Modelling pitting corrosion of stainless steel

in atmospheric exposures to chloride

containing environments
D. Krouse*1, N. Laycock2 and C. Padovani3
We describe work in modelling pitting corrosion of 300 series stainless steel during exposure to
chloride containing environments, with a focus on atmospheric conditions. It is well known that,
under certain conditions of temperature and humidity, pitting corrosion can initiate on stainless steel
if sufficient quantities of chloride containing salts are deposited on its surface. One of the key
hypotheses that this work intended to test is the existence of bounds on maximum pit depths in the
presence of fixed and possibly limited amounts of cathodic current. In the presence of the relatively
high cathodic currents associated with high relative humidity and high surface contamination by
hygroscopic salts, which are most likely to be representative of outdoor environments, maximum pit
depths are predicted to be of the order of 150300 mm, developed over periods of several months.
In conditions more likely to be representative of indoor conditions with corrosion occurring under
thin moisture layers, pit propagation is severely inhibited and very high chloride concentrations are
required that favour the development of shallow, dish shaped pits.
Keywords: Atmospheric pitting corrosion, Modelling, Stainless steel

This paper is part of a special issue on Long-Term Prediction of Corrosion Damage in Nuclear Waste Systems

Introduction not only in electrochemistry and associated mechanistic

In this paper, we describe work in modelling pitting
corrosion of 300 series stainless steel during exposure to
chloride containing environments, with a focus on atmo-
spheric conditions. Among other applications, such work
is important to understand the durability of thin walled
stainless steel containers for intermediate level radioactive
wastes, such as those used in the UK. These containers
will have to maintain their integrity during prolonged
periods of interim storage in surface facilities potentially in
coastal locations and, later on, during the operational
period of a geological disposal facility in which radioactive
wastes would be finally disposed.1 It is well known that,
under certain conditions of temperature and humidity,
pitting corrosion can initiate on stainless steel if sufficient
quantities of chloride containing salts are deposited on its
surface.2 Stress corrosion cracking can also develop from
pits,3,4 so understanding the factors affecting both the
long term stability and growth rate of pits is important for
a robust assessment of container integrity. The time scales
relevant to the management of radioactive wastes,
expected to be many decades,1 pose significant challenges
1
Callaghan Innovation, PO Box 31-310, Lower Hutt, Wellington, New
Zealand
2
School of Metallurgy and Materials, University of Birmingham,
Edgbaston, Birmingham B15 2TT, UK
3
Nuclear Decommissioning Authority, Radioactive Waste Management
Directorate, Harwell, UK
Corresponding author, email donal.krouse@callaghaninnovation.govt.nz
understanding but also in computer modelling.
One of the key hypotheses that this work intended to
test is the existence of bounds on maximum pit depths in
the presence of fixed and possibly limited amounts of
cathodic current, such as those likely to be present in
atmospheric conditions where only relatively small
cathodic areas develop in the presence of droplets or
thin moisture films. This hypothesis is currently sup-
ported by observation of pits formed in atmospheric
conditions over periods of up to a couple of decades.5
We use deterministic model simulations to show that a
pit has a finite lifetime, making it possible to predict
maximum pit depths over the long term. This approach
can be contrasted with stochastic predictions in which a
population of propagating pits is considered. Classically,
an extreme value distribution is used to predict the
maximum pit depth over an extended area of exposed
metal using only information from a small sample
area.27 Methods such as deterministic extreme value
statistics (DEVS) extend the classical method by
incorporating mechanisms for passivity breakdown such
as the point defect model and also repassivation.28 Using
such methods, it is possible to make longer term
predictions of maximum pit depths based on short term
data, and DEVS has been considered elsewhere in the
context of evaluating its applicability to intermediate
level radioactive waste containers.29 Given the relative
scarcity of data pertinent to the problem in question
(most studies reporting long term pitting data for
relevant stainless steel grades focus on outdoor exposure

2014 Institute of Materials, Minerals and Mining

Published by Maney on behalf of the Institute
Received 15 February 2014; accepted 11 July 2014
DOI 10.1179/1743278214Y.0000000221 Corrosion Engineering, Science and Technology 2014 VOL 49 NO 6 521
Krouse et al. Modelling pitting corrosion of stainless steel
to marine environments), this approach has not been
pursued in this study, since a fully deterministic
treatment requires some knowledge of the distribution
of pit growth rates from relevant measurements on an
adequate sample size. Moreover, it is known that under
current limitation, pits do not grow independently and
that after a time there is a champion pit that eventually
consumes all the current.22,30,31 For this reason, we do
not consider that a stochastic prediction is necessary for
determining pit depth bounds.
Following other workers,6,7 our modelling strategy
has been to deconstruct the overall model into two
component models: one for the anode (i.e. the pit where
the metal dissolves to produce metal ions) and one for
the cathode (i.e. the unpitted surface of the metal where
oxygen is consumed). We then couple the two models
using the principles of charge and mass conservation.
For the anode, we refined the pit propagation model
previously developed by Laycock and White (LW).8 For
the cathode, we consider that an oxygen reduction
reaction (ORR)6 occurs on the metal surface, which is
immersed in a thin film volume of electrolyte. For all
simulations, we have used NaCl solutions, for which
experimental data on both conductivity and expected
corrosion behaviour are readily available.
Fundamental to this work has been the inclusion in the
model of the effects of the local chemistry on the
electrochemical behaviour of the metal as well as
transport limitations associated with the development of
a lacy cover, not fully considered in the previous work
(section on Mechanisms for pit repassivation). Here,
chemistry refers to the composition of the principal
chemical species in solution. The anode model (section on
Anode model) shows that both the lifetime and
maximum size of pits are determined by a number of
other key parameters, especially the critical current
density icrit, which is the current density at the critical
concentration of electrolyte required to hamper repassi-
vation.8 We have then calibrated the model with better
data (section on Model calibration), in particular using
kinetic information obtained from in situ, real time two-
dimensional pit cross-sections imaged using synchrotron
radiography. We have also developed a method for
incorporating in the model concentrated solution effects,
including non-linear conductivity, not previously con-
sidered (section on Modification of LW model for high
chloride). The latter effects are likely to be particularly
important in atmospheric conditions where highly con-
centrated solutions can form in specific conditions as also
described elsewhere.9 With our present simulation soft-
ware, it is infeasible to run simulations of pitting
corrosion in long term atmospheric exposures. In order
to extrapolate mechanistic information to sufficiently
long time scales, we also present (section on Long term
atmospheric pitting of stainless steels) a long term
simulation method based on analytical approximations
of the LW model considering a simplified pit morphology
in which there is no undercutting at the metal surface.
Mechanisms for pit repassivation
Classic criterion of pit stability
A widely accepted view in the relevant scientific
literature is that a pit continues to grow so long as its
pit stability product, xi, maintains a critical value.10
522 Corrosion Engineering, Science and Technology 2014
Here, xi is the numerical product of the pit depth x and
the current density i. Values of between 0?3 and
0?6 A m22 have been reported for stainless steels.1113
The xi criterion was originally developed by Galvele14
for model one-dimensional pits by solving the steady
state mass transport equations for the main ionic
species.15 The product xi is numerically proportional
to the total metal ion concentration at the bottom of the
pit and so corresponds to a critical concentration of
dissolved metal. Accordingly, for the pit to grow, the
metal must dissolve at a rate sufficient to maintain the
critical concentration of metal ions, which, as a result of
metal ion hydrolysis, corresponds to a critical chemistry
in terms of pH and chloride ion concentration.
The xi criterion has been used to assess the stability of
three-dimensional pits including hemispherical7 and dish
shaped.16 In these cases, in order to apply the stability
criterion, the current density is generally assumed to be
constant along the pit surface. Since x cannot decrease,
repassivation of the pit can only occur by a decrease in
pit current. Moreover, the drop in current required to
stifle the pit can be calculated knowing the repassivation
potential Erp, which is the potential at which a stably
growing pit will cease to grow.17 The current drop is
given by integrating the current density as a function of
potential between the limits Erp and the potential of the
pit. In any estimates of pit stability yielded by this
method, however, a key limitation is that it neglects the
effect of the local chemistry in determining the actual
dissolution kinetics as well as the stabilising effect of
remnants of dissolution processes such as lacy covers,
which provide a diffusive barrier able to stabilise pits.
Effect of local chemistry and lacy cover
formation
Using artificial one-dimensional pits, which are not able
to develop a lacy cover, Gaudet et al.18 showed how to
back calculate the concentration of metal ions within a
pit assuming effective diffusive transport. The boundary
conditions are given by the pit current density and the
effective potential, which is the applied potential
adjusted by the solution potential. By this method, they
were able to characterise the dissolution kinetics,
developing empirical correlations between the current
density and the metal ion concentration and potential.
Gaudet et al. found that a strong correlation exists
between the metal concentration and the anodic current
density at the corroding interface at the pit bottom,
which indicated a very fast dissolution above a critical
concentration and much slower dissolution or repassiva-
tion below such concentration. Examples of Gaudet
curves are shown later on in this paper.
Simulations of real pits, both two- and three-dimen-
sional, suggest that the local metal ion concentration
along the corroding interface of a propagating pit can
change significantly depending on the relative ability of
transport processes to locally reduce the concentration
of metal ions released by dissolution and in particular
enhanced transport closer to the pit mouth. In these
conditions, as a result of the characteristics observed
by Gaudet et al., local changes in concentrations of
dissolving metal ions can result in drastic changes in
anodic current density, with areas corroding actively to
form lobes and other areas being effectively passivated.
Using the results of Gaudet et al.,18 as an approximation
VOL 49 NO 6
 Krouse et al. Modelling pitting corrosion of stainless steel

1 Model pit propagating at 20 mA in 1M NaCl after a 15 s, b 20 s and c 50 s; undissolved metal is white region and elec-
trolyte lls remainder; contours give concentration of dissolved metal ions (mol L21), which increases with depth into
pit; pit cavity grows in form of expanding elliptical lobe at pit bottom

to the dissolution kinetics, Laycock and White8 have
proposed that the current density along the pit surface
can be modelled by
 where there is a metal salt film and the metal ion
ia , CCcrit
iC ~ (1)
0, CvCcrit
where C is the total metal ion concentration, ia is the
anodic current density assumed to be independent of C
and Ccrit is a critical metal ion concentration, defined as
the value of C in the Gaudet curve above which the
anodic current density increases sharply. The pit surface
is therefore passive wherever the metal ion concentration
lies below Ccrit. Together with Ccrit, the LW model
defines also a critical current density, icrit, which is an
increasing function of the metal ion concentration
that intersects the pit current density at Ccrit. From
equation (1), the intersection occurs at a definite
potential Ecrit given by
ia Ecrit ~icrit Ccrit (2)
Owing to the solution resistance, the effective metal
potential increases from the bottom of the pit to the rim
and, as a result, the passive region of a propagating pit is
delimited by a critical boundary where C5Ccrit and the
potential is E5Ecrit. Below this boundary, the pit current
density decreases to its mass transport limited value ilim,
concentration is at its saturation value Csat. Above the
critical boundary, the pit current density drops steeply
to zero where the pit surface is passive.
An additional complication that needs to be con-
sidered when modelling pit propagation is the formation
of lacy covers observed by a number of authors. The
mechanism of formation of these covers has been
attributed to local dilution of the aggressive local
chemistry required to support metal dissolution once,
during pit growth, upward propagating lobes intersect
the environment external to the pit that is more
dilute.19,20 In modelling terms, according to the math-
ematical expressions described above, the metal forming
a lacy cover repassivates when C,Ccrit similarly to other
surfaces within the pit.
Anode model
The LW model of pits propagating under constant
potential has been described elsewhere.8,21 Here, we
present unpublished material on the LW model applied

Corrosion Engineering, Science and Technology 2014 VOL 49 NO 6 523

Krouse et al. Modelling pitting corrosion of stainless steel

to pits growing at a constant current, which is closer to  

zeff F dQ
the reality of pits growing under atmospheric conditions; JM ~{D 1z C +C (5)
in this case, the cathodic current is determined by the RT dC
size of wet areas forming the cathode and the thickness where zeff is the effective charge of the metal ions given
of moisture layer available, which decreases the rate of by
oxygen reduction. The main features of model pits
developed under a constant current are illustrated in 1X
zeff ~ metal ions zi ci &z (6)
Fig. 1. Initially, the pit grows and undercuts the metal at C
the pit mouth but, in doing so, ionic transport outside Applying equation (5), the effective diffusion coefficient
the pit is enhanced and, locally, the metal ion of the LW model is seen to be
concentration falls below the critical value. Further  
growth can only occur where the metal ion concentra- z2 C
Deff ~D 1z (7)
tion is sufficiently high for active dissolution. As the 2 cb zC
figure shows, the pit must grow by forming an where D is assumed to be constant (1025 cm2 s21).
expanding elliptical lobe at the pit bottom where the Thus, the effect of the electric field is to enhance
metal ion concentration exceeds the critical value. diffusion of the metal ions and increase conductivity
As reported by Laycock and White,8 the simulation with a relative magnitude that decreases with increasing
time step is constrained, and in our simulations, 45 s of bulk chloride concentration.
computer time is needed to simulate 1 s of pit growth. It
is therefore infeasible to run simulations over long time Spacetime variation of concentration
scales without suitable approximations. These are Because the transport of the metal ions is by effective
discussed later on in this paper. diffusion with a variable diffusion coefficient that
depends on the concentration, the similarity transfor-
Conductivity submodel for NaCl solution mation method can be applied to determine the space
In NaCl solutions, species contributing to the conduc- time variations of metal ion concentrations. Applying
tivity of the electrolyte are Naz and Cl2, which are both this analysis, the concentration must take the following
monovalent and assumed to behave as in an ideal functional form
solution, and the dissolved metal ions Mzz, which are r{r z{z

assumed to behave as in a non-ideal solution. In this
case, the solution conductivity takes the form
DF 2    
s~ 2cb z z2 zc C
RT
where cb is the bulk chloride (and sodium) concentra-
tion, c is a free parameter to correct the non-ideality of
the solution with respect to metal ions, and C is the
concentration of Mzz ions. D is the diffusion coefficient,
F is the Faraday constant, R is the gas constant and T is
the absolute temperature.
In these conditions, the steady state solution poten-
tial as defined by Ohms law is then given by
 
zRT cC
Q~ ln 1z
cF 2cb
This equation results by integrating the ionic current
density divided by the conductivity, with the boundary
condition that the potential is zero where the metal ion
concentration is zero. In the original computer pro-
gramme that generated the LW model results,8 the
conductivity is calculated using equation (3) with c52.
However, with this value, there is a 64 mV/decade
dependence of the potential drop on the bulk chloride
concentration, whereas it is well known that the
dependence should be ,90 mV/decade. A value of
c51?45 is required for the correct value and has been
employed in the latest simulations.
Equivalent diffusive transport
The fact that the potential is a function of metal ion
concentration, logarithmic or otherwise, leads to a
simplification of the transport equations, which are
equivalent to diffusive transport with a concentration
dependent diffusion coefficient
0 0
CM ~C , n (8)
tm t
where r and s are the lateral dimension and depth of the
pit cavity, r0 and s0 are time dependent offsets, t is the
(3)
time, and m and n are numerical values correlating the
metal ion concentration with space and time. Short time
simulation results suggest m5n51/2 for a pit growing
under constant current, which are representative of
parabolic behaviour in both the depth and lateral
dimension. This behaviour has been observed in 304
SS foils under galvanostatic conditions in 0?010?1M
NaCl.22
Lacy cover and repassivation under constant
current conditions
When there is a lacy cover, in the LW model, the
(4)
potential at the pit mouth is spatially constant, since the
metallic nature of such a feature is assumed to make it
sufficiently conductive to dissipate any differences in
potential. This situation also occurs when the solution
conductivity is high due to high bulk chloride concen-
trations. By equation (4), there is also a spatially
constant concentration of metal ions Cmouth at the pit
mouth corresponding to the potential.
In the revised version of the model, this boundary
condition leads to the following estimate of the
repassivation potential
 
zRT cb zCcrit
Erp &DEcorr zba logicrit z ln (9)
2F cb zCmouth
where the quantity
DEcorr :Ecorr {ba logicorr (10)
is fixed (y20?35 V for model 316 SS). Moreover, in
these circumstances, the difference between the potential
of the pit and the repassivation potential is directly
proportional to the current but inversely proportional to

524 Corrosion Engineering, Science and Technology 2014 VOL 49 NO 6

 Krouse et al. Modelling pitting corrosion of stainless steel

2 a radiograph of pit grown for 400 s on 20 mm foil 304 SS in 0?005M NaCl at 650 mV(Ag/AgCl), b triangular mesh
required for model and c representative metal ion concentration (mol L21) calculated from model

the diameter of the pit mouth dm. Consistent with this,

short time simulations, which show variations of dm with
t0?5 as confirmed experimentally,22 imply that the time to
achieve repassivation varies as the square of the current.

Model calibration
The meaning of calibration in this context is that we
are using the experimental data to provide or refine
inputs to the model as opposed to using them to test
the outcomes of the model. In our simulations, pit
morphology is ultimately determined by the evolution of
the current density with time, from which the velocity of
the pit surface can be calculated and integrated over
time by application of Faradays second law. To
accurately predict the pit morphology, especially over
long time scales, it is necessary to obtain accurate
estimates of the model parameters that determine the
local current density and its relationship to metal
potential and metal ion concentration. In order to
calibrate the model, we have used direct measurements 3 Simulation of current density i as function of metal ion
of the local current density derived from two-dimen- concentration and potential E (corrected for potential
sional cross-sections of pits in metal foils obtained using drop within pit, y130 mV) also known as Gaudet
fast in situ synchrotron radiography. The experimental curves, obtained using LW model (304 SS in 0?1M NaCl
work was undertaken by Ghahari and Davenport, and 10 mA current); data are for two positions within
detailed in a PhD thesis.22 The calibration procedure is pit (at rim where C is lowest and at bottom where C is
described in limited detail below. highest) and three different times (indicated by lines of
same style); range of E occurs as both applied poten-
Back calculation of parameters used for model
tial and potential drop in solution adjust to match
calibration applied current
Model parameters are obtained by a process of back
calculation that is a generalisation of the methodology
of Gaudet et al. for one-dimensional artificial pits.18 We velocity of the corroding interface and hence the local
start from in situ, real time measurements of two- current density can be measured accurately. We then use
dimensional pit growth in foils from which the local a finite element model to back calculate the metal ions

Corrosion Engineering, Science and Technology 2014 VOL 49 NO 6 525

Krouse et al. Modelling pitting corrosion of stainless steel
and solution potential taking the local current density as
a boundary condition. This process is illustrated in
Fig. 2. From this work, we have verified that mass
transport can be treated with an effective diffusivity and
we were able to reproduce curves of a similar shape to
those of Gaudet et al., as shown in Fig. 3.
Modification of LW model for high
chloride
From Laycock and White,8 the icrit parameter in the LW
model is defined by the empirical equation
icrit ~x10yC=Csat
where x and y are chosen such that Ccrit is between ,50
and 80% of the saturation value Csat which is assumed to
be independent of the bulk chloride concentration.
However, as the bulk chloride concentration increases,
the saturation concentration of dissolved metal ions
reduces due to a common ion effect.9 As a result of this,
as the chloride concentration increases, there is a
concomitant reduction in the ratio Ccrit/Csat. In order
to capture these bulk chloride effects, we modified the
icrit parameter9 according to the following equation
icrit ~x10yDC=DCsat
in which DCsat decreases with increasing bulk chloride.
Here, x and y are chosen to satisfy equation (2) using
experimental values of DCcrit and DCsat from Ernst and
Newman23 over the range of bulk chloride concentra-
tions from 1 to 9 mol L21.
In the present work, we have conducted simulations
of atmospheric pitting in open circuit conditions using
the amended critical current density defined in equa-
tion (12). Simulations with an initial pit cavity of radius
3 mm and mouth diameter 1?6 mm below a 50 mm film of
electrolyte with an area 3 cm2 were performed. In these
simulations, the ORR kinetics are those described by
Chen et al.6 Model parameters and data are given in
Table 1. The variation of the free corrosion potential
with time, for the pits grown in 1, 3 and 5M NaCl, is
shown in Fig. 4. All pits started with a relatively low
potential, indicative of active anodic dissolution. After
relatively short periods, however, all pits eventually
attained a higher potential equal to typical values of the
Table 1 Model parameters and data used in simulations of atmospheric pitting (section on Modication of LW model for
high chloride)
Parameter Description
D Metal ion diffusion coefficient
F Faraday constant
R Gas constant
T Temperature
z Metal ion charge
ba Anodic Tafel slope
icorr Anodic corrosion current density
Ecorr Anodic corrosion potential
cb Bulk chloride concentration
Csat Saturation concentration of metal ion
Ccrit Critical concentration of metal ion
DO2 Oxygen diffusion coefficient
bc Tafel slope (ORR)
io,c Exchange current density (ORR)
Eo,c Reversible potential (ORR)
ip Passive current density
526 Corrosion Engineering, Science and Technology 2014
(11)
4 Simulation of open circuit potential of stainless steel
(316 SS) under thin moisture layer (50 mm thick) and at
concentrations of 1, 3 and 5 mol L21 NaCl; horizontal
dotted line is open circuit potential when there is no
pitting, i.e. when only passive current on metal con-
sumes cathodic current
(12)
corrosion potential observed on passive stainless steel
surfaces in aerated solutions that are indicative of
repassivation.
Differently from the others, the pit grown in a 7M
NaCl solution propagated stably during the simulation
period (y30 min in real time). Aside from their lifetime,
however, a key difference between the morphology
observed in these cases was that pits grown in 5M NaCl
indicated the formation of a lacy cover, while pits
developed in 7M NaCl did not; instead, they showed a
salt film. These simulations suggest that increasing
chloride concentration increases the lifetime of the
metastable pits and that, in NaCl, 5 mol L21 chloride
appears to be a critical concentration for stable pit
propagation (note that a concentration of 5 mol L21 in
an electrolyte with a surface of 3 cm2 and an electrolyte
thickness of 50 mm would be achieved with an average
surface contamination of ,1 mg cm22). This concentra-
tion is also just below the saturation concentration of
Value
1029 m2 s21
96 485 C mol21
8.314 J mol21 K21
298.15 K
2.2
0.1 V/decade
1 A m22
20.35 V(SCE)
17M NaCl
(5.98983150.63919546cb)/1.25M
(0.6676944160.0309266246cb0.0023897516cb2)6Csat/M
1029 m2 s21
0.1 V(SCE)
1025 A m22
20.05 V(SCE)
1024 A m22
VOL 49 NO 6

sodium chloride at room temperature. If this result was
verified experimentally, this would indicate that, in the
presence of very thin moisture layers y50 mm as in this
simulation, pit propagation is very severely inhibited. It
is interesting to rationalise this result in the context of
experimental observations from long term exposure tests
such as described below. These tests show very limited
propagation in indoor environments when only thin
moisture layers are likely to form as opposed to the
deeper propagation observed in outdoor environment
where bulk condensation or even wetting due to
rainwater can occur. It is also interesting to note that
conditions of high chloride concentration under thin
moisture layers are conditions in which stress corrosion
cracking is observed, i.e. close to the deliquescence point
of available salts.
Long term atmospheric pitting of
stainless steels
The risk factors for pitting corrosion under natural
atmospheric exposures are well known to be moisture,
chloride and hygroscopic particulates. These factors are
linked: ultimately, the main parameters determining the
ability of pits to initiate and grow are the geometry and
chemistry of the layer of liquid that forms on the metal
surface under a humid atmosphere. In order to simulate
long term atmospheric exposure of stainless steel, we
have coupled the LW anode model8 adapted to high
chloride conditions with a cathode model similar to that
developed by Chen et al.6 Our main difficulty has been
to extend the time scales of computer simulations from
tens of seconds to the much longer times necessary to
assess whether a pit has indeed reached a maximum size.
We have been able to overcome some of these problems
using a method of quasi-steady state simulation in
which the pit grows at the current required to maintain a
chloride dependent critical concentration of metal ions
on the bare metal surface.
Model of long term pit propagation under
atmospheric conditions
We have used a finite element model to represent the
system. The geometry is three-dimensional but sym-
metric about the axis of the pit. The main parameters
employed in the model are the liquid film characteristics
(i.e. chloride concentration, oxygen solubility, conduc-
tivity, thin film electrolyte thickness and radius of the
thin film rc); the electrochemical parameters Ecorr, Erp
and ip (the passive current); an empirical parameter DC
that determines the diffusion limited current of the pit;
and an assumed diffusion coefficient D. All parameters
are calculated at room temperature. The DC parameter
is a key parameter that both sets the time scale of the
simulations and captures the effects of the high chloride
concentration. This parameter is calculated from the
chloride concentration9 using a curve based on experi-
mental values from Ernst and Newman.23 The liquid
film characteristics except rc are derived from known
correlations with NaCl deposition density (DD) and
relative humidity (RH) as reported by Chen et al.6
Finally, rc and the electrochemical parameters used are
those reported by Chen and Kelly.7
The results of simulations run employing this model
indicate that a pit growing under a thin liquid film
adapts to the cathodic current limitation by adopting a
Corrosion Engineering, Science and Technology
Krouse et al. Modelling pitting corrosion of stainless steel
5 Model results for depth of propagating pit in 300 series
stainless steels growing under atmospheric conditions
with varying NaCl DD and RH; uctuations in calcu-
lated depth arise from errors in nite element approxi-
mation used to implement model
shallow dish shaped morphology, in which the current
becomes confined to a decreasing lateral area of the pit
as the bottom of the pit repassivates, while the lateral
surfaces are able to grow. The model does not include
the effects of the wetting and drying cycle and corrosion
product formations, which are likely to hinder the long
term propagation of single pits; instead, we use relatively
aggressive conditions of continuous high DD and high
RH to simulate a worst case scenario. Under such
conditions, the model predicts a maximum depth of pits
of the order of 100300 mm attained over periods of
several months, as shown in Fig. 5. The maximum depth
increases with the DD, and for a given density, this
depth increases with the RH.
Both the time scales and depths predicted by the
model are consistent with available field data, although
such data are insufficiently time resolved to be able to
attempt a rigorous validation of the model. In parti-
cular, quantitative data on long term pit growth for
stainless steel under atmospheric conditions show a
relatively limited variability in maximum pit depths in
similar exposure conditions. Chen and Kelly7 collated
field measurements of pits exposed under atmospheric
conditions for up to 26 years. These data show
maximum pit depths of 80 mm for type 316 SS and
,150 mm for type 304 SS. The maximum pit depth in
304 SS was attained in only 0?4 year (i.e. 150 days). In
other reports of exposure tests of 304 SS and 316 SS in
outside environments, even in the most aggressive areas,
heavy industrial and marine, a maximum of only 80 mm
after 18 years was observed.5 However, in short term
tests, maximum depths of 55 mm have been observed in
sheltered, external marine environments after 1 years
exposure.24 By contrast, monitoring data for an
uncontrolled indoor, inland environment showed a
maximum pit depth of ,5 mm over 511 years (see
Smart24 and Smart et al.25), while inspections of stainless
tubing on offshore platforms showed deep pits of several
millimetres within a few years.26
2014 VOL 49 NO 6 527
Krouse et al. Modelling pitting corrosion of stainless steel
Conclusions
We have presented a modelling strategy to evaluate the
development of pitting corrosion of the widely used 300
series stainless steel grades in atmospheric conditions. We
used separate models of the anode and the cathode,
coupled to conserve mass and charge. Based solely on the
anode model, we concluded that a pit attains a maximum
depth given a fixed current. When coupled to the cathode
under atmospheric exposure, the pit maximum depth
could be predicted and correlated with chloride DD and
the RH. Consistent with outdoor field data, simulations
in NaCl solutions predict maximum pit depths to be of
the order of 150300 mm under conditions of relatively
high cathodic current associated with high RH and
chloride DD that are most likely to be representative of
outdoor environments.
For simulations under thin layers (y50 mm) of NaCl
solutions, possibly more representative of indoor envir-
onments, our simulations show that development of
stable pits is very difficult and can only occur at very high
chloride concentrations. At such concentrations, the
morphology of pits switches from undercutting beneath
the metal surface to lateral growth without undercutting.
With standard ORR kinetics and an assumed open circuit
potential of 20?15 V (SCE), this critical concentration is
y5 mol L21 chloride. This effect may be important if
particularly concentrated solutions were to develop on
surfaces contaminated with chloride salts at an RH close
to the deliquescence point, since these are conditions in
which stress corrosion cracking tends to be also observed.
Acknowledgements
This work was supported by the UK Nuclear Deco-
mmissioning Authority under contract no. NR 3274.
Mechanistic studies underpinning model developments
were designed and undertaken by Alison Davenport and
Majid Ghahari of the University of Birmingham School of
Metallurgy and Materials.
References
1. Nuclear Decommissioning Authority: Geological disposal: pack-
age evolution status report, NDA/RWMD/031, 2010. www.nda.
gov.uk/documents/biblio
2. N. Mi, M. Ghahari, T. Rayment and A. Davenport: Corros. Sci.,
2011, 53, 3114312.
3. A. Simoes and M. G. S. Ferreira: Key Eng. Mater., 1987, 20, 3125
3128.
4. P. C.-S. Chen, T. Shinohara and S. Tsujikawa: Zairyo-to-Kankyo,
1997, 46, 313320.
528 Corrosion Engineering, Science and Technology 2014 VOL
5. N. R. Smart, C. C. Naish and A. M. Pritchard: Corrosion
principles for the assessment of stainless steel radioactive waste
containers, Serco, 2006, www.nda.gov.uk/documents/biblio
6. Z. Y. Chen, F. Cui and R. G. Kelly: J. Electrochem. Soc., 2008,
155, (7), C360C368.
7. Z. Y. Chen and R. G. Kelly: J. Electrochem. Soc., 2010, 157, C69.
8. N. J. Laycock and S. P. White: J. Electrochem. Soc., 2001, 148,
B264B275.
9. N. J. Laycock, D. P. Krouse, S. M. Ghahari, A. J. Davenport, T.
Rayment and C. Padovani: ECS Trans., 2012, 41, 316.
10. G. S. Frankel: Pitting corrosion, in Metals handbook, (ed. S. D.
Cramer and B. S. Covino, Jr), Vol. 13A; 2003, Materials Park, OH,
ASM International.
11. G. S. Frankel, L. Stockert, F. Hunkeler and H. Behni: Corrosion,
1987, 43, 429.
12. C. Pistorius and G. T. Burstein: Philos. Trans. R. Soc. London A,
1992, 341A, 531.
13. E. Williams, J. Stewart and P. H. Balkwill: Corros. Sci., 1994, 36,
1213.
14. J. R. Galvele: J. Electrochem. Soc., 1976, 123, 464474.
15. R. Newman: Electrochem. Soc. Interface, 2010, 3338.
16. M. H. Moayed and R. C. Newman: J. Electrochem. Soc., 2006, 153,
(8), B330B335.
17. A. Anderko, N. Sridhar and D. S. Dunn: Corros. Sci., 2004, 46,
15831612.
18. G. T. Gaudet, W. T. Mo, T. A. Hatton, J. W. Tester, J. Tilly, H. S.
Isaacs and R. C. Newman: AIChE J., 1986, 32, 949.
19. P. Ernst, N. J. Laycock, M. H. Moayed and R. C. Newman:
Corros. Sci., 1997, 39, 11331136.
20. N. J. Laycock, S. P. White, J. S. Noh, P. T. Wilson and R. C.
Newman: J. Electrochem. Soc., 1998, 145, 1101.
21. N. J. Laycock, J. S. Noh, S. P. White and D. P. Krouse: Corros.
Sci., 2005, 47, 31403177.
22. S. M. Ghahari: In situ synchrotron X-ray characterisation and
modelling of pitting corrosion of stainless steel, PhD thesis,
University of Birmingham, AL, USA, 2012.
23. P. Ernst and R. C. Newman: Corros. Sci., 2007, 49, 3705.
24. N. R. Smart: Atmospheric pitting corrosion of stainless steel
radioactive waste containers, Serco, 2005. www.nda.gov.uk/docu
ments/biblio
25. N. R. Smart, R. J. Winsley, P. A. H. Fennel, B. Reddy, A. P. Rance
and C. Padovani: Miner. Mag., 2012, 76, (8), 28912899.
26. G. Schiroky, A. Dam, A. Okeremi and C. Speed: Pitting and
crevice corrosion of offshore stainless steel tubing, www.digitalre
fining.com/article/1000087
27. N. J. Laycock, P. J. Laycock, P. Scarf and D. P. Krouse:
Applications of statistical analysis techniques in corrosion science
and engineering, in Shreirs corrosion, 4th edn; 2010,
Amsterdam, Elsevier B.V.
28. D. D. Macdonald and G. R. Engelhardt: Predictive modeling of
corrosion, in Shriers corrosion, (ed. J. A. Richardson et al.),
Vol. 2, 16301679; 2010, Amsterdam, Elsevier.
29. C. P. Jackson and N. R. Smart: Assessment of pit propagation
rates in atmospheric conditions, AMEC report D.006809/001,
2012, www.nda.gov.uk
30. M. I. Suleiman and R. C. Newman: Corros. Sci., 1994, 36, 1657
1665.
31. D. P. Krouse, P. McGavin and N. J. Laycock: Computer
simulation of pitting corrosion: multiple interacting pits, Proc.
48th Annual Conf. of the Australasian Corrosion Association
2008: Corrosion and Prevention 2008, Wellington, New Zealand,
November 2008, Australasian Corrosion Association, 203.
49 NO 6