Cement and Concrete Research 95 (2017) 257–269

Contents lists available at ScienceDirect

Cement and Concrete Research

journal homepage: www.elsevier.com/locate/cemconres

Effect of crack openings on carbonation-induced corrosion

Rita Maria Ghantous a,c,⁎, Stéphane Poyet a, Valérie L'Hostis a, Nhu-Cuong Tran b, Raoul François c
a
Den-Service d'Etude du Comportement des Radionucléides (SECR), CEA, Université Paris-Saclay, F-91191, Gif-sur-Yvette, France
b
EDF, R&D, MMC, F-77818 Moret-sur-Loing Cedex, France
c
LMDC, Université de Toulouse, INSA, UPS, Toulouse, France

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

Article history: Reinforced concrete is widely used in the construction of buildings, historical monuments, infrastructures and
Received 9 November 2016 nuclear power plants. For a variety of reasons, many concrete structures are subject to unavoidable cracks that
Received in revised form 14 February 2017 accelerate the diffusion of atmospheric carbon dioxide to the steel/concrete interface. Carbonation at the inter-
Accepted 23 February 2017
face induces steel corrosion that could cause the development of new cracks in the structure, a determining factor
for its durability. The aim of this article is to study the effect of existing cracks on the development of carbonation-
Keywords:
Reinforced concrete
induced corrosion. The results indicate that, after the initiation phase, the corrosion kinetics decreases with time
Cracks and the free corrosion potential increases independently of the crack opening. In addition, the corroded zone
Carbonation-induced corrosion matches the carbonated one. The interpretation of these results allows the authors to conclude that, during the
Environmental conditions corrosion process, corrosion products seal the crack and act as a barrier to oxygen and water diffusion. Conse-
quently, the influence of crack opening on corrosion development is masked and the corrosion development is
limited.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction alkalization of carbonated mortar. In the same study, results of experi-

Corrosion of the reinforcing steel is the main pathology affecting re-
inforced concrete structures and is a determining factor for their dura-
bility. These structures are subject to unavoidable cracks which create
a pathway for atmospheric carbon dioxide, oxygen, water and chlorides
to the steel/concrete interface and subsequently reduce the time to cor-
rosion initiation. Both laboratory studies [1–6] and in situ observations
[7–11] have noted earlier corrosion initiation in cracked concrete struc-
tures. Load-induced cracking is accompanied by interfacial slipping and
separation between concrete and steel [12–15]. It has also been shown
that both chloride- and carbonation-induced corrosion start and devel-
op for a few millimeters around the rebar at its interception with the
preexisting crack [16,17]. While there is a general consensus in the sci-
entific community on the deleterious effects of cracks and steel/mortar
interface quality on the initiation of reinforcement corrosion, the effects
of such phenomena on the propagation of reinforcement corrosion in-
duced by carbonation are still open to debate.
Tuutti [18] proposed a corrosion model for cracked and carbonated
concrete, which supposes that steel repassivation is possible due to
the sealing properties of the corrosion products and to a potential re-
⁎ Corresponding author at: Den-Service d'Etude du Comportement des Radionucléides
(SECR), CEA, Université Paris-Saclay, F-91191, Gif-sur-Yvette, France.
E-mail addresses: rita-maria.ghantous@yncrea.fr (R.M. Ghantous),
stephane.poyet@cea.fr (S. Poyet), valerie.lhostis@cea.fr (V. L'Hostis),
nhu-cuong.tran@edf.fr (N.-C. Tran), Raoul.francois@insa-toulouse.fr (R. François).
ments exposing cracked and carbonated specimens to different relative
humidities, validate the repassivation assumption. Tremper [19] inves-
tigated cracked specimens after 10 years of outdoor exposure and iden-
tified the presence of corrosion near the cracks and their immediate
surroundings. He also added that the corrosion detected did not have
deleterious consequences because of its minor degree. Additionally,
Vesikari [20] investigated the corrosion development in the cracked
concrete of 50 year old bridges and found that very few cracks had
given rise to serious corrosion damage. This observation supports the
idea that a reduction of corrosion kinetics occurs after the initiation
phase, due to the corrosion products developed deep in the crack.
On the other hand, Dang et al. [21], studying carbonation-induced
corrosion propagation, used ring-shaped mortar specimens with
preexisting cracks induced mechanically by internal pressure (expan-
sive core method). After accelerated carbonation (50% CO2 ‐ 65% RH),
the specimens were subjected to humidification/drying cycles to accel-
erate corrosion development. It was observed that the CO2 spread along
the total length of the steel/mortar interface and induced corrosion ini-
tiation on the entire circumference of the rebar. Thereafter, new cracks
were detected in the concrete cover characterized by a low tensile
strength [22]. Nevertheless, because corrosion was initiated on the en-
tire length of the rebar, it was not possible to conclude whether carbon-
ation-induced corrosion in cracked structures would lead to corrosion
cracks and thus to corrosion propagation.
In the field of nuclear energy, reinforced concrete is used for struc-
tures that have an impact on safety (containment buildings, cooling

http://dx.doi.org/10.1016/j.cemconres.2017.02.014
0008-8846/© 2017 Elsevier Ltd. All rights reserved.
258 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269

Fig. 1. Schematic representation of the reinforced mortar specimen.

towers), it is therefore important to complete the literature and under-
stand the effect of cracking and steel/concrete interface damage, in
terms of both corrosion initiation and corrosion propagation in cracked
specimens. In this project, specimens were exposed to raining/drying
cycles to simulate the outside environment and accelerate corrosion
naturally. The ultimate goal was to find answers to the following
questions:
• Does crack opening influence the kinetics of carbonation-induced cor-
rosion?
• Will carbonation-induced corrosion deep in the crack induce the de-
velopment of corrosion cracks and threaten the durability of the struc-
ture or it will seal the cracks and slow down corrosion propagation?
In the following sections, the experimental protocols developed to
provide answers to the above questions are detailed and then the re-
sults obtained are discussed.
2. Experimental program
2.1. Materials and specimens
The specimens tested were 70 × 70 × 280 mm prisms. A 6 mm cor-
rugated rebar was positioned in the middle of each specimen as shown
in Fig. 1. All the specimens were prepared with a mortar mix containing
three parts sand, two parts cement and one part water. Ordinary
Portland cement (CEM I 52.5) and siliceous sand (according to CEN EN
196-1) were used. The yield strength of the steel used was 500 MPa.
Mortar was poured in prismatic molds in two layers, each of them
being vibrated to eliminate air bubbles. After 24 h, the specimens
were unmolded and then cured for 28 days in water with calcium hy-
droxide. The mechanical properties of the mortar mix measured follow-
ing current standards and recommendations are shown in Table 1. The
specimens tested in this study were made from 2 consecutive batches
of 60 l each.
2.2. Cracking

Table 1 Immediately after curing, cracks of different widths were generated

Mechanical characteristics of the mortar mix. at mid-span of the prismatic specimens using a three-point bending
Compressive strength fc (MPa) 55 test. The cracking protocol consisted in applying the load at the mid
Tensile strength fctk (MPa) 3.5 span of the simply supported specimen using a hydraulic pump as
Young’s Modulus E (GPa) 33.3 shown in Fig. 2(a). The crack was usually obtained in the cross section
Poisson’s ratio ν 0.21
subjected to the maximal tensile stress (mid-span). In order to quantify

Fig. 2. Three-point bending test on 70 × 70 × 280 mm specimens.

 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269
Fig. 3. Range of residual crack openings obtained by the three point bending test.
the crack opening, a Linear Variable Differential Transformer (LVDT)
was placed where the crack was expected to appear as shown in Fig.
2(b). It measured the horizontal displacement between two points
located 20 mm from the specimen center on either side. Throughout
the test, the applied load and the displacement measured by the LVDT
were logged. Note that the displacement registered by the LVDT
before the appearance of the crack was due to flexural deformation of
the beam and cannot be considered as a crack opening. Therefore, to
calculate the crack opening, these initial displacements were set equal
to zero.
At the beginning of the bending test, the load was increased slowly
until a crack was detected. Then, by performing loading\unloading cy-
cles and by increasing the maximum applied load from one cycle to an-
other, the residual crack opening was increased in a controlled way and
a range of crack opening values was obtained (Fig. 3).
It was important, here, to limit the length of the zone in which the
bond between the steel and the mortar was altered (to prevent corro-
sion initiation all along the steel/mortar interface as in [22]). A previous
study proved that the three point bending test did not lead to total steel/
mortar interface damage [23]. In the present study, three residual crack
Fig. 4. Custom rain chambers.
259
openings, of 100 μm, 300 μm and 500 μm, were performed on several
prismatic specimens.
2.3. Corrosion
To trigger corrosion, the specimens were kept in an accelerated car-
bonation chamber at 3% CO2, 55% relative humidity and 25 °C for one
month. It had been verified that these accelerated carbonation condi-
tions were representative of the natural conditions in a previous study
[23,24], in which it was also shown that the carbonation duration was
long enough to ensure carbonation of the alteration zone [23,24]. Car-
bonation induces depassivation of the steel intercepting a crack and
thus active corrosion could start once the presence of the required re-
agents (water and oxygen) was ensured. It is important to note that, be-
fore carbonation, the cracked specimens were pre-conditioned at (20 ±
1 °C; 60 ± 5% RH) for one month. This step was important to reduce the
water content of the cementitious materials and thus ensure CO2 diffu-
sion in them during the carbonation test.
After carbonation, active corrosion was accelerated by performing
multiple simulations of rain/drying cycles. To this end, a custom rain
chamber was designed and manufactured. It was equipped with three
rows, in order to test the required number of specimens. The ‘rainfall’
was ensured using a micro drip system (Fig. 5(b)). On each row,
water was recovered in a collector (yellow in Fig. 4) and was evacuated
from the three rows through a common pipe (grey pipe linked to the
yellow collector in Fig. 4). Tap water was used to represent rainfall. To
avoid clogging of the sprinklers due to hard water, the water was fil-
tered to remove impurities (filters in green in Fig. 4). The pH of the
water was 7.5 and its chemical composition is given in Table 2.
Raining/drying cycles were programmed by an automated control
unit. In each row of each rain chamber and in the laboratory, a ther-
mo-hygrometer continuously measured temperature and relative hu-
midity. Dedicated software was used to log all the data each minute. A
photo of the small chamber is shown in Fig. 5(a). This experiment was
performed in an air-conditioned laboratory in which the temperature
was regulated at 20 °C ± 1 °C and the RH at 60 ± 5%.
Each raining phase lasted 30 min and was followed by a 72 h drying
phase. During the raining period, the RH value increased to 95% ± 5%.
Once the drying phase started, the RH took around 12 h to drop to
50% ± 10% (Fig. 6).
260 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269

Table 2
Elements present in water.

Element NH4-N NH4 NO2 HCO3 Cl NO3 SO4 CO3

Concentration (mg/L) b0.039 b0.05 b0.01 200 12 17 45 b1

Fig. 5. Real device for raining/drying cycles.

2.4. Corrosion analyses
After various exposure durations, three specimens were used for
physical-chemical analysis and one other specimen was used for
surface analysis (optical microscopy and Raman micro-spectroscopy).
Specimens were analyzed after 0, 15, 30, 60, 119 raining/drying cycles
corresponding to 0, 1.5, 3, 6, 12 months respectively. The free corrosion
potential of the rebars was monitored continuously during the experi-
ment on other specimens exposed to the same corrosion conditions.
Each of these analysis methods was applied to specimens having three
different residual crack openings (100, 300 and 500 μm).
The experimental program was summarized in Table 3 keeping in
view the parameters of the specimens studied and the corrosion analy-
ses methods.
rebar was inspected visually. Corroded lengths were measured on its
upper and lower parts with respect to the casting direction.
After steel extraction, phenolphthalein was pulverized onto the
fresh surface to measure the carbonated length along the steel/mortar
interface and compare it to the corroded length (Fig. 7).
b). Gravimetric measurements
The corrosion products attached to the extracted rebar were removed
by chemical attack (Sb2O3 + SnCl2 + HCl) following the standard ISO
8407 [25]. The difference between the initial mass of the steel (before cor-
rosion) and its final mass (after chemical treatment) gave the steel mass
lost during corrosion (ΔmS). In this study, gravimetric measurements
were repeated on three replicates.

2.4.1. Physico-chemical analysis c). Estimation of corrosion kinetics

a). Extraction of rebar and visual inspection The corrosion kinetics was calculated according to Eq. (1) [26].

The specimen was first split into two parts. Before the reinforcement mm ΔmS 3:65  10−2
¼  ð1Þ
was extracted, the distribution of corrosion products along the steel year ρS  SSL Δt

Fig. 6. Relative humidity variation during raining/drying cycles in the reference protocol.
 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269 261

Table 3
Summary of the experimental program.

Specimens Crack Number of Time period for Method of

characteristics openings specimens corrosion analysis
(μm) analysis

70 × 70 × 280 100 22 0, 1.5, 3, 6, 12 - Gravimetric

mm (CEM I 300 22 measure-
mortar) 500 22 ments
- Optical mi-
croscopy Fig. 9. Electrochemical measurement.
- Raman
spectrosco-
py
- Free corro-
sion 2.4.2. Surface analysis
potential Optical microscopy and Raman micro-spectroscopy were used to
identify and to measure the thickness of the corrosion products formed
at various locations in the specimens. From each specimen, a sample of
120 × 70 × 15 mm was extracted as shown in Fig. 8(a) and mounted in
epoxy resin at room temperature (20 °C). The sample was then polished
until the mid-section of the steel/reinforcement was reached (according
to the polishing direction shown in Fig. 8(a). Then, the surface of this
section was polished by grinding with SiC paper, grades 80 to 4000
and with 3-μm diamond paste under ethanol (Fig. 8(b)).
The morphology of the corrosion products layer and its thickness
along the entire length of the rebar were obtained by optical microsco-
py. Raman micro-spectroscopy analyses were performed on two differ-
ent zones: deep in the crack (called zone 2 in Fig. 8(b)) and at the end of
Fig. 7. Steel corrosion length with respect to the length of the carbonated steel/mortar
interface.
the corrosion layer (called zone 1 in Fig. 8(b)). The phases were identi-
fied by comparison with spectra in the literature [27–29].

where:ΔmS is the iron mass lost (mg);ρS is the volumetric mass density
2.4.3. Electrochemical analysis
of steel (g/cm3).SSL is the surface area of the corroded steel reinforce-
The evolution of the free corrosion potential was monitored by
ment (dm2).Δt is the duration of the active corrosion period (days).
means of an embedded manganese dioxide reference electrode (ERE
X-ray Computed Tomography (X-ray CT) was performed on the cor-
20) that was placed in the mortar cover of some specimens before cast-
rugated 6 mm steel rebar in order to determine its external area. The
ing. It was positioned at 50 mm from the crack (Fig. 9). In this configu-
rebar was examined using the 3D μCT scanner at BAM (Federal Institute
ration, the reference electrode measured a coupled corrosion potential
for Materials Research and Testing in Berlin). This scanner is equipped
(between active and passive parts of the steel bar), which was continu-
with a 225 kV micro focus X-ray tube and a flat panel detector with
ously logged using a data acquisition switch unit.
2048 × 2048 pixels. The spatial resolution was approximately 3 μm spe-
cial voxel (volumetric picture element) when scanning the 6 mm corru-
gated rebar. The surface area obtained by μCT was 19.54 ± 3. Results
0.08 mm2/mm.
Consequently, the area of the corroded steel reinforcement was ob- 3.1. Distribution of the corrosion products
tained according to Eq. (2). In this equation, Lcorr is the average corro-
sion length determined during the visual inspection of the corroded a). Length of the corrosion product layer
rebar (mm).
The evolution of the average corrosion length measured on three rep-
SSL ¼ 19:54  Lcorr  10ð−4Þ ð2Þ licates with respect to the number of raining/drying cycles is given in

Fig. 8. Sample preparation.

262 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269

Fig. 10. Corrosion length evolution with respect to raining/drying cycles and residual crack width (visual inspection).

Fig. 10. It should be noted that, in each specimen, there was no difference
in the corrosion lengths measured on the upper and the lower part of the
rebar with respect to the casting direction. Consequently, the corroded
length in each rebar was considered to be equal to the average value of
these two corrosion lengths measured on its upper and lower part.
The corrosion length was observed to increase with the residual
crack opening and tended to stabilize after 30–60 raining/drying cycles.
This observation is consistent with the study by Tremper [19] in which
the size of the corroded area increased with the crack width.
b). Development of corrosion products
Fig. 11 shows the development of the corrosion to carbonation
length ratios with respect to raining/drying cycles applied to specimens
having 100, 300 and 500 μm residual crack openings. Each point corre-
sponds to the average corrosion length to carbonation length ratio mea-
sured on the upper and lower steel/mortar interfaces of different
specimens subjected to the same exposure conditions for the same
duration.
It can be seen that the corrosion products filled the carbonated steel/
mortar interface zone in the first thirty raining/drying cycles and did not
spread farther thereafter. Moreover, some ratios were greater than 1,
reaching a value of 1.6. This may have been related to the accuracy of
the phenolphthalein pH indicator test and not to corrosion propagation.
According to Glasser and Matschei [30], the carbonated area detected by
the phenolphthalein test is limited to the totally carbonated zone where
the pH is lower than 8 and where portlandite and C-S-H are depleted.
This pH is lower than the value needed to depassivate the steel, which
is between 9.4 and 10 [31]. Parrott et al. [32] divided the carbonated
depth in two zones: the first totally carbonated and the second a
mixed zone where calcium carbonate and portlandite existed. A color
change in the phenolphthalein solution was visible in the first zone
and covered only half of the mixed zone. Moreover, by detecting car-
bonation products in zones where the phenolphthalein result was neg-
ative, Demoulin et al. [31] confirmed the limited ability of the
phenolphthalein test to detect carbonation, especially at its beginning.
Therefore, in this study the carbonation length measured using the
phenolphthalein test may have underestimated the real carbonated
length and this could explain the presence of some corrosion to carbon-
ation ratios that were slightly higher than 1.
c). Thickness of the layer of corrosion products
Fig. 12 provides information on the corrosion layer thickness on ei-
ther side of the crack in specimens showing different residual crack
openings (100, 300 and 500 μm) and subjected to different numbers
of raining/drying cycles (30, 60 cycles). The letter “S” on each graph rep-
resents the area between the curve and the x-axis. It corresponds to the

Fig. 11. Evolution of the corrosion over carbonation length ratio with respect to raining/drying cycles and residual crack widths (visual inspection).
 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269 263

Fig. 12. Corrosion products distribution obtained on specimens showing different residual crack openings (optical microscopy).

overall corroded surface measured on the upper part of the steel/mortar
interface and could be used as an indicator of the amount of corrosion
products having developed at this part of the interface.
It can be noted that the corrosion layer thickness measured on the
upper steel/mortar interface increases with the residual crack opening.
It is obvious that the length of the corrosion product layer also increases
with the residual crack opening, which is in agreement with Fig. 10. More-
over, the amount of corrosion products increases with the residual crack
opening. Concerning the corrosion evolution, no significant difference in
the corrosion length is visible between 30 and 60 cycles for the three re-
sidual crack openings. This is in agreement with Fig. 10 and Fig. 11.
triplicates subjected to the same conditions. An increase in the iron
mass loss with the number of cycles and with crack opening is obvious.
Whatever the crack opening, the iron mass loss tends to stabilize after
30–60 raining/drying cycles. This stabilization in the loss of iron mass
is consistent with the observation by Tuutti [18] (who exposed cracked
specimens to corrosion at different relative humidities). Corrosion at-
tack was fast at the beginning of the corrosion process, then was seen
to stabilize after this time and remained the same until the end of the
test.
b). Corrosion rate

3.2. Gravimetric measurements
a). Iron mass loss
Fig. 13 presents the results of the gravimetric measurements. Each
point in this figure is an average value of iron mass loss measured on
It is clear from Fig. 14 that the corrosion kinetics decreased with the
number of cycles for the three residual crack openings. This was due to
the formation of porous oxide layer at the steel surface. The decrease in
the corrosion kinetics could not be related to an autogenous healing of
cracks in concrete because the occurrence of the latter phenomena re-
quests a fully saturated crack [33,34] which is not the case in this
study. In addition, the optical microscopy analyses performed along

Fig. 13. Iron mass losses in specimens having different residual crack openings with respect to raining/drying cycles.
264 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269
Fig. 14. Corrosion rate evolution in specimens having different residual crack openings with respect to raining/drying cycles.
the total duration of the experiments confirm the absence of a calcite
crystal growth in the crack with the raining/drying cycles. For instance,
Fig. 15 shows an optical microscopy photo of a crack in a specimen hav-
ing a 100 μm surface residual crack opening (smallest crack width tested
in this study) after 6 months of exposure to raining/drying cycles. It can
be seen that no calcite crystal seems to plug the crack path while the
corrosion kinetics after this exposure duration is reduced. Therefore,
the reduction in the corrosion kinetics is certainly due to corrosion
products that seal the crack and not to an autogenous healing of crack
in concrete.
It can also be noted that, near 60 cycles, the corrosion rate seems
to be independent of the residual crack opening and is around 10 μm/
year after 119 cycles. In addition, the difference between the
corrosion kinetics measured at 6 and 12 months is negligible.
Consequently, an estimation of the corrosion kinetics at 12 months
relative to that at 6 months should logically give a value comparable
to the passive corrosion kinetics (0.1–1 μm/year [35,36]). However,
it was not possible to determine the exact value in this study because
the specimens analyzed at 6 months were not the same as those
analyzed at 12 months.
The observed decrease in the corrosion kinetics corresponds to the
decrease proposed in Tuutti's model [18]. According to Tuutti, this is
linked to the steel repassivation and he assumes that the corrosion
products seal cracks and inhibit the access of carbon dioxide to the
Fig. 15. Optical microscopy photo of a crack in a specimen showing 100 μm residual crack
opening and exposed to 60 raining/drying cycles.
rebar. However, substances that increase the pH value can diffuse
from the mortar to the steel, leading to a repassivation. This assumption
was also proposed by Schießl [10]. The phenomenon was verified here
by monitoring the evolution of the carbonated steel/mortar interface
length at different times during the corrosion test. It was found that
the carbonated length remained unchanged for the entire exposure du-
ration and did not decrease (Fig. 16). This does not exactly fit the theory
of Tuutti and Schießl and the explanation could be linked to the neutral
water that fills the cracked area every three days, during the raining
phase. The absorption of water by the mortar may inhibit a possible in-
crease in its pH in the cracked zone surrounding the rebar. In addition,
this could also be related to the macrocell process that leads to the cre-
ation of hydroxyl ions outside the carbonated (and corroded) area. Nev-
ertheless, despite all these assumptions, an increase in pH cannot be
totally discounted because it could occur without bringing the pH
above 9, which would make it undetectable by the phenolphthalein
pH indicator test. The theory of Tuutti and Schießl would probably
apply for long exposure durations (more than 12 months).
3.3. Free corrosion potential measurements
The evolution of the measured corrosion potentials is given in
Fig. 17. Two distinct periods can be observed in the corrosion process.
• The first period, during the first month of exposure, corresponds to
the initiation of corrosion, in which a drop in the corrosion potential
from −70 mV/SCE to −350 mV/SCE indicates corrosion initiation at
the crack location.
At the beginning of the corrosion test, the steel deep in the crack
intercepted carbonated mortar. Therefore, this part of the steel was
able to be oxidized and the corrosion potential thus fell.
In this project, the rebar was surrounded by a heterogeneous mortar
(carbonated and non-carbonated). This heterogeneity induced a differ-
ence in the potential and led to the formation of a cell between these
two zones. The passivated steel intercepting a non-carbonated mortar
and located far from the crack could act as a cathode while the steel
intercepting a carbonated mortar and located deep in the crack could
act as an anode. Therefore, during corrosion initiation, in addition to
the microcell mechanism, a macrocell corrosion mechanism directly de-
pendent on the chemical characteristics of the surrounding electrolyte
(pH of the mortar) also came into play. According to [37], it is likely
that macrocell corrosion would predominate.
• The second part of the process concerns corrosion propagation deep in
the crack. In this process, the corrosion potential starts to increase from
- 350 mV/SCE and almost reaches the corrosion potential of the
 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269 265

Fig. 16. Evolution of the carbonated steel/mortar interface length in specimens having different residual crack openings and corroded in the reference test.

uncracked specimen (around −70 mV/SCE), which is an indication of process at the crack location due to the oxide layer developed deep in
the reduction of the macro-cell process of corrosion. This increase in the crack. The difference in the potential between the zone of the
the corrosion potential could be related to the limitation of the anodic steel intercepting a carbonated mortar and the one intercepting a

Fig. 17. Variation of free corrosion potential (versus SCE) in specimens showing different residual crack openings and exposed to raining/drying cycles for 310 days.
266 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269
Fig. 18. Representative Raman spectra of corrosion products obtained on specimens having three different residual crack openings.
non-carbonated mortar would then be strongly reduced. At this mo-
ment, the macrocell corrosion process would be limited and the
microcell corrosion process would dominate.
3.4. Type of corrosion products
Representative Raman spectra obtained in three different zones of
the upper part of the steel/mortar interface are given in Fig. 18. A sche-
matic summary of the types of corrosion products identified is given in
Fig. 19. Ferrihydrite (5Fe2O3.9H2O), a poorly crystallized phase whose
chemical formula is still under discussion [38,39], is always present. In
addition to ferrihydrite, maghemite is also detected. These phases are
known to be chemically and electrochemically reactive1 [40]. Poorly
1
In this case, reactivity means the ability of identified phases constituting the rust layer
to be reduced
crystallized goethite is also detected and, as pointed out in the study
by Monnier et al. [41], the electrochemical reactivity of this phase is
linked to its crystallinity. Therefore, the goethite detected in the corrod-
ed specimens can also be a reactive phase. Lepidocrocite is also detected
in the corroded specimens and, according to Misawa et al. [42], this
phase appears in more aerated conditions. Despite the fact that this
phase is not usually shown in the corrosion of steel in concrete [31,
43], it has been detected in long-term corrosion layers developed on re-
bars embedded in concrete but only where the steel/concrete interface
contains air bubbles [44]. The presence of these voids may increase the
aeration, which may justify the appearance of lepidocrocite in [44]. Its
presence in this study could be correlated with the presence of cracks
in the concrete. It may also have been an artifact due to the preparation
of the sample. The cause of the lepidocrocite detection could not be con-
firmed but, independently of its origin during the wetting stages of
raining/drying cycles, it could be reduced and be involved in the follow-
ing cycle as proven in [45]. Antony et al. [46,47] confirm that
 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269 267

Fig. 19. Summary of the corrosion products identified in different zones of specimens having different residual crack openings.

lepidocrocite can be one of the reactive phases but they add that ferrihy-
drite is much more reactive.
In some Raman spectra, a double peak is visible at 340, 395 cm−1. It
was impossible to determine the corresponding phase on the basis of
our literature review. It usually appears with ferrihydrite and it may
be a precursor phase developing during the formation or the transfor-
mation of ferrihydrite.
It can be seen that the types of corrosion products are independent
of the residual crack openings.
4. Discussion and conclusions
In this research study, the effect of crack opening on the initiation
and propagation of carbonation-induced corrosion was tested and

Fig. 20. Corrosion model supposing corrosion propagation induced by a total carbonation of the concrete cover (corresponding to the corrosion model proposed by François & Arliguie
[50]).
268 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269
Fig. 21. Corrosion model supposing corrosion propagation induced by a corrosion crack.
analyzed. The results indicate a decrease in the corrosion kinetics with
the number of raining/drying cycles. The effect of the crack opening
on the corrosion kinetics was masked after several such cycles and the
corrosion products did not spread beyond the limits of the carbonated
zone. No corrosion cracks were visible on the outside surface of the
specimens. Furthermore, an increase in the free corrosion potential
was observed after several raining/drying cycles. All these results
allow the authors to infer that the corrosion products seal the cracks
and limit the access of oxygen and water to the rebar, thus slowing
down the propagation of corrosion This result is in agreement with
that obtained by a numerical simulation in the study by Millard et al.
[48]. It has also been proven that even a crack of 500 μm residual
width (in un-loaded conditions) does not lead to a dangerous corrosion
state. This result was unexpected since the European standard EN 1992-
1-1 recommends a maximum crack opening of 300 μm (in loaded con-
ditions) for the XC4 class.
Additionally, the free corrosion potential measurements suggest
that two corrosion mechanisms may occur consecutively. The first
mechanism is globally macrocell corrosion that occurs during the initi-
ation phase. Afterwards, the second mechanism, which acts during the
propagation phase, corresponds mainly to microcell corrosion.
The corrosion rate measured in this study decreased with time and,
for the duration of this study, the corrosion was in its induction phase
(reduction in the corrosion rate induced by the development of a porous
corrosion layer on the steel surface). The duration of this phase cannot
be predicted and, in the long term, two possibilities exist:
• The first is that corrosion will not have any detrimental effect on the
durability of structures (Fig. 20(a)).
In this case, the parameter most harmful for the sustainability of a
structure is the concrete cover carbonation. The model proposed by
Tuutti [18] to describe the steel corrosion sequence in uncracked con-
crete could be applicable in the uncracked zones. It is presented in Fig.
20(b) and consists of two phases (initiation and propagation). During
the initiation phase, carbon dioxide penetrates the concrete cover,
eventually leading, at a certain depth of carbonation, to a situation
where corrosion initiation is thermodynamically favored. Thereafter,
the corrosion propagation phase begins and is marked by an acceler-
ation in the corrosion rate and consequently in the deterioration
rate. The duration of the initiation and the propagation phases are de-
pendent on the concrete cover properties (thickness, w/c ratio) [6,10,
11,19,49].
By combining, the corrosion processes deep in the crack and far from
the crack, it can be deduced that the corrosion model proposed in the
study by François & Arliguie [50] may describe the global behavior of a
structure (Fig. 20(c)). A similar corrosion model for cracked and car-
bonated concrete is also proposed by Tuutti [18].
The results shown in Schießl's study [10] support this first hypothesis
concerning corrosion propagation. The author investigated cracked
specimens after four and ten years of exposure and reported the pres-
ence of corrosion in the uncracked zones due to the penetration of the
carbonation front to the steel.
• The second possibility assumes a creation of corrosion cracks before
the total carbonation of the concrete cover. In this condition, the cor-
rosion cracks may be the underlying cause of the corrosion propaga-
tion process. However, it should be noted that the new corrosion
cracks intercept rebar embedded in an alkaline medium. Therefore,
a certain duration of carbonation is needed before the corrosion can
propagate along the rebar. In fact, corrosion cracks are generally par-
allel to the rebar and this promotes homogenous carbonation deep
in the crack. Therefore, a microcell corrosion mechanism may exist
during the propagation phase. A schematic sketch of steel corrosion
development in this second possibility is given in Fig. 21. It should
be noted that this second possibility has a small probability of occur-
rence, especially in structures that are not subjected to dynamic
loading.
Finally, concerning the types of corrosion products, it can be de-
duced that there is no influence of the crack opening on the type of cor-
rosion products. The main observation is that the corrosion products
developed are known to be reactive phases that are electrochemically
and chemically unstable and have tendency to be reduced
(lepidocrocite, ferrihydrite, poorly crystallized goethite, etc.). Therefore,
these phases may act as a cathode during corrosion due to the microcell
mechanism, in which the iron is the anode.
Acknowledgment
The experiments reported in this paper were conducted in the
LECBA Laboratory at CEA Saclay in collaboration with the LMDC Labora-
tory of INSA Toulouse. The authors gratefully acknowledge the support
of the industrial partner EDF. The authors also gratefully acknowledge
the Federal Institute for Materials Research and Testing in Berlin for
the X-ray Computed Tomography measurement. Finally, there is also
an acknowledgment for Etienne Amblard, from the LECBA laboratory,
for his help during the realization of the experiments.
References
[1] C. Arya, F.K. Ofori-Darko, Influence of crack frequency on reinforcement corrosion in
concrete, Cem. Concr. Res. 26 (1996) 345–353.
[2] N.S. Berke, M.P. Dallaire, M.C. Hicks, R.J. Hoopes, Corrosion of steel in cracked con-
crete, Corros. Sci. 49 (1993) 934–943.
[3] R. Francois, G. Arliguie, Effect of micro-cracking and cracking on the development of
corrosion in reinforced concrete members, Mag. Concr. Res. 51 (1999) 143–150.
[4] O. Gautefall, O. Vennesland, Effects of cracks on the corrosion of embedded steel in
silica-concrete compared to ordinary concrete, Nord. Concr. Res. 2 (1983) 17–28.
[5] T.U. Mohammed, N. Otsuki, M. Hisada, T. Shibata, Effect of crack width and bar types
on corrosion of steel in concrete, J. Mater. Civ. Eng. 13 (2001) 194–201.
 R.M. Ghantous et al. / Cement and Concrete Research 95 (2017) 257–269
[6] P. Schießl, M. Raupach, Laboratory studies and calculations on the influence of crack
width on chloride-induced corrosion of steel in concrete, ACI Mater. J. 94 (1997)
56–61.
[7] P. Fidjestol, N. Nilson, Field test of reinforcement corrosion in concrete, ACI Spec.
Publ. 65 (1980) 205–217.
[8] K. Katawaki, Corrosion of steel in the concrete exposed to seawater spray zone,
Symp. Proc. Crack. Concr. Struct. (1977) 133–136.
[9] G. Rehm, H. Moll, Versuche zum studium des einflußes der rissbreite auf die
rostbilding an der bewehrung von stahlbeton-bauteilen, Deutscher Ausschuss für
Stahlbeton 1964, p. 169.
[10] P. Schießl, Zur frage der zulassigen rissbreite und der enforderlichen betondeckung
im stahlbetonbau unter besonderer berucksichtigung der karbonatisierung des
betons, Deutscher Ausschuss für Stahlbeton 1976, p. 255.
[11] E.F. O'Neil, Study of reinforced concrete beams exposed to marine environment, ACI
Spec. Publ. 65 (1980) 113–132.
[12] R. François, I. Khan, N.A. Vu, H. Mercado, A. Castel, Study of the impact of localised
cracks on the corrosion mechanism, Eur. J. Environ. Civ. Eng. 16 (2012) 392–401.
[13] A. Michel, A.O.S. Solgaard, B.J. Pease, M.R. Geiker, H. Stang, J.F. Olesen, Experimental
investigation of the relation between damage at the concrete-steel interface and ini-
tiation of reinforcement corrosion in plain and fibre reinforced concrete, Corros. Sci.
77 (2013) 308–321.
[14] B.J. Pease, Influence of Concrete Cracking on Ingress and Reinforcement
Corrosion(Thesis) Technical University of Denmark, 2010.
[15] B. Pease, M. Geiker, H. Stang, J. Weiss, The design of an instrumented rebar for as-
sessment of corrosion in cracked reinforced concrete, Mater. Struct. 44 (2011)
1259–1271.
[16] R. Durton, A. Mommens, Corrosion des armatures dans le béton armé, 1964.
[17] R. François, J.C. Maso, Effect of damage in reinforced concrete on carbonation or
chloride penetration, Cem. Concr. Res. 18 (1988) 961–970.
[18] K. Tuutti, Corrosion of Steel in Concrete, Cement and Concrete Research Institute,
1982.
[19] B. Tremper, The corrosion of reinforcing steel in cracked concrete, J. Proc. 43 (1947).
[20] E. Vesikari, Corrosion of reinforcing steels at cracks in concrete, NASA STI/Recon
Tech. Rep. N 83 (1981).
[21] V.H. Dang, R. François, V. L'Hostis, D. Meinel, Propagation of corrosion in pre-cracked
carbonated reinforced mortar, Mater. Struct. 48 (2015) 2575–2595.
[22] V.H. Dang, Initiation and Propagation Phases of re-Bars Corrosion in Pre-Cracked Re-
inforced Concrete Exposed to Carbonation or Chloride Environment(PhD thesis)
Institut National des Sciences Appliquées de Toulouse, 2013.
[23] R.M. Ghantous, A. Millard, S. Poyet, R. François, V. L'Hostis, N.C. Tran, Experimental
and numerical characterization of load-induced damage in reinforced concrete
members9th International Conference on Fracture Mechanics of Concrete and Con-
crete Structures 2016, pp. 1–7.
[24] R. M. Ghantous, S. Poyet, V. L'Hostis, N. C. Tran, and R. François, “Effect of accelerated
carbonation conditions on the characterization of load-induced damage in rein-
forced concrete members,” Submitt. to Mater. Struct., 2017.
[25] NF EN ISO 8407, Corrosion des métaux et alliages - Élimination des produits de cor-
rosion sur les éprouvettes d'essai de corrosion, 2014.
[26] D. Landolt, Corrosion and Surface Chemistry of Metals, CRC Press, 2007.
[27] D. Neff, P. Dillmann, L. Bellot-Gurlet, G. Beranger, Corrosion of iron archaeological ar-
tefacts in soil: characterisation of the corrosion system, Corros. Sci. 47 (2005)
515–535.
[28] D. Neff, S. Reguer, L. Bellot-Gurlet, P. Dillmann, R. Berthelon, Structural characteriza-
tion of corrosion products on archeological iron: an integrated analytical approach
to establish corrosion forms, J. Raman Spectrosc. 35 (2004) 739–745.
269
[29] M. Hanesh, Raman spectroscopy of iron oxides and (oxy)hydroxides at low laser
power and possible applications in environmental magnetic studies, Geophys. J.
Int. 177 (2009) 941–948.
[30] F. Glasser, T. Matschei, Interactions between Portland cement and carbon dioxide,
Proceedings of the 12th International Congress on the Chemistry of Cements, 2007.
[31] A. Demoulin, C. Trigance, D. Neff, E. Foy, P. Dillmann, V. L'hostis, The evolution of the
corrosion of iron in hydraulic binders analysed from 46- and 260-year-old buildings,
Corros. Sci. 52 (2010) 3168–3179.
[32] L. Parrott, D. Killoh, Carbonation in a 36 year old, in-situ concrete, Cem. Concr. Res.
19 (1989) 649–656.
[33] C. Edvardsen, Water permeability and autogenous healing of cracks in concrete, ACI
Mater. J. 96 (1999) 448–454.
[34] N. Hearn, Self-sealing, autogenous healing and continued hydration: what is the dif-
ference? Mater. Struct. 31 (1998) 563–567.
[35] A. Rosenberg, C. Hansson, C. Andrade, Mechanisms of corrosion of steel in concrete,
Mater. Sci. Concr. 1 (1989) 285–314.
[36] C.M. Hansson, Comments on electrochemical measurements of the rate of corrosion
of steel in concrete, Cem. Concr. Res. 14 (1984) 574–584.
[37] P. Bamforth, Probalistic performance based durability design of concrete structures,
Management of Concrete Structures for Long-Term Serviceability, Proceedings of
the Int. Seminar. Sheffield, UK, 1997.
[38] U. Cornell, Rochelle M. Schwertmann, The Iron Oxides: Structure, Properties, Reac-
tions, Occurrences and Uses, John Wiley & Sons, Weinheim, 2003.
[39] U. Schwertmann, R.M. Cornell, Iron Oxides in the Laboratory: Preparation and Char-
acterization, John Wiley & Sons, Weinheim, 2008.
[40] R.M. Cornell, U. Schwertmann, The Iron Oxides: Structure, Properties, Reactions, Oc-
currences and Uses, Wiley-VCH, 2003.
[41] J. Monnier, et al., A corrosion study of the ferrous medieval reinforcement of the
Amiens cathedral. Phase characterisation and localisation by various microprobes
techniques, Corros. Sci. 52 (2010) 695–710.
[42] T. Misawa, K. Hashimoto, S. Shimodaira, The mechanism of formation of iron oxide
and oxyhydroxides in aqueous solutions at room temperature, Corros. Sci. 14
(1974) 131–149.
[43] B. Huet, Comportement à la corrosion des armatures dans un béton carbonaté. Influ-
ence de la chimie de la solution interstitielle et d ‘une barrière de transport(PhD the-
sis) Insitutat National des Sciences Appliquées de Lyon, 2005.
[44] V. L'Hostis, L. Vincent, V. Praca, D. Neff, L. Bellot-Gurlet, P. Dillmann, Characterization
of long-term corrosion of rebars embedded in concretes from French Historical
Buildings aged from 50 to 80 years, Eurocorr 2007, pp. 1–10.
[45] M. Stratmann, K. Hoffmann, In situ Mössbauer spectroscopic study of reactions
within rust layers, Corros. Sci. 29 (1989) 1329–1352.
[46] H. Antony, L. Legrand, L. Maréchal, S. Perrin, P. Dillmann, A. Chaussé, Study of
lepidocrocite γ-FeOOH electrochemical reduction in neutral and slightly alkaline so-
lutions at 25 °C, Electrochim. Acta 51 (2005) 745–753.
[47] H. Antony, S. Perrin, P. Dillmann, L. Legrand, A. Chaussé, Electrochemical study of in-
door atmospheric corrosion layers formed on ancient iron artefacts, Electrochim.
Acta 52 (2007) 7754–7759.
[48] A. Millard, V. L'Hostis, Modelling the effects of steel corrosion in concrete, induced
by carbon dioxide penetration, Eur. J. Environ. Civ. Eng. 16 (2012) 375–391.
[49] R. François, G. Arliguie, Influence of service cracking on reinforcement steel corro-
sion, J. Mater. Civ. Eng. 10 (1998) 14–20.
[50] R. François, G. Arliguie, Durability of loaded reinforced concrete in chloride environ-
ment, ACI Spec. Publ. 145 (1994) 573–596.