Cement and Concrete Composites 135 (2023) 104837

Contents lists available at ScienceDirect

Cement and Concrete Composites

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

High-strength engineered cementitious composites with nanosilica

incorporated: Mechanical performance and autogenous
self-healing behavior
Zhigang Zhang a, b, Zhipeng Li a, Jialuo He a, Xianming Shi a, *
a
National Center for Transportation Infrastructure Durability and Life Extension, Department of Civil and Environmental Engineering, Washington State University,
Pullman, WA, 99164, USA
b
Key Laboratory of New Technology for Construction of Cities in Mountain Area (Chongqing University), Ministry of Education, Chongqing, 400045, PR China

A R T I C L E I N F O A B S T R A C T

Keywords: It is highly desirable to achieve better self-healing performance of high-strength engineered cementitious com­
ECC posite (ECC) incorporating hydrophobic polyethylene (PE) fibers. This work demonstrated that admixing
PE fiber nanosilica was able to make the ECC matrix denser and strengthen the interfacial bond between PE fiber and
High strength concrete
matrix. The nano-modified mixtures achieved enhanced mechanical performances, narrowed the crack width
Nanosilica
Self-healing
and facilitated autogenous self-healing of PE-ECC. For instance, at 28 days, admixing nanosilica at 1 wt%
High-volume fly ash improved the tensile strength, first-cracking strength, and strain capacity of M1 series ECC by 24.5%, 23.4%, and
58.1%, which reached 9.97 MPa, 7.03 MPa and 5.17%, respectively. The incorporation of 1 wt% NS also reduced
the average crack width of PE-ECC by 59.5% and 59.4%, which reached 58.1 μm and 23.3 μm, for the M1 and M2
series, respectively. The narrower crack widths facilitated autogenous self-healing of PE-ECC specimens sub­
jected to 30 wet-dry cycles, where particle-shaped product (mainly calcite) and dense product (mainly C–S–H
gel) formed on the M1 and M2 series PE-ECC, respectively. It is intriguing that the high-volume fly ash PE-ECC
(M2-1%) achieved the compressive strength of 80.7 MPa and 97.9 MPa at 28 days and 90 days, respectively. This
environmentally sustainable mixture also achieved the first-cracking strength, tensile strength and strain ca­
pacity of 3.98 MPa, 7.78 MPa, and 5.28%, respectively, at 28 days of age.

1. Introduction hundred times the maximum tensile strength strain of conventional

concrete [4,5]. Under loading, instead of one single localized crack on
Concrete is the most widely consumed human-made construction normal concrete, multiple micro-cracks develop sequentially on ECC
material and serves as a crucial material enabling the development of specimens. Such micro-cracks can be restrained below 100 μm in width,
modern society. Concrete is good in resisting compression but weak in upon the fiber-bridging effect. This kind of outstanding capacity in
resisting tension and thus prone to cracking under mechanical and crack-opening control of ECC not only endows the concrete with better
environmental loads. Such drawback of concrete poses a risk to the life- ductility, but also greatly improves the transport properties of concrete
cycle performance of concrete infrastructure, because the cracks likely (e.g., reduced water permeability and lower chloride ion diffusivity).
create preferable paths for the intrusion of aggressive agents (water, Moreover, such tiny cracks on the ECC can heal themselves autoge­
chloride, sulfate, CO2, etc.) and thus aggravate other durability issues of nously under moist condition, which is expected to further benefit the
the concrete. To overcome the inherent brittleness, Li et al. developed a durability of concrete infrastructure [6–8].
type of highly ductile concrete in the 1990s named engineered cemen­ In recent years, the development of ECC materials is moving towards
titious composites (ECC), through a premeditated design of fiber, ma­ high strength to cater to the increasing demand for complex or mega
trix, and interface in between at a moderate fiber dosage, based on the structures and extend the application fields of ECC. For instance, the
micro-mechanics theory [1–3]. teams led by Yu and Zhang developed ECC mixtures featuring a 28-day
ECC materials are characteristic of high ductility, with several compressive strength of 60–120 MPa [4,9–12]. Recently, Dai’s team

* Corresponding author.
E-mail address: xianming.shi@wsu.edu (X. Shi).

https://doi.org/10.1016/j.cemconcomp.2022.104837
Received 21 June 2022; Received in revised form 9 October 2022; Accepted 29 October 2022
Available online 2 November 2022
0958-9465/Published by Elsevier Ltd.
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

further enhanced the design of ECC to achieve a 28-day compressive Table 1

strength up to 210 MPa [13–15]. Chemical compositions of binder materials.
To produce high-strength ECC, polyethylene (PE) fibers with high SiO2 Al2O3 Fe2O3 CaO MgO SO3
strength and high modulus are preferred over the commonly used
Cement 22.60% 4.30% 2.40% 64.40% 2.10% 2.30%
polyvinyl alcohol (PVA) fibers. However, different from hydrophilic Fly ash 35.64% 18.31% 5.63% 25.44% 6.04% 1.98%
PVA fibers, the PE fibers are intrinsically hydrophobic, which weakens Silica fumea 93.47% – – – – –
the bonding between the fiber and the cementitious matrix. This likely a
Specific surface of 22.28 m2/g, moisture content: 0.27%.
causes widening of the micro-cracks on the ECC specimen [9], which
tends to undermine the self-healing ability of the ECC and adversely
affect its transport properties. Note that the water permeability coeffi­
cient of concrete is proportional to its crack width to the third power
[16]. It is thus highly desirable to improve the crack-width-control ca­
pacity and recover the self-healing ability of high-strength ECC
(HS-ECC), especially for the hydraulic structures.
Over the past decade or so, the introduction of nano-sized materials
in concrete has attracted widespread attention by virtue of its great
benefits to the mechanical and durability performances of concrete. The
nanomaterials in concrete can offer nucleation sites for hydration of
cement and regulate the process to form more and better hydration
products [17,18]; moreover, they can serve as nanofillers that fill the
space between gel particles; both of which result in a denser cement
paste [19]. Due to its unique chemistry and large specific surface area,
nanosilica (NS) also exhibits a high pozzolanic activity with portlandite
(CH) to form dense C–S–H gel [20]. Cumulative studies have demon­
strated that the admixing of NS (when well dispersed) can significantly
improve the mechanical properties and durability of concrete [21–23].
Nanosilica, admixed at an appropriate dosage, has been demon­
strated to be generally beneficial to fiber reinforced concrete (FRC). Wu Fig. 1. The particle size distribution of solid ingredients.
et al. [24] reported that NS can substantially enhance the interfacial
bond between fiber and surrounding matrix in ultra-high performance 2. Experimental
concrete (UHPC). Yesilmen and Najjar et al. [25,26] concluded that the
NS led to an increase of frictional bond between the PVA fiber and the 2.1. Materials and specimen preparations
cementitious matrix, and more percentage of PVA fibers were thus
ruptured, resulting in an improvement of the flexural strength of ECC. In this study, the binder ingredients included Type I-II Portland
Bashar and Razavi et al. [27,28] reported that NS is favorable to the cement, Class C fly ash, silica fume (SF), and nanosilica (NS). The
enhancement of compressive and flexural strengths and elastic modulus chemical compositions of the binder materials are summarized in
of PVA-ECC while reducing its tensile strain capacity. Fu and Zhou et al. Table 1. The NS has a purity of 99.8%, median particle size of 11 nm, and
[29–31] studied the influence of NS on mechanical performance of PE a specific surface area of 175–225 m2/g. The quartz sand in the 70# to
fiber reinforced ECC (PE-ECC). Fu and Zhou et al. [29,30] claimed that 100# mesh range was used as fine aggregate. The size distributions of
the NS significantly improved the compressive strength and tensile the main solid ingredients are displayed in Fig. 1. Table 2 lists the mix
properties of lightweight PE-ECC incorporating fly ash cenosphere. proportions of PE-ECCs tested in this work. In the mixtures, there are
Nonetheless, Zhou [31] found that the addition of NS more than 1% by two series, with the fly ash-to-cement mass ratio of 0.8 in M1 and 1.8 in
weight of binder adversely affects the ductility of PE-ECC incorporating M2, respectively. For each series, a very small amount of cement was
100% recycled fine aggregate. replaced by NS, and the dosages of NS were 0.5% and 1% of total binder
Investigations on self-healing performance of high-strength ECC are materials by weight. The mass ratios of water and sand to binder stayed
very limited, let alone the influence of NS addition on self-healing at 0.204 and 0.30, respectively. For the NS-modified PE-ECC mixtures,
behavior of PE-ECC. The existing studies relevant to NS addition and due to the extreme high surface area of NS, the amount of super­
self-healing of ECC mainly focused on ECC with compressive strength plasticizer (SP) is more than the reference ones to achieve a desired
less than 60 MPa, which is the lower limit of high strength concrete [32]. flowability. The dosage of PE fiber in ECC mixtures was adopted at 2%
According to the literature review, one can hypothesize that admixing by volume. The PE fiber has a diameter and length of 26 μm and 12 mm,
NS at no more than 1% by weight of binder would make the ECC matrix respectively, and its tensile strength and modulus are 3000 MPa and
denser and strengthen the interfacial bond between PE fiber and matrix. 100 GPa, respectively.
As a result, it is likely to achieve enhanced compressive and tensile During the preparation of PE-ECC specimens, all dry ingredients
strengths of PE-ECC, narrow the crack width and facilitate autogenous were pre-mixed for 3 min using a Hobart mixer (Model A-200, The
self-healing of PE-ECC. Hobart Mgf. Co., Troy, OH). The NS along with the polycarboxylate-
Using PE fibers and silica fume, this work developed two HS-ECC based superplasticizer were first dispersed in water and then the
mixtures with a compressive strength exceeding 75 MPa. A type of NS aqueous suspension was sonicated for 40 min with amplitude of 40% by
was admixed into the two mixtures, and the effects of NS on mechanical a 400-W Digital Sonifier (Model 450, Branson Ultrasonic SA, Geneva).
properties were investigated via compressive and uniaxial tensile tests. This resulted in very good dispersion of the NS in the suspension. The
To shed light on the mechanisms underlying the observed alteration in sonicated suspension was then added into the well-mixed dry in­
the HS-ECC properties due to the admixed NS, the non-evaporable water gredients and mixed for another 3 min to reach a homogeneous mortar
amount was measured and thermogravimetric analysis (TGA) was con­ mix. Finally, the PE fibers were slowly added into the mortar and mixed
ducted. Finally, the autogenous self-healing phenomenon on HS-ECC for another 3 min until no cluster of fibers was observed, i.e., all fibers
was microscopically examined. are well dispersed. Subsequently, the fresh mortar was poured into the
molds. All specimens were demolded after 24 h and cured in ambient
temperature (around 25 ◦ C) and relative humidity of 55 ± 5% until

2
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

Table 2
The PE-ECC mix proportions (kg/m3).
ECC Mixtures Cement Fly ash SF NS Sand Water SP PE fiber

M1 series Ref.-M1 718 574 128 0 430 290 18 19

M1-0.5% 710 574 128 7.1 430 290 20 19
M1-1% 704 574 128 14.2 430 290 23 19
M2 series Ref.-M2 480 862 128 0 430 290 18 19
M2-0.5% 473 862 128 7.1 430 290 20 19
M2-1% 466 862 128 14.2 430 290 23 19

Fig. 2. Dimensions of the dogbone specimen and the tensile test set-up.

testing. Compressive tests were performed on cubic specimens to gain the

compressive strength of ECCs in accordance with ASTM C109 [33]. The
side length of the cubic specimens was 50.8 mm. The specimens were
2.2. Testing procedures tested at 3, 7, 28, 60, and 90 days, and the compressive strength was
calculated by averaging the value of three specimens for each ECC
In this study, for measuring the non-evaporable water content and mixture.
conducting the TGA, the neat paste without sand and fibers was pre­ Uniaxial tensile testing was conducted to assess the tensile perfor­
pared and poured into a cubic mold. After 28 days of curing, the cubic mance of ECCs at the age of 90 days of curing. Fig. 2 illustrates the di­
samples were crushed via compressive test, whereafter the small pieces mensions of the specimens and the test set-up. This work adopted the
in the core of samples were selected and immersed in anhydrous ethanol dogbone-shaped specimen in which the cross-sectional area within
for termination of further hydration. Finally, the samples were grounded gauge length is relatively smaller to ensure the occurrence of cracking in
into powder before testing, and all the powder was sieved by mesh #200 this section. Two external Linear Variable Differential Transformer
to ensure a size below 75 μm. (LVDT) sensors were fixed on the dogbone-shaped specimens to record
A loss-on-ignition (LOI) method was adopted to measure the non- the displacement for the calculation of tensile strain of each ECC spec­
evaporable water content. Each powder sample weighted approxi­ imen. After the tensile test, the dogbone-shaped specimens were sub­
mately 1 g was dried in an oven at 105 ◦ C until a constant weight was jected to wet-dry curing cycles to allow autogenous self-healing, where
reached, and then heated to 1050 ◦ C for 2 h in a muffle furnace. The non- each cycle consisted of immersing the specimen in water for 12 h and
evaporable water content was calculated using eqs. (1) and (2). then drying it in air for 12 h. After 30 cycles, small samples were cut
Wn = (LOI − Lc ) / (1 − Lc ) (1) from each dogbone-shaped specimen for observations by scanning
electron microscopy (SEM).
LOI = (m105◦ C − m1050◦ C ) / m105◦ C (2) A three-point bending test was carried out to measure the fracture
toughness (Km) of each ECC matrix at the age of 90 days of curing, ac­
where Wn is the non-evaporable water content; LOI is the mass loss of cording to ASTM E399 [34]. For this test, the specimen features a
powder sample; Lc is the ignition loss of binder materials. The ignition dimension of 330 mm in length, 35 mm in width, and 70 mm in height;
loss of cement, fly ash, silica fume, and NS was 0.61%, 2.07%, 2.16%, the full span length is 200 mm; and the depth of the notch is half of the
and 0.11%, respectively. specimen height.
Thermogravimetric analysis (TGA) was performed on each powder
sample. Some small pieces from the core of a dried cubic sample were
obtained and ground into powder before testing. Fifty mg of the powder
sample was heated at a rate of 10 ◦ C/min from 25 ◦ C to 900 ◦ C in a
nitrogen atmosphere at a flow rate of 30 ml/min.

3
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

gravimetric (DTG) curves of the ECC mixtures. It can be seen that the
mass loss of samples can be divided into three stages. The mass loss in
the first stage up to 380 ◦ C is tied to decomposition of C–S–H, ettringite,
and AFm phases; the loss in the second stage between 380 ◦ C and 450 ◦ C
is due to dehydroxylation of portlandite (CH); the mass loss in the third
stage between 600 ◦ C and 800 ◦ C is due to decarbonation of CaCO3.
Fig. 5 indicates that the CH content in the PE-ECC mixtures decreases
as NS dosage increases and the reduction can be as much as 50.1%. This

Fig. 3. Non-evaporable water content in hardened ECC paste.

3. Results and discussion

3.1. Non-evaporable water content

The increased bound water content in NS-modified ECC implies

higher compressive strength and higher first-cracking strength under
tension. This is because that the non-evaporable water is the chemically Fig. 5. The CH content in the ECC mixtures.
bonded water in the hydrated phase, which can represent the amount of
hydration products in a hardened paste. Fig. 3 plots the non-evaporable
content (Wn) in the ECC mixtures. The results show that, for both ECC
series, the bound water content increases as the dosage of NS increases,
which implies that adding NS resulted in the production of more hy­
dration products during the binder hydration process. This is because
the NS provided additional nucleation sites for the growth of C–S–H, and
thus accelerated the hydration of the cementitious binder. Moreover, the
NS features high pozzolanic activity and can react with CH (portlandite)
to form additional C–S–H gel.
For the M1 series, the Wn value increased to 11.2% and 12.0% as the
dosage of NS rose to 0.5% and 1.0%, respectively, i.e., an increase of
9.5% and 17.4% from that of Ref.-M1. In contrast, the incorporation of
NS only increased the Wn value in M2 series by 3.5% and 5.4%,
respectively. The relative lower Wn increases induced by NS in the M2
series can be attributed to the lower cement content that lowered the
availability of CH due to the higher fly ash to cement mass ratio in the
mix design.

3.2. Thermal gravimetric analysis

Fig. 6. Compressive strength of the ECCs as function of curing age.
Fig. 4 illustrates thermogravimetric (TG) and differential thermal

Fig. 4. TG and DTG curves of the ECC mixtures at 28 days: (a) M1 mixes; (b) M2 mixes.

4
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

Fig. 7. Tensile stress-strain curves of the PE-ECC mixtures.

implies the role of NS in promoting the pozzolanic reaction with cement, with the dosage of NS, at all curing ages, thanks to the combined
due to the extremely high pozzolanic activity of NS. The more consumed nucleation, pozzolanic, and nanofiller effects of NS [17,18]. At 28 days,
CH indicates the production of more C–S–H gel, which likely translated the incorporation of NS at 1% by weight of binder increased the
to better strengths of the ECCs. The CH content (see Fig. 5) in the ECC compressive strength of PE-ECC by 15.8% and 8.0%, for the M1 and M2
mixtures was obtained from mass loss between 380 ◦ C and 450 ◦ C in the series, respectively.
TG curves. The CH in ECCs is normally produced as a hydration product At early age, the compressive strength of the M1 series developed
of cement particles, and can be consumed to form C–S–H gel upon rapidly as it can achieve more than 54 MPa at 3 days. In contrast, due to
pozzolanic reaction of supplementary cementitious materials (e.g., fly the insufficient cement content, the gain of compressive strength of the
ash, SF and NS). Fig. 5 reveals that the CH content in the M2 series is M2 series was much lower than that of the M1 series.
lower than that of the M1 series, which is because the relatively lower At 28 days, the compressive strength of M1 and M2 series ranges at
cement content and higher fly ash content in the M2 series. 84.9–94.9 MPa and 75.8–80.7 MPa, respectively, all of which exceed the
threshold strength value of high strength concrete (60 MPa) [32].
Therefore, the ECC mixtures developed in this work belong to the family
3.3. Compressive strength of ECCs of high strength concrete.
At 90 days, the compressive strength of M1-1% and M2-1% reach
Fig. 6 shows the compressive strength of the PE-ECC mixtures at 105.7 MPa and 97.9 MPa, respectively, which is a rather high value for
various curing ages. The compressive strength of all ECC mixtures concrete.
increased as the age advanced, due to the continuous hydration of It is worth noting that, after 28 days, the compressive strength of the
cementitious binder. For both series, the compressive strength increased

5
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

Table 3 saturated-multiple- cracking behavior of ECCs if the fiber-bridging ca­

The summary of tensile parameters of the ECC mixtures. pacity was not strengthened. For the above reasons, a weakened matrix
Mixtures Ref.- M1- M1-1% Ref.- M2- M2-1% toughness or stronger fiber/matrix interfacial bond is desired for a
M1 0.5% M2 0.5% higher ductility of PE-ECC.
First- 5.71 ± 5.85 ± 7.03 ± 3.54 ± 3.93 ± 3.98 ± As illustrated in Fig. 8, the incorporation of NS notably increased the
cracking 0.49 0.08 0.19 0.32 0.47 0.49 first-cracking (tensile) strength for both series of PE-ECC, especially for
strength the M1 series. Admixing NS at 0.5 wt% and 1 wt% improved the 28-day
(MPa) first-cracking strength of M1 series ECC by 2.4% and 23.4%, which
Tensile 8.01 ± 9.17 ± 9.97 ± 6.11 ± 6.65 ± 7.78 ±
strength 0.25 0.11 0.42 0.05 0.05 0.18
reached 5.85 MPa and 7.03 MPa, respectively. The likely mechanism is
(MPa) as follows: the combined nucleation and pozzolanic effects of NS facil­
Strain 3.27 ± 5.13 ± 5.17 ± 4.48 ± 5.44 ± 5.28 ± itated the formation of additional C–S–H gel, which along with the
capacity 0.26 0.38 0.27 0.19 0.43 0.37 nanofiller effect resulted in more densified ECC matrix. The higher first-
(%)
cracking strength of the ECC is desirable for structural members,
Modulus 26.10 28.32 30.42 24.13 26.54 29.07
(GPa) ± 1.63 ± 1.98 ± 1.68 ± 1.31 ± 1.97 ± 2.35 because the occurrence of cracking can be delayed. Note that the M2
series has a lower first-cracking strength than that of the M1 series, due
to the relative lower cement amount.
M2 series is comparable with that of the M1 series, howbeit with one It should be cautioned that the high first-cracking strength of ECC is
third less cement amount. This denotes that one can develop a cement- usually accompanied with high fracture toughness (Km) of the matrix. As
efficient high-strength ECC, by leveraging the pozzolanic reaction of plotted in Fig. 9, the value of crack tip toughness (Jtip = K2m/Em) follows
high volume of class C fly ash. the same trend with the Km value. In other words, the inclusion of NS in
ECC also caused the increment of Jtip, which implies that more energy is
3.4. Tensile performance of ECCs required to initiate the cracking of the concrete.
From Fig. 8 and Table 3, one can see that the tensile strength of the
Fig. 7 displays the tensile stress-strain relationship of the ECC mix­ PE-ECC mixtures showed the similar tendency to that of the first-
tures. From the curves, the tensile parameters, including first-cracking cracking strength, in terms of their influence by the admixed NS and
strength, elastic tensile modulus, tensile strength, strain capacity, and by the higher fly ash content. For instance, admixing NS at 0.5% and
tensile elastic modulus of ECCs can be gained. The first-cracking 1.0% by the weight of binder improved the tensile strength of M1 series
strength of ECC is defined as the stress value before the first drop of ECC by 14.5% and 24.5%, which reached 9.17 MPa and 9.97 MPa,
tensile stress, which corresponds to the formation of first micro-crack. respectively. On the other hand, the higher fly ash content and lower
The elastic tensile modulus is determined as the slope of the initial cement content caused a significant reduction on the tensile strength of
linear elastic stage in the curves. The tensile strength of ECC is the ul­ ECC mixtures; for instance, the tensile strength of PE-ECC was decreased
timate tensile stress in the curves, and its corresponding strain is defined
as the strain capacity. The value of tensile parameters of the ECC mix­
tures were averaged from three curves, and the results are summarized
in Table 3.
As revealed in Fig. 7, all PE-ECC mixtures exhibited high tensile
ductility accompanied with a conspicuous strain-hardening phenome­
non. The strain capacity of Ref.-M1 is 3.27%, which is about 300 times
that of conventional concrete; as such, Ref.-M1 can be considered as
high-ductility concrete. Nonetheless, this level of strain is considerably
less than that of PE-ECC, as reported in literature [9–15,35]. Moreover,
the cracking pattern that will be illustrated in a later section shows that
it is far from saturated-multiple-cracking status of ECC. Also, the crack
width on Ref.-M1 reached nearly 100 μm, which is undesirably high and
not beneficial for the occurrence of self-healing. This unsatisfactory
ductility in Ref.-M1 is likely a result of premature failure of the two core
criteria (strength and energy criteria) for proper design of ECC [36,37].
The high strength of plain cement-based materials normally comes with
high fracture toughness (Km). This likely postpones the initiation of
Fig. 9. The fracture properties of the ECC matrices.
cracks at a new site in the ECC specimen, sequentially hindering the

Fig. 8. Tensile parameters of the ECC mixtures: (a) M1 series; (b) M2 series.

6
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

Fig. 10. Representative SEM images of the fiber/matrix interface. Note: At least 5 points were selected within the interface area for the EDS analysis.

from 8.01 MPa in Ref.-M1 to 6.11 MPa in Ref.-M2. cement ratio) caused the fiber/matrix interface in Ref.-M2 to feature a
The tensile strength of ECCs is intrinsically associated with the peak more porous microstructure, which weakened the fiber/matrix
fiber bridging stress (σcu), above which the cumulative tensile force is interface.
able to pull plenty of a single fiber out from its surrounding block, i.e., Fig. 8 also shows that the tensile strain capacity of HS-ECCs was
overcoming the fiber/matrix interfacial frictional bond (τ0). Based on considerably increased upon the incorporation of NS, up to 5.17% and
the micromechanics theory [38], the σcu is proportional to τ0, as 5.44% for both M1 and M2 series, respectively. Admixing NS at 0.5 wt%
expressed in Eq. (3) in which fiber rupture and slip-hardening are and 1 wt% improved the 28-day strain capacity of M1 series ECC by
ignored for simplicity. The coefficient (g) is the snubbing factor that can 56.9% and 58.1%, respectively. The increment in strain capacity
be considered as a constant for a specific fiber type. [Note: the symbols induced by the admixing of 0.5 wt% NS can be attributed to the afore­
are summarized in Appendix I]. According to Eq. (3), the change of mentioned strengthening of the fiber/matrix interface as well as the
τ0 value in the PE-ECC mixtures follows the same trend as that of tensile subsequent fiber-bridging capacity. Based on the ECC design theory
strength; that is, the admixed NS strengthened the fiber/matrix interface [38], the complementary energy (Jb’) can be simplified as Eq. (5) for
whereas the increase in fly ash to cement ratio weakened it. fibers of hydrophobic nature. From Eq. (5), one can deduce that the
( ) incorporation of NS into HS-ECC benefits the increment of Jb’ value
Vτ L
σ cu = f 0 f ⋅g (3) since the τ0 value was enhanced. This indicates more sufficient residual
2 df
energy is available to trigger cracks at new sites, which is conducive to
the occurrence of saturated-multiple cracking. As mentioned previously,
2 ( )
g= 1 + eπf /2 (4) the Jtip of ECC mixtures was also improved by the admixed NS. To
4 + f2
facilitate the saturated multiple cracking behavior (i.e., ductility) of
The frictional bond between fiber and cementitious matrix is closely ECC, one needs to improve the value of pseudo strain-hardening index
correlated with the stiffness and packing density of the interfacial (PSH = Jb’/Jtip); that is, the degree of increase in Jb’ should prevail over
transition zone [39]; the alteration in the τ0 value can thus be demon­ that in Jtip.
strated by the morphology of the fiber/matrix interface. As revealed in ( )
Fig. 10, for both series of HS-ECC mixtures, the incorporation of NS Lf τ20 L2f
(5)
′
J b = Vf
decreased the Ca/Si ratio of hydration products on the fiber surface, df 6df Ef
which indicates a higher polymerization degree of C–S–H [40]. As a
It is interesting to note that the strain capacity showed a comparable
result, a more refined and denser fiber/matrix interface can be observed
level between the PE-ECC mixtures with NS dosage of 0.5 wt% and 1 wt
in the NS-modified ECC mixtures, as compared with that of the reference
% (see Fig. 8), although more cracks were initiated on M1-1% and M2-
ones. On the other hand, by comparing Ref.-M1 and Ref.-M2, one can
1%. This can be explained by the narrowed crack width (see Table 4).
observe that the insufficient cement hydration (due to the high fly ash to

7
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

Table 4 3.5. Autogenous self-healing phenomenon of the PE-ECCs

The crack information of ECC mixtures.
Ref.- M1- M1- Ref.- M2- M2- Fig. 13 presents the visual observation of autogenous self-healing
M1 0.5% 1% M2 0.5% 1% phenomenon on the pre-damaged PE-ECC specimens after undergoing
Number of 25 46 53 58 72 77 30 wet-dry cycles. The representative observation on a portion of ECC
cracks specimens was selected to demonstrate the extent of self-healing. As
Average crack 97.6 64.6 58.1 57.4 46.5 23.3 illustrated in Fig. 13, the self-healing in Ref.-M1 was rarely observed,
width μm μm μm μm μm μm and this is due to its large crack width. Upon incorporating NS, the crack
width on the ECC specimens was greatly reduced; hence white residues
Compared with that of 0.5 wt% NS case, adding NS at 1 wt% into PE-ECC were found within cracks in M1-0.5% and M1-1%. Compared with Ref.-
can further enhance the fiber/matrix interfacial bond, which in turn M1, the crack width was also narrowed in Ref.-M2, because of the
restrained the crack width more effectively. It is worth noting that, the increased fly ash content. As a result, the cracks were filled by self-
strain capacity of the M2 series is higher than that of the M1 series, and healing products, howbeit to a limited extent. Among all PE-ECC mix­
this is due to the relative lower matrix toughness of the PE-ECC mixtures tures in this work, the most efficient self-healing was observed in M2-1%
with the higher fly ash to cement ratio. because the vast majority of the cracks were no wider than 30 μm.
Fig. 11 presents the crack pattern of the ECC specimens within the To further analyze the chemical composition of self-healing prod­
gauge length, after the tensile test. The number of cracks on the speci­ ucts, the HS-ECC specimens were examined using SEM and EDS. As
mens denotes the degree of saturated-multiple-cracking behavior, while displayed in Fig. 14, some discrete stone-like particles filled in the cracks
the maximum and average crack widths directly determine the transport on M1-0.5%, M1-1%, Ref.-M2%, and M2-0.5%, whereas self-healing
properties and are critical factors to self-healing of the ECC. The detailed products with dense microstructure were observed in the cracks on
crack information is summarized in Table 4. M2-1%. For each ECC specimen, at least five different points on the self-
As shown in Fig. 11 and Table 4, the admixed NS resulted in more
cracks being triggered on the ECC specimens. The number of cracks on
M1-0.5% and M1-1% was 46 and 53, which is 84% and 112% more than
that of 25 on Ref.-M1, respectively. The higher degree of saturated-
multiple-cracking behavior of NS-modified ECC confirms that the
incorporation of NS improved more in Jb’ than that in Jtip.
In addition, narrower cracks were triggered on the ECC specimens
incorporating NS. The average crack width in Ref.-M1 is 97.6 μm, which
is not conducive to the occurrence of self-healing of the HS-ECC. As
illustrated in Fig. 12, the admixed NS effectively narrowed down the
cracks. For instance, the average crack width in M1-1% and M2-1% was
reduced to 58.1 μm and 23.3 μm, respectively. Notably the maximum
crack width in M2-1% ECC specimens was only 40 μm. The tighter crack
widths are desirable for facilitating the self-healing of the HS-ECC.
Furthermore, the narrowed crack width helps explain why the strain
capacity in M1-1% and M2-1% ECC specimens was not better than that
of their M1-0.5% and M2-0.5% counterparts, howbeit with the forma­
tion of more cracks.

Fig. 12. Gaussian fitting of crack distribution as a function of crack width.

Fig. 11. Representative crack pattern in the ECC specimens.

8
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

Fig. 13. Autogenous self-healing in the ECC specimens after 30 wet-dry cycles.

Fig. 14. SEM images of self-healing products on the HS-ECC specimens.

product on M2-1% was mainly C–S–H gel. This finding agrees well
Table 5
with that reported in Ref. [41], in which different kinds of self-healing
EDS element analysis of the self-healing products.
products were found in cracks with various widths.
Point C O Mg Al Si Ca In summary, the admixed NS resulted in the reduction of crack width
1 8.06 56.02 – – 1.48 34.44 in HS-ECC specimens, and as a consequence, improved the autogenous
2 – 32.67 – 1.49 23.17 31.41 self-healing behavior of the HS-ECC.
3 14.87 36.52 2.32 – 8.65 32.91

4. Conclusion and outlook

healing products were analyzed using EDS. For example, the chemical
composition of the self-healing products on M1-1% and M2-1% is pre­ This work experimentally investigated the mechanical properties of
sented in Table 5. The results reveal that the particle-shaped self-healing high-strength ECCs (water/binder ratio of 0.204) incorporating hydro­
product on M1-1% was mainly CaCO3, whereas the dense self-healing phobic PE fibers and different levels of nanosilica (0%, 0.5 wt%, and 1
wt%) and class C fly ash (40.4% and 58.5% by weight of cementitious

9
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

binders). Subsequently, this work microscopically examined the autog­ National Natural Science Foundation of China (Grant No. 52078083),
enous self-healing behavior of the ECCs. The main conclusions can be the 111 Project of China (Grant No. B18062), and the Fundamental
drawn as follows. Research Funds for the Central Universities (Grant No. 2021CDJQY-
008).
(1) Admixing NS at either 0.5% or 1% by weight of binder was able to
make the ECC matrix denser and strengthen the interfacial bond Appendix I. Notation
between PE fiber and matrix. The nano-modified mixtures ach­
ieved enhanced mechanical performances, narrowed the crack
width and facilitated autogenous self-healing of PE-ECC. The symbols are used in equations
(2) At 28 days, the compressive strength of all PE-ECC mixtures σ cu peak fiber bridging stress
exceeded 75 MPa, and the incorporation of NS at 1% by weight of τ0 fiber/matrix interfacial frictional bond
binder increased the compressive strength of PE-ECC by 15.8% g snubbing factor
and 8.0%, for the M1 and M2 series, respectively. At 28 days, f snubbing coefficient
admixing NS at 1 wt% improved the tensile strength, first- Vf volume of fibers
cracking strength, and strain capacity of M1 series ECC by Lf fiber length
24.5%, 23.4%, and 58.1%, which reached 9.97 MPa, 7.03 MPa df fiber diameter
and 5.17%, respectively. Ef fiber-elastic modulus
(3) Admixing NS in the PE-ECC mixtures at 0.5 wt% and 1 wt% both Em matrix-elastic modulus
increased the number of cracks and narrowed the width of cracks Km matrix-fracture toughness
formed on the specimens. For instance, the incorporation of 1 wt Jtip: crack-tip fracture toughness
% NS reduced the average crack width of PE-ECC by 59.5% and J b’ complementary energy
59.4%, which reached 58.1 μm and 23.3 μm, for the M1 and M2
series, respectively. The higher degree of saturated-multiple- References
cracking behavior of NS-modified ECC confirms that the incor­
poration of NS improved more in Jb’ than that in Jtip. The nar­ [1] V.C. Li, C.K. Leung, Steady-state and multiple cracking of short random fiber
composites, J. Eng. Mech. 118 (1992) 2246–2264.
rower crack widths facilitated autogenous self-healing of PE-ECC
[2] V.C. Li, D.K. Mishra, H.C. Wu, Matrix design for pseudo strain-hardening fiber
specimens subjected to 30 wet-dry cycles, where particle-shaped reinforced cementitious composites, EILEM J. Mater. Struct. 28 (183) (1995)
product (mainly calcite) and dense product (mainly C–S–H gel) 586–595.
formed on the M1 and M2 series PE-ECC, respectively. [3] V.C. Li, C. Wu, S. Wang, A. Ogawa, T. Saito, Interface tailoring for strain-hardening
polyvinyl alcohol-engineered cementitious composite (PVA-ECC), ACI Mater. J. 99
(4) The hardened PE-ECC specimens (at 28 days of age) exhibited (5) (2002) 463–472.
higher bound water contents (up to 17.4% increase) induced by [4] Z. Zhang, S. Liu, F. Yang, Y. Weng, S. Qian, Sustainable high strength, high ductility
the admixed NS, along with lower CH contents (up to 50.1% engineered cementitious composites (ECC) with substitution of cement by rice husk
ash, J. Clean. Prod. 317 (2021), 128379.
decrease). The admixed NS induced the formation of a denser [5] M. Li, V.C. Li, High-early-strength engineered cementitious composites for fast,
microstructure at the fiber/matrix interface (along with a denser durable concrete repair-material properties, ACI Mater. J. 108 (1) (2011) 3–12.
and stronger cementitious matrix), resulting in better interfacial [6] M.D. Lepech, V.C. Li, Water permeability of engineered cementitious composites,
Cement Concr. Compos. 31 (2009) 744–753.
bond that accounts for the improved tensile performance of the [7] A. Alyousif, M. Lachemi, G. Yildirim, M. Şahmaran, Effect of self-healing on the
PE-ECC specimens. different transport properties of cementitious composites, J. Adv. Concr. Technol.
13 (3) (2015) 112–123.
[8] Z. Zhang, Q. Zhang, V.C. Li, Multiple-scale investigations on self-healing induced
To calculate the PSH index, the complementary energy (Jb’) is
mechanical property recovery of ECC, Cement Concr. Compos. 103 (2019)
needed. Future work should explore the single fiber pull-out test from 293–302.
which one can obtain the fiber/matrix interfacial parameters and then [9] Z. Zhang, A. Yuvaraj, J. Di, S. Qian, Matrix design of light weight, high strength,
high ductility ECC, Construct. Build. Mater. 210 (2019) 188–197.
calculate the theoretical Jb’ value. Future work may also explore the
[10] K. Yu, Y. Wang, J. Yu, S. Xu, A strain-hardening cementitious composites with the
selection of nanomaterials other than nanosilica to facilitate the satu­ tensile capacity up to 8, Construct. Build. Mater. 137 (2017) 410–419.
rated multiple cracking behavior of PE-ECC, balancing the consider­ [11] Y. Ding, J.T. Yu, K.Q. Yu, S. Xu, Basic mechanical properties of ultra-high ductility
ations over first cracking strength, Jtip, and Jb’. This work, however, has cementitious composites: from 40 MPa to 120 MPa, Compos. Struct. 185 (2018)
634–645.
demonstrated the concept of nano-modification of PE-ECC to achieve [12] Z. Zhang, F. Yang, J. Liu, S.P. Wang, Eco-friendly high strength, high ductility
outstanding autogenous self-healing as well as use of relatively high engineered cementitious composites (ECC) with substitution of fly ash by rice husk
volume of supplementary cementitious materials. ash, Cement Concr. Res. 137 (2020), 106200.
[13] L.Y. Xu, B.T. Huang, J.C. Lao, J.G. Dai, Tailoring strain-hardening behavior of high-
strength Engineered Cementitious Composites (ECC) using hybrid silica sand and
Declaration of competing interest artificial geopolymer aggregates, Mater. Des. 220 (2022), 110876.
[14] L.Y. Xu, B.T. Huang, V.C. Li, J.G. Dai, High-strength high-ductility Engineered/
Strain- Hardening Cementitious Composites (ECC/SHCC) incorporating
The authors declare that they have no known competing financial geopolymer fine aggregates, Cement Concr. Compos. 125 (2022), 104296.
interests or personal relationships that could have appeared to influence [15] B.T. Huang, J.X. Zhu, K.F. Weng, V.C. Li, J.G. Dai, Ultra-high-strength engineered/
the work reported in this paper. strain- hardening cementitious composites (ECC/SHCC): material design and effect
of fiber hybridization, Cement Concr. Compos. 129 (2022), 104464.
[16] C. Edvardsen, Water permeability and autogenous healing of cracks in concrete,
Data availability ACI Mater. J. 96 (4) (1999) 448–455.
[17] X. He, X. Shi, Chloride permeability and microstructure of Portland cement mortars
incorporating nanomaterials, Transport. Res. Rec. 2070 (1) (2008) 13–21.
Data will be made available on request.
[18] S. Du, J. Wu, O. AlShareedah, X. Shi, Nanotechnology in cement-based materials: a
review of durability, modeling, and advanced characterization, Nanomaterials 9
Acknowledgements (9) (2019) 1213.
[19] K. Behfarnia, N. Salemi, The effects of nano-silica and nano alumina on frost
resistance of normal concrete, Construct. Build. Mater. 48 (2013) 580–584.
The authors acknowledge the financial support of this work by the U. [20] A.M. Said, M.S. Zeidan, M.T. Bassuoni, Y. Tian, Properties of concrete
S. Department of Transportation through the National University incorporating nano-silica, Construct. Build. Mater. 36 (2012) 838–844.
Transportation Center, National Center for Transportation Infrastruc­ [21] H. Li, H. Xiao, J. Yuan, J. Ou, Microstructure of cement mortar with nano-particles,
Composites Part B 35 (2) (2004) 185–189.
ture Durability and Life-Extension (TriDurLE) (USDOT, Grant No. [22] H.J. Du, S.H. Du, X.M. Liu, Effect of nano-silica on the mechanical and transport
69A3551947137). Z. Zhang acknowledges the financial support by the properties of lightweight concrete, Construct. Build. Mater. 82 (2015) 114–122.

10
Z. Zhang et al. Cement and Concrete Composites 135 (2023) 104837

[23] M. Rupasinghe, R. Nicolas, P. Mendis, M. Sofi, T. Ngo, Investigation of strength and [31] Y. Zhou, S. Zheng, et al., Performance enhancement of green high- ductility
hydration characteristics in nano-silica incorporated cement paste, Cement Concr. engineered cementitious composites by nano-silica incorporation, Construct. Build.
Compos. 80 (2017) 17–30. Mater. 281 (2021), 122618.
[24] Z. Wu, K.H. Khayat, C. Shi, Effect of nano-SiO2 particles and curing time on [32] (includes amendment No. 3, May 2013 (incorporating amendments 1 and 2)), TR
development of fiber-matrix bond properties and microstructure of ultra-high 49 Design Guidance for High Strength Concrete, 1998.
strength concrete, Cement Concr. Res. 95 (2017) 247–256. [33] ASTM C109, Standard Test Method for Compressive Strength of Hydraulic Cement
[25] S. Yeşilmen, Y. Al-Najjar, M.H. Balav, M. Şahmaran, et al., Nano-modification to Mortars, ASTM International, West Conshohocken, PA, 2008.
improve the ductility of cementitious composites, Cement Concr. Res. 76 (2015) [34] ASTM E399-12, Standard Test Method for Linear-Elastic Plane-Strain Fracture
170–179. Toughness KIc of Metallic Materials, 2014.
[26] Y. Al-Najjar, S. Yeşilmen, A.M. Al-Dahawi, M. Şahmaran, et al., Physical and [35] L. Li, Z. Cai, K. Yu, Y.X. Zhang, Y. Ding, Performance-based design of all-grade
chemical actions of nano-mineral additives on properties of high-volume fly ash strain hardening cementitious composites with compressive strengths from 40 MPa
engineered cementitious composites, ACI Mater. J. 113 (6) (2016) 791–801. to 120 MPa, Cement Concr. Compos. 97 (2019) 202–217.
[27] B.S. Mohammed, B.E. Achara, M.S. Liew, The influence of high temperature on [36] T. Kanda, V.C. Li, A new micromechanics design theory for pseudo strain
microstructural damage and residual properties of nano-silica-modified (NS- hardening cementitious composite, J. Eng. Mech. ASCE 125 (4) (1999) 373–381.
modified) self-consolidating engineering cementitious composites (SC-ECC) using [37] T. Kanda, V.C. Li, Multiple cracking sequence and saturation in fiber reinforced
response surface methodology (RSM), Construct. Build. Mater. 192 (2018) cementitious composites, JCI Concr. Res. Technol. 9 (2) (1998) 19–33.
450–466. [38] S. Wang, V.C. Li, Engineered cementitious composites with high-volume fly ash,
[28] S.M. Razavi, H. Nazarpour, M.H. Beygi, Investigation of the efficacy of nano-silica ACI Mater. J. 104 (3) (2007) 233–241.
on mechanical properties of Green-Engineered Cementitious Composite (GECC) [39] O. Kayali, Effect of high volume fly ash on mechanical properties of fiber
containing high volume natural zeolite, Construct. Build. Mater. 291 (2021), reinforced concrete, Mater. Struct. 37 (269) (2004) 318–327.
123246. [40] Z. Wang, Y. Chen, L. Xu, et al., Insight into the local CSH structure and its evolution
[29] C. Fu, R. Guo, Z. Lin, H. Xia, Y. Yang, Q. Ma, Effect of nanosilica and silica fume on mechanism controlled by curing regime and Ca/Si ratio, Construct. Build. Mater.
the mechanical properties and microstructure of lightweight engineered 333 (2022), 127388.
cementitious composites, Construct. Build. Mater. 298 (2021), 123788. [41] L.L. Kan, H.S. Shi, A.R. Sakulich, V.C. Li, Self-healing characterization of
[30] B. Xi, Y. Zhou, K. Yu, et al., Use of nano-SiO2 to develop a high performance green engineered cementitious composite materials, ACI Mater. J. 107 (6) (2010)
lightweight engineered cementitious composites containing fly ash cenospheres, 617–624.
J. Clean. Prod. 262 (2020), 121274.

11