Construction and Building Materials 112 (2016) 925–932

Contents lists available at ScienceDirect

Construction and Building Materials

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

Distance-associated chloride binding capacity of cement paste subjected

to natural diffusion
Zijian Song ⇑, Linhua Jiang, Ziming Zhang, Chuansheng Xiong
College of Mechanics and Materials, Hohai University, Nanjing 210098, PR China

h i g h l i g h t s

Specimens undergoing natural diffusion are used for measuring the chloride binding capacity.

Difference between conventional CBIs and natural-diffusion-determined CBIs is found.

CBPs are proposed to quantify the hydroxyl-dependent chloride binding capacity.

A hydroxyl-dependent chloride binding model is established.

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

Article history: Chloride binding capacity is usually quantified by chloride binding isotherms (CBIs). However, an impor-
Received 4 January 2016 tant factor influencing chloride binding capacity, namely the internal chemical environment (ICE), cannot
Received in revised form 21 February 2016 be represented in the conventional CBIs. Knowing the drawbacks of the conventional CBIs, this paper
Accepted 5 March 2016
straightly put the emphasis on the distance-associated chloride binding capacity. Specimens undergoing
Available online 9 March 2016
natural diffusion were used for measuring the chloride binding capacity at different distance. CBIs,
chloride binding profiles (CBPs), chemical profiles and microscopy result were performed to analyze
Keywords:
the rules and mechanisms. From the results of CBIs and CBPs, it is found that the conventional
Cement paste
Chloride binding
laboratory-determined CBIs would overestimate the chloride binding capacity beyond a certain distance
Distance-associated from the surface, and underestimate the chloride binding capacity near the surface. Based on the
Diffusion chemical profiles and the microscopy results, it is found that ICE changes greatly along with natural dif-
Model fusion, and meanwhile the chloride binding capacity at different distance develops discriminatively with
the ICE change. Plus, a hydroxyl-dependent chloride binding model is established to quantify the
distance-associated chloride binding capacity. The model explains the difference between the conven-
tional laboratory-determined CBIs and the natural-diffusion-determined CBIs, and it fits well with the
test results.
Ó 2016 Published by Elsevier Ltd.

1. Introduction chlorides generally exist in two main forms in concrete, including

Chloride-induced corrosion is one of the major causes of
degradation of reinforced concrete structures [1–4]. It is known
that a sufficient amount of chloride ions in concrete are able to
break the passive layers (c-Fe2O3) of reinforcements, thereby sub-
stantially degrading the durability and service life of reinforced
concrete structures [5]. According to the two-stage durability
model [6], the induction stage, also known as the longest and the
critical stage, is defined as a duration from the time chlorides
firstly penetrated into concrete to the time a threshold level of
chlorides reached the steel bar. At this stage, the penetrated
⇑ Corresponding author.
E-mail address: hhulhjiang@gmail.cn (Z. Song).
free chlorides and bound chlorides. It is a widespread view that
only free chlorides are immediately detrimental with regard to
the reinforcement corrosion risk though some bound chlorides
may be released at later ages [7–11]. As a result of binding, the
time needed for free chlorides’ reaching a threshold level has a
considerable growth [12,13]. Knowing this, chloride binding capac-
ity is commonly regarded as one of the most important factors
deciding the length of service life of reinforced concrete structures.
Chloride binding includes physical binding and chemical bind-
ing. Neither of the two binding processes is a result of single reac-
tion, but both contain various physical or chemical mechanisms
[7,14]. Chemical binding in an early aged cement hydrate system
mainly refers to the reaction between chlorides and unhydrated
C3A/C4AF to form a Friedel’s salt or its analogue [15,16]. In a mature

http://dx.doi.org/10.1016/j.conbuildmat.2016.03.020
0950-0618/Ó 2016 Published by Elsevier Ltd.
926 Z. Song et al. / Construction and Building Materials 112 (2016) 925–932
cement hydrate system, however, chemical binding is possibly due
to the anion exchange between chloride ions and AFm (Al2O3-
Fe2O3-mono) phases [17–19]. Actually, Friedel’s salt or its analogue
is also a specific kind of AFm phases. It is known that the AFm
phases have a layered structure, which contains a positively
charged octahedral principal layer [Ca2(Al,Fe)(OH)6]+ and a nega-
tively charged interlayer [x.nH2O] [19,20]. In cement chemistry,
the common species of the interlayer x anions are usually Cl (Frie-
del’s salt), 0.5CO2
3 (Monocarboaluminate), OH (Hydroxy-AFm),
0.5SO24 (Monosulfoaluminate), 0.5Cl + 0.25SO2
4 (Kuzel’s salt)
etc. [21–24]. An AFm crystal may also contain more than one spe-
cies of X anions to form solid solution [21]. The point is that the
hydrogen bonds between principal layer and interlayer anions
are weak and therefore the intruded chlorides can replace other
species of interlayer anions of AFm phases to form Friedel’s salt
or analogue in the mature cement hydrated system [18]. Physical
binding is neither unitary. Ramachandran pointed out three kinds
of binding with C-S-H: in a chemisorbed layer on hydrated calcium
silicates, present in the C-S-H interlayer spaces, and intimately
bound in the C-S-H lattice [25].
It is because of the various binding mechanisms that chloride
binding capacity is affected by many factors, e.g., content of alumi-
nates, C/S ratio of C-S-H gel, sulfate ion concentration, hydroxyl ion
concentration and cation type of pore solution [26–30]. In other
words, chloride binding capacity is strongly impacted by internal
chemical environment (abbreviated ‘‘ICE” below), which is known
as the composition of pore solution and solid phases surrounding
the location where chloride binding takes place. Likewise, some
external interactions being able to influence the ICE could also
impact the chloride binding, e.g. sulfate attack, leaching, carbona-
tion [7,31,32]. Even in a normal chloride diffusion process, ICE
would also gradually change, from the surface to the deep, along
with chloride transport so as to keep electrical neutrality.
Chloride binding is often quantified at thermodynamic equilib-
rium by chloride binding isotherms (CBIs), which are known as
the relationships between free and bound chloride ions over a range
of chloride concentrations at a given temperature [8]. CBIs are usu-
ally provided with clear mathematical formulae and therefore they
can be easily applied into the chloride transport models. Consider-
ing this, CBIs have attracted considerable attention from the
researchers. Up to present, four types of binding isotherms (i.e., lin-
ear, Langmuir, Freundlich and BET binding isotherm) have been
proposed to describe the chloride binding [33]. In order to deter-
mine suitable types and parameter values of CBIs, two main labora-
tory methods are frequently used in the existed studies [15,27,34].
According to the two methods, researchers need not to take sam-
ples from natural diffusion specimens but directly use powder or
thin disc specimens, which are soaked in different concentrations
of chloride solutions, instead. In that way, however, the chemical
environment of chloride binding will not match the actual ICE of
the concrete experiencing a natural diffusion. Likewise, the impact
of the gradual ICE variation on chloride binding capacity cannot be
precisely presented at different transport distance either. Therefore,
it is perhaps not very proper to straightly use the conventional
laboratory-determined CBIs to predict the chloride binding capacity
at different distance of the concrete undergoing natural diffusion.
Knowing the drawbacks of the conventional laboratory-
determined CBIs, this paper straightly put the emphasis on the
distance-associated chloride binding in natural diffusion. A com-
parison between the conventional laboratory-determined CBIs
and the natural-diffusion-determined CBIs is presented. A CBPs
(Chloride binding profiles) method is also proposed to clearly ana-
lyze the chloride binding capacity at different distance of cement
pastes undergoing chloride diffusion. The main purpose of this
paper is to contribute to improving the understanding of
distance-associated chloride binding capacity in a natural diffusion
process, and thereby promoting better evaluation of chloride bind-
ing capacity in practical engineering.
2. Materials and method
2.1. Materials
Ordinary Portland cement (OPC) was used and the detailed chemical composi-
tion of the cement is presented in Table 1. Neat cement pastes were prepared at
water-to-cement (w/c) ratios of 0.30, 0.40, 0.50 and 0.60.
2.2. Specimens preparation and immersion test
6 cylindrical specimens (50 mm in diameter and 100 mm in height) and 30 disc
specimens (50 mm in diameter and 3 mm in height) were casted for each w/c ratio.
After curing in saturated limewater at 20 ± 1 °C for 28 days, the specimens were
taken out for an immersion test.
The cylindrical specimens were used for measuring the chloride binding capac-
ity at different distance after the immersion test. Prior to the immersion test, the
cylindrical specimens underwent a surface treatment to implement the one-
dimensional diffusion. In this work, the side and bottom surfaces of each specimen
were sealed with epoxy resin, and only the top surface was kept unsealed. After
that, the cylindrical specimens were exposed to a 0.5 mol/L NaCl solution for a per-
iod of 180 d. The solution tanks are covered with polythene sheets to avoid evapo-
rations and carbonizations. In addition, the solutions were renewed every month
for maintaining the chloride concentrations.
The disc specimens were used to determine the CBIs in a conventional way (tak-
ing no account of the distance). They were averagely divided into five groups and
immersed in 0.1, 0.2, 0.3, 0.4 and 0.5 mol/L NaCl solutions, respectively. Considering
that the thin discs are possibly easy to reach saturation, a much shorter immersion
period, about one month, was set for these specimens. Polythene sheet covering
was also used here to avoid evaporations and carbonizations.
2.3. Measurement of free chloride and other chemical concentrations
After immersion, the 6 cylindrical specimens were cut layer-by-layer from the
exposed surface. The nominal thickness of each layer was 5 mm. The actual thick-
ness of each layer, excluding the cut abrasion, is controlled within 3–4 mm. Before
pressing, the specimen layers were converted to a saturated, surface-dry condition.
Then each specimen layer was crushed into four or five small pieces. One or two
pieces of each layer were randomly selected and temporarily stored for the follow-
ing total chloride concentration measurement, and the rest pieces were used here
for pressing the pore solution.
The small pieces from the same-distance layers of the six cylinders were
stacked together and put into a presser. At a pressure up to 300 MPa, the pore solu-
tion was then slowly extracted out. When sufficient pore solution was obtained, it
was divided into two parts and separately stored in two test-tubes. For one tube of
solution, the free chloride concentration was measured using an automatic titrime-
ter. For the other, the chemical concentration of the pore solution was measured by
flame photometry. Prior to titration, the PH value was also tested by a glass PH elec-
trode in the first tube of solution.
The disposal of the 30 disc specimens was much easier because they need not
be re-cut before pressing. After retaining samples for the total chloride concentra-
tion measurement, the crushed discs in the same group were straightly stacked in
the presser to extract the pore solution. Later, the free chloride concentration, the
PH value and the other chemical concentrations were measured using the above
methods.
2.4. Measurement of total chloride concentrations
The restored specimen pieces were ground into powder using a mortar and pes-
tle. Then an amount of powdered specimen was soaked in dilute nitric acid solution
to determine the total chloride concentration. Each test result was given as an aver-
age of three replicates.
2.5. Microscopy test
A small flake sample was typically taken from the cylinder specimen with 0.5
w/c ratio. The flake was selected nearly parallel to the diffusion direction and per-
pendicular to the exposed surface. With the HITACHI S-3400N Scanning Electron
Microscope (SEM), the micro morphology of the cement hydration was analyzed
at different distances from the exposed surface. Besides, a linear Energy Dispersive
X-ray Spectroscopy (EDS) scanning was also carried out in distance of 0–3 mm to
analyze the chemical element profile in the surface region.
 Z. Song et al. / Construction and Building Materials 112 (2016) 925–932 927

Table 1
Composition of cement (wt.%).

Composition SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O TiO2 Mn2O3 LOI
Content 21.52 5.13 5.25 61.86 1.46 2.28 0.21 0.83 0.14 0.10 0.05

3. Results and discussion
3.1. CBIs
As previously mentioned, half the CBIs were determined
according to the conventional laboratory methods as a comparison.
Fig. 1 presents the conventional laboratory-determined CBIs for
cement pastes of four different w/c ratios. In Fig. 1, the amount
of bound chlorides is an average of three replicates, so the error
bars are used to represent the measured ranges. The two ends of
each error bar represent the max and the min value of the three
replicates, respectively. The plots of Fig. 1 show that the amount
of bound chlorides increases with the free chloride concentration
in the range of 0–0.6 mol/L for all the paste materials used in this
investigation. It also can be clearly seen that the relationship
between free and bound chloride ions is nonlinear, which is in
accordance with the results of most researchers [15,34,35]. As
reported by Tang and Nilsson, the Freundlich binding isotherm
(cb = acbf ) is the most frequently-used binding isotherm for chlo-
rides in high concentration (free chloride ion concentration
>0.01 mol/L) [34]. In the present paper, this type of binding iso-
therm is also used for fitting. The fitting curves, the parameter val-
ues and the correlation coefficient (r2) for the four pastes are all
given in Fig. 1. It can be seen that the test results fit well with
the Freundlich binding isotherms. In all cases, the correlation coef-
ficients are >95%. As observed from the fitting curves, the slopes
decrease markedly along x-axis. This indicates that the chloride
chloride concentration in the range of 0–0.6 mol/L for all the paste
materials; 2) the chloride binding capacity increases with the w/c
ratio. However, the difference is also obvious. It can be intuitively
seen that the relationships between free and bound chloride ions
here are much more similar to the linear rule comparing with
those in Fig. 1. Considering that, both the linear and the Freundlich
binding isotherms are used here for fitting, which are presented in
Fig. 2. The values of parameters and the correlation coefficient (r2)
for the four pastes are also given in Fig. 2. From the fitting results, it
is found that the test plots match well with both the linear and the
Freundlich binding isotherms. The correlation coefficients of the
two types of fitting curves are almost >95%. Even if the Freundlich
binding isotherm is adopted, it can be found that the value of
parameter b becomes much larger than that in Fig. 1. This evidence
confirms that the binding rules obtained in Fig. 2 are strictly more
similar to the linear rule (b up to 1 = linear). This phenomenon
explains why a few of researchers found similar linear binding iso-
therms in field engineering or natural diffusion test. Mohammed
and Sandberg considered that the linear rules obtained in field
engineering and natural diffusion test are due to the leaching of
hydroxyl ions [36,37]. The specific mechanisms will be discussed
in the later sections.
(a) 1.2
y = 1.4104x0.6943, R² = 0.9822
Bound chloride (% paste sample)

binding capacity decreases markedly with the increasing free chlo- 1

y = 1.188x0.7064, R² = 0.9920
ride concentration. Plus, it is still necessary to mention that the
y = 1.2384x0.8227, R² = 0.9925
chloride binding capacity increases with the w/c ratio though this 0.8
has been a widespread viewpoint. y = 0.9683x0.8339, R² = 0.9959
The rest half of the CBIs were determined using natural diffu- 0.6
sion specimens. As introduced in Section 2.3, the test samples were
taken from different distances of six natural diffusion specimens. 0.4
w/c0.3
Fig. 2 presents the CBIs determined by this method for the cement w/c0.4
pastes of different w/c ratios. It can be easily found that most rules w/c0.5
0.2
observed in Fig. 2 are similar to those observed in Fig. 1, namely, 1)
w/c0.6
the amount of bound chlorides increases markedly with the free
0
0 0.2 0.4 0.6 0.8 1 1.2
Free chloride (mol/L pore solution)
1 (b) 1.2
y = 0.9811x0.3274, R² = 0.9724 y = 1.8928x, R² = 0.9131
Bound chloride (% paste sample)

0.9
Bound chloride (% paste sample)

y = 1.6039x, R² = 0.9556
y = 0.9430x0.3629, R² = 0.9595 1
0.8 y = 1.4336x, R² = 0.9846
0.7 y = 0.8519x0.4642, R² = 0.9739
y = 1.1717x, R² = 0.9931
0.8
0.6 y = 0.7009x0.4536, R² = 0.9367

0.5 0.6
0.4 w/c0.3
w/c=0.3
0.4 w/c0.4
0.3 w/c=0.4
w/c0.5
0.2 w/c=0.5
0.2 w/c0.6
0.1 w/c=0.6

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Free chloride (mol/L pore solution) Free chloride (mol/L pore solution)

Fig. 1. Conventional laboratory-determined CBIs for cement pastes of different w/c Fig. 2. Natural-diffusion-determined CBIs for cement pastes of different w/c ratios,
ratios, and comparison with Freundlich binding isotherms. and comparison with (a) Linear, (b) Freundlich binding isotherms.
928 Z. Song et al. / Construction and Building Materials 112 (2016) 925–932

3.2. CBPs
Despite the numerous advantages, CBIs still have some draw-
backs. For example, they cannot intuitively present how the chlo-
ride binding capacity varies with the distance from the exposed
surface to the deep. Thus, CBPs (Chloride binding profiles) are pro-
posed here to improve the drawbacks. In this paper, CBP is defined
as a relationship between chloride binding capacity and the dis-
tance over a range from the exposed surface to the deep. Before
applying the CBPs, an important issue facing us is how to quantita-
tively characterize the chloride binding capacity. From our point of
view, there are two possible forms to settle it. One is to use the
ratio of bound to total chloride concentration (cb/ct) and the other
is to use the ratio of bound to free chloride concentration (cb/cf). By
unifying the units, the two forms are given in Eqs. (1) and (2),
respectively.
cb
r cb =ct ¼  100%
cb þ cf M Cl um =qw
cb
r cb =cf ¼  100%
cf M Cl um =qw
where, cb is the bound chloride content (%), cf is the free chloride
concentration (mol/L), Mcl is the molar mass of chloride ions
(35.45 g/mol), um is the mass porosity (%), qw is the water density
(1000 g/L).
By using the two forms given in Eqs. (1) and (2), the CBP figure
is presented. Fig. 3 shows the measured CBPs and the ‘CBI-
recalculated’ CBPs for the cement pastes with different w/c ratios.
The solid symbols represent the measured CBPs which were
obtained in the natural diffusion specimens, and the hollow ones
are the CBI-recalculated CBPs which were recalculated on the basis
of the conventional laboratory-determined CBIs and the measured
(a) 1.05
1.00
Chloride binding capacity, cb/ct
free chloride profiles. In Fig. 3(a), the ratio of bound to total chlo-
ride concentration (Eq. (1)) is used to characterize the chloride
binding capacity. It is found that the chloride binding capacity gen-
erally increases with the distance from the exposed surface,
whether in the measured or the calculated CBPs. This kind of
increase is generally in accordance with the previously mentioned
fact that the chloride binding capacity increases with the decreas-
ing free chloride concentration, considering the free chloride con-
centration also decrease with the distance. It is also noticed that
the increasing rates of the CBI-calculated CBPs are larger than
those of the measured CBPs. In detail, the chloride binding capac-
ities obtained from the CBI-calculated CBPs within a certain dis-
tance from the surface are a little larger than those obtained
from the measured CBIs, and beyond the distance the rule is oppo-
site. Considering the fact that the CBI-recalculated CBPs were recal-
culated from the CBI data, it is safe to draw the conclusion that CBIs
would overestimate the chloride binding capacity beyond a dis-
ð1Þ tance from the exposed surface of OPC pastes, and a little underes-
timate the chloride binding capacity near the surface.
In Fig. 3(b), the ratio of bound to free chloride concentration
(Eq. (2)) is used. As can be seen, the difference between the mea-
ð2Þ
sured CBPs and the CBI-calculated CBPs at the deep distances is
much more obvious. The front cb/cf in the CBI-calculated CBPs is
about 3–8 times of that in the measured CBPs.
3.3. Chemical profiles of pore solution
Fig. 4 typically shows the chemical profiles of pore solution
pressed from different distances of the w/c 0.5 cement paste.
Among the data, the OH concentrations were converted from
the pH values while the others were directly determined using
the previously mentioned methods.
Owing to the concentration difference, the K+ and OH ions near
the surface greatly leached out. Thus, there was a sharp concentra-
tion drop of these ions near the surface. It can be seen from Fig. 4
that the K+ and OH concentrations decline gradually from the

w/c0.3 meas. deep to the surface. Likewise, it seems reasonable to suppose that
0.95 w/c0.3 calc. the Ca2+ profile also has a descendent front as observed in the K+
0.90
w/c0.4 meas.
and OH profiles. However, the evidence shows that the Ca2+ con-
0.85 centration does not decline but rather has a little growth within a
w/c0.4 calc.
0.80 certain distance. In fact, the similar phenomenon has been
w/c0.5 meas. observed by the previous researchers without giving an explana-
0.75
0.70
w/c0.5 calc. tion [38]. It is known that the leaching of the Ca2+ and OH ions
w/c0.6 meas. would lead to the dissolution/desorption of portlandite and other
0.65
calcium compounds. The concentration of original Ca2+ ions is
0.60 w/c0.6 calc.
too small (about 0.01 mol/L in this paper) to contend against to
0.55
0 10 20 30 40 50 the dissolution/desorption process [39], so the Ca2+ concentration
Distance from the exposed surface (mm)
0.45
(b) 80
0.4
Chloride binding capacity , cb/cf

70 w/c0.3 meas.
Concentraon (mol/L)

0.35
60 w/c0.3 calc.
0.3 OH-
50 w/c0.4 meas.
0.25 Ca2+
w/c0.4 calc.
40 Na+
0.2
w/c0.5 meas.
30 0.15 K+
w/c0.5 calc.
20 0.1
w/c0.6 meas.
10 w/c0.6 calc. 0.05
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Distance from the exposed surface (mm) Distance (mm)

Fig. 3. Measured and ‘CBI-recalculated’ CBPs for cement pastes with different w/c Fig. 4. Chemical profiles in pore solution pressed from different distances of w/c 0.5
ratios using (a) cb/cf, (b) cb/cf to represent chloride binding capacity. cement paste.
 Z. Song et al. / Construction and Building Materials 112 (2016) 925–932
in the certain distance becomes higher than the original value
under the combined action of leaching and dissolution/desorption.
Owing to the concentration difference, the Na+ ions in the
source solutions also penetrated into the cement pastes along with
the Cl ions. It is found that the Na+ ions penetrated much slower
than Cl. This is much different from the rule obtained in the dif-
fusion process of NaCl electrolyte solution. In the electrolyte solu-
tion, Na+ ions will diffuse at the same speed as Cl ions so as to
keep the electrical neutrality [40]. In the cement hydrate system,
not only the Na+ ions but also other species of ions and solid phases
will have a response to the chloride diffusion. Thus, it is perhaps
nothing unusual to find the Na+ ions penetrated much slower than
Cl in cement hydrate systems considering the fact that the bulk
diffusion coefficient of Na+ itself is much smaller than Cl.
It can be concluded that the chemical composition of pore
solution changes discriminatively and obviously at different
distance when natural chloride diffusion takes place. The change
of the pore solution composition is also potential to influence the
solid phases. That is to say the entire ICE will change along with
chloride diffusion. It is because of the ICE change that the chloride
binding capacity also changes discriminatively at different dis-
tance. In Section 3.5, a model is built up to describe chloride bind-
ing capacity considering the change of an important ICE factor.
3.4. Microscopy test
Fig. 5 presents the morphology of cement hydrates of the w/c
0.5 cement paste. It can be seen from Fig. 5 that the microstructure
of the near-surface hydrates changes a lot after natural diffusion, as
expected. In Fig. 5(a), a narrow degraded zone (about 0–0.2 mm
from the surface) is found near the surface at low magnification
(30 times). Fig. 5(b) presents the morphology of the degraded zone
at a higher magnification (2.0 k times). It is found that the
degraded zone is totally amorphous. None of the familiar crystals
can be observed in that zone. For comparison, Fig. 5(c) presents
the morphology of the undegraded zone at 2.0 k magnification.
The common crystals existed in cement hydrates can be easily
found in the undegraded zone, such as the hexagonal structural
Fig. 5. Morphology of cement hydrates of w/c 0.5 cement paste: (a) overview, (b) degraded zone, (c) transitional zone, (d) undegraded zone.
929
crystals (AFm, CH) and the needle-like structural crystals (AFt
[19]). Moreover, the features of the crystals in this zone remain
intact and the crystal edges are also clear. Fig. 5(d) presents the
morphology of the transitional zone between degraded and unde-
graded zone. Most familiar crystals can also be found here. How-
ever, in this zone, the AFt crystals almost disappear and the
edges of the crystals become somewhat unclear.
Fig. 6 presents the EDS linear scanning results for the w/c 0.5
cement paste. Owing to the non-uniformity of cement paste, the
chemical element profiles are found a little disordered, especially
in some small-scale regions. At large enough scales, however, clear
trends can be observed from the large amounts of data. As
observed in Fig. 6, the calcium content generally decreases from
the deep to the surface of the specimen. In the degraded zone, this
kind of decrease is more evident. This indicates that the leaching of
calcium ions in pore solution had been influencing the solid phases
near the surface. Further, it can be reckoned that the portlandite,
C-S-H gels, AFm phases and other calcium compounds near the
surface had been largely dissolved. Due to the dissolution of C-S-
H gels and AFm phases, chloride ions lost their binding matrix
and therefore the content of bound chlorides essentially decreased.
Thus, it is found in Fig. 6(c) that the chlorine profile has a drop near
the surface. It is noticed that, if only considering the impact of the
free ions (e.g. the hydroxyl ions) on chloride binding, this kind of
chloride drop would be unexpected. It is known that, as a result
of leaching, the hydroxyl ion concentration near the surface largely
decreased [41,42]. The concentration decrease would also reduce
the binding competition of hydroxyl ions with chloride ions, and
thereby promoting the chloride binding [17,43]. According to this
theory, the content of bound or total chlorides near the surface
should be the largest. Considering the opposite rule observed in
EDS results, it can be assumed that the solid phases (being an
important element of ICE) will also impact the chloride binding
capacity.
In this investigation, the degrading degree is negligibly small
due to the short immersion duration and perhaps the non-flowing
solution. Therefore, the zone where the bound chloride is evidently
influenced by the solid phase degradation (i.e. the chloride drop zone)
930 Z. Song et al. / Construction and Building Materials 112 (2016) 925–932
Fig. 6. EDS linear scanning for w/c 0.5 cement paste (a) scanning position, (b) oxygen profile, (c) chlorine profile, (d) calcium profile.
is also small (about 0.75 mm in depth). It is noticed that the small
chloride drop zone cannot even be presented by the traditional
sampling and titration method since the coarse sampling intervals
(usually 3–5 mm). In fact, in most cases, the chloride drop zone is in
small depth and can be ignored comparing with the long chloride
diffusion depth. This does not mean that the impact of the solid
phase degradation on chloride binding is small enough to be
ignored in all cases. In some long duration tests, the chloride drop
zone has been found much obvious [44]. Perhaps in the flowing
or aggressive environment (e.g. the flowing seawater, NH4Cl-rich
underground water), the impact of the solid phase degradation on
chloride binding would be much stronger. Anyway, the small but
truly existed chloride drop zone found in EDS results illustrate that
the chloride binding capacity is definitely impacted by solid phase
degradation (or change).
3.5. Modeling
Chloride binding capacity is strongly impacted by ICE, which
contains many ions and solid phases in the cement hydrate system.
However, the influence of the ICE factors is not able to be repre-
sented in the conventional CBIs. Thus, it is urgent to build up a
multifactorial chloride binding model to apply it into practical
use. In most situations, it is difficult and unnecessary to consider
all the ICE factors when describing the chloride binding capacity.
Before creating an applicable model, the preliminary work to be
done is to analyze the influence degrees of the multi ICE factors
in line with the specific situation, and then select one or a few of
the ICE factors as the determining factors for modeling. From our
point of view, it can still be a kind of progress even if only one
ICE factor is taken into account.
As previously mentioned, in most cases, the impact of solid
phase variation is very small and can be ignored. Considering the
anion exchange mechanism [17,44], the hydroxyl ion concentra-
tion is perhaps the most important factor influencing the
distance-associated chloride binding among the ICE factors. Thus,
a hydroxyl-dependent chloride binding model is established here.
As we know, the ratio of free chloride concentration to hydroxyl
concentration is frequently used to estimate the critical value for
steel bar corrosion [45]. Cooperating with this type of critical value,
the hydroxyl-dependent chloride binding model is probably very
compatible to predict the service life of reinforced structures.
Eq. (3) shows the general reaction of chloride binding under the
chloride-hydroxyl ion exchange mechanism.

R  OH þ Cl () R  Cl þ OH ð3Þ
where R-OH represents the binding matrix including CSH gels and
AFm phases, R-Cl represents all the solid chloride phases. On the
basis of the mass action law and the empirical adsorption kinetics,
the positive and negative reaction rates of Eq. (3) are assumed as:
r1 ¼ k1 cROH cm
f ; r 2 ¼ k2 c b cOH
n
ð4Þ
where, cR-OH is the content of binding matrix; cOH is the hydroxyl
ion concentration; k1, k2, m and n are constants. When reaching
equilibrium, the content of bound chlorides can be calculated by
letting r1 = r2. Eq. (5) shows the detailed formula and the simplify-
ing process. The term in the box is the final model used in the pre-
sent paper, which is deduced on the basis of the understanding that
the content of binding matrix changes little with chloride binding.
k1 c m
f c ROH Cm C OH const:
cb ¼  k Cn f  ƒƒƒƒƒƒ!
ƒƒƒƒƒƒ cb ¼ acbf ð5Þ
k2 cnOH OH
It is found that, if the hydroxyl concentration is constant, the
model in the box is the same as the Freundlich binding isotherm.
This more or less reflects the rationality of the hydroxyl-
dependent model.
 Z. Song et al. / Construction and Building Materials 112 (2016) 925–932
0.50
0.45
0.40
0.35
0.30 Measured plots:
cOH-
Fitting curve:
0.25
y = 0.125x0.7087
0.20 R² = 0.9047
0.15
0.10
0.05
0 2 4 6 8
cfm/cb
Fig. 7. Fitting curve of hydroxyl-dependent chloride binding model.
The final model clearly shows the impact of hydroxyl concen-
tration on chloride binding capacity and explains why the
natural-diffusion-determined CBIs are different from the conven-
tional CBIs. In the conventional laboratory methods, the hydroxyl
concentrations of the parallel disc specimens are similar to each
other after the same duration of leaching, so the Freundlich bind-
ing isotherm is usually found by using these methods. In the natu-
ral diffusion method, the hydroxyl concentration increases with
the distance while the free chloride concentration decreases with
the increasing distance. Therefore, the larger the free chloride con-
centration is, the smaller the decreasing scope of the chloride bind-
ing capacity is. Thus, the natural-diffusion determined CBIs are
found more linear.
According to Eq. (5), the parameter m is the same as the param-
eter b. Thus, the value of m can be evaluated from the conventional
laboratory-determined CBIs. Then the parameters k and n can be
solved by fitting the equation to the experimental results deter-
mined using natural diffusion specimens. For the w/c 0.5 cement
paste, the m is about 0.3629. Using cm f /cb as the independent vari-
able and cOH as dependent variable, Eq. (5) are converted into a
simple power function for fitting. The measured plots, the fitting
curve, the obtained power function and the correlation coefficient
(r2) are all given in Fig. 7. It can be seen that the new model fit well
with the test results (r2 > 90%). From the fitting result, the new
model for the w/c 0.5 cement paste is:
c0:3629
f external chloride environment – a review, Constr. Build. Mater. 23 (1) (2009)
cb ¼ 0:0532
c1:4110
OH
4. Conclusion
This study dealt with the distance-associated chloride binding
capacity of cement pastes including four different w/c ratios from
0.30 to 0.60. Specimens undergoing natural diffusion were used for
measuring the chloride binding capacity at different distance. CBIs,
CBPs, chemical profiles and microscopy results were performed to
analyze the influence rules and mechanisms of distance on chlo-
ride binding capacity, and then the conclusions are presented
below.
The conventional laboratory-determined CBIs fit well with the
Freundlich binding isotherms. In comparison, the natural-
diffusion-determined CBIs are much more similar to the linear
rules. By analyzing the CBPs results, it is found that the conven-
tional laboratory-determined CBIs would overestimate the chlo-
ride binding capacity beyond a certain distance from the exposed
surface of OPC pastes, and underestimate the chloride binding
capacity near the surface.
On the basis of the chemical profiles and the microscopy results,
it is safe to draw the conclusion that the entire ICE changes when
931
natural diffusion takes place. It is because of the change of the ICE
that the chloride binding capacity at different distance also devel-
ops discriminatively. In this study, the hydroxyl concentration is
considered to be the most important factor among the ICE factors
influencing the chloride binding capacity of the specimens under-
w/c 0.5 going natural diffusion. Therefore, a hydroxyl-dependent chloride
binding model is established to quantify the distance-associated
chloride binding capacity. The model explains the difference
between the conventional laboratory-determined CBIs and the
CBIs determined using natural diffusion specimens, and it fits well
with the test results.
Acknowledgements
The authors wish to acknowledge the financial support by the
National Natural Science Foundation of China under Project Nos.
51278167 and 51479051 as well as the Natural Science Foundation
of Jiangsu province under Project No. BK20131374. The supports of
the Experiment Center of Mechanics and Materials in Hohai
University have been most helpful and are appreciated. The State
Key Laboratory of Hydrology-water Resources and Hydraulic Engi-
neering is also appreciated for providing some important test
equipments.
References
[1] M. Dobrescu, M. Vasilescu, Reinforcing bar corrosion for buildings
construction, Adv. Mater. Res. 1114 (2015) 219–223.
[2] J. Zhu, Y. Zhang, D. Zhao, Durability assessment of an RC railway bridge pier
under a chloride-induced corrosion environment, Fifth International
Conference on Transportation Engineering (2015).
[3] L. Zeng, R. Song, Controlling chloride ions diffusion in concrete, Sci. Rep. 3
(2013) 1–7.
[4] Zijian Song, Linhua Jiang, Jianzhong Liu, Jiaping Liu, Influence of cation type on
diffusion behavior of chloride ions in concrete, Constr. Build. Mater. 99 (2015)
150–158.
[5] E. Badogiannis, E. Aggeli, V.G. Papadakis, S. Tsivilis, Evaluation of chloride-
penetration resistance of metakaolin concrete by means of a diffusion–Binding
model and of the k-value concept, Cement Concr. Compos. 63 (2015) 1–7.
[6] K. Tuutti, Corrosion of Steel in Concrete, Swedish Cement Concrete Research
Institute, Stockholm, 1982.
[7] V. Baroghel-Bouny, X. Wang, M. Thiery, M. Saillio, F. Barberon, Prediction of
chloride binding isotherms of cementitious materials by analytical model or
numerical inverse analysis, Cem. Concr. Res. 42 (9) (2012) 1207–1224.
[8] Q. Yuan, C. Shi, et al., Chloride binding of cement-based materials subjected to
ð6Þ 1–13.
[9] A. Neville, Chloride attack of reinforced concrete: an overview, Mater. Struct.
28 (176) (1995) 63–70.
[10] A.K. Suryavanshi, J.D. Scantlebury, S.B. Lyon, Corrosion of reinforcement steel
embedded in high water–cement ratio concrete contaminated with chloride,
Cement Concr. Compos. 20 (3) (1998) 263–269.
[11] B. Reddy, G.K. Glass, P.J. Lim, N.R. Buenfeld, On the corrosion risk presented by
chloride bound in concrete, Cem. Concr. Compos. 24 (1) (2002) 1–5.
[12] P.F. McGrath, Development of test methods for predicting chloride penetration
into high performance concrete (Ph.D. thesis), Department of Civil
Engineering, University of Toronto, Canada, 1996.
[13] B. Martin-Perez, H. Zibara, R.D. Hooton, M.D.A. Thomas, A study of the effect of
chloride binding on service life predictions, Cem. Concr. Res. 30 (2000) 1215–
1223.
[14] H.A. Berman, Determination of chloride in hardened Portland cement paste,
mortar, and concrete, J. Mater. 7 (3) (1972) 330–335.
[15] A. Delagrave, J. Marchand, J. Olliver, S. Julien, K. Hazrati, Chloride binding
capacity of various hydrated cement systems, Adv. Cem. Based Mater. 6 (1997)
28–35.
[16] H. Justnes, A review of chloride binding in cementitious systems, Nord. Concr.
Res. 21 (1998) 1–6.
[17] A.K. Suryavanshi, J.D. Scantlebury, S.B. Lyon, Mechanism of Friedel’s salt
formation in cements rich in tri-calcium aluminate, Cem. Concr. Res. 26 (5)
(1996) 717–727.
[18] Y. Chen, Z. Shui, W. Chen, G. Chen, Chloride binding of synthetic Ca–Al–NO3
LDHs in hardened cement paste, Constr. Build. Mater. 93 (2015) 1051–1058.
[19] H.F.W. Taylor, Cement Chemistry, 2nd ed., Thomas Telford Publishing, London,
1997.
[20] L.G. Baquerizo, T. Matschei, K.L. Scrivener, M. Saeidpour, L. Wadsö, Hydration
states of AFm cement phases, Cem. Concr. Res. 73 (2015) 143–157.
932 Z. Song et al. / Construction and Building Materials 112 (2016) 925–932
[21] Luis G. Baquerizo a, Thomas Matschei a, Karen L. Scrivener b, Mahsa Saeidpour
c, Lars Wadsö c, Hydration states of AFm cement phases, Cem. Concr. Res. 73
(2015) 143–157.
[22] R. Allmann, Refinement of the hybrid layer structure [Ca2Al(OH)6]+[1/
2SO43H2O], Neues Jahrbuch für Mineralogie Monatshefte 136–144 (1997).
[23] T. Matschei, B. Lothenbach, F.P. Glasser, The AFm phase in Portland cement,
Cem. Concr. Res. 37 (2007) 118–130.
[24] M. Francois, G. Renaudin, O. Evrard, A cementitious compound with
composition 3CaOAl2O3CaCO311H2O, Acta Crystallogr., C 54 (9) (1998)
1214–1217.
[25] V.S. Ramachandran, Possible states of chloride in the hydration of tricalcium
silicate in the presence of calcium chloride, Mater. Struct. 4 (1) (1971) 3–12.
[26] A.K. Suryavanshi, J.D. Scantlebury, S.B. Lyon, The binding of chloride ions by
sulphate resistant cement, Cem. Concr. Res. 25 (3) (1995) 581–592.
[27] C. Arya, N.R. Buenfeld, J.B. Newman, Factors influencing chloride binding in
concrete, Cem. Concr. Res. 20 (2) (1990) 291–300.
[28] D.F.H. Rasheeduzzafar, S.S. Al-Saadoun, A.S. Al-Fahtani, F.H. Dakhil, Effect of
tricalcium alumina content of cement on corrosion of reinforcing steel in
concrete, Cem. Concr. Res. 20 (5) (1990) 723–738.
[29] J.J. Beaudoin, V.S. Ramachandran, R.F. Feldman, Interaction of chloride and C–
S–H, Cem. Concr. Res. 20 (6) (1990) 875–883.
[30] P.W. Brown, S. Badger, The distributions of bound sulfates and chlorides in
concrete subjected to mixed NaCl, MgSO4, NaSO4 attack, Cem. Concr. Res. 30
(2000) 1535–1542.
[31] A.K. Suryavanshi, R. Narayanswamy, Stability of Friedel’s salt in carbonation in
carbonated concrete structural elements, Cem. Concr. Res. 26 (5) (1996) 717–
727.
[32] A. Neville, Chloride attack of reinforced concrete: an overview, Mater. Struct.
28 (176) (1998) 63–70.
[33] Y. Zhang, Mechanics of chloride ions transportion in concrete (Ph.D. thesis),
College of Civil Engineering and Architecture, Zhejiang University, Hangzhou,
China, 2008.
[34] L. Tang, L.-O. Nilsson, Chloride binding capacity and binding isotherms of OPC
pastes and mortars, Cem. Concr. Res. 23 (2) (1993) 247–253.
[35] W. Sergi, S.W. Yu, C.L. Page, Diffusion of chloride and hydroxyl ions in
cementitious materials exposed to a saline environment, Mag. Concr. Res. 44
(158) (1992) 63–69.
[36] T.U. Mohammed, H. Hamada, Relationship between free chloride and total
chloride contents in concrete, Cem. Concr. Res. 33 (2003) 1487–1490.
[37] P. Sandberg, Studies of chloride binding in concrete exposed in a marine
environment, Cem. Concr. Res. 29 (1999) 473–477.
[38] E. Samson, J. Marchand, J.L. Ropert, J.P. Bournazel, Modeling ion diffusion
mechanisms in porous media, Int. J. Numer. Meth. Eng. 46 (12) (1999) 2043–
2060.
[39] L. Jiang, Z. Song, H. Yang, Q. Pu, Q. Zhu, Modeling the chloride concentration
profile in migration test based on general Poisson Nernst Planck equations and
pore structure hypothesis, Constr. Build. Mater. 40 (1) (2013) 596–603.
[40] Y. Li, J. Lu, Theory of Electrolytic Solution, Tsinghua University Press, Beijing,
2005.
[41] Y. Maltaisa, E. Samsona, J. Marchanda, Predicting the durability of Portland
cement systems in aggressive environments—laboratory validation, Cem.
Concr. Res. 34 (2004) 1579–1589.
[42] K. Yokozeki, K. Watanabe, N. Sakata, N. Otsuki, Modeling of leaching from
cementitious materials used in underground environment, Appl. Clay Sci. 26
(1) (2004) 293–308.
[43] J. Tritthart, Chloride binding in cement II. The influence of the hydroxide
concentration in the pore solution of hardened cement paste on chloride
binding, Cem. Concr. Res. 19 (5) (1989) 683–691.
[44] B. Johannesson, K. Yamada, L.O. Nilsson, Y. Hosokawa, Multi-species ionic
diffusion in concrete with account to interaction between ions in the pore
solution and the cement hydrates, Mater. Struct. 40 (7) (2007) 651–665.
[45] Y. Wang, G. Fang, W. Ding, N. Han, Xing F, Self-immunity microcapsules for
corrosion protection of steel bar in reinforced concrete, Sci. Rep. 5 (2015).