Corrosion Science 106 (2016) 1–15

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

Correlation between hydration of cement and durability of natural

fiber-reinforced cement composites
Jianqiang Wei a,b,∗ , Siwei Ma a , D’Shawn G. Thomas b
a
Department of Civil Engineering and Engineering Mechanics, Columbia University, New York, NY 10027, USA
b
The Lyles School of Civil Engineering, Purdue University, West Lafayette, IN 47907, USA

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

Article history: The influence of cement hydration on the durability of natural fiber-reinforced cementcomposites and
Received 15 June 2015 the deterioration of the embedded natural fibers were investigated by incorporating four supplementary
Received in revised form 13 January 2016 cementitious materials (SCMs). Cement hydration was presented to be a crucial factor in understanding
Accepted 21 January 2016
natural fiber degradation behavior. The results indicated that the incorporation of SCMs slowed down
Available online 24 January 2016
both mineralization and alkaline hydrolysis of fiber cell walls by promoting the hydration of cement.
This is attributed to the reduced alkalinity of pore solution and calcium hydroxide content. Degrada-
Keywords:
tion of natural fiber was significantly mitigated and the durability of cement composites was improved
A. Concrete
B. SEM
consequently.
B. X-ray diffraction © 2016 Published by Elsevier Ltd.
C. Alkaline corrosion
C. Inclusion

1. Introduction [2–5]. Cellulose, which is a kind of linear hemi-crystal phase is

Cement based composite materials, especially concrete, play an
important role in the history of human development in the last 2000
years. Attributed to excellent compressive strength, low-cost, long
service life and relatively low maintenance requirements, cement
composites have already been applied in various infrastructures.
The main disadvantages of cement composites are their low tensile
strength, cracking resistance and fracture toughness. Steel reinforc-
ing bars and various fibers, such as steel fiber, mineral fibers, glass
fiber and plastic fibers, have been used to reinforce mortar and con-
crete to control the initiation and growth of microcracks [1]. The
desirable improvements of tensile behavior, flexural strength and
toughness, and ductility can all be realized by using fiber reinforce-
ment. The increase in fracture energy is caused by the fiber itself and
the bond between the fiber and matrix. To meet the requirement of
sustainable development of construction materials, natural fibers
attract more attention as reinforcement of cementitious materials.
Owing to its high tensile strength, high modulus and low cost,
sisal fiber has proven to be a suitable natural reinforcement to
cement based materials. It is well known that natural fiber con-
sists of three main components: cellulose, hemicellulose and lignin
∗ Corresponding author at: HAMP Hall of Civil Engineering, G144, 550 Stadium
Mall Drive, West Lafayette, IN 47907, USA.
E-mail addresses: jw2938@columbia.edu, wei68@purdue.edu (J. Wei).
the main structural components for natural fiber [6]. Hemicellu-
lose and lignin are amorphous phases and are binding phases for
cellulose [5,7]. The distinct cell structure endows sisal fiber excel-
lent mechanical properties with low density, meanwhile it is also
the reason for natural fiber’s poor durability in alkaline environ-
ments, such as the cement matrix. When exposed to alkaline pore
solution and the mineral-rich environment of the cement matrix,
natural fiber will experience severe degradation and will become
brittle due to two aging mechanisms: alkaline hydrolysis and cell
wall mineralization. This significantly affect the fiber’s reinforc-
ing role in cement composites. Furthremore, cement composites
can suffer premature deterioration including a decrease in post-
cracking toughness and cracking when they undergo various aging
processes.
Sisal fiber degradation can be mitigated through two ways:
fiber pretreatment and cement matrix modification. Pretreatment
of natural fiber includes chemical, physical and physicochemical
methods. Silane coating, hornification, autoclave, sodium silicate,
potassium silicate [8–11] have been shown to improve the mechan-
ical properties and durability of natural fiber in cement based
materials. Bastidas et al. [12] and Zornoza et al. [13] reported that
the corrosion resistance of steel embeded in mortar exposed to
chloride can be substantially improved by incorporating fly ash.
Moreover, the partial cement replacement by slag [14] and silica
fume [15] has also been proven to be an effective way to miti-
gate the corrosion of rebar in concrete. So it can be anticipated

http://dx.doi.org/10.1016/j.corsci.2016.01.020
0010-938X/© 2016 Published by Elsevier Ltd.
2 J. Wei et al. / Corrosion Science 106 (2016) 1–15

that the aging behavior of natural reinforcements can be arrested Table 1

Chemical composition of the used materials.
through cement modification using these supplumentary materi-
als. Tolêdo’s work indicates that natural fiber degradation can be Compositions Cement MK RHA SF FA
effectively mitigated by producing a calcium hydroxide (Ca(OH)2 ) CaO 63.7 0.0707 0.51 4.89 27.31
free cement matrix through high volume cement replacement with SiO2 12.9 51.8 90.45 92.3 32.25
calcined clay (metakaolin (MK) and calcined waste crushed clay Fe2 O3 7.97 4.15 0.015 0.84 6.85
brick) [16–18] or MK [19]. Ground granulated blast furnace slag SO3 5.25 0.105 0.037 0.30 2.82
Al2 O3 4.14 42.4 0.016 0.665 17.41
[20,21], silica fume, fly ash, and MK [22,23] have similar effects. In
MgO 3.50 – 0.24 0.525 5.32
the author’s previous work, both fiber pretreatment and cement SrO 0.915 0.0397 – 0.0398 0.51
matrix modification by using various supplementary cementitious K2 O 0.907 0.218 1.67 0.263 0.37
materials, such as metakaoline (MK), nanoclay (NC), limestone (LS), Na2 O – – 0.031 – 1.67
TiO2 0.279 1.07 0.009 – 1.52
rice husk ash (RHA), fly ash (FA) and diatomite have been inves-
ZnO 0.251 – – 0.105 0.11
tigated. The results indicate that both of the two methods can ZrO2 0.119 0.0884 – – 0.15
effectively arrest the natural fiber’s degradation. However, pre- P2 O5 – – 0.76 – 0.52
treatment of natural fiber may require more effort, can increase MnO – – 0.07 – 0.10
cost, and needs to consider the compatibility between modifying Cl – 0.0457 – 0.0465 –
C – – 6.19 – 3.09
agents and the cement matrix, as well as its effect on the interfacial
Other 1.21 0.0125 0 0.0257 0
properties of fiber-cement. Therefore, it is more logical to improve
the initial mechanical properties and durability of natural fiber-
reinforced cement composites through modifying the hydration of of laser diffraction are shown in Fig. 1. The median particle size
cement. of cement (D50) is 9.57 ␮m and its surface weighted mean (the
Altough cement matrix modification promises to be an effc- diameter of a sphere that has the same surface area ratio as a par-
tieve way to improve the durability of natural fiber in cement. ticle of interest) and volume weighted mean (the diameter of a
The modification mechanims of these SCMs in cement and corre- sphere that has the same volume ratio as a particle of interest) are
lation between cement hydration and fiber degradation, which are 3.5 ␮m and 12.3 ␮m, respectively. MK shows a higher fraction of
the foremost objectives of this study, are not yet fully understood. particles with a diameter smaller than 5 ␮m compared to PC. The
In this study, in order to get the cement martices with different Brunauer–Emmett–Teller (BET) specific surface area (surface areas
degrees of hydration and to investigate the influence of the alka- of solids by physical adsorption of gas molecules based on BET multi
linity of pore solution on fiber degradation behavior, four SCMs, layer adsorption theory; the specific surface area is obtained from
such as MK, RHA, silica fume (SF) and FA, were selected to partially total surface area divided by mass) of MK (2.93 m2 /g) is higher than
replace cement with different mixing amounts. Given the differ- that of PC (1.7 m2 /g), which means it can modify the cement matrix
ences in their chemical composition, particle size and pozzolanic in both hydration products and microstructure as pozzolanic sub-
activity, the cement substitution level was adjusted for each SCM stitution material and fine filler, respectively. Although having a
to ensure the diversity in cement hydration. The effect of these similar SiO2 content, RHA’s median particle size and surface area
SCMs on cement hydration kenetics and hydration products, as are 8.93 times and 24.02% of those of SF, respectively. It can be
well as their effect on durability of natural fiber-reinforced cement anticipated that SF will improve cement hydration and consume
composites were investigated. The correlation between degree of Ca(OH)2 more effectively than RHA, and consequently enhance the
cement hydration and natural fiber degradation behavior was also durability of natural fiber-reinforced cement composites. However,
determined. the distinct porous structure of RHA (Fig. 2-a) endows it a great
potential to serve as external water carrier to realize the internal
curing, which is beneficial to enhance the hydration of cement. The
2. Experimental particle size distribution of FA is more decentralized than those of
other SCMs. A median particle size of 13.2 ␮m was obtained from
2.1. Raw materials FA, with a BET-specific surface areas of 1.46 m2 /g.
Sisal fibers produced from Tanzania, with an average diameter
The ASTM Type I Portland cement (PC), used in this paper, with of 202.5 ␮m and a tensile strength of 605 MPa, was provided by Bast
a Blaine fineness of 385 m2 /kg, was sourced from Lafarge North Fibers LLC of Creskill, New Jersey. Fig. 2b shows the microstructures
America Inc., Whitehall, PA, USA. Aggregate used to prepare mor- of the surface of a sisal fiber, respectively. The fiber’s rough surface
tar specimens consisted of natural river sand that had a maximum is beneficial to enhance interfacial bonding properties between the
aggregate size of 4.7 mm, specific gravity of 2.6 and fineness mod- fiber and cement matrix, but it also provides space for the precipi-
ulus of 2.8. Metakaolin (MK) was sourced from Pivorpozz Inc. Rice tation of cement hydration products on fiber surface.
husk ash was supplied by Agrilectric Research Co., Lake Charles,
LA. For comparison, SF and FA were also incorporated into cement 2.2. Mixture proportion and specimens preparation
as SCM to investigate their effects on cement hydration and fiber
degradation. The chemical compositions of cement and SCMs deter- Pure cement matrix and blends with cement substitutions by
mined by means of X-ray fluorescence (XRF) are summarized in MK, RHA, SF and FA at 30 wt.% (MK30), 20 wt.% (RHA20), 10 wt.%
Table 1. (SF10), and 30 wt.% (FA30) levels, respectively, were investigated
From Table 1, it can be seen that MK has a higher aluminate con- in this study. A water/cementitious materials/sand ratio of 0.4:1:1
tent than other SCMs. So it can be predicted that MK will enhance was used for all mortar mixtures. If necessary, the workability was
the hydration of cement more significantly at early age. The silicate adjusted when necessary using ADVA 408 superplasticizer up to
mineral contents of RHA and SF are much higher than that of other 3 wt.% of binder, which is not enough to significantly impact the
SCMs. FA, which is a kind of high-calcium Class C fly ash shows hydration kinetics of cement [24]. The cement pastes were mixed
higher lime content than SF. The silicate content of FA is lower than with gradually added sand in a mechanical mortar mixer at 60 rpm
other SCM, but it has a higher aluminate amount than SF and RHA. for 2 min followed by a 1 min rest and 3 min of further mixing at
The particle size distributions (PSD) of cement and the sup- 120 rpm. Cement pastes prepared for chemical analysis were cast
plementary cementitious materials (SCM) determined by means in sealed bottles and then stored at 23 ± 2 ◦ C until to be tested.
 J. Wei et al. / Corrosion Science 106 (2016) 1–15 3

Fig. 1. Particle size distributions of cement and the SCMs: (a) relative frequency of particles in dependence of the diameter (noncumulative), and (b) volume of the particles
smaller than a certain diameter (cumulative).

Fig. 2. The microstructure of RHA (a) and sisal fiber surface (b).

Fig. 3. The process of casting sisal fiber-reinforced mortar beams (a) pave long aligned sisal fiber layer and (b) mortar layer.

Three 200 mm × 50 mm × 12 mm beams for four-point flexural
test were cast for each testing point. Two layers of long aligned
sisal fiber with a total volume fraction of 2% were used to rein-
force cement mortar beams. As shown in Fig. 3a, a mortar layer
of 2 mm thickness was placed at the bottom of the mold then a
layer of sisal fiber (1% volume fraction) was carefully laid. A sec-
ond layer of mortar of approximately 3–4 mm thickness was then
spreaded followed by the second layer of sisal fiber. The fabric was
fully saturated with paste by a roller and for each specimen, the
final layer was cement mortar (Fig. 3b). A vibration table was used
to consolidate the mortar beams after filling each layer of mortar.
The duration and frequency of vibration were established accord-
ing to American Society for Testing Material (ASTM) C192 [25]. The
specimens were demolded after 1 day and immersed in Ca(OH)2 -
saturated water at 23 ± 2 ◦ C for 28 days prior to accelerated aging
treatment.
Three 30 mm × 30 mm × 300 mm prismatic samples for each
testing point were prepared for fiber pull-out tests. Sisal fibers were
vertically embedded in the matrix with an embedment length of
30 mm. The samples were demolded after 24 h. Five specimens per
data point were tested after 1, 7 and 28 days of curing at 23 ◦ C with
a relative humidity of 50%.
4 J. Wei et al. / Corrosion Science 106 (2016) 1–15
2.3. Accelerated aging condition
The durability of sisal fiber reinforced cement based compos-
ites was determined by the degradation of mechanical properties
of specimens subjected up to 50 cycles of wetting and drying, which
can accelerate the deterioration of both sisal fiber and the cement
matrix. For the drying and wetting cycles, the specimens were
submerged in sealed tap water at 70 ◦ C and then oven dried in a
circulating air environment at 70 ◦ C alternately. In order to deter-
mine the duration of each cycle, the development of mass change as
a function of immersing or drying time was investigated. Increase
and decrease of specimen mass subjected to wetting and drying
cycles were measured every 10 min for the first 2 h, every 1 h subse-
quently and every 3 (for wetting) or 4 (for drying) hours after 12 h.
Samples were considered at equilibrium when the mass change
was less than 2% over 3 consecutive measurements.
2.4. Flexural properties of natural fiber-reinforced cement mortar
The durability of sisal fiber-reinforced cement mortar was eval-
uated by means of flexural properties of the beams after accelerated
aging. Three identical beam specimens with a span of 150 mm were
bended on an Instron 5984 34k Universal Testing Machine with a
crosshead rate of 0.2 mm/min. The displacements at midspan were
determined using linear variable differential transformers (LVDTs).
According to ASTM C 1185 [26], two indices, post-cracking strength
( p ) and post-cracking toughness (Ip ) were selected to evaluate
the durability of sisal fiber-reinforced cement mortars. The post-
cracking strength was determined using the equation:
Fmax L
p = (1)
bh2
where L is the span of specimens (150 mm), b and h are the width
and thickness of specimens, respectively. For the beams subjected
to a larger number of wetting and drying cycles, and in absence
of post-cracking toughening, the peak strength will be equal to
the first crack strength. Post-cracking toughness is defined as the
amount of energy absorption after an initial crack, which is calcu-
lated as the area under a load-deflection curve after first cracking
up to specified deflection criterion. In this study, the test was termi-
nated when deflection reaches 16.5 mm. For the aged sample, the
maximum deflection was less than 16.5 mm. The test was ended
when the load drop to 1/5 of its maximum load. The post-cracking
toughness was determined using the equation:
 16.5or0.2εT
Ip = Fdε (2)
ε0
where F is load and ε is deflection in bending test. ε0 and εT are
deflections at first cracking and maximum load, respectively.
2.5. Uniaxial tensile behavior of embedded fiber and interfacial
bonding property
In accordance to ASTM D 3822 [27], the uniaxial tensile proper-
ties of single raw and immersed fibers, which were separated out
from cement pastes after aging as described in the author’s previ-
ous work [28], were determined using a micro-force Instron 5948
testing system with a gauge length of 20 mm and a loading rate of
0.2 mm/min. Twenty single fibers were tested for each group and
the scattered data of the tensile tests were analysed using Weibull
distribution model. Probability function of this double-parameter
distribution is given by:
 
f (x) = m (x)m−1 exp −xm (3)
where f(x) is the frequency distribution of the variable x, and m is
the Weibull modulus, which indicates the scatter degree of testing
data.
The failure probability P can be calculated as cumulative distri-
bution function by defining the variable x as / 0 and integrating
both sides of Eq. (3):
  m 

P = 1 − exp − (4)
0
where  is the measured failure stress of each fiber and  0 is a
normalizing parameter, which can be defined as a characteristic
strength of each group.
By taking logarithm of both sides twice, it yields:
 1 
lnln = mln − mln0 (5)
1−P
The Weibull modulus and characteristic strength can be deter-
mined by plotting the straight line fitting ln vs. lnln ((1 − P)−1 ).
The failure probability P is actually a provability index:
R
P= (6)
N+1
where R is the number of fibers broken at stress  R , and N is the
total number of sisal fibers that have been tested (here N = 20).
The interfacial bonding properties between fiber and cement
matrices were determined by means of fiber pull-out tests, which
were carried out using the same micro-force device for fiber tensile
testing with a loading rate of 0.2 mm/min. It is well known that the
typical force-displacement curve of fiber pull-out test consists of
four stages: (i) in the linear region of the typical force-displacement
curve (when the load is lower than the maximum load Fd of linear
region), the interface remains intact and the linear curve is gen-
erated by the elastic deformation of the fiber, (ii) when the load
becomes higher than Fd but lower than the maximum pull-out load
(Fmax ), the fiber begins to debond from the matrix through interfa-
cial crack propagation in the interfacial zone [29,30], (iii) after the
peak load value Fmax , the whole embedded length fully debonds
from the matrix through unstable crack propagation, and (iv) the
last stage indicates the failure of interfacial zone and the fiber is
pulled out without resistance. Here, the interfacial bond strength
depending on the pull-out behavior of the fibers from the matrix
was determined by the maximum pull-out load Fmax , as following:
Fmax
= (7)
2dLf
where Fmax is the maximum pull-out load, 2dLf is the contact
area between the fiber and cement matrix, d is the radius of interfa-
cial zone (equals to the radius of fiber) and Lf = 30 mm is the length
of interfacial zone or embedment length of the fiber. The pull-out
energy, which indicates energy absorption needs to be dissipated
to pull out a fiber from the matrix, which was calculated as the area
under a load-deflection curve up to specified deflection criterion.
Here the test was terminated up to a pullout distance of 2 mm. The
pull-out energy Epull was determined using Eq. (8):
 2
Epull = Fdx (8)
0
where F is pull-out load and x is the pullout distance of fiber from
the cement matrix.
2.6. Hydration heat
The heat flow of the hydration of cement and blended pastes
was determined using TAM Air iso-thermal calorimeter at 25 ◦ C. In
accordance to ASTM C305, pastes with a water to binder ratio of
0.5 were mixed by hand with a spatula for 1 min at a temperature
 J. Wei et al. / Corrosion Science 106 (2016) 1–15
of 23 ± 2 ◦ C. 5 g of paste was cast in glass cylinders using a plastic
dropper within 2 min after initial contact of cement and water. Then
the sealed samples were placed in the calorimeter, which records
the hydration heat flow of the sample up to 45 h.
2.7. Thermal analysis
Cement hydration products and sisal fiber components were
determined by means of thermogravimetric analysis (TGA) and
derivative thermogravimetry (DTG) using a TA Instruments Q50
Thermogravimetric Analyzer and a PerkinElmer Pyris1 Thermo
Gravimetric Analyzer, respectively. For the TGA investiagtion of
cement, the ground hardened cement powder was heated at a heat-
ing rate of 10 ◦ C/min from room temperature to 150 ◦ C and then
15 ◦ C/min to 800 ◦ C in N2 -atmosphere. The measured content of
Ca(OH)2 (CCH ) is expressed as percentage of the dry sample weight
at 500 ◦ C as follows [31,32]:
W400 − W500 MCa(OH)2 W400 − W500
CCH = × = × 4.1 (9)
W500 MH2 O W500
where WT is the dry sample weight at the temperature T (◦ C).
The bound water content (Cbw ) was determined as the differ-
ence between the mass measurements at 105 ◦ C and 800 ◦ C and
corrected for the ignition losses of the unhydrated cement powder
and the substitution cementitious materials [31–33]:
W105 − W800
Cbw = − (fc × ILuc + fSCM × ILSCM ) (10)
W800
where fc , fSCM are the mass fractions of cement (PC) and SCM in
the dry mix, respectively; ILuc and ILSCM are the ignition losses of
unhydrated cement and SCM, respectively.
The degree of hydration (DOH) () for the cement pastes was
calculated using the following formula:
Cbw(t)
= × 100% (11)
Cbw(∞)
where Cbw(t) is the bound water content of cement paste at time
t normalized to the fraction of cement in a paste; Cbw(∞) is the non-
evaporable water corresponding to full hydration of 1 g cement
(0.23 g [34]). For the TGA investigation of sisal fiber, the ground
sisal powder samples of 25 mg were placed in an alumina crucible
and tests were carried out in a N2 atmosphere with a heating rate of
10 ◦ C/min from room temperature to 600 ◦ C. The DTG curves, which
are the derivative of TGA curves with respect to the temperature,
were employed to read the exact boundaries for the temperature
intervals.
2.8. X-ray diffraction analysis
The crystallinity fraction of the embedded natural fibers after 30
wetting and drying cycles was analyzed by X-ray diffraction (XRD)
carried out on ground powders with an X2 Scintag X-Ray diffrac-
tometer in a –2 configuration using CuK␣ source ( = 1.54 Å) at
−40 kV and 35 mA. The samples were scanned in step mode with a
step size of 0.02◦ (2) in the angular range of 5–25◦ . Two indexes
were calculated from the X-ray diffraction spectra of sisal fibers
embedded in the cement matrices after aging: the percentage of
crystallinities using the peak height method [35] (%Cr) and crys-
tallinity index (Cr.I.). The Segal empirical %Cr (˛) was calculated
using the following equation [36]:
I002
˛= × 100% (12)
I002 + Iam
where I002 is the maximum intensity of diffraction of the (0 0 2)
lattice peak at 2 = 22.5◦ , Iam is the intensity of diffraction contribut-
5
ing to the amorphous fraction, which is taken at a 2 angle between
18◦ and 19◦ where the intensity is a minimum [37–39].
The Cr.I. (ˇ), which indicates the embedded fiber’s crystallinity
degree [35], was evaluated by using the following equation
[37,40–42]:
I002 − Iam
ˇ= (13)
I002
3. Results and discussion
3.1. Determination of aggressive condition
The mass changes of mortar beams (for flexural test) reinforced
with natural fiber in the first wetting and drying cycle as functions
of immersion and drying times are shown in Fig. 4. The compari-
son between Fig. 4a and b indicates that the weight change is more
sensitive to water immersion than drying environments and that
the amount of water absorbed at equilibrium is less than that loss
in drying process, which may be caused by the hysteresis between
water absorption and desorption and can be explained by the ink-
bottle effect in capillary pores [43]. Due to large particle size (as
shown in Fig. 1) and specific porous structure (Fig. 2a), the incor-
poration of 20 wt.% RHA leads to a dramatic increase of free water
in the capillary pores and, thus, an evident increase of weight loss
in drying stage. Similar effect can also be noticed in FA30, which
yields a severer mass change than that of neat PC mortar. Owing
to their fine particle size and high pozzolanic activity, the partial
replacements of cement by MK and SF make the cement matrix
denser than neat PC. As a result, less weight losses were observed
in SF10 and MK30. According to the longest required period of time
(25 h for drying (MK30) and 15 h for wetting (SF20)) to reach mass
equilibrium, the accelerated aging condition for all samples were
determined to test their durability all at same age. A duration of 42 h
was selected for each wetting and drying cycle: 15 h immersion in
tap water at 70 ◦ C and 25 h drying in an oven at 70 ◦ C, with 1 h air
drying under 23 ◦ C and 10% humidity between each half-cycle.
3.2. Kinetics of cement hydration
The development of normalized specific heat flow and released
heat of hydration of neat PC and the blends at early age (up to
40 h) are shown in Fig. 5. Each curve represents the average of two
measurements, the error of which is less than 0.2 mW/g. From the
curves, it can be seen that the heat flow of each blend is almost the
same in the first hydration stage: pre-induction period. Both the
neat PC and cement blends show a strong heat evolution generated
by the quick conversion of aluminate phases in the first minutes.
The main differences occur during the following three hydration
stages: induction, acceleration and deceleration periods. The cur-
tate induction period indicates that the incorporations of SF, FA
and MK effectively accelerated the hydration process of cement.
The induction period for all groups starts after 0.5 h of hydration.
The acceleration periods of the neat PC and MK30 begins after
2 h, however in the SF10 and FA30 this period starts 0.75 h ear-
lier, which means the formation rate of C–S–H was accelerated.
However, RHA20 presented a longer induction period than neat
PC: the acceleration periods begins after 4.5 h of hyradation and its
first heat flow peak does appear until 15 h. This may be due to its
large particle size and surper water absoprtion, which significantly
arrest the cement hydration at early age (first 6 h) and the release
of which will provide additional water for the unhydrated cement
and enhence cement hydration at a later stage.
There are two peaks (shoulders) on the PC curve, namely a
silicate and an aluminate peak, due to the silicate reaction, and tri-
calcium aluminate reactions to form ettringite and subsequently
6 J. Wei et al. / Corrosion Science 106 (2016) 1–15

Fig. 4. Mass change of natural fiber-reinforced mortar beams in the drying (a) and wetting (b) stage.

8 200

a 180 b
7 PC
MK30
160
Normalized heat flow (mW/g)

RHA20
6
SF10
Released heat (J/g)

140
FA30
5
120

4 100

80
3
PC
60 MK30
2 RHA20
40 SF10
1 FA30
20

0 0
0 5 10 15 20 25 30 35 40 45 0 5 10 15 20 25 30 35 40
Time (h) Time (h)

Fig. 5. Specific normalized heat flow (a) and released heat (b) with a w/b ratio of 0.5 at 25 ◦ C.

conversion into monosulfate, respectively. The two peaks, with
maximum heat flows of 2.3 mW/g and 2.0 mW/g, were observed
after 9.7 h and 16.5 h of hydration. There are at least three contribu-
tions of SCMs to the hydration heat of Portland cement: providing
extra space for the hydration products, enhanced nucleation, and
the reaction between SCMs and clinkers of cement [44]. The heat
flow peaks in Fig. 5a demonstrates that among the SCMs in this
study, MK shows the most remarkable impact on cement hydration
at early age. Compared with its contribution to silicate reaction, The
effect of MK on the aluminate reaction was much more significant:
with the addition of MK, the second peak occurred much earlier
and became much narrower and higher (the maximum second heat
flow: 7.22 mW/g for MK30 and 1.98 mW/g for PC). Because of the
low tricalcium aluminate (C3 A) amount in PC, the second shoulder
peak was lower than the silicate peak for neat PC, but the oppo-
site held for PC-MK blends. Corresponding to the second formation
of ettringite, besides the nucleation role, the aluminates originat-
ing from MK also possibly participate in the reaction at this stage
[24]. The fact that this “shoulder” can be found in RHA20 as well,
but both of its two peaks appeared with an evident delay as ana-
lyzed above. It can be noted that the second heat flow of RHA20
maximum reaches 2.9 mW/g, which is higher than that of FA30
and PC. This may correlate to the high armorphous silicate content
and the release of water from the specific porous structure of RHA,
which can enhance cement hydration significantly at a later stage.
Owing to high pozzolanic activity, the incorporation of 10 wt.% SF
enhanced the silicate reaction most considerably, with a maximum
heat flow of 3.3 mW/g after 7.5 h of hydration. The incorporation of
30 wt.% FA leads to an earlier acceleration period and a maximum
heat flow of 2.5 mW/g. In the subsequent deceleration period, FA30,
FA10 and MK30 present a similar heat flow development, which
decreases significantly faster than the rest groups and transfers to
diffusional period after 25 h.
Fig. 5b shows the normalizad cumulative released heat up to
40 h of hydration. The improvement in both silicate and aluminate
reactions cause a high hydration heat release at early hydration
time. It can be seen that SF10 releases more hydration heat than
neat PC before 22.5 h, and then PC releases the maximum heat
among the groups. Due to the dramatic second heat flow peak as
shown in Fig. 5a, the release heat of MK30 presents a rapid growth
after 9.5 h of hydration and then gently increases and become lower
than that of neat PC after 19 h. Although the acceleration period
appeared later, RHA20 presents a similar normalizad released heat
as MK30 after 40 h of hydration. Among all the blends, due to the
 J. Wei et al. / Corrosion Science 106 (2016) 1–15
lower pozzolanic activity and quick end of both the acceleration
and deceleration period, FA30 releases a lowest hydration heat after
1 day.
3.3. Effect of SCMs on cement hydration products
It is well known that, by incorporating SCMs, cement hydra-
tion can be effectively enhanced owing to pozzolanic reactions
[24,31,45,46]. When materials rich in amorphous silica (pozzolans)
are added to the Portland cement, they react with the Ca(OH)2 , lib-
erated during the formation of C–S–H by the hydration of C3 S and
ˇC2 S, thereby generating an extra production of C–S–H with higher
Si/Ca ratio [47]. As a result, the alkalinity of cement paste is reduced
and a high early-age strength can be obtained. TGA investigation of
cement blends containing the SCMs was performed after 28 days of
hydration to determine their influence on cement hydration prod-
ucts. The Ca(OH)2 and bound water contents of neat PC and the
blends are shown in Fig. 6.
From Fig. 6a, it can be noted that the amount of Ca(OH)2 is
very sensitive to SCM’s incorporation in cement. After 28 days
of hydration, neat cement yielded a Ca(OH)2 content of 23.19%.
By partially replacing cement with SCMs, the Ca(OH)2 content of
cement paste experienced a considerable reduction. The smallest
Ca(OH)2 content (5.63%) and Ca(OH)2 content normalized to the
amount of Portland cement (7.67) were achieved by MK30 and
SF10, respectively. Due to the high content of Al2 O3 , MK has a
higher pozzolanic activity than SF at early stages, which leads to
a fast Ca(OH)2 consumption. It is also important to note that SF has
a higher portlandite consumption ability than MK at later hydra-
tion processes, because of its higher SiO2 content. By considering
the cement replacement level, silica fume is more effective in con-
suming Ca(OH)2 than other SCMs. Owing to the high content of
amorphous silicate, RHA also showed an effective role in reduc-
ing Ca(OH)2 . By replacing 20 wt.% cement with RHA, the blend
yields a Ca(OH)2 content of 10.34% and a normalized Ca(OH)2
content of 12.93, which are 55.41% and 44.24% lower than those
of neat cement, respectively. With a 30% corporation of FA, the
Ca(OH)2 content decreased by 37.04%, however, by considering the
high cement substitution level, the normalized Ca(OH)2 content of
cement only decreased by 10.05%.
The amount of ‘bound’ water present in the hardened Portland
cement paste depends on the phase composition of the original
cement, on the degree of hydration and also on the way in which
the term ‘bound water’ has been defined [48]. Bound water includes
contributions from all the hydration products, such as Ca(OH)2 ,
C–S–H, ettringite, and monocarbonate, which are all water rich
products of cement hydration. Fig. 6b shows the influences of
SCM incorporations on bound water content and degree of cement
hydration after 28 days. The bound water contents of MK30, RHA20
and FA30 are 8.81%, 6.56%, and 23.45% lower than that of neat PC.
This may be attributed to two causes: (i) the stabilization of water
rich hydration products, i.e. ettringite and monocarbonate [31],
and (ii) the lower cement fractions in these blends due to cement
replacement. It can be seen that the contents of bound water in
cement blends are lower than that in the neat PC, but are greater if
they are referred to the Portland cement fraction. This is believed
to relate to the fact that pozzolanic reaction is a water-intake pro-
cess other than from Ca(OH)2 [49], so more water can be fixed. In
addition, in cement blends, the hydration of PC can be enhanced
substantially and an increased amount of water rich hydration
products with better stabilization is thus obtained. The comparison
between neat PC and SF10 demonstrates the slightly higher content
of bound water in the SF-blend after 28 days of hydration, and thus
a higher normalized bound water content. Among all the pastes,
MK30 and neat PC show the highest and the lowest normalized
bound water contents, respectively.
7
According to the bound water content normalized to dry cement
fraction, the degree of cement hydration of each blend was calcu-
lated to evaluate the efficiency of these SCMs in enhancing cement
hydration. By replacing 30 wt.% cement, MK30 shows the highest
cement DOH among all pastes, which is 28.70% higher than that of
neat PC. This may be attributed to the high initial pozzolanic reac-
tivity of MK owing to its Al2 O3 phases, which are involved in the
formation of gehlenite (C2 ASH8 ) and a small amount of crystalline
C4 AH13 phase [50]. Therefore the amount of calcium suffered a sig-
nificant reduction and thus the induction period of cement paste
was shortened. Moreover, the incorporation of MK provides more
siliceous surface for early precipitation of hydration products of
cement (mainly C–S–H), which also results in a short induction
period [51]. With same mechanism, silica fume, which has a high
silicate content and large surface area, also shows a high efficiency
factor in enhancing cement hydration. It can be seen that, after 28
days, the degree of cement hydration in SF10 is 21.02% higher than
that of neat PC and is slightly lower than that of MK30. By replacing
20 wt.% and 30 wt.% cement by RHA and FA, the degree of cement
hydration was improved by 16.69% and 9.22%, respectively. It can
be noted that the efficiencies of these two ashes, RHA and FA, is
lower than those of MK and SF in cement hydration modification.
Owing to higher silicate amount and the specific porous structure,
RHA enhanced cement hydration more significantly than FA.
3.4. Interfacial bonding properties
The flexural properties of fiber-reinforced composites depend
on several factors, such as fiber strength, fiber modulus, properties
of the matrix, and interfacial bonding properties. Under tension,
natural fiber reinforced cement composites fail by a combination
of fiber fracture and pull-out [52,53], and the fiber pull-out is the
principal mechanism of failure [54]. In this study, effects of SCM
incorporation on interfacial bonding properties between natural
fiber and cement matrices were investigated by means of single
fiber pull-out testing. Fig. 7 shows the bond strength and pull-out
energy after 1, 7 and 28 days. It can be noticed that both bond
strength and energy are very sensitive to curing age, and natu-
ral fiber-reinforced neat PC yields the worst values. Among all the
composites, MK30 shows the best interfacial properties, especially
at early age. This may be attributed to the high content of Al2 O3 in
MK, which could consume Ca(OH)2 more effectively and the cement
matrix become dense at early age. By replacing 30 wt.% cement with
metakaolin, the 1-day, 7-day and 28-day bond strength is improved
by 156.25%, 102.49%, and 43.40%, respectively. Similar trends can
also be found in pull-out energy. Although there is only 10 wt.%
cement replacement by SF, SF10 yields a comparable bond strength
and pull-out energy as MK30 after 7 days. The 28-day bond strength
and pull-out energy increased by 26.42% and 50.79%, respectively,
in the presence of 10 wt.% SF. By incorporating 30 wt.% fly ash, the
28-day bond strength was only improved by 11.32%. Although the
modifying effect of RHA is not comparable with metakaolin and
silica fume, it leads to an enhancement in improving the bond-
ing behavior of natural fiber in a cement matrix, which is more
considerable than that of fly ash.
It can be noticed that cement substitution results in an increase
of interfacial bonding properties between natural fiber and cement
matrices. By reducing Ca(OH)2 , ettringite and forming more C–S–H,
mainly porosity and thickness of the transition zone [55,56] adja-
cent to the fibers were reduced and more cement products adhered
on the fiber’s surface, thereby the bonding strength was improved
and the fibers’ slipping needed more energy. It is anticipated that
the initial flexural properties of natural fiber-reinforced cement
mortar, before exposed to accelerated aging, will be improved sig-
nificantly owing to the improved interfacial bonding properties.
In addition, by incorporating these SCMs, the durability of sisal
8 J. Wei et al. / Corrosion Science 106 (2016) 1–15

25 100 100
Bound water
a b Degree of hydration
Ca(OH)2 content 90 90

20 Normalized Ca(OH)2 content

Content of bound water (%)

80 80
Ca(OH)2 content (%)

Degree of hydration (%)

70 70

15 60 60

50 50

10 40 40

30 30

5 20 20

10 10

0 0 0
PC MK30 RHA20 SF10 FA30 PC MK30 RHA20 SF10 FA30

Fig. 6. The effects of SCMs on cement hydration: (a) original and normalized Ca(OH)2 contents, and (b) degree of hydration determined by means of normalized bound water
contents.

0.9 22
a b
0.8 20
1d
18 1d
0.7 7d
Pull-out energy (N·mm)

28d 7d
Bond strength (MPa)

16
0.6 28d
14
0.5 12

0.4 10

8
0.3
6
0.2
4
0.1 2

0.0 0
PC MK30 RHA20 SF10 FA30 PC MK30 RHA20 SF10 FA30

Fig. 7. Interfacial bonding properties between natural fiber and cementitious matrices: (a) bond strength (MPa) and (b) pull-out energy (N mm).

fiber-cement composites will also be improved attributing to the A significant increase in durability for natural fiber-reinforced
reduction of Ca(OH)2 content and alkalinity in the cement matrices. cement composites was achieved by incorporating SCMs. As wet-
ting and drying cycles proceeded, natural fiber-reinforced PC
yielded the most dramatic decrease of flexural properties: after
3.5. Durability of natural fiber-reinforced mortar
50 wetting and drying cycles, strength and toughness decreased
by 85.58% and 99.18%, respectively. This means, among the five
Owing to crack bridging and stress transfer between fiber and
groups, natural fiber-reinforced neat PC has the worst durability,
the cement matrix, mortar beams reinforced with long aligned nat-
which may be caused by the severe degradation of natural fibers in
ural fibers yielded remarkable post-cracking flexural behavior. Sole
PC. By replacing 30 wt.% cement with MK, the flexural properties
big-scale crack was replaced by multiple fine cracks, consequently
of sisal fiber-reinforced mortar did not experience evident decline
more energy was consumed and the brittle failure of cement mortar
after the first 5 and 15 cycles. The high Al2 O3 content in MK results
was prevented. This makes it more reasonable here to determine
in an enhanced cement hydration and Ca(OH)2 consumption at
the durability of natural fiber-reinforced cement composites by
early age, consequently the degradation of natural fiber in MK30
means of flexural properties after exposed to accelerated aging. The
can be mitigated more effectively than other groups. After 50 wet-
durability of natural fiber-reinforced mortar, after each aging stage,
ting and drying cycles, the flexural strength and toughness of MK30
determined by post-cracking flexural strength and toughness, are
decreased by 33.48% and 37.44%, respectively. For SF, after 5 cycles,
presented in Fig. 8a and b, respectively. From Fig. 8, it can be noticed
the decrease of flexural behavior is slightly more dramatic than that
that, after 28 days of curing, MK30, RHA20 and SF10 show sim-
of MK30. This indicates that silicate is not as effective as aluminate
ilar initial post-cracking strength, which are 35.73%, 32.97% and
in enhancing cement hydration at early age, and the natural fibers
34.73% higher than that of natural fiber-reinforced neat PC mortar,
suffered certain degree of deterioration before Ca(OH)2 was con-
respectively. The incorporation of 30 wt.% FA leads to an increase of
sumed. The final strength and toughness of SF10 were 52.57% and
flexural strength by 14.56%, which is not as considerable as those of
57.66% lower than its initial values. Although RHA20 can lead to a
MK, RHA and SF. Similar trends can be found in flexural toughness,
significant increase of initial properties, its strength and toughness
in which MK30 and SF10 yielded the best values among the groups.
 J. Wei et al. / Corrosion Science 106 (2016) 1–15 9

14
0 Cycle 0 Cycle
Post-cracking flexural strength (MPa) a 5 Cyles
7 b 5 Cyles
12 15 cycles 15 Cycles

Post-cracking toughness (N·m)

30 Cycles 6 30 Cycles
50 Cycles 50 Cycles
10
5

8
4

6
3

4
2

2 1

0 0
PC MK30 RHA20 SF10 FA30 PC MK30 RHA20 SF10 FA30

Fig. 8. Durability of natural fiber-reinforced cement mortar by means of four-point flexural properties, (a) post-cracking flexural strength, and (b) post-cracking flexural
toughness.

Fig. 9. Tensile properties of single natural fibers: (a) Weibull plots and (b) tensile strength.

experienced an evident decrease after exposed to wetting and dry-
ing cycles. This may be caused by the remarkable water penetration
in RHA20 samples attributed to RHA’s porous structure. More pore
solution, which plays an important role in alkaline hydrolysis and
cell walls mineralization, was produced, especially in the transition
zone between natural fiber and the cement matrix. Similar phe-
nomenon was observed in FA30: after 50 wetting and drying cycles,
the post-cracking strength and toughness decreased by 84.03% and
95.80%, respectively. Although RHA and FA modified samples show
obvious property decline subjected to accelerated aging, they still
lead to an increase in durability of cement composites reinforced
with natural fiber. The final flexural strength of RHA20 and FA30
is 165.04% and 26.83% higher, respectively, than that of natural
fiber-reinforced neat PC.
3.6. Degradation of the embedded natural fibers
3.6.1. Tensile behavior
The uniaxial tensile behavior of raw and embedded sisal fibers in
the five different cement matrices, after 5 and 30 wetting and dry-
ing cycles, were investigated. The Weibull plots of testing results
are shown in Fig. 9a, from which the scatter natural of tensile test-
ing data for natural fibers and the Weibull modulus can be clearly
observed from the fitting lines and their slopes. The raw natural
fiber and the embedded fiber in PC yield similar Weibull modu-
lus, however, they show the highest and lowest tensile strength,
respectively, among the groups. This indicates that the data of a
single fiber tensile test are concentrated when the tensile strength
is high or low, and apparently scattered for strengths in between.
The raw fibers, without any degradation, show a tensile strength
of 605.5 MPa. The tensile strength of the natural fiber embedded in
the neat PC matrix decreased by 35.07% and 86.19% after 5 and 30
wetting and drying cycles, respectively. The dramatic reduction of
strength demonstrates that natural fiber experienced a sever degra-
dation in the alkaline environment of pure cement matrix. A result
of interest can be noticed that, after 5 wetting and dying cycles,
the fiber embedded in MK30 yields a higher tensile strength than
the raw natural fiber. This may be attributed to the slight removal
of amorphous components (i.e. lignin and hemicellulose) and the
rearrangement of cellulose microfibrils in the mild alkaline matrix
of MK30, which could improve the mechanical behavior of natural
fibers [57,58]. From, Fig. 9b, it can be seen that the replacement of
30 wt.% cement by MK is the most effective way to mitigate natural
fiber degradation: after 30 wetting and drying cycles, the tensile
10
20
19
Young's Modulus(GPa)
18
17
16
15
Raw PC MK30 RHA20
Fig. 10. Young’s modulus of single natural fibers in uniaxial tensile test.
strength of natural fiber in MK30 decreased by 19.48%. Due to
higher Ca(OH)2 content and lower degree of cement hydration, the
final strength of fiber in SF10 is 21.58% lower than that in MK30. The
fibers embedded in RHA20 and FA30 show similar tensile behavior,
which is lower than those in MK30 and SF10 but much better than
that in neat PC, especially after 30 cycles.
As shown in Fig. 10, the raw natural fiber shows a lower Young’s
modulus for the tensile test, which indicates that cement matri-
ces lead to an increase of fiber embrittlement. Although, the fiber
embedded in MK30 shows a higher tensile strength than the raw
fiber after 5 cycles, its modulus increased due to the reduction
of non-cellulose components. In the author’s previous work [59],
it has been analysed and proven that both the ingress of cement
hydration (especially Ca(OH)2 ) and alkaline hydrolysis of the amor-
phous components can lead to the embrittlement of natural fiber.
The mineralization of cell wall caused by these two mechanisms is
another important degradation way, in addition to alkaline degra-
dation. The fiber embedded in neat PC matricies show the highest
modulus, which is 12.08% and 28.51% higher than that of raw fiber
after 5 and 30 cycles, respectively. Among the four modified cement
matrices, MK30 shows the best effect in mitigating fiber embrittle-
ment. After 30 wetting and drying cycles, the modulus of fiber in
MK30 is 6.49% higher than that of raw fiber but is 17.13% lower
than that of the fiber embedded in neat PC. Although worse than
MK30, SF10 and RHA20, FA30 also show effective impacts in inhibit-
ing fiber mineralization. It can be noticed that, except for MK30,
in other cement matrices, the Young’s modulus of natural fiber
increased with decreasing tensile strength. This in turn demon-
strates that the cement matrix modification using SCMs, such as
MK, SF, RHA and FA, is beneficial to mitigate natural fiber degrada-
tion, in both strength and toughness.
3.6.2. TGA analysis
In order to figure out the reason for strength depression in the
embedded natural fibers, the component variation of fibers was
investigated. The cell wall of natural fibers primarily consist of sev-
eral bio-polymers, such as pectin, cellulose, hemicellulose, lignin,
pigments and other extractives. Three of them, cellulose, hemicel-
lulose and lignin, are the main components. Furthermore, the ratio
of which can affect natural fiber’s mechanical properties consid-
erably. Cellulose and hemicelluloses are the major components of
the secondary layers of the cell wall in lignocellulosic fibers [60].
The hemicelluloses are present in the surface layer as adhesives
to strengthen the bonds between individual fibers in a three-
J. Wei et al. / Corrosion Science 106 (2016) 1–15
dimensional network of lignocellulosic [61,62]. It also appears to
form a link between the cellulose microfibrils and the lignin [63].
5 Cycles
30 Cycles
Cellulose is the main structural component of natural fiber. In gen-
eral, the higher content of cellulose, the stronger the natural fiber
that can be obtained [64]. Lignin is a three-dimensional polyphe-
nolic polymer, built up of phenylpropane units, most of which
are located in the middle lamella. This amorphous hydrocarbon
polymer provides compressive strength to the slender microfib-
rils and prevents them from buckling under compressive loads
[63]. Therefore, their relationship is similar as that among ossat-
ure, connective tissue and muscle of the human body. Lignin has
an extensive range of the degradation temperature in two or three
stages [65,66], which make it difficult to determine the exact con-
tent of this component using TGA. In this study, the contents of
cellulose and hemicellulose were roughly evaluated.
From the TGA and derivative thermogravimetry (DTG) carves of
raw natural fiber shown in Fig. 11a, it can be noticed that there are
three main stages of weight loss: (i) slight weight loss from 50 to
SF10 FA30 110 ◦ C due to evaporation of moisture in the fibers; (ii) weight loss
for decomposition of hemicelluloses and the main part of lignin
under temperature between 270 and 365 ◦ C; (iii) the maximum
weight loss under higher temperature (375–500 ◦ C) accompanying
the degradation of cellulose. Here, the weight loss in two temper-
ature ranges, 300–360 ◦ C (range #2 in Fig. 11a) and 370–500 ◦ C
(range #3 in Fig. 11a), were calculated to determine the amount
of hemicellulose and cellulose, respectively. The raw natural fiber
shows a hemicellulose content of 13.58% and a cellulose content of
58.88%. Due to alkaline hydrolysis of hemicellulose and the amor-
phous region of cellulose chains, the second stage of DTG carves for
the emdedded fibers merged with the thermal decomposition peak
of cellulose. The contents of hemicellulose and celullose of the fibers
embedded in PC decreased by 10.82% and 46.60%, respectively, after
30 wetting and drying cycles.
From Fig. 11b, it can be seen that, after accelerated aging, the
natural fibers embedded in MK30 yield the highest cellulose con-
tent (45.59%), which is 16.45%, 7.63% and 42.87% higher than those
of fibers embedded in RHA20, SF10 and FA30. From the comparison
between the raw fiber and fiber embedded in neat PC (Fig. 11a), as
well as the comparison among the embedded fibers in Fig. 11b,
it can be noticed that, with severe degradation, the DTG peaks
shifted to lower temperatures, which indicates that the thermal
stability of natural fiber was reduced. These results agree well with
the tensile properties of natural fiber in Section 3.6.1: the lower
the amount of cellulose, the lower the tensile strength of natural
fiber was detected. In addition to the mineralization of cell walls,
the degradation of hemicellulose, lignin, as well as the stripping
of cellulose microfibrils are also the cause of increase of the fiber’s
Young’s modulus in tensile testing.
3.6.3. XRD analysis
The XRD analysis of raw and the embedded fibers after 30
wetting and drying cycles was carried out to determine their degra-
dation degree. Fig. 12a shows the XRD patterns of natural fibers,
from which it can be seen that cellulose crystallographic plane
(1̄ 1 0), (1 1 0) and (0 0 2) and the amorphous scatter between 18◦
and 19◦ were detected from the raw natural fiber. Due to the severe
deterioration of amorphous components, the intensity of cellulose
become greater and the peaks get sharper. The peaks centered
around 29.2◦ (2), which is the dominant degree for calcite, calcium
silicate hydrate (C–S–H) and calcium–alumino–silicate–hydrate
(C–A–S–H) [67], were detected in the X-ray patterns of fiber embed-
ded in MK30 after 30 cycles. This may be caused by the residual
cement hydrates on sisal fiber’s surface, since MK reacts with the
calcium hydroxide producing more C–S–H and C–A–S–H [68]. This
in turn indicates that the interfacial bonding strength between nat-
 J. Wei et al. / Corrosion Science 106 (2016) 1–15 11

Fig. 11. TGA and DTG curves of the raw and embedded natural fibers in cement matrices after 30 wetting and drying cycles.

Fig. 12. XRD analysis of the raw and embedded natural fibers after 30 wetting and drying cycles: (a) XRD pattern, (b) %Cr. and Cr.I.
−
1, 2, 4: cellulose crystallographic plane (1 1 0), (1 1 0) and (0 0 2); 3: Ca(OH)2 ; 5: ettringite; 6: Calcite; 7: C–S–H, C–A–S–H.

ural fiber and the cement matrix is better in MK30 than other
matrices, which agrees well with the results shown Fig. 7.
Fig. 12b shows the Segal empirical percentage of crystallinity
(%Cr.) and crystallinity index (Cr.I.), which were calculated to quan-
tify the fiber’s crystalline properties. Raw natural fiber show the
lowest %Cr. and Cr.I. among all groups, and these two indices of fiber
increased by 12.98% and 20.31%, respectively, after 30 cycles in PC.
Although the embedded fiber’s %Cr. and Cr.I. are 8.94% and 13.70%
higher than those of raw natural fiber, MK30 shows the best effect
in mitigating natural fiber crystallization. Compared with natural
fibers embedded in neat PC, the fiber’s %Cr. and Cr.I. decreased by
3.02% and 3.77%, respectively, by replacing 10 wt.% cement with SF.
Due to lower Ca(OH)2 consumption and degree of cement hydra-
tion, the fiber embedded in RHA20 yields higher indices than those
embedded in MK30 and SF10. By replacing the same amount of
cement, the fiber embedded in FA30 shows %Cr. of 86.70% and Cr.I.
of 0.85, which are 2.46% and 4.49% higher than those of fiber in
MK30, but are 1.20% and 1.24% lower than those of fiber embedded
in neat PC, respectively.
3.6.4. Microstructure analysis
The microstructure of natural fibers (Fig. 13) were analyzed on
the fracture surfaces of fiber reinforced cement mortar samples
after 30 wetting and drying cycles to visually evaluate the influ-
ence of SCMs on fiber degradation. Due to the alkaline hydrolysis
of lignin and hemicellulose, the natural fiber’s surface was fully
12 J. Wei et al. / Corrosion Science 106 (2016) 1–15
Fig. 13. SEM micrographs of the natural fibers on fracture surface of fiber-reinforced cement mortar after 30 wetting and drying cycles: (a) and (b) PC, (c) MK30, (d) RHA20,
(e) SF10 and (f) FA30.
decomposed and all the cellulose micro-fibrils are stripped from
each other. The space of lumen was saturated with the cement
hydration products, which provides evidence for cell wall min-
eralization. As a result, fibers become more brittle and lose their
reinforcement capacity in cement based materials. Attributed to
the reduction of the amount of Ca(OH)2 and alkalinity depression
in the MK modified cement matrix, the fibers still retain their initial
morphology with a good integrity, except for several cracks on their
cross section. In addition, due to the retention of lignin and hemi-
cellulose, which act as protective layers for cellulose micro-fibrils,
no cellulose chain stripping and cement products precipitation was
detected in natural fibers. The fiber embedded in RHA20 experi-
enced a slight severer degradation of non-cellulose components
than that in MK30. It can be seen that the surface of the fiber has
been partial corroded and the cement hydration products have pre-
cipitated on the fiber periphery. The open lumens also indicate the
deterioration of armorphous components and provide additional
space for cement hydration products migration. A better structure
of natural fiber was detected in SF10. Although a small part of fiber
suffered cellulose miro-fibrils stripping, the remaining portion of
the fiber is still in good integrity. In addition to the fiber, the strip-
ping space between fiber and the cement matrix in MK30 is much
smaller than those in other groups, this may be attributed to lower
autogenous shrinkage of the MK modified cement and less dry-
ing shrinkage of natural fiber in the low alkaline environment. In
FA30, after 30 wetting and drying cycles, the degradation degree
of natural fiber is worse than that in RHA20. The surface of fiber
has been completely broken and more than half of the cellulose
fibrils were stripped. Compared with the degraded fiber in neat
PC, the stripped fibrils are still clinging together and less hydration
products are found in between. SEM micrographs of the embed-
ded natural fiber clearly demonstrate the degradation degree of
the fiber and agree well with the results in TGA and XRD analysis,
as well as the durability of fiber-reinforced cement composites.
3.7. Correlation between durability and cement hydration
From the discussion above, hydration of cement and durabil-
ity of natural fiber-reinforced cement composites, as well as the
degradation of embedded natural fibers have been investigated
separately. The incorporations of SCMs not only enhanced the
hydration of cement but also play important roles in mitigating
natural fiber’s deterioration owing to their pozzolanic activity in
the cement matrix. It is believed, the reduction of Ca(OH)2 and the
 J. Wei et al. / Corrosion Science 106 (2016) 1–15 13

Fig. 14. Correlations between degree of hydration, Ca(OH)2 content and durability of natural fiber-reinforced cement composites: (a) post-cracking flexural strength and (b)
flexural toughness.

Fig. 15. Correlations between cement hydration and degradation of embedded natural fibers: (a) effect of degree of hydration on fiber tensile behavior (b) effect of Ca(OH)2
content on fiber’s crystalline indices.

improvement of DOH are the root causes for the increased dura-
bility. Here, the correlations between durability of composites and
hydration of cement, and the correlation between degradation of
natural fiber and cement hydration products, are analyzed.
Fig. 14 a and b shows the effects of cement hydration on
post-cracking flexural strength and the toughness of natural fiber-
reinforced cement mortar after 30 and 50 cycles, respectively. It can
be noticed that the flexural strength of the composites increases
with increasing DOH of the cement matrix, which means the nat-
ural fiber-reinforced cement composites with full hydration can
obtain improved durability. The correlation between Ca(OH)2 con-
tent and flexural strength may be used to explain this phenomenon.
The incorporations of SCMs not only enhanced the hydration of
cement, but also reduced the Ca(OH)2 content in the cement matrix.
It is well known that Ca(OH)2 is the main reason for the fiber degra-
dation [59], including mineralization and alkaline hydrolysis. From
Fig. 14a and b, it can be found that, in this study, the higher the
DOH in the cement matrix the lower Ca(OH)2 content that can be
detected, thus the better durability obtained. From Fig. 14b, it can
be seen that the toughness of natural fiber-reinforced cement mor-
tar is more sensitive to DOH and Ca(OH)2 content than flexural
strength, especially at higher DOH. When DOH is less than 80%,
the effect of DOH on post-cracking flexural toughness (at both 30
and 50 cycles) is neglectable. However, when the DOH of cement
becomes higher than 80%, the flexural toughness becomes very
sensitive to DOH. Again, this can be explained by the variation of
Ca(OH)2 content: with increasing DOH, the reduction of Ca(OH)2
content becomes dramatic and its effect on the degradation of nat-
ural fiber decreases.
The effects of DOH and Ca(OH)2 content on the embedded fiber’s
tensile behavior and crystalline properties after 30 wetting and dry-
ing cycles are shown in Fig. 15a and b, respectively. The durability
of natural fiber-reinforced cement composites is affected by sev-
eral factors, such as durability of the cement matrix, interfacial
bonding properties and fiber degradation. The incorporations of
SCMs not only improved the initial flexural behavior of compos-
ites but also lead to improved interfacial properties between fiber
and cement matrices. It is evident that the decreased durability of
modified cement composites is caused by the degradation of nat-
ural fiber. From Fig. 15a, the correlation between DOH and tensile
14 J. Wei et al. / Corrosion Science 106 (2016) 1–15
properties of the embedded natural fibers can be determined: with
DOH increasing, the depression of tensile strength decreased and
the fibers’ toughness increased. The curves in Fig. 15b indicate that
both %Cr. and Cr.I. decrease with decreasing Ca(OH)2 content in
the cement matrix. The deterioration of non-cellulose components
and the amorphous regions of cellulose chains may be the reasons
for the increased %Cr. and Cr.I., and they are also important causes
for the increased embrittlement of the embedded natural fiber, in
addition to fiber cell wall mineralization.
4. Conclusions
At an early age (first 1.5 d), the incorporation of SCMs lead to a
promoted cement hydration reaction and lower total cumulative
released heat. Owing to the pozzolanic avidity and seeding effect
of SCMs, silicate reaction and tricalcium aluminate reactions were
enhanced.
The SCMs in this study clearly reacts with hydration products,
especially Ca(OH)2 , in cement to form C–S–H gel. A consequence of
this is that the content of Ca(OH)2 in the cement matrix, which is the
main reason causing natural fiber degredation and mineralization,
was effectively reduced. Owing to the high Al2 O3 and SiO2 content,
and their large particle surface area, MK and SF show higher poz-
zolanic activities and better modification effects than RHA and FA in
improving degree of cement hydration and Ca(OH)2 consumption.
Cement substitution resulted in increases of interfacial bonding
properties between natural fiber and cement matrices. The porosity
and thickness of the transition zone were reduced and more cement
hydration products adhered on the natural fiber’s surface. Both
interfacial bonding strength and pull-out energy were improved.
As a result, the initial flexural properties of natural fiber-reinforced
cement composites were improved significantly.
A significant increase in durability for natural fiber-reinforced
cement composites was achieved by incorporating SCMs. The worst
and best durability, determined by means of post-cracking flexural
strength and toughness, were yielded by neat PC and MK30, which
showed the lowest and highest degree of cement hydration, respec-
tively. SF, RHA and FA also showed acceptable effects in mitigating
deterioration of the composites.
In good agreement with the durability of composites, the degra-
dation of the embedded natural fiber was effectively mitigated by
modifying the cement matrix. After 5 and 30 wetting and drying
cycles, the tensile strength of natural fibers embedded in the mod-
ified cement matrices were much higher than that of the fiber in
neat PC. The Young’s modulus was also decreased. The results of
TGA, XRD and microstructure analysis indicated that the alkaline
hydrolysis of amorphous components of natural fiber, the mineral-
ization of fiber cell walls and the stripping of cellulose micro-fibrils
were mitigated in the modified cement matrices, especially MK30
and SF10.
The degradation of natural fiber shows a strong dependence on
degree of cement hydration, especially when the DOH becomes
higher than 80%. Therefore, it can be concluded that the durabil-
ity of natural fiber-reinforced cement composites is proportional to
the cement hydration. The reduction of Ca(OH)2 content plays an
important role in mitigating natural fiber degradation and improv-
ing the durability of composites.
In summary, the durability of natural fiber-reinforced cement
composites show a strong correlation with the hydration behav-
ior of the cement matrix and the modification of cement hydration
using SCMs, which is an effective way to mitigate fiber degrada-
tion. At present, the degree of cement hydration is simply modified
by partially replacing cement using SCMs; indeed, more investiga-
tions in DOH variation by using chemical admixtures or through
other physical or chemical approach should be considered to
systematically study the effect of cement hydration on natural
fiber durability. In this regard, subsequently, the effects of chem-
ical admixtures, thermal treatment and pre-carbonation on the
hydration of cement matrix and the degradation behavior of the
embedded natural fibers will be investigated.
Acknowledgements
The authors would like to express their sincere thanks to
the Erbilgin Family, for their consistent financial support for the
author’s doctorate research through the Erbilgin Family fellowship
at Columbia University. The authors also wish to acknowledge Mr.
Hagh Mckee from Bast Fibers LLC, Creskill, New Jersey, for his supply
of sisal fiber.
References
[1] P.N. Balaguru, S.P. Shah, Fiber-reinforced Cement Composites, McGraw-Hill,
New York, 1992.
[2] M.C. Popescu, C.M. Popescu, G. Lisa, Y. Sakata, Evaluation of morphological
and chemical aspects of different wood species by spectroscopy and thermal
methods, J. Mol. Struct. 988 (2011) 65–72.
[3] K. Slopiecka, P. Bartocci, F. Fantozzi, Thermogravimetric analysis and kinetic
study of poplar wood pyrolysis, Appl. Energ. 97 (2012) 491–497.
[4] M. Poletto, H.L.O. Júnior, A.J. Zattera, Thermal decomposition of natural fibers:
kinetics and degradation mechanisms, in: Reactions and Mechanisms in
Thermal Analysis of Advanced Materials, John Wiley & Sons, Inc., 2015, pp.
515–545.
[5] J. Morán, V. Alvarez, V. Cyras, A. Vázquez, Extraction of cellulose and
preparation of nanocellulose from sisal fibers, Cellulose 15 (2008) 149–159.
[6] M.G.A. Pickering, A review of recent developments in natural fibre composites
and their mechanical performance, Compos. Part A: Appl. Sci. Manufact.
(2015), in press.
[7] M.Z. Rong, M.Q. Zhang, Y. Liu, G.C. Yang, H.M. Zeng, The effect of fiber
treatment on the mechanical properties of unidirectional sisal-reinforced
epoxy composites, Compos. Sci. Technol. 61 (2001) 1437–1447.
[8] K. Bilba, M.A. Arsene, Silane treatment of bagasse fiber for reinforcement of
cementitious composites, Compos. Part A: Appl. Sci. Manufact. 39 (2008)
1488–1495.
[9] J. Claramunt, M. Ardanuy, J.A. García-Hortal, R.D.T. Filho, The hornification of
vegetable fibers to improve the durability of cement mortar composites, Cem.
Concr. Compos. 33 (2011) 586–595.
[10] A.M. Cooke, Durability of autoclaved cellulose fiber cement composites, 7th
Inorganic-Bonded Wood and Fiber Conference (2000).
[11] J.L. Pehanich, P.R. Blankenhorn, M.R. Silsbee, Wood fiber surface treatment
level effects on selected mechanical properties of wood fiber–cement
composites, Cem. Concr. Res. 34 (2004) 59–65.
[12] D.M. Bastidas, A. Fernández-Jiménez, A. Palomo, J.A. González, A study on the
passive state stability of steel embedded in activated fly ash mortars, Corros.
Sci. 50 (2008) 1058–1065.
[13] E. Zornoza, J. Payá, P. Garcés, Chloride-induced corrosion of steel embedded in
mortars containing fly ash and spent cracking catalyst, Corros. Sci. 50 (2008)
1567–1575.
[14] M. Holloway, J.M. Sykes, Studies of the corrosion of mild steel in
alkali-activated slag cement mortars with sodium chloride admixtures by a
galvanostatic pulse method, Corros. Sci. 47 (2005) 3097–3110.
[15] M. Manera, Ø. Vennesland, L. Bertolini, Chloride threshold for rebar corrosion
in concrete with addition of silica fume, Corros. Sci. 50 (2008) 554–560.
[16] R.D. Toledo Filho, F.d.A. Silva, E.M.R. Fairbairn, J.A.M. d. Filho, Durability of
compression molded sisal fiber reinforced mortar laminates, Constr. Build.
Mater. 23 (2009) 2409–2420.
[17] F. Silva, J.A. Melo Filho, R.D. Toledo Filho, E.M.R. Fairbairn, Mechanical
behavior and durability of compression moulded sisal fiber cement mortar
laminates (SFCML), in: 1st International RILEM conference on textile
reinforced concrete (ICTRC), Rilem Publications S.A.R.L, 2006, pp. 171–180.
[18] A.R. Martin, M.A. Martins, O.R.R.F. da Silva, L.H.C. Mattoso, Studies on the
thermal properties of sisal fiber and its constituents, Thermochim. Acta 506
(2010) 14–19.
[19] A.L.F.S. D’Almeida, J.A. Melo Filho, R.D. Toledo Filho, Use of curaua fibers as
reinforcement in cement composites, Chem. Eng. Trans. 17 (2009) 1717–1722.
[20] V. Agopyan, H. Savastano, V. John, M. Cincotto, Developments on vegetable
fibre–cement based materials in São Paulo, Brazil: an overview, Cem. Concr.
Compos. 27 (2005) 527–536.
[21] H. Savastano Jr., P.G. Warden, R.S.P. Coutts, Potential of alternative fibre
cements as building materials for developing areas, Cem. Concr. Compos. 25
(2003) 585–592.
[22] R.M. de Gutiérrez, L.N. Díaz, S. Delvasto, Effect of pozzolans on the
performance of fiber-reinforced mortars, Cem. Concr. Compos. 27 (2005)
593–598.
 J. Wei et al. / Corrosion Science 106 (2016) 1–15
[23] B.J. Mohr, J.J. Biernacki, K.E. Kurtis, Supplementary cementitious materials for
mitigating degradation of kraft pulp fiber-cement composites, Cem. Concr.
Res. 37 (2007) 1531–1543.
[24] M. Antoni, J. Rossen, F. Martirena, K. Scrivener, Cement substitution by a
combination of metakaolin and limestone, Cem. Concr. Res. 42 (2012)
1579–1589.
[25] ASTM C192 Standard Practice for Making and Curing Concrete Test Specimens
in the Laboratory, ASTM International, West Conshohocken, Pennsylvania,
United States, 2015.
[26] ASTM C1185-08, Standard Test Methods for Sampling and Testing
Non-Asbestos Fiber-Cement Flat Sheet, Roofing and Siding Shingles, and
Clapboards, ASTM, International, West Conshohocken, Pennsylvania, United
States, 2012.
[27] ASTM D3882/D3822M-14, Standard Test Method for Tensile Properties of
Single Textile Fibers, ASTM International, West Conshohocken, PA, 2014.
[28] J. Wei, C. Meyer, Degradation rate of natural fiber in cement composites
exposed to various accelerated aging environment conditions, Corros. Sci. 88
(2014) 118–132.
[29] S. Zhandarov, E. Mäder, Characterization of fiber/matrix interface strength:
applicability of different tests, approaches and parameters, Compos. Sci.
Technol. 65 (2005) 149–160.
[30] C.-H. Liu, J.A. Nairn, Analytical and experimental methods for a fracture
mechanics interpretation of the microbond test including the effects of
friction and thermal stresses, Int. J. Adhes. Adhes. 19 (1999) 59–70.
[31] F. Deschner, F. Winnefeld, B. Lothenbach, S. Seufert, P. Schwesig, S. Dittrich, F.
Goetz-Neunhoeffer, J. Neubauer, Hydration of Portland cement with high
replacement by siliceous fly ash, Cem. Concr. Res. 42 (2012) 1389–1400.
[32] K. De Weerdt, M.B. Haha, G. Le Saout, K.O. Kjellsen, H. Justnes, B. Lothenbach,
Hydration mechanisms of ternary Portland cements containing limestone
powder and fly ash, Cem. Concr. Res. 41 (2011) 279–291.
[33] P. Hou, S. Kawashima, D. Kong, D.J. Corr, J. Qian, S.P. Shah, Modification effects
of colloidal nanoSiO2 on cement hydration and its gel property, Compos. Part
B—Eng. 45 (2013) 440–448.
[34] H.F.W. Taylor, Cement Chemistry, 2nd edition, Thomas Telford Service Ltd.,
London, 1997.
[35] L. Segal, J.J. Creely, A.E. Martin, C.M. Conrad, An empirical method for
estimating the degree of crystallinity of native cellulose using the X-ray
diffractometer, Textile Res. J. 29 (1959) 786–794.
[36] A.K. Vijay, K. Kaushik, Susheel Kalia, Effect of mercerization and benzoyl
peroxide treatment on morphology, thermal stability and crystallinity of sisal
fibers, Int. J. Textile Sci. 1 (2012) 101–105.
[37] M.F. Dasong Dai, Characteristic and performance of elementary hemp fibre,
Mater. Sci. Appl. 1 (2010) 336–342.
[38] M.B. Roncero, A.L. Torres, J.F. Colom, T. Vidal, The effect of xylanase on
lignocellulosic components during the bleaching of wood pulps, Bioresour.
Technol. 96 (2005) 21–30.
[39] L.Y. Mwaikambo, M.P. Ansell, Chemical modification of hemp, sisal, jute, and
kapok fibers by alkalization, J. Appl. Polym. Sci. 84 (2002) 2222–2234.
[40] R.G. Elenga, G.F. Dirras, J. Goma Maniongui, P. Djemia, M.P. Biget, On the
microstructure and physical properties of untreated raffia textilis fiber,
Compos. Part A: Appl. Sci. Manufact. 40 (2009) 418–422.
[41] M. Le Troedec, D. Sedan, C. Peyratout, J.P. Bonnet, A. Smith, R. Guinebretiere,
V. Gloaguen, P. Krausz, Influence of various chemical treatments on the
composition and structure of hemp fibres, Compos. Part A: Appl. Sci.
Manufact. 39 (2008) 514–522.
[42] A. Roy, S. Chakraborty, S.P. Kundu, R.K. Basak, S. Basu Majumder, B. Adhikari,
Improvement in mechanical properties of jute fibres through mild alkali
treatment as demonstrated by utilisation of the Weibull distribution model,
Bioresour. Technol. 107 (2012) 222–228.
[43] R.M. Espinosa, L. Franke, Inkbottle pore-method: prediction of hygroscopic
water content in hardened cement paste at variable climatic conditions, Cem.
Concr. Res. 36 (2006) 1954–1968.
[44] B. Lothenbach, K. Scrivener, R.D. Hooton, Supplementary cementitious
materials, Cem. Concr. Res. 41 (2011) 1244–1256.
15
[45] P. Fidjestøl, R. Lewis, 12—microsilica as an addition, in: P.C. Hewlett (Ed.),
Lea’s Chemistry of Cement and Concrete, Fourth Edition,
Butterworth-Heinemann Oxford, 1998, pp. 679–712.
[46] H. Madani, A. Bagheri, T. Parhizkar, The pozzolanic reactivity of
monodispersed nanosilica hydrosols and their influence on the hydration
characteristics of Portland cement, Cem. Concr. Res. 42 (2012) 1563–1570.
[47] A. Neville, P.-C. Aïtcin, High performance concrete—an overview, Mat. Struct.
31 (1998) 111–117.
[48] P. Hewlett, Lea’s Chemistry of Cement and Concrete,
Butterworth-Heinemann, San Diego, 2004.
[49] J.I. Escalante-Garcia, Nonevaporable water from neat OPC and replacement
materials in composite cements hydrated at different temperatures, Cem.
Concr. Res. 33 (2003) 1883–1888.
[50] S.T. Lee, H.Y. Moon, R.D. Hooton, J.P. Kim, Effect of solution concentrations and
replacement levels of metakaolin on the resistance of mortars exposed to
magnesium sulfate solutions, Cem. Concr. Res. 35 (2005) 1314–1323.
[51] A. Korpa, T. Kowald, R. Trettin, Hydration behaviour, structure and
morphology of hydration phases in advanced cement-based systems
containing micro and nanoscale pozzolanic additives, Cem. Concr. Res. 38
(2008) 955–962.
[52] M.D. Campbell, R.S.P. Coutts, Wood fibre-reinforced cement composites, J.
Mat. Sci. 15 (1980) 1962–1970.
[53] R.S.P. Coutts, P. Kightly, Microstructure of autoclaved refined wood-fibre
cement mortars, J. Mat. Sci. 17 (1982) 1801–1806.
[54] R. Andonian, Y.W. Mai, B. Cotterell, Strength and fracture properties of
cellulose fibre reinforced cement composites, Int. J. Cement Compos. 1 (1979)
151–158.
[55] P.J.M. Monteiro, O.E. Gjorv, P.K. Mehta, Effect of condensed silica fume on the
steel-cement paste transition zone, Cem. Concr. Res. 19 (1989) 114–123.
[56] M. Avi, Steel-concrete bond in high-strength lightweight concrete, Mater. J.
89 (1993).
[57] H. Gu, Tensile behaviours of the coir fibre and related composites after NaOH
treatment, Mater. Design 30 (2009) 3931–3934.
[58] A.K. Bledzki, J. Gassan, Composites reinforced with cellulose based fibres,
Prog. Polym. Sci. 24 (1999) 221–274.
[59] J. Wei, C. Meyer, Degradation mechanisms of natural fiber in the matrix of
cement composites, Cem. Concr. Res. 73 (2015) 1–16.
[60] T.E. Timell, Recent progress in the chemistry of wood hemicelluloses, Wood
Sci. Technol. 1 (1967) 45–70.
[61] O. Dahlman, A. Jacobs, J. Sjöberg, Molecular properties of hemicelluloses
located in the surface and inner layers of hardwood and softwood pulps,
Cellulose 10 (2003) 325–334.
[62] J.X. Sun, X.F. Sun, H. Zhao, R.C. Sun, Isolation and characterization of cellulose
from sugarcane bagasse, Polym. Degrad. Stab. 84 (2004) 331–339.
[63] W. Hao, K.T.L. Alan, Hemp and hemp-based composites, in: Green Composites
from Natural Resources, CRC Press, 2013, pp. 63–94.
[64] A. Dufresne, Nanocellulose From Nature to High Performance Tailored
Materials, De Gruyter, Berlin, 2012.
[65] E. Jakab, O. Faix, F. Till, Thermal decomposition of milled wood lignins studied
by thermogravimetry/mass spectrometry, J. Anal. Appl. Pyrolysis 40–41
(1997) 171–186.
[66] R. Sun, Q. Lu, X.F. Sun, Physico-chemical and thermal characterization of
lignins from Caligonum monogoliacum and Tamarix spp, Polym. Degrad. Stab.
72 (2001) 229–238.
[67] W.A. Hunnicutt, Characterization of Calcium-silicate-hydrate and
Calcium-alumino-silicate-hydrate, University of Illinois at
Urbana-Champaign, 2013.
[68] S. Jalali, A. Peyroteo, R.M. Ferreira, From Waste Product to Metakaolin:
Beneficial Impact on Performance Parameters, Repositório Cientfico de Acesso
Aberto de Portugal, 2006.