Corrosion Science 46 (2004) 2681–2699

www.elsevier.com/locate/corsci

Corrosion of reinforcing steel

in simulated concrete pore solutions
Effect of carbonation and chloride content
M. Moreno a, W. Morris b, M.G. Alvarez c, G.S. Duffo

b,c,*

a
Instituto de Tecnolog
ıa, Comisi
on Nacional de Energ
ıa At
omica,
Univ. Nac. Gral. San Mart
ın, Argentina
b
CONICET, Argentina
c
Comisi
on Nacional de Energ
ıa At
omica, Depto. Materiales (CAC);
Av. Gral. Paz 1499, (1650) San Mart
ın, Buenos Aires, Argentina
Received 24 March 2003; accepted 19 March 2004
Available online 12 May 2004

Abstract

The corrosion susceptibility of as-received reinforcing steel bars (rebars) in solutions sim-
ulating the pore liquid of alkaline and carbonated concrete has been studied by means of
potentiodynamic polarisation tests and polarisation resistance measurements. The effect of
different degrees of carbonation and the presence of several chloride contents in the simulated
pore solutions was investigated. Results show the beneficial effect of high alkalinity on the
localised corrosion of steel caused by chloride ions. From the results of the potentiodynamic
tests a critical chloride concentration above which pitting could take place was evaluated for
each solution. The chloride threshold values here found are of the same order than those
previously reported in the literature for film-free steel. The results obtained in solutions
simulating carbonated concrete showed that under weak carbonation conditions carbon steel
does not passivate while in the presence of high levels of carbonate and bicarbonate the
resistance to localised corrosion is improved.

2004 Elsevier Ltd. All rights reserved.

Keywords: A. Steel; B. Polarisation; C. Passivity; Pitting corrosion

*
Corresponding author. Address: Comisi
on Nacional de Energ
ıa At

omica, Depto. Materiales (CAC);
Av. Gral. Paz 1499, (1650) San Mart
ın, Buenos Aires, Argentina. Tel.: +54-11-6772-7403/4754-7403;
fax: +54-11-4754-7388.
E-mail address: duffo@cnea.gov.ar (G.S. Duff

o).

0010-938X/$ - see front matter
2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2004.03.013
2682 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

1. Introduction

The corrosion of reinforcing steel bars embedded in concrete is most frequently

the result of the chloride-induced breakdown of the passive film formed in the high
alkaline condition of the concrete. Passivity breakdown occurs when a critical
chloride concentration, which depends on the potential [1,2] is reached on the
reinforcement surface. So far, numerous investigations [3–6] have been carried out to
determine the effect of the chloride concentration on the corrosion behaviour of
reinforcing steel in alkaline solutions and simulated concrete pore solutions, with the
purpose to establish a unique critical chloride concentration for pitting initiation.
However, the chloride threshold depends on several variables and, for this reason,
values reported by different researchers showed a significant scatter. By using a
statistical interpretation of the pitting potential, Li and Sag€ues [3] found that in
simulated concrete pore solutions with pH ¼ 13.6 the critical chloride concentration
would be 0.4–0.6 M while in saturated Ca(OH)2 the chloride threshold would be
somewhat between 0.01 and 0.04 M. According to Bertolini et al. [4] the critical
chloride content of carbon steel was between 0.028 and 0.17 M at pH ¼ 12.6 and
increased to 0.28 M in a NaOH solution at pH ¼ 13.9. Hausmann [5] found that steel
immersed in a pH ¼ 13.2 NaOH solution with the addition of 0.25 M NaCl remained
in the passive state while Gouda [6] reported that the maximum sodium chloride
concentration that can be tolerated in a NaOH solution with pH ¼ 13.9 was 0.12 M.
Finally, Gouda [6] found that additions of sodium chloride up to 0.007 M did not
affect the passivity of the steel in a saturated Ca(OH)2 solution whereas Gonz
alez
Fern
andez and Andrade [7] determined that in that medium pitting of steel could
appear at a concentration of 0.0028 M.
The objective of the present work was to examine the effects of the chloride
concentration on both the polarisation characteristics and the corrosion rate of as-
received reinforcing steel immersed in solutions simulating the pore liquid of alkaline
and carbonated concrete, with particular reference to the critical chloride concen-
tration. The electrochemical processes affecting rebar corrosion in concrete were
analysed by means of potentiodynamic polarisation curves. The corrosion rate of
steel was monitored in time by means of polarisation resistance measurements. The
effect of different levels of carbonate plus bicarbonate in the simulated pore solution
was also evaluated.

2. Experimental

Specimens 35-mm long were cut from 4.5-mm-diameter carbon steel rods (AISI
1010). The chemical composition of the steel was (in wt.%): 0.134% C; 0.82% Si;
0.69% Mn; 0.43% Al; 0.18% Cu; 0.04% S; 0.016% P; 0.07% Ni; <0.1% Cr. A copper
wire lead was soldered to one end of the samples. Both ends of the samples were
covered with an epoxy resin, leaving a free surface of about 3.5 cm2 exposed. The
material was tested in the as-received condition. Before the measurements, the
specimens were ultrasonically degreased with acetone, rinsed with alcohol, and dried
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2683

Table 1
Composition and pH of the simulated concrete pore solutions tested
Solution Composition pH Chloride concentration (%)
1 0.9 M NaOH 13.9 0
2 Saturated Ca(OH)2 12.5 0
3 0.3 M NaHCO3 + 0.1 M Na2 CO3 9.0 0
4 0.015 M NaHCO3 + 0.005 M Na2 CO3 9.0 0
5 0.9 M NaOH + x% Cl
13.9 1.0; 3.0; 5.0; 10.0
6 Saturated Ca(OH)2 + x% Cl
12.5 0.02; 0.05; 0.10; 0.20
7 0.3 M NaHCO3 + 0.1 M Na2 CO3 + x% Cl
9.0 0.01; 0.05; 0.10; 0.50; 1.0
8 0.015 M NaHCO3 + 0.005 M Na2 CO3 + x% Cl
9.0 0.001; 0.005; 0.01; 0.05; 0.10

in hot air. No further surface treatment was applied to the samples in order to check
the material in a condition similar to the state in which it is going to be actually used.
Table 1 gives the composition and pH of the solutions tested. Solutions (1)–(4)
were selected to simulate the pore liquid of alkaline and carbonated concrete while
solutions (5)–(8) were used to evaluate the effect of different levels of chloride con-
tamination on the corrosion behaviour of the steel [4]. The chloride concentrations
employed for the several simulated pore solutions were selected with the purpose to
evaluate for each solution the highest chloride concentration that can be allowed
without causing pitting corrosion of the steel. The solutions were prepared with
analytical grade reagents and double distilled water. The chloride concentration was
adjusted by the addition of NaCl.
All measurements were performed at room temperature in a Pyrex glass cell with
a platinum counterelectrode. Potentials were measured through a Luggin capillary
with a saturated calomel reference electrode. All potentials are reported in the
standard hydrogen electrode scale (SHE).
Potentiodynamic polarisation curves were obtained at a scan rate of 0.5 mV/s
using a PAR 173 potentiostat with a PAR 175 voltage scan generator. The current
and potential were continuously recorded with a Houston 2000 X–Y recorder. Be-
fore starting the measurements, the samples were allowed to reach a stationary
potential by exposure to the solution for 1 h at open circuit potential. The experi-
ments were repeated 3–5 times in all the solutions tested, using a new specimen each
time.
In another set of experiments the corrosion process of the steel in solutions (1)–(8)
was followed by means of polarisation resistance (LPR) measurements. In this tests,
the potential of the samples was decreased to a potential 20 mV lower than the
corrosion potential at a 0.1 mV/s scan rate while the changes in current density were
recorded. When applying this method, the E vs. i curve was found to obey a quasi-
linear relationship and the results of independent experiments showed good repro-
ducibility. The LPR value was determined from the slope of the curve:
LPR ¼ DE=Di

The instantaneous corrosion rate of the metal can then be obtained from the linear
polarisation resistance through the Stern–Geary equation [8]:
2684 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

icorr ¼ B=LPR

being B a value that is function of the Tafel slopes. In the present work, the B values
suggested by Andrade et al. [9] for passive iron and active iron embedded in concrete
were used for the calculations (B ¼ 52 and 26 mV, respectively). Since the conduc-
tivity of the solutions tested was high, no ohmic drop compensation was necessary.
To follow the corrosion process, periodic measurements of LPR were performed
during one week. The corrosion potential of the samples (Ecorr ) was also monitored
within the same period of time. Duplicate tests were performed in each of the
solutions proven.
After the electrochemical tests, the specimens were rinsed in a 50% phosphoric
acid solution at room temperature to remove the corrosion products that covered the
metallic surface and observed with a Philips SEM 500 scanning electron microscope.

3. Results

3.1. Anodic polarisation curves

3.1.1. 0.9 M NaOH (pH ¼ 13.9) + x% Cl
(x ¼ 0 to 10) solutions

Fig. 1 shows examples of the anodic polarisation curves obtained for carbon steel
in 0.9 M NaOH solution and in the same solution with the addition of different
concentrations of chloride ion. Rebar segments immersed in 0.9 M NaOH solution
were in the passive state at the free corrosion potential. Upon increasing the po-
tential above the corrosion potential, a passive region was found where the current
density was of the order of 10
6 A/cm2 . At potential values higher than 0.5 Vshe an
increase in the current density was observed resulting from the onset of the oxygen
evolution reaction. Fig. 1 shows that the anodic polarisation curve obtained in the

-2
10
-
0% Cl
-3 -
10 1% Cl
-
3% Cl
Current Density, A/cm 2

-
-4 5% Cl
10 -
10% Cl
-5
10

-6
10

-7
10

-8
10 0.9M Na(OH)
pH 13.9
-9
10
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8
Potential, Vshe

Fig. 1. Potentiodynamic anodic polarisation curve for carbon steel in 0.9 M NaOH solution, pH ¼ 13.9,
containing different concentrations of chloride ion.
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2685

presence of 1% Cl
was similar to the one obtained in pure NaOH solution. On the
other hand, the presence of higher chloride concentrations has a substantial effect on
the anodic behaviour of the steel. The anodic polarisation curves showed a passive
zone, where the passive current density increased with increasing chloride concen-
tration and a passivity breakdown potential or pitting potential above which high
current increases were observed and pitting nucleated on the metallic surface. The
pitting potential was a function of the chloride concentration. The higher the
chloride concentration, the lower the pitting potential. Also the corrosion potential
shifted toward lower values with increasing chloride concentration. The microscopic
observation of the specimens after polarisation revealed the presence of pits in those
samples tested in solutions with a chloride content equal to or higher than 3% (Fig.
2). No pits or generalised corrosion was found at lower chloride concentrations.

3.1.2. Saturated Ca(OH)2 (pH ¼ 12.5) + x% Cl
(x ¼ 0 to 0.2) solutions

The anodic polarisation curve obtained for carbon steel in saturated Ca(OH)2
solution was similar to those found in 0.9 M Na(OH) solution (Fig. 3). A passive
region, where the current density was lower than 10
6 A/cm2 , was observed before
reaching the onset of the oxygen evolution reaction. The major difference was a
displacement of the curve towards higher potentials that was most probably asso-
ciated to the lower pH of the Ca(OH)2 solution. The addition of 0.02% Cl
has no
significant effect on the anodic behaviour of the material. It was only when the
chloride concentration was equal to or higher than 0.05% that the existence of a
passivity breakdown potential was observed. The occurrence of pitting was con-
firmed by the observation of the samples in the SEM (Fig. 4).

3.1.3. 0.3 M NaHCO3 + 0.1 M Na2 CO3 (pH ¼ 9.0) + x% Cl
(x ¼ 0 to 1.0) solutions
The rebar polarisation behaviour in the solution containing high levels of car-
bonates and bicarbonates was also found to be analogous to that observed in the

Fig. 2. Morphology of the attack observed on carbon steel after anodic polarisation in 0.9 M NaOH + 3%
Cl
solution, pH ¼ 13.9.
2686 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

-2
10
Ca(OH)2 saturated solution
-3
10 pH=12.5

2
Current Density, A/cm
-4
10

-5
10

-6
10
-
-7 0.00 % Cl
10 0.02 % Cl-
0.05 % Cl-
10
-8 0.10 % Cl-
0.20 % Cl-
-9
10
-0.2 0.0 0.2 0.4 0.6 0.8
Potential, VSHE

Fig. 3. Potentiodynamic anodic polarisation curve for carbon steel in saturated Ca(OH)2 solution,
pH ¼ 12.5, containing different concentrations of chloride ion.

Fig. 4. Morphology of the attack observed on carbon steel after anodic polarisation in saturated
Ca(OH)2 + 0.05% Cl
solution, pH ¼ 12.5.

more alkaline solutions (Fig. 5). A passive region was observed followed by a current
density increase at potentials higher than 0.8 Vshe associated with the onset of the
oxygen evolution reaction. In the presence of chloride ions passivity breakdown was
found when the chloride concentration in the solution was equal to or higher than
0.05%. The polarisation curves obtained in the solution containing 0.05% Cl
was
quite similar to the one obtained in the absence of chlorides. The current density only
showed a high increase at potentials closed to the onset of the oxygen evolution
reaction and no definite. pitting potential value could be distinguished. Nevertheless,
the microscopical examination of the samples after polarisation revealed the pres-
ence of few small pits covered with corrosion products (Fig. 6), indicating that
passivity breakdown had already occurred. The polarisation curve obtained at a
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2687

-2
10
0.1 M Na 2CO3 + 0.3 M NaHCO3
-3 pH=9.0
10

Current Density, A/cm 2

-4
10

-5
10

-6
10 -
0.00 % Cl
-
-7
0.01 % Cl
-
10 0.05 % Cl
-
0.10 % Cl
-
-8 0.50 % Cl
10 -
1.00 % Cl
-9
10
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
Potential, VSHE

Fig. 5. Potentiodynamic anodic polarisation curve for carbon steel in 0.3 M NaHCO3 + 0.1 M Na2 CO3
solution, pH ¼ 9.0, containing different concentrations of chloride ion.

Fig. 6. Morphology of the attack observed on carbon steel after anodic polarisation in 0.3 M
NaHCO3 + 0.1 M Na2 CO3 + 0.05% Cl
solution, pH ¼ 9.0.

higher chloride concentration, 0.1%, shows an abrupt increase of the current density
at a potential value of 0.45 Vshe . At higher potentials a limiting current density of the
order of 10
4 A/cm2 was found. In this potential range isolated spots of corrosion
products were seen to form on the surface of the samples. After removing those
deposits irregularly shaped small pits were detected. Finally, the polarisation curves
obtained in the solutions with chloride contents higher than 0.1% exhibited a clear
passivity to pitting transition. The formation of green corrosion products on the
metallic surface was seen above the pitting potential. Similarly to the behaviour in
more alkaline solutions, the pitting potential was a function of the chloride ion
concentration, and the higher the concentration the lower the pitting potential. Also
the corrosion potential shifted to the less noble direction with increasing chloride
2688 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

concentration. Isolated, irregularly shaped crystallographic pits, covered with large

amounts of corrosion products, were observed on the specimen surface after po-
larisation (Fig. 6).

3.1.4. 0.015 M NaHCO3 + 0.005 M Na2 CO3 (pH ¼ 9.0) + x% Cl
(x ¼ 0 to 0.1)
solutions
The anodic polarisation curve (Fig. 7) of the rebar steel in the solution containing
low levels of carbonates and bicarbonates was different from that obtained in the
more concentrated solution. In this medium, the corrosion potential was approxi-
mately 300 mV lower than the one measured in solution (3) and no passivity was
found. As the potential was raised from the corrosion potential, instead of a low
passive current density, a current density that increased rapidly with the potential,
attaining a value of the order of 10
5 A/cm2 , was observed. A wide and blunt anodic
peak appeared at a potential of about )0.1 Vshe . Further increase in potential led to a
limiting current density. This limiting current was attributed to the limiting value of
the transport of ions across a thick film of green corrosion products, clearly seen on
the metallic surface.
The addition of 0.001% Cl
has no significant effect on the anodic behaviour of
the alloy. In the more concentrated chloride solutions the corrosion potential shifted
to a value approximately 100–150 mV lower than that in the plain carbonate/
bicarbonate solution. The shape of the polarisation curves was similar at all chloride
concentrations above 0.005%. A limiting current density of the order of 10
5 A/cm2
was observed when the potential was increased from the corrosion potential. By
further increasing the potential, a current density increment was seen. If, after the
current density increase, the direction of the potential scanning was inverted, the
current returned to the limiting value, showing a pronounced hysteresis, typical of a
passivity breakdown processes [10]. The microscopical observation of the samples
after polarisation revealed that in all the solutions tested the steel has undergone

-2
10
0.0015 M NaHCO3 + 0.005 M Na2 CO3
pH=9
-3
10
2
Current Density, A/cm

-4
10
-5
10
-6
10 -
0.000 % Cl
-
-7 0.001 % Cl
10 0.005 % Cl
-
-
0.010 % Cl
-8 -
10 0.050 % Cl
-
0.100 % Cl
-9
10
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
Potential, VSHE

Fig. 7. Potentiodynamic anodic polarisation curve for carbon steel in 0.015 M NaHCO3 + 0.005 M
Na2 CO3 solution, pH ¼ 9.0, containing different concentrations of chloride ion.
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2689

Fig. 8. Morphology of the attack observed on carbon steel after anodic polarisation in 0.015 M
NaHCO3 + 0.005 M Na2 CO3 + 0.01% Cl
solution, pH ¼ 9.0.

generalised corrosion. Irregularly shaped crystallographic pits were also observed in

the specimens tested in the more concentrated chloride solutions (Fig. 8).

3.2. Polarisation resistance and corrosion potential measurements

3.2.1. 0.9 M NaOH (pH ¼ 13.9) + x% Cl
(x ¼ 0 to 10) solutions

Fig. 9 shows the changes in the corrosion current density (icorr ) values derived
from the LPR measurements and in the corrosion potential (Ecorr ) observed for as-
received carbon steel during a one-week exposure in 0.9 M NaOH solution and in the
same solution with the addition of different concentrations of chloride ion. In 0.9 M
NaOH solution, icorr decrease with time and, after about 100 hours, a stationary
corrosion current density lower than 10
7 A/cm2 was obtained. This current density
value was found to be equivalent to a penetration rate as low as 1 lm/year. Similarly,
when plotting the changes in potential, Ecorr initially decreased, but after nearly 70 h,
it began to increase and attained a stationary value of about 0.030 Vshe , which
corresponds to the passive region.
When a chloride concentration equal to or higher than 1% (0.28 M) was added to
the solution, the corrosion rate was found to reach a value more than one order of
magnitude higher than that in pure sodium hydroxide solution. Correspondingly, the
Ecorr decreased during the exposition, achieving values lower than )0.200 Vshe . Pit-
ting was observed on the metallic surface after the experiment.

3.2.2. Saturated Ca(OH)2 (pH ¼ 12.5) + x% Cl
(x ¼ 0 to 0.2) solutions

The variations of icorr and Ecorr observed for carbon steel in saturated Ca(OH)2
solution were similar to those found in 0.9 M Na(OH) solution (Fig. 10). The
instantaneous corrosion rate diminished during the exposition, reaching a value
lower than 10
7 A/cm2 , whereas Ecorr exhibited a value around 0.040–0.030 Vshe . Also
in this media the corresponding penetration rate was 1 lm/year.
2690 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

Fig. 9. Development of the corrosion potential (a) and the corrosion rate (b) for carbon steel during a one-
week exposure in a 0.9 M NaOH solution, pH ¼ 13.9, containing different concentrations of chloride ion.

In the chloride containing solutions, at concentrations equal to or higher than

0.02% (0.056 M) the icorr raised to values more than one order of magnitude above
that found in the absence of chlorides while the Ecorr showed a decreasing trend and
attained after a 160 h exposure a steady value near to )0.200 Vshe .

3.2.3. 0.3 M NaHCO3 + 0.1 M Na2 CO3 (pH ¼ 9.0) + x% Cl
(x ¼ 0 to 1.0) solutions
The changes in Ecorr and Icorr found for the rebar steel in the solution containing
high levels of carbonates and bicarbonates were close to those detected in more
alkaline solutions (Fig. 11). The instantaneous corrosion rate, after a 50 h exposure,
attained a stationary value of approximately 5 · 10
7 A/cm2 , which was equivalent to
a penetration rate close to 6 lm/year. Ecorr showed few variations throughout the
experiment, exhibiting values between 0.04 and 0.09 Vshe .
The addition of chloride ions produces a noticeable increase in the instantaneous
corrosion rate and a shift of Ecorr towards lower values. As was shown in Fig. 11, in
the presence of 0.1% Cl
the corrosion rate was of the order of 10
6 A/cm2 while the
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2691

Fig. 10. Development of the corrosion potential (a) and the corrosion rate (b) for carbon steel during a
one-week exposure in a saturated Ca(OH)2 solution, pH ¼ 12.5, containing different concentrations of
chloride ion.

Ecorr attained a stationary value close to 0.02 Vshe . In a more concentrated chloride
solution, 1,0%, the corresponding values were icorr ¼ 8  10
6 A/cm2 and Ecorr ¼

0:23 Vshe .

3.2.4. 0.015 M NaHCO3 + 0.005 M Na2 CO3 (pH ¼ 9.0) + x% Cl
(x ¼ 0 to 0.1)
solutions
The corrosion behaviour of the carbon steel in the solution containing low levels
of carbonates and bicarbonates (Fig. 12) showed a remarkable difference from that
obtained in the other solutions tested. Even in the chloride-free solution, the samples
exhibited quite high values of the instantaneous corrosion rate, near 7 · 10
6 A/cm2
or 75 lm/year, and a low value of Ecorr (approx. )0.27 Vshe ). The addition of the
aggressive ion had no noticeable effect on the icorr value while the corrosion potential
of the steel, after a 160 h exposure in these solutions, was found to reach values
between )0.25 and )0.35 Vshe .
2692 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

Fig. 11. Development of the corrosion potential (a) and the corrosion rate (b) for carbon steel during a
one-week exposure in a 0.3 M NaHCO3 + 0.1 M Na2 CO3 solution, pH ¼ 9.0, containing different con-
centrations of chloride ion.

4. Discussion

4.1. Potentiodynamic experiments in high alkaline solutions

The low instantaneous corrosion rate values calculated from polarisation resis-
tance measurements and the corrosion potential values measured in saturated
Ca(OH)2 and 0.9 M NaOH solutions, indicated that in these highly alkaline solu-
tions the rebar steel was in the passive state at the free corrosion potential. This
result is in agreement with the predictions of the equilibrium potential–pH [11]
diagram for the Fe–H2 O system. According to this diagram a passive oxide film
should be stable in both solutions. The corresponding anodic polarisation curves
showed that passivity was maintained even when potentials values several hundreds
of millivolts higher than the corrosion potential were attained. In highly alkaline
solutions, such as saturated Ca(OH)2 solutions, NaOH, KOH or mixtures of theses
solutions, the passive state of steel is provided by the presence of a thin oxide film
that protects the metal from active corrosion. Nagayama and Cohen [12] as well as
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2693

Fig. 12. Development of the corrosion potential (a) and the corrosion rate (b) for carbon steel during a
one-week exposure in a 0.015 M NaHCO3 + 0.005 M Na2 CO3 solution, pH ¼ 9.0, containing different
concentrations of chloride ion.

Kruger and Calvert [13] suggested that the main component of the passive film
formed in alkaline solutions were Fe3 O4 or c-Fe2 O3 oxides while Smialowska and co-
workers [14,15], by using ellipsometry, concluded that the composition of the passive
film formed in both calcium and sodium hydroxide solutions was similar to an oxi-
hydroxide (FeOOH) rather than to Fe3 O4 or c-Fe2 O3 . Finally, Yonezawa et al. [16]
found that in saturated Ca(OH)2 solution a discontinuous layer of Ca(OH)2 was
present on the metal surface.
The addition of chloride ions has a noticeable effect on the corrosion behaviour of
carbon steel in high alkaline solutions, since in the presence of this ion a passivity
breakdown process took place above a certain potential that decreases with the
chloride concentration.
A critical chloride concentration above which pitting occurred on the metallic
surface was evaluated from the results of the potentiodynamic tests. The critical
chloride content was defined as the highest chloride concentration that can be tol-
erated without causing pitting corrosion of the steel up to the oxygen evolution
2694 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

potential (0.6–0.8 Vshe ). The chloride threshold was 1% (0.28 M) in the solution with
pH 13.9 (Fig. 1) but decreased to 0.02% (0.0056 M) in the saturated Ca(OH)2
solution at pH 12.5 (Fig. 3). Thus, the critical chloride concentration for pitting
initiation increases when the solution pH increases, as has been previously observed
[4,6]. The beneficial effect of alkalinity on chloride induced localised corrosion of
carbon steel is accounted for by the pitting mechanism developed by Galvele [17–19],
in which the maintenance of a local acidification on the metal solution interface is
required as the necessary condition for pitting initiation and propagation. The local
acidification is caused by the hydrolysis of metal ions inside of microcracks already
existing in the passivating film. For each metal and alloy, a critical acidification is
necessary to render repassivation unlikely and sustain pit activity. The presence of
high concentrations of OH
ions in the solution hinders the formation of the
localised acidification, producing inhibition of the pitting process.
The critical chloride concentration values here measured for as-received carbon
steel in alkaline solutions are of the same order than those previously reported for
samples with film-free surface [3–6]. As far as the threshold [Cl
]/[OH
] ratio nec-
essary for pitting to occur is concerned, the values found were 0.313 and 0.178 for 0.9
M NaOH and saturated Ca(OH)2 solutions, respectively. Both values are lower than
the [Cl
]/[OH
] ¼ 0.6 limit proposed by Haussman [5]. However, in those solutions
where the condition of [Cl
]/[OH
] ratio P 0.6 was accomplished, such as 0.9 M
NaOH + 3% Cl
([Cl
]/[OH
] ¼ 0.94) and saturated Ca(OH)2 + 0.1% Cl
([Cl
]/
[OH
] ¼ 0.89), pitting corrosion was initiated at 0.2 and 0.1 V, respectively, i.e. at
potential values several hundred of millivolts lower than potentials typical for
oxygen evolution.
It is worthwhile to mention that the present experiments were a rough represen-
tation of the actual concrete conditions. The initiation of chloride induced corrosion
of steel in concrete, and hence the chloride threshold, depends on several factors,
such as concrete mix proportions, blended materials, water content of the concrete,
temperature, type of cation, C3 A content of cement, among others. Attempts to
relate the critical [Cl
/[OH
] ratio values obtained in solution to the total chloride
threshold in concrete showed that, in general, the chloride threshold is higher for
mortars and concretes than for solution [1,20].

4.2. Potentiodynamic experiments in solutions simulating carbonated concrete

The results obtained in solutions simulating carbonated concrete pore solution

showed that, at constant carbonate/bicarbonate ratio and constant pH, the corro-
sion behaviour of the steel varied with the concentration of these ions. The instan-
taneous corrosion rate and the Ecorr values measured in the solution containing high
levels of carbonates and bicarbonates suggested that a protective passive film was
formed on the metal surface at the free corrosion potential. Likewise, the poten-
tiodynamic polarisation curve exhibited a large potential range in which the current
density was relatively low, of the order of 10
6 A/cm2 , and no attack was detected on
the surface of the samples after polarisation. On the other hand, the behaviour of the
steel during anodic polarisation in 0.015 M NaHCO3 + 0.005 M Na2 CO3 solution
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2695

indicated that the rebar steel undergone generalised corrosion when exposed to a
diluted carbonate/bicarbonate solution. The high corrosion rate values attained as
well as the relatively low value of the corrosion potential (Fig. 12) during the long-
term exposition to this solution confirmed this conclusion. In addition, the corrosion
potential of the steel in the 0.015 M NaHCO3 + 0.005 M Na2 CO3 solution was found
to fall into the generalised corrosion zone of the Pourbaix diagram for the Fe–H2 O
system [11].
The present results concerning the effect of the concentration on the corrosion
behaviour of the steel are in agreement with data reported in the literature [21–23].
Thomas and Davies [23] suggested that the increase in the stability of the Fe(III)
passive film on iron with increased bicarbonate concentration was due to the shift of
the Flade potential towards lower values. Thus, a high bicarbonate concentration
will prevent the dissolution of the passive film.
Several authors [24–26] have reported the existence of two current density peaks
in the anodic polarisation curve of carbon steel in bicarbonate solutions. The first
peak, at E <
0:2 V was associated with the dissolution of metallic iron (Fe) to
ferrous ion (Fe2þ ) and the formation on the metallic surface of a pseudo-passive
layer of Fe(OH)2 and/or FeCO3 [26,27]:
Fe þ 2H2 O ! FeðOHÞ2 þ 2Hþ þ 2e
ð1Þ

Fe þ HCO
þ
3 ! FeCO3 þ H þ 2e


ð2Þ

By further increasing the potential, a second current density peak was observed, at
0.1 V, and a passive film was formed
3FeðOHÞ2 ! Fe3 O4 þ 2H2 O þ 2Hþ þ 2e
ð3Þ

The current density peaks were not observed in the polarisation curves shown in
Fig. 5 because, as a consequence of the surface conditions of the samples, the open
circuit potentials measured in the present work were more positive than the potential
range in which they could be detected. On the other hand, the shape of the poten-
tiodynamic polarisation curves of the steel in the solution containing low levels of
carbonate and bicarbonate (Fig. 7) was comparable to the one reported by Mao
et al. [27]. These authors suggested that the changes in the shape of the polarisation
curves with the bicarbonate concentration were mainly related to differences in the
stability of the FeCO3 pseudo-passive film. When the HCO
3 concentration is low,
the pseudo-passive film of FeCO3 could be formed on the metal surface as a result of
local saturation. However, the local equilibrium of Fe2þ and CO2
3 with FeCO3 could
be broken by the diffusion of the Fe2þ ion away from the surface, resulting in the
dissolution of the film. If this was the case, the FeCO3 film would have gone through
a competition between formation and dissolution processes, and thus, the polari-
sation curve in low-concentration HCO
3 solutions showed a broad anodic current
peak of relatively high current density value. If at higher potentials, passivation did
not occur (Reaction 3), the metal would continue to dissolve as Fe(II) under diffu-
sion control and the current density would remain high.
2696 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

In the presence of chloride ions, pitting occurred in the bicarbonate plus car-
bonate containing solutions (Figs. 5 and 7). However, from a mechanistic point of
view, the pitting process in the more concentrated solution is different from that in
the less concentrated solution. In the first solution pitting was originated by the
localised rupture of passivity on an otherwise passive metal surface. In the less
concentrated solution, pitting occurred not on a passive metal surface, but on a
metal that was undergoing generalised dissolution. This type of pitting was also
found for nickel in NiCl2 [28,29] solutions and for ferritic stainless steel in HCl
solution [30]. There is no information available about its mechanism but in all the
above-mentioned systems pitting corrosion was initiated on a metal surface covered
by a salt layer. For carbon steel in dilute bicarbonate plus carbonate solution, the
salt film is the pseudo-passive layer of Fe(OH)2 and/or FeCO3 formed on the metal
surface at potentials lower than 0.1 V.
The potentiodynamic tests showed that in the 0.015 M NaHCO3 + 0.005 M
Na2 CO3 (pH 9) solution, where carbon steel did not show a passive behaviour, the
critical chloride concentration for pitting was found to be 0.001% (0.00028 M). On
the other hand, in the 0.3 M NaHCO3 + 0.1 M Na2 CO3 (pH 9) solution the chloride
threshold was 0.01% (0.0028 M). This value is comparable to the 0.02% (0.0056 M)
critical chloride concentration obtained from the polarisation curves in the more
alkaline saturated Ca(OH)2 solution, pH ¼ 12.5. The above observation can also be
accounted for by the localised acidification mechanism for pitting. Transport process
analysis inside a pit [17,19] showed that anions of weak acids salts, because of their
buffer properties, act as an impediment for the development of the locally acidified
zone in the same way as OH
ions. Therefore, in the presence of anions of weak acid
salts, it is more difficult to reach the low pH conditions necessary to promote pitting.

4.3. LPR and Ecorr measurements

The results of LPR measurements showed that when carbon steel is exposed at the
free corrosion potential in high alkaline solutions and in a solution simulating the
pore liquid of a high carbonated concrete the metal remain passive. The instanta-
neous corrosion rate diminishes during the exposition attaining values of the order of
10
7 A/cm2 while the Ecorr increases to values close to 0.030 Vshe or above (Figs. 9–
11). In the chloride containing solutions the icorr values attained after 160-h exposure
were at least one order of magnitude higher than those measured in the passive state.
Such increase is the signal that passivity rupture have already occurred. Corre-
spondingly, the Ecorr showed a decreasing trend during the exposition, achieving
steady values near )0.200 Vshe or below. Figs. 9–11 show that in these solutions,
passivity rupture was detected in the whole range of chloride concentration tested.
Therefore, no chloride threshold values were obtained in these experiments.
It may be noted that during the long term exposition at the free corrosion po-
tential passivity rupture was detected even at chloride concentrations lower than the
critical chloride concentration evaluated from the potentiodynamic tests. This dif-
ference is probably associated to the criteria used to determine the chloride threshold
from the polarisation curves since owing to the existence of an induction time for
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2697

pitting, the chloride level below which steel was not found to undergo pitting cor-
rosion up to the oxygen evolution potential may be overestimated in a potentiody-
namic measurement.
In Fig. 13, the Ecorr values measured in the different solutions tested have been
drawn in a potential–pH diagrams for the Fe–Cl
–H2 O system [31] constructed for
different chloride concentrations from data reported in the literature [3,31,32]. It was
reported by Pourbaix [33] that the pits, which will start to grow at a potential equal
to or higher than the pitting potential, will keep growing even when the potential is
lowered below the pitting potential. The pits will stop growing only when the po-
tential is lower than a certain potential called repassivation potential, or protection
potential. For this reason, the passive zone of the Pourbaix diagram for the Fe–Cl
–
H2 O system was divided in three regions, namely:

(a) Pitting zone, the lower limit of which was given by the pitting potential values.
(b) Imperfect passive zone, limited by the pitting potential as the upper limit and the
repassivation potential as the lower limit.
(c) Perfect passive zone, limited by the repassivation potential and the equilibrium
potential, as upper and lower limit respectively.

As shown in Fig. 13, the corrosion potential values measured in chloride con-
taining solutions were found to fall into the imperfect passive zone of the diagram,
i.e. between the pitting potential and the repassivation potential. Pitting corrosion

Fig. 13. Potential–pH diagram for the Fe–Cl
–H2 O, for different chloride concentration at 25 C. Line a
corresponds to the hydrogen reaction equilibrium and line b corresponds to the oxygen reaction equi-
librium. (h) Ecorr in pH ¼ 9, 0.3 M NaHCO3 + 0.1 M Na2 CO3 + 0.028 M Cl
solution. (j): Ecorr in pH ¼ 9,

0.3 M NaHCO3 + 0.1 M Na2 CO3 + 0.28 M Cl
solution. ( ) Ecorr in pH ¼ 12.5 saturated Ca(OH)2 + x%

Cl
solution (x ¼ 0:056, 0.014 and 0.028 M). ( ) Ecorr in pH ¼ 13.9, 0.1 M Na(OH) + x% Cl
solution
(x ¼ 0:28, 0.84 and 2.8 M).
2698 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699

will then may initiate on irregularities always present on the metal surface during the
initial period of exposition at the free corrosion potential. As the potential decreases,
no new pits will be nucleated on the metal surface but growing of the already existing
pits will occur.
In the more diluted carbonate plus bicarbonate solution, where the rebar steel
does not passivate, quite high icorr values (near 7 · 10
6 A/cm2 ) were derived from the
LPR measurements (Fig. 7). The occurrence of pitting has no noticeable effect on
the final instantaneous corrosion rate values. However, due to the localisation of the
attack, the penetration rate within the pits should be higher than the corrosion rate
on the remainder metallic surface.

5. Conclusions

• In high alkaline solutions and in the more concentrated carbonate plus bicarbon-
ate solution carbon steel remains passive even if the potential is raised up to the
potential range of the oxygen evolution reaction. In the presence of low levels of
carbonate and bicarbonate carbon steel does not passivate and generalised corro-
sion occurred at all potential higher than the free corrosion potential.
• In all the solutions tested a critical chloride concentration was found, above which
the presence of chlorides leads to the initiation of pitting on the metal surface. The
chloride threshold values estimated from the results of the potentiodynamic tests
were found to be of the same order than those previously reported for film-free
steel.
• In high alkaline solutions and in the more concentrated carbonate plus bicarbon-
ate solution, where the carbon steel exhibit passivation, the occurrence of chlo-
ride-induced pitting produces a significant increase in the icorr values derived
from the LPR measurements. On the other hand, in the less concentrated carbon-
ate plus bicarbonate solution pitting initiation has no noticeable effect on the
instantaneous corrosion rate values.
• From a mechanistic point of view, the pitting process in high alkaline solutions
and in the more concentrated carbonate plus bicarbonate solution is different
from that in the less concentrated solution. In the first solutions pitting was orig-
inated by the localised rupture of a passive film whereas in the less concentrated
solution, pitting occurred not on a passive metal surface, but on a metal that is
undergoing generalised corrosion.
• An increase in pH as well as the presence of high levels of carbonate and bicar-
bonate has a beneficial effect on the chloride threshold and improves the resistance
of carbon steel to localised corrosion.

Acknowledgement

The present research has been supported by the Agencia Nacional de Promoci

on
Cient
ıfica y Tecnol

ogica (ANPCyT), Argentina.
 M. Moreno et al. / Corrosion Science 46 (2004) 2681–2699 2699

References

[1] C. Alonso, C. Andrade, M. Castellote, P. Castro, Cemt. Concr. Res. 30 (2000) 1047.
[2] C. Alonso, M. Castellote, C. Andrade, Electrochim. Acta 47 (2002) 3469.
[3] L. Li, A.A. Sag€ ues, CORROSION/99, paper No. 567, NACE, Houston, TX, 1999.
[4] L. Bertolini, F. Bolzoni, T. Pastore, P. Pedeferri, Br. Corros. J. 31 (1996) 218.
[5] D.A. Hausmann, Mater. Prot. 6 (1967) 19.
[6] V.K. Gouda, Br. Corros. J. 5 (1970) 198.
[7] J.A. Gonz
alez Fern
andez, M.C. Andrade, in: La corrosi
on en estructuras de hormig
on armado,
Forma de evaluarla y prevenirla, Teor
ıa y Pr
actica de la lucha contra la corrosi

on, Consejo Superior
de Investigaciones Cient
ıficas, Centro Nacional de Investigaciones Metal
urgicas, Madrid, 1984.
[8] J.A. Gonz
alez Fern
andez, in: Control de la Corrosi
on: Estudio y Medida por T
ecnicas
Electroqu
ımicas, Consejo Superior de Investigaciones Cient
ıficas, Centro Nacional de Investigaciones
Metal
urgicas, Madrid, 1989.
[9] M.C. Andrade, V. Castelo, C. Alonso, J.A. Gonzalez, ASTM STP 906, American Society of Testing
and Materials, Philadephia, 1986, p. 43.
[10] J.R. Galvele, in: R.P. Frankenthal, J. Kruger (Eds.), Passivity of Metals, The Electrochemical Society,
Inc., NJ, 1978, p. 285.
[11] M. Pourbaix, Atlas of Electrochemical Equilibrium in Aqueous Solutions, Pergamon Press, Oxford,
1966, p. 307.
[12] M. Nagayama, M. Cohen, J. Electrochem. Soc. 110 (1963) 670.
[13] J. Kruger, J.P. Calvert, J. Electrochem. Soc. 114 (1967) 43.
[14] H. Oranowska, S. Saklarska-Smialowska, Corros. Sci. 21 (1981) 735.
[15] T. Zacroczymski, C.J. Fan, Z. Saklarska-Smialowska, J. Electrochem. Soc. 132 (1985) 2826.
[16] T. Yonezawa, V. Ashworth, R.P. Procter, Corrosion 44 (1988) 489.
[17] J.R. Galvele, J. Electrochem. Soc. 123 (1976) 464.
[18] J.R. Galvele, Corros. Sci. 21 (1981) 551.
[19] S.M. Gravano, J.R. Galvele, Corros. Sci. 24 (1984) 517.
[20] L. Li, A.A. Sagu
es, Corrosion 57 (2001) 19.
[21] D.M. Brasher, Br. Corros. J. 4 (1969) 122.
[22] M.C. Alonso, M.C. Andrade, Corros. Sci. 29 (1989) 1129.
[23] J.G.N. Thomas, J.D. Davies, Br. Corros. J. 12 (1977) 108.
[24] J. Jelinek, P. Neufeld, Corros. Sci. 20 (1980) 489.
[25] J.G.N. Thomas, T.J. Nurse, R. Walker, Br. Corros. J. 5 (1970) 87.
[26] D.H. Davies, G.T. Burstein, Corrosion 36 (1980) 416.
[27] X. Mao, X. Li, R.W. Revie, Corrosion 50 (1994) 651.
[28] I. Garz, H. Worch, W. Schatt, Werkstoffe u. Korr 27 (1969) 71.
[29] M.G. Alvarez, Doctoral Thesis, Universidad Nacional de Buenos Aires, 1977.
[30] J.R. Galvele, J.B. Lumsden, R.W. Staehle, J. Electrochem. Soc. 125 (1978) 1204.
[31] M. Moreno, Master in Science and Technology Thesis, Instituto de Tecnolog
ıa ‘‘Jorge S
abato’’,
Universidad de Gral San Mart
ın, 2000.
[32] M. Pourbaix, Corros. Sci. 26 (1970) 431.
[33] M. Pourbaix, Corrosion 14 (1974) 25.