Você está na página 1de 9

Electrochimica Acta 77 (2012) 339–347

Contents lists available at SciVerse ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

A transient multi-ion transport model for galvanized steel corrosion protection


V. Topa a , A.S. Demeter a,∗ , L. Hotoiu b , D. Deconinck b , J. Deconinck b
a
Technical University of Cluj Napoca, Faculty of Electrical Engineering, 26–28 George Baritiu Street, 400027 Cluj-Napoca, Romania
b
Vrije Universiteit Brussel, Researc Group Electrochemical and Surface Engineering, Pleinlaan 2, 1050 Brussels, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: A transient multi-ions transport and reaction model for galvanized steel corrosion in a 10 mol m−3 NaCl
Received 30 December 2011 is proposed. The pH and the current density evolutions are simulated as a function of the immersion
Received in revised form 8 June 2012 time and the results are compared with measurements. The simulations are two-dimensional and for a
Accepted 8 June 2012
time up to 20 000 s. When considering no flow in the 10 mm electrolyte layer above the electrodes, it was
Available online 15 June 2012
impossible to find agreement with the measured pH profile that showed a diffusion layer. The addition
of natural convection, which is mimicked by a 0.1 mm/s fluid flow towards the electrodes, leads to a
Keywords:
much better qualitative agreement of the measured and simulated pH profile. Even in the presence of
Multi-ion transport model
Zinc corrosion
natural convection, the model predicts a low pH area only above the zinc and the zinc/steel interface. The
FEM approach insertion of two insulating pieces, representing the passivation due to zinc hydroxide deposit extends
the low pH area up to 4 mm beyond the edges of the anode and shows that the passivating layer can still
grow further. The thus obtained plateau of low pH around the anode matches closely the measurements,
while the predicted high pH matches the measurements also far away from the anode.
© 2012 Elsevier Ltd. All rights reserved.

1. Introduction zinc-steel model bi-electrodes [3,4,9,11,13,20–24] and industrial


samples in model corrosive solutions [6,10,11,13,16–18].
Steel sheets galvanized by hot-dip or by an electrochemical Corrosion products, which can be found as precipitates of dis-
process, to which an organic coating is added, are common raw solving metal ions onto the metal, often play a key role in the
material for metallic constructions. The defects created in the corrosion protection mechanism. Ogle et al. [23] studied the influ-
protective coating by cuttings before use (e.g. cut edges), or by acci- ence of precipitated zinc based corrosion products on the local
dental scratches, create a situation where both the steel substrate pH and current density distribution. They observed that above
and the protective zinc coating are exposed to the environment. the areas covered by corrosion products, identified as ZnO and
Since the protection offered by the organic coating is no longer 3Zn(OH)2 ·2ZnCO3 , the pH remains near-neutral and the current
available, the corrosion protection needs to perform a “self-healing” density tends to stay anodic (positive) even under slight cathodic
mechanism, involving the deposition of passivating zinc-based cor- polarization. A larger number of zinc compounds, arranged in suc-
rosion products. In what follows we focus on this kind of cut-edge cessive layers, have been identified on the surface of the zinc
corrosion. [25,26], but not all of them are present in marine exposures.
The cut-edge corrosion phenomenon consists of a series of elec- Moreover, in sulfate containing electrolytes, like NH4 SO4 , no pre-
trochemical and chemical reactions, often confined in less than a cipitation is observed [6].
1-mm range around the cut edge. In order to properly investigate At the same time and in support of understanding and predict-
these reactions, spatially resolved electrochemical techniques have ing the behavior of the studied corrosion systems, a number of
been applied. Scanning Vibrating Electrode Technique (SVET) [1,2] mainly stationary mechanistic models have been developed and
was used to measure the spatial distribution of the current densi- validated against experimental data [10,15,18,27,28]. A wide range
ties in the solution above a corroding sample [3–18]. pH microscopy of corrosion models is presented in the literature. The potential
[19] was used to map the local pH variations above a cut edge sam- model (PM) solves the Laplace equation with proper boundary
ple [6,9,20,21] and the distribution of the Zn2+ ions concentration conditions. Analytical solutions are evidently restricted to simple
was investigated by means of ion selective microelectrodes [20,21] geometries [29]. Numerical solutions of the PM deal with linear [30]
as well as by noninvasive methods like optical visualization [22]. or nonlinear [28] boundary conditions and have for instance been
The above mentioned techniques have been used to investigate used to investigate the effect of the electrolyte thickness on the
distribution of the corrosion current density. A more elaborated,
but one-dimensional model, combines the PM with steady state
∗ Corresponding author. oxygen diffusion to simulate mixed-potential uniform corrosion
E-mail address: andrei.demeter@et.utcluj.ro (A.S. Demeter). [31].

0013-4686/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.electacta.2012.06.021
340 V. Topa et al. / Electrochimica Acta 77 (2012) 339–347

Fig. 1. Geometry (dimensions in mm) and mesh for the multi-ions simulations in the cases where the zinc and the iron are in direct contact, as in the case of a fresh cut. The
vertical boundaries are insulators.

More complex models consider Multi-Ion Transport and homo- This choice is fully acceptable on the considered scale. Mod-
geneous Reactions in the electrolytic solution (MITRe Models) and els for more concentrated solutions considering ion interactions
have been proposed e.g. for pitting corrosion [32,33]. A similar, but (Mean Spherical Approximation [37,38]; additional potential equa-
more complete model was proposed to describe the cut edge corro- tion [39]) are not considered.
sion of galvanized sheet steel. A four-stage mechanism of cathodic The local current density is given by the sum of the ionic fluxes
inhibition in NaCl solutions is proposed [16,18]. Local polarization
curves, along with measurements of the current density distri- 
I

bution and the pH, surface analysis and stationary numerical J=F ni Ni . (4)
simulations have been used to confirm the proposed mechanism. i=0

Although this numerical model takes into account the most impor-
tant chemical species and homogenous reactions present in the 2.2. Geometry and meshing
aqueous environment, it cannot predict the evolution of the system
in time since it is solved stationary. Furthermore, the influence of The setup consists of a 3 mm wide strip of Zn deposited in the
the precipitated species is considered only in the post-processing. middle of a 39 mm wide coupon of iron (steel), submerged under a
We present a time dependent model for zinc-based precipitates 10 mm 10 mol m−3 NaCl solution.
on iron corrosion inhibition. Two different situations are considered. In the first case, the
The model considers the current density distribution and the initial state of the corrosion cell is studied: the zinc strip and the
concentration profiles of all involved species and the homogenous surrounding steel (iron) are in direct contact. The two-dimensional
reactions in between them at each time step. The role of the pre- geometry with the boundaries and the mesh used to simulate this
cipitated corrosion products on the electrode reaction rate is not case are shown in Fig. 1. In the second case, inserting two seg-
considered. Also the electrocatalytic properties of the precipitate to ments of a 6.5 mm wide insulating pieces in between the anode and
support the Oxygen Reduction Reaction (ORR) have been neglected cathodes mimics the inhibiting effect of the zinc-based precipitates
in the model [34,35]. The position and the passivating effect of the deposited in the vicinity of the anode. As one may observe in Fig. 2,
precipitated zinc-based products are simulated by insertion of an the insertion of the insulating pieces does not change the relative
insulating segment in between the anode and cathodes. position of the electrodes, nor the mesh. Using the same electro-
chemical system and identical boundary conditions, the result will
2. Modeling be influenced only by the presence of the insulators.
For the Finite Element Method, this geometry was discretized
2.1. Governing equations in linear triangular elements with a refinement towards the elec-
trodes down to 0.5 ␮m, resulting in a mesh of 5508 elements and
Due to the complexity of the corrosion process that involves 2560 nodes.
mainly mass transport because of diffusion, migration, and convec-
tion, along with homogenous reactions, the full multi-ion transport 2.3. Boundary conditions
and reaction model (MITReM) is considered. The dilute solutions
theory [36] is applied and the flux Ni of each species is given by At the liquid–air interface, also the inlet, constant concentra-
tions of oxygen and carbon dioxide are imposed and equal to those
Ni = vci − Di ∇ ci − ni ui Fci ∇ U, (1) reported in the literature for water in direct contact with atmo-
with ci , Di , ni and ui , respectively, the concentration, diffusion con- spheric air in normal conditions of temperature and pressure [40].
stant, charge and mechanical mobility of species i. F is Faraday’s At insulators, the normal gradients of electric potential and all
constant, U the solution potential and v the velocity of the solvent. species fluxes are zero.
The mass balance for any species is written as At electrodes, the normal fluxes are given by the rates of the
electrochemical reactions
∂ci
= −∇ · Ni + Ri , (2) 
E
∂t
Ni = ve sie , (5)
where ∇ · Ni is the divergence of the flux and Ri the produc-
e=1
tion/destruction rate of species i due to homogenous reactions.
The bulk solution is considered electro-neutral, yielding with ve the rate of the electrochemical reaction, and sie the stoi-
chiometric coefficient of the species i in reaction e.

I
At the anode Zn is dissolved
ci ni = 0. (3)
i=1 Zn  Zn2+ + 2e− . (6)
V. Topa et al. / Electrochimica Acta 77 (2012) 339–347 341

Fig. 2. Geometry (dimensions in mm) and mesh identical to the first case. A piece of insulator of 6.5 mm is placed in between the anode and cathodes, representing the status
of dense corrosion products that are precipitated around the anode.

A linear relation approximates this electrochemical reaction grade NaCl. The polarization curves were recorded in quiescent
rate: electrolyte, exposed to air, at room temperature, 23 ± 1 ◦ C.
a(V − U) + b
ve = , (7) 2.4. The electrolyte solution
nF
with n = 2, a = 22.982 (S/m2 ) and b = 26.255 (A/m2 ), the parameters The electrolyte solution, as used in [21], contains 10 mol m−3
of the reaction. This linearization is justified as a linear fit represents NaCl, is at 25 ◦ C and pH = 5.4, and the soluble zinc corrosion prod-
fairly well the measured polarization curve in the working range ucts are similar with [18]. The diffusivity and charge number of all
between 2 and 10 A/m2 (see Fig. 3). species considered are given in Table 1.
On the cathode, the oxygen reduction reaction, Eq. (8), is taking The equilibrium solution composition was computed, prior to
place the simulation, using the homogenous reactions equilibrium con-
kred stants and the initial values for the concentrations.
4OH− ←−2H2 O + O2 + 4e− . (8) The result of this computation, also given in Table 1, is used to
initialize the concentrations of all species in solution.
The rate of this reaction is described by the Tafel kinetic [36]
In between the ionic species present in the solution, we con-
 a nF 
red sidered eight homogenous reactions given in Table 2, leading to
ve = −kred cox exp − (V − U) (9)
RT soluble reaction products. The precipitation reactions and the solid
corrosion products are not considered at this point.
with ˛red = 0.096 the apparent charge transfer coefficients for
Due to the fact that, in the literature, usually the equilibrium
reduction, cox the concentration of the dissolved O2 , kred = 6 × 10−9
constants K are given, and only seldom forward and backward
(m/s) the reaction rate constants for reduction, n the number of
reaction rate constants are published, the values shown in Table 2
electrons exchanged in reaction, R the universal gas constant, T the
without a literature reference were fitted to match the measure-
temperature, in K, U the solution potential, in V, V the electrode
ments.
potential, in V, ve the reaction rate, in mol m−3 s−1 .
The kinetic parameters for the electrode reactions have been
fitted on polarization curves recorded with a Metrohm Autolab 2.5. Numerical aspects
PGSTAT302N Potentiostat/Galvanostat using a platinum counter
electrode and saturated calomel electrode as reference. The work- Eq. (1) combined with Eq. (2) and the Nernst–Einstein relation
ing electrodes (Fe = 1 cm2 , Zn = 0.785 cm2 ) were embedded in epoxy (Di = RTui [36]) yields a mass balance equation for each of the chem-
polymer and ground down to 1200 SIC paper. The 10 mol m−3 NaCl ical species [41]
solution was prepared from demineralized water and analytical ∂ci nDF
= Di ∇ · (∇ ci ) − v · ∇ ci + i i ∇ · (ci ∇ U) (10)
∂t RT
Together with the electro-neutrality condition given in Eq. (3),
a closed system of equations is formed in the 13 unknown concen-
trations ci given in Table 2 and the potential U in the electrolyte.
This system is discretized using the Finite Element Method (FEM).
The unknowns, in total 35 840, are placed in the vertices of
the triangles of the computational domain (cf. Figs. 1 and 2) and
the variation of the unknowns is linear within each triangle. The
final set of non-linear equations is solved using a Newton–Raphson
iteration procedure and the time integration was handled with a
Crank–Nicholson second order scheme.

2.6. Flow field

In order to mimic the natural convection caused by possible tem-


perature and/or concentration variations in the fluid bulk, a small
uniform descending flow of 0.1 mm/s (see below) is imposed at the
top of the geometry (the inlet) and a zero pressure is imposed at
Fig. 3. Anodic polarization. A linear fit (solid line) coincides with the measured the side walls, being two outlets. The bottom of the geometry is a
polarization curve (dashed line) in the range between 2 and 10 A/m2 . non-slip wall.
342 V. Topa et al. / Electrochimica Acta 77 (2012) 339–347

Table 1
The ion composition of the electrolyte, with their charge number, diffusivity and initial equilibrium concentration.

Chemical species Charge number Diffusivity 109 m2 s−1 Initial concentration, mol m−3 Reference
+
Na +1 1.33 10.00 [21]
Cl− −1 2.03 10.00 [21]
H+ +1 9.31 2.71 × 10−3 [21]
OH− −1 5.30 3.50 × 10−6 [21]
O2 0 2.40 0.26 [18]
CO2 0 1.91 1.31 × 10−2 [23]
H2 CO3 0 1.60 3.35 × 10−5 [23]
HCO3 − −1 1.10 2.71 × 10−3 [21]
CO3 2− −2 0.92 6.41 × 10−8 [23]
Zn2+ +2 0.71 9.88 × 10−12 [21]
ZnOH+ +1 1.00 1.15 × 10−13
Zn(OH)2 0 10−11 5.33 × 10−17
Zn(OH)3 − −1 0.10 3.13 × 10−22
ZnCO3 0 0.10 1.27 × 10−16

Table 2
Homogenous reactions and their kinetic parameters. The units are given as footnotes below the table.

Reaction K kf kb Reference

H2 O ↔ H+ + HO− 1.4 × 10−08 a 7.08 × 10−4 b 7.85 × 104 c [24]


H2 O + CO2 ↔ H2 CO3 2.85 × 10−3 2.6 × 10−2 d 9.125d [25]
H2 CO3 ↔ H+ + HCO3 − 1.98 × 10−4 e 7.14 × 106 d 3.6 × 107 c [23]
HCO3 − ↔ H+ + CO3 2− 6.41 × 10−8 e 109 d 1.56 × 1019 c [23]
Zn2+ + H2 O ↔ ZnOH+ + H+ 3.16 × 10−5 e 3.16 × 10−2 d 103 c
ZnOH+ + H2 O ↔ Zn(OH)2 + H+ 1.25 × 10−6 e 1.25 × 10−4 d 102 c
Zn(OH)2 + H2 O ↔ Zn(OH)3 + + H+ 1.59 × 10−8 e 1.59 × 10−10 d 10−2 c
Zn2+ + CO3 2− ↔ ZnCO3 2− 200f 2 × 10−7 c 10−9 d
a
mol2 m−6 .
b
mol m−3 s−1 .
c
mol−1 m3 s−1 .
d
s−1 .
e
mol m−3 .
f
mol−1 m3 .

Table 3 3. Results and discussions


The assumptions made for the four cases under consideration.

Case Electrode Homogenous Insulators Flow In order to evaluate the effect of the insulators and the flow field,
reactions reactions present four cases were investigated (see Table 3). In all cases the full model
Case 1 (C1) Yes Yes No No is solved in time.
Case 2 (C2) Yes Yes Yes No In the first two cases, as a first approximation for the quiescent
Case 3 (C3) Yes Yes No Yes liquid in the real experimental setup, no convection was consid-
Case 4 (C4) Yes Yes Yes Yes
ered. For the second two cases we assumed that the boundary layer
observed above the cathodes in the measurements of Tada et al. [21]
is the result of a small convective flow, caused by either tempera-
ture and/or concentration gradients in the bulk of the solution, i.e.
the natural convection. This assumption is based on the fact that
The flow field is obtained by solving the laminar Navier–Stokes for this geometry the Grashof number (Gr) is equal to 3340 at a
and the continuity equations on the same mesh. The velocity profile temperature difference of 0.1 ◦ C in between the top and the bot-
obtained in this manner is given in Fig. 4 and is used in all electro- tom of the geometry. This value is higher than the critical value of
chemical simulations. Diffusion layers of about 3 mm are formed. Gr = 1000 for the free surface case [42] such that natural convection
This approach assumes that the flow is not changing in time and is very probable.
also avoids entering into the detail and the quantification of the On the other hand, a 0.1 ◦ C temperature gradient over a 10 mm
flow inducing force (see further). thick layer of water has a negligible effect on transport properties

Fig. 4. The flow field and the associated boundary conditions; the dimensions are in mm.
V. Topa et al. / Electrochimica Acta 77 (2012) 339–347 343

Fig. 5. Simulated pH distribution above the electrodes at 3600 s for case 1. The position of the electrodes is marked as in Fig. 1; dimensions in mm. The solid black lines are
lines of constant pH with their values.

Fig. 6. Simulated pH distribution above electrodes and insulators at 3600 s for case 2. The position of the electrodes and insulators is marked as in Fig. 2; dimensions in mm.

Fig. 7. Simulated pH distribution distribution above electrodes at 3600 s for case 3. The position of the electrodes is marked as in Fig. 1; dimensions in mm.

and reaction kinetics. Therefore, the use of an isothermal model is The transient simulations for all four cases ran up to 20 000 s.
still valid. That is the time that was necessary in [21] to completely consume
The inlet velocity was chosen in a way to keep the Gr/Re2 ratio the zinc strip such that the galvanic protection ceased. In what
below 1 such that the natural convection has the major contribu- follows, the simulation results will be presented and compared
tion to the convective mass transport [42]. After some trials, an with the measurements provided in [21].
inlet velocity of 0.1 mm/s was taken. Lower velocities result into
larger diffusion layer thickness, whereas higher values induce a 3.1. Spatial pH distribution
too high removal of reaction products. We performed also simu-
lations with flow from bottom to top. In this case the mixing of The color plots presented in Figs. 5–8 give a complete view of the
the electrolyte leads to a higher pH above the anode, which is not spatial pH distributions above the electrodes and insulators after
measured. 3600 s, for each of the four cases.

Fig. 8. Simulated pH distribution distribution above electrodes and insulators at 3600 s for case 4. The position of the electrodes and insulators is as in Fig. 2; dimensions in
mm.
344 V. Topa et al. / Electrochimica Acta 77 (2012) 339–347

Fig. 9. Comparison between the measured (thick, solid line) and simulated vertical pH profiles after 3600 s exposure to the corrosive environment at three different positions
along the x axis.

As one may observe in Figs. 5 and 6, for cases 1 and 2, where no A series of 100 equidistant points have been extracted at 3600 s
convection is considered and the steady state is not yet reached, from the color plots given in Figs. 5–8 along the vertical for three
the OH− produced by the cathodic reaction diffuses away from the different positions of the x-axis: 0 mm, 4 mm and 13.5 mm (see thin
cathodes, creating a smooth pH gradient in the vertical direction. At vertical lines plotted in Fig. 5). In Fig. 9 the pH values in these points
the boundary between the anodic and cathodic zones, the steeper have been plotted against the distance from the electrode together
gradient is caused by the H+ produced by the Zn2+ hydrolysis reac- with the measured values [21].
tions, which counterbalances the pH increase. When the electrodes As observed in Fig. 9, in case 1 the vertical pH profile above
are in direct contact (case 1, Fig. 5), the pH shift takes place at the the middle of the anode (x = 0) is over predicted with more than 1
interface between the anodic and cathodic zones, but above the pH unit. The increase in pH towards the top of the cell is caused
cathode. by the diffusion fronts of OH− ions produced at cathodes, which
In Fig. 6, one can see that the presence of the insulating zones tend to merge above the anode. The insertion of insulator pieces
pushes the pH shift above the insulator, closer to the cathode. At in case 2 shows to give the best agreement with the measurement
the same time, the pH above the anode is with one unit lower than profile [11]. The addition of flow, cases 3 and 4, slightly decreases
in case 1. This is caused by the higher diffusion rate of the H+ ions, the predicted pH from 5.6 down to 5.2 at the electrode surface.
which can travel further than the OH− ions in the same time. Except near the top of the cell, at 4 mm from the center of the
For both cases 1 and 2, at the liquid/air interface, the pH is anode, case 1 displays pH values of typically 10.4 along most of
slightly decreased by the H+ ions produced by the dissociation of the height of the cell. This is not observed in the measurements
the carbonic acid and bicarbonate ions. The “islands” of low pH [21] with pH values in between 5 and 6, typical for the anodic zone,
above the anode are the remains of the initial low pH of the bulk although this is above the steel surface, which is expected to behave
electrolyte, which is slowly alkalinized by the OH− production from cathodically. The pH decreases near the top of the cell due to the
cathodes. H+ ion produced by the homogenous reactions of carbonic acid and
In the cases 3 and 4, given in Figs. 6 and 8 the pH increase above bicarbonate dissociation, which is expected. The insertion of the
the cathodes is confined by the flow to less than 3 mm in the ver- insulators, in case 2, decreases the pH value close to the electrode
tical direction. In the vicinity of the anodes, a slight pH decrease is surface with only 3 pH units, not enough to match the measure-
observed, while the bulk solution remains at the initial pH. ments. The pH increase around the middle of the cell height is

Fig. 10. Measured end simulated horizontal pH profiles at 0.3 mm from the electrodes surface after 4500 s exposure to the corrosive environment.
V. Topa et al. / Electrochimica Acta 77 (2012) 339–347 345

Fig. 12. The mean cathodic current density (in absolute values) evolution as a func-
tion of time.

Fig. 11. Evolution of the pH in the middle of the zinc strip (x = 0 mm) at different
heights in electrolyte as a function of the immersion time for case 4. As an example the pH values in case 4 were extracted at several
heights above the center of the zinc surface and plotted against
immersion time in Fig. 11.
caused by diffusion of the OH− ions produced by the cathodic reac- A fast pH decrease is observed in the first 100 s for all 5 moni-
tion. At the top of the cell, as in case 1, the homogenous reactions tored positions (z = 0, 0.1, 0.5, 1, and 2 mm). This fast acidification
decrease the pH again. The addition of flow in case 3 decreases the is caused by the rapid diffusion of the H+ ions, produced at the
pH at the cathode surface down to 8.4. This value decreases linearly zinc surface by the hydrolysis reactions. The OH− ions produced on
to 6.9 in the first 1 mm, and then decreases rapidly to the bulk value. cathodes are mostly consumed by the Zn2+ hydrolysis reactions, so
The insertion of the insulators, case 4, produces pH values that are they do not reach the vicinity of the anode to compensate for the
in agreement with [21] along the entire height of the cell. This is a local acidification.
strong indication that the Zn-based corrosion products precipitate The pH does not increase after a certain amount of time
on the surface of the cathode and inhibit the cathodic reaction, as (15 000–20 000 s) as is observed in [21] because our model does
proposed in [18]. not account for the consumption of the Zn anode and is always
Further away from the anode, at x = 13.5 mm, the model pro- present. In reality the anode gets consumed and the direct corrosion
duces results in agreement with [21] at the cathode surface, for all protection ceases.
the four cases. Away from the surface of the cathode, the agree-
ment with the measurements is broken in cases 1 and 2 where the
3.3. The current density evolution
pH drop is only due to the homogenous reactions at the top. Cases
3 and 4 follow the trend of the measurements and reach the bulk
Transient simulations predict also the evolution of current den-
value at about the same height as the measurements.
sity along the electrodes. Unfortunately no currents were reported
For all four cases the simulated results at 4500 s have been
by Tada. Fig. 12 shows the evolution of the mean cathodic cur-
probed at 0.3 mm from the surface of the electrodes, along the hor-
rent density i(t) (A/m2 ), being the integral of the calculated current
izontal line indicated in Fig. 5, and plotted together with the results
density over the electrode divided by its length.
of the measurements from [21] in Fig. 10. We observe that the four
One may observe that the initial current is high, between −0.7
cases tend to agree with the measurements, far from the anode.
and −1 A/m2 in all cases. This is due to the initial bulk concentration
The situation is, however, clearly different above the anode. Cases
of the dissolved O2 at the surface of the cathodes. After this initial
1 and 2 display a “well” shape above the anode, with a fast increase
consumption, the oxygen needs to be transported to the surface,
towards the maximum value, respectively, at about 3.5 and 4.5 mm.
which causes a rapid drop of the current density in all four cases.
With the lowest simulated pH value of 7 above the center of the
This drop is less pronounced when flow is present (cases 3 and 4)
anode, case 1 gives the worst match with the measurements. The
and fully diffusion controlled in the cases 1 and 2.
insertion of insulators in case 2 results into a minimal pH value
The insulating segments in case 2 and case 4 introduce a small
that matches the measurements, and widens the pH well, but not
supplementary electrical resistance caused by the longer path
enough to match the recorded plateau [21]. In case 3 one can best
the ions have to travel between electrodes. This effect is more
observe the limited buffering effect of the Zn2+ ions, mentioned in
[10]. The right minimum pH value is observed above the center of
the anode. The pH increases rapidly towards the edges of the anode
due to the flux of OH− ions produced here by the cathodic reaction
and is followed by a lower increase rate for about another 4 mm.
Case 4 shows the best match with the measurements along the
whole profile. It has a plateau of low pH and slightly under predicts
the measurements for about 5 mm from the edges of the anode, in
good agreement with [21]. The zone where the pH rises is situated
between the left and right measurements because of the equally
chosen insulating pieces. By taking different insulator lengths one
could improve the agreement with the measurements.

3.2. Evolution of pH in time

Transient simulations show the evolution in time of the corro- Fig. 13. Current density distribution along the iron cathode at the right after 10 000 s
sion process. exposure time.
346 V. Topa et al. / Electrochimica Acta 77 (2012) 339–347

• A set of well-chosen forward and backward reaction rate con-


stants for the homogenous reactions was needed to reproduce
adequately the results of Tada et al. at the two different times.

It is believed that these aspects are valid for many corrosion


systems that are measured in presumed absence of flow.
We finally show that in this configuration, the average corrosion
current density is dominated by the concentration of the dissolved
oxygen at the surface of the cathodes, and that the Iron electrodes
are cathodically protected as long as a zinc electrode is present.
Further developments of this model shall account for the evolu-
tion in time of the cathodic activity, due to the local precipitation
of the Zn-based corrosion products. Also the total Zn consumption
needs to be tracked in time until full consumption.
Fig. 14. The dissolved oxygen concentration at the surface of the iron cathode at
the right after 10 000 s exposure time.
Acknowledgments

pronounced for case 1 and case 2, in which also no real stationary The authors would like to thank A.C. Bastos and M. Zheludke-
regime is observed, even after 20 000 s. vich from University of Aveiro, CICECO, Department of Ceramics
Figs. 13 and 14 show, respectively, the current density distri- and Glass Engineering, and S. Lamaka from Instituto Superior Téc-
bution and the oxygen distribution along the surface of the iron nico, ICEMS, Department of Chemical Engineering for providing the
electrode at the right and at t = 10 000 s. One observes that the cur- polarization curves used for the electrochemical reactions and their
rent density is negative along the entire electrode, indicating that valuable input concerning the homogenous reactions.
it is protected cathodically along the whole surface. Fig. 14 shows The present work was supported by the project “Improve-
clearly that the oxygen concentration drops to very low values such ment of the doctoral studies quality in engineering science for
that one can conclude that the total current is controlled by the development of the knowledge based society – QDOC” contract
oxygen transport along the iron cathode. No. POSDRU/107/1.5/S/78534, project co-funded by the European
As expected, the highest current density is observed at the com- Social Fund through the Sectorial Operational Program Human
mon edge with the anode, in cases 1 and 3, and at the edge towards Resources 2007–2013. Additionally, it was supported by CNC-
the anode, in cases 2 and 4. The presence of the insulator leads to SIS – UEFISCSU under IDEI research program, with grant number
higher current densities on the cathode’s edge in contact with the ID2538/2008.
insulator. This is surprising but explained by the fact that the solu-
tion above the insulator is richer in dissolved oxygen and enables References
easier access of oxygen towards the cathode edge. This is seen in
Fig. 14 where the oxygen drops slower in the x-direction when the [1] L.F. Jaffe, R. Nuccitelli, Journal of Cell Biology 63 (1974) 614.
[2] C. Scheffey, Review of Scientific Instruments 59 (1988) 787.
insulator is present.
[3] S. Bohm, H.N. McMurray, S.M. Powell, D.A. Worsley, Electrochimica Acta 45
Finally, Fig. 14 shows that in the absence of flow, cases 1 and 2, (2000) 2165.
the dissolved oxygen concentration at the surface of the cathode [4] D.A. Worsley, D. Williams, J.S.G. Ling, Corrosion Science 43 (2001) 2335.
[5] A.C. Bastos, M.G. Ferreira, A.M. Simöes, Corrosion Science 48 (2006) 1500.
is depleted, the ORR is running on diffusion controlled regime. The
[6] K. Ogle, S. Morel, D. Jacquet, Journal of the Electrochemical Society 153 (2006)
addition of flow, in cases 3 and 4 leads to a 2 orders of magnitude B1.
higher oxygen concentration at the surface, which translates into a [7] A.M. Simöes, A.C. Bastos, M.G. Ferreira, Y. Gonzalez-Garcia, S. Gonzalez, R.M.
1 order of magnitude higher current density, see Fig. 13. Souto, Corrosion Science 49 (2007) 726.
[8] R.M. Souto, Y. Gonzalez-Garcia, A.C. Bastos, A.M. Simoes, Corrosion Science 47
(2007) 4568.
[9] S.V. Lamaka, O.V. Karavai, A.C. Bastos, M.L. Zheludkevich, M.G.S. Ferreira, Elec-
4. Conclusions trochemistry Communications 10 (2008) 259.
[10] F. Thébault, B. Vuillemin, R. Oltra, K. Ogle, C. Allely, Electrochimica Acta 53
(2008) 5226.
A time dependent multi-ions transport and reaction model for [11] A.M. Simöes, J. Torres, R. Picciochi, J.C.S. Fernandes, Electrochimica Acta 54
galvanized steel corrosion in NaCl solution is presented. The model (2009) 3857.
takes into account all relevant ionic species which may be present [12] K.B. Deshpande, Corrosion Science 52 (2010) 3514.
[13] A.M. Simöes, J.C.S. Fernandes, Progress in Organic Coatings 69 (2010) 219.
in this corrosion system in the pH range of 4–12 and at 25 ◦ C. The [14] M. Yan, V.J. Gelling, B.R. Hinderliter, D. Battocchi, D.E. Tallman, G.P. Bierwagen,
finite element method is used to solve this model in time during Corrosion Science 52 (2010) 2636.
20 000 s for a two-dimensional configuration that was measured by [15] K.B. Deshpande, Electrochimica Acta 56 (2011) 1737.
[16] F. Thébault, B. Vuillemin, R. Oltra, C. Allely, K. Ogle, Corrosion Science 53 (2011)
Tada et al. In order to obtain an overall agreement between mea-
201.
surements and simulations for a moderated exposure time, it was [17] A.P. Yadav, H. Katayama, K. Noda, H. Masuda, A. Nishikata, T. Tsuru, Electrochim-
necessary to take into account the following: ica Acta 52 (2007) 3121.
[18] F. Thébault, B. Vuillemin, R. Oltra, C. Allely, K. Ogle, Electrochimica Acta 56
(2011) 8347.
• Insertion of two insulating segments in between the anode and [19] E. Klusmann, J.W. Schultze, Electrochimica Acta 42 (1997) 3123.
[20] A.C. Bastos, M.G. Taryba, O.V. Karavai, M.L. Zheludkevich, S.V. Lamaka, M.G.S.
cathode is needed to retrieve the passivating effect of the precip- Ferreira, Electrochemistry Communications 12 (2010) 394.
itated zinc-based corrosion products. This assumption proves to [21] E. Tada, K. Sugawara, H. Kaneko, Electrochimica Acta 49 (2004) 1019.
be valid, since the low pH area extends over the insulator also. [22] E. Tada, H. Kaneko, Corrosion Science 52 (2010) 3421.
[23] K. Ogle, V. Baudu, L. Garrigues, X. Philippe, Journal of the Electrochemical Soci-
In absence of these insulating segments the low pH is seen only ety 147 (2000) 3654.
above the anode. [24] E. Tada, S. Satoh, H. Kaneko, Electrochimica Acta 49 (2004) 2279.
• The addition of a small vertical flow field meant to mimic a small [25] I.S. Cole, T.H. Muster, S.A. Furman, N. Wright, A. Bradbury, Journal of the Elec-
trochemical Society 155 (2008) C244.
natural convection. This flow results into the settling of an about
[26] I.S. Cole, T.H. Muster, D. Lau, N. Wright, N.S. Azmat, Journal of the Electrochem-
3 mm thick diffusion layer. ical Society 157 (2010) C213.
V. Topa et al. / Electrochimica Acta 77 (2012) 339–347 347

[27] C.R. Crowe, R.G. Kasper, Journal of the Electrochemical Society 133 (1986) 879. [37] S. Van Damme, J. Deconinck, Journal of Physical Chemistry B 111 (2007)
[28] J.-M. Lee, Electrochimica Acta 51 (2006) 3256. 5308.
[29] E. Kennard, J.T. Waber, Journal of the Electrochemical Society 117 (1970) 880. [38] S. Van Damme, J. Deconinck, Computation in Modern Science and Engineering
[30] E. McCafferty, Journal of the Electrochemical Society 124 (1977) 1869. 2 (Pts A and B) (2007) 504.
[31] M.S. Venkatraman, I.S. Cole, B. Emmanuel, Electrochimica Acta 56 (2011) 7171. [39] S. Sarkar, W. Aquino, Electrochimica Acta 56 (2011) 8969.
[32] J.R. Galvele, Journal of the Electrochemical Society 123 (1976) 464. [40] W. Stumm, J.J. Morgan, Aquatic Chemistry, 3rd ed., John Wiley & Sons, New
[33] G.R. Engelhardt, H.H. Strehblow, Journal of Electroanalytical Chemistry 365 York, 1996.
(1994) 7. [41] L. Bortels, J. Deconinck, Van den Bossche, Baart, Journal of Electroanalytical
[34] M. Stratmann, H. Streckel, Corrosion Science 30 (1990) 697. Chemistry 404 (1996) 15.
[35] M.S. Venkatraman, I.S. Cole, B. Emmanuel, Electrochimica Acta 56 (2011) 8192. [42] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomena, 2nd ed., John
[36] J. Newman, K.E. Thomas-Aleya, Electrochemical Systems, 3rd ed., John Wiley & Wiley and Sons, New York, 2002.
Sons, Inc., Hoboken, NJ, 2004.

Você também pode gostar