Corrosion Science 46 (2004) 2909–2924

www.elsevier.com/locate/corsci

Role of chlorides in hot corrosion

of a cast Fe–Cr–Ni alloy.
Part II: thermochemical model studies
David A. Shores *, Bani P. Mohanty
Department of Chemical Engineering and Materials Science, Corrosion Research Center,
University of Minnesota, Minneapolis, MN 55455, USA
Received 2 December 2002; accepted 27 April 2004
Available online 20 July 2004

Abstract

In many high temperature applications severe degradation of alloys is caused by thin

deposits of molten salts, especially alkali metal sulfates, alkali metal chlorides and mixtures of
these salts. Calculations of multi-component thermochemical equilibria in systems involving
(initially) metal/salt/gas as a function of local oxygen activity can help identify the important
hot corrosion reactions. Such calculations for four pure salts (KCl, NaCl, Na2 SO4 , and
K2 SO4 ) and a mixture of these salts in contact with an Fe–20%Cr alloy at 800
C are pre-
sented. The results predict that the compositions of gas, oxide, (sulfide, when sulfate was
input) and salt phases depend strongly upon the salt chosen and upon the local oxygen
activity. In some cases the equilibrium salt composition is significantly changed by reactions
with metal or oxide phases. The calculated results for the salt mixture were compared with
experimental data from part I of this paper. The model calculations have led to the identifi-
cation of two new factors that support faster hot corrosion rates in an alkali chloride + sulfate
salt compared to that in alkali sulfate alone. First, alkali chlorides, unlike sulfates, support a
continuous salt pathway from ambient to the metal/scale interface, allowing oxidant to be
efficiently transported to oxidize metal. Secondly, under oxidizing conditions alkali chlorides
have a higher solubility of dissolved Fe- and Cr-containing species than that in alkali sulfates.
Both of these factors support higher transport rates, which according to the fluxing theory of
hot corrosion will lead to faster corrosion.
 2004 Elsevier Ltd. All rights reserved.

Keywords: High temperature corrosion; Stainless steel; Molten salt corrosion; HSC Chemistry

*
Corresponding author. Tel.: +1-612-625-4048; fax: +1-612-626-7246.
E-mail address: dshores@umn.edu (D.A. Shores).

0010-938X/$ - see front matter  2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2004.04.014
2910 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924

1. Background

Part I of this paper described the hot corrosion behavior of a cast, high-chromium
stainless steel (HH) that was induced by thin deposits of a mixture of sodium and
potassium sulfates and chlorides. It was observed that alkali chlorides in the salt
deposit changed the corrosion product morphology compared to corrosion by alkali
sulfate deposits. More importantly, the addition of chlorides caused the corrosion
rate to increase by a factor of roughly 100 compared to the rate in alkali sulfate
deposits. To help understand the basis for the faster corrosion processes, we have
sought to identify in Part II the important participating chemical reactions by the use
of a thermochemical equilibrium model.

2. Local equilibrium model

Hot corrosion involves reactions between a molten salt and an alloy undergoing
oxidation. Ideally, in a ‘‘clean’’ environment like air, the normal oxide film that
forms at the surface of an alloy acts as an effective barrier to further oxidation. The
continuing growth of the oxide film requires the transport of metal species or oxidant
species through the film via solid state diffusion. In practice, however, the oxide film
often becomes porous and cracked, allowing much faster transport along these
channels and leading to a heterogeneous film morphology. Transport is typically the
rate-limiting step, and therefore steady state conditions establish a gradient in
oxygen activity across the scale. When a molten salt is also present, the salt wets the
surfaces of the oxides and is able to penetrate into and through the pores and cracks
by capillary action. Since transport via diffusion through a molten salt will be much
faster than via solid state diffusion through the oxide scale, the scale will grow much
faster, and the relevant gradients and chemical reactions will be those involving salt,
scale and metal phases. Therefore, we visualize a simplified, steady state model in
which the gaseous environment is air or oxygen, the oxide scale contains intercon-
nected cracks and pores which are filled with salt, and chemical reactions are fast
compared to transport. A steady-state gradient in oxygen activity is established and
maintained along the salt phase pathway. In turn, this gradient supports the
transport of species from the gas inward toward the metal and possibly the counter
transport of species outward from the corroding sites toward the gas phase. The
local activities of oxidant, salt, corrosion products and/or metal are taken to be in
equilibrium with the local oxygen activity. Thus, at the gas/scale interface the oxygen
activity is taken to be the ambient, and at the metal/scale interface the oxygen
activity decreases to a very low value corresponding to the equilibrium between
metal and oxide. Between these two interfaces, the local oxygen activity is assumed
to vary in a continuous manner, but the actual variation with distance is complex
because of the complexity of the scale morphology.
Molten salt provides a medium for the transport of oxidants inward toward the
metal and of dissolved metal ions outward. Salt permeating through the porous scale
will encounter a wide range of local oxygen activities, including local conditions near
 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924 2911

the metal/scale interface sufficiently reducing to destroy sulfate salts. Reactions that
form metal chlorides or metal sulfides near the oxide/metal interface will consume
alloy components, esp. chromium, that are intended to build or repair the protective
scale. Metal sulfides will tend to be precipitated near the metal/scale interface, where
the activity of sulfur is high and that of oxygen is low. Iron and chromium chlorides,
which may form at or below the metal/scale interface, can migrate towards the scale/
gas interface as dissolved species in the molten salt phase. At higher local oxygen
activities the cations may precipitate as oxides and thereby contribute to scale
growth. As a consequence the physical structure of the scale will not be a continuous
barrier, but rather it will be a porous and chaotic layer, providing pathways for the
continuation of salt permeation and relatively fast corrosion reactions.
As noted in the Literature Review in Part I, the existence of volatile metal
chlorides, such as CrCl2 (g), was predicted from thermodynamic calculations based
on single chemical reactions. Such species were postulated to enhance the transport
of metal through the corrosion product layer [1]. By defining the reacting system
more broadly to include gas, salt, oxide and metal phases, as has been done in the
present model cases, gas phase activities of metal chlorides may be calculated in a
more realistic setting. Typically, such calculations have shown gas phase activities to
be more constrained than might be inferred from considering single reactions in
isolation. Metal chloride activities were typically much less than 1.0, and were found
to depend strongly on oxygen activity. Nevertheless, it is clear from experimental
corrosion studies, as well as model studies, that volatile species can be active par-
ticipants by transporting material away from the corroding specimen, as evidenced
by condensation of metal salts in cooler parts of the reaction system.
In the hot corrosion experiments described in Part I both alkali sulfates and alkali
chlorides were present. The presence of both salt species increases chemical com-
plexity and makes it imperative to consider the problem of competitive reactions.
For example, in the presence of a limited quantity of salt, containing both alkali
sulfate and alkali chloride, how will Cr partition among several possible phases and
species, such as Cr (in metal), Cr2 O3 (s), CrCl3 (g), CrCl2 (g), CrS(s) and Na2 CrO4
(dissolved)? This and related questions can be addressed by calculating the multi-
component equilibrium for a model system that simulates the local metal–salt-cor-
rosion product environment. To help clarify reactions in the complex salt mixture,
we have also calculated equilibria in single salts (KCl, NaCl, Na2 SO4 , and K2 SO4 )
under comparable conditions of temperature, pressure and alloy composition.

2.1. Computational details

The program ‘‘HSC Chemistry for Windows’’ [2] has been employed to calculate
equilibrium compositions in several sets of systems related to hot corrosion reac-
tions. The first step in the equilibrium calculation is the definition of the system. The
base model system consists of an Fe–20 mol%Cr alloy at a temperature of 800
C
and a total pressure of 1 atm, together with one salt component (KCl, NaCl,
Na2 SO4 , or K2 SO4 , or one special case salt mixture whose composition is
(Na,K)2 SO4 ––42 mol% (Na,K)Cl with the K/Na ratio held at 1.4). This salt mixture
2912 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924

was used in the experimental work reported in Part I. Two ratios of salt-to-Fe + Cr
was considered, as will be discussed further below. The major variable was the ratio
of oxygen to Fe + Cr.
For each system the input parameters consisted of a fixed temperature and
pressure, specified kmol of Fe, Cr, salt and oxygen, and a list of species representing
potential reaction products. Operationally, the input metals were held constant at 0.8
kmol Fe and 0.2 kmol Cr, the moles of salt were fixed at 0.05 or 0.15 kmol, and the
input oxygen kmol were stepped from a small value to a large value with 8–10 steps.
The small value of oxygen was chosen to yield an equilibrated system that included a
metallic phase (typically Fe-rich). This system was intended to simulate the quite
reducing conditions at the metal/scale interface, with salt components present. The
large value of oxygen was chosen such that all the metal was oxidized and the gas
phase had an oxygen activity on the order of 0.5 or greater, which simulates con-
ditions near the gas/scale interface with salt components present. Between these two
extremes were several systems with intermediate oxygen activities, which were in-
tended to simulate the range of local oxygen activities across the scale. One aspect of
these calculations that deviates from experiment is that the ratios of the input ele-
ments (Fe–Cr–salt components) were maintained constant for each set of systems
from reducing to oxidizing. Obviously, this ignores the possibility of selective
transport of some species across the scale. With this caveat in mind this sequence of
equilibrated systems allows us to approximate changes in the compositions of the
salt and other phases as a function of oxygen activity. We presume that such changes
also correspond to compositional changes across the thickness of the oxide scale.
The amount of salt relative to the amount of metal is an important variable. A
thin salt deposit was simulated by choosing the input amount of salt to be small
compared to the amount of reacted metal, i.e., the ratio of salt-to-reacted metal was
chosen to be similar to that observed experimentally. A rough assessment of an
appropriate value of this ratio can be obtained from Fig. 3 in Part I, which shows
that at 800
C the sample gained about 2.5 mg/cm2 after 20 h with an initial salt
deposit of 2.5 mg/cm2 . A mass balance can be expressed as: Dweight ðmeasuredÞ ¼
weight gain of oxygen ðas oxidesÞ  weight loss by salt evaporation or decomposi
tion.
An estimate of the loss of salt via the gas phase can be derived from Fig. 5, Part I.
A detailed accounting is complex, but a rough approximation that is sufficient for the
present purposes can be based on losses of sulfate and chloride. This amounts to a
loss of about 40% of the salt, or 1 mg/cm2 . Converting to moles and applying the
mass balance yielded a salt/metal reacted ratio of 0.08. Accordingly, calculations
have been carried out at ratios of 0.05 and 0.15, bracketing the estimated value. In
most cases changing the amount of salt over this range had little effect on the cal-
culated equilibrium.
All possible equilibrium species must be identified; fortunately, HSC has a fairly
comprehensive database. Because several elements were involved for each case a
relatively large number of potential species was considered; e.g., for the special case
of the mixed salt––Fe, Cr, Na, K, S, O and Cl––over 150 potential species were
identified in the HSC database. This number was reduced to approximately 40 or less
 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924 2913

by eliminating species that were found in early calculations to appear in trivial

concentrations in the calculated equilibria. A small amount of nitrogen gas was
added to ensure a computationally a stable gas phase. By excluding all nitrogen-
containing species except N2 , the nitrogen behaved as an inert gas. A check on the
validity of the gas phase activities may be carried out by separately calculating the
equilibrium constant for a simple reaction, for example,
1
FeOðsÞ ¼ O2 ðgÞ þ FeðsÞ
2
At 800
C pO2 ¼ 1.2 · 1019 when the activities of Fe(s) and FeO(s) are 1.0. This
value is reported in Tables 1, 2, 4, and 5 for the systems where the activities of Fe and
FeO were found to be 1.0, i.e., at low values of input oxygen.
The selected species were grouped into phases: gas, metal, oxide(s), sulfide or salt.
This grouping necessarily involves a presumption about solubilities. For example,
placing Fe and Cr into the ‘‘metal’’ phase is computationally equivalent to assuming
they are soluble and exist as a single phase. Correspondingly, sulfide species and salt
species were grouped into respective separate phases. Each oxide, on the other hand,
was placed in a separate phase. As a simplification, activity coefficients were taken as
1.0. There is no presumption about whether the salt phase is liquid or solid. The
melting temperatures of KCl, NaCl and mixed salt systems are close to or below 800
C, but those of K2 SO4 and Na2 SO4 are above 800
C. Obviously, the issue of
whether a liquid salt phase exists at the given temperature is important to the kinetics
of corrosion, but not to the calculation of an equilibrium composition. While a more
refined approach would consider activity coefficients, the simpler approach adopted
here clearly identifies the broad aspects of the appearance or disappearance of
species.
To calculate an equilibrium composition the HSC program then re-distributed the
input elements among the various species such that the total Gibbs energy of the
system was minimized, subject to the constraints that all interacting species were in
equilibrium with each other and that the mass of the elements was conserved.
Thermodynamically, this corresponded to a closed system calculation. Many species
virtually disappear (their activities are set to 1036 to avoid computational difficulties
from the use of zero) if they are relatively unstable. For example, at very low oxygen
activity, where the activities of Fe and FeO were 1, the activities of Fe3 O4 and
Fe2 O3 were 0.0, (<1036 ).

3. Results

Two modes of displaying results were used: (a) mole fraction (equivalent to
activity for ideal solutions) of each species within a designated phase, and (b) kmol
of each species. The results of the equilibrium calculations are summarized in Tables
1–5 in terms of equilibrium mole fractions. Table 1 (case 1) reports the results for an
Fe–20%Cr alloy with a small amount of KCl salt (0.05 kmol) for a range of oxygen
input values. Between 0.7 and 0.8 kmol input oxygen the oxygen activity increased
2914 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924

Table 1
Equilibrium mole fractions of species for the system Fe–Cr–K–Cl–O with 0.05 kmol salt at 800
C (Case 1),
with 0.15 kmol salt at 800
C (Case 2)
Input oxygen 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
(kmol)
Case 1
Gases
O2 (g) 1.2E)19 1.2E)19 1.2E)19 1.2E)19 5.0E)19 1.4E)15 0.97 0.99
N2 (g) 1.000 1.000 1.000 1.000 1.000 1.000 0.020 0.007
Cl2 (g) 1.2E)18 4.1E)21 4.1E)21 4.1E)21 8.2E)21 4.3E)18 0.008 0.005
CrCl2 (g) 1.9E)16 3.6E)20 3.6E)20 3.6E)20 2.5E)20 3.4E)20 8.2E)15 5.1E)15
CrCl3 (g) 1.7E)18 1.8E)23 1.8E)23 1.8E)23 1.8E)23 5.6E)22 5.9E)09 2.9E)09
CrO3 (g) 2.4E)23 1.4E)24 1.4E)24 1.4E)24 4.0E)24 1.5E)21 3.6E)09 3.7E)09
CrO2 Cl(g) 4.2E)20 1.4E)22 1.4E)22 1.4E)22 2.8E)22 4.7E)20 1.8E)07 1.5E)07
CrO2 Cl2 (g) 3.0E)25 5.7E)29 5.7E)29 5.7E)29 1.6E)28 6.1E)25 1.0E)04 6.6E)05
FeCl2 (g) 2.5E)09 8.4E)12 8.4E)12 8.4E)12 8.4E)12 8.3E)11 1.2E)06 7.2E)07
FeCl3 (g) 2.0E)16 3.9E)20 3.9E)20 3.9E)20 5.5E)20 1.2E)17 7.4E)06 3.7E)06
Salt
KCl 1.00 1.00 1.00 1.00 1.00 1.00 0.98 0.98
CrCl2 2.6E)12 5.1E)16 5.1E)16 5.1E)16 3.6E)16 4.8E)16 1.2E)10 7.3E)11
CrCl3 4.0E)18 4.4E)23 4.4E)23 4.4E)23 4.4E)23 1.4E)21 1.4E)08 7.1E)09
K2 CrO4 1.9E)10 3.2E)09 3.2E)09 3.2E)09 9.2E)09 3.6E)07 0.012 0.018
FeCl2 3.3E)08 1.1E)10 1.1E)10 1.1E)10 1.1E)10 1.1E)09 1.5E)05 9.4E)06
FeCl3 5.9E)17 1.1E)20 1.1E)20 1.1E)20 1.6E)20 3.6E)18 2.2E)06 1.1E)06
FeOCl 3.8E)07 2.2E)08 2.2E)08 2.2E)08 3.1E)08 7.2E)07 0.006 0.005
Oxides
Cr2 FeO4 0.00 1.00 1.00 1.00 1.00 1.00 0.00 0.00
Cr2 O3 1.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00
FeO 1.00 1.00 1.00 1.00 1.00 0.00 0.00 0.00
Fe2 O3 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00
Fe3 O4 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00
KCrO2 1.00 1.00 1.00 1.00 1.00 0.00 0.00 0.00
KFeO2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Metal
Cr 2.7E)07 1.5E)08 1.5E)08 1.5E)08 0.00 0.00 0.00 0.00
Fe 1.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00

Case 2
Gases
O2 (g) 1.2E)19 1.2E)19 1.2E)19 1.2E)19 5.0E)19 1.1E)08 0.968 0.985
N2 (g) 1.000 1.000 1.000 1.000 1.000 1.000 0.020 0.007
Cl2 (g) 1.2E)18 4.1E)21 4.1E)21 4.1E)21 8.2E)21 1.3E)10 0.012 0.008
CrCl2 (g) 1.9E)16 3.6E)20 3.6E)20 3.6E)20 2.5E)20 4.9E)17 1.2E)14 8.3E)15
CrCl3 (g) 1.7E)18 1.8E)23 1.8E)23 1.8E)23 1.8E)23 4.4E)15 1.1E)08 6.0E)09
CrO3 (g) 2.4E)23 1.4E)24 1.4E)24 1.4E)24 4.0E)24 1.6E)15 3.6E)09 3.7E)09
CrO2 Cl(g) 4.2E)20 1.4E)22 1.4E)22 1.4E)22 2.8E)22 9.7E)14 2.2E)07 1.8E)07
CrO2 Cl2 (g) 3.0E)25 5.7E)29 5.7E)29 5.7E)29 1.6E)28 6.9E)15 1.5E)04 1.1E)04
FeCl2 (g) 2.5E)09 8.4E)12 8.4E)12 8.4E)12 8.4E)12 1.6E)08 1.7E)06 1.2E)06
FeCl3 (g) 2.0E)16 3.9E)20 3.9E)20 3.9E)20 5.5E)20 1.3E)11 1.3E)05 7.5E)06
 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924 2915

Table 1 (continued)
Input oxygen 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
(kmol)
Salt
KCl 1.000 1.000 1.000 1.000 1.000 1.000 0.985 0.982
CrCl2 2.6E)12 5.1E)16 5.1E)16 5.1E)16 3.6E)16 7.0E)13 1.7E)10 1.2E)10
CrCl3 4.0E)18 4.4E)23 4.4E)23 4.4E)23 4.4E)23 1.1E)14 2.6E)08 1.5E)08
K2 CrO4 1.9E)10 3.2E)09 3.2E)09 3.2E)09 9.2E)09 3.7E)05 0.008 0.012
FeCl2 3.3E)08 1.1E)10 1.1E)10 1.1E)10 1.1E)10 2.1E)07 2.2E)05 1.5E)05
FeCl3 5.9E)17 1.1E)20 1.1E)20 1.1E)20 1.6E)20 3.9E)12 3.9E)06 2.2E)06
FeOCl 3.8E)07 2.2E)08 2.2E)08 2.2E)08 3.1E)08 7.3E)05 0.007 0.006

Oxides
Cr2 FeO4 0.00 1.00 1.00 1.00 1.00 1.00 0.00 0.00
Cr2 O3 1.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00
FeO 1.00 1.00 1.00 1.00 1.00 0.00 0.00 0.00
Fe2 O3 0.00 0.00 0.00 0.00 0.00 1.00 1.00 1.00
Fe3 O4 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00
KCrO2 1.00 1.00 1.00 1.00 1.00 0.00 0.00 0.00
KFeO2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Metal
Cr 2.7E)07 1.5E)08 1.5E)08 1.5E)08 0.00 0.00 0.00 0.00
Fe 1.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00

abruptly, and the activities of many other species also changed abruptly. For
example, the mole fractions of Cl2 (g) and Fe2 O3 (s) increased and those of Fe3 O4 (s),
FeCl2 (g) and FeCl2 (salt) decreased abruptly between 0.7 and 0.8 kmol input oxy-
gen. Because of their relevance to the corrosion processes, the mole fractions of
several other species, such as chromium chlorides, are included in the table to show
that these species were present at low concentrations and are unlikely to be active
participants. In reducing conditions the salt phase consisted of nearly pure KCl, with
the primary dissolved species being FeCl2 at a concentration of a few ppbs. Under
oxidizing conditions, (input oxygen P 0.8 kmol) dissolved potassium chromate and
FeOCl replaced FeCl2 as the primary solutes, and the concentration of solute in-
creased to about 2%.
Table 1 (case 2) reports equilibrium calculations for the Fe–Cr–K–Cl–O system
with a larger (0.15 kmol) amount of KCl, but with the other parameters the same as
for case 1. The results are only incrementally different from case 1. The results for
NaCl were similar; for brevity only one reducing and one oxidizing condition is
shown at two different input kmol of NaCl in Table 2. Under reducing conditions the
major dissolved species in the NaCl phase were Na2 CrO4 and FeOCl, but at the ppb
level. In oxidizing conditions, the same species were dominant at a level of about 2%.
The behavior of the sulfate salts was quite different. Table 3 shows equilibrium
compositions as a function of oxygen for the system with K2 SO4 . Under reducing
conditions, (pO2 ¼ 1.2 · 1019 ) Fe and FeO are stable, but the salt is completely re-
duced to sulfides and KCrO2 . At the next level of oxygen, K2 SO4 is stable, as are the
metal sulfides and KCrO2 ; but Fe is not stable. At an intermediate oxygen level
2916 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924

Table 2
Equilibrium mole fractions of species for the system Fe–Cr–Na–Cl–O at 800
C
Input oxygen 0.05 kmol salt 0.15 kmol salt
(kmol) (Reducing) 0.4 (Oxidizing) 0.9 (Reducing) 0.4 (Oxidizing) 0.9
Gas
O2 (g) 1.2E)19 0.989 1.2E)19 0.987
N2 (g) 1.000 0.007 1.00 0.007
Cl2 (g) 8.8E)21 0.004 1.3E)21 0.006
CrCl2 (g) 1.3E)18 4.1E)15 1.2E)20 6.5E)15
CrCl3 (g) 9.9E)22 2.1E)09 3.3E)24 4.1E)09
CrO2 Cl(g) 3.5E)21 1.3E)07 7.9E)23 1.6E)07
CrO2 Cl2 (g) 2.1E)27 5.2E)05 1.8E)29 8.2E)05
FeCl2 (g) 1.8E)11 5.7E)07 2.7E)12 9.0E)07
FeCl3 (g) 1.2E)19 2.6E)06 7.0E)21 5.2E)06
NaCl(g) 3.6E)04 3.6E)04 3.6E)04 3.6E)04
Salt
NaCl 1.00 0.981 1.00 0.985
CrCl2 1.9E)14 5.7E)11 1.6E)16 9.1E)11
CrCl3 2.4E)21 5.0E)09 8.0E)24 1.0E)08
Na2 CrO4 1.6E)08 0.015 6.3E)09 0.009
FeCl2 2.3E)10 7.4E)06 3.5E)11 1.2E)05
FeCl3 3.6E)20 7.6E)07 2.1E)21 1.5E)06
FeOCl 3.2E)08 0.004 1.3E)08 0.005
Oxide
Cr2 FeO4 0.00 0.00 1.00 0.00
Cr2 O3 1.00 1.00 0.00 1.00
FeO 1.00 0.00 1.00 0.00
Fe2 O3 0.00 1.00 0.00 1.00
Fe3 O4 0.00 0.00 0.00 0.00
CrNaO2 0.00 0.00 0.00 0.00
NaFeO2 0.00 0.00 0.00 0.00
Metal
Cr 2.7E)07 0.00 1.5E)08 0.00
Fe 1.00 0.00 1.00 0.00

(pO2 ¼ 1.2 · 108 ) K2 SO4 is stable, but with dissolved chromate and ferrous ions in
the range of about 10 ppm, along with oxides; only FeS among the sulfides remains.
Under oxidizing conditions all the sulfides and KCrO2 have disappeared, and the
predominant dissolved species in the salt is potassium chromate at 0.2%. Table 4
illustrates the behavior of Na2 SO4 , which is quite similar to that of K2 SO4 .
In Table 5 for the mixed salt, the combined effects of sulfate and chloride salts are
evident. Under reducing conditions the sulfate salts have largely been reduced, with
the sulfur entering into FeS and the alkali metals forming chromites and, to a lesser
extent, ferrites. The salt phase has become primarily KCl. Under oxidizing condi-
tions the sulfide species are unstable, and the salt phase is similar to the input salt.
The dissolved components in the salt phase are primarily chromates, which become
more stable as the oxygen activity increases, and a smaller concentration of ferrous
ions. The change in the composition of the salt phase is quite substantial as the
 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924 2917

Table 3
Equilibrium mole fractions of species for the system Fe–Cr–K–S–O with 0.15 kmol K2 SO4 at 800
C
Input oxygen 0.2 0.4 0.6 0.7 0.8 0.9
(kmol)
Gas
O2 (g) 1.2E)19 2.0E)17 1.9E)16 1.1E)08 9.80E)01 9.93E)01
N2 (g) 1.000 1.000 1.000 0.999 0.020 0.007
S2 (g) 3.2E)10 2.4E)07 8.2E)06 1.6E)18 1.7E)34 1.4E)34
SO2 (g) 1.4E)10 6.3E)07 3.5E)05 8.6E)04 8.2E)04 7.3E)04
Salt
K2 SO4 0.00 1.000 1.000 1.000 0.997 0.997
K2 CrO4 0.00 1.2E)06 4.9E)07 7.6E)06 0.002 0.002
KCr(SO4 )2 0.00 4.4E)23 1.3E)18 4.4E)10 9.2E)04 7.8E)04
FeSO4 0.00 1.7E)12 2.0E)10 1.9E)06 1.7E)04 1.6E)04
Oxide
Cr2 FeO4 0.00 0.00 1.00 1.00 0.00 0.00
Cr2 O3 0.00 0.00 0.00 0.00 1.00 1.00
FeO 1.000 0.00 0.00 0.00 0.00 0.00
Fe2 O3 0.00 0.00 0.00 1.00 1.00 1.00
Fe3 O4 0.00 1.000 1.00 1.00 0.00 0.00
KCrO2 1.000 1.000 1.00 0.00 0.00 0.00
KFeO2 0.00 0.00 0.00 0.00 0.00 0.00
Sulfide
CrS 3.2E)08 3.12E)08 0.00 0.00 0.00 0.00
FeS 0.667 0.750 1.00 0.00 0.00 0.00
FeS2 4.5E)06 1.37E)04 0.00 0.00 0.00 0.00
K2 S 0.333 0.250 0.00 0.00 0.00 0.00

Metal
Cr 1.2E)09 0.00 0.00 0.00 0.00 0.00
Fe 1.000 0.00 0.00 0.00 0.00 0.00

oxygen activity changes from oxidizing to reducing. These changes are illustrated in
Figs. 1 and 2, which show as a function of oxygen activity, the composition of the
major salt components, and the concentrations of the minor dissolved species
involving Cr and Fe.
These results are broadly in accord with the experimental results reported in part
I, which showed large changes in the salt compositions. However, the experimental
analyses could not sample salt from reducing and oxidizing regions separately,
therefore, they must be viewed as composite analyses representing some average of
reducing and oxidizing conditions. 1
It is useful to distill a broader, qualitative view of the major phases present in the
equilibrated systems using the detailed information in Tables 1–5 expressed in terms

1
Neither NaFeO2 or KFeO2 were observed experimentally by X-ray diffraction of the corrosion
products in part I.
2918 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924

Table 4
Equilibrium mole fractions of species for the system Fe–Cr–Na–S–O at 800
C
Input oxygen 0.05 kmol salt 0.15 kmol salt
(kmol) (Reducing) 0.4 (Oxidizing) 0.9 (Reducing) 0.2 (Oxidizing) 0.9
Gas
O2 (g) 1.2E)19 9.93E)01 1.2E)19 0.993
N2 (g) 1.00 0.007 1.00 0.007
S2 (g) 6.7E)10 1.9E)35 3.2E)10 5.3E)35
SO2 (g) 2.0E)10 2.7E)04 1.4E)10 4.6E)04
SO3 (g) 5.8E)20 2.2E)04 4.0E)20 3.7E)04
Na2 SO4 (g) 3.4E)16 6.0E)09 2.7E)15 6.0E)09
Salt
Na2 SO4 0.00 0.998 0.00 0.999
Cr2 (SO4 )3 0.00 1.2E)12 0.00 5.7E)12
Na2 CrO4 0.00 0.002 0.00 9.3E)04
FeSO4 0.00 5.8E)05 0.00 9.7E)05
Fe2 (SO4 )3 0.00 4.1E)12 0.00 1.9E)11
Na2 SO3 0.00 8.5E)11 0.00 8.5E)11
Oxides
Cr2 FeO4 1.00 0.00 0.00 0.00
Cr2 O3 0.00 1.00 0.00 1.00
FeO 1.00 0.00 1.00 0.00
Fe2 O3 0.00 1.00 0.00 1.00
Fe3 O4 0.00 0.00 0.00 0.00
NaCrO2 1.00 0.00 1.00 0.00
NaFeO2 0.00 0.00 0.00 0.00
Sulfide
CrS 6.1E)07 0.00 1.2E)07 0.00
FeS 0.959 0.00 0.667 0.00
FeS2 9.3E)06 0.00 4.5E)06 0.00
Na2 S 0.041 0.00 0.333 0.00
Metal
Cr 1.5E)08 0.00 4.5E)09 0.00
Fe 1.00 0.00 1.00 0.00

of kmol instead of mole fractions. In Tables 6–10 these results are organized in two
ways: (a) the major species existing in each condensed phase are listed in the order of
decreasing amount down to about 10% and (b) the ‘‘disposition’’ of each element is
summarized, i.e., the dominant species into which each element has segregated. For
example, Table 6 for KCl input salt shows that the oxide phase is primarily FeO at
low pO2 and primarily Fe2 O3 at high pO2 . The salt phase is primarily KCl at both
low and high pO2 . Chromium segregates to the spinel oxide at low pO2 and to Cr2 O3
at high pO2 . Table 7 shows that the disposition of elements and phases in the Fe–Cr–
Na–Cl–O systems is very similar to that in the Fe–Cr–K–Cl–O system.
As would be expected, the behavior of systems with alkali sulfate salts has both
differences and similarities compared to systems with alkali chloride salts. Table 8
(K2 SO4 input) shows that at low pO2 , the sulfate decomposed, yielding KCrO2
 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924 2919

Table 5
Equilibrium mole fractions of species for the system Fe–Cr–K–Na–S–Cl–O (mixed salt) at 800
C
Input oxygen 0.05 kmol salt 0.15 kmol salt
(kmol) (Reducing) 0.4 (Oxidizing) 0.9 (Reducing) 0.3 (Oxidizing) 0.9
Gas
O2 (g) 1.2E)19 0.992 1.2E)19 0.991
N2 (g) 1.000 0.007 1.000 0.007
Cl2 (g) 1.1E)20 0.001 1.0E)20 0.002
SO2 (g) 2.1E)10 1.0E)04 2.1E)10 1.5E)04
SO3 (g) 6.0E)20 8.3E)05 6.0E)20 1.2E)04
CrO2 Cl2 (g) 1.5E)28 1.7E)05 1.4E)28 2.6E)05
FeCl2 (g) 2.3E)11 1.9E)07 2.1E)11 2.9E)07
FeCl3 (g) 1.7E)19 4.9E)07 1.5E)19 9.2E)07
Na2 SO4 (g) 7.3E)17 1.5E)09 8.4E)17 1.5E)09
Salt
K2 SO4 1.0E)07 0.327 1.1E)07 0.327
Na2 SO4 1.2E)08 0.254 1.4E)08 0.253
KCl 0.810 0.258 0.803 0.260
NaCl 0.190 0.152 0.197 0.153
K2 CrO4 7.8E)10 0.005 8.3E)10 0.003
KCr(SO4 )2 1.4E)32 2.3E)05 1.5E)32 4.2E)05
Na2 CrO4 2.7E)11 0.001 3.1E)11 7.1E)04
CrCl2 1.4E)15 1.9E)11 1.3E)15 2.9E)11
CrO2 Cl2 (l) 2.4E)29 2.7E)06 2.2E)29 4.1E)06
FeCl2 3.0E)10 2.4E)06 2.8E)10 3.7E)06
FeCl3 5.1E)20 1.5E)07 4.5E)20 2.7E)07
FeOCl 3.6E)08 0.002 3.5E)08 0.003
FeSO4 8.0E)17 2.1E)05 8.0E)17 3.2E)05
Oxides
Cr2 FeO4 1.00 0.00 1.00 0.00
Cr2 O3 0.00 1.00 0.00 1.00
FeO 1.00 0.00 1.00 0.00
Fe2 O3 0.00 1.00 0.00 1.00
Fe3 O4 0.00 0.00 0.00 0.00
KCrO2 0.493 0.00 0.507 0.00
KFeO2 0.011 0.00 0.00 1.00
NaCrO2 0.458 0.00 0.493 0.00
NaFeO2 0.038 0.00 0.00 1.00

Sulfide
CrS 6.3E)07 0.00 6.3E)07 0.00
FeS 0.990 0.00 0.989 0.00
FeS2 1.0E)05 0.00 9.9E)06 0.00
K2 S 6.9E)04 0.00 7.3E)04 0.00
Na2 S 0.009 0.00 0.010 0.00
Metal
Cr 1.5E)08 0.00 1.5E)08 0.00
Fe 1.00 0.00 1.00 0.00
2920 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924

Salt Composition in Mixed Salt - 0.15 kmoles

1

Equilibrium mole fractions

K2SO4
0.8 Na2SO4
KCl
NaCl
0.6

0.4

0.2

0
-20 -15 -10 -5 0
LOG pO2

Fig. 1. Plot of calculated mole fractions of the (major) components of the salt phase as a function of
oxygen activity for the case of mixed salt on Fe–20 mol%Cr alloy at 800
C. The initial salt composition
was is (Na,K)2 SO4 –42 mol%(Na,K)Cl with the K/Na ratio held at 1.4. The input quantities defining the
system were 0.15 kmol of mixed salt, 0.8 kmol Fe, 0.2 kmol Cr and variable oxygen ranging from 0.3 to 0.8
kmol.

Dissolved Species in Mixed Salt - 0.15 kmoles

0
Log (equilibrium mole fractions)

-5

-10

-15

-20
log (K2CrO4)
log (KCrCrO4)
-25 log (Na2CrO4)
log (CrCl2)
log (FeCl2)
-30 log (FeOCl)
log (FeSO4)
-35
-20 -15 -10 -5 0
Log pO2

Fig. 2. Plot of calculated mole fractions of dissolved (minor) species in the salt phase as a function of
oxygen activity for the case of mixed salt on Fe–20 mol%Cr alloy at 800
C. The initial salt composition
was is (Na,K)2 SO4 –42 mol%(Na,K)Cl with the K/Na ratio held at 1.4. The input quantities defining the
system were 0.15 kmol of mixed salt, 0.8 kmol Fe, 0.2 kmol Cr and variable oxygen ranging from 0.3 to 0.8
kmol.

and FeS + K2 S. At high pO2 , K2 SO4 is stable and displaces KCrO2 . The Fe–Cr–Na–
S–O system (Na2 SO4 input), described in Table 9, parallels the behavior of the Fe–
 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924 2921

Table 6
Predominant phases and segregation of elements for the system Fe–Cr–K–Cl–O with 0.15 kmol salt at
800
C
Low pO2 High pO2
Oxide FeO, Cr2 O3 Fe2 O3 , Cr2 O3
Metal Fe None
Salt KCl KCl

K KCl KCl
Cl KCl KCl
Fe FeO, metal Fe2 O3
Cr Cr2 O3 Cr2 O3

Table 7
Predominant phases and segregation of elements for the system Fe–Cr–Na–Cl–O with 0.15 kmol salt at
800
C
Low pO2 High pO2
Oxide FeO, Cr2 FeO4 Fe2 O3 , Cr2 O3
Metal Fe None
Salt NaCl NaCl + Na2 CrO4 (minor)

Na NaCl NaCl
Cl NaCl NaCl
Fe FeO, metal Fe2 O3
Cr Cr2 FeO4 Cr2 O3

Table 8
Predominant phases and segregation of elements for the system Fe–Cr–K–S–O with 0.15 kmol salt at
800
C
Low pO2 High pO2
Oxide FeO, KCrO2 Fe2 O3 , Cr2 O3
Metal Fe None
Salt None K2 SO4
Sulfide FeS, K2 S None
K KCrO2 , K2 S K2 SO4
S FeS, K2 S K2 SO4
Fe FeO, metal, FeS Fe2 O3
Cr KCrO2 Cr2 O3

Cr–K–S–O system, with minor variations based on the relative stabilities of Na2 S vs.
K2 S and KCrO2 vs. NaCrO2 . In both systems, because the alkali sulfate salts are
unstable at low pO2 , the alkali chromites are promoted to the primary alkali-metal
containing species.
This interplay becomes more complex when the system includes Na, K, S, and Cl,
as described in Table 10. Broadly speaking, however, these interactions appear to be
2922 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924

Table 9
Predominant phases and segregation of elements for the system Fe–Cr–Na–S–O with 0.15 kmol salt at
800
C
Low pO2 High pO2
Oxide FeO, NaCrO2 Fe2 O3 , Cr2 O3
Metal Fe None
Salt None Na2 SO4
Sulfide FeS, Na2 S None
Na NaCrO2 , Na2 S Na2 SO4
S FeS, Na2 S Na2 SO4
Fe metal, FeO Fe2 O3
Cr NaCrO2 Cr2 O3

Table 10
Predominant phases and segregation of elements for the system Fe–Cr–K–Na–S–Cl–O with 0.15 kmol
(mixed salt) at 800
C
Low pO2 High pO2
Oxide FeO, KCrO2 , NaCrO2 Fe2 O3 , Cr2 O3
Metal Fe None
Salt 0.8 KCl + 0.2 NaCl K2 SO4 , KCl, Na2 SO4 , NaCl
Sulfide FeS None
K KCrO2 , KCl K2 SO4 , KCl
Na NaCrO2 , NaCl Na2 SO4 , NaCl
S FeS K2 SO4 , Na2 SO4
Cl KCrO2 , KCl KCl, NaCl
Fe FeO, metal, FeS Fe2 O3
Cr KCrO2 , Cr2 FeO4 Cr2 O3

a superposition of the chemical interactions expressed by the simpler systems in

Tables 6–9. For example, since the sulfate component is unstable at low pO2 relative
to sulfides, the salt phase near the metal/scale interface is comprised mainly of
chlorides. KCl dominates because Na becomes sequestered to some degree in chr-
omites. The amount of salt is also reduced to about 40% of the input of 0.15 kmol of
mixed salt (at pO2 ¼ 7 · 1020 ). As the pO2 increases the amount of salt increases with
K2 SO4 returning first (at pO2 ¼ 2 · 1019 ), then both Na2 SO4 and NaCl returning (at
pO2 ¼ 2 · 1018 ), until at high pO2 essentially all the input of 0.15 kmol mixed salt
appears in the equilibrated mixture. This behavior is also illustrated in Fig. 2.
A further comparison of sulfate vs. chloride salts on the basis of dissolved cor-
rosion products (compiled from Tables 1–5) is summarized in Table 11. First, it is
striking that the total dissolved corrosion species in chloride salts is significantly
larger, by a factor of 3–5, than that in sulfate salts. Secondly, it is also clear that the
concentrations of dissolved Cr-containing species greatly exceeds that of Fe-con-
taining species in all the salts, and Cr exceeds Fe by more than a factor of 10 in 4 of
the 5 examples. Thirdly, Na-based salts support a lower solubility for Cr and Fe
species than K-based salts, but the differences in solubilities based on alkali cations is
less striking than the comparison based the anion.
 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924 2923

Table 11
Summary of dissolved Fe- and Cr-containing species dissolved in salt phase oxidizing conditions/0.15 mole
fraction input salt
System Dissolved Fe and Cr species (ppm) Total (ppm)
Fe–Cr–K–Cl–O Fe––17
Cr––11,500 11,517
Fe–Cr–Na–Cl–O Fe––14
Cr––9270 9284
Fe–Cr–K–S–O Fe––155
Cr––2840 2995
Fe–Cr–Na–S–O Fe––97
Cr––930 1027
Fe–Cr–K–Na–S–Cl–O Fe––3060
Cr––4120 7180

4. Discussion

4.1. Comparison of equilibrium model with experimental data

The results of the equilibrium calculations, especially Tables 5 and 10 may be

compared with experimental results in Part I. As predicted by the model, and often
observed experimentally with hot corrosion by sulfate salts, sulfides are formed in
the corrosion product layer. The model suggests that in mixed sulfate + chloride
salts, there will still be salt in contact with the metal, and that salt is expected to be
essentially pure chlorides. The data shown in Fig. 7 of Part I, i.e., in the ‘‘spongy’’,
Cr-depleted metal region where the channels were found to contain salt yielding a
high Cl signal, are consistent with this suggestion.
With regard to the overall composition of the salt phase, the comparison of model
and experiment is more difficult. Figure 5, Part I reports analysis of water-soluble
salt after an exposure of 24 h at 800
C showing chromate as the major dissolved
corrosion product species. However as noted above, these data correspond to salt
compositions averaged over the thickness of the corrosion layer, and more partic-
ularly, to salt compositions exposed to a range of oxygen activities. A direct and
detailed comparison of these results with experimental data is not possible without
additional experimental information about local salt compositions. Nevertheless,
Table 5 for mixed salts (oxidizing conditions/0.15 mole fraction) predicts Cr-con-
taining dissolved species to be predominant, but at a lower concentration.
With regard to the question about why chloride + sulfate salts were much more
aggressive than sulfate salts, two factors have emerged from the equilibrium calcu-
lations. First, the presence of chlorides has allowed a salt phase to exist at the metal/
scale interface. Importantly, this also means that a continuous salt phase pathway
exists to deliver oxidant directly to the metallic phase, and in parallel, to move
dissolved metal species away from the reaction site. Secondly, the equilibrium cal-
culations show that chloride salts and even the chloride + sulfate mixture supported a
higher concentration of dissolved Cr and Fe than sulfate only salts. Since high
temperature corrosion rates typically are expected to be controlled by transport
2924 D.A. Shores, B.P. Mohanty / Corrosion Science 46 (2004) 2909–2924

according to the fluxing model [3], both of these factors argue for a faster corrosion
rate in chlorides.

References

[1] H.J. Grabke, Fundamental mechanisms of the attack of chlorine, HCl and chlorides on steels and high
temperature alloys in the temperature range 400 to 900
C, in: R.W. Bryers (Ed.), Incinerating
Municipal and Industrial Wastes, Hemisphere, New York, 1991, pp. 161–176.
[2] A. Roine, Outokumpu HSC Chemistry for Windows, Ver. 5.1, 1999, Outokumpu Research Oy: Pori,
Findland.
[3] K.N. Lee, D.A. Shores, Transport considerations in the hot corrosion of Ni by Molten alkali
carbonates, J. Electrochem. Soc. 137 (1990) 859–871.