Construction and Building Materials 165 (2018) 801–811

Contents lists available at ScienceDirect

Construction and Building Materials

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

Self-healing properties of cement-based and alkali-activated slag-based

fiber-reinforced composites
Huy Hoàng Nguyễn a, Jeong-Il Choi a, Keum-Il Song a, Jin-Kyu Song a, Jungwon Huh b, Bang Yeon Lee a,⇑
a
School of Architecture, Chonnam National University, 77 Yongbong-ro, Buk-gu, Gwangju 61186, Republic of Korea
b
Department of Ocean Civil Engineering, Chonnam National University, 50 Daehak-ro, Yeosu 59626, Republic of Korea

h i g h l i g h t s

The self-healing properties of cement-based and slag-based fiber-reinforced composites are investigated.

Slag-based composites have advantages in terms of a reduction of relative crack width.

Slag-based composites have low resonant frequency recovery.

Calcium carbonate is the dominant healing material for cement-based and slag-based composites.

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

Article history: This paper presents an experimental study of the self-healing properties of cement-based and alkali-
Received 30 October 2017 activated slag-based fiber-reinforced composites with controlled crack width. Two types of binder, i.e.
Received in revised form 2 January 2018 cement and alkali-activated slag-based polyethylene fiber-reinforced composites with identical water-
Accepted 3 January 2018
to-binder ratios, were designed. Compressive strength and uniaxial tension tests were performed to mea-
Available online 20 February 2018
sure the mechanical properties of the composites, and the self-healing performance was investigated by
observation of the crack width and by measuring resonance frequency. Scanning electron microscopy and
Keywords:
energy-dispersive X-ray spectroscopy were also adopted to analyze the morphology and chemical com-
Cement
Composite
position of the healing materials. The test results showed that alkali-activated slag-based composites
Crack width have advantages compared to cement-based composites in terms of a reduction of relative crack width;
Fiber however, cement-based composites have a higher resonant frequency recovery than alkali-activated slag-
Self-healing based composites. It is also observed that calcium carbonate is the dominant healing material of cement-
Slag based and slag-based composites.
Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction
Along with the development of humanity, upgrading infrastruc-
ture is a strategically important issue. Furthermore, improving the
quality of concrete structures is a matter of extreme urgency. Con-
crete is the world’s most widely used construction material on the
basis of its reliability and commercial value [1]. However, the dete-
rioration of concrete in term of cracks happening over time is one
of the major factors causing a reduction of the quality of concrete
structures. Cracks on concrete structures can occur during the ser-
vice life due to external and internal factors such as over-load,
environmental exposure, shrinkage, or design error [2]. The cost
of repair and maintenance of construction projects is quite large,
and thus concepts to develop a new building material that recovers
⇑ Corresponding author.
E-mail address: bylee@jnu.ac.kr (B.Y. Lee).
damage automatically to ensure the service life have been exten-
sively studied in recent years [3].
A self-healing characteristic is a universal phenomenon that
occurs in the human body, and occurs naturally without being
affected by any external factors [3]. Concrete is also a special mate-
rial with the ability to heal cracks by itself under natural conditions
over time [2]. The target of designing autogenic healing concrete is
to overcome problematic constructions to give a longer lifespan.
Neville (1995) and Hearn (1997) pointed out that the tiny cracks
in concrete could be sealed completely under a moist environment
due to the delayed hydration of cement [4,5]. Edvardsen (1999)
mentioned that calcium carbonate crystals are the main healing
material within the crack width [6]. From a review of the literature
[4–6], there are two primary definitions that should be noted on
the self-healing mechanism of deteriorating concrete structures,
autogenous healing and engineered repair. Autogenous healing is
a phenomenon that takes place by a natural process [7]. It can be

https://doi.org/10.1016/j.conbuildmat.2018.01.023
0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
802 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811
described by the hydration process of the unhydrated cementitious
materials, blockage of cracks caused by solid substances in the
water, swelling of CASAH in the crack flanks, and crystallization
of calcium carbonate [8]. Meanwhile, engineered repair is activated
by artificial healing materials or through intentional methods from
human intervention to accomplish the healing process [7,9]. There
are various approaches to improve the self-healing capacity of a
concrete structure that has suffered damage including the use of
bacteria, chemical admixtures, or microencapsulation [10–12].
In recent years, engineered cementitious composite (ECC) has
been recognized as a unique construction material with the forma-
tion of multiple fine cracks, providing remarkable potential to
achieve self-healing efficiency [8]. Developing enhanced tech-
niques to upgrade the self-healing ability of ECC has been an
important topic of study. Yang et al. (2009) introduced the
mechanical recovery and healing rate of ECC through wet-dry
cycles [13]. Sisomphon et al. (2013) clarified the self-healing mech-
anism of ECC incorporating a cementitious admixture [14]. Hung
et al. (2017) reported that self-healing can be achieved in ECC
through the effect of natural weathering with intrinsic tight crack
width [15]. Qiu et al. (2016) developed an advanced method of
improving the self-healing capacity of a cementitious composite
based on the addition of blast furnace slag and the alkalinity con-
dition [16]. They all concluded that the permissible crack width,
which was smaller than 50 mm, achieved healing cracks [13–16].
Diminished self-healing was observed in the range of cracks
between 50 mm and 150 mm; nevertheless if the crack width
exceeded 150 mm, the self-healing ability vanished due to the
appearance of serious damage to the matrix structure.
Previous studies developed high performance fiber-reinforced
cementless composites called alkaline activated slag (AAS) based
composites, characterized by high ductility and material greenness
[17–19]. This technology also has played a role in innovating car-
bon dioxide emission control and sustainable green infrastructure
due to the elimination of cement binder. While knowledge of self-
healing in normal concrete is available, the self-healing of fiber-
reinforced alkali-activated slag-based cementless composite has
been fairly limited. In this light, it is necessary to develop a new
construction material in term of enhanced ductility, greenness,
and high self-healing potential based on an alkali-activated slag-
based fiber-reinforced composite.
The purpose of this paper is to experimentally investigate the
self-healing capacity of cement-based and alkali-activated slag-
based fiber-reinforced composites with an identical water-to-
binder ratio. The self-healing performance of the composites was
evaluated by visual observation of crack width and measurement
of the resonant frequency with time. Scanning electron microscope
(SEM) and energy-dispersive X-ray spectroscopy (EDS) were also
adopted to analyze the microstructure and chemical composition
of the healing materials.
2. Materials and experimental methodology
2.1. Materials and mixture proportions
Two types of binder, i.e. cement and alkali-activated slag, were
used in this study. The chemical compositions of cement and
Table 1
Chemical compositions of cement and GGBS.
Binder Chemical composition (%)
SiO2 Al2O3 Fe2O3 CaO
Cement 18.5 4.5 3.3 65.8
GGBS 30.6 13.8 0.5 40.4
ground-granulated blast-furnace slag (GGBS) are listed in Table 1.
Type I Portland cement was used, and the specific surface area
and density of cement were 3297 cm2/g and 3.14 g/cm3, respec-
tively. The specific surface area and density of GGBS were 4320
cm2/g and 2.91 g/cm3, respectively. The GGBS is the source mate-
rial of a cementless composite and it was activated by alkali activa-
tors. The calcium hydroxide in a powder form was used as an alkali
activator. The amount of calcium hydroxide was 11.1% of GGBS in
terms of the mass ratio. Polyethylene (PE) fiber was used as a rein-
forcing fiber. The physical properties of PE fiber used in this study
were a diameter of 12 mm, a fiber length of 18 mm, a tensile
strength of 2700 MPa, a density of 0.97 g/cm3, and an elastic mod-
ulus of 88 GPa. Superplasticizer (SP), viscosity modifying admix-
ture (VMA), and anti-foaming agent are all in powder form, and
are used as the additives for the mixtures. A superplasticizer (SP)
and viscosity modifying admixture (VMA) were used to achieve
controlled workability as well as to ensure good fiber dispersion
of the fresh mixture. An anti-foaming agent was included in the
mixture to diminish the amount of air bubbles. Table 2 lists the
mixture proportions. The water-to-binder ratio of all mixtures
was 0.3.
2.2. Mixing procedure and specimen preparation
Each powder type binder was mixed in a commercial planetary
mixer for three minutes. Water was then slowly added and the
mixture was mixed for another five minutes. Next, the SP and
VMA were also added to the mixture. Once a homogeneous mix-
ture was attained, PE fibers were slowly inserted into the mixture.
The fresh mixture was then poured into molds for a compressive
strength test (three 50 mm cube specimens for each mixture)
and a self-healing test including uniaxial tension tests (four dog-
bone shaped specimens for each mixture), and it was covered with
plastic sheets to prevent the evaporation of water in air at the
room temperature (23 °C ± 3 °C) for two days. Finally, the hardened
specimens were removed from the molds and cured in water at a
temperature of 23 °C ± 3 °C until the age of 28 days.
2.3. Mechanical tests and self-healing capacity evaluation
The compressive strength was measured using 50 mm cube
specimens according to the ASTM C109 with three specimens for
each mixture [20]. The tensile load by displacement control in
accordance with a constant loading speed of 0.1 mm/min. was
applied to each specimen to create cracks and investigate the ten-
sile behavior using an electrical uniaxial testing machine with a
capacity of 20 kN at the age of 28 days. Fig. 1 shows the specimen
geometry and the test setup [21]. The specimens have a gauge
length of 80 mm and the cross-sectional area within the gauge
length is 390 mm2 (30 mm  13 mm). The tensile load was mea-
sured by a load cell attached to the top of the jig and the deforma-
tion within the gauge length was measured by two linear variable
differential transducer attached to both sides of the specimens.
According to the highly ductile characteristic of cement-based
and slag-based composites, cracks of all specimens can be self-
controlled below 100 mm. However, it should be noted that to eval-
uate the maximum self-healing capacity, most specimens were
MgO SO3 TiO2 K2O MnO Na2O etc
3.4 2.2 0.3 1.1 – – 0.9
8.0 4.0 0.9 0.5 0.5 0.4 0.4
 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811
Table 2
Mixture proportions (proportion by weight except fiber).
Mixture ID Binder Water
Cement AAS
C30 1 0.30
S30 1 0.30
Fig. 1. (a) The specimen geometry and (b) uniaxial tension test setup to create
cracks and to investigate tensile behavior.
applied under tensile loading with reference to the extended time
in order to generate larger cracks, which are over 350 mm.
After the cracking event, the tensile load was released and the
initial crack widths were examined immediately using an optical
microscope. Furthermore, the number of cracks for each specimen
was counted manually along both sides of the specimens to
identify the cracking pattern at tensile strength and tensile strain
Fig. 2. The resonant frequency test set-up.
803
SP VMA Anti-foaming agent Fiber (vol%)
0.001 0.001 0.001 1.75
0.003 0.0004 0.001 1.75
capacity. The crack patterns consisting of the number of cracks,
crack width, and crack spacing were quantitatively monitored
within the gauge length; for instance, the crack spacing was
obtained by dividing the gauge length by the number of cracks. It
was assumed that all deformation concentrated at the cracks due
to the higher elastic deformation of the crack width compared to
the relative deformation of the matrix. Dividing the total deforma-
tion at maximum tension loading by the number of cracks thus
gives the formula to calculate the average crack width on the spec-
imens at tensile strength and tensile strain capacity of the
composite.
In this study, water is the particular conditioning regime to pro-
mote the autogenous healing ability of the composites. It is neces-
sary to mention that the water was replaced every 7 days. The
residual crack widths were scrutinized once at 4, 7, 14, 28 and
36 days of age regarding the submerging period of each specimen
in water. In addition, to evaluate the environmental condition dur-
ing the self-healing process, the pH values of the water were also
measured after the self-healing process at the healing age of 36
days.
Resonant frequency is one of the widely used measuring criteria
to determine the self-healing degree in concrete. Previous studies
showed that the relationship between representative resonant fre-
quency and tensile behavior can be used to assess the self-healing
capacity of concrete [13–16]. The stiffness recovery of cracked dog-
bone specimens was characterized by the recovery performance of
resonant frequency as a criterion of the self-healing process in
composites. The resonant frequency setup was adopted with refer-
ence to ASTM C215 [22]. The experimental devices include a signal
804 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811

Table 3
Mechanical properties of cement (C30) and slag (S30)-based composites.

Mixture ID Compressive strength (MPa) First cracking strength (MPa) Tensile strength (MPa) Tensile strain capacity (%)
C30 86.3 ± 1.8 3.10 ± 0.58 8.17 ± 0.92 2.66 ± 0.33
S30 46.1 ± 3.3 1.91 ± 0.33 8.55 ± 0.83 5.46 ± 0.22

accelerometer, a data acquisition system, and a steel ball mounted

on a thin rod. In detail, the dog-bone specimens were put on a
shockproof film in order to ensure that the transmission of the
time signal is totally transferred as well as no slight joggle hap-
pened to affect the reliability of testing. As shown in Fig. 2, the
specimens were impacted by an impactor, the solid steel ball, in
the longitudinal direction. The accelerometer was attached to the
other side of the specimen at its end to measure the signal. During
the test, a Fast Fourier transformation (FFT) was adopted to obtain
the frequency domain signal by converting from the time domain
signal. Similar to the crack width reduction, the resonant frequency
was checked once at 4, 7, 14, 28, and 36 days of age with reference
to the immersing timeline.
In order to determine the morphology of healing materials cov-
ering the cracks, SEM is used to observe the crack surface at differ-
ent level of magnification. EDS is also operated to analyze the
chemical composition of the healing product with reference to
the change of element contents in the material matrix. All SEM
and EDS samples were removed from the healing zone of the spec-
imen surface, and honed to about 1 cm2 for the testing.

3. Results and discussion

3.1. Compressive strength and tensile behavior

The mechanical properties including compressive strength, first

cracking strength, tensile strength, and tensile strain capacity of
C30 and S30 are listed in Table 3. The first cracking strength, tensile
strength, and tensile strain capacity were identified from the ten-
sile stress and strain curves (Fig. 3). The S30 mixture showed
Fig. 3. Tensile stress-strain curves of: (a) C30 and (b) S30.
46.6% lower compressive strength compared to that of the C30
mixture with an identical water-to-binder ratio. A previous study
also reported that the compressive strength of an AAS-based com-
posite is approximately 50% of that of a cement-based composite
Table 4
[23]. The differences between the theoretical values and the mea- Cracking patterns of C30 and S30 mixtures.
sured values of the densities were lower than 3.1%, which means
Mixture Average crack width Number of Crack spacing
that the structure of the slag-based composite is as compact as that
ID (mm) cracks (mm)
of the cement-based composite [23]. However, it was not clearly
C30 38.0 ± 4.6 55.3 ± 7.8 1.45 ± 0.20
verified that the reason for the lower compressive strength of the
S30 54.7 ± 3.2 79.9 ± 7.0 1.00 ± 0.09
AAS-based composite might be the weaker bond between the
hydration products of the AAS-based composite than cement-
based composite.
Fig. 3 shows the tensile stress and strain curves of cement and
alkali-activated slag-based composites. All specimens were sub- higher than that of the C30 mixture due to the higher ratio of ten-
jected to fracture limit to create the largest crack width on the sile strength to cracking strength. Furthermore, the tensile strain
specimen surface under the tensile load and showed clear strain- capacity of the S30 mixture in this study is relatively high com-
hardening behavior with multiple cracks. Although the first crack pared to past studies on self-healing characteristics of fiber-
strength of the S30 mixture was 1.91 MPa, which is 38.4% lower reinforced composites [13–16].
than that of the C30 mixture, the tensile strength of S30 was Table 4 listed the cracking pattern of the S30 and C30 mixtures
4.65% higher than that of the C30 mixture. Moreover, the tensile composed of average crack width, number of cracks, and cracks
strain capacity of S30 mixture was 5.46%, which is 2.05 times spacing. The average crack width of the S30 mixture, which was
higher than that of the C30 mixture. The ratio of tensile strength 54.7 mm, is higher than that of the C30 mixture by 43.9%. The
to first crack strength could be a criterion to evaluate the domi- S30 mixture showed a 44.5% higher number of cracks compared
nance of the high tensile strain capacity of the composites [24]. to that of C30. On the other hand, the crack spacing of the S30 mix-
Based on the data in this study, this ratio of the S30 mixture was ture is 31.0% more narrow compared to the C30 mixture. Overall, it
4.48, which is 69.9% higher than that of the C30 mixture. It is worth appears that the S30 mixture was damaged more than the C30
mentioning that the tensile strain capacity of the S30 mixture was mixture.
 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811
3.2. Crack width reduction
Crack width reduction could be considered as an indicator for
the self-healing capacity of cement-based and slag-based compos-
ites. Six cracking measurement positions were randomly selected
for each specimen, and crack width variation was observed at 4,
7, 14, 28, and 36 days of age. Fig. 4 shows the mapping of cracks
and images of magnified zones.
Fig. 5 shows the self-healing phenomenon of the C30 mixture
and S30 mixture based on the reduction of crack width under
water immersion for 36 days. The crack width, which is around
45 mm, is evaluated through the optical microscope at the C3 posi-
tion for the C30 mixture and the S6 position for the S30 mixture. As
can be observed, the cracks were sealed by an abundant white for-
mation in the form of crystal-like materials sprouting from both
sides of the cracks. It appears that the healing products were cre-
ated by fine crystals bridging together to fill the cracks. In the case
of the S30 mixture, the crack appears to be healed completely
while some voids remained for the C30 mixture with respect to
the residual crack width of about 4 mm. The healing materials were
located along the crack mouths. Moreover, the healing products
were not only visible on the crack surface, but also seen on the
areas nearby crack lines.
The initial and healed crack widths of specimens before and
after the wetting period were directly observed by the optical
microscope at 84 magnification. Fig. 6 shows the reduction of
Fig. 4. Mapping of cracks and image of magnified zones of: (a) C30 mixture and (b) S30 mixture.
805
the crack width with respect to the downward movement of each
solid dot denoting the crack width measurement locations on the
surface of the specimens. In other words, the full healing process
will be achieved if the dots lie on the X-axis. A 45-degree dashed
line was marked to show the zero-healing area. It has to be noted
that the data in Fig. 6 only show the representative crack widths,
not all crack widths for each specimen. The maximum crack width
observed in the C30 and S30 mixtures was up to 350 mm, which
indicated high specimen damage. As can be seen, smaller crack
width has more self-healing potential compared to larger crack
width. According to the data in Fig. 6, the maximum crack width
calculated from the fully healing positions was 55.0 mm for the
S30 mixture and 54.0 mm for the C30 mixture, respectively. There-
fore, it is worth mentioning that the fine crack width around 50 mm
exhibited full healing capacity within 36 days of wetting due to the
zero width of the cracks obtained for both the C30 and S30 mix-
tures. On the other hand, the crack width over 50 mm showed a
partial healing phenomenon regarding the residual crack width.
Therefore, the crack width of 50 mm can be seen as the healing
threshold of C30 and S30 mixtures in case of crack width reduction.
It is also notable that the healing threshold of the slag-based com-
posite (S30) is similar to that of the cement-based composite (C30)
and other previous studies [13–16].
Fig. 7 shows the change of relative crack width (W c ) within 36
wetting days. It was calculated by the average value of crack range
collected from the data of Fig. 6 to the initial crack width.
806 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811

Pre-

loading

(wc: 45 μm) (wc: 49 μm)

14th day of

submerged

age

(wc: 32 μm) (wc: 38 μm)

28th day of

submerged

age

(wc: 18 μm) (wc: 22 μm)

36th day of

submerged

age

(wc: 4 μm) (wc: 0 μm)

(a) (b)
Fig. 5. Self-healing phenomenon of (a) C30 and (b) S30 (at positions of C3 and S6).

Wt 3.3. Resonant frequency

Wc ¼  100 ð%Þ
Wi
where W t and W i are the crack width at a certain time and the ini-
tial crack width, respectively. The initial crack width was catego-
rized into three classes, smaller than 50 mm (Wc1), in a range of
50–150 mm (Wc2), and larger than 150 mm (Wc3). It can be obviously
demonstrated for the C30 and S30 mixtures that the relative crack
width reduces over time and the crack width smaller than 50 mm
has the highest healing ability, in contrast to the other two classes.
On the other hand, during the timeline of submerging, the S30 mix-
ture has a slight healing advantage compared to the C30 mixture
due to the lower relative crack width. At the final stage of 36 sub-
merged days, the relative crack width of the S30 mixture with ref-
erence to Wc1, Wc2, and Wc3 was 0%, 23.4%, and 41.0%, which are
5.5%, 2.1% and 5.5% lower than the data of the C30 mixture,
respectively.
ð1Þ
The resonant frequency (RF) was directly measured before and
after the cracking process with the submerging time. Fig. 8 shows
the normalized resonant frequency with respect to all tensile spec-
imens for each group of C30 and S30 mixtures. The normalized RF
was obtained by dividing the RF (RFself-healed) at the self-healed
condition by the original RF (RForiginal). The normalized RF can be
described by the following equation:
RFselfhealed
Normalized RF ¼ ð2Þ
RForiginal
As can be observed, the normalized RF of the C30 and S30 mix-
tures dramatically dropped after making cracks by a loading event,
but was quickly retained and steadily increased according to the
wetting period of the conditioning process. This implies that the
stiffness of the specimens was recovered by the rise of the normal-
 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811
Fig. 6. The crack width reduction of: (a) C30 and (b) S30.
ized RF. The normalized RF values of the C30 and S30 mixtures
were relatively similar, being around 20%. This is the fundamental
criterion to establish the normalized average RF based on the nor-
malized RF subsets, as shown in Fig. 9.
It was clearly seen that the normalized average RF of the C30
mixture at 36 submerged days was 77.8% which is 13.5% higher
than that of the S30 mixture, thus showing the good recovery abil-
ity of the C30 mixture in terms of mechanical stiffness. This is likely
due to the smaller crack width as well as the number of cracks of the
C30 mixture compared to the S30 mixture, as reported in Table 4.
The investigation of Kan et al. (2012) revealed that the resonant fre-
quency is generally affected by the crack width [25]. The larger
crack width might make the structure less dense and weaken the
oscillation waves transmitted through the specimen caused by an
impactor and significantly decrease the resonant frequency. Over-
all, the S30 mixture has advantages in reduction crack width, but
displays a limitation for resonant frequency recovery.
3.4. Healing materials analysis
Fig. 10 shows the morphology of partial healing cracks observed
by SEM at 1200 magnification. As can be seen, a huge number of
807
Fig. 7. The relative crack width of: (a) C30 mixture and (b) S30 mixture.
small crystal-like materials covers both sides along the cracks to
fill the gaps. The morphology of the healing materials is described
by white crystals precipitated on the crack mouth and even outside
the healing zone, which is near the crack face. It is possible that the
healing materials were crystallized to seal the crack. On the other
hand, the crack line was filled by the concentration of abundant
crystal-like materials growing in both crack flanks to narrow the
crack width. The cracks can then be divided into two zones, healed
and unhealed crack zones. The healed crack zone might be under-
stood as the area sealed completely whereas the unhealed crack
zone is likely the area showing partial healing ability regarding
some existing voids or no healing appearance. In Fig. 10, a crack
width of about 45 mm is considered for the C30 and S30 mixtures.
In general, the unhealed crack zone of the S30 mixture is quite nar-
row compared to that of the C30 mixture. This implies that the
crack width reduction of the S30 mixture is higher than that of
the C30 mixture, and this is satisfied according to the results in
Fig. 5. Moreover, as mentioned in the aforementioned section,
the resonant frequency of the C30 and S30 mixtures could not be
obtained at the highest level of recovery of 100% due to the less
dense matrix caused by remaining voids inside the cracks.
Fig. 11 shows the healing materials covering the PE fibers visu-
alized by the SEM observation for the C30 and S30 mixtures. It can
be hypothesized that the healing products are precipitated in the
form of crystals on the fiber surface. Although the healing products
did not fully adhere to the surface of the fibers, the fibers were able
to facilitate the precipitation of healing materials. Based on this
viewpoint, fibers bridging cracks can be seen as one of the healing
characteristics that promote healing capacity for cement-based
and slag-based composites, indicating that the healing products
are not only formed on the crack surface but also are embedded
in the fibers after the self-healing process.
808 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811
Fig. 8. Normalized resonant frequency (RF) of: (a) C30 and (b) S30 over submerging
time.
Fig. 9. Normalized average resonant frequency of: (a) C30 mixture and (b) S30
mixture over submerging time.
Based on the shape of the material surface, the healing materi-
als were distinguished by the gray soft line plotted as the boundary
between the original zone of the matrix and the healing material
area, as shown in Fig. 12. There are two positions on the healed
crack zone, which were highlighted to investigate the chemical
compositions of the C30 and S30 mixtures via an EDS analysis.
As indicated by the results in Table 5, a strong peak of calcium con-
tent is characterized for all located positions of the C30 and S30
mixtures, and there is a little silicon concentration. According to
previous studies [26,27], the ratio of Ca/Si could be used as a crite-
rion to determine the appearance of CaCO3 or CAS-H as healing
products. In this study, the composition of calcium is overwhelm-
ing compared to silicon, indicating that calcium carbonate is the
dominant healing material of the cement-based and slag-based
composites (C30 and S30 mixtures). On the other hand, based on
the atomic ratio, it is simple to estimate the chemical formula of
calcium carbonate with respect to one calcium atom with one car-
bon atom and three oxide atoms. The residual atoms are likely a
small amount of CASAH present in the healing products.
Table 6 shows the pH values of clean water and water after sub-
merging the specimens of cement-based and alkali-activated slag-
based composites. As shown in this table, the pH values of water
after submerging the specimens were higher than those of clean
water, which means that alkaline compounds existed in water. In
other words, Ca2+ ion remained in the matrix due to leakage and
diffusion from Ca(OH)2 and CAS-H in the specimens [6]. A rise of
alkalinity also promotes calcium carbonate precipitation [28].
The calcite formation dependent on the pH can be described as
the following chemical reaction [8]:
Ca2þ þ CO2
3 ! CaCO3 ðpH > 8Þ
Ca2þ þ HCO3 ! CaCO3 þ Hþ ð7:5 < pH < 8Þ
Following that, it is essential to remark that the pH of water
containing C30 and S30 mixtures is basically satisfied with the cal-
cium carbonate formation.
According to the self-healing phenomenon of the slag-based
composite, the crack closing was schematically illustrated in
Fig. 13. After the hydration process, calcium ions remained in the
matrix from the residual alkali activator of Ca(OH)2 and CASAH
in the matrix. In fact, carbon dioxide (CO2) in the atmosphere dis-
solves in the water, and simultaneously reacts with the water
molecules to create carbonic acid (H2CO3). This form of fragile
chemical bonds easily decomposes, so that CO2 3 and HCO3 ions
are easily separated from H2CO3, concomitantly combining with
Ca2+ diffused from the flowing water to precipitate calcium carbon-
ate (CaCO3). The chemical process would be described as the fol-
lowing formulae [8,29]:
H2 O þ CO2 $ H2 CO3
H2 CO3 $ HCO3 þ Hþ
HCO3 $ CO2
3 þH
þ
þ
H2 CO3 $ CO2
3 þ 2H
CaðOHÞ2 $ Ca2þ þ 2OH
Ca2þ þ CO2
3 ! CaCO3
With respect to previous research [6,8,28], CO2 in water pene-
trates the crack wall to promote the diffusion of Ca2+ inside the
hardened composite. This guarantees that the self-healing process
takes place continuously with time due to sufficient Ca2+ ions after
consuming the previous crystal growth.
4. Conclusion
This study experimentally investigated the self-healing proper-
ties of cement-based and alkali-activated slag-based fiber-
reinforced composites with high ductility and controlled crack
width. A series of experiments was performed to measure the
mechanical properties and to observe the self-healing phe-
nomenon and healing materials of the composites. From the cur-
rent test results, the following conclusions were drawn:
 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811 809

Fig. 10. The healed and unhealed zones of (a) C30 mixture and (b) S30 mixture observed by SEM.

Fig. 11. The healing materials cover on the fiber of (a) C30 mixture and (b) S30 mixture observed by SEM.

Fig. 12. SEM images of highlighted positions for (a) C30 mixture and (b) S30 mixture to observe the chemical compositions of healing materials.

1. Although the alkali-activated slag-based composite showed tensile strain capacity than the cement-based composite. Fur-
lower compressive strength than the cement-based composite thermore, the slag-based composite showed higher average
with an identical water-to-binder ratio, the slag-based com- crack width and number of cracks than the cement-based
posite showed higher tensile strength and two times higher composite.
810 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811

Table 5
Chemical compositions of healing materials of C30 and S30.

Position Ratio Chemical composition (%)

C O Na Mg Al Si Ca
a-1 Weight 19.42 48.93 1.19 0.39 0.68 1.60 27.78
Atomic 29.30 55.42 0.94 0.29 0.46 1.04 12.56
a-2 Weight 15.79 52.95 – – 0.10 0.37 30.80
Atomic 24.30 61.18 – – 0.07 0.24 14.21
b-1 Weight 12.15 46.62 – 0.32 0.19 0.24 40.49
Atomic 20.38 58.70 – 0.26 0.14 0.17 20.35
b-2 Weight 14.14 49.31 – 0.27 0.12 0.10 36.06
Atomic 22.73 59.53 – 0.21 0.09 0.07 17.38

Table 6 through resonant frequency recovery, which was 13.5% higher

pH values of water before and after the self-healing process. than that of slag-based composite. This is likely due to the smal-
ler initial crack width as well as the number of cracks of the
Mixture pH
cement-based composite compared to the slag-based
Clean water Water after submerging the specimens
composite.
C30 7.17 9.84 4. SEM observation clearly visualized the morphology of self-
S30 7.19 10.69
healing products. Specifically, the healing crack width was
described by healed and unhealed crack zones, which were dis-
tinguished by abundant white tiny crystals embedded along the
2. Smaller crack width has more self-healing potential compared crack boundary of the specimens. CaCO3 was the major healing
to larger crack width for both cement-based and slag-based material for both cement-based and slag-based composites as
composites. The threshold crack width for full healing of the indicated by the high amount of calcium in the matrix structure
slag-based composite was similar to that of the cement-based after the self-healing process.
composite. The slag-based composite has a slight healing
advantage compared to the cement-based composite in terms The findings presented in this research provide a preliminary
of lower crack width at 36 days of wetting. approach for developing a new self-healing material based on
3. The normalized resonant frequency of both composites alkali-activated slag. According to the unique characteristics of this
increased over time after the cracking event. The composite including high ductility, greenness, and autogenous
cement-based composite showed better healing potential

Fig. 13. The self-healing mechanism of the slag-based composite.

 H.H. Nguyễn et al. / Construction and Building Materials 165 (2018) 801–811
healing, the alkali-activated slag-based fiber-reinforced composite
is a promising intelligent material for the future.
Acknowledgements
This research was supported by a grant (17SCIP-B103706-03)
from Construction Technology Research Program funded by Min-
istry of Land, Infrastructure and Transport of Korean government
and also supported by Basic Science Research Program through
the National Research Foundation of Korea (NRF) funded by the
Ministry of Science and ICT (NRF-2016R1A1A1A05005208).
References
[1] J.M. Crow, The concrete conundrum, Chem. World 5 (3) (2008) 62–66.
[2] S. Zwaag, Self healing materials: an alternative approach to 20 centuries of
materials science, Springer Science+ Business Media BV2008.
[3] M. de Rooij, K. Van Tittelboom, N. De Belie, E. Schlangen, Self-healing
phenomena in cement-Based materials: state-of-the-art report of RILEM
technical committee 221-SHC: self-healing phenomena in cement-based
materials, Springer, 2013.
[4] N. Hearn, Self-sealing, autogenous healing and continued hydration: what is
the difference?, Mater Struct. 31 (8) (1998) 563–567.
[5] A.M. Neville, Properties of concrete, Longman, London, 1995.
[6] C. Edvardsen, Water permeability and autogenous healing of cracks in
concrete, Mater. J. 96 (4) (1999) 448–454.
[7] M. Roig-Flores, F. Pirritano, P. Serna, L. Ferrara, Effect of crystalline admixtures
on the self-healing capability of early-age concrete studied by means of
permeability and crack closing tests, Constr. Buildi. Mater. 114 (2016) 447–
457.
[8] M. Wu, B. Johannesson, M. Geiker, A review: self-healing in cementitious
materials and engineered cementitious composite as a self-healing material,
Constr. Buildi. Mater. 28 (1) (2012) 571–583.
[9] M. Roig-Flores, S. Moscato, P. Serna, L. Ferrara, Self-healing capability of
concrete with crystalline admixtures in different environments, Cem. Concr.
Compos. 86 (2015) 1–11.
[10] Y.Ç. Ersßan, E. Hernandez-Sanabria, N. Boon, N. De Belie, Enhanced crack closure
performance of microbial mortar through nitrate reduction, Cem. Concr.
Compos. 70 (2016) 159–170.
[11] D. Jaroenratanapirom, R. Sahamitmongkol, Self-crack closing ability of mortar
with different additives, J. Met. Mater. Miner. 21 (1) (2011) 9–17.
[12] L. Lv, Z. Yang, G. Chen, G. Zhu, N. Han, E. Schlangen, F. Xing, Synthesis and
characterization of a new polymeric microcapsule and feasibility investigation
in self-healing cementitious materials, Constr. Buildi. Mater. 105 (2016) 487–
495.
811
[13] Y. Yang, M.D. Lepech, E.-H. Yang, V.C. Li, Autogenous healing of engineered
cementitious composites under wet–dry cycles, Cem. Concr. Res. 39 (5) (2009)
382–390.
[14] K. Sisomphon, O. Copuroglu, E. Koenders, Effect of exposure conditions on self
healing behavior of strain hardening cementitious composites incorporating
various cementitious materials, Constr. Buildi. Mater. 42 (2013) 217–224.
[15] C.-C. Hung, Y.-F. Su, H.-H. Hung, Impact of natural weathering on medium-
term self-healing performance of fiber reinforced cementitious composites
with intrinsic crack-width control capability, Cem. Concr. Compos. 80 (2017)
200–209.
[16] J. Qiu, H.S. Tan, E.-H. Yang, Coupled effects of crack width, slag content, and
conditioning alkalinity on autogenous healing of engineered cementitious
composites, Cem. Concr. Compos. 73 (2016) 203–212.
[17] J.-I. Choi, B.Y. Lee, R. Ranade, V.C. Li, Y. Lee, Ultra-high-ductile behavior of a
polyethylene fiber-reinforced alkali-activated slag-based composite, Cem.
Concr. Compos. 70 (2016) 153–158.
[18] B.Y. Lee, C.-G. Cho, H.-J. Lim, J.-K. Song, K.-H. Yang, V.C. Li, Strain hardening
fiber reinforced alkali-activated mortar – A feasibility study, Constr. Buildi.
Mater. 37 (2012) 15–20.
[19] Y. Lee, J.-I. Choi, H.-K. Kim, B.Y. Lee, Effects of a defoamer on the compressive
strength and tensile behavior of alkali-activated slag-based cementless
composite reinforced by polyethylene fiber, Compos. Struct. 172 (2017)
166–172.
[20] ASTM, Standard test method for compressive strength of hydraulic cement
mortars (using 50 mm [2 in.] Cube Specimens), American Society for Testing
and Materials, 2007, ASTM C109/C109M.
[21] JSCE, Recommendations for design and construction of high performance fiber
reinforced cement composites with multiple fine cracks (HPFRCC), Japan
Society of Civil Engineers, Japan, 2008.
[22] ASTM, Standard Test Method for Fundamental Transverse, Longitudinal, and
Torsional Frequencies of Concrete Specimens, American Society for Testing
and Materials, 2014, ASTM C215.
[23] J.-I. Choi, K.-I. Song, J.-K. Song, B.Y. Lee, Composite properties of high-strength
polyethylene fiber-reinforced cement and cementless composites, Compos.
Struct. 138 (2016) 116–121.
[24] T. Kanda, V.C. Li, Practical design criteria for saturated pseudo strain hardening
behavior in ECC, J. Adv. Concr. Technol. 4 (1) (2006) 59–72.
[25] L.-L. Kan, H.-S. Shi, Investigation of self-healing behavior of Engineered
Cementitious Composites (ECC) materials, Constr. Buildi. Mater. 29 (2012)
348–356.
[26] H. Huang, G. Ye, D. Damidot, Characterization and quantification of self-
healing behaviors of microcracks due to further hydration in cement paste,
Cem. Concr. Res. 52 (2013) 71–81.
[27] H. Huang, G. Ye, D. Damidot, Effect of blast furnace slag on self-healing of
microcracks in cementitious materials, Cem. Concr. Res. 60 (2014) 68–82.
[28] K. Sisomphon, O. Copuroglu, E. Koenders, Self-healing of surface cracks in
mortars with expansive additive and crystalline additive, Cem. Concr. Compos.
34 (4) (2012) 566–574.
[29] J. Pichtel, Mine water: hydrology, pollution, remediation, J. Environ. Qual. 32
(2) (2003) 739.