thermodynamics of a

meet these requirements for high temperature oxida- Single-Oxidant Reaction
tion resistance, with chromia-forming alloys being the
Consider the high temperature reaction of a metal,
most extensively used in high temperature industrial
M, with an oxidant gas, in this case oxygen. The metal
applications.
initially absorbs oxygen and then chemical reaction
Almost all theoretical treatments of high
ensues to form an oxide. For the metals of relevance
temperature corrosion assume local equilibrium. In
to this review, the resulting oxide is solid. Thus, the
the majority of cases, this proves to be a reasonable
oxide first nucleates and then grows to form a scale on
assumption, although it is usually insufficient for pre-
the metal surface. Depending on its growth kinetics, the
dicting the nature and phase constitution of the reaction
scale may or may not protect the underlying metal.
product. This is because reaction kinetics dictate the
The formation of an oxide may be generally
pathway for scale development. The following discusses
described by the reaction
key fundamental aspects associated with the thermo-
dynamics and kinetics of high temperature corrosion. 2x 2
M þ O 2 ¼ Mx Oy ½1
From a thermodynamic standpoint, it will be shown that y y
relative partial pressures of the gaseous components in Presented in this manner, the high temperature oxida-
the reacting atmosphere are important for predicting tion of metals may seem to be among the simplest of
the composition of the product scale. From a kinetic reactions; however, the reaction path and behavior often
standpoint, it will be shown that limiting equations can involve a number of phenomena and processes that
be established for predicting both scaling kinetics and depend on a variety of factors.3,4 In fact, at the same
critical concentrations in the alloy for transitioning to temperature for a given gas–metal reaction, one can
and sustaining protective scaling behavior. observe drastically different rates of reaction (linear,
parabolic, or any other rate) when the gaseous composi-
tion of the atmosphere is altered and the relative partial
pressures of the gaseous components are different.5
1.09.2 Pure Metal Reactions Under equilibrium conditions, the law of mass
action for reaction [1] gives6
In the context of this discussion, the metal and gas 2=y
react to form a solid surface scale. The prototypical gas aMx Oy
K1 ¼ ½2
will be oxygen; however, it should be realized that the aM2x=y aO2
fundamental treatments equally apply to other gases, where K1 is the temperature-dependent equilibrium
such as sulfur and nitrogen. Moreover, the oxidant may constant and ai is the chemical activity of species i. In
be simple, like O2, S2, and N2, or it may be more complex, most cases, the solids (metal and oxide) are assumed to
like H2O, H2S, and NH4. For the sake of clarity, the be in their pure standard state, so that their activities are
following discussion will consider oxide-scale formation. defined as unity. At relatively high temperatures and
An oxide scale may be a single layer or it may be moderate pressures, the oxidant gas can be treated as
comprised of two or more layers of varying compositions being ideal; that is, the activity of oxygen can be approxi-
that depend on the temperature and oxidizing condi- mated by its partial pressure in atmospheres. Thus, eqn
tions. There are two important factors in discussing the [2] simplifies to
oxidation of metals: thermodynamics and kinetics.
Metallic elements react with oxygen to form oxides if it K1 ¼ 1=PO2 ½3
is energetically feasible. Thermodynamics show whether where PO2 is the oxygen partial pressure.
or not a reaction can take place. When the oxidation Thermodynamically, reaction [1] for any metal
reaction is possible, kinetics shows how fast the reaction can take place spontaneously from left to right
will be. In practical applications, kinetics is of more when its overall Gibbs free energy change, DG, is
importance because it determines the extent of metal negative. For reaction [1], the Gibbs free energy
consumption and the overall reaction pathway (i.e., change under isobaric conditions is given as:
assemblage and structure of the reaction product). Oxi-
DG ¼ DG  þ RT lnK1 ½4
dation theory of pure metals provides the foundation

for understanding the more complicated processes where DG is the standard Gibbs free energy
associated with alloy oxidation. of formation of the oxide at absolute temperature

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

182 Principles of High Temperature Corrosion

T and R is the gas constant. If DG ¼ 0, the system is at usually called an Ellingham diagram,6 summarizes the
equilibrium, and if DG > 0, the reaction is thermody- temperature dependence of DG  for various common
namically unfavorable. At equilibrium (DG ¼ 0), oxidation reactions at a standard state (pO2 ¼ 1 atm).
Such a diagram is shown in Figure 1 for oxides.
DG  ¼
RT ln K1 ¼ RT ln PO2 ½5
The Ellingham diagram in Figure 1 shows the
Thus, knowing that at equilibrium both the forward relative thermodynamic stability of the indicated oxi-
and reverse reaction rates are equal, the dissociation des. The lower the line on the diagram, the more
pressure of the oxide can be defined from [5] as: negative the standard free energy of formation and,

 hence, the more stable the oxide. For example, the
DG
PO2 ¼ exp
diss
½6 lines for Al2O3, SiO2, and Cr2O3 are lower than those
RT for FeO, NiO, and CoO in the Ellingham diagram, so
The metal M can only be oxidized to the oxide MxOy the former oxides are more stable and therefore meet
at the temperature T if the ambient partial pressure an important criterion for being protective scales. It is
of oxygen is larger than the dissociation pressure further seen in Figure 1 that the DG  versus T lines for
defined by eqn [6]. most of the oxides are parallel and positively sloped.
For the usual conditions of constant temperature This is a consequence of the fact that the entropy of a
and pressure, the auxiliary function DG  is usually gas is much larger than that of a solid. Thus, for the
described by the simple relation metal oxidation reactions represented by eqn [1],
DG  ¼ DH 
T DS  ½7 2  2x 
DS  ¼ SM
SM
SO 2 
SO 2 : ½8
 
y x Oy
y
where DH is the enthalpy of reaction and DS is the 
entropy change under standard-state conditions. Tabu- It follows from eqn [7] that qDG =qT ¼
DS   SO 2 ,
lated values of DH  and DS  for the determination of which is greater than zero.
DG  at any given temperature are readily available.7–9 The PO2 can be read directly from Figure 1 by
A small selection of useful values is given in Table 1. using the PO2 scale along the bottom and the right
Equation [7] shows that a plot of DG  versus T side of the diagram. A straight line drawn from the
gives a straight line. This line would change in slope index point labeled ‘O’ (at DG  ¼ 0, T ¼ 0 K) at the
when a new phase forms (i.e., at melting or boiling upper left of the diagram, through a specific temper-
temperature). A Gibbs energy–temperature diagram, ature point on an oxide line, intersects the PO2 scale
at the dissociation oxygen partial pressure (POdiss
2
) for
Table 1 Standard energies of reaction7,8 for pure solid
that oxide at that particular temperature. Accordingly,
metal (except liquid Al) and various common gasses the oxides lower on the diagram are more stable and
consequently have lower POdiss values. For instance,
DG = DH – TDS
2
Reaction from Figure 1, it is found that the dissociation pres-
(J mol
1)
sure for NiO is 10
10 atm at 1000 C, while that for
DH
DS Cr2O3 is 10
22 atm, and for SiO2 and Al2O3, it is
(J mol
1) (J mol
1 K) 10
26 and 10
34 atm, respectively. The significance of
2
þ 12 O2 ¼ 13 Al2 O2
565 900 this is that it is difficult thermodynamically to pre-
3 AlðlÞ 128
Co þ 12 O2 ¼ CoO
233 886 70.7 clude the oxidation of Cr, Si, and Al.
3CoO þ 12 O2 ¼ Co2 O4
183 260 148.1 Similar free-energy diagrams, which can be inter-
3 Cr þ 2 O2 ¼ 3 Cr2 O2
373 420
2 1 1
86 preted in exactly the same way, have been constructed
Fe þ 12 O2 ¼ FeO
264 890 65.4
3FeO þ 12 O2 ¼ Fe2 O4
312 210 125.1
for sulfides, carbides, and nitrides.5 Moreover, Lou
2Fe2 O4 þ 12 O2 ¼ 3Fe2 O2
249 450 140.7 et al.10 presented an excellent review on the use of
Mn þ 12 O2 ¼ MnO
412 304 72.8 Ellingham diagrams for treating gas–solid reactions.
Ni þ 12 O2 ¼ NiO
234 345 84.3 Low oxygen partial pressures are in practice
2 Si þ 2 O2 ¼ 2 SiO2
451 040
1 1 1
86.8 achieved using oxygen-bearing gas mixtures. The
H2 þ 12 O2 ¼ H2 O
246 440 54.8
CO þ 12 O2 ¼ CO2
282 420 86.8
most common gas mixtures are H2O/H2 and CO2/
O2 þ 12 S2 ¼ SO2
362 420 72.4 CO. The partial pressures of oxygen are then estab-
NiO þ Cr2 O2 ¼ NiCr2 O4
1 376 880 332 lished from the equilibria:
NiO þ Al2 O2 ¼ NiAl2 O4
1 933 667 408
1=2
1 PH2 PO2
The calculated DG value would be for the mole numbers shown in H2 O ¼ H2 þ O2 ; K9 ¼ ½9
the particular reaction considered. 2 PH2 O

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

Thermodynamics and Theory of External and Internal Oxidation of Alloys 183

H2/H2O ratio 10–8 10–6 10–4 pO2

CO/CO2 ratio 10–8 10–6 10–4 10–2
Temperature (F) 2
10
390 750 1110 1470 1830 2190 2550 2910 3270 3630 3990 4350
0 0 1
O3
Fe 2
2
=6 1
–100 +O M
CO 2 CO 2
e 3O
4
M M =2 =2 10–2
4F + O2 + O2
O O
2C 2C O 1
–200 u 2O iO = 2H 2
= 2C 2N H + O2
+ O2 = M 2 2
4Cu i+ O2 10–4
2N
–300 102

C + O2 = CO2 10–6
2
–400 10
ΔG = RT ln po2 (kJ mol–1 O2)

M
O 10–8
Zn B 2C
H –500 =2 O3 +O
+O 2 2 /3Cr 2 2 =2 104
n = CO
2Z + O2
C 4 /3Cr
M M B 10–10
–600
104
iO 2
=S
i+O
2
–700 S 10–12
106

–800 l 2O 3
/A
2 3
= 10–14
+ O 2
106
4 3 / Al
–900 B
O
Mg
O Ca B 10–16
M
=2 =2 108
+O
2 a +O2
–1000 g 2C
2M
M Change of state Element Oxide
M 10–18
–1100 Melting point M M 108

Boiling point B B
1010
–1200 10–20
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Temperature (C)
CO/CO2 ratio 1014 1012 10–22
1010
OK H2/H2O ratio
po2 (atm)
10–200 10–100 10–70 10–60 10–50 10–42 10–38 10–34 10–30 10–28 10–26 10–24
Figure 1 Ellingham diagram showing the standard Gibbs energies of formation of selected oxides as a function of
temperature. Reproduced from Gaskall, D. R. Introduction to the Thermodynamics of Materials, 5th ed.; Taylor & Francis:
New York, 2008.

and these equilibrium constants if the equilibrium PCO/

1=2 PCO2 or PH2/PH2O ratios are known. In oxygen-lean
1 PCO PO2
CO2 ¼ CO þ O2 ; K10 ¼ ½10 gases containing both H2O and CO2, the PO2 is usu-
2 PCO2 ally determined by the H2–H2O reaction [9] since
where K9 and K10 are the temperature-dependent steam is more reactive than CO2. Moreover, for a
equilibrium constants for reactions [9] and [10], controlled laboratory experiment, it is preferred
respectively. Thus, the PO2 may be determined from practice to facilitate the equilibrium by using a

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

184 Principles of High Temperature Corrosion

platinum-containing catalyst in the reaction zone of of which type of corrosion should predominate. In
the furnace. such cases, it may be necessary to conduct field-
In addition to the direct oxidation with oxygen, exposure tests to properly evaluate the corrosion
the overall metal oxidation reactions in these gas behavior of the candidate alloys. Even so, the assump-
mixtures are: tion of thermodynamic equilibrium provides a rea-
sonable starting point for assessment.
xM þ yH2 O ¼ Mx Oy þ yH2 ½11a
Quite often, two-dimensional phase stability dia-
and grams are used to assess the high temperature corrosion
of a metal M exposed to a dual-oxidant atmosphere.
xM þ yCO2 ¼ Mx Oy þ yCO ½11b
To illustrate this, the possible reactions of M exposed to
From Figure 1, the H2/H2O and CO/CO2 ratios oxidizing–sulfidizing atmosphere are determined by
can be obtained by using the same method as considering the following:
for determining POdiss
2
, using the index points labeled 1
‘H’ and ‘C’ at the left of the diagram instead of M þ O2 $ MO ½13
2
point ‘O.’ 1
M þ S2 $ MS ½14
The growth of a scale is usually sufficiently slow 2
1 1
for local equilibrium to be closely approached. In MO þ S2 $ MS þ O2 ½15
fact, it is found that the Gibbs phase rule6 can be 2 2
applied to rationalizing the phase assemblage of a The equilibrium from reactions [13] and [14] defines
growing scale. Specifically, the Gibbs phase rule the critical PO 2 and PS2 values (i.e., dissociation pres-
under isothermal and isobaric conditions is given as sures) for M/MO and M/MS equilibrium, respectively,
while reaction [15] defines critical PS2/PO2 ratios for
PþF ¼C ½12 MS/MO equilibria. With regard to the latter, it can be
where P is the number of phases at a given location in easily shown for reaction [15] that log(PS2) ¼ log(PO2) –
the scale, F represents the degrees of freedom, and 2logK15, where K15 is the equilibrium constant for this
C number of components in the system. For the reaction. Thus, a plot of log(PS2) versus log(PO2) should
oxidation of a pure metal, C ¼ 2 (i.e., metal and oxy- have a line of slope equal to the unity that separates MS
gen), and for diffusion to occur there must be a stability from MO stability. Figure 2 shows the general
gradient in the chemical activity across the scale, so construction of an M–S–O phase-stability diagram. It is
that F ¼ 1. Thus, P ¼ 1, meaning that only one phase noted that a more complete diagram may include
can exist at a given plane (parallel to the metal/scale higher-order oxides and sulfides, as well as sulfates
interface) within the scale. In practical terms, this (MSO4).11 Only the simpler diagram is discussed here
means that a metal capable of forming multiple oxi- since the main important points can still be made.
des must form those oxides as distinct scale layers. The diagram in Figure 1 shows the stability range
For example, at 1 atm O2 and temperatures above of a metal and its oxide and sulfide products as a
560  C, pure iron oxidizes to form a triple-layered function of the two principal reactants: oxygen and
scale in the sequence FejFeOj Fe3O4 jFe2O3 jO2, with
the oxide position being dictated by the necessity for
the oxygen content to progressively increase when
traversing from the iron to the oxygen at the scale MS
surface. A similar line of reasoning can be used to
show that a pure metal oxidized isothermally will not
2
log pS

form internal oxide precipitates, but instead must

form an external product. MO

M
1.09.2.2 Thermodynamics of Dual Oxidant
Reactions
The thermodynamic aspects of multioxidant corro-
log pO2
sion have been discussed by Giggins and Pettit.11
Often, however, the complexity of a given process Figure 2 Schematic representation of a simple
environment precludes an accurate determination phase-stability diagram for a metal and its oxide and sulfide.

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

Thermodynamics and Theory of External and Internal Oxidation of Alloys 185

sulfur. A given atmosphere is defined by either equi- type of scale formed on 310 under different PO2–PS2
librium or nonequilibrium PO2 and PS2 values and combinations. It is seen that sulfide-to-oxide transi-
would therefore be represented by a point in the tion at a given PS2 actually occurs at a higher PO2 than
stability diagram. The location of that point identifies that predicted from equilibrium calculations. The
the phase that is in stable equilibrium with that par- experimentally-determined boundary is dictated by
ticular atmosphere. However, as shown in Figure 3, kinetic factors and, accordingly, is referred to as the
other phases can form even if only oxide stability kinetic boundary. For the case of the 310 stainless
is predicted. This depends on the reacting gas, the steel shown in Figure 4, the actual PO2 values for the
PO2–PS2 combination, and whether the scale devel- transition from chromium–sulfide to chromium–
ops open pathways for gaseous penetration (e.g., oxide formation are about three orders of magnitude
microcracks). Internal sulfidation is particularly higher than the equilibrium values. The kinetic fac-
apt to occur in SO2-containing atmospheres due tors which influence the location of the kinetic
to the local equilibrium dictated by the reaction boundary include composition and surface finish of
O2 þ ‰ S2 ! SO2, which has an equilibrium constant the alloy, and gas composition. Although theoretical
denoted as K3. As indicated in Figure 3, the formation prediction of the location of a kinetic boundary is
of an internal sulfide beneath an external oxide scale not possible, LaBranche and Yurek13 showed that
can only be completely avoided if for H2–H2O–H2S gas mixtures there is a critical
qffiffiffiffiffiffiffi H2O/H2S ratio associated with the kinetic boundary.
PSO2 <K3 PO 2 PS2 ½16 The value of this ratio is dictated by the competitive
formation of the oxide and the faster-growing sulfide.
Although phase-stability diagrams are very useful for It was found by LaBranche and Yurek that oxide
interpreting reaction products and gaining insights formation on pure chromium at 900 C could only
into reaction pathways, they do not have any predic- occur when the area fraction of Cr2O3 was greater
tive capabilities from a practical standpoint. than 0.9 in the early stages of exposure, which
On the basis of equilibrium thermodynamics of corresponds to H2O/H2S > 10.
reaction [15], the transition from sulfidation to oxi-
dation of metal M should occur when

1=2
PO2
> K15 ½17
P S2
The thermodynamic boundary separating sulfide
and oxide stability in a phase stability diagram is 0
determined by replacing ‘>’ with ‘=’ in eqn [17].
Figure 4 shows the phase-stability diagram for the
FeS
Type 310 stainless steel at 875  C.12 Superimposed on Kinetic
–2
this diagram are experimental data indicating the boundary
log pS2 (Pa)

Cr3S4
min.
pSO = K3⋅pO
* ⋅ pS* –4
2 2 2
1 Cr2O3
SO2 or S2
MS CrS
1 MS
Sulfide-containing
log pS2

scale
MO 2 Alloy Pure oxide
SO2

M 2 MS
3 –20 –15 –10
3 log pO2 (Pa)
log pO2 MO Figure 4 Thermodynamic phase-stability diagram for
type 310 stainless steel at 875 C, showing the
M
experimentally-determined kinetic boundary. Reproduced
Figure 3 Application of the phase-stability diagram in from Stroosnijder, M. F.; Quadakkers, W. J. High Temp.
identifying possible modes of attack. Technol. 1986, 4, 141.

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

186 Principles of High Temperature Corrosion

1.09.2.3 Kinetics of Scale Formation thickness, x, increases with time, t, and, since this
corresponds to the increasing diffusion distance,
The reaction of a metal with an oxidant to form a
the oxidation rate decreases. Thus, the instantaneous
product scale is concomitant with a weight gain due
oxidation rate is quite simply inversely proportional
to pick-up of the oxidant (e.g., grams of oxygen
to the oxide thickness, that is,
gained). Accordingly, the scaling rate may be quanti-
0
fied by the change in the scale thickness, x (cm), or by dx kp
weight gain, DW (mg cm
2). In the case of oxidation, ¼ ½20
dt x
these two parameters are directly related by 0
where kp is a proportionality constant. Integration of
Vox eqn [20] gives
x¼ DW ½18
yMO x 2 ¼ kp t þ C ½21

1
where Vox is the molar volume of oxide in cm mol ,
3
where kp is taken to be the parabolic rate cons-
y is the stoichiometric amount of oxygen in oxide 0
tant (kp ¼ 2 kp) with typical units of cm2 s
1. Another
MxOy , and Mo is the atomic weight of oxygen. form of the parabolic rate equation is given by the
The formation of oxide scale is also related to the weight gain (g cm
2):
consumption of metal, and the relationship between
DW and the thickness of the metal consumed (X) is DW 2 ¼ kp t þ C ½22
given by where the units of kp in this case are g2 cm
4 s.
Vm The parabolic rate law is the standard for ana-
X¼ DW ½19
gMO lysis of high temperature oxidation kinetics, in which
diffusion through the relatively thick scale con-
where Vm is the molar volume of metal in cm3 mol
1
trols reaction rates. The diffusion-controlled scale-
and g is the stoichiometric factor for the oxide scale
thickening process is thermally activated, meaning
product (i.e., g ¼ y/x for MxOy). For most metals, the
that the rate increases exponentially with temperature,
oxidation rates follow one or more of the three possi-
as given by the Arrhenius equation:
ble kinetic laws: linear, logarithmic, and parabolic.

Q 
These kinetic laws are discussed in the following.
kp ¼ ko exp ½23
Formation of an oxide scale will separate the two RT
reactants, metal and gaseous oxygen. In order for the Here, ko is a constant that is a function of the oxide
reaction to proceed further, at least one of the reac- composition and the gas pressure, and Q is the activa-
tants must progress through the scale to form more tion energy for oxide-scale growth. Figure 51 shows
oxide at the oxide/gas, oxide/metal, or both inter- the temperature dependence of the kp values for the
faces. The mechanisms by which the reactants prog- oxides of Fe, Co, Ni, Cr, Al, and Si. The figure shows a
ress through the scale can therefore be an important range of kp values for Al2O3 and Cr2O3 growth because
part of the overall mechanism and kinetics by which these oxides do not show intrinsic behavior and are
high temperature oxidation reaction proceeds. Another instead very sensitive to impurities (i.e., doping).
aspect of the oxidation process, which can sometimes Deviations from parabolic kinetics are generally
be rate controlling, is the kinetics of the interfacial analyzed in terms of chemical and metallurgical effects
reaction steps. on the rates of the relevant diffusing process(es). The
parabolic rate law will not hold in the very early stages
1.09.2.3.1 Parabolic rate law of oxidation, before the scale has developed sufficient
At high temperature, initial scale growth is usually continuity and thickness. In the case of hafnium and
very rapid; however, the reaction rate will eventually zirconium, the time exponent in eqn [21] is found to be
decrease when scale thickness reaches 0.5 mm and 3 (cubic kinetics) rather than 2 (parabolic kinetics) at
the transport of reacting species through the scale high temperatures. This has been attributed to the
becomes rate controlling. When the rate-controlling simultaneous dissolution of oxygen into the substrate
step in the oxidation process is the diffusion of reac- metal during oxidation.14 It has also been shown for
tant(s) through the oxide layer and with the boundary NiO-15 and Al2O3-scale16 growth that subparabolic
conditions for diffusion being time independent, the kinetics can occur if both short-circuit diffusion
scaling kinetics will follow the parabolic rate law. through the scale predominates and the average grain
Parabolic kinetics results from the fact that the scale size of the scale increases with oxidation time.

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

Thermodynamics and Theory of External and Internal Oxidation of Alloys 187

T (C) microcracking and porosity may develop as the scale

1200 1100 1000 900 800 700 600 thickens, reducing the protectiveness of the oxide.
10–5 The parabolic rate law may then fail, and the kinetics
approaches linearity at some time after the start of
Parabolic scaling constant, k (g2 cm–4 s–1)

10–7 FeO/Fe3O4/Fe2O3 reaction and scale growth. That is, a constant oxida-
tion rate can develop after a period of parabolic
behavior. In this special case of parabolic scaling
10–9 superimposed on a relatively constant rate of scale
cracking and healing, the overall kinetics is said to
CoO
be ‘paralinear.’20
10–11
NiO
Cr2O3
1.09.2.3.3 Logarithmic rate law
At low temperatures (e.g., T < 300–400  C), oxidation
10–13 SiO2 on Al2O3 rates are often inversely proportional to time, that is,
MoSi2
dx k
¼ ½25
10–15 dt t
where k is a constant. Integration of [25] leads to the
10–17 logarithmic rate law
6 7 8 9 10 11 12 13
104/T (K–1) x ¼ ka logðkb t þ 1Þ ½26
Figure 5 Parabolic oxidation rate constant for various where ka and kb are constants. Logarithmic oxidation is
oxide scales as a function of temperature. Reproduced from usually obeyed for relatively thin scales at low
Gleeson, B. In Corrosion and Environmental Degradation, temperatures.
Vol. II: Volume 19 of the Materials Science and Technology
Series; Schütze, M., Ed.; Wiley-VCH: Weinheim,
Germany, 2000. 1.09.2.4 Transport Properties of Metal
Oxides
1.09.2.3.2 Linear rate law Metallic oxides are seldom stoichiometric, meaning
Under certain conditions, the oxidation of a metal that the metal-to-oxygen atom ratio is not exactly
proceeds at a constant rate according to a linear rate that given by the stoichiometric chemical formula,
law, that is, even though the compound is electrically neutral.
The same can be stated for sulfides and nitrides,
x ¼ kl t ½24
but the focus here will be on oxides as the prototypi-
where x is the scale thickness and kl is the linear rate cal systems. The ionic charge imbalance in a non-
constant. A linear rate law may result when a phase- stoichiometric oxide is compensated by electronic
boundary reaction controls the kinetics rather than a charges, that is, electron and electron holes. As a
transport process. An example is CO2 dissociation at consequence, nonstoichiometric oxides exhibit both
the scale surface controlling the oxidation kinetics of electronic and ionic conductivities. These conductiv-
steel in a CO2-rich atmosphere.17 ities are temperature dependent, which, from the
Linear kinetics is also possible if the oxide is electronic standpoint, classifies the nonstoichio-
volatile or molten, if the scale spalls or cracks, or if metric oxides as semiconductors.
a porous, nonprotective oxide forms on the metal.4 Electronic semiconductors are categorized as
Since the rate of oxidation never slows down, con- n-type (excess of electrons) or p-type (excess of elec-
sumption of the metal occurs in a relatively short tron holes). The n-type oxides may have either an
time in comparison to a metal scaling according to excess of cation interstitials or an excess of oxygen
parabolic kinetics. Examples of metals that scale in a vacancies (i.e., a deficiency in filled sites on the oxy-
nonprotective, linear manner due to continual scale gen sublattice, MO1
x) as the principal ionic defects.
cracking are Nb and Ta.18 Some examples of n-type oxides are TiO2, Fe2O3,
In the early stages of a metal oxidation process, the NiFe2O4, ZnO, Al2O3, and SiO2.18 For n-type oxides,
scale may be sufficiently thin that linear oxidation the principal ionic defect concentration, C, is found
kinetics prevails. As the scale thickens, a transition to to be proportional to the negative power of the oxy-
parabolic kinetics will usually ensue.19 Conversely, gen partial pressure21:

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

188 Principles of High Temperature Corrosion

C / PO2

1=n
½27  The addition of substitutionally dissolved foreign
cations of lower valence increases the concentra-
where n is a positive integer which depends upon the tion of electron holes and decreases the concentra-
charge of the ionic defect in the oxide and the oxide’s tion of cation vacancies.
stoichiometry.  The addition of substitutionally dissolved foreign
The p-type oxides may have either an excess of cations of higher valence decreases the concentra-
cation vacancies (i.e., a deficiency in filled sites on the tion of electron holes and increases the con-
metal sublattice, M1
xO) or an excess of oxygen centration of cation vacancies.
interstitials. Some common p-type oxides include
NiO, CoO, FeO, FeCr2O4, CoCr2O4, and NiAl2O4. Thus, for the p-type, metal-deficient NiO, doping
For p-type oxides, the principal ionic defect concen- with Cr3+, will cause an increase in the nickel vacancy
tration is related to the oxygen partial pressure by21: concentration, which in turn will cause an increase
in the effective diffusivity of nickel in the NiO. This
1=n
C / PO2 ½28 increase in diffusivity is manifested as an increase
in the kp for NiO-scale growth. Indeed, it is found
During the oxidation process, ions transport by two
experimentally that NiO grows faster on dilute
ways: the cation migrates outward to scale/gas inter-
Ni–Cr alloys than on pure Ni.22
face and the anion migrates inward to metal/scale
interface (Figure 6). Thus, two reactions could
1.09.2.5 Wagner’s Theory of Metal Oxidation
potentially happen at the two interfaces to result in
scale growth, although it is typically found that one Carl Wagner’s theory of metal oxidation23,24 provides
reaction predominates. In other words, either metal a fundamental understanding of the essential features
or oxygen diffusion predominates in the thickening of of the high temperature growth of a continuous scale.
a given oxide scale. The model ideally assumes that the scale is dense,
For the defective structures treated so far, it has single-phase, continuous, and adheres to the metal over
been assumed that the oxides are pure; however, it is the entire metal surface. The basic assumption of the
thermodynamically impossible to produce perfectly theory is that lattice diffusion of the reacting atoms or
pure compounds or materials. It is therefore neces- ions through the dense scale is rate controlling (i.e.,
sary to consider the effects of impurities on the the microstructure of the oxide scale was not consid-
defective structure of oxides. It is noted, however, ered). Figure 7 gives a set-up of the model.14
that the presence of impurities can sometimes be As indicated in Figure 7, the scale growth involves
neglected if the intrinsic defect concentration in the fluxes of both ionic and electronic charged species.
host oxide is relatively large. As an example of the The driving forces for these fluxes are related to
so-called ‘doping effect,’21 the following is found for a the chemical potential gradient and electrostatic
p-type, metal-deficient oxide M1
xO: field that develop in the growing scale. The relative

M MO O2 M MO O2

M2+ O2–

2e– 2h+

M = M2+ + 2e– M2+ + 2e– + ½O2 = MO M + O2– + 2h+ = MO ½O2 = O2– + 2h+

Figure 6 Ionic and electronic transport processes and interfacial reactions in the growth of a nonstoichiometric oxide
scale MO.

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

Thermodynamics and Theory of External and Internal Oxidation of Alloys 189

Metal M Oxide MO Gas O2

Cations

Anions

Electrons

Electron holes

aM = 1 PO2


ΔGMO
 = exp
aM
ΔG  2RT
pO 2 = exp MO
RT ( pO2) 1/2

M = M2+ + 2e– or M2+ + 2e– + ½O2 = MO or

M + O2– = MO + 2e– ½O2 + 2e– = O2–

Figure 7 Transport processes according to Wagner’s theory. Adopted from Birks, N.; Meier, G. H. Introduction to High
Temperature Oxidation of Metals; Edward Arnold: London, 1983.

migration rates of cations, anions, electrons, and elec- Because the diffusion flux within the oxide is
tron holes must also be balanced such that no net proportional to the defect concentration, eqn [28]
charge build up occurs within the scale. can be used for a p-type oxide to arrive at the follow-
Wagner derived an expression for the parabolic ing relation for the oxidation rate:
rate constant in terms of the electronic and ionic 0
conductivities of the oxide or, alternatively, in terms k / ½ðPOo 2 Þ1=n
ðPOi 2 Þ1=n  ½31a
of the self-diffusion coefficients of the reacting ions in
where POo 2 and POi 2 are the oxygen pressures at the
which parameters can be measured relatively easily.
scale/gas interface and metal/scale interface respec-
Limiting cases of Wagner’s derivation are as follows14:
tively, and n is an integer related to the cation vacancy
mð0M
0 1 or oxygen interstitial charge. For an n-type oxide, use
k ¼ DM dmM ½29 of eqn [27] gives:
RT
m00M
k / ½ðPOi 2 Þ
1=n
ðPOo 2 Þ
1=n 
0
½32a
and
mð0X where in this case n is related to the oxygen vacancy
1
0
k ¼ DX dmX ½30 or cation interstitial charge. In most cases, the ambi-
RT ent oxygen pressure POo 2 is much greater than POi 2 ,
m00X
which is the dissociation pressure (POdiss
2
). Thus, eqns
where k 0 is the parabolic rate constant with units of [31a] and [32a] can be approximated to give
cm2s
1, DM and DX are the self-diffusion coefficients for 0
metal, M, and nonmetal, X, through the scale, respec- k / ðPOo 2 Þ1=n for a p-type scale ½31b
tively, and mM and mX are the chemical potentials for the
metal and nonmetal. Equation [29] is valid when cation and
diffusion predominates and eqn [30] is valid when anion k / ðPOi 2 Þ
1=n for a n-type scale
0
½32b
diffusion predominates. The good agreement between
parabolic rate constants calculated from diffusivities and These equations show that the growth rate of a
the oxidation rate constants measured experimentally p-type oxide is directly dependent on the oxygen
have provided validation for Wagner’s theory.25,26 partial pressure in the atmosphere. By contrast, the

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

190 Principles of High Temperature Corrosion

growth rate of an n-type oxide is independent of the 1.09.3 Alloy Reactions

external oxygen partial pressure.
Oxides formed in practice are usually more com- The fundamentals of pure-metal oxidation provide a
plex than what was assumed by Wagner. First, a multi- basic understanding of alloy oxidation, but the latter
layer scale can form on a number of metals, such as is generally much more complex as a result of some,
Fe, Cu, and Co. For instance, and as discussed earlier, a or all, of the following14:
three-layered FeO/Fe3O4/Fe2O3 scale forms on iron
above 570  C at atmospheric pressure. Treatments (a) The various metal components in an alloy will
of the growth kinetics of multilayered scales have been have different affinities for oxygen, as indicated
presented by a number of authors.27–30 In general, these in the Ellingham diagram in Figure 1.
treatments relate overall oxidation scaling kinetics to (b) More oxides may be formed, including ternary
the intrinsic growth rates of the individual layers in the and higher oxides.
scale. Second, scales forming on metal surfaces are (c) A degree of solid solubility may exist between
usually polycrystalline in structure, and in many cases, the oxides.
these structures are fine-grained. The activation energy (d) The various metal ions will have different mobi-
for grain-boundary diffusion is lower than that for lattice lities in the oxide phases.
diffusion by up to a factor of three. As a consequence, (e) The various metals will have different diffusiv-
grain-boundary diffusion will tend to predominate at ities in the alloy.
lower temperatures and can cause the scaling kinetics to (f) The extra degree of thermodynamic freedom
be orders of magnitude higher than what would be provided by an additional metal component in
predicted based on lattice diffusion.26 the alloy may result in subsurface oxide precipi-
As indicated above, Wagner’s theory considered tation (i.e., internal oxidation).
the ideal case of scale formation in which the scale is When it is considered that scales can crack, contain
assumed to be a compact and perfectly adherent voids, spall, and even form multiple layers of irregu-
barrier, that is, free of voids, pores, and fissures. The lar thickness, the situation becomes even more com-
assumptions made by Wagner were indeed appropri- plex. The following provides a brief overview of the
ate; however, deviations from Wagner’s theory have fundamental aspects of alloy oxidation.
been shown.31 Many studies have shown that pores
form and develop preferentially along grain bound-
aries in the scale.32,33 The presence of voids in scales 1.09.3.1 Thermodynamics of Alloy Oxidation
clearly represents a deviation from ideal scale growth. As in the case of a pure metal, alloy oxidation is
In some extreme cases, it may be necessary to take driven by an overall decrease in free energy. How-
into account the void volume fraction in Wagner’s ever, the alloy components are not in their pure state,
model; but often the void formation may be consid- which corresponds thermodynamically to their chem-
ered to be a secondary process that does not signifi- ical activities being less than unity. For dilute solution
cantly affect the overall scaling kinetics. of metallic component B in A, the average solvent
As the scale grows, stresses develop due to differ- A atom exists essentially in the same chemical sur-
ences in the molar volume of metal and oxide(s). The rounding environment as in its pure state, with only a
resulting growth stresses are typically compres- small number of neighboring solute B atoms. Thus, in
sive.34,35 More significant stresses can develop under the limit of a very dilute solution, the A atoms follow
thermal cycling conditions due to a mismatch in the Raoult’s law,6 such that:
coefficient of thermal expansion (CTE) between
oxide scale and metal. Increasing stresses may even- lim aA ¼ NA ½33
NA !1
tually result in crack formation and even scale
detachment.35 Cracking of a protective oxide scale where aA and NA are the chemical activity and mole
can result in the parabolic oxidation kinetics being fraction of solvent A, respectively. Also, according to
interrupted by a sudden increase in rate when the gas Henry’s law, the chemical activity of solute B is given as:
can react directly with the metal surface. As oxide lim aB ¼ gB NB ½34
NB !0
begins to cover the metal surface again, parabolic
oxidation is resumed. The overall oxidation of the where gB is the activity coefficient of solute B for the
metal becomes approximately a paralinear process of A–B system. It can be easily shown that the dissociation
periodic cracking and healing of a protective oxide. PO2 for an AO scale in contact with an A–B alloy is

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

Thermodynamics and Theory of External and Internal Oxidation of Alloys 191

greater than what would be calculated for AO in con- remainder of this chapter in order to provide more
tact with pure A. Specifically, general kinetic-based expressions.) An example of

2 this could be the preoxidation of an alloy at a low
PO2 1
¼ ½35 PO2 followed by in-service exposure to a higher PO2
PO 2 aA
and/or different temperature. The sustained growth
where aA is the activity of A in the A–B alloy, and PO2 of a continuous BOu scale requires a sufficient supply
and PO 2 are the dissociation pressures for the A–B alloy of B from within the alloy to the alloy/scale interface.
and pure A, respectively. The resulting subsurface concentration gradient of
It is typically the activity of the solute that is solute B is represented schematically in Figure 8, in
most important when assessing the thermodynamics which NBi is the mole fraction of B in the alloy at the
of alloy oxidation; although, the results are often alloy/scale interface. The maximum possible rate of
of limited utility. This can be shown as follows. B supply in the alloy can be imposed by setting NBi
Consider the oxidation of an A–B alloy in which the equal to zero, thereby producing the steepest possible
solute B forms a more stable and protective oxide. diffusion gradient of B. Under steady-state conditions,
Assuming that the only stable oxides are AO and BO this maximum rate of B supply would have to equal
and that these oxides are insoluble in one another, the the rate of B consumption due to BOu scale growth.
important reaction to consider is Determination of the minimum B content in the alloy,
NBðminÞ
o
, necessary for the sustained exclusive growth
AO þ B ! BO þ A ½36
of a BOu scale on an A–B alloy was originally consid-
where B and A correspond to B and A in the ered by Wagner,38 who further assumed that: (1) the
alloy solution. The law of mass action for this reaction diffusion coefficient of B in the alloy, DB, is indepen-
gives dent of concentration; (2) the BOu scale obeys para-
aA NA bolic thickening kinetics with a rate constant kp;
K36 ¼  ½37 (3) solvent metal A is insoluble in BOu; and (4) the
aB gB NB
recession of the alloy/scale interface may be neglected.
Thus, reaction [36] will only proceed to the right Wagner derived the criterion
if aaAB < K36 , which means that there is a critical aB,
1
or conversely NB, for this to occur. In most practical Vm pkp 2
NBðminÞ >
o
½38
cases of interest, the solute–metal oxide BO is signifi- nMO 2DB
cantly more stable than the base-metal oxide AO, where Vm is the equivalent molar volume of the alloy
causing K36 to be much larger than unity. This, in and MO is the atomic weight of the oxidant, which is
turn, results in a very small minimum NB value for oxygen in this discussion (M ¼ 16 g mol
1). Many
BO to be thermodynamically stable in contact with researchers have used the criterion in eqn [38]
the A–B alloy. In the case of an Ni–Cr alloy, NCr must to predict the minimum content of B necessary for a
be greater than 10
9 at 1000  C for the displace- bare alloy to form an exclusive BOu scale layer. However,
ment reaction 3NiO þ 2Cr ! Cr2O3 þ 3Ni to be this is not a correct use of the criterion, since its deriva-
thermodynamically stable.36 Such a low Cr content is tion was based on supply rather than establishment. The
orders of magnitude below what is necessary kineti-
cally to sustain Ni–Cr/Cr2O3 stability, let alone to
kinetically establish a continuous Cr2O3 scale. For the A-B O2
BOν
latter, the critical Cr mole fraction, NCr
crit
, in Ni–Cr

alloy exposed to air at 1000 C is experimentally
found to be 0.2.37 N 0B
JB

1.09.3.2 Criterion for the Sustained

Exclusive Growth of a Protective Scale
If a continuous BOu scale on an A–B alloy is estab- N iB
lished under a particular set of conditions, it may be
necessary to determine if its growth can be sustained Figure 8 Schematic representation of the concentration
under a different set of conditions. (The stoichiomet- profile of B in a binary alloy A–B which is forming an
ric factor u for the oxide BOu will be used for the exclusive scale layer of BOn.

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

192 Principles of High Temperature Corrosion

criterion given in [38] gives only the minimum possible diffusion-controlled and the depth of the internal
B content in the alloy necessary to supply B at a suffi- oxidation zone, x, in an alloy undergoing no external
cient rate to the alloy/scale interface for the sustained scale formation can be described by the following
growth of an established BOu scale layer. The actual kinetic expression, assuming no enrichment of B in
B content necessary for both the establishment and the internal oxidation zone44,45:
sustained growth of a BOu scale will very likely be higher
s 
than NBðminÞ
o 2NO DO t 1=2
owing to transient and kinetic effects. x¼ ½41
nNBo
where NOs is the solubility of the oxidant in the alloy,
1.09.3.3 Internal Oxidation DO is the diffusivity of the oxidant in the alloy, NBo is
the initial solute concentration in the A–B alloy, and
Internal oxidation is used here in a generic sense to t is the time. The product NOs DO is the oxygen per-
represent a process in which a diffusing oxidant from meability in alloy. The solubility of oxygen in the
the surface reacts with a less-noble solute component alloy depends upon the oxygen partial pressure at the
in the alloy to form discrete particles.39 Internal oxi- alloy surface in accordance with Sievert’s law,6 that is,
dation is not desired because it changes the optimized
1=2
mechanical properties of an alloy and may result in NOs / PO2 ½42
internal stress and weakening of the grain boundary.
Thus, the depth of internal oxidation would be
For a dilute binary alloy A–B, the dissolved oxygen
expected to decrease with decreasing PO2 in the man-
atoms can react with less-noble B atoms in the alloy in 1=4
ner x / PO2 .
the manner,
The above discussion considers only internal pre-
BþuO ! BOu ½39 cipitation in the absence of external scale formation.
In the case of AOu scale formation, the value of NOs is
The necessary condition for BOu formation in the no longer fixed by the environment but is instead
alloy may be formulated in terms of the equilibrium fixed by the equilibrium between at the alloy/scale
solubility product Ksp , such that: interface (assuming no through-scale access by the
oxidant) according to the reaction,
½aB ½aO n > Ksp ½40
n
A þ O2 ¼ AOn ½43
The degree to which the solubility product must be 2
exceeded (i.e., the degree of supersaturation) before The effective thickness of the internal precipitation
precipitation occurs depends on a number of factors, zone in the situation of concurrent AOu-scale forma-
such as the stability, composition, and crystal structure tion decreases in a most pronounced manner from
of the precipitating BOu phase. For instance, if the that predicted in the absence of an external scale at
crystal structure and lattice dimensions of BOu are large kp and NBo values.46 A large kp value for AOu-
such that it can form a coherent interface with the scale growth corresponds to a large amount of metal
alloy matrix, then only a small degree of supersatura- recession, which means the total attack is still high,
tion is likely to be necessary for BOu precipitation. In even though the internal precipitation zone is rela-
general, the greater the stability of BOu , the lower will tively low.
be its Ksp value. It is for this reason that the internal The preceding theoretical considerations are
precipitates commonly observed in oxidized high based on models assuming predominant lattice diffu-
temperature alloys are of the stable oxides Al2O3, sion of both oxygen and the alloying element. As
SiO2, TiO2, and Cr2O3.40,41 Multiple internal oxida- such, the theoretical treatments are only expected
tion zones can also develop if more than one reaction to apply to relatively high temperatures or to very
product is stable.42,43 The sequence of the thermody- large-grained materials, with the extreme being sin-
namically possible phases progresses from metal-rich gle crystals. As the temperature is reduced, grain-
in the innermost zone of the alloy to oxidant-rich at boundary effects will become increasingly important.
the surface. This is due to relatively rapid diffusion along grain
The internal oxidation zone extends to the depth boundaries, enhanced concentration or segregation of
at which the activity of dissolved oxygen becomes too the oxidant and the alloying element at grain bound-
small for the formation of the oxide BOu . The kinet- aries, and a preferred tendency for nucleation at grain
ics of internal oxidation are generally found to be boundaries.4

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

Thermodynamics and Theory of External and Internal Oxidation of Alloys 193

1.09.3.4 Transition from Internal Oxidation experimental studies conducted over the past
to External Scale Formation 40 years have validated this criterion.1
The transition from internal to external oxide forma-
tion typically occurs with a relatively small increase 1.09.4 Epilogue
in the alloy content of the less-noble component B in
the A–B alloy, at which oxidation leads to the forma- Advances in the high temperature stability of materi-
tion of a surface scale of BOu, and the alloy no longer als are critically needed to realize the full perfor-
undergoes internal oxidation. Considering again the mance and potential of many current and future
alloy system A–B, there are two limiting possible commercial systems. These advances must lead to
situations: one is when only one of the components significantly enhanced capabilities that will allow
can oxidize and the other is when both of the com- high temperature components to operate robustly
ponents can oxidize. For the first case, if the oxygen for prolonged periods in harsh environments, such
partial pressure is low (i.e., below the stability of AOu as those involving aggressive gases, deposits, photon
formation), the component B in the alloy surface or radiation fluxes, stresses, high or low pressures,
nucleates as BOu in an A-rich matrix. If B can diffuse or some combination of these conditions. For exam-
fast enough to reach the surface and maintain a ple, coal gasification, biomass conversion, and gas-
sufficient supply for BOu formation, a complete sur- cooled nuclear reactor systems typically produce
face layer of BOu will be established. However, if this complex, multioxidant gaseous environments that
condition is not met, then BOu will precipitate as can be highly aggressive from the standpoint of
internal particles within the alloy. So, the internal or surface degrading structural components. The result-
external formation of BOu depends on the balance ing multioxidant process environments are often non-
between the outward flux of B and the inward flux of equilibrium and can involve both gaseous and deposit-
atomic oxygen into the alloy. induced attack. Fundamentally, the high temperature
Wagner47 proposed that the condition for the stability of a material in an aggressive environment
transition from internal to external BOu formation relates to reactions and transport at and across its
occurs when a critical volume fraction of BOu, f *, is external surfaces. Similar to what was discussed in
attained. Under this condition, the influx of oxygen is this chapter, these reactions are defined in terms of
so restricted that sideways growth of the internal BOu some combination of chemical potential, temperature
precipitates is kinetically favored to the extent that and pressure and can be highly complex. Different
BOu eventually forms as a continuous layer on the processes, many of them coupled, are involved from
alloy surface. Under conditions of no or a negligibly the onset of reaction, that is, the absorption and dis-
small rate of metal recession, the following criterion sociation of gaseous molecules at the surface, to the
for the transition from internal to external BOu for- steady-state growth of a protective surface scale that
mation was obtained: develops. Indeed, the ability to control the growth and

 1=2 stability of this scale and to predict its behavior under
 Vm NOS DO
N B f
o
p ½44 different types of extreme chemical environments for
Vox 2nDB extended periods of time will require a much greater
where NBo is the critical mole fraction of B in the basic understanding of the underlying reactions and
alloy, Vm is molar volume of the alloy, Vox is molar transport processes involved. The basic starting points
volume of the oxide, and DB is the diffusion coeffi- have been presented in this chapter, but still much
cient of solute B in the alloy. Rapp44 reported excel- more research and development are needed to improve
lent agreement between experimental and reasonably the reliability and durability of materials exposed to
predicted values of NIno  as a function of oxygen aggressive conditions.
partial pressure in Ag–In alloys oxidized at 550 C.
The value of f * is usually taken to be 0.3 in accor-
dance with the results from this study by Rapp.
From the criterion given by eqn [44], it is seen that References
NBo can be decreased by decreasing the product
of NOs and DO (i.e., decreasing the permeability of 1. Gleeson, B. In Corrosion and Environmental Degradation,
Vol. II: Volume 19 of the Materials Science and Technology
oxygen) and/or increasing DB. Indeed, the direct Series; Schütze, M., Ed.; Wiley-VCH: Weinheim, Germany,
and indirect control of these variables in numerous 2000.

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

 Author’s personal copy

194 Principles of High Temperature Corrosion

2. Lai, G. Y. High-Temperature Corrosion of Engineering 21. Kofstad, P. Nonstoichiometry, Diffusion and Electrical
Alloys; ASM International: Materials Park, OH, 1990. Conductivity in Binary Metal Oxides; Wiley-Interscience:
3. Kofstad, P. High Temperature Oxidation of Metals; John New York, 1972.
Wiley and Sons: New York, 1966. 22. Giggins, C. S.; Pettit, F. S. Trans. AIME 1969, 245, 2495.
4. Young, D. High Temperature Oxidation and Corrosion of 23. Wagner, C. Z. Phys. Chem. 1933, 21, 25.
Metals; Elsevier: Amsterdam, 2008. 24. Wagner, C. Atom Movements; ASM: Cleveland, 1951;
5. Alcock, J. C. B.; Easterbrook, E. Thermodynamics and p 153.
Kinetics of Gas-Metal Systems. In Corrosion, Volume 1: 25. Raynaud, G. M.; Clark, W. A. T.; Rapp, R. A. Metall. Trans. A
Metal/Environment Reactions, 3rd ed.; Shreir, L. L., 1984, 15A, 573.
Jarman, R. A., Burstein, G. T., Eds.; Butterworth- 26. Atkinson, A. Rev. Mod. Phys. 1985, 57, 437.
Heinemann: London, 1994. 27. Garnaud, G.; Rapp, R. A. Oxid. Met. 1977, 11, 193.
6. Gaskall, D. R. Introduction to the Thermodynamics of 28. Hsu, H. S. Oxid. Met. 1986, 26, 315.
Materials, 4th ed.; Taylor & Francis: New York, 2008. 29. Gesmundo, F.; Viani, F. Corros. Sci. 1978, 18, 217.
7. Kubaschewski, O.; Alcock, C. B. Metallurgucal 30. Wang, G.; Gleeson, B.; Douglass, D. L. Oxid. Met. 1989,
Thermochemistry, 5th ed.; Pergamon Press: Oxford, 1979. 31, 415.
8. Barin, I.; Platzki, G. Thermochemical Data of Pure 31. Yurek, G. J. Corrosion Mechanisms; Marcel Dekker Inc.:
Substances; VCH: Weinheim, 1995. New York, 1987; p 397.
9. JANAF Thermochemical Data, Army-Navy-Air Force 32. Dravinieks, A.; McDonald, H. J. Electrochem. Soc. 1948,
Thermochemical Panel, Dow Chemical Company, 94, 139.
Midland, MI, 1962–1963. 33. Mrowec, S. In Proceedings of JIM International
10. Lou, V. L. K.; Mitchell, T. E.; Heuer, A. H. J. Am. Ceram. Symposium on High Temperature Corrosion of
Soc. 1984, 68, 49. Metals and Alloys; Mt Fuji, Japan, 17–20 November
11. Giggins, C. S.; Pettit, F. S. Oxid. Met. 1980, 14, 363. 1982; p 115.
12. Stroosnijder, M. F.; Quadakkers, W. J. High Temp. 34. Stringer, J. Corros. Sci. 1970, 10, 1970, 513.
Technol. 1986, 4, 141. 35. Schütze, M. Protective Oxide Scales and Their
13. LaBranche, M. H.; Yurek, G. J. Oxid. Met. 1987, Breakdown; Institute of Corrosion and Wiley Series on
28, 73. Corrosion and Protection; John Wiley & Sons Ltd:
14. Birks, N.; Meier, G. H. Introduction to High Temperature England, 1991.
Oxidation of Metals; Edward Arnold: London, 1983. 36. Birks, N.; Rickert, H. J. Inst. Met. 1963, 91, 1963, 308.
15. Peraldi, R.; Monceau, D.; Pieraggi, B. Oxid. Met. 2002, 58, 37. Wood, G. C.; Wright, I. G.; Hodgkiess, T.; Whittle, D. P.
275. Werkst. Korros. 1970, 20, 900.
16. Naumenko, D.; Gleeson, B.; Wessel, E.; Singheiser, L.; 38. Wagner, C. J. Electrochem. Soc. 1952, 99, 369.
Quadakkers, W. J. Metall. Mater. Trans. A 2007, 38A, 39. Douglass, D. L. Oxid. Met. 1995, 44, 81.
2974. 40. Gleeson, B.; Harper, M. A. Oxid. Met. 1998, 49, 373.
17. Lee, V. H. J.; Gleeson, B.; Young, D. J. Oxid. Met. 2005, 41. Wei, F. I.; Stott, F. H. High Temp. Technol. 1989, 7, 59.
63, 15. 42. Schnaas, A.; Grabke, H. J. Oxid. Met. 1978, 12, 387.
18. Bradford, S. A. Fundamentals of Corrosion in Gases. In 43. Kane, R. H. Corrosion 1981, 37, 187.
Metals Handbook, 9th ed.; Corrosion; ASM International: 44. Rapp, R. A. Corrosion 1965, 21, 382.
Metals Park, OH, 1987; Vol. 13. 45. Wagner, C. Z. Elektrochem. 1959, 63, 772.
19. Pettit, F. S.; Wagner, J. B. Acta Met. 1964, 12, 35. 46. Gesmundo, F.; Viani, F. Oxid. Met. 1986, 25, 269.
20. Sheasby, J. S.; Smeltzer, W. W. Oxid. Met. 1981, 15, 215. 47. Wagner, C. Z. Elektrochem. 1959, 63, 772.

Shreir’s Corrosion, Fourth Edition (2010), vol. 1, pp. 180-194

View publication stats