Magazine of Concrete Research Magazine of Concrete Research, 2016, 68(7), 353–363

http://dx.doi.org/10.1680/jmacr.15.00234
Volume 68 Issue 7
Paper 1500234
Effect of carbonation on release of Received 06/06/2015; revised 17/07/2015; accepted 28/07/2015
Published online ahead of print 23/09/2015
bound chlorides in chloride-contaminated
ICE Publishing: All rights reserved
concrete
Geng, Easterbrook, Liu and Li

Effect of carbonation on release

of bound chlorides in chloride-
contaminated concrete
Jian Geng Long-Yuan Li
Associate Professor, School of Civil Engineering and Architecture, Professor of Structural Engineering, School of Marine Science and
Ningbo Institute of Technology, Zhejiang University, Ningbo, People’s Engineering, University of Plymouth, Plymouth, UK
Republic of China
Dave Easterbrook
Associate Professor, School of Marine Science and Engineering, University
of Plymouth, Plymouth, UK
Qing-Feng Liu
Assistant Professor, School of Marine Science and Engineering, University
of Plymouth, Plymouth, UK; State Key Laboratory of Ocean Engineering,
School of Naval Architecture, Ocean and Civil Engineering, Shanghai Jiao
Tong University, Shanghai, People’s Republic of China

In this paper an experimental study is presented on the interaction between concrete carbonation and chloride
attack. The experiments are carried out on chloride-contaminated cement pastes that have been exposed to a carbon
dioxide environment. Free and total chloride contents at different carbonation times are measured in the tested
specimens using traditional leaching and acid-soluble methods. Bound chloride contents are calculated from the
measured total and free chlorides. The experimental data show that the carbonation of cement paste results in a
release of bound chlorides, which is related to not only the decomposition of Friedel’s salt, but also the
decomposition of calcium-silicate-hydrate (C-S-H) gel. Based on the obtained experimental results, a numerical model
is also developed for simulating the carbonation process of the chloride-contaminated cement paste. The model can
be used to explain how the bound chlorides are released in a chloride-contaminated cement paste under the
influence of carbonation reactions.

Notations t time (s)

B mass concentration of bound chlorides (mg/g) x distance from exposed surface (m)
Beq mass concentration of bound chlorides when xco depth of carbonation (m)
equilibrium between free and bound chlorides is α constant used in chloride binding isotherm
reached (mg/g) β constant used in chloride binding isotherm (g/mg)
Bo initial mass concentration of bound chlorides (mg/g)
Bso modified initial mass concentration of bound Introduction
chlorides (mg/g) The corrosion of reinforcing steel in concrete structures has
C mass concentration of free chlorides (mg/g) become a major concern for infrastructure owners and oper-
Co initial mass concentration of free chlorides (mg/g) ators. Two major causes that are connected to corrosion of
Cs mass concentration of free chlorides in surface reinforcing steel in concrete structures are carbonation and
layer (mg/g) chloride attacks (Page and Page, 2007). Carbonation is the
D diffusion coefficient of free chloride ions in cement reactions between atmospheric carbon dioxide and the alkaline
paste (m2/s) components of concrete, producing a carbonated surface layer
Dco diffusion coefficient of carbon dioxide (m2/s) in which the pore solution pH value is depressed to near-
erfc complementary error function neutral levels. A secondary effect of carbonation, which is also
f(kc, xco) reduction function due to carbonation significant in terms of its influence on reinforcing steel cor-
kb rate constant (1/s) rosion, is that it can impede chloride binding and cause the
kc reduction factor due to carbonation release of bound chlorides into the pore solution phase of
kt carbonation coefficient (1/sn) concrete that contains a modest level of chloride salts as a con-
L length of specimen (m) taminant, thus exacerbating the corrosive nature of the electro-
n constant lyte. In addition, carbonation can also influence the diffusion

353
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research
Volume 68 Issue 7
of chloride ions in concrete as it can alter the pore volume and
pore structure and thus the transport properties of the con-
crete. In certain cases the permeability of the carbonated con-
crete may increase, as in the case of concrete made with
blended cements such as blast furnace slag and fly ash con-
crete; in others, it may decrease as in ordinary Portland
cement concrete (Dias, 2000; Saillio et al., 2014). In urban and
industrial areas, where environmental pollution results in a sig-
nificant concentration of carbon dioxide, carbonation-initiated
reinforcement corrosion prevails (Yoon et al., 2007). However,
in highways and marine or coastal structures, chloride ions,
originating from deicing salts or seawater, are the primary
cause of reinforcing steel corrosion (Thomas and Jones, 1996).
The chlorides that are transported through the concrete pore
network and micro-cracks depassivate the oxide film covering
the reinforcing steel and accelerate the reaction of corrosion
(Ryou and Ann, 2008; Yu et al., 1993). The worst situation is
when a concrete structure is subjected to both carbonation and
chloride attacks (Backus et al., 2013). Surveys on concrete
structures have shown that the presence of moderate levels of
chloride contamination (0·4–1·0% by mass of cement) can
have a significant influence on the rates of corrosion that may
be induced in carbonated concrete (Aramaki et al., 2006). This
illustrates the difficulty that arises when the presence of more
than one corrosion-inducing agency contributes to the aggres-
siveness of the exposure conditions for a particular structure.
Extensive research has been carried out over the last two
decades on the corrosion of reinforcing steel induced by con-
crete carbonation (Bertos et al., 2004; Papadakis et al., 1991;
Yuan et al., 2009). The work includes the experimental investi-
gation and numerical modelling of the carbonation process in
reinforced concrete structures (Chang and Chen, 2006; Isgor
and Razaqpur, 2004; Saetta and Vitaliani, 2004, 2005; Saetta
et al., 1993), experimental investigation of the effect of con-
crete carbonation on pore structure and diffusion properties of
hydrated cement pastes (Anstice et al., 2005; Ngala and Page,
1997) and the experimental and material analyses of the effect
of carbonation on chloride binding capacity in cementitious
materials (Arya et al., 1990; Chávez-Ulloa et al., 2013;
Delagrave et al., 1997; Fabbri et al., 2009; Glass and Buenfeld,
2000; Goñi and Guerrero, 2003; Hassan, 2001; Ioannou et al.,
2015; Jirickova and Cerny, 2006; Mesbah et al., 2012; Reddy
et al., 2002; Suryavanshi and Swamy, 1996; Suryavanshi et al.,
1996). Note that in the service life prediction of a concrete
structure it is necessary to distinguish the free and bound
chlorides due to the retardation effect of bound chlorides
(Mangat and Molloy, 1995). If the release of bound chlorides
due to concrete carbonation were ignored in the prediction
model, the service life predicted would be inaccurate and
unsafe. The X-ray diffraction (XRD) and differential thermal
analysis (DTA) results reported by Suryavanshi and Swamy
(1996), obtained from the experiments of four different types
of concrete slabs subjected to chloride penetration over a total
period of about 3 years and then exposed to atmospheric
354
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
Effect of carbonation on release of
bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li
carbonation for almost the same period, showed that the solu-
bility of Friedel’s salt increases with the degree of concrete car-
bonation and the stability of both Friedel’s salt and ettringite
is pH dependent. Experimental study on the stability of
Friedel’s salt in calcium aluminate cement pastes containing
sodium chloride (NaCl) with respect to carbonation was con-
ducted by Goñi and Guerrero (2003), in which the carbonation
was carried out on a powdered sample in flowing 5% carbon
dioxide gas at 65% relative humidity (RH) to accelerate the
process. It was found that the carbonation of Friedel’s salt
did not produce a significant increase of soluble chloride,
suggesting that the chlorides released by carbonation of
Friedel’s salt might be trapped by the solid. Similar exper-
iments in which the specimens were carbonated first and
then followed by the chloride attack were conducted by
Hassan (2001) to investigate the chloride binding capacity of
carbonated concrete. In his experiments three different pre-
carbonated pastes were immersed in chloride solutions with
different concentrations. He found that the pre-carbonated
pastes had almost no binding capacity.
The above survey of the literature shows that, despite the con-
siderable amount of work, both experimental and numerical,
that has been published on concrete carbonation and its influ-
ence on chloride binding, there is not much work focusing on
the quantification of release of bound chlorides in chloride-
contaminated concrete caused by concrete carbonation. In this
paper an experimental study is presented on the release of
bound chlorides in chloride-contaminated cement paste speci-
mens when they are exposed to a carbon dioxide environment.
The free and total chloride contents at various different carbo-
nation times were measured in the tested specimens using
leaching and acid-soluble methods. The bound chloride con-
tents were calculated from the measured total and free chlor-
ides. Based on the obtained experimental results, a numerical
model was also developed for simulating the carbonation
process of the chloride-contaminated cement paste.
Experimental programme
The cement used in this experiment was Type 42·5 ordinary
Portland cement. In order to reduce the experimental running
time but still be able to achieve good and representative results,
chloride binding was achieved by using 0·5 mol/l sodium chlor-
ide solution as the mixing water for the casting of the cement
paste specimens. The water to cement ratio used in the mix
was 0·5.
In total, ten cement paste specimens of dimensions 40 mm

40 mm
160 mm were cast and then cured at a standard
curing condition (20 ± 2°C and 95% RH) for a period of 56 d.
After this, all specimens were dried at room temperature
(20 ± 2°C and 60% RH) for 1 d. After the standard curing and
drying processes, two specimens were analysed and the remain-
ing eight specimens were used for further accelerated
 Magazine of Concrete Research
Volume 68 Issue 7
carbonation tests. In order to investigate the effect of carbona-
tion depth on the release of bound chlorides, each specimen
was sealed on five of its six surfaces by paraffin wax and only
one 40 mm
40 mm surface was left untouched. All eight
specimens were placed into an accelerated carbonation
chamber, which was set to about 20% carbon dioxide gas with
a temperature of 20–25°C and a RH of 65–75%. During the
accelerated carbonation test, two specimens were removed for
analysis after the carbonation times of 14, 28, 56 and 90 d.
Four slices, each 1 cm thick, were cut from the carbonated
specimens, starting from the exposed surface (see Figure 1).
Each slice was broken into small blocks, which were then
immersed in anhydrous ethanol for 7 d to terminate hydration.
These small blocks were ground into fine powder (passing
through a 0·15 mm mesh sieve). The powders were then stored
in a desiccator with silica gel and soda lime at 11% RH to
minimise any further carbonation before they were used for the
determination of total and free chloride contents and other
material characterisation analyses.
The free chloride content of the fine powder was determined by
using the leaching method. The total chloride content of the
fine powder was determined using the acid-soluble method. The
bound chloride content was calculated based on the measured
total and free chloride contents. Note that the initial total chlor-
ide content in each specimen cured after the standard curing
condition can be calculated based on the mixing water of
0·5 mol/l sodium chloride solution, which is 8·863 mg/g of
10 mm
1 2 3 4
Exposed surface for
carbonation experiment
160 mm
(a)
4
3
2
1
(b)
Figure 1. Schematic view of specimens investigated in
carbonation tests: (a) dimensions and exposed conditions;
(b) layers used in chloride content analysis
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
Effect of carbonation on release of
bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li
cement paste. However, owing to the transport of free chloride
ions taking place in the specimen during the carbonation
period, the total chloride content measured at each time in each
layer may not be equal to this initial total chloride content.
The XRD analysis was carried out using a D8 Advance instru-
ment (Bruker AXS) with a Cu Kα radiation generated with
40 kV and 30 mA.
Experimental results
Chloride binding in cementitious materials is dominated
by the content of tricalcium aluminate (C3A) and tetra-
calcium aluminoferrite (C4AF) to form Friedel’s salt
(3CaO•Al2O3•CaCl2•10H2O) (Florea and Brouwers, 2012). In
this study, sodium solution was used as the mixing water, so
Friedel’s salt is formed from the chemical reactions not only
between sodium chloride and tricalcium aluminate but also
between sodium chloride and mono-sulfoaluminate, which is
the hydration product of tricalcium aluminate. The forming
process of Friedel’s salt can be interpreted by following
reactions
CaðOHÞ2 þ 2NaCl ! CaCl2 þ 2Naþ þ 2OH
3CaO  Al2 O3 þ CaCl2 þ 10H2 O
! 3CaO  Al2 O3  CaCl2  10H2 O
Sealed by paraffin wax
Sealed by paraffin wax
40 mm
Sealed by paraffin wax
355
 Magazine of Concrete Research Effect of carbonation on release of
Volume 68 Issue 7 bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li

3CaO  Al2 O3  CaSO4  12H2 OðAFmÞ þ 2NaCl

10 Total chlorides
! 3CaO  Al2 O3  CaCl2  10H2 O þ 2Naþ þ SO2 Free chlorides

Mass concentration of chlorides: mg/g

4
Bound chlorides
þ 2H2 O
8

After the carbonation, hydration products are converted to 6

calcium carbonate (CaCO3), silica gel and alumina gel, and
the pH value of the pore solution drops to about 9 (Bertos
4
et al., 2004). The drop in pH value may decrease the degree of
competition offered by hydroxyl, but increase the solubility of
Friedel’s salt and thus causes the release of bound chlorides 2
(Goñi and Guerrero, 2003; Page and Venneland, 1983;
Suryavanshi and Swamy, 1996). The decomposition of
0
Friedel’s salt due to reduced alkalinity can be interpreted by 0 14 28 56 90
the following reactions Time: d

3CaO  Al2 O3  CaCl2  10H2 O Figure 2. The mass concentration of free, bound and total
! 4Ca2þ þ 2Cl þ 4OH þ 2AlðOHÞ3 chlorides in surface layer

12
AlðOHÞ3 þOH ! AlðOHÞ
4
Total chlorides
Mass concentration of chlorides: mg/g

Free chlorides
10
Bound chlorides
Carbonation changes the nature of the hydration products and
thus affects the chloride binding ability of cement pastes. 8
Figure 2 shows the variation of the measured bound, free and
total chlorides with the carbonation time, obtained from the 6
surface layer of the tested specimens. It can be seen from the
figure that there is almost no change in bound chlorides
4
during the first 14 d of carbonation. This is mainly because the
thickness of layers chosen in the tests is relatively large (1 cm),
while the penetration of carbon dioxide in cement paste is very 2
slow. However, with the further increase of carbonation time,
there is a clear decrease of bound chlorides. The longer the 0
carbonation time is, the more the bound chlorides reduce. This 1 2 3 4

finding is consistent with what is reported in literature (Goñi Layer

and Guerrero, 2003; Suryavanshi and Swamy, 1996). As is
expected, the decrease of bound chlorides results in an increase
of free chlorides. However, the rate of the decrease of bound
chlorides is not the same as the rate of the increase of free
chlorides. This is demonstrated by the content of total chlor-
ides shown in Figure 2, which decreases with the carbonation
time. Note that during the carbonation test the specimen was
not exposed to water-saturated air, meaning that the leaching
of free chlorides through the exposed surface would be very
small and could be ignored. This suggests that some of the free
chlorides in the surface layer have diffused into inner layers
during the carbonation period.
Figure 3 shows the variation of the measured bound, free and
total chlorides with the distance from the exposed surface,
obtained at 56 d of carbonation. It can be seen from the
Figure 3. The mass concentration of free, bound and total
chlorides at 56 d
figure that the content of bound chlorides increases with the
distance from the exposed surface, which is correlated to the
general one-dimensional diffusion profile of carbon dioxide,
whereas the content of free chlorides decreases with the dis-
tance from the exposed surface, which is associated with the
release of bound chlorides. The content of total chlorides exhi-
bits a convex shape, indicating that the release rate of bound
chlorides is slightly greater than the diffusion rate of free chlor-
ides. The sum of the total chlorides in these four layers
is 34·244 mg/g, which is slightly less than the initial total
chlorides 4
8·863 = 35·452 mg/g, indicating that only a very

356
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research Effect of carbonation on release of
Volume 68 Issue 7 bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li

12
Total chlorides C
Free chlorides
Mass concentration of chlorides: mg/g

10 Bound chlorides

8
V
V 90 d
6 V C C
V C

4 56 d

F CH
2 28 d

E C-S-H 0d
0
1 2 3 4
Layer 5 10 15 20 25 30 35 40 45 50 55 60
2θ
Figure 4. The mass concentration of free, bound and total
chlorides at 90 d Figure 5. XRD patterns of surface layer at 0, 28, 56 and 90 d of
carbonation (E: AFt; F: FS; CH: calcium hydroxide; V: vaterite; CSH:
C-S-H gel; C: calcite)

small amount of chlorides has leached from the exposed

surface or gone further into the specimen.
that the intensity of vaterite increases with the carbonation
time. Vaterite is a kind of transformation of calcite and its
Figure 4 shows the variation of the measured bound, free and
presence is the result of the carbonation of C-S-H gel in
total chlorides with the distance from the exposed surface,
cement-based materials. This suggests that there is an increase
obtained at 90 d of carbonation. The main features shown in
in the content of C-S-H gel decomposed due to carbonation.
Figure 4 are similar to those found in Figure 3, except that
Thus, it can be concluded that the release of bound chlorides
there are more bound chlorides released and more free chlor-
due to carbonation is related to not only the decomposition of
ides accumulated in each layer. Also, it can be seen from
FS but also the decomposition of C-S-H gel. In the literature
Figure 4 that the content of total chlorides in the inner layers,
it was also reported that C-S-H in hardened cement paste can
particularly in layers 3 and 4, exceeds their initial value of
be decomposed through carbonation to be altered into calcium
8·863 mg/g. This again demonstrates that there is a transport
carbonate, silicon dioxide (SiO2) and water (H2O) (Kobayashi
of free chlorides, in addition to the release of bound chlorides.
et al., 1994).

The XRD results of the surface layer of the specimens sub-

Figure 6 shows the XRD results of different layers in the speci-
jected to 0, 28, 56 and 90 d of carbonation are shown in
men subjected to 56 d of carbonation. It can be seen from the
Figure 5. From the plotted XRD patterns it is possible to
figure that there is an obvious diffraction peak of FS in the
identify ettringite (AFt), Friedel’s salt (FS) and calcium
second layer, although its intensity is lower than that found in
hydroxide (CH) in the sample before the specimen was carbo-
the non-carbonated specimen. Also, AFt is not detected in the
nated (0 d) with clear diffraction peaks at around 9°, 11° and
second layer of the carbonated specimen. This indicates that
18° (2θ). However, after 28 d of carbonation the diffraction
the carbonation reaction has taken place in the second layer.
peak of AFt disappears and the intensity of diffraction peaks
In contrast, the XRD pattern of the fourth layer is found to be
of FS and CH reduces. When the carbonation time reaches
similar to that found in the non-carbonated specimen. This
90 d, AFt, FS and CH are hardly identified in the XRD
suggests that there is no significant carbonation reaction
pattern. Also, it is observed from Figure 5 that the intensity of
taking place in the fourth layer before 56 d of carbonation.
diffraction peak of calcite increases with the carbonation time.
Consequently, there is limited release of bound chlorides in
These results are consistent with what was reported in the
that layer. This is consistent with the results shown in Figure 4.
literature, for example, Goñi and Guerrero (2003) and
Suryavanshi and Swamy (1996). The decrease in the intensity
of diffraction peak of FS indicates that there is a decompo- Numerical modelling
sition of FS due to carbonation reaction, which leads to the The experimental results described in the preceding section
decrease of bound chlorides and the increase of free chlorides indicate that there is an interaction between the release
as demonstrated by Figure 2. It can be noticed from Figure 5 of bound chlorides and the transport of free chlorides.

357
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research Effect of carbonation on release of
Volume 68 Issue 7 bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li

Note that the cement paste is a porous material, whose pores

C are partially filled with water and air. When the specimen is
exposed to a carbon dioxide environment, carbon dioxide will
V penetrate into the specimen from its exposed surface through
both the gaseous and liquid phases of the cement paste.
However, after they have penetrated into the cement paste, part
F
of the carbon dioxide reacts with the hydrated cement paste,
CH V
E 56 d – fourth layer which is known as carbonation, whereas the other part con-
V
tinues to penetrate into the cement paste. The carbonation not
only tends to neutralise the alkalinity of the pore solution, but
56 d – second layer also leads to the release of bound chlorides in the chloride-
contaminated cement paste. The release of bound chlorides
C will increase the content of free chlorides and thus leads to the
C 56 d – first layer transport of free chlorides through the liquid phase of the
V C
cement paste.

5 10 15 20 25 30 35 40 45 50 55 60
According to the mass conservation of chlorides in a unit
2θ
volume of cement paste, the following mass conservation
Figure 6. XRD patterns of first, second and fourth layers after
equation can be established
56 d of carbonation (E: AFt; F: FS; CH: calcium hydroxide; V:

@C @B @ @C
vaterite; CSH: C-S-H gel; C: calcite) 1: þ ¼ D
@t @t @x @x

To simulate this interaction, a transport model that considers
the transport of free chloride ions and carbon dioxide as well
as the adsorption/desorption of bound chlorides induced by
carbonation is developed and presented in this section (see
Figure 7). Since the specimen is exposed to the carbon dioxide
environment only on its one surface and all other surfaces are
sealed, the transport of free chloride ions and carbon dioxide
in the specimen can be treated as a one-dimensional transport
problem.
where C is the mass concentration of free chlorides, B is the
mass concentration of bound chlorides, t is the time, x is the
distance from the exposed surface and D is the diffusion coeffi-
cient of free chloride ions in the cement paste. The mass con-
centration of bound chlorides can be described using the
following first-order reaction equation (Wang, 2001)
@B
2: ¼ kb ðB  Beq Þ
@t

where kb is the rate constant and Beq is the mass concentration

of bound chlorides at the time when the equilibrium between
CO2 CO2 Gaseous phase
CO2 free and bound chlorides has been reached. It is obvious from
Equation 2 that the bound chlorides will be released to
CO2
become free chlorides if B > Beq; otherwise the free chlorides
CO2 will be adsorbed to become bound chlorides.

Cl Cl Liquid phase For non-carbonated concrete the bound chlorides at the equili-
Cl
Cl brium state can be expressed in terms of the free chlorides by
CO2
Cl using a chloride-binding isotherm such as a linear, Freundlish
Cl Cl
or Langmuir isotherm (Tang and Nilsson, 1993). If the
Langmuir isotherm is used (Li and Page, 2000), Beq can be
Cl Cl Cl Cl Solid phase
expressed as follows

Cl αC
Cl 3: Beq ¼
Cl 1 þ βC
Cl Cl

where α and β are the chloride-binding constants, which are

Figure 7. One-dimensional transport model proposed for dependent on the type of cement used and the water to cement
concrete under carbonation ratio employed in the specimens. For the present experiments,
α and β can be determined based on the initial mass

358
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research Effect of carbonation on release of
Volume 68 Issue 7 bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li

concentrations of the free and bound chlorides (see Figure 2),

which gives α = 1·6 and β = 0·0212 g/mg. 1·0

Reduction function, f (kc, xco)

For fully carbonated concrete the mass concentration of bound 0·8
chlorides at the equilibrium state can be expressed using a
similar expression as Equation 3 but with a reduction factor as
follows (Hassan, 2001; Saillio et al., 2014) 0·6

kc αC 0·4
4: Beq ¼
1 þ βC
14 d
0·2 28 d
where kc is the reduction factor due to carbonation, which 56 d
90 d
depends on the type of binders and mixture of the concrete
and the concentration of carbon dioxide. 0
0 0·5 1·0 1·5 2·0 2·5 3·0 3·5 4·0 4·5 5·0
Distance from exposed surface: cm
For the present one-dimensional accelerated carbonation
problem, the depth of carbonation can be approximated as Figure 8. Reduction function for carbonated concretes (Dco =
follows (Broomfield, 2006; Chávez-Ulloa et al., 2013; Villain 1·2
10−11 m2/s, kc = 0·01, n = 0·7, kt = 92·36 mm/year0·7, vertical
et al., 2007) dash lines are the depths of carbonation xco calculated from
Equation 5)
5: xco ¼ kt tn

where xco is the depth of carbonation, kt is the carbonation smooth increase of f (kc, xco) from kc to 1 in the transition
coefficient and n is a constant. Theoretically, Equation 3 is zone. Substituting Equations 5 and 7 into Equations 1 and 2
applied to the non-carbonated zone of x > xco and Equation 4 yields
is applied to the carbonated zone of x < xco. This means that
there will be a discontinuity at the junction between the carbo- @C @2C
¼D 2
nated and non-carbonated zones. In reality, however, there @t @x 
 
8:
will be a transition zone between the carbonated and non- 1  kc x  kt tn αC
þ kb B  1  erfc pffiffiffiffiffiffiffiffiffi
carbonated zones, in which the concentration of carbon 2 2 Dco t 1 þ βC
dioxide changes gradually and smoothly from the carbonated
zone to the non-carbonated zone. To take into account this
transition zone, the following modified reduction function is
proposed which can be applied to both carbonated and non-  
 
@B 1  kc x  kt tn αC
carbonated zones 9: ¼ kb B  1  erfc pffiffiffiffiffiffiffiffiffi
@t 2 2 Dco t 1 þ βC


1  kc x  xco
6: f ðkc ; xco Þ ¼ 1  erfc pffiffiffiffiffiffiffiffiffi
2 2 Dco t

The solution of Equations 8 and 9 requires the following

where f (kc, xco) is the reduction function due to carbonation,
initial and boundary conditions
erfc is the complementary error function and Dco is the diffu-
sion coefficient of carbon dioxide. In this case the mass con- 10: Bðx; 0Þ ¼ Bo and Cðx; 0Þ ¼ Co
centration of the bound chlorides at the equilibrium state can
be expressed as follows


 @Cð0; tÞ @CðL; tÞ
1  kc x  xco αC 11: ¼0 and ¼0
7: Beq ¼ 1  erfc pffiffiffiffiffiffiffiffiffi @x @x
2 2 Dco t 1 þ βC
where Bo is the initial mass concentration of bound chlorides,
Co is the initial mass concentration of free chlorides and L is
Figure 8 graphically shows the variation of the reduction func- the length of the specimen. Equation 11 indicates that the free
tion with the distance at various different times. It can be seen chloride ions do not enter or exit the specimen at its two ends.
from the figure that f (kc, xco) = kc in the fully carbonated zone Equations 8 and 9 can be solved using numerical methods
and f (kc, xco) = 1 in the non-carbonated zone. There is a from which the mass concentrations of free and bound

359
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research Effect of carbonation on release of
Volume 68 Issue 7 bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li

chlorides can be obtained. Numerical schemes for solving a set
of partial differential equations can be found in many math-
ematical textbooks and thus are not provided or discussed
further here.
Numerical results
Before Equations 8 and 9 can be solved numerically, one has
to know the values of all parameters involved in the model.
The initial mass concentrations of bound and free chlorides
can be obtained from the experimental data shown in
of Cs(t) is shown in Figure 9, from which a reasonably good
agreement between them is observed.
After Bo, Co, D, Dco, kt, kc and kb are defined, Equations 8
and 9 can be solved numerically. Figure 10 and Figure 11
show the predicted mass concentration distribution profiles of
bound and free chlorides in the specimen at three different
times (0, 56 and 90 d of carbonation). To demonstrate the
6

Mass concentration of bound chlorides: mg/g

Figure 2, which gives Bo = 5·297 mg/g and Co = 3·566 mg/g.
5 Equation 13
The diffusion coefficient of free chloride ions in the liquid Experimental data
phase of the cement paste is assumed as D = 2·032

10−11 m2/s, which is two orders of magnitude smaller than that 4
in water. The diffusion coefficient of carbon dioxide used in
Equation 6 is assumed to be Dco = 1·2
10−11 m2/s. 3

The carbonation coefficient kt can be determined experimen-

2
tally by measuring the carbonation depth. The reported value
of kt in the literature for the case of n = 0·5 varies from
2·0 mm/√year for good concrete exposed to the natural 1
environment to 80 mm/√year for normal concrete exposed in
an accelerated carbonation test chamber (Broomfield, 2006; 0
0 10 20 30 40 50 60 70 80 90 100
Chávez-Ulloa et al., 2013). In the present study, the values n =
Time: d
0·7 and kt = 92·36 mm/year0·7 are assumed.
Figure 9. Variation of bound chlorides in surface layer with time
The reduction factor kc varies from 0 to 0·2, depending on the (α = 1·6, β = 0·0212 g/mg, kc = 0·01, kb = 3·125
10−7/s, Cs = 4·95
type of concrete (Hassan, 2001; Saillio et al., 2014). In the mg/g, Bso = 5·3 mg/g)
present study, kc is determined based on the best fit of pre-
dicted profiles of free and bound chlorides with the experimen-
tal data shown in Figure 3 and Figure 4, which gives kc = 0·01. 6
Mass concentration of bound chlorides: mg/g

The rate constant kb can be estimated based on the experimen-

5
tal data of the free and bound chlorides in the surface layer
shown in Figure 2. Note that when Equation 9 is applied to
4
the surface layer, it can be simplified as follows


dB kc αCs 3
12: ¼ kb B 
dt 1 þ βCs
2
where Cs is the mass concentration of free chloride ions in the
0d
surface layer. If Cs does not vary with time, Equation 12 can 1 56 d
be integrated directly and it yields 90 d

 0
kc αCs kc αCs 0 0·005 0·010 0·015 0·020 0·025 0·030 0·035 0·040 0·045
13: BðtÞ ¼ þ Bso  ekb t
1 þ βCs 1 þ βCs Distance from exposed surface: m

Figure 10. Mass concentration distribution profiles of

where Bso = αCs/(1 + βCs) is the modified initial mass concen-
tration of bound chlorides. By the best fit of predicted B(t)
using Equation 13 with the experimentally obtained B(t)
shown in Figure 2, kb = 3·125
10−7 s−1 = 0·0275 d−1 =
9·86 year−1 is identified. A comparison of the predicted B(t)
with the experimental data in which Cs is taken as the average
bound chlorides in specimen (α = 1·6, β = 0·0212 g/mg,
D = 2·032
10−11 m2/s, Dco = 1·2
10−11 m2/s, n = 0·7,
kt = 92·36 mm/year0·7, kc = 0·01, kb = 3·125
10−7/s, circular
and square data points are the experimental data at 56
and 90 d, respectively)

360
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research Effect of carbonation on release of
Volume 68 Issue 7 bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li

8 contamination in cement pastes. The experiments were carried

out on chloride-contaminated cement pastes in an accelerated
Mass concentration of free chlorides: mg/g

7
carbonation chamber. The experimental data showed that the
carbonation of cement paste results in a release of bound
6
chlorides, which is related to not only the decomposition of
5 Friedel’s salt but also the decomposition of C-S-H gel. The
release of bound chlorides leads to an increase of free chlorides
4 and the corresponding diffusion of free chlorides from a region
of high concentration to a region of low concentration.
3

2 Based on the obtained experimental results, a numerical model

0d has also been developed for simulating the carbonation process
1 56 d of the chloride-contaminated cement paste. Detailed discussion
90 d
of the determination of parametric values involved in the
0
0 0·005 0·010 0·015 0·020 0·025 0·030 0·035 0·040 0·045 model has been provided. The reliability of the model was
Distance from exposed surface: m tested using the experimental data. The model can be used to
explain how the bound chlorides are released in a chloride-
Figure 11. Mass concentration distribution profiles of free contaminated cement paste and how the free chlorides trans-
chlorides in specimen (α = 1·6, β = 0·0212 g/mg, D = 2·032
10−11 port under the influence of the carbonation reaction.
m2/s, Dco = 1·2
10−11 m2/s, n = 0·7, kt = 92·36 mm/year0·7, kc =
0·01, kb = 3·125
10−7/s, circular and square data points are the
experimental data at 56 and 90 d, respectively) Acknowledgements
The authors wish to acknowledge the financial support
received for this work from EU Research Council (FP7-
numerical model developed here, experimental data measured PEOPLE-2011-IRSES-294955). The first author wishes to
at 56 and 90 d of carbonation for bound and free chlorides are acknowledge the National Natural Science Foundation of
also superimposed in these figures. It is evident from the com- China (no. 51108415) and China Scholarship Council (no.
parison shown in the figures that the present transport model 201208330319) for supporting his research work conducted in
is able to reflect the main features of the interaction between Plymouth.
the release of bound chlorides and the transport of free chlor-
ides in the specimen when it is subjected to an accelerated car-
bonation test. The predicted profiles show that the release of
bound chlorides increases with time and more bound chlorides REFERENCES
were released in the place near the exposed surface where Anstice DJ, Page CL and Page MM (2005) The pore solution
carbon dioxide concentration was higher. The variation of free phase of carbonated cement pastes. Cement and Concrete
chlorides is opposite to that of bound chlorides, which indi- Research 35(2): 377–383.
cates that the release of bound chlorides leads not only to an Aramaki S, Kakio T, Tanimura Y and Miyagawa T (2006) The
increase in, but also the transport of, free chlorides in the mechanical properties of corroded reinforcing steel bars
specimen. collected from the railways reinforced concrete viaducts
deteriorated by carbonation and mixed chloride. Journal of
Note that, although the model developed herein is for the Japan Society of Civil Engineers, Series E2 (Materials and
experiments in which the chloride was initially mixed in Concrete Structures) 62(4): 777–789.
the concrete specimens, most parameters determined by the Arya C, Buenfeld NR and Newman JB (1990) Factors influencing
model, such as the rate constant kt, the reduction factor kc due chloride binding in concrete. Cement and Concrete
to carbonation and the carbonation coefficient kb, are indepen- Research 20(2): 291–300.
dent of the initial condition and can be used for general pur- Backus J, McPolin D, Basheer M, Long A and Holmes N (2013)
poses. The model itself can be applied to the case where both Exposure of mortars to cyclic chloride ingress and
chloride and carbon dioxide ingress into concrete from carbonation. Advances in Cement Research 25(1): 3–11.
exposure to the environment if the initial and boundary con- Bertos MF, Simons SJR, Hills CD and Carey PJ (2004) A review of
ditions for the chlorides are properly modified. accelerated carbonation technology in the treatment of
cement-based materials and sequestration of CO2. Journal
of Hazardous Materials B 112(3): 193–205.
Conclusions Broomfield JB (2006) Corrosion of Steel in Concrete,
This paper has presented an experimental study on the inter- Understanding, Investigation and Repair. E & F N Spon
action between concrete carbonation and chloride level of Ltd, London, UK.

361
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
 Magazine of Concrete Research
Volume 68 Issue 7
Chang CF and Chen JW (2006) The experimental investigation
of concrete carbonation depth. Cement and Concrete
Research 36(9): 1760–1767.
Chávez-Ulloa E, Camacho-Chab R, Sosa-Baz M, Castro-Borges P
and Pérez-López T (2013) Corrosion process of reinforced
concrete by carbonation in a natural environment and an
accelerated test chamber. International Journal of
Electrochemical Science 8(7): 9015–9029.
Delagrave A, Marchand J, Olliver JP, Julien S and Hazrati K
(1997) Chloride binding capacity of various hydrated
cement systems. Advanced Cement Based Materials 6(1):
28–35.
Dias WPS (2000) Reduction of concrete sorptivity with age
through carbonation. Cement and Concrete Research 30(8):
1255–1261.
Fabbri A, Corvisier J, Schubnel A et al. (2009) Effect of
carbonation on the hydro-mechanical properties of
Portland cements. Cement and Concrete Research 39(12):
1156–1163.
Florea MVA and Brouwers HJH (2012) Chloride binding related
to hydration products: part I: ordinary Portland cement.
Cement and Concrete Research 42(2): 282–290.
Glass GK and Buenfeld NR (2000) The influence of chloride
binding on the chloride-induced corrosion risk in
reinforced concrete. Corrosion Science 42(2): 329–344.
Goñi S and Guerrero A (2003) Accelerated carbonation of
Friedel’s salt in calcium aluminate cement paste. Cement
and Concrete Research 33(1): 21–26.
Hassan Z (2001) Binding of External Chloride by Cement
Pastes. PhD thesis, Department of Building Materials,
University of Toronto, Toronto, Ontario, Canada.
Ioannou S, Paine K, Reig L and Quillin K (2015) Performance
characteristics of concrete based on a ternary calcium
sulfoaluminate–anhydrite–fly ash cement. Cement and
Concrete Composites 55: 196–204.
Isgor OB and Razaqpur AG (2004) Finite element modeling of
coupled heat transfer, moisture transport and carbonation
processes in concrete structures. Cement and Concrete
Composites 26(1): 57–73.
Jirickova M and Cerny R (2006) Chloride binding in building
materials. Journal of Building Physics 29(3): 189–200.
Kobayashi K, Suzuki K and Uno Y (1994) Carbonation of
concrete structures and decomposition of CSH. Cement
and Concrete Research 24(1): 55–61.
Li LY and Page CL (2000) Finite element modelling of chloride
removal from concrete by electrochemical method.
Corrosion Science 42(12): 2145–2165.
Mangat PS and Molloy BT (1995) Chloride binding in concrete
containing PFA, gbs or silica fume under sea water
exposure. Magazine of Concrete Research 47(171):
129–141, http://dx.doi.org/10.1680/macr.1995.47.171.129.
Mesbah A, Cau-dit-Coumes C, Renaudin G, Frizon F and Leroux F
(2012) Uptake of chloride and carbonate ions by calcium
monosulfoaluminate hydrate. Cement and Concrete
Research 42(8): 1157–1165.
362
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
Effect of carbonation on release of
bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li
Ngala VT and Page CL (1997) Effects of carbonation on
pore structure and diffusion properties of hydrated
cement pastes. Cement and Concrete Research
27(7): 995–1007.
Page CL and Page MM (2007) Durability of Concrete and
Cement Composite. Woodhead Publishing in Materials,
Oxford, UK.
Page CL and Venneland O (1983) Pore solution composition
and chloride binding capacity of silica-fume cement pastes.
Materials and Structures 16(1): 19–25.
Papadakis VG, Vayenas CG and Fardis MN (1991)
Fundamental modeling and experimental investigation
of concrete carbonation. ACI Material Journal
88(4): 363–373.
Reddy B, Glass GK, Lim PJ and Buenfeld NR (2002) On the
corrosion risk presented by chloride bound in concrete.
Cement and Concrete Composites 24(1): 1–5.
Ryou JS and Ann KY (2008) Variation in the chloride
threshold level for steel corrosion in concrete arising from
different chloride sources. Magazine of Concrete Research
60(3): 177–187, http://dx.doi.org/10.1680/macr.2008.60.
3.177.
Saetta AV and Vitaliani RV (2004) Experimental investigation
and numerical modeling of carbonation process in
reinforced concrete structures. Part I: theoretical
formulation. Cement and Concrete Research 34(4):
571–579.
Saetta AV and Vitaliani RV (2005) Experimental investigation
and numerical modeling of carbonation process in
reinforced concrete structures. Part II: practical
applications. Cement and Concrete Research 35(35):
958–967.
Saetta AV, Schrefler BA and Vitaliani RV (1993) The carbonation
of concrete and the mechanism of moisture, heat, carbon
dioxide flow through porous materials. Cement and
Concrete Research 23(4): 761–772.
Saillio M, Baroghel-Bouny V and Barberon F (2014) Chloride
binding in sound and carbonated cementitious materials
with various types of binder. Construction and Building
Materials 68: 82–91.
Suryavanshi AK and Swamy RN (1996) Stability of Friedel’s salt
in carbonated concrete structural elements. Cement and
Concrete Research 26(5): 729–741.
Suryavanshi AK, Scantlebury JD and Lyon SB (1996)
Mechanism of Friedel’s salt formation in cements rich in
tri-calcium aluminate. Cement and Concrete Research
26(5): 717–727.
Tang L and Nilsson LO (1993) Chloride binding capacity and
binding isotherms of OPC pastes and mortars. Cement and
Concrete Research 23(2): 247–253.
Thomas MDA and Jones MR (1996) A critical review of service
life modeling of concretes exposed to chlorides. In
Concrete in Service of Mankind: Radical Concrete
Technology (Dhir RK and Hewlett PC (eds)). E & F N
Spon Ltd, London, UK, pp. 723–736.
 Magazine of Concrete Research
Volume 68 Issue 7
Villain G, Thiery M and Platret G (2007) Measurement
methods of carbonation profiles in concrete:
thermogravimetry, chemical analysis and
gammadensimetry. Cement and Concrete Research
37(8): 1182–1192.
Wang Y (2001) Mathematical Modelling of Chloride Ingress
into Concrete and Electro-chemical Chloride Removal
from Concrete. PhD thesis, Aston University,
Birmingham, UK.
Yoon IS, Copuroglu O and Park KB (2007) Effect of
global climate change on carbonation progress
Downloaded by [ Purdue Univ Lib TSS] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
Effect of carbonation on release of
bound chlorides in chloride-contaminated
concrete
Geng, Easterbrook, Liu and Li
of concrete. Atmospheric Environment 41(34):
7274–7285.
Yu SW, Sergi G and Page CL (1993) Ionic diffusion across an
interface between chloride-free and chloride-containing
cementitious materials. Magazine of Concrete Research
45(165): 257–261, http://dx.doi.org/10.1680/macr.1993.45.
165.257.
Yuan Q, Shi C, De Schutter G, Audenaert K and Deng D (2009)
Chloride binding of cement-based materials subjected to
external chloride environment – a review. Construction and
Building Materials 23(1): 1–13.
WHAT DO YOU THINK?
To discuss this paper, please submit up to 500 words to
the editor at journals@ice.org.uk. Your contribution will
be forwarded to the author(s) for a reply and, if con-
sidered appropriate by the editorial panel, will be pub-
lished as a discussion in a future issue of the journal.
363