Construction and Building Materials 185 (2018) 508–518

Contents lists available at ScienceDirect

Construction and Building Materials

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

The role of seawater in interaction of slag and silica fume with cement in
low water-to-binder ratio pastes at the early age of hydration
Huang Li, Nima Farzadnia ⇑, Caijun Shi ⇑
Key Laboratory for Green & Advanced Civil Engineering Materials and Application Technology of Hunan Province, College of Civil Engineering, Hunan University,
Changsha 410082, China

h i g h l i g h t s

Seawater changed the interaction of SF and slag with cement in favor the autogenous shrinkage.

Seawater increased the role of cement and slag in the strength development.

Seawater increased the role of slag in the hydration of the cement.

The reactivity of the SF particles was decreased in seawater blended paste.

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

Article history: Very low water-to-binder ratio of ultra-high performance concrete (UHPC) may facilitate using seawater
Received 17 April 2018 as the mixing water, as there may not be sufficient moisture and air in its dense matrix to initiate the
Received in revised form 10 July 2018 corrosion of steel reinforcements. However, extensive research is needed to investigate the effects of
Accepted 12 July 2018
intermixed chloride from seawater on fresh and hardened properties UHPC. This study investigated
Available online 19 July 2018
the role of seawater in the interaction of slag and silica fume with cement in very low water-to-binder
paste of UHPC at early ages of hydration. The results showed that seawater increased the reactivity of slag
Keywords:
and enhanced its interaction with cement due to triggered reactions between seawater components and
Seawater
Hydration kinetics
rich Al-phases in cement and slag. On the contrary, seawater decreased the interaction of silica fume with
Autogenous shrinkage cement. The overall influence of seawater in the low water-to-binder paste enhanced the hydration
Compressive strength kinetics of binary and ternary binders and restrained the autogenous shrinkage. Furthermore, the early
Low w/b paste age compressive strength was increased. The findings of the study showed that using seawater was in
favor of early age properties of paste with very low water-to-binder ratio.
Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction Some scholars reported that employing seawater as mixing

water in concrete changes the strength development of the matrix.
The needs for concrete infrastructure expend abundant amount For example, it attributes to the early age strength gain and
of natural resources like aggregates and freshwater, which is more reduces the setting time [2–4]. Furthermore, the concrete
evident in fast developing countries like China. In coastal areas, microstructure is improved due to the acceleration of the hydra-
where the quantity of fresh water is very limited, its usage as mix- tion process by chloride ions input, the induced chemical reac-
ing water will be extremely costly and the work efficiency during tions; and formation of precipitations in the pore structure [4,5].
the construction will be very low. Hence, incorporation of seawater The inclusion of seawater in concrete with supplementary cemen-
as mixing water can facilitate construction of marine structures titious materials (SCMs) can further refine the pore structure. Shi
more efficiently and at lower costs, while cuts down on the exces- et al. [5] found that the combined effect of Metakaolin (MK) as
sive usage of fresh water resources. However, there is a debate on an alumina rich SCM and seawater was beneficial to the
applicability of the seawater usage as it contains aggressive microstructure of conventional concrete, due to formation of pre-
compounds [1]. cipitations and filler compounds through chemical reactions of salt
ions in seawater and alumina phases of MK. They also showed that
⇑ Corresponding authors. the hybrid system of MK and seawater improved the chloride resis-
E-mail addresses: [email protected] (N. Farzadnia), [email protected]
tance of concrete by increasing the amount of chemically bound
(C. Shi). chloride through the formation of Friedel’s salt. Also, Li et al. [6]

https://doi.org/10.1016/j.conbuildmat.2018.07.091
0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
 H. Li et al. / Construction and Building Materials 185 (2018) 508–518 509

reported that the compressive strength of concrete mixed with macro-level. A factorial design system based on Wadsworth’s
seawater exhibited the highest value in MKa containing concrete study [30], was applied to quantify the effect of each component
specimens as compared with plain ones. on the examined properties at the early age of hydration. To elab-
Although some research affirms the advantages of using seawa- orate the underlying mechanisms, Thermogravimetric Analysis
ter as mixing or curing water in concrete [5,6], many studies (TGA), Fourier transform infrared spectroscopy (FTIR), X-ray
reported its deteriorating effect due to the formation of expansive diffraction (XRD) and Nitrogen adsorption and desorption (NAD)
compounds in reaction with hydration products [7–12]. More were applied to assess the role of seawater in the interaction of
importantly, the use of seawater as mixing water will escalate cor- slag and silica fume with cement in very low water-to-binder ratio
rosion risk of the embedded steel in reinforced concrete [1]. pastes.
Reportedly, the corrosion of conventional concrete exposed to sea-
water resulted in serious problems with millions of dollars spent 2. Raw materials and experimental methods
for maintenance, repair or replacement of damaged structures
[13,14]. Mohammed et al. [15] showed that the seawater mixed 2.1. Raw materials
concrete exhibited higher re-bar corrosion in tidal environments
Portland cement P.I 42.5 complying with the Chinese Standards GB175-2007
when compared with fresh water mixed concrete. Some other and silica fume (SF) with 63% particle of 0.1–0.5 lm and specific surface area of
studies also confirmed that the usage of seawater as mixing water 18500 m2/kg, and slag with specific surface area of 446 m2/kg were used. Table 1
accelerated the corrosion of reinforcement in concrete and caused presents the chemical compositions of the cement, SF and slag. A polycarboxylates
the spalling and cracking of structural members [16–18]. There- based superplasticizer (SP) with solid content of 49% was used to guarantee a good
flowability of the paste.
fore, the high risk of corrosion in seawater mixed reinforced con-
The seawater used in this study was prepared by synthetic method as complied
crete has led to its prohibition by various international standard with ASTM D-14. The chemical composition of seawater is presented in Table 2. The
regulations. However, it has been allowed in plain concrete manu- pH of seawater was 8.4.
facture under the proviso that it complies with the given standards
criteria of the pertinent regulation [19,20]. 2.2. Mixture proportion and specimen preparation
Unlike the conventional concretes, the incorporation of seawa-
To characterize and quantify the interaction of SF, slag and cement with the
ter as the mixing water in concretes with high density and low per-
presence of seawater, the seven-batch factorial design method was used in this
meability such as ultra-high performance concrete (UHPC) can be study. Two series of specimens were cast; with fresh water or seawater. Mixture
of less harm. UHPC is a cement-based composite with a compres- proportion of cement paste with w/b ratio of 0.2 was used in this study and the
sive strength typically higher than 150 MPa and tensile strength SP dosage was kept at 2% by the mass of binder. The detailed proportions are shown
greater than 20 MPa. The raw materials of UHPC include cement, in Table 3. Initially, the binary and ternary mixtures were mixed in a mixer for 1
min at high speed; water and SP were added into the mixer and mixed for 3 min
supplementary cementitious materials, quartz sand, superplasti-
under low speed and high-speed conditions, respectively. When pastes were ready,
cizer, and fibers. The use of low water-to-binder ratio (0.18) they were cast into 40  40  160 mm molds and vibrated for 1 min. After the
and SCMs in the UHPC decreases the porosity and refines the pore vibration, all the specimens were moved to a curing chamber under temperature
structure to denser and less interconnected voids [21]. Alkaysi of 20 °C and RH of 95% ± 5 for 24 h. Then, the specimens were demolded and kept
in curing conditions to testing ages. The sample preparation and curing regime
et al. [22] reported that the porosity and the charge pass in UHPC
complied with Chinese standard GB/T50081-2002.
were as low as 5.9% and 89 C, respectively. In another study by
Tafraoui et al. [23], the gas permeability of UHPC and chloride dif-
Table 1
fusion coefficient were reported to be lower than 1.5  1019 m2
Chemical compositions of cementitious materials (%).
and 1.7  1014 (m2/s), respectively. Therefore, it is assumed that
the dense microstructure of UHPC along with the high chloride Chemical Composition SiO2 Al2O3 Fe2O3 CaO MgO

binding capacity due to incorporation of SCMs can immobilize Cement 25.26 6.38 4.05 64.67 2.68
the corrosive components available in seawater. Furthermore, SF 90.82 1.03 1.50 0.45 0.83
Slag 33 13.91 0.82 39.11 10.04
there is not sufficient moisture and air in the dense matrix of
low water-to-binder ratio bulk paste to initiate the corrosion of
steel reinforcements in UHPC. However, few issues need to be care-
fully addressed in the usage of seawater as mixing water in the Table 2
The artificial seawater specified by ASTM D-14 (g/L).
production of UHPC. Firstly, the role of seawater in interaction of
slag and silica fume with cement and its effect on the hydration Types NaCl MgCl26H2O Na2SO4 CaC12 KCl
kinetics needs to be studied. In the low water-to- paste, the hydra- Content 24.5 2.54 4.1 1.2 0.1
tion process can directly affect the autogenous shrinkage, which is
a critical issue in UHPC [24,25]. The interaction of slag and silica
fume with cement in fresh water mixed systems and its effect on
Table 3
autogenous shrinkage of UHPC is well documented [26–29]. How- Mixture design of pastes.
ever, the relatively higher pH and the existence of salts in the sea-
Number Cement (%) SF (%) Slag (%) SP (%) W/B ratio Water type
water can change the interaction and the consequent early age
properties. There is little information on the hydration kinetics of SW1 100 0 0 2 0.20 Seawater
SW2 85 15 0 2 0.20 Seawater
the binary and ternary cementitious systems mixed with seawater.
SW3 75 0 25 2 0.20 Seawater
Secondly, the possibility of corrosion of reinforcement in the sea- SW4 72.2 11.1 16.7 2 0.20 Seawater
water mixed UHPC needs further attention. SW5 70 30 0 2 0.20 Seawater
In this study, the role of seawater in the interaction of slag and SW6 58.3 16.7 25 2 0.20 Seawater
silica fume with cement in the binding matrix of UHPC will be SW7 50 0 50 2 0.20 Seawater
FW1 100 0 0 2 0.20 Freshwater
studied. The overarching purpose of this study is to investigate FW2 85 15 0 2 0.20 Freshwater
the effect of seawater as the mixing water on the binary and tern- FW3 75 0 25 2 0.20 Freshwater
ary binder of slag, silica fume and cement with a very low water- FW4 72.2 11.1 16.7 2 0.20 Freshwater
to-binder ratio. Accordingly, heat evolution, autogenous shrinkage FW5 70 30 0 2 0.20 Freshwater
FW6 58.3 16.7 25 2 0.20 Freshwater
and compressive strength tests were used to show the synergic
FW7 50 0 50 2 0.20 Freshwater
effect of seawater and cementitious materials in the paste at the
510 H. Li et al. / Construction and Building Materials 185 (2018) 508–518

2.3. Experimental methods

2.3.1. Heat of hydration

The hydration heat was tested using an isothermal calorimeter. About 4 g of the
mixtures were taken from the prepared fresh paste and filled into a glass bottle,
then moved into TAM Air isothermal calorimeter immediately for testing. All mea-
surements were lasted for 70 h.

2.3.2. Autogenous shrinkage

A horizontal test setup was used to test the autogenous shrinkage of the mix-
tures. The fresh paste was cast into corrugated pipe molds with inner diameter of
20 mm and length of 35 mm, and then sealed with steel cap at each side to avoid
moisture loss. During the experiment, one side of the steel cap was fastened in a
steel frame, and the other side of the specimen can be freely moved without
restrain. The autogenous shrinkage was monitored by a current displacement sen-
sor with resolution of 0.5um and accuracy of 0.05% at the side without restrain. All
the specimens were deposited in a chamber with temperature of 20 ± 1 °C and a rel-
ative humidity of 65 ± 5%. During the test, the length of corrugated pipe was
recorded every 1 h at the first 24 h and 3 h at the following 48 h. Fig. 2. Expression apparatus.

2.3.3. Compressive strength

The compressive strength was tested by using 40  40  40 mm cubic speci- 2.3.7. Free chloride concentration
mens complying with Chinese Standards GB/T 17671-1999. The compressive The free chloride concentration of samples was measured by pore solution
strength test was conducted on three specimens for the same mixture, and the final expression and titration. The pore solution was obtained by pore solution expres-
results were taken as the average. sion method as in [31]. The expression apparatus (Fig. 2) consisted of a hollow steel
cylinder where paste was contained. A piston, which was located on the top of the
paste sample inside the hollow cylinder, and a base plate with fluid drainage chan-
2.3.4. Calcium hydroxide (CH) content
nels for gathering the extracted the pore solution. The surface of paste samples
The calcium hydroxide (CH) content of the specimens was assessed quantita-
(total of 4 cement paste discs for each test series) were dried using a cloth to reach
tively by thermogravimetric analysis (TGA). Specimens were broken to pieces,
a saturated surface dry condition. The samples were later crushed, placed in the
and then immerged in absolute alcohol liquid for 24 h to terminate the hydration.
expression apparatus and loaded to 550 MPa at a rate of 2.0–3.0 MPa/s. The pase
After the immersion, the samples were ground to powder until all passed through
samples were held at this pressure for 1 min, then unloaded rapidly and reloaded
the 0.045 mm sieve. The collected powder was then moved to a vacuum oven at
again. The expressed pore solution was collected with an injection syringe. The pin-
60 °C for 24 h. Then, the prepared powder samples were heated at 10 °C/min from
head was covered with a latex tubing to prevent leaking, then was injected into a
room temperature to 1000 °C in a nitrogen atmosphere in a TGA instrument. The
plastic container and sealed immediately to avoid carbonation. The pore solution
quantitative analysis of CH content was carried out based on the dehydration of
expression apparatus was cleaned with anhydrous ethanol between every two
CH as in Eq. (1). The mass loss ‘mH2 O ’ was obtained from the TGA curves at the range
expressions to avoid cross contamination. The reported results were averages of
of 400–450 °C, at which the CH is dehydrated (Fig. 1). The CH contents ‘mCH’ was
three concentrations of expressed pore solution.
then calculated by Eq. (2):

CaðOHÞ2 ! CaO þ H2 O ð1Þ 2.3.8. Nitrogen adsorption and desorption (NAD)

NAD analysis were conducted with the gas volumetric analyzer Quantachrome
mCH ¼ mH2 O  MCH =MH2 O ð2Þ Instruments QuadraSorb Station 1. The cement pastes were preheated at 60 °C for
24 h under vacuum condition to eliminate the influence of free water.
where mH2 O is mass loss of water, MCH is molar mass of Ca(OH)2, and MH2 O is molar
mass of H2O.
3. Results
2.3.5. X-ray diffraction (XRD)
The preparation of samples for XRD was the same as those prepared for TGA. 3.1. Hydration kinetics
The prepared powders were tested by using Philips X-ray diffractometer with CuKa
radiation. The samples were step-scanned from 10 to 70° (2h) at a rate of 5°/min.
The heat of hydration is a good indication to the role of different
SCMs in hydration kinetics of cement. Previously, the hydration of
2.3.6. Fourier transform infrared spectroscopy (FTIR)
The preparation of samples for FTIR was the same as those prepared for TGA.
cement in UHPC was comprehensively studied by Shi et al. [32].
The FTIR spectra were tested by Thermo-Scientific IS10 FTIR workstation. About They investigated the influence rate of slag and SF in the UHPC bulk
1 mg of sample and 100 mg of IR-grade KBr were ground together for 5 min and paste using the factorial design method. They reported that the
pressed into a thin and transparent disc. Each sample was tested at a resolution role of SF in the hydration was more highlighted than that of slag
of 2 cm1 with 32 scans.
in a low water-to-binder paste. The present study, though, focused
on the role of seawater in the interaction of slag and SF with
cement. Fig. 3 shows the heat flow of cementitious materials in
the paste at the first 70 h. The solid curves represent the hydration
100 0.2
Mass/% scans of the specimens mixed with fresh water (FW), and the dash
DTG/(%/min) 0 curves are indicative of the seawater mixed specimens (SW). In
95 general, the hydration in the low water-to-binder paste was
DTG (% /min)
Mass loss (% )

-0.2 retarded in comparison to the conventional paste because of the

higher content of superplasticizer [33]. Nevertheless, the dosage
90 -0.4
of superplasticizer was kept constant throughout the study, so
-0.6 the retarding effect can be considered the same for all the speci-
85 mens. Besides, the accelerating effect of slag and silica fume on
CH -0.8 the hydration kinetics of cement with fresh water is very well doc-
umented [34,35], However, Fig. 3 shows an obvious tendency of
80 -1
0 200 400 600 800
seawater to accelerate the hydration process of the binder from 2
Temperature (°C) to 5 h comparing to those cast with fresh water. For pure cement
(SW1), the acceleration rate was much less than that of SF and slag
Fig. 1. Quantitative analysis of CH content. containing specimens, showing the role of seawater on the
 H. Li et al. / Construction and Building Materials 185 (2018) 508–518 511

12 (Fig. 4). He reported that the kinetics of cement hydration may

SW1
SW2 be described by the following Eq. (3):
SW3
10 t  t0
SW4 P ¼ P0  ð3Þ
SW5 ðt  t 0 Þ þ t50
SW6
8 where,
Heat f low(KJ/Kg*h)

SW7
FW1
FW2
FW3 P = Heat evolution (KJ/Kg) at the time t,
6 FW4 t0 = Ending time of induction period,
FW5
FW6
t50 = The time for 50% reaction degree,
4 FW7 P0 = Ultimate heat evolution when the time is infinitely long.

2 If the experimental results of heat evolution are figured in

reciprocal diagrams of P verses 1/(t  t0), linear curves could be
obtained; t50 and 1/P0 can be determined by the extrapolating
0 method. The values of P0, t0, and the equations are given in Table 4.
0 10 20 30 40 50 60 70 Accordingly, the seven-batch factorial design method, known as
Time(hour) ‘‘simplex-centroid” design, was applied [30]. The values of P0 and
Fig. 3. Heat flow of different pastes.
t0 from Table 4 were used in the factorial design to calculate the
influential coefficient of each constituent. Assuming the mixture
is composed of three components x1, x2 and x3, then:
interaction of slag and SF with cement. The incorporation of
Y ¼ b1 x1 þ b2 x2 þ b3 x3 þ b12 x1 x2 þ b23 x2 x3 þ b13 x1 x3
seawater in SF containing specimens accelerated the occurrence
of the main hydration peak for almost 2 h for SW2 (15% SF) and þ b123 x1 x2 x3 ð4Þ
5 h for SW5 (30% SF), respectively. Comparatively, the effect of sea- Where,
water on the slag containing specimens was the highest. With the
replacement of 25% slag, the hydration time was accelerated for Y = Responses, including any structural and property character-
about 4 h for SW3 (25% slag), and more than 10 h for SW7 (50% istics of ternary cementitious materials,
slag). The remarkable hydration acceleration of slag containing b1, b2, b3, is the influential coefficient related to the cement, SF
specimens may be mainly related to the complex reactions and slag in this study,
between Al-rich phases of slag and components in the seawater. x1, x2 and x3 are the portions of the three components.
On one hand, the seawater provides an alkaline environment dur-
ing the hydration of slag, which can activate the slag [36]; thus, it Based on the parameters shown in Table 4, the relationship
accelerated the hydration of slag at the early age. Shi et al. [5] between P0, t0 and the ternary binder composition of fresh water
reported the increased hydration products in Metakaolin contain- and seawater mixed specimens were obtained as the following
ing specimens (w/b = 0.4) at ages as early as 3 days as an indication equations:
of hydration acceleration. Nevertheless, there is still limited
YFP0 ¼ 2:55x1 þ 3:93x2 þ 2:95x3  0:031x1 x2  0:29x2 x3
knowledge on how seawater can accelerate the hydration process
and hydration kinetics in binary and ternary systems.  0:024x1 x3 þ 0:0040x1 x2 x3 ð5Þ
The incorporation of seawater also accelerated the hydration of
ternary mixtures of slag and SF and increased the total heat evolu- YSP0 ¼ 2:23x1 þ 3:16x2 þ 4:23x3  0:0013x1 x2  0:39x2 x3
tion comparing to the fresh water mixed specimens (Fig. 4). To  0:036x1 x3 þ 0:0067x1 x2 x3 ð6Þ
scrutinize the effect of seawater on the interaction of slag and SF
with cement, the hydration kinetics were calculated based on an YFt0 ¼ 0:10x1 þ 0:89x2 þ 0:39x3  0:010x1 x2  0:035x2 x3
equation proposed by Knudsen [37] using total heat evolution
 0:0031x1 x3 þ 0:00037x1 x2 x3 ð7Þ

Table 4
250 SW1 Values of P0, t0, t50, and the equations.
SW2
SW3 NO. P0 (kJ/kg) t0 (h) t50 (h) Equation
Heat of hydration(KJ/Kg)

SW4
200 SW1 223 9.80 29.92 P ¼ 223  ðt9:8Þþ29:92
t9:80
SW5
SW6 SW2 220 8.51 25.77 P ¼ 210  ðt8:51Þþ25:77
t8:51

SW7
SW3 210 11.21 35.45 P ¼ 195  ðt11:21Þþ35:45
t11:21

150 FW1
FW2 SW4 233 9.54 32.72 P ¼ 233  ðt9:54Þþ32:72
t9:54

FW3 SW5 223 9.66 31.33 P ¼ 223  ðt9:66Þþ31:33

t9:66
FW4
100 FW5 SW6 225 9.59 27.27 P ¼ 225  ðt9:59Þþ27:27
t9:59

FW6 SW7 239 10.02 28.99 P ¼ 239  ðt10:02Þþ28:99

t10:02
FW7
FW1 255 10.23 30.99 P ¼ 255  ðt10:23Þþ30:99
t10:23

50 FW2 236 8.84 28.89 P ¼ 236  ðt8:84Þþ28:89

t8:80

FW3 220 11.56 37.56 P ¼ 224  ðt11:56Þþ37:56

t11:56

FW4 234 10.20 32.42 P ¼ 250  ðt10:20Þþ32:42

t10:20
0
0 10 20 30 40 50 60 70 FW5 231 12.12 36.43 P ¼ 231  ðt12:12Þþ36:43
t12:12

Time(hour) FW6 223 10.24 29.34 P ¼ 223  ðt10:24Þþ29:34

t10:24

FW7 215 16.79 40.22 P ¼ 215  ðt16:79Þþ40:22

t16:79

Fig. 4. Total heat of different pastes.

512 H. Li et al. / Construction and Building Materials 185 (2018) 508–518

YSt0 ¼ 0:098x1 þ 0:47x2  0:0016x3  0:0054x1 x2 can be well explained by the lower reactivity of particles and the
 0:0011x2 x3 þ 0:020x1 x3  0:000053x1 x2 x3 ð8Þ delayed hydration in slag-cement system as shown in Fig. 3. Exten-
sive information on the formation mechanism of autogenous
YFP0 and YSP0 represent the heat evolution of specimens mixed shrinkage by capillary tension and influence of relative humidity
with fresh water and seawater when the time extends to an infinite and pore structure can be found in [25].
length, respectively. YFt0 and YSt0 represent the ending time of For seawater mixed specimens, the autogenous shrinkage was
induction period. From the Eq. (5), it can be seen that b1, b2, b3 restrained remarkably for some, especially for the ones with SF.
(as the influential coefficients of the equation) are positive, indicat- The seawater decreased the strain values of SW5 (30% SF) and
ing that the three cementitious materials conducted positively to SW7 (50% slag) by 51.8%, and 31.5% comparing to their fresh water
the P0, however, the impact of SF on P0 was more remarkable than mixed counterparts at 72 h, respectively. However, seawater with
that of cement and slag. It has very good consistency with a study pure cement experienced an increase of 48.5% of the shrinkage
by Shi et al. [31]. The factorial design method confirms the major when compared to FW1. Furthermore, the same trend was
role of SF particles in the hydration of cement. The incorporation observed for the specimens with lower replacement level of SF,
of seawater in the matrix (Eq. (6), though, increased the b3 showing namely, SW2 (15% slag). The role of seawater on the interaction
the increased influence of slag on the hydration kinetics with the of cementitious binder and its effect on the autogenous shrinkage
presence of the seawater. Comparing b1 and b3 in Eq. (6) indicates was assessed using the factorial design method. The relationship
that seawater had more influence on the reactivity of the slag than between autogenous shrinkage strain at 72 h and the ternary bin-
that of cement. With respect to t0 (Eq. (7), the b3 value in Eq. (8) der composition of fresh water and seawater mixed specimens
turned negative in seawater mixed specimens, showing that sea- were obtained as in the following equations:
water changed the role of slag to an accelerating one in the hydra-
YFL ¼ 0:0035x1 þ 0:055x2  0:0017x3  0:00038x1 x2
tion of cement at the end of the induction period. The significant
accelerating effect of seawater with presence of slag can be clearly þ 0:00086x2 x3 þ 0:00032x1 x3  0:000023x1 x2 x3 ð9Þ
observed for SW7 comparing to its counterpart (FW7) as well as
the control specimen (SW1) as shown in Fig. 3. It can be stated that YSL ¼ 0:0052x1  0:12x2  0:0022x3 þ 0:0018x1 x2
slag retarded the hydration process of the paste with fresh water
þ 0:0056x2 x3 þ 0:00018x1 x3  0:000078x1 x2 x3 ð10Þ
but accelerated it when mixed with seawater. The results from
simplex-centroid design show that the role of slag on hydration From Eqs. (9) and (10), it can be seen that the SF and cement
kinetics was more highlighted as compared with SF and cement had a direct effect on autogenous shrinkage of low w/b paste
in seawater mixed specimens. mixed with fresh water, while the incorporation of slag inversely
affected the autogenous shrinkage as shown by the negative sign
3.2. Autogenous shrinkage of the b3. At the same time, the coefficient associate to SF (b2) is
illustrative of the most dominant effect of the SF on the autogenous
Fig. 5 shows the autogenous shrinkage strains of the pastes shrinkage of the fresh water mixed specimens. It shows that SF was
mixed with fresh water and seawater. As seen, the autogenous the major reason of the increased autogenous shrinkage of the low
shrinkage strains ranged between 0.4 mm/m and 1.1 mm/m for water-to-binder paste, while slag had a decreasing effect. However,
all the specimens, which show relatively high shrinkage comparing the replacement of fresh water with the seawater, changed the b2
to those reported for mixtures with high water-to-binder ratio from positive to negative, which indicates that the seawater chan-
[38,39]. The results also revealed that the incorporation of seawa- ged the negative influence of SF in the autogenous shrinkage. On
ter had a significant role in the interaction of slag and SF with the contrary, using seawater as the mixing water increased the role
cement in terms of autogenous shrinkage. For fresh water mixed of cement in the induction of shrinkage, illustrated from the
specimens, the incorporation of SF exhibited significant autoge- increase of b1 in Eq. (10). However, the effect of seawater was
nous shrinkage, which is in good agreement with previous studies more remarkable on the interaction of SF with cement than that
[40,41]. With the replacement of slag, though, the autogenous of slag in terms of autogenous shrinkage for low water-to-binder
shrinkage was much lower than that of SF containing specimens. paste.
When the slag content reached 50%, the autogenous shrinkage
decreased by 19% compared to that of SF containing specimens 3.3. Compressive strength
at 72 h. The lower autogenous shrinkage strains caused by slag
Fig. 6 shows the compressive strength of fresh water and sea-
water mixed pastes with different cement, SF and slag contents
1.2 FW1 at 3 days of hydration. As seen for fresh water mixed specimens,
FW2
the incorporation of SF at lower contents increased the compres-
Autogenous shrinkage(mm/m)

FW3
1.0 FW4 sive strength at the early age, while the effect of slag on the early
FW5
FW6 age strength development was negligible. The similar results were
FW7
0.8 SW1
observed for the UHPC binding matrix in previous studies [42,43].
SW2 Results also showed that the increase in the replacement of slag to
SW3
0.6 SW4 50% decreased the compressive strength to a level below the con-
SW5 trol specimen (FW1). This is mainly related to the low reactivity of
SW6
0.4 SW7 slag, which can lower the hydration rate especially when incorpo-
rated at high volumes. The decelerating effect of slag was also
found from the hydration scan as illustrated in Fig. 3.
0.2
Fig. 6 shows that using seawater for casting low w/b pastes
increased the compressive strength of all specimens, which was
0.0
0 6 12 18 24 30 36 42 48 54 60 66 72 more remarkable for slag containing mixtures. The increasing rates
Time(h) of each specimen regarding its freshwater mixed counterparts are
annotated in percentage. Previously, Shi et al. [5] and Li et al. [8]
Fig. 5. Autogenous shrinkage of different pastes. confirmed that seawater increased the compressive strength of
 H. Li et al. / Construction and Building Materials 185 (2018) 508–518 513

95 terms of the final heat evolution and initiation of the acceleration

FW peak, while it decreased the interaction of SF with cement. For
SW
the autogenous shrinkage, the seawater reduced the effect of SF
Compressive strength Mpa

on the shrinkage remarkably, while the interaction of slag and

2.2% cement was kept almost unchanged. For the compressive strength,
1.7%
90 2.4%
the cement and slag were mainly effective on the strength devel-
4.0 opment of seawater mixed specimens, whereas the SF showed less
influence in the strength at 3 days.
4.9% To elaborate the role of seawater in the interaction of slag and
3.0% SF with cement in terms of the early age properties, TGA, XRD,
85 FTIR, and NAD were applied. As for an accepted index for the inter-
action of slag, SF and cement, the CH contents of the low w/b paste
4.8% at 3 days were calculated from TGA curves and are shown in Fig. 7.
The highest content of CH (about 9%) was for control specimens,
which is relatively low comparing to that of conventional cement
80
1 2 3 4 5 6 7 mixtures [46]. It is mainly related to the low water-to-binder ratio
Number applied in the specimens, which hinders the complete hydration of
cementitious materials in the matrix. On the other hand, the incor-
Fig. 6. Compressive strength of different pastes at 3 days.
poration of SF and slag further decreased the content of CH reflect-
ing the pozzolanic reactivity of the incorporated slag and SF. The
high reactivity of the ternary binder was also evident from the
the Metakaolin containing paste (w/b of about 0.4) at the early
rapid initiation of the autogenous shrinkage for the slag and SF
ages, which was related to the accelerated formation of hydration
incorporated specimens (Fig. 5). However, the influence of SF in
products and the presence of Al-rich phases in MK. A study by Roy
the restriction of CH crystal growth and its consumption was
and Idorn [44] explained that the hydration of slag can be triggered
higher than that of slag in the freshwater mixed specimens. The
by alkalis released during the hydration process, as well as the
influential coefficients from the hydration kinetics also illustrated
associated temperature rise. Besides, the higher alkalinity of sea-
that the obtained b2 was larger than b3 for the fresh water mixed
water (as compared to fresh water) can further dissolve cementi-
specimens. However, the increase of the SCMs contents lowered
tious materials in the pore solution [45]. This can attribute to
their effect on the CH content.
acceleration of hydration and the increase of compressive strength
For the seawater mixed specimens, the concentration of CH
at early ages. However, results of this study revealed that the incor-
showed a slight increase for control specimen (SW1), while it
poration of seawater affected the interaction of SF and slag with
decreased for other specimens as compared with the fresh water
cement in different fashions in terms of the early age compressive
mixed counterparts. For pure cement paste, previous studies
strength of low water-to-binder paste. Based on the compressive
showed that the seawater increased the hydration products in
strength values, the factorial design equations were constructed
the matrix and was related to acceleration of hydration [6], which
and the influential coefficients of each constituents in compressive
is in good agreement with the obtained hydration scans in Fig. 3.
strength of fresh water specimens (YF) and seawater specimens
The accelerating effect of the seawater on cement hydration may
(YS) were obtained as shown in Eqs. (11) and (12):
be explained by relatively smaller ionic size of Cl comparing to
YF ¼ 0:85x1 þ 0:40x2 þ 0:63x3 þ 0:0089x1 x2  0:055x2 x3 OH [47], which facilitates its diffusivity into the passivating layer
(metastable C-S-H). It can result in rapid rupturing of the passivat-
þ 0:0028x1 x3 þ 0:0010x1 x2 x3 ð11Þ
ing layer on the surface of the cement clinker [48] and increase the
dissolution of cement compounds such as C3S. On the other hand,
YS ¼ 0:87x1 þ 0:36x2 þ 0:61x3 þ 0:0093x1 x2  0:018x2 x3 the negative electron layer (Cl) formed on the surface of the clin-
þ 0:0042x1 x3 þ 0:00031x1 x2 x3 ð12Þ ker can facilitate further dispersion and separation of cement clin-
ker and enhances the hydration [49]. Besides, NaCl in the seawater
As seen from Eq. (11), the obtained values of b1, b2, b3 are pos- can consume the CH in the pore solution as in Eq. (13) and increase
itive, indicating that all binding constituents, namely, cement, SF the CaCl2 content in the matrix [50]. However, In the low water-to-
and slag content contributed to the early age compressive strength binder cement paste, where the low contents of Cl salts are avail-
of the specimens. However, the coefficients values showed that the able, the formation of CH due to hydration overtook its
influence of cement as the main binder was more effective than the consumption.
SCMs followed by slag and SF. Eq. (12) also shows that all the bind-
ing materials contributed to the early age compressive strength of CaðOHÞ2 þ 2NaCl ! CaCl2 þ 2Naþ þ 2OH ð13Þ
the specimens mixed with seawater. However, seawater increased
the influence of the cement on early strength gain, while it The possible increase in the amount of CaCl2 in the pore solu-
decreased the interaction of SF with cement in terms of strength tion due to reaction of NaCl and CH (Eq. (13)) can further accelerate
development. the hydration [51]. CaCl2 can flocculate the C-S-H layers, facilitat-
ing diffusion of ions and water through the metastable C-S-H layer,
4. Discussion allowing a higher rate of diffusion during the early diffusion-
controlled period [52,53].
Results from the hydration kinetics, autogenous shrinkage and Higher hydration rate of seawater mixed cement can increase
the compressive strength at the initial 72 h of hydration showed the self-desiccation, which is followed by an increase in the auto-
that the incorporation of seawater in the low water-to-binder genous shrinkage (SW1) as seen in Fig. 5. For the specimens with
paste changed the influence rate of slag and SF in the cement paste. SF and slag, the CH content reduced to a level lower than that of
The major influential coefficients from the factorial design method the fresh water mixed specimens. It can be related to the modified
are tabulated in Table 5. Overall, using seawater as the mixing crystallinity of the CH by SCMs [54], which can facilitate its con-
water increased the role of slag in the hydration of the binder in sumption by seawater components in the matrix. However, the
514 H. Li et al. / Construction and Building Materials 185 (2018) 508–518

Table 5
The values of major influential coefficients.

Hydration Autogenous shrinkage Compressive strength

Total heat Ending time of induction period
b1 b2 b3 b1 b2 b3 b1 b2 b3 b1 b2 b3
FW 2.55 3.93 2.95 0.10 0.89 0.39 0.0035 0.055 0.0017 0.85 0.40 0.63
SW 2.23 3.16 4.23 0.098 0.47 0.0016 0.0052 0.12 0.0022 0.87 0.36 0.61

10 The second mechanism is dominated by the role of calcium

hydroxide and un-hydrated cement on the formation of Friedel’s
SW
FW salt [59]. Eqs. (15) and (16) show the reaction of the second
9 mechanism mainly in pure cement paste [5]:

C3 A þ CaCl2 þ 10H2 O ! C3 A  CaCl2  10H2 O ð15Þ

CH Content(%)

8
C3 A þ C3 S þ 2NaCl þ 15H2 O ! C3 A  CaCl2  10H2 O
þC2 S  4H2 O þ 2NaOH ð16Þ
7 With the presence of Al phases in the slag, an accumulated
formation of Friedel’s salt can occur as in Eq. (17) [61]:

6 3CaðOHÞ2 þ CaCl2 þ Al2 O3 þ 7H2 O ! 3CaO  Al2 O3  CaCl2  10H2 O

ð17Þ

5 Previously, Justnes [62] stated that the chemical chloride bind-

1 2 3 4 5 6 7 ing is dominated by the content of calcium aluminates and calcium
Number aluminoferrites. Neville [63] argued that the above hypothesis is
rather valid only in the case of intermixed chlorides. However, both
Fig. 7. CH content of different pastes at 3 days.
mechanisms can be considered to be acting in this study, because
seawater was used as the mixing water and high contents of free
chloride ions were introduced to the matrix during the hydration.
CH contents for slag were relatively lower than specimens with SF, Accordingly, the XRD was conducted to analyze phase changes
which reflect different role of seawater on the interaction of SF and in the slag containing specimens. Eminent peaks of C3S, C2S, C3A,
slag with cement in terms of hydration kinetics. CH and Friedel’s salts were observed. In this study, no peaks related
Influential coefficients derived from the factorial design method to brucite and M-S-H were detected owing to either the low
(Table 5) showed that the seawater decreased the influence of SF in amount of MgCl2 in the matrix or their occurrence at later stages
the hydration as the b2 decreased from 3.93 to 3.16. The decreased [64]. Besides, the formation of Kuzel’s salt was not observed. Balo-
influence of SF can be explained by the accelerated hydration of nis [65] and Glasser [66] reported that Kuzel’s salt was formed at a
cement clinker by seawater. The accelerated hydration causes quite low chloride concentration. Hence, the Friedel’s salt was the
rapid render of C-S-H, which can form a passive layer on the sur- main product in this study due to the high concentration of Cl- in
face of SF due to its high surface energy (the nucleation effect) the seawater.
[55,56]; and may lead to reduction of SF and cement interaction. Fig. 8 shows that the peak assigned to CH (d = 0.486 nm) exhib-
This can well explain the reduced influence of SF on the autoge- ited higher intensity for SW1 compared to FW1. It shows a good
nous shrinkage. On the other hand, the influential coefficients conformity with TGA results. This was resulted from the accelera-
derived from hydration kinetics also indicated that the b3 increased tion of hydration process by using seawater as the mixing water,
from 2.95 to 4.23 as seawater incorporated, indicating a higher role which is also reflected by a decrease of peaks associated to C2S
of slag in the hydration kinetics. A study by Chen et al. [57] and C3S. Shi, et al. [5] reported a similar trend in the conventional
reported that sulfates in seawater can act as activators to dissolve concrete; the presence of seawater promoted the production of
slag in the matrix. Moreover, the alkaline environment of the pore Portlandite. The formation of Friedel’s salt and a consequent
solution by formation of calcium hydroxide during the hydration of reduction of C3A were also observed for SW1 as in Eq. (14). For
cement can trigger the rapid dissolution of slag and increases its slag containing specimens (SW3 and SW7), Friedel’s salt intensity
reactivity to consume CH. (d = 0.787) increased, which was along with the consumption of
The chloride binding capacity of slag and SF containing speci- the CH as in Eq. (16). With the increase of the slag content to
mens can also show their interaction degree with cement in sea- 50% (SW7), the peaks of Friedel’s salt extended confirming the
water mixed specimens. In general, the chemical chloride promoting role of slag in the formation of Friedel’s’ salt in seawater
binding is mostly related to two mechanisms; the ion exchange mixed specimens.
mechanism [58,59], and the dissolution and precipitation mecha- Formation of the chloride binding was also observed through
nism [59]. The ion exchange mechanism is the chemical substitu- the free chloride content in the matrix. Fig. 9 illustrates the con-
tion of sulfates, hydroxyl or carbonate groups in AFm phases centration of free chloride in seawater specimens. The original con-
with Cl, forming Friedel’s salt (Eq. (14)) [60]. tent of Cl introduced to the matrix by seawater was 2.76 mg/g by
mass of the paste. As can be seen, the calculated chloride in
9
C3 A  CaSO4  14H2 O >
= liberated form was 1.75 mg/g , which shows that 34% of injected
 chloride in the form of mixing water was bound to the system.
C3 A  CaðOHÞ2  12H2 O þ Cl ! C3 A  CaCl2  10H2 OðFriedel’ssaltÞ
>
; Comparing to conventional concrete, the binding capacity was
C3 A  CaCO3  10H2 O
lower. Li et al. [6] reported that the total bound chloride was 48%
ð14Þ in the cement paste with w/b ratio of 0.45. The higher binding
 H. Li et al. / Construction and Building Materials 185 (2018) 508–518 515

Friedels's salt Portlandite C A C3S C S capacity of the conventional concrete may be related to the avail-
3 2
ability of higher amount of water to intrigue the chemical reactions
between the cementing compounds and Cl- in seawater. Lollini
SW7
FW7
SW3
FW3 et al. [67] found that the existing alumina phases did not partici-
SW1
FW1 pate effectively in the formation of Friedel’s salt at the low w/b.
In this study, the free chloride concentration decreased from
SW7
1.75 mg/g to 1.46 mg/g and 1.33 mg/g, respectively, for SW3 (slag
FW7 25%) and SW7 (slag 50%). The higher alkalinity of intermixed sea-
water comparing to fresh water can enhance the reactivity of the
slag at the early ages and increase the reaction rate of Al phases
SW3
of slag with Cl- in the seawater [68]. The XRD patterns also con-
FW3 firmed the higher formation of Friedel’s salt in the matrix with
the presence of higher contents of slag (Fig. 8). The increased for-
mation of Friedel’s salt well elaborate the role of seawater and
SW1 the increased interaction of slag with cement.
Fig. 9 shows that SF also increased the chloride binding capacity
FW1
of the matrix. The free chloride concentration decreased to 1.62
mg/g and 1.46 mg/g for SW2 (SF 15%) and (SW3 30%) at the age
10 20 30 40 50 60 70 of 3 days, respectively. SF can increase the dissolution of C3S and
o
2 Theta( ) C3A [69] and facilitate the reactions as in Eqs. (14) and (15) and
contribute to higher chemical chloride binding capacity. The con-
Fig. 8. XRD patterns of control and slag containing pastes at 3 days.
tribution of SF to the chloride binding can also correspond to the
physical binding of the free chloride during the early C-S-H forma-
tion. However, the role of SF on the chloride binding was less than
2.0 slag.
Free chloride concentration(mg/g)

The role of seawater in the interaction of SF and slag with

cement was also studied in terms of C-S-H formation using FTIR.
Fig. 10 shows the deconvolution of the main band associated to
C-S-H over the range from 800 cm1 to 1200 cm1. The results
showed that seawater had a different effect on the C-S-H formation
for SF and slag containing specimens.
1.5 The band at around 810 cm1 is due to Si-O stretching vibration
at Q1 site as the main element of un-hydrated cement paste
[70,71]. The main band at around 970 cm1 is assigned to Si-O
stretching vibrations at Q2 site [70,71]. The Q2 tetrahedron indi-
cates a wide range of C-S-H (Ca/Si  2:1) and is the most common
tetrahedron in well hydrated cement paste. The two other bands at
1089 cm1 and 1135 cm1 are typical bands of Q3 and Q4 sites
1.0 assigned to C-S-H with higher SiO2 content [72,73]. As can be seen,
SW1 SW2 SW3 SW4 SW5 SW6 SW7 the absorption peak related to the Q4 is higher in SF containing
Number specimens than slag, which is due to effect of SF on the formation
of low Ca/Si C-S-H. For SF containing specimens, the pore solution
Fig. 9. Free chloride concentration of seawater mixed pastes. is rich in Si comparing to the Ca content, which leads to formation

2.5 2.5
FW2 FW3
SW2 a SW3 b
Q4
2.0 1135 2.0
Q2
CaCO3 962
1.5 1.5 973
Q1
Q2 Q3
831
962 1089
CaCO3
1.0 Q3 Q4
973 1.0
Q1 1089 1135
831
0.5
0.5

0.0
800 900 1000 1100 1200 0.0
800 900 1000 1100 1200
Wave number(cm-1) Wave number(cm-1)
Fig. 10. The FTIR results of SF and slag containing pastes.
516 H. Li et al. / Construction and Building Materials 185 (2018) 508–518

of low Ca/Si C-S-H. The formation of C-S-H with low Ca/Si was also formation of high Ca/Si C-S-H further densified the microstructure
observed in a study by Rossen et al. [54]. They reported that the Ca/ of slag containing specimens.
Si in the inner product C-S-H decreased at the early age as CH was The effect of seawater on the interaction of slag, SF and cement
consumed. After the complete consumption of CH, the Ca/Si in terms of C-S-H formation was also studied by the changes in the
dropped to a ratio as low as 1. Using seawater as the mixing water pore structure. The pore structure of fresh water and seawater
accelerated this process and the Ca/Si of the C-S-H was lower for mixed specimens with pure cement, slag (25%) and SF (15%) were
the SF specimens at the 3 days as seen from Fig. 10 (a). On the analyzed using NAD and the results are shown in Fig. 11. As seen,
other hand, the alkaline environment of the pore solution can the pore size distribution of slag containing specimens (SW3)
accelerate the dissolution of slag, and hence higher amount of Ca revealed lower pore volume at any pore diagram range. It can be
was released to the pore solution, which increased the Ca/Si of related to the formation of high Ca/Si C-S-H as well as formation
the C-S-H. This can well explain the higher compressive strength of Friedel’s salt as showed in Fig. 8. The densification of microstruc-
obtained for the slag containing specimens at 3 days. The ture due to formation of Friedel’s salt was also reported in a study

18 120
FW1 FW1
16 SW1 SW1
100
Adsorption volume(cc/g)

14
80

dV(cc/nm*g)
12

10
60
8

6 40
4
20
2

0 0
1 2 4 8 16 32 64 128 1 2 4 8 16 32 64 128
Pore Diameter (nm) Pore Diameter (nm)
20 120
FW2
FW2
18 SW2
SW2
Adsorption volume(cc/g)

100
16
dV(cc/nm*g)

14
80
12
10 60
8
40
6
4
20
2
0 0
1 2 4 8 16 32 64 128 1 2 4 8 16 32 64 128
Pore Diameter (nm) Pore Diameter (nm)
20 140
FW3
18 SW3 FW3
120 SW3
Adsorption volume(cc/g)

16
14 100
dV(cc/nm*g)

12
80
10
8 60
6 40
4
2 20

0 0
1 2 4 8 16 32 64 128 1 2 4 8 16 32 64 128
Pore Diameter (nm) Pore Diameter (nm)

Fig. 11. NAD results of pastes at 3 days.

 H. Li et al. / Construction and Building Materials 185 (2018) 508–518 517

by Yuan et al. [74]. On the contrary, with the incorporation of sea- is a great concern in the manufacture of UHPC. It also enhanced the
water in pure cement and specimens 15% SF containing, the densi- early age compressive strength of all specimens. However, the cor-
fication was in the form of refinement of the pores to smaller nano rosion potential of the steel fiber in the seawater mixed UHPC
capillaries. The control specimens (SW1) exhibited higher pore needs further attention.
fraction at the range below 4 nm as compared to FW1. However,
within the pore size range of 4 nm to 128 nm, the pore fraction Conflict of interests
decreased; the same trend was observed for the SF containing
specimens but in larger size pores. The shift in the pore size of The authors of this work have no competing interests.
the control and the specimens with 15% SF to the smaller pores
caused higher autogenous shrinkage at the early ages of hydration
Acknowledgments
(Fig. 5). However, results showed that seawater mixed specimens
with higher contents of SF had lower autogenous shrinkage com-
The financial support from the National Natural Science Foun-
paring to those mixed with fresh water. As discussed earlier, the
dation of China under project No. 51638008 is gratefully acknowl-
role of SF in the hydration kinetics was reduced with presence of
edged. The project was also supported by Key Laboratory of
seawater, which can be mostly related to the decreased pozzolanic
advanced Civil Engineering Materials (Tongji University), Ministry
behavior and nucleation effect of micro silica particles. The low-
of Education, China (No. 201703).
ered hydration kinetics can reduce the self-desiccation for the SF
containing seawater mixed specimens and contribute to restrained
References
autogenous shrinkage.
[1] J. Xiao, C. Qiang, A. Nanni, et al., Use of sea-sand and seawater in concrete
construction: current status and future opportunities, Constr. Build. Mater. 155
5. Conclusion (2017) 1101–1111.
[2] T. Nishida, N. Otsuki, H. Ohara, Z.M. Garba-Say, T. Nagata, Some considerations
for applicability of seawater as mixing water in concrete, J. Mater. Civ Eng
The role of seawater as mixing water in the interaction of slag (2013).
and SF with cement in very low water-to-binder ratio pastes at [3] T.U. Mohammed, H. Hamada, T. Yamaji, Performance of seawater-mixed
concrete in the tidal environment, Cem. Concr. Res. 34 (2004) 593–601.
the early age was studied. Results from the hydration kinetics, [4] N. Otsuki, T. Saito, Y. Tadokoro, Possibility of sea water as mixing water in
autogenous shrinkage and the compressive strength at the initial concrete, J. Civ. Eng. Archit. 6 (2012) 1273–1279.
72 h of hydration showed that the incorporation of seawater chan- [5] Z. Shi, Z. Shui, Q. Li, H. Geng, Combined effect of metakaolin and seawater on
performance and microstructures of concrete, Constr. Build. Mater. 74 (2015)
ged the interaction of slag and SF with cement which affected the
57–64.
early age properties of the matrix. The role of seawater on the [6] Q. Li, H. Geng, Y. Huang, et al., Chloride resistance of concrete with metakaolin
interaction of slag, SF and cement was investigated through their addition and seawater mixing: A comparative study, Constr. Build. Mater. 101
effects on hydration products, namely, CH and C-S-H as well as for- (2015) 184–192.
[7] G.L. Kalousek, E.J. Benton, Mechanism of seawater attack on cement pastes, ACI
mation of Friedel’s salt. Accordingly, the hydration kinetics and the Mater. J. (February 1970) 187–192.
chloride binding were studied, and the changes of FTIR spectra and [8] M. Regourd, Physico-chemical studies of cement pastes, mortars and concretes
pore structure were used to interpret the role of seawater in the exposed to seawater Performance of concrete in marine environment, ACI
Publ. 65 (1980) 63–82.
interaction of slag, SF with cement during the C-S-H formation. [9] M. Regourd, Chemical durability of concrete, Swedish Cement and Concrete
Seawater increased the role of slag in the hydration of the binder Research Institute, Contemporary European Concrete Research, 1982, pp. 121–
in terms of the final heat evolution and initiation of the acceleration 140.
[10] J.M. Sayward, Special Report 84–25 US Army Corps of Engineers, Salt Action on
peak. The obtained CH contents indicated that the slag containing Concrete, Hanover, New Hampshire, Cold regions Research & Engineering
mixtures consumed more CH during the hydration of cement com- Laboratory, 1984, p. 76.
pared to SF containing specimens when mixed with seawater. Also, [11] R.N. Swamy, M.M. Al-Asali, New test methods for alkali–silica reaction, in: P.E.
Grattan-Bellew (Ed.), Proceedings of the 7th International Conference on
higher chemical chloride binding was formed for the slag contain- Alkali-Aggregate Reactions in Concrete, Ottawa, Canada, 1987, pp. 324–329.
ing specimens with the presence of seawater. Results also showed [12] A. Shayan, A. Xu, G. Chirgwin, et al., Effects of seawater on AAR expansion of
that the reactivity of the SF particles decreased for seawater mixed concrete, Cem. Concr. Res. 40 (2010) 563–568.
[13] R. Victor, G. Baskir, J. Camp, et al., Report Card for America’s Infrastructure,
pastes. This decreased the negative effect of silica fume on autoge-
Reston Va Am. Soc. Civil Eng. (2013).
nous shrinkage of low water-to-binder paste in ternary mixtures [14] G.H. Koch, M.P.H. Brongers, N.G. Thompson, et al., Corrosion cost and
especially when higher contents of SF were incorporated. preventive strategies in the united states, J. Endocrinol. 122 (2002) 23–31.
Seawater increased the role of cement and slag in the strength [15] T.U. Mohammed, H. Hamada, T. Yamaji, Performance of seawater-mixed
concrete in the tidal environment, Cem. Concr. Res. 34 (2004) 593–601.
development. Seawater also showed to affect the C-S-H formation [16] Z.L. Chen, X.N. Tang, G.F. Sun, Y.L. Liu, Research on durability and application of
for slag and SF containing specimens differently. Seawater seawater concrete, Ocean Eng. 26 (4) (2008) 102–106 (in Chinese).
increased the formation of low Ca/Si C-S-H for specimens with [17] M. Makita, Y. Mori, K. Katawaki, Marine corrosion behaviour of reinforced
concrete exposed at Tokyo bay, Am. Concr. Inst. Spec. Publ. 65 (1980) 271–289.
SF, while it increased C-S-H gel with higher Ca/Si for slag contain- [18] R. Shalon, M. Rapheal, Influence of sea water on corrosion of reinforcement,
ing mixture. This led to denser microstructure of slag containing Am. Concr. Inst. 55 (6) (1959) 1251–1268.
specimens in which nano-capillary decreased at all sizes. Forma- [19] IS 456–2000, Plain and Reinforced Concrete – Code of Practice, Indian Stand,
2000.
tion of Friedel’s salt also helped to densify the matrix. Overall, [20] EHE, Instrucción Del, Hormigón Estructural (Spanish Concrete Structural
the densification helped to decrease the autogenous shrinkage Code), 2008.
and increased the compressive strength. As for control and speci- [21] S. Pyo, S. El-Tawil, A.E. Naaman, Direct tensile behavior of ultra high
performance fiber reinforced concrete (UHP-FRC) at high strain rates, Cem.
mens with low contents of SF, the densification was in the form Concr. Res. 88 (2016) 144–156.
of refinement in which there was a shift in the pore size to smaller [22] R. Yu, P. Spiesz, H.J.H. Brouwers, Effect of nano-silica on the hydration and
nano-capillaries, which induced autogenous shrinkage. On the con- microstructure development of Ultra-High Performance Concrete (UHPC) with
a low binder amount, Constr. Build. Mater. 65 (2014) 140–150.
trary, autogenous shrinkage was restrained for specimens with
[23] A. Tafraoui, G. Escadeillas, T. Vidal, Durability of the Ultra High Performances
higher contents of SF as well as all ternary blend mixtures as the Concrete containing metakaolin, Constr. Build. Mater. 112 (2016) 980–987.
interaction of SF with cement was reduced. [24] D. Wang, C. Shi, Z. Wu, J. Xiao, Z. Huang, Z. Fang, A review on ultra high
In general, findings of the study showed that using seawater as performance concrete: Part II. Hydration, microstructure and properties,
Constr. Build. Mater. 96 (2015) 368–377.
the mixing water changed the interaction of SF and slag with [25] L. Wu, N. Farzadnia, C. Shi, et al., Autogenous shrinkage of high performance
cement in favor of hydration and the autogenous shrinkage, which concrete: A review, Constr. Build. Mater. 149 (2017) 62–75.
518 H. Li et al. / Construction and Building Materials 185 (2018) 508–518

[26] O. Mejlhede Jensen, P. Freiesleben Hansen, Autogenous deformation and [50] J. Geng, D. Easterbrook, Q. Liu, L. Li, Effect of carbonation on release of bound
change of the relative humidity in silica fume-modified cement paste, ACI chlorides in chloride contaminated concrete, Mag. Concr. Res. 1–11 (2015).
Mater. J. 93 (6) (1996) 539–543. [51] H.C. Vollmer, Calcium chloride in concrete, Bibliography 13, Highway Research
[27] E. Ghafari, S.A. Ghahari, H. Costa, E. Júlio, A. Portugal, L. Durães, Effect of Board, Washington, D.C, 1952, p. 58.
supplementary cementitious materials on autogenous shrinkage of ultra-high [52] M. Juenger, P. Monteiro, E.M. Gartner, G.P. Denbeaux, A soft X-ray microscope
performance concrete, Constr. Build. Mater. 127 (2016) 43–48. investigation into the effects of calcium chloride on tricalcium silicate
[28] Y. Yang, R. Sato, K. Kawai, Autogenous shrinkage of high-strength concrete hydration, Cem. Concr. Res. 35 (2005) 19–25.
containing silica fume under drying at early ages, Cem. Concr. Res. 35 (3) [53] V.K. Peterson, M. Juenger, Hydration of tricalcium silicate: effects of CaCl2 and
(2005) 449–456. sucrose on reaction kinetics and product formation, Chem. Mater. 18 (2006)
[29] K.M. Lee, H.K. Lee, S.H. Lee, G.Y. Kim, Autogenous shrinkage of concrete 5798–5804.
containing granulated blast-furnace slag, Cem. Concr. Res. 36 (7) (2006) 1279– [54] J.E. Rossen, B. Lothenbach, K.L. Scrivener, Composition of C-S-H in pastes with
1285. increasing levels of silica fume addition, Cem. Concr. Res. 75 (2015) 14–22.
[30] M.H. Wadsworth, Handbook of statistical methods for engineers and scientists, [55] H.N. Stein, J.M. Stevels, Influence of silica on the hydration of 3CaOSiO2, J.
McGraw-Hill, New York, 1990. Appl. Chem. 14 (1964) 338–346.
[31] X. Hu, C. Shi, Q. Yuan, et al., Influences of chloride immersion on zeta potential [56] H.M. Jennings, P.L. Pratt, An experimental argument for the existence of a
and chloride concentration index of cement-based materials, Cem. Concr. Res. protective membrane surrounding portland cement during the induction
106 (2018) 49–56. period, Cem. Concr. Res. 9 (1979) 501–506.
[32] C. Shi, D. Wang, L. Wu, et al., The hydration and microstructure of ultra high- [57] W. Chen, Hydration of slag cement: theory, modeling and application. PhD
strength concrete with cement–silica fume–slag binder, Cem. Concr. Compos. thesis, University of Twente, 2007.
61 (2015) 44–52. [58] A.K. Suryavanshi, J.D. Scantlebury, S.B. Lyon, Mechanism of Friedel’s salt
[33] D. Jansen, J. Neubauer, F. Goetz-Neunhoeffer, R. Haerzschel, W.D. Hergeth, formation in cements rich in tri-calcium aluminate, Cem. Concr. Res. 26 (5)
Change in reaction kinetics of a Portland cement caused by a superplasticizer— (1996) 717–727.
calculation of heat flow curves from XRD data, Cem. Concr. Res. 42 (2012) [59] C. Arya, Y. Xu, Effect of cement type on chloride binding and corrosion of steel
327–332. in concrete, Cem. Concr. Res. 25 (1995) 893–902.
[34] J. Zelić, D. Rušić, D. Veza, et al., The role of silica fume in the kinetics and [60] H. Hirao, K. Yamada, H. Takahashi, H., Zibara Chloride binding of cement
mechanisms during the early stage of cement hydration, Cem. Concr. Res. 30 estimated by binding isotherms of hydrates, J. Adv. Concr. Technol. 3 (1)
(10) (2000) 1655–1662. (2005) 77–84.
[35] K. Scrivener, B. Lothenbach, N. De Belie, et al., TC 238-SCM: hydration and [61] M. Thomas, R.D. Hooton, A. Scott, H. Zibara, The effect of supplementary
microstructure of concrete with SCMs State of the art on methods to cementitious materials on chloride binding in hardened cement paste, Cem.
determine degree of reaction of SCMs, Mater. Struct. 48 (4) (2015) 835–862. Concr. Res. 42 (2012) 1–7.
[36] C. Shi, R.L. Day, A calorimetric study of early hydration of alkali-slag cements, [62] H. Justnes, A review of chloride binding in cementitious systems, Nordic Concr.
Cem. Concr. Res. 25 (6) (1995) 1333–1346. Res. 21 (1997).
[37] T. Knudsen, 7th ICCC, vol. II, Parigi-Paris, 1980, 170–175. [63] A.M. Neville, Properties of Concrete: Fourth and Final Edition, fourth ed.,
[38] M.S. Meddah, A. Tagnit-Hamou, Pore structure of concrete with mineral Wiley, 1996.
admixtures and its effect on self-desiccation shrinkage, Aci Mater. J. 106 (3) [64] D.R.M. Brew, F.P. Glasser, Synthesis and characterization of magnesium silicate
(2009) 241–250. hydrate gels, Cem. Concr. Res. 35 (2005) 85–98.
[39] H. Wee, N. Lim, Autogenous shrinkage of ground-granulated blast-furnace slag [65] M. Balonis, B. Lothenbach, G. Le Saout, F.P. Glasser, Impact of chloride on the
concrete, Aci Mater. J. 97 (5) (2000) 587–593. mineralogy of hydrated Portland cement systems, Cem. Concr. Res. 40 (2010)
[40] P. Lura, O.M. Jensen, K. Breugel, Autogenous shrinkage in high performance 1009–1022.
cement paste: an evaluation of basic mechanisms, Cem. Concr. Res. 33 (2) [66] F.P. Glasser, A. Kindness, S.A. Stronach, Stability and solubility relationships in
(2003) 223–232. AFm phases Part I. Chloride, sulfate and hydroxide, Cem. Concr. Res. 29 (1999)
[41] I. Maruyama, A. Teramoto, Temperature dependence of autogenous shrinkage 861–866.
of silica fume cement pastes with a very low water–binder ratio, Cem. Concr. [67] F. Lollini, E. Redaelli, L. Bertolini, Investigation on the effect of supplementary
Res. 50 (2013) 41–50. cementitious materials on the critical chloride threshold of steel in concrete,
[42] T. Oertel, F. Hutter, R. Tanzer, U. Helbig, G. Sextl, Primary particle size and Mater. Struct. 49 (2016) 4147–4165.
agglomerate size effects of amorphous silica in ultra-high performance [68] D. Furuya, N. Otsuki, T. Saito, A. Lee Yun, Study on the effects of seawater as
concrete, Cem. Concr. Comp. 37 (2013) 61–67. mixing water on the hydration characteristics of blast-furnace slag cement, in:
[43] J.J. Thomas, H.M. Jennings, J.J. Chen, Influence of nucleation seeding on the 34th Conference on Our World In Concrete & Structure: 16-18 August 2009,
hydration mechanisms of tricalciumsilicate and cement, J. Phys. Chem. C 113 Singapore.
(2009) 4327–4334. [69] H. Cheng-yi, R.F. Feldman, Hydration reactions in Portland cement–silica fume
[44] D.M. Roy, G.M. Idorn, Hydration, structure, and properties of blast furnace slag blends, Cem. Concr. Res. 15 (1985) 585–592.
cements, mortars, and concrete, ACI J. (1982) 444–457. [70] P. Yu, R.J. Kirkpatrick, B. Poe, P.F. McMillan, X. Cong, Structure of calcium
[45] K. Renichi, S. Chong Tak, G. Seishi, D. Masaki, The latent hydraulic property of silicatehydrate (C-S-H): near-, mid-, and far-infrared spectroscopy, J. Am.
granulated blast furnace slag by various activators, Tetsuto-Hagane 65 (13) Ceram. Soc. 82 (1999) 742–748.
(1979) 1825–1829. [71] I. García-Lodeiro, A. Fernández-Jiménez, M.T. Blanco, A. Palomo, FTIR study of
[46] P. Ghoddousi, L.A. Saadabadi, Study on hydration products by electrical the sol–gel synthesis of cementitious gels: C-S–H and N–A–S–H, J. Sol-Gel Sci.
resistivity for self-compacting concrete with silica fume and metakaolin, Technol. 45 (2008) 63–72.
Constr. Build. Mater. 154 (2017) 219–228. [72] I.G. Lodeiro, D. Macphee, A. Palomo, A. Fernández-Jiménez, Effect of alkalis on
[47] J. Skalny, J. Maycock, Mechanisms of acceleration by calcium chloride: a fresh C-S–H gels. FTIR analysis, Cem. Concr. Res. 39 (2009) 147–153.
review, J. Test. Eval. 3 (4) (1975) 303–311. [73] S.-Y. Hong, F. Glasser, Alkali binding in cement pastes: Part I. The CSH phase,
[48] K.L. Scrivener, P. Juilland, P.J. Monteiro, Advances in understanding hydration Cem. Concr. Res. 29 (1999) 1893–1903.
of Portland cement, Cem. Concr. Res. 78 (2015) 38–56. [74] Q. Yuan, C. Shi, G.D. Schutter, K. Audenaert, D. Deng, Chloride binding of
[49] Y. Lu, G. Shi, Y. Liu, et al., Study on the effect of chloride ion on the early age cement-based materials subjected to external chloride environment – A
hydration process of concrete by a non-contact monitoring method, Constr. review, Constr. Build. Mater. 23 (2009) 1–13.
Build. Mater. 172 (2018) 499–508.