Journal of The Electrochemical Society, 161 (6) C321-C329 (2014) C321

0013-4651/2014/161(6)/C321/9/$31.00 © The Electrochemical Society

Mathematical Models for Under-Deposit Corrosion

I. Aerated Media
Ya-Chiao Chang,a Richard Woollam,b and Mark E. Orazema,∗,z
a Department of Chemical Engineering, University of Florida, Gainesville, Florida 32611, USA
b BP America, Inc., Houston, Texas 77079, USA

A 2-D axial symmetric time-dependant mathematical model was developed to investigate conditions for onset of under-deposit
corrosion. A model for corrosion in fluids containing dissolved oxygen is presented. The presence of anodic and cathodic regions
was not assumed a priori, but was rather the result of numerical simulations, which revealed galvanic coupling caused by the
differential aeration cells. In contrast to models presented in the literature, the conservation equation for each ionic species was
employed in this work rather than Laplace’s equation. The model included coupled, nonlinear, convective diffusion equations for
ionic species, which included the contribution of migration; local electroneutrality; homogeneous reactions; formation of primary
precipitates; and anodic and cathodic reactions written explicitly in terms of local concentration and potential driving forces.
© 2014 The Electrochemical Society. [DOI: 10.1149/2.034406jes] All rights reserved.

Manuscript submitted October 18, 2013; revised manuscript received March 11, 2014. Published April 30, 2014. This was Paper
2211 presented at the Honolulu, Hawaii, Meeting of the Society, October 7–12, 2012.

Under-deposit corrosion is a type of localized corrosion that Mathematical Model Development

can lead to a catastrophic consequence in the gas and oil indus-
The model development can be classified into three categories:
try. CORMED,1 Norsok,1 and deWaards2 built empirical and semi-
electrochemical reactions, conservation equations and chemical reac-
empirical models that can provide accurate interpolation prediction
tions, and formation and influence of precipitates.
for corrosion rate; however, more complicated correction factors are
required to account for situations where extrapolation or prediction
fails. Mechanistic models developed by Gray,3 Nesic,4,5 and Anderko6 Electrochemical reactions.— The cathodic reaction was assumed
are based on fundamental theories, but the assumption of pre-defined to be the reduction of oxygen
anodic and cathodic regions introduces errors in the predicted corro-
sion rate. O2 + 2H2 O + 4e− → 4OH− [1]
As a step in the development of a model for under-deposit cor-
The corresponding current density can be expressed as
rosion, models for two geometries were developed in this work. The
first is a droplet of water on a metal surface surrounded by air, i.e.,  
2.303
the Evans drop experiment.7,8 Models for atmospheric corrosion often i O2 = −K 0,O2 cO2 (0) exp − (m − 0 ) [2]
βO2
use a similar droplet geometry, but in addition to accounting for elec-
trochemical reactions, such models generally must account for evap- where βO2 is the Tafel slope for the oxygen reduction reaction and
oration and condensation,9 diurnal cycles, and surface-tension driven K 0,O2 is a rate constant with units of A m/mol. The current density i O2
motion of the droplet air interface.10,11 Muto and Sugimoto described is a function of both the local potential driving force (m − 0 ) and
an environment model to simulate the corrosion behavior of Type the concentration of oxygen at the electrode surface cO2 (0).
304 stainless steel in a subtropical marine environment in Okinawa.12 Since iron is a major component in pipeline steel, the anodic reac-
Venkatraman et al.13 describe a model for corrosion under a droplet tion was assumed to be iron dissolution, given by14,15
that accounts for effects of parameters such as exchange current densi-
ties, initial concentrations, shape and size of the droplet, diffusivity of Fe → Fe2+ + 2e− [3]
oxygen and ionic species, and potential and current distributions. As
compared to these other models for atmospheric corrosion, the present The anodic reaction showing the active to passive transition was as-
model employed a fixed droplet size, accounting for transient diffu- sumed to be the consequence of the formation of corrosion products.
sion of oxygen from the air-water interface. The presence of anodic In the absence of films associated with corrosion products, the kinetic
and cathodic regions was determined by local potential distribution current density i k was given by
caused by the formation of differential aeration cells, and the location  
of the anodic and cathodic regions was not assumed a priori, but 2.303
i k,Fe = K 0,Fe exp (m − 0 ) [4]
was instead the result of the numerical simulation. The active-passive βFe
transition for the anodic reaction was also was not assumed a priori,
but was instead obtained from coupling of hypothesized surface and where K 0,Fe is a rate constant with units of A/m2 . Passivation was
homogeneous reactions. included in the kinetic expression by allowing the fractional surface
The second geometry, representing a semi-spheroidal solid deposit coverage of corrosion products γ to weight the contributions of the
surrounded by water under turbulent flow, was developed to explore passive and active current densities. Thus,
the coupling between the deposit in which diffusion was constrained  
and the surrounding area where diffusion was enhanced by turbulent 2.303
i Fe = (1 − γ) K 0,Fe exp (m − 0 ) + γ · i p,Fe [5]
flow. This model is intended to provide a framework for addressing βFe
coupled heterogeneous and homogeneous reactions in systems subject
to corrosion. While the present system assumes an aerated electrolyte where i p,Fe is the value of the passive current density. The value of γ
in which oxygen reduction plays a significant role, future work will was determined from the reactions considered in the model.
explore the role of systems more relevant to the oil and gas industry,
such as HCO− 2−
3 and CO3 in deaerated electrolytes. Conservation equations and chemical reactions.— The model ac-
counted for conservation of ionic species and dissolved oxygen, i.e.,

∗
Electrochemical Society Fellow. ∂ci
z
E-mail: meo@che.ufl.edu = −∇ · Ni + Ri [6]
∂t
Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 C322 Journal of The Electrochemical Society, 161 (6) C321-C329 (2014)

where Ri is the net rate of production of species i by homogeneous a value of 0.16 Pa in this work and is below the critical value such that
reactions. The homogeneous rate Ri can be expressed as the initiation of localized corrosion at carbon steel is independent of
  flow under CO2 corrosion condition.20
Ri = k f cr (r, θ, t) − kb cp (r, θ, t) [7]
r p
Influence and formation of precipitates.— The relationship be-
where r and θ are radial and angular coordinates in a spherical coordi- tween current density and charged species is given by
nate system, kf and kb are respectively the rate constants for forward
and backward reactions, and cr and cp are the concentrations for reac- 
tants and products, respectively. i=F z i Ni [18]
The homogeneous reactions considered in the present model in- i

cluded the dissociation of water and formation of the ferrous hydroxide

precipitates. The rate of the water dissociation reaction At the metal surface, i.e., θ = 90◦ , the fluxes of Fe2+ and OH− can be
written as
H2 O  H+ + OH− [8]
was expressed by i Fe (0)
NFe2+ |θ=90o = [19]
F
RH+ = ROH− = kb (Kw − cH+ cOH− ) [9]
where Kw is the dissociation constant for water. The formation of the and
ferrous hydroxide precipitates was assumed to follow
i O2 (0) i Fe(OH)3
Fe2+ + 2OH−  Fe(OH)2(s) [10] NOH− |θ=90o = − − [20]
F F
The reaction rate was expressed as
⎧ The surface coverage (γ) is the ratio of the apparent surface area of
⎨0 if Fe(OH)2(s) ≤ 0 deposited precipitates to the surface area of metal surface and can be
RFe(OH)2 = expressed as
⎩k
Fe(OH)2 cFe(OH)2 (cOH ) − Ksp,Fe(OH)2 if Fe(OH)2(s) > 0
2

[11] AFe(OH)2 cFe(OH)2 MFe(OH)2 Vmesh

γ= = [21]
where Ksp,Fe(OH)2 is the solubility-product constant. Ametal surface ρs θs Amesh
Under the assumption that convection can be neglected, the con-
centration of dissolved oxygen inside the deposit followed where Vmesh and Amesh are the volume and surface area, respectively,
∂cO2 of the finite element mesh directly above the metal surface, ρs is the
= ∇ 2 cO2 [12] density of precipitated Fe(OH)2 , MFe(OH)2 is the molecular weight of
∂t Fe(OH)2 , and θs is the thickness of a monolayer deposit. Equation 21
The flux of species i was expressed by provides for a local value of surface coverage, and a value of γ greater
F than unity corresponds to a deposit that is thicker than a monolayer.
Ni = − z i Di ci ∇ − Di ∇ci [13] The surface coverage was employed to calculate the passive current
RT
density associated with the formation of Fe(OH)2 . For the less protec-
Within the deposit, convection term was not employed; whereas, out- tive corrosion product case, the anodic current density was modified
side the deposit, diffusion was assumed to be enhanced by turbulent following
flow, as is described later. The potential distribution was assumed to
be governed by ⎧

 ⎪
⎪
⎪
⎪ (1 − γ)K 0,Fe exp 2.303 ( −  )
∇ · (κ∇) + F z i ∇ · (Di ∇ci ) = 0 [14] ⎪
⎪ βFe
m 0
⎪
⎨

i
i Fe = +γK po,Fe exp βFe (m − 0 ) if γ ≤ 1
2.303
[22]
Equation 14 reverts to Laplace’s equation under assumption of a uni- ⎪
⎪
⎪
⎪

form conductivity and in the absence of concentration gradients. ⎪
⎪
⎪
⎩ K po,Fe exp βFe (m − 0 )
2.303
if γ > 1
The geometry comprising two concentric quarter-ellipses was used
to represent the deposits surrounded by a bulk solution. The cov-
ered area was treated as a porous medium, and the effective diffu-
where K po,Fe , the rate constant for iron dissolution at the metal-
sion coefficient of species Di,covered was expressed by the Bruggeman
precipitate interface, is smaller than K 0,Fe for iron dissolution at the
approximation16,17
metal surface. For the protective corrosion product case, the anodic
Di,covered = 1.5 Di [15] current density was modified following
where  is the porosity. In the bulk region, the eddy diffusivity term ⎧

Di(t) was used to account for the influence of turbulence on mass ⎪
⎪
⎪ (1 − γ)K 0,Fe exp 2.303
⎪ ( −  )
transfer.18,19 ⎪
⎨
βFe m 0

Di,bulk = Di + Di(t) i Fe = +γi po,Fe if γ ≤ 1 [23]

[16] ⎪
⎪
⎪
⎪
The corresponding expressions for the eddy diffusivity are ⎪
⎩ i po,Fe if γ > 1
⎧
⎪
⎪ 4A1 z + 3 − 5A2 z + 4
⎪
⎪ for z + ≤ 20
(t) ⎨ where i po,Fe , the potential-independent corrosion current density for
Di 1 − 4A1 z + 3 + 5A2 z + 4
= [17] the passivated electrode covered with a protective deposit, is extremely
ν ⎪
⎪ z+ small as compared to the corrosion current of the bare metal surface.
⎪
⎪
⎩ −1 for z + > 20 A similar approach was employed for the oxygen reduction reac-
2.5
tion with the exception that the reaction on the passive layer surface
where √A1 and A2 are constants, ν is the kinematic viscosity and was assumed to be less inhibited to account for the semiconductive
z + = τ0 ρ/ν is the dimensionless distance from the metal surface, nature of the precipitates. Thus, the oxygen reduction could still occur
where ρ is the density of the fluid. The wall shear stress τ0 was assigned on the precipitates even when the iron dissolution was passivated. The

Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 161 (6) C321-C329 (2014) C323

cathodic current density was expressed as calculated from

⎧  F2  2
⎪
⎪ − (1 − γ)K 0,O2 + γK oxide,O2 × κ= z Di ci [27]
⎪
⎪ RT i i
⎪
⎪
⎨ 

i O2 = cO2 (0) exp − 2.303
βO2
( m − 0 ) if γ ≤ 1 to be 9.3 × 10−3 S/m, a value between drinking water and seawater.
⎪
⎪ The parameters used in the calculations are presented in Table I.
⎪
⎪

⎪
⎪ For the Evans Drop model, the boundary condition at the electrode
⎩ −K oxide,O2 cO2 (0) exp − 2.303
βO
(m − 0 ) if γ > 1
2 was that the solution was electrically neutral, the fluxes of ferrous and
[24] hydroxide ions were given by equations 19 and 20, respectively, and
where K oxide,O2 is the rate constant for oxygen reduction at precipitate- the fluxes of other ionic species were equal to zero. The initial condi-
solution interface. tion at the surface was that the electrode was uncovered by films, i.e.,
The ferrous hydroxide can react with hydroxide ions to produce γ = 0. With the exception of dissolved oxygen, the flux of all species
ferric hydroxide. The net reaction can be written as21 normal to the water=air interface was set to zero. A counter electrode
was placed at the top of the droplet where the value of potential could
Fe(OH)2 (s) + OH−  Fe(OH)3 (s) + e− [25] be specified. Calculations were performed at applied potentials and at
The current density was expressed as the corrosion potential (open-circuit). For calculations at open-circuit,
  an iterative solution at each time set was performed in which the inte-
i Fe(OH)3 2.303   grated current along the electrode surface was set equal to zero. The
= K Fe(OH)3 cOH− exp m − 0 − UFe(OH)3
F βFe(OH)3 equations solved are nonlinear, and a Newton Raphson convergence
[26] was achieved at each time step.
The formation of Fe(OH)3 did not contribute significantly to the value For the Deposit model, the boundary condition at the electrode
of fractional surface coverage γ. was that the solution was electrically neutral, the fluxes of ferrous
and hydroxide ions were given by equations 19 and 20, respectively,
and the fluxes of other ionic species were equal to zero. The ini-
Numerical Method
tial condition at the surface was that the electrode was uncovered by
A Finite-Element Method (FEM, COMSOL Multiphysics, films, i.e., γ = 0. Continuity of flux, concentration, and potential was
http://www.comsol.com) was used to solve the time-dependent equa- imposed at the solid-electrolyte interface, and open boundary con-
tions governing this system, posed in spherical coordinates, e.g., r, θ, ditions were employed at the outer ellipse to ensure that the outer
and φ. Under the assumption of axial symmetry, a one-quarter-ellipse boundary did not influence the results. Thus, under the assumption
geometry was used to represent a water droplet placed on the metal that the curvature of the pipe may be neglected, the model could be
surface as shown in Figure 1a. A geometry consisting of two concen- envisioned as pertaining to a deposit inside a pipe, as is represented in
tric quarter-ellipses was used to represent solid deposits surrounded Figure 2. As potential was obtained by electroneutrality and con-
by electrolyte, as is shown in Figure 1b. servation equations, the results did not require assumption of
The mesh used for the FEM simulation was nonuniform with a a fixed solution conductivity. The conductivity was allowed to
tighter spacing of nodes at the electrode surface and at the boundaries. vary in accordance with the variation of concentration of ionic
The mesh size was selected to be sufficiently small that decreasing species, and the potential was influenced by formation of diffusion
the mesh size further did not change the simulation results. The time potentials.
step t = 1 s was selected such that the simulations for transient
diffusion in spherical and slab geometries gave results that were in
Simulation Results
good agreement with analytic solutions. In addition, simulations were
performed for the present system with values of 0.1, 1, 10, and 100 s The corrosion rate was found to be strongly dependent on the pres-
for t to ensure that the results reported were independent of the ence and properties of surface films and associated chemical and elec-
value used for t ≤ 1. trochemical reactions. In the present work, iron was used as the react-
For the water droplet problem, the water-air interface was assumed ing species as it is the principle material used in gas and oil pipelines.
to be equilibrated with air at all times, such that the the concentra- Iron, chromium, nickel, titanium and their alloys also demonstrate
tion of dissolved oxygen was cO2 = 0.284 mol/m3 . For both Evans active-passive transitions, which in this paper was assumed to be the
drop and predefined deposit models, the initial concentration of dis- consequence of the presence of the corrosion products. Under certain
solved oxygen was assumed to be cO2 = 0.284 mol/m3 , the value conditions, the surface under the drop can experience both active and
corresponding to equilibration with air. The initial concentrations of passive behavior. To explore simultaneous active and passive behavior,
ionic species were cNO−3 (0) = 1.7 × 10−1 mol/m3 , cH+ (0) = 1 × 10−4 parameters were chosen to allow the open-circuit potential distribu-
mol/m3 , cOH− (0) = 1 × 10−4 mol/m3 , and, by electroneutrality, cK+ (0) tion on the metal surface to encompass the region between passive
= 1.7 × 10−1 mol/m3 . The initial conductivity of the solution can be and active behavior.

Air Electrolyte
Electrolyte Solids

(a) (b)

Figure 1. Schematic representations of: a) a water droplet on a metal surface; and b) deposits surrounded by electrolyte.

Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 C324 Journal of The Electrochemical Society, 161 (6) C321-C329 (2014)

Table I. Parameters for the simulations in two different geometries.

Deposit
Evans Drop two-quarter ellipse, Fig. 1b

Geometry one-quarter ellipse, Fig. 1a Less Active Passive

K 0,O2 7.0 × 10−3 A m/mol 1.0 A/m2 3.38 × 10−4 A/m2
D O2 1.97 × 10−9 m2 /s 1.97 × 10−9 m2 /s
cO2 (∞) 0.284 mol/m3 0.284 mol/m3
βO2 0.12 V/decade22 0.12 V/decade22
K oxide,O2 7.0 × 10−3 A m/mol 7.0 × 10−1 A m/mol 2.63 × 10−6 A m/mol
K 0,Fe 1.00 × 10−4 A/m2 5.0 × 10−6 A/m2 1.0 × 10−5 A/m2
DFe2+ 7.20 × 10−10 m2 /s 7.20 × 10−10 m2 /s
βFe 0.06 V/decade 0.06 V/decade
K po,Fe 1.00 × 10−8 A/m2 4.96 × 10−7 A/m2 2.54 × 10−11 A/m2
K Fe(OH)3 1.59 × 106 A m/mol 1.59 × 106 A m/mol
βFe(OH)3 0.12 V/decade23 0.12 V/decade23
UFe(OH)3 0.56 V(NHE) 0.56 V(NHE)
κ(0) 4.42 × 10−4 S/m 4.42 × 10−4 S/m
D H+ 9.31 × 10−9 m2 /s 9.31 × 10−9 m2 /s
DOH− 5.26 × 10−9 m2 /s 5.26 × 10−9 m2 /s
DNO− 1.90 × 10−9 m2 /s 1.90 × 10−9 m2 /s
3
DK+ 1.98 × 10−9 m2 /s 1.98 × 10−9 m2 /s
kb,H2 O 500 500
Kw 1.00 × 10−8 mol2 /m6 1.00 × 10−8 mol2 /m6
Ksp ,Fe(OH)2 4.87 × 10−6 mol2 /m6 4.87 × 10−6 mol2 /m6
 0.2 0.2
Vmesh 7.20 × 10−6 cm3 /s 7.20 × 10−6 cm3 /s
ρs 3,400 kg/m3 3,400 kg/m3
θ 7.0 × 10−9 m 7.0 × 10−9 m
Amesh 7.20 × 10−6 cm2 /s 7.20 × 10−6 cm2 /s
Ksp ,Fe(OH)3 6.3 × 10−38 6.3 × 10−38
cNO− (0) 1.7 × 10−1 mol/m3
3
cH+ (0) 1.0 × 10−4 mol/m3
cOH− (0) 1.0 × 10−4 mol/m3
t 1.0 s
ν 1.05 × 10−6 m2 /s
A1 1.0972 × 10−4
A2 3.295 × 10−6

For the one-quarter-ellipse geometry (see Evans Drop in
Table I), the shift of passivated area can be illustrated by changes
of the total current density. The local pH differences can only be ob-
served when Fe(OH)2 reacts further with OH− to form Fe(OH)3 . The
two-concentric-quarter-ellipse geometry (see Deposit in Table I) was
used to evaluate under-deposit corrosion and the influences of both
the formation of precipitates and the surrounding turbulent flow.24,25
The results are presented in terms of two sets of calculations. The
parameters marked Less Active in Table I lead to a corrosion prod-
uct that still allows iron dissolution and oxygen reduction reaction;
whereas, the parameters marked Passive in Table I lead to a corro-
sion product that strongly inhibits anodic reaction. The results were
used to study the influence of the protectiveness and properties of
precipitates.
Active-passive transitions.— Calculated local anodic and cathodic
current densities are shown in Figure 3 as functions of electrode po-
tential for the deposit model at a position r = 0. The anodic reac-
tion shows an active-passive transition in which the active corrosion
is replaced by passive behavior as the electrode potential increases.
The resulting surface-averaged current, corresponding to the elec-
trode area under the predefined deposit, is presented in Figure 4 as
individual points corresponding to each applied potential. The pas-
sivation condition calculated for more anodic potentials represents

Pipe

Turbulent Flow Figure 2. Schematic representations of pipe with an

embedded deposit model of Figure 1b. The use of open
boundary conditions for the outer ellipse ensures that
the outer boundary did not influence the results.
Open boundary
conditions

Deposit

Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 161 (6) C321-C329 (2014) C325

0.1
0.10 1.5

itotal / Am-2
0.0
-ic -0.1 1.0
0.08 ia -0.2
-0.3 0.5
-0.4
t=1s t=1s 0.0
-0.5
0.06 0.0 0.2 0.4 0.6 0.8 1.00.0 0.2 0.4 0.6 0.8 1.0
0.1 1.5

itotal / Am-2
0.0
0.04 -0.1 1.0
-0.2
-0.3 0.5
0.02
Eavg / V

-0.4
t = 10 t = 10 0.0
-0.5
0.0 0.2 0.4 0.6 0.8 1.00.0 0.2 0.4 0.6 0.8 1.0
0.00 0.1 1.5

itotal / Am-2
0.0
-0.1 1.0
-0.02 -0.2
-0.3 0.5
-0.4 t = 20
t = 20 0.0
-0.5
-0.04 0.0 0.2 0.4 0.6 0.8 1.00.0 0.2 0.4 0.6 0.8 1.0
0.1 1.5

itotal / Am-2
0.0
-0.06 -0.1 1.0
-0.2
-5 -4 -3 -2 -0.3 0.5
10 10 10 10 -0.4 t = 23 t = 23
-0.5 0.0
0.0 0.2 0.4 0.6 0.8 1.00.0 0.2 0.4 0.6 0.8 1.0
ilocal / A m-2 0.1 1.5

itotal / Am-2
0.0
-0.1 1.0
-0.2
Figure 3. Calculated polarization curves showing anodic and cathodic current -0.3 0.5
densities at r = 0 for the Deposit model. -0.4
-0.5
t = 25 t = 25
0.0
0.0 0.2 0.4 0.6 0.8 1.00.0 0.2 0.4 0.6 0.8 1.0
0.1 1.5

itotal / Am-2
0.0
-0.1 1.0
-0.2
a dynamic balance between the passive current and oxygen reduc- -0.3 0.5
-0.4 t = 30
tion, which, as shown in Figure 4 occasionally became negative. The -0.5
0.0 0.2 0.4 0.6 0.8
t = 30
1.00.0 0.2 0.4 0.6 0.8 1.0
0.0

current presented is that averaged over electrode area under the prede- itotal / Am-2
0.1 1.5
0.0
fined deposit, and the negative current was balanced by anodic current -0.1 1.0
-0.2
on the uncovered area. A similar negative current on passivated sur- -0.3 0.5
-0.4
faces at the open-circuit or corrosion potential was seen only when -0.5
t = 35 t = 35
0.0
0.0 0.2 0.4 0.6 0.8 1.00.0 0.2 0.4 0.6 0.8 1.0
balanced by an anodic current on a different portion of the electrode. 0.1 1.5
itotal / Am-2

0.0
The role of film formation passivation is evident in Figure 5 in -0.1 1.0
-0.2
which distributions of current density (Figure 5a) and surface cover- -0.3 0.5
-0.4
age (Figure 5b) are presented at different times for the Evans Drop -0.5
t = 40 t = 40 0.0
0.0 0.2 0.4 0.6 0.8 1.00.0 0.2 0.4 0.6 0.8 1.0
model. These simulations were performed such that the system was r / r0 r / r0
at the corrosion potential and the net current, integrated over the (a) (b)

electrode surface, was equal to zero. At short times, due to the greater
accessibility of oxygen, the cathodic reaction dominated at the periph- Figure 5. Radial distributions calculated at the open-circuit or corrosion po-
tential for the Evans Drop model with time as a parameter: a) total current
ery of the droplet. Ferrous ions are produced by the anodic reaction density; and b) fractional surface coverage.
and react with hydroxide ions produced by the cathodic reaction to
form ferrous hydroxides. Fe(OH)2 further reacts with the excess OH−
to form Fe(OH)3 , which is an insoluble iron hydroxide complex and
could slowly dehydrate to form rust. The Pourbaix diagram presented As time increases, the surface coverage exceeded the full coverage
in Figure 6 shows that Fe(OH)3 is the major precipitate under the value, i.e., γ = 1, starting from the periphery and extending to the
given conditions. As shown in Figure 7, a slightly alkaline region is center until the surface was all covered with precipitates. The anodic
seen at the periphery due to predominance of the cathodic reaction reaction was inhibited by the formation of Fe(OH)3 , but the cathodic
in that region; whereas, the center shows slight acidity as the conse-
quence of the formation of precipitates. The formation of Fe(OH)3
can be indicated by the surface coverage which is shown in Figure 5b. 2.0

1.5 FeO42-
3+ Fe(OH)2+
Fe
0.04 1.0 FeOH2+
Fe(OH)3(aq)
0.03
0.5
Fe2O3(S)
0.02
E (SHE)

0.0 2+
Fe(OH)4-
Fe
0.01
itotal / A m-2

-0.5 Fe3O4(S)
0.00
Fe(OH)3-
FeOH+

-0.01 -1.0
Natural pH

Fe(S)
-0.02 -1.5

-0.03 -2.0
0 2 4 6 8 10 12 14
-0.04
-0.05 0.00 0.05 0.10 pH

m
/V
Figure 6. Calculated Pourbaix diagram representing the equilibrium potential
between iron and its various species as a function of pH at 25◦ C. The diagrams
Figure 4. Surface-averaged current density calculated at the open-circuit or were generated by CorrosionAnalyzer 1.3 Revision 1.3.33 by OLI Systems
corrosion potential for the Deposit model as a function of applied potential. Inc.
Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 C326 Journal of The Electrochemical Society, 161 (6) C321-C329 (2014)

12
t = 92 s t = 92 s
7.12 10 1.05

itotal / Am-2
8
1.00
6
0.95
7.10 4
2 0.90
0
0.85
7.08 0.0
12
0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

10 t = 94 s t = 94 s 1.05

itotal / Am-2
8
7.06 6
1.00

4 0.95

7.04 2 0.90
pH

0
0.85
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
12
7.02 10 t = 95 s t = 95 s 1.05

itotal / Am-2
8
1.00
6
7.00 4 0.95
2 0.90
0
6.98 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5
0.85
3.0
12
10 t = 96 s t = 96 s 1.05
6.96

itotal / Am-2
8
1.00
6
4 0.95
0.0 0.2 0.4 0.6 0.8 1.0 2 0.90
0
r / mm 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5
0.85
3.0
12
10 t = 97 s t = 97 s 1.05
Figure 7. Radial distributions of pH value calculated at the open-circuit or

itotal / Am-2
8
1.00
6
corrosion potential for the Evans Drop model. 4 0.95
2 0.90
0
0.85
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
12
reaction was still active. The active-passive transition was, therefore, itotal / Am-2 10 t = 98 s t = 98 s 1.05
8
a consequence of the formation of precipitates. 6
1.00

4 0.95
2 0.90
Less-Protective Precipitates (Fe(O H )2 ).— In this calculation, the 0
0.85
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
passivation was assumed to result from the less-protective precipitates, r / mm r / mm
(a) (b)
Fe(OH)2 . Fe(OH)3 accumulates on top of Fe(OH)2 and, therefore, does
not play a part in passivation. Figure 8. Radial distributions calculated at the open-circuit or corrosion po-
The influence of turbulent flow for the Deposit model can be ob- tential for the Deposit model with time as a parameter: a) total current density
served in Figure 8a. At t = 92 s, the concentration of oxygen at the and b) fractional surface coverage.
electrode surface approached zero. At times t > 92 s, the transport of
oxygen to the reacting surface in the region with turbulent flow was
greater than in the covered region. As a consequence, the cathodic re-
action was favorable in the surrounding area, while the anodic reaction exchange of fluid with the bulk solution caused the concentrations
dominated in the center. The shift from active to passive behavior was of Fe2+ and OH− to fall below the super-saturated value, thus fa-
observed as the result of the deposition of precipitates, as is shown in voring dissolution of the film. The dissolution rate of Fe(OH)2 was
Figure 8b. assumed to be dependant on the solution composition, temperature,
The corresponding changes of potential distribution are presented and pressure. Two different values of solubility product were used in
in Figure 9. The maximum potential difference of 0.1 V was ob- this paper. Whitman26 calculated the solubility product of Fe(OH)2
tained when the outer region was completely covered with precipitates to be 8.7 × 10−14 in the ferrous-ion concentration in 20% potassium
(t = 96 s). The precipitates continued to accumulate on the metal sur- hydroxide solution. A value of Fe(OH)2 = 8.0 × 10−16 was obtained
face, and the corrosion remained active under the predefined covered by Leussing27 at a temperature of 25◦ C. The impact of the solubil-
region as a result of the differential oxygenation cell driven by the ity product value can be shown in Figure 10 where the fractional
surrounding turbulent flow. surface coverage calculated for the deposit model is presented as a
The stability of films in the exposure of iron to aerated water function of radial position with time as a parameter. The film dis-
at room temperature may be influenced by the solubility of ferrous solved for the solubility product of 8.7 × 10−14 , but remained intact
hydroxide. As the film grew and reduced the rate of corrosion, the for Ksp = 8.0 × 10−16 . Careful examination of Figure 10b shows a

0.1129
3.0
92 s t = 96 s
0.10 0.1
94 s
2.5
95 s
96 s
0.08 0.08
97 s 2.0
98 s
z / mm

1600 s
Figure 9. Potential distributions for the results
\V

0.06 1.5
0.06 presented in Figure 8: a) potential along the elec-
0

trode surface with time as a parameter; b) false

color representation of potential for t = 96 s
0.04 1.0 0.04
with superimposed current lines.

0.02 0.5 0.02

0.0 0
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 -3.579e-3
r / mm r / mm
(a) (b)

Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 161 (6) C321-C329 (2014) C327

1.4 1.4 50 50

1.2 t = 1000 s 1.2 t = 1000 s t = 26000 s t = 12000 s

40 40
1.0 1.0
30 30
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
1.4 1.4
20 20
1.2 t = 1280 1.2 t = 1280
10 10
1.0 1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
1.4 1.4
r / mm r / mm
1.2 t = 1320 1.2 t = 1320
50 50
1.0 1.0 t = 30000 s t = 14000 s
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 40 40
1.4 1.4
30 30
1.2 t = 1800 1.2 t = 1800
20 20
1.0 1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
10 10
1.4 1.4
0 0
1.2 t = 1900 1.2 t = 1900 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
r / mm r / mm
1.0 1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
1.4 1.4 Figure 12. Radial distributions of fractional surface coverage calculated at
1.2 t = 1950 1.2 t = 1950
the open-circuit or corrosion potential for the Deposit model with time as a
parameter: a) Ksp,Fe(OH)3 = 6.3 × 10−36 ; and b) Ksp,Fe(OH)3 = 6.3 × 10−38 .
1.0 1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
1.4 1.4

1.2 t = 2000 1.2 t = 2000

slight thinning at r = 1 mm. For this condition, the anodic reaction
occurred near the intersection of predefined porous medium and bulk
1.0 1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 solution.
1.4 1.4

1.2 t = 44500 1.2 t = 44500 Protective Precipitates (Fe(O H )3 ).— Ferrous hydroxides can re-
1.0 1.0
act further with the excess OH− , producing the secondary precipitates,
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 ferric hydroxides. Ferrous hydroxide was assumed to be in the form
r / mm r / mm
(a) (b) of a colloidal gel that did not passivate the metal surface. Only ferric
hydroxide was assumed to shield completely the active metal surface.
Figure 10. Radial distributions of fractional surface coverage calculated at The cathodic reaction was preferred in the surrounding area, while the
the open-circuit or corrosion potential for the Deposit model with time as a anodic reaction was favored in the center, as shown in Figure 11. As
parameter: a) Ksp,Fe(OH)2 = 8.7 × 10−14 ; and b) Ksp,Fe(OH)2 = 8.0 × 10−16 . a consequence of protective precipitates, the corrosion rate decreased
1.5 after the metal surface was fully-covered.
10
t = 10500 s t = 10500 s However, for the protective case, the influence of solubility
1.0 8
product value is not significant. When the secondary precipitates
itotal / Am-2

6
0.5 (Fe(OH)3 ) is the source of passivation, the change of solubility prod-
4
0.0
uct only affected the time frame for building up deposits, as shown in
2
Figure 12.
0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

1.5 Influence of Uncovered Region.— The ratio of the radius covered

t = 10900 s 10 t = 10900 s
by the deposits to the outer bulk solution region radius (rs /rb ) was
1.0 8
studied since it could be an influential factor for the initiation of
itotal / Am-2

6
0.5 pitting. Deposit to open area ratios of 1:1, 1:2, 1:3, 1:4, 1:14, and
4
0.0 1:39 were investigated. The maximum potential drop was obtained
for rs /rb = 4. As illustrated in Figure 13, for rs /rb = 1, 2, and 4,
2

0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 the potential variation depended strongly on rs /rb . A large potential
1.5 variation at large values of rs /rb as shown in Figure 13 may result in
10
t = 11500 s t = 11500 s pitting and, thus, under-deposit corrosion.
1.0 8
itotal / Am-2

6
0.5
4 Conclusions
0.0
2

0
This work provides a framework to study interactions among
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 chemical reactions, electrochemical reactions, deposition of films,
1.5 and transport. While the emphasis of the present work is on sys-
10
t = 12000 s t = 12000 s tems containing dissolved oxygen, a similar approach will be applied
1.0 8
for more complex systems containing dissolved CO2 and H2 S. Some
itotal / Am-2

6
0.5 unique features of the present work are that concentrations of species
4
0.0
were obtained from solution of conservation equations, conditions
2
for localized corrosion and galvanic effects were seen to result from
0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 the formation of precipitates, and the locations of regions associ-
r / mm r / mm
(a) (b) ated with anodic and cathodic behaviors were not assumed a priori,
but were instead obtained as a consequence of the calculations. The
Figure 11. Radial distributions calculated at the open-circuit or corrosion active-passive transition was treated as a result of deposition of films
potential for the Deposit model with time as a parameter: a) total current composing of corrosion products. The precipitate formation model
density and b) fractional surface coverage. can illustrate the importance of each property in the system, i.e., the

Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 C328 Journal of The Electrochemical Society, 161 (6) C321-C329 (2014)

0.12 i p,Fe current density for the corrosion reaction 3 in the pres-
rs / r b = 1 ence of surface films, A/m2 , see equation 5
i po,Fe potential-independent corrosion current density for the
0.10 rs / r b = 2 passivated electrode covered with a protective deposit,
rs / r b = 4 A/m2 , see equation 23
0.08 i O2 current density for oxygen reduction, A/m2 , see
reaction 1
kb backward rate constant for a generic homogeneous reac-
/V

0.06 tion such as reaction 7 with units dependant on reaction

stoichiometry
0

kf forward rate constant for a generic homogeneous reac-

0.04
tion such as reaction 7 with units dependant on reaction
stoichiometry
0.02 K 0,Fe rate constant for the corrosion reaction, A/m2 , see equa-
tion 4
K sp,Fe(OH)2 solubility product for formation of Fe(OH)2 precipitate,
0.00 mol2 /m6 , see reaction 10
0.0 0.2 0.4 0.6 0.8 1.0
K po,Fe rate constant for the corrosion reaction at the metal-
r / ro deposit interface, A/m2 , see equation 22
K 0,O2 rate constant for oxygen reduction, A m/mol, see equa-
Figure 13. Potential distributions along electrode surface calculated at the tion 2
open-circuit or corrosion potential at the time of maximum potential variation K oxide,O2 rate constant for the oxygen reduction reaction at the
with rs /rb as a parameter. deposit-electrolyte interface, A/m2 , see equation 24
Kw dissociation constant for water, mol2 /m6 , see reaction 8
MFe(OH)2 molecular weight of Fe(OH)2 , g/mol
protectiveness of film, the solubility product, and the ratio of the Ni flux density of ionic species, mol/m2 s
covered and uncovered region. R universal gas constant, 8.3143 J/mol K
Differential concentration cells were shown to be formed in both r radial position in a spherical coordinate system, m
the Evans drop model and the model of a deposit in a pipe with Ri rate of homogeneous production of species i, mol/m3 s
turbulent flow. The differences of the diffusion behaviors in adjacent T temperature, K
areas created two distinct environments and caused galvanic coupling. t time, s
UFe(OH)3 equilibrium potential for reaction 25, V, see equation 26
Acknowledgments Vmesh volume of the finite element mesh directly above the
electrode surface, m3
Financial support from BP America is gratefully acknowledged. z+ dimensionless
√ position away from the wall in turbulent
flow,z + = τ0 ρ/ν
zi charge number for species i, m2 /s
List of Symbols
A1 constant used in equation 17, A1 = 1.0972 × 10−4 Greek
A2 constant used in equation 17, A2 = 3.295 × 10−6
βFe Tafel slope for the corrosion reaction, V/decade, see
Aelectrode electrode area, m2
equation 4
AFe(OH)2 apparent area of a monolayer deposition of Fe(OH)2 ,
βFe(OH)3 Tafel slope for production of ferric hydroxide, reaction
m2
25, A/m2 , see equation 26
Amesh area of the finite element mesh directly above the elec-
βO2 Tafel slope for oxygen reduction, V/decade, see equa-
trode surface, m2
tion 2
ci concentration of species i, mol/m3
γ fractional coverage of the electrode surface with corro-
cO2 oxygen concentration, mol/m3 , see equation 2
sion products used to weight the contributions of i k,Fe
cp concentration of species participating as products in
and i p,Fe
a generic homogeneous reaction such as reaction 7,
θ angular position in a spherical coordinate system
mol/m3
θs thickness of a monolayer, m
cr concentration of species participating as reactants in
κ electrolyte conductivity, S/m
a generic homogeneous reaction such as reaction 7,
ν kinematic viscosity, m2 /s
mol/m3
ρ electrolyte density, kg/m3
Di diffusion coefficient for species i, m2 /s
ρs density of precipitates, kg/m3
Di,bulk effective diffusion coefficient for species i within region
τ0 wall shear stress, Pa
of turbulent flow, m2 /s, see equation 16
φ electrolyte potential, V
Di,covered diffusion coefficient for species i within deposit, m2 /s,
φ0 potential of the electrolyte outside the diffuse part of the
see equation 15
double layer, V
Di(t) eddy diffusivity for species i within region of turbulent
φm potential of the electrode, V
flow, m2 /s, see equation 16
F Faraday’s constant, 96,487 C/equiv.
i current density, A/m2
Subscripts
i Fe current density for the net corrosion reaction 3, A/m2 ,
see equation 5 Fe2+ pertaining to ferrous ions
i k,Fe current density for the corrosion reaction 3 in the ab- Fe(OH)2 pertaining to ferrous hydroxide precipitate
sence of surface films, A/m2 , see equation 4 Fe(OH)3 pertaining to ferric hydroxide precipitate
i Fe(OH)3 current density for production of ferric hydroxide, reac- H+ pertaining to hydrogen ions
tion 25, A/m2 , see equation 26 OH− pertaining to hydroxide ions

Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 161 (6) C321-C329 (2014)
References
1. R. Nyborg, in Corrosion/2002, 02-233 (Houston, TX: NACE International, 2002),
1–16.
2. C. de Waard and D. E. Milliams, Corrosion, 31, 177 (1975).
3. L. G. S. Gray, B. G. Anderson, M. J. Danyxh, and P. G. Tremaine, in Corrosion/1989,
89-464 (Houston, TX: NACE International, 1989), 1–19.
4. S. Nešić, M. Nordsveen, R. Nyborg, and A. Stangeland, Corrosion, 59, 489 (2003).
5. S. Nešić, J. Postlethwaite, and S. Olsen, Corrosion, 52, 280 (1996).
6. A. Anderko and P. J. Shuler, in Corrosion/1998, 98-064 (Houston, TX: NACE Inter-
national, 1998), 1–15.
7. U. R. Evans, The Corrosion and Oxidation of Metals: Scientific Principles and Prac-
tical Applications (London: E. Arnold Ltd., 1960).
8. E. McCafferty, Introduction to Corrosion Science (New York, NY: Springer-Verlag
New York Inc., 2010).
9. I. Cole, R. Holgate, P. Kao, and W. Ganther, Corrosion Science, 37, 455 (1995).
10. S. B. Lyon, C. W. Wong, and P. Ajiboye, in Atmospheric Corrosion, W. W. Kirk and
H. H. Lawson, editors, STP1239 (West Conshohocken, PA: ASTM International,
1995), 26–37.
11. I. S. Cole, D. Lau, and D. A. Paterson, Corrosion Engineering, Science and Technol-
ogy, 39, 209 (2004).
12. I. Muto and K. Sugimoto, Zairyo to Kankyo/ Corrosion Engineering, 47, 519 (1998).
Accelerated corrosion test; Dew point.
Downloaded on 2019-04-02 to IP 212.118.132.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
C329
13. M. S. Venkatraman, I. S. Cole, D. R. Gunasegaram, and B. Emmanuel, Materials
Science Forum, 654-656, 1650 (2010).
14. J. O. Bockris, D. Drazic, and A. R. Despic, Electrochim. Acta, 4, 325 (1961).
15. S. Nešić, N. Thevenot, J. Crolet, and D. Drazic, in Corrosion/1996, 96-003 (Houston,
TX: NACE International, 1996), 1–23.
16. J. S. Newman and K. E. Thomas-Alyea, Electrochemical Systems (John Wiley &
Sons, Inc., 2004).
17. A. C. West, J. Electrochem. Soc., 140, 403 (1993).
18. M. Nordsveen, S. Nešić, R. Nyborg, and A. Stangeland, Corrosion, 59, 443
(2003).
19. S. Nešić and K.-L. J. Lee, Corrosion, 59, 616 (2003).
20. G. Schmitt and M. Mueller, in Corrosion/1999, 99044 (San Antonio, TX: NACE
International, 1999).
21. K. R. Applin and N. Zhao, Ground Water, 27, 168 (1989).
22. C. Picioreanu and M. C. M. van Loosdrecht, Journal of the Electrochemical Society,
149, B211 (2002).
23. G. Meruane, C. Salhe, J. Wiertz, and T. Vargas, Biotechnology and Bioengineering,
80, 280 (2002).
24. F. LaQue, Corrosion, 13, 33 (1957).
25. F. L. LaQue, J. Electrochem. Soc., 116, 73C (1969).
26. W. G. Whitman, R. P. Russell, and G. H. B. Davis, J. Am. Chem. Soc., 47, 70
(1925).
27. D. L. Leussing and I. M. Kolthoff, J. Am. Chem. Soc., 75, 2476 (1953).