Construction and Building Materials 155 (2017) 413–426

Contents lists available at ScienceDirect

Construction and Building Materials

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

Hydration for the Alite mineral: Morphology evolution, reaction

mechanism and the compositional influences
Zhang Lina a, Cheng Xin a, Hou Dongshuai b,⇑, Guo Siyao b
a
University of Jinan, No. 336 Nanxinzhuang West Road, Shandong 250022, China
b
Qingdao University of Technology, Qingdao, Shandong 266033, China

h i g h l i g h t s g r a p h i c a l a b s t r a c t

 C-S-H gel showed morphology of foil- Morphology of the calcium silicate hydrate from micro-meter to nano-meter.
like wrinkled clusters.
 The C3S hydration product possessed
poor crystallined feature.
2+
 Ca ions restrict the silicate
polymerization during hydration.
 Mean silicate chain length decreases
with increasing Ca/Si ratio.
 The C-S-H gel varies from branch
structure to ellipsoid cluster.

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

Article history: Hydration of minerals, playing fundamental role in many technological applications, involves multi-stage
Received 30 May 2016 chemical reactions, morphology changes and compositional variations. For the first time, the hydration
Received in revised form 9 July 2017 mechanism for the tricalcium silicate was investigated by the combination of C3S hydration experimen-
Accepted 16 August 2017
tally and polymerization of silicate monomers via molecular dynamics simulation. Tricalcium silicate was
Available online 23 September 2017
self-synthesized and utilized to hydrate with water at the water/solid ratio of 50. Samples hydrated from
one day to one year were obtained and characterized by XRD, IR, SEM, NMR and TEM techniques. Based
Keywords:
on SEM observation, the morphology of hydrated C3S transformed from small localized to foil-like wrin-
Hydration
Polymerization
kled clusters with fibrillar-like wrinkles distributed randomly. Wrinkled product of C-S-H possessed
Molecular dynamics poor-crystallined or amorphous feature based on XRD and TEM analysis. Such was also evidenced by
Branched cluster the fact that no layered crystal phase was observed in simulated calcium silicate gel clusters. Besides,
the simulated pair distribution function of C-S-H gel demonstrating the structure feature of short range
order and long range disorder, also confirmed glassy structural nature for the C-S-H gel. Furthermore,
based on NMR and SEM results C/S ratio changed to 1.2 and the mean silicate chain length in C-S-H
was obtained at around 4 after hydration for one year with Q1 and Q2 species representing linear silicate
chain structure present dominantly. The molecular dynamics helped illuminate the hydration mecha-
nism that the morphology of C-S-H gel was closely related with C/S ratio. At C/S ratio of 1, solid branched
3-dimmensional cluster was segregated from the solution while ellipsoid colloid particle with the size of
4 nm  4 nm  8 nm was formed at C/S ratio of 1.75. It was also found that calcium ions prevented the

⇑ Corresponding author.
E-mail address: dshou@outlook.com (D. Hou).

http://dx.doi.org/10.1016/j.conbuildmat.2017.08.088
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
414 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426
polymerization reaction of silicate monomers and silicate chains were shortened significantly at high Ca/
Si ratio, amplifying the predominate percentage of Q1 species in the C-S-H gel.
1. Introduction
Hydration reactions of minerals have been intensively investi-
gated in many natural and industrial processes. The formation of
hydration productions, modifying the local chemical and physical
environment, for example, can improve the colonization by micro-
bial communities in the soil sediment formation [1], stabilize the
detrimental ions in radioactive waste repositories [2], play catalyz-
ing role in water-oxidation reactions [3], enhance the microstruc-
ture and durability of the marine concrete structures [4] and so
forth. One important example is the dissolution of Alite mineral
to recombine and form in cement paste that is ubiquitously uti-
lized as the construction materials. Meanwhile, considering 5%
emission of global green housing gases due to the manufacture
of cement production, investigation on the Alite mineral, involving
the hydration mechanism and processing improvement, is of cru-
cial importance to the environmental sustainability development.
The reaction of Alite mineral produces Calcium silicate hydrate
(C-S-H), which makes dominant contribution to mechanical prop-
erties of cement-based materials with its great significance empha-
sized. However, after decades of research and study, the insights of
C3S hydration were continuously developing [5] and there were
still some fundamental questions about C-S-H to be explicated
[6]. Such was due to the complex components of cement as well
as the intrinsic complicated structure and properties of C-S-H as
gel product. Large amount of studies have been conducted based
on cement hydration [7–10]. But such was not that easy to grasp
the nature of calcium silicate hydrate in a multi-complex reacting
system. Consequently, the hydration studies based on C3S and C2S
were carried out by many researchers [11–13]. As was widely
accepted, the reactivity of C3S was higher than that of C2S [14].
Hence, it was more effective and reasonable to study C-S-H from
the hydration of C3S.
Up till now, some valuable results have been acquired for the
hydration evolution of C3S and C-S-H produced. Considering the
hydration mechanism of C3S at early period of hydration, some
theoretical studies have been released. Juilland and co-authors
proposed the dissolution theory of C3S. They emphasized the role
of saturation state of the solution and proposed the dominate dis-
solution control transited from etch pits formation initially to step
retreats of such etch pits later. And this was the onset of induction
period mechanism as well [15]. Fernandez conducted particle size
effect of C3S to the hydration kinetics and formation of C-S-H
microstructure [16]. Based on her study, the hydration kinetics
was closely affected by the size of C3S particles. The smaller ones
resulted in higher dissolution peak, more heat release as well as
earlier nucleation and growing process. On the contrary, there
did not exhibit much difference on the microstructure formed,
amplifying the independency between particle size and
microstructure. Research on middle stage of C3S hydration was also
carried out. Kjellsen and Lagerblad compared hydration product of
cement with C3S, in which Hadley grain phenomenon could be
observed. [17] In terms of structure characterization, various
experiments were carried out by XRD, NMR, Raman, SEM, XPS,
thermal analysis and so forth [18–21] Moreover, there were also
quite amount of studies carried out on the influence of C3S hydra-
tion with other components or elements incorporated [22–24].
In addition to experimental efforts, computational methods,
such as molecular dynamic, played supplement role in understand-
Ó 2017 Elsevier Ltd. All rights reserved.
ing experimental findings and provided molecular insights on the
hydration of minerals. The C-S-H gel investigation has made great
progress with the assistance of such computational method. The
simulation of silicate sol-gel reaction has been developed by Feus-
ton and Garofalini [25] for more than 20 years. They first simulated
the oligomerization in silica sols by modified BMH potential and
RSL2 potential. The basic methodology was that randomly dis-
tributed silicic monomers and water molecules were heated to
2500 K, thus the reaction rate could be accelerated. Although the
silicate sol-gel simulation has been developed for quite a long time,
the introduction of sol-gel reaction to C-S-H gel is recent-year’s
effort. Dolado first simulated the silicate sol-gel reaction with the
existence of calcium ions [26]. Polymerization of silicic acid (Si
(OH)4) in calcium-riched region was taken as the process of cemen-
titious hydration. Their study demonstrated that the addition of cal-
cium ions slowed down the polymerization process of silica.
Additionally, by counting SiAOH and CaAOH bonds, C-S-H gel
was showed similarity with tobermorite and jennite at different
Ca/Si(C/S) ratio. Furthermore, based on the model constructed, they
simulated the hydration process with the existence of aluminum
atoms which indicated aluminum elongated the silicate chain and
contributed to the 3-dimensional silicate-aluminum structures.
Currently, 3  3  6 nm calcium silicate segmental branches have
been discovered when the simulation system was enlarged to
nearly 9 nm in one dimension after running for 2.3 ns [27].
However, most of the studies by experiment and computation
up to date were mainly focused on the early stage of the hydration
and the water to solid ratios (w/s) mainly equaled to smaller values
from 0.2 to 1. There were quite a few studies designed for high w/s
ratio in long-term hydration. HAGA conducted quite high w/s ratio
from 10 to 2000 for the hydration understanding of Ordinary Port-
land Cement [28]. The aim of designing higher w/s ratio was to
help promote full hydration of C3S as well as induce the separation
of calcium hydroxide and C-S-H which both incorporated with
each other and could not be easily distinguished to some extent
[29]. To understand the hydration process of C3S and acquire C-
S-H as single phase as possible to answer some basic questions
of C-S-H, including composition, crystal phase, structure and so
forth, higher w/s ratio was utilized for the hydration of C3S.
In this paper, the hydration of C3S at high w/s ratio of 50 was
conducted from one day to one year to follow the structure evolu-
tion of C3S and C-S-H produced. Synthesized C3S was utilized as the
reactant. Hydrated product was characterized by various tech-
niques to trace the hydration process of C3S as well as the proper-
ties of C-S-H. Furthermore, to better understand the hydration
mechanism, under the environment of high water concentration,
the silicic polymerization with different amounts of calcium ions
introduced was simulated by performing molecular dynamics
method. The silicate morphology, the influence of C/S ratio and
the reaction mechanism has been analyzed in accordance with
the atomic trajectory.
2. Methodology
2.1. Experimental section
2.1.1. Material
C3S was self-synthesized by traditional method. CaCO3 (A.R.,
Alfa Aesar) and SiO2 (4N, Sinopharm Chemical Reagent Co. Ltd)
 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426 415

Table 1 were first grinded (molar ratio of 3) to pass 63 lm sieves and

Information of the simulation box. evenly mixed for two hours. The obtained fine powder was com-
C/S Number of atoms Size (Å) pressed into pancakes and put into Pt crucible for calcinations at
Ca Si O H 1500 °C for five hours and then quickly cooled down to ambient
temperature within 5 min and subsequently grinded into fine pow-
1 3375 3375 33,750 47,250 105
1.25 4096 5102 47,104 67,584 112.96
der, compressed and recalcined. After repeating the above process
1.5 2662 3993 34,606 50,578 100.1 for 4–5 times, pure C3S powder was attained and characterized
1.75 2401 1372 29,498 48,706 97 (Appendix). Ultra pure water (U.P. water, resistivity >18.2 MX
cm)

Fig. 1. SEM images for hydrate C3S from 1 d to 365 d (a) 1 d (b) 3 d (c) 7 d (d) 14 d (e) 28 d (f) 180 d (g) 365 d (h) Enlargement of product of (g).
416 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426
was utilized throughout the whole experiment conducted. PVC
bottles were washed, dried and sealed.
2.1.2. Apparatus
Power XRD (X’pert Pro) was carried out with all the samples
grinded in agate thoroughly, transferred into Al holder and then
flatted. The 2h was scanned from 5° to 80° at the scan rate of 2°/
min. FT-IR (Bruker Vertex 70) was carried out with KBr as refer-
ence. Grinded powder was mixed with KBr and then pressed into
transparent plate for the tests. SEM (JEOL6300, JEOL6700) was car-
ried out with the product coated with Au. 29Si NMR (Advance III
900 WB with 3.2 mm MAS probe head) tests were carried out at
the rotation speed of 10 kHz. The scans of 640 were accumulated
and 120 s delay was used. TEM (JEOL2010F) was conducted on cop-
per grid with amorphous carbon layer. EDS (X-Max 80 mm2,
Oxford instruments) was utilized for composition analysis.
2.1.3. Hydration process
Weighed C3S were mixed with UP water at w/s ratio of 50, and
then sealed in PVC bottle for 1 day (d), 3 d, 7 d, 14 d, 28 d, 180 d
and 365 d, respectively. Vacuum filtration was subsequently car-
ried out quickly to separate solid with liquid, followed by fast (less
than 10 min) drying of solid product in 80 °C vacuum oven with
the powder attained for further characterization.
2.2. Simulation methodology
The hydration of tricalcium silicate is a multiple step-coupled
chemical reaction, challenging the computational modeling of
cement hydrate system. This study focused on sol-gel reaction of
silicate monomers in the presence of calcium ions in water solu-
tion. FG model, initially developed for the silica material [25],
was employed herein for sol-gel molecular dynamics simulation,
which contains three interaction potentials: Born-Mayer-Huggins
(BMH) potential, Stillinger-Weber (SW) potential and Rahman-
Stillinger-Lemberg (RSL2) potential. Besides, the interaction
parameters among Ca and Si, O, H were obtained from the previous
study on C-S-H (I) phases [30].
It should be noted that the sol-gel reaction of silicate monomers
were quite slow due to the high energy barrier for the reaction. In
order to model the time-consuming reaction in a nanosecond scale,
high temperature (1800 K  2500 K) simulation environment had
been preferred to accelerate the reaction rate. Polymerization at
room temperature is quite low, so most of the researchers investi-
gated silicate polymerization by MD method at high temperature
run. Garofalini and Martin [31] studied the polymerization of the
silicic acid molecules and network formation from 300 K to
2500 K and found that the high temperature increased the poly-
merization rate significantly. More importantly, the reaction path-
way for the high temperature reaction remained same as observed
in the room temperature. The dimerization reaction pathway sim-
ulated at high temperature is consistent with the ab initio calcula-
tion result that water is removed from the five-coordinate silicon
species during the polymerization process [32]. The produced Qn
species in the reaction products matches well with the NMR exper-
imental data. In current simulation, 1800 K was adopted as the
simulation temperature to contribute the silicic polymerization.
MD simulations were performed to study the evolution of cal-
cium silicate cluster with C/S ratio varying from 1.0 to 1.75. The
initial configuration was obtained by adding Si(OH)4, Ca(OH)2
and water molecules randomly in a box with the size of
100 Å  100 Å  100 Å, to achieve the overall density of about
1.2 g/cm3. The detailed information of the simulation systems
was listed in Table 1. Different from previous simulation by Dolado
and coworkers [26], high H2O/Si ratio allows for the complete dis-
solution of calcium ions and silicate monomers.
The simulation process was implemented in the following pro-
cedure. Initial configurations were relaxed at 300 K for 100 ps to let
the atoms diffuse to random positions. Aiming at increasing the
reaction rate, the system was heated for 100 ps, linearly increasing
the temperature to 1800 K. The reactions were implemented at
1800 K for 10,000 ps, followed by the linearly decreasing tempera-
ture to 300 K in 1000 ps. Finally, the systems were maintained at
300 K and further equilibrated for 50 ps before long production
run for 500 ps. Atom coordination information was recorded every
1 ps for later analysis. A time step of 1  10 3 ps was utilized.
3. Results and discussion
3.1. Experiment
3.1.1. Morphology evolution
Fig. 1 illustrated the morphology of hydrated C3S varying from
1 d to 365 d. Only after 24 h (Fig. 1(a)), the surface of C3S became
rough with some regular-shaped product, including rectangular,
round and fibrillars covered on the top. Such was characterized
first by TEM-EDS, which was shown in Table 2 that the dominant
element was calcium and oxygen, indicating their nature of CH
related. Consequently, instead of spalling from the initial C3S par-
ticles, the C-S-H produced stayed with the original C3S particles.
From such point of view, the morphology variation of C-S-H could
be described as follows. At one day the surface of reacting particles
was only a little bit rough compared with unreacted one. After
hydration for 3 d the entire surface covered distinct wrinkles
which were distributed evenly. After 14 d, the morphology of C3S
particles was significantly changed. A closely round shape became
clearly seen and the extent of the fold was greatly increased. Such
round-shaped of the particles were even easier to be observed with
the folds in the product greatly enhanced. After one year, the over-
all product was performed in relatively uniform folded and wrin-
kled clusters with the diameter less than 10 lm packed
randomly together. Compared with the folds at the early stage of
the hydration, such wrinkles at one year was much sharper and
clearer with the contour of the pleat easier to be distinguished
(Fig. 1(h)). Similar morphology could be observed under TEM
shown in Fig. 2 with large cluster possessing sharp wrinkles clearly
shown. Meanwhile, obscure ring was obtained, implying the ill-
crystallined properties of C-S-H. By overviewing the morphology
variation, the hydration seemed to react from outer to inner with
edge gradually curved, resulting in the evolution of C3S particles
to packed porous clusters of C-S-H.
As is known, drying method could influence the morphology
and properties of the C-S-H attained [20,33,34]. The drying manner
utilized herein was direct drying. The morphology results coin-
cided with those obtained by Fonseca and Jennings [20]. At
early-age of hydrated C3S, when subjected to rapid drying rates,
C-S-H was attained as fibrillar while very slowly drying was con-
ducted, it was less distinct. As a consequence, early-age wrinkled
Table 2
Composition of differed morphologies of CH.
Type Rectangular Ball Fibrillar
Element C Si Si
A.W.% 20.86 0.96 2.67
Element Si Ca Ca
A.W.% 0.72 67.54 40.90
Element Ca O O
A.W.% 24.73 31.50 53.02
Element O Al
A.W.% 53.69 3.41
Where A.W.: atomic weight.
 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426 417

Fig. 2. TEM image, localized structure and corresponding diffraction pattern attained for hydrated C3S product at 365 d (a) overview of hydrated C3S particle (b) enlargement
of localized area (c) diffraction pattern of (b).

C-S-H obtained in Fig. 1 could be mostly attributed to the rapid
drying of the samples. But coming to long-term of hydration, it
has been previously reported that a fibrillar product might natu-
rally form over time at 21 d. Similarly, it could be seen from
3.0
0.6
Rate of C/S/d-1
Fig. 1(e–g) as well as Fig. 2 that after 28 d, along with the incre-
ment of the duration of hydration, fibrillar structures were present
and obvious, which to some extent might be the natural formation
of C-S-H, which could be explained by whisker growth theory at
very high relative humidity. More protective drying method was
essential to be conducted in the future to figure out the natural
state of C-S-H after one year’s hydration, such as solvent exchange
and supercritical drying [35]. Concerning the pores formed, appar-
ently with the proceeding of hydration reaction, hydrated C3S was
shown to be more porous. As was reported by Zhang and Scherer,

2.5 0.4 different drying pathways could induce the discrepancy of the
porosity at micro and nano scale [33]. Hence, the specific porosity
C/S ratio

0.2 and distribution was not mentioned. But the trend of the transfor-
2.0 mation of solid particle initially to loose and porous product with
0.0 the growing degree of hydration could be gained roughly.
0 50 100 150 200 250 300 350
The C/S variation of the hydrated particles along with the
1.5 Duration/d
hydration process was measured by SEM/EDX with the results
shown in Fig. 3. The basic trend of C/S evolution performed contin-
1.0 uous decrement. The ratio of C/S dropped quickly from 3 to about
2.4 within 1 d, indicating the fast migration of CH out of reactant
0 50 100 150 200 250 300 350 particles. At the very beginning, CH was required to achieve its
own dissolution equilibrium with surroundings, and thus easily
Time/d
to migrate, resulting large amount of CH produced as well as
Fig. 3. C/S variation of hydrated C3S with duration of hydration. decrease of C/S. From 1 d to 3 d, the rate of such variation along

Fig. 4. Illustration of Qn sites (Red: oxygen, yellow: silicon). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this
article.)
418 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426

with the time was steeply dropped to very slow rate. Starting from
3 d to 28 d, the drop of C/S variation was kept plateau to around
1.4. From 28 d to 365 d, C/S was decreased to around 1.2 in an
extremely slow rate. The decreasing rate at longer period of hydra-
tion on one hand was due to C-S-H surrounded as barrier for the
income of water molecules. On the other hand, the removal of
CH from inner to outer surroundings became more difficult. And
therefore, the rate of such reactions along with C/S variation was
lagged in accompany with the prolonging of hydration.
3.1.2. NMR
In silicate composite structure, the connectivity factor Qn was
an important parameter to estimate the silicate connection. There
were five types of silicate connection, which could be denoted as
Qn, where n represented one silicon atom bonded to n other sili-
cons via SiAOASi bonds, the illustration of which was shown in
Fig. 4. Q0 was the monomer; Q1 represented the dimmer structure
60 70 80 90 100
120
f Fig. 6. Q0 exhibited monotonous decreasing from 100% to around
80
M
(two connected silicate tetrahedrons); Q2 was the long chain; Q3
was the branch structure; Q4 was the network structure [25]. The
chemical shift mentioned for different types of silicate sites varied.
Theoretically five categories of Qn sites possessed the chemical
shift of (centered at) 70 ppm (Q0), 80 ppm (Q1), 88 ppm
(Q2), 98 ppm (Q3), and 110 ppm (Q4) [36]. 29Si spectra obtained
for hydration product at 1 d, 7 d, 14 d, 28 d, 180 d and 365 d were
shown in Fig. 5. The chemical shift centered at around 70 ppm
representing Q0 was clearly observed that was originated from
C3S. Besides, there were two peaks accompanied with the occur-
rence of hydration appeared with the chemical shift centered at
about 80 ppm and 88 ppm corresponding to Q1 and Q2 silicate
sites respectively. Such appearance of Q1 and Q2 validated the
on-going polymerization of silicate structure in hydrated C3S. No
Q3 and Q4 sites appeared during the hydration process, indicating
the formation of short and linear silicate chains dominantly with-
out network structure present in hydration product. Such observa-
tion of silicate sites Q0, Q1 and Q2 were in good accordance with IR
results (Appendix). By quantitative analysis, the variation of sili-
110 cate sites along with the duration of hydration was illustrated in
14% after 365 d, illuminating the continuous consumption of C3S
during hydration and the large hydration degree. Certain amount

of Q0 remained, indicating the hydration degree did not reach

40 100%, which was in consistency with the result gained by XRD.
Q2 exhibited a monotonous increase in proportion, indicating the
0
120 polymerization of long chains lasting. The proportion of Q2 finally
e achieved at 42.2% at 365 d. Q1 increased up to 11.6% within the ini-
tial 14 d, decreased to about 6% at 28 d and bounced back contin-
80 uously to 43.7% at 365 d, illustrating the existence of consumption
K

period for Q1. The rate of Q0 varying along with the hydration was
40 also included in Fig. 6. A sudden jump was performed on Q0 at 24 h
but such rate quickly went down, which became gradually slower
120
0 with the prolonging of the hydration, indicating the fast reaction at
d the initial and the obstruction in the longer time of the hydration.
80
3.2. The network and ellipsolid particles for the C-S-H gel
I

40
During around 10 ns of sol-gel reaction in solution, calcium sil-
icate clusters with different morphologies were segregated from
0
120 water solution. The hydration reaction was chemical liquid-solid
c
separation process, resulting in the solid precipitation from solu-
80 tion. As shown in Fig. 7(a) and (b), when the C/S ratio was smaller
than 1.5, the calcium silicate cluster grew across the simulation
G

40
Q0
7
Rate of Q0 variation/d -1

0
120 100 6 Q1
b
5 Q2
80 4
80 3
E

2
40 1
60 0
Qn/%

0 50 100 150 200 250 300 350

120
0 Duration/d
a
40
80
B

20
40

0
0
60 70 80 90 100 110 0 50 100 150 200 250 300 350
Frequency (PPPM) Duration/d
Fig. 5. NMR spectra for hydrated C3S at varied duration of hydration (a) 1 d (b) 7 d Fig. 6. Qn variations along with duration of hydration (Insert: rate of Q0 variation
(c) 14 d (d) 28 d (e) 180 d (f) 365 d. along with duration of hydration).
 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426 419

Fig. 7. (a) The final configuration of the C-S-H gel at different C/S ratio in
10 nm  10 nm  10 nm box; (b) 2  2  2 super-cell of the final configuration
(Green: calcium atoms, yellow: silicon atoms). (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of this
article.)

box with branch size larger than 3.5 nm. When the simulation box
was periodically extended, honeycomb-like network and gel pores
filled with water could be observed. More interestingly, the net-
work phase, growing in radiated style, demonstrated amorphous
feature similar to the silica glass achieved from sol-gel reaction
[31].
On the other hand, the cluster with C/S ratio equaling to 1.75
nucleated and developed to an ellipsoid particle with the diameter
approximately 5 nm in Fig. 7. At high C/S ratio, due to long simula-
tion time, the hydration degree was supposed to be higher than
that of Dolado’s work [27], which could transform to an egg-like
particle. Compared with the amorphous network structure, the
Fig. 8. (a) 4 nm  6 nm  8 nm ellipsoid particle; (b) Surface adsorbed water and
water penetrated in the C-S-H gel; (c) short silicate chains and calcium atoms
distributed in the C-S-H gel. (Red: oxygen, green: calcium, yellow: silicon, white:
hydrogen). (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
hydration product at high C/S ratio demonstrated more order state,
which was quite similar to the crystallization process. The different
cluster morphologies transformation would be discussed in details
420 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426

in later section about cluster evolution. Herein, the egg-like phase phous C-S-H gel with fibrillar-like morphology was widely pro-
and network phase indicated that composition of C-S-H gel, espe- duced and packed near the surface of C3S, implying that the
cially C/S ratio, was an important factor influencing the morpholo- increasing silicon content transformed ordered structure to amor-
gies of C-S-H gel at nano-scale. phous state. Such two distinguished clusters at nano-scale con-
According to previous experimental findings, with progressively firmed the transformation trend due to C/S composition variation
increasing of time, C/S ratio gradually decreased and the amor- from current experiment and previous work [37]. The branches

150 2
o o Ca/Si=1 Exp
29 ~31.5 Ca/Si=1.25 1
XRD intensity (ab. units)

120 Ca/Si=1.5 1.25

Ca/Si=1.75
1.5

x ray PDF (ab.unit)

o 0 1.75
90 7.4

60
-2

30

0 -4
10 20 30 40 1 2 3 4 5
2 Theta (degree) Distance (Å)
(a) (b)

15 25
CaO CSH SiO CSH
CaO glass SiO Tober14
12 SiO glass
RDF of CaO (ab intensity)

CaO tober14 20
RDF of SiO (ab units)

9
15

6
10

3
5

0
0
0 2 4 6 0 2 4 6
Distance (Å) Distance (Å)
(c) (d)
8.0 1.0
this work Si-OH/Si
Si-OH/Si and Ca-OH/Ca ratio

Experiment Ca-OH/Ca
7.5
0.8 exp Ca-OH/Ca
Coordination number

7.0
0.6
6.5
0.4
6.0

0.2
5.5

5.0 0.0
0.6 0.8 1.0 1.2 1.4 1.6 1.8 1.0 1.2 1.4 1.6 1.8
Ca/Si ratio Ca/Si
(e) (f)
Fig. 9. (a) simulated XRD curves of the C-S-H clusters (b) comparison between simulated PDF and experimental one obtained from reference (c) partial PDF of SiAO (d) partial
PDF of CaAO in simulated C-S-H cluster, tobermorite 14 Å and silicate glass obtained from Pellenq’s work; (e) coordination number of CaAO (d) SiAOH/Si and CaAOH/Ca ratio
evolution with C/S ratio.
 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426 421

and ellipsoid particles could be thought as two intrinsic represen-
tative seeds to construct various morphologies at larger scale. The
ellipsoid particle with high C/S ratio was more likely to assemble
together to construct high dense phase, while the network struc-
ture with low C/S ratio might transform into loose fibril
morphology.
3.3. Local structure of the C-S-H gel
It was observed in Fig. 8 that the ellipsoid structure was com-
posed of short silicate chains with various lengths and calcium
atoms, which were linked together by ion-covalent bonds. As
shown in Fig. 8(b), near the surface, CaAOH and SiAOH could be
seen due to the interaction between the surface of calcium and sil-
icate elements and water. The ellipsoid was enclosed by dense
water layer due to the strong physical adsorption. It should also
be noted that during the hydration, the hydroxyl groups were pro-
duced inside the C-S-H gel, developing H-bonds linkage between
neighboring silicate chains. The size of the cluster could be deter-
mined by the following procedure: the center of the particle was
easily to be determined firstly, followed by many straight lines
across the center points drawn and a series of lines were cut into
various lengths by particle boundary. The maximum and minimum
‘diameter’ was around 8 nm and 4 nm respectively. Because such
particle was not symmetrical in any direction, it was hard to be
described by accurate ellipsoid function. In the plane perpendicu-
lar to the axis with 8 nm length, 6 nm was the longest axis length.
Hence the geometry of the structure was estimated to be
4 nm  6 nm  8 nm. Such particle size was quite similar to basic
element proposed in Jennings model [38], proving the colloid char-
acteristic for the C-S-H gel.
The X-ray spectra curves for the simulated cluster with C/S ratio
from 1.0 to 1.75 were calculated by the assistance of the XRD pack-
age from Debyer software. Previous experiments proposed that the
broadening peaks at around 29.5° implied the ill-crystalline phase
for the C-S-H gel. As shown in Fig. 9(a), the simulation results
herein demonstrated one wide peak centered ranging from 29°
to 31.5°. For the C-S-H gel of C/S ratio 1.75, one well defined peak
at 7° could be observed. Consistent with the experimental finding,
the broad former peak indicated the amorphous properties for the
inner structure of the C-S-H colloid. The later peak, existing in all
the curves, corresponded to 1.2 nm spacing, which was a typical
layer distance in tobermorite and jennite like minerals [39]. It
could be attributed to the fact that inside the particle exist hydro-
xyl groups. Due to the high reaction temperature, the integrate
water molecules did not penetrated into silicate calcium clusters.
Hence, the ellipsoid colloid, produced in water environment, was
not densified during the hydration process. SiAOH and CaAOH
structure and surface adsorbed water could, in some degree,
enlarge inner space of the calcium silicate cluster.
To better understand the local structure for the glass-like cal-
cium silicate hydrate, the X-ray Pair Distribution Functions (PDF)
for C-S-H gels were computed and compared with previous exper-
imental observation with C/S ratio of 1.6 by Soyer-Uzun [40]. As
shown in Fig. 9(b), a good agreement between the simulation
and experiment work could be observed, implying that the molec-

100 100
Q0 Q0
Ca/Si=1 Q1 Q1
Q2 Q2
80 80
Qn percentage (%)

Q3 Q3
Q4 Q4
Qn (%)

60 60

40 40

20 20

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Time(ns) Time (ns)
(a) (b)
100 100
Q0 Q0
Q1 Q1
Q2 Q2
Q n percentage (%)

80 80
Q3 Q3
Q percentage (%)

Q4 Q4

60 60

40 40

20 20

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Time (ns) Time(ns)
(c) (d)
Fig. 10. Q species evolution with time at C/S ratio (a) 1.0 (b) 1.25 (c) 1.5 (d) 1.75.
422 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426

ular model can describe well the inner amorphous structure. The
peaks located at 1 Å, 1.64 Å, 2.45 Å, 3.15 Å and 3.60 Å were corre-
sponding to OAH bond, SiAO bond, CaAO bond, SiASi(SiACa)
and SiACa (SiAO) spatial correlation, respectively. It should be
noted that simulated CaAO bond slightly shifted to higher distance
compared with the experimental one. In order to characterize the
local structure, partial PDFs of SiAO and CaAO in the simulated
C-S-H cluster were calculated to compare with that in the crys-
talline tobermorite 14 Å and silicate glass modified by the calcium
ions, which were obtained from Pellenq’s work [41]. As shown in
Fig. 9(c) and (d), similar position and integrals of the first peak
indicated that the short range order of Si and Ca atoms in the C-
The percentage of (100-Q0) reflected the polymerization degree
of silicate morphology. With C/S ratio increased, the decreasing
rate of Q0 species became slower and the final Q0 percentage at
10 ns gradually increased from around 20% to more than 50%, indi-
cating that calcium ions decayed the polymerization of silicate
tetrahedrons. Considering the experimental results in previous
section, after hydration for one year, there were still over 15%??
50
Q0
Q1
40
Q2

Qn percentage (%)
Q3
S-H gel resembled those in both crystal and silicate glass. Never- Q4
theless, at media range the discrepancy of PDFs could be observed
30
between C-S-H cluster and the crystal. The second peak of SiAO
partial PDF of tobermorite crystal demonstrated bimodal distribu-
tion, while the glass and simulated cluster only presented a 20

smooth broadening peak at around 4 Å. The missing of the bimodal

distribution in the second coordinated shell for the Si species indi- 10
cated that the C-S-H structure could not maintain the layered order
in the medium range. Additionally, the second small peak posi- 0
tioned at 3.2 Å of CaAO partial PDF in the crystal could not be 1.0 1.2 1.4 1.6 1.8

observed in that of both C-S-H clusters and the silicate glass. Ca/Si ratio
Hence, the simulated cluster with short range order and long range (a)
disorder presented glassy structural nature.
Furthermore, as shown in Fig. 9(e), the coordination number of
Ca atoms, ranging from 6.5 to 6.4, was slightly higher than the
experimental observation on the C-S-H gel (6.2–6.0) with C/S ratio
higher than 1.5. In the simulation work, the coordinated oxygen
atoms for the calcium atoms were composed of non-bridging oxy-
gen atom (ONB), bridging oxygen (OB) atom and oxygen atom in
water or hydroxyl group (Ow). The CaAO bond length ranked in
the following order: Ca-Ow > Ca-ONB > Ca-OB. The slightly overesti-
mated CaAO bond length and coordination number in the simula-
tion work indicated that Ca-Ow connection occupied dominant
ratio in the local structure for the calcium atoms in the molecular
model.
In IR analysis (Appendix), hydroxyl bonds were important com-
ponent in the nano-structure of the C-S-H gel. The molecular
model further categorized the protons into SiAOH, CaAOH and
surface adsorbed water molecules. As shown in Fig. 9(f), the simu-
lated CaAOH/Ca ratio first provided a quantitative agreement with
the experimental finding by INS technique [42]. With the incre-
ment of C/S ratio, the ratio of SiAOH/Si gradually decreased, while
CaAOH/Ca continued increasing. The hydroxyl bonds transforma- (b)
tion trend matched well with the charge balance calculation 10
obtained by Cong and Kirkpatrick [43], which indicated that the exp1
exp2
proton in SiAOH groups was gradually substituted by Ca and then this model
8
Mean chain length

by CaAOH group with the addition of C/S ratio. However, com- this exp
model2
pared with Cong and Kirkpatrick’s calculation, SiAOH/Si ratio
was overestimated which was mainly attributed to the large 6
amount of un-hydrated silicate monomers in the C-S-H gel, which
would be discussed in the following section.
4

3.4. Mechanism for polymerization

2
In order to better interpret the change of intrinsic molecular
0.5 1.0 1.5 2.0 2.5 3.0
structures, Q species evolution should be taken into consideration.
As shown in Fig. 10, Q0 species decreased monotonously and the Ca/Si ratio
(c)
growing rates of Q1, Q2 were larger than that of Q3 and Q4. In
despite of the large discrepancy of timescale, the simulated reac- Fig. 11. (a) Qn percentage evolution along with C/S ratio; (b) molecular configu-
tion process resembled the real hydration process that was charac- ration of C-S-H gel at C/S ratio 1.0, 1.25, 1.5 and 1.75, the highlight circle included
terized by the gradual consumption of monomers and production Q3 structures; (c) the mean silicate chain length evolution with C/S ratio both from
current results and previous experimental results [46] and modeling [26]. (Red:
of silicate chains. Such evolution process indicated that the sol- oxygen, green: calcium, yellow: silicon, white: hydrogen). (For interpretation of the
gel reaction was the structural transformation from monomer references to color in this figure legend, the reader is referred to the web version of
structures to silicate chains, cross-linkage structures or networks. this article.)
 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426 423

of monomers in the C-S-H gel. Correspondingly, the percentage of
un-hydrated Q0 species of the simulated gel obtained during only
10 ns was reasonable. For all cases, the ratio of Q0 species reduced
dramatically at the initial 1.5 ns, implying that chemical reactions
dominantly happened at this stage. Such could be comparable to
the hydration of C3S at initial 28 d by experiments. During this
stage, the major silicate skeleton had been constructed, which con-
tributed to the strength development in cement hydrates. From
1.5 ns to 10 ns, the number of Q0 species decreased in slow rate,
similar with the long term hydration of cement. In this period,
the hydration products were further densified, leading to better
durability for the cement based materials.
The main difference in Q species between high C/S phase and
low C/S phase was the percentage of Q3 + Q4 species. As shown in
Fig. 10(a), at C/S ratio of 1, Q3 and Q4 species took up more than
12% of the whole gel, which matched well with previous experi-
mental findings by Cong and Kirkpatrick [43]. The ratio of Q3 and
Q4 in their synthesized C-S-H gel reached more than 15% as the
C/S ratio reduces to 0.79, indicating that long chain structures
could interconnect with each other, which was similar to con-
nected silicate chains in Merlino’s tobermorite model [44]. It
should also be noted that Q3 and Q4 species were the dominated
phases in the modified silicate glass with the addition of cation
atoms [45]. The existence of Q3 and Q4 as well as the three dimen-
sional skeleton for cement hydrates, implied the glassy nature for
C-S-H gel at low C/S ratio. Nevertheless, it could be observed in
Fig. 11(a) that the ratio of Q3 and Q4 reduced to zero along with
the increment of C/S ratio. Also the branch structure progressively
transformed to the long silicate chains and further to short ones,
amplifying that the presence of calcium ions contributed to the for-
mation of linear structures rather than branch structures in the
polymerization process of Si(OH)4. Such was consistent with the
experimental findings that in both cementitious and synthetic C-
S-H gel, short silicate chains occupied predominate percentage
[43,46].
Mean chain length (2(Q1 + Q2 + Q3)/Q1) was an important struc-
tural factor to estimate the connectivity of silicate chains. In Fig. 11
(c), it could be observed that the mean silicate chain length
decreased with C/S ratio increased, indicating the de-
polymerization role of calcium atoms in the formation of silicate
gel. Such simulated results have been compared with experimental
findings and previous results. The trend of chain length in simu-

Fig. 12. Chemical reaction pathway for the polymerization reactions between two monomers. (Red: oxygen, yellow: silicon, white: hydrogen). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)
424 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426
lated system was in accordance with that obtained from experi-
ments. In particular, when C/S ratio higher than 1.25, the silicate
chain length calculated from the simulation matched well with
those from NMR tests and previous experimental results [46],
implying that the simulated C-S-H gel could describe the inner sil-
icate structure well at high C/S ratio region. Furthermore, it could
be observed from Fig. 11 that the 10 ns hydration simulation
showed higher polymerized degree than that obtained from 2 ns
simulation by Dolado and co-workers [26].
The degrading role that calcium atoms played in silicate struc-
ture de-polymerization had been discussed. Behind the observa-
tion was the mechanism of chemical reaction between silicate
monomers and calcium hydroxyls. Two chemical reactions played
predominate role in cement hydration: the first one was the sili-
cate monomer condensation with water reduction and the second
one was caused by the SiAOACa formation.
At low C/S ratio, the first reaction widely occurred during the
hydration process. The sequence for the first mechanism was
demonstrated in Fig. 12: a silicate monomer diffused and moved
close to the neighboring silicon atoms; proton dissociated; bond
formed between neighboring monomers and fivefold silicate struc-
ture formed; proton transferred and SiAOw formed; SiAOw bond
broke and water diffused. In the polymerization reaction, the
SiAOH dissociation led to the formation of a penta-coordinated sil-
icon structure, which was energetically unfavorable. The unstable
structure subsequently relaxed and dissociated with the Ow atom.
In this way, the oligomerization reaction produced long silicate
Fig. 13. chemical reaction pathway for the CaAOASi bond formation (Red: oxygen, green: calcium, yellow: silicon, white: hydrogen) (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article.).
chains and water molecules. Furthermore, the newly formed dim-
mer structure could connect with neighboring monomers to
enhance the silicate chain length or produce branch structures.
On the other hand, at high C/S ratio, the second reaction became
predominate. The second reaction mechanism was listed in the fol-
lowing sequence, as shown in Fig. 13: the hydrated calcium ion
was initially moving close to silicate monomer; CaAOw bond broke
and water molecule diffused away; CaAOASi bond formed and
SiAOH de-protonated; the free dissociated proton associated with
the hydroxyl group, forming CaAOw. Compared with silicate tetra-
hedrons, the hydrated calcium ions, on average, had 6–8 coordi-
nated oxygen atoms in the water solution. The binding energy of
SiAO was higher than that of CaAO. Hence, the second reaction
was more probable to happen than the first one.
From the two reaction mechanisms, during the hydration pro-
cess, the SiAOACa bond formation and polymerization reaction
were interplayed with each other. At high C/S ratio, the silicate
monomers needed to compete with calcium atoms for the connec-
tion with neighboring silicate structures. Previous calcium silicate
glass simulation proposed that the Non bridging oxygen (NBO)
energetically preferred calcium atoms [45]. Consequently, the
incorporation of more calcium contributed to more SiAOACa
bonds formed to replace the SiAOASi bonds. On the other hand,
from dynamic respect, calcium atoms slowed down the diffusion
of silicon atoms. The restricted silicon atoms had slower diffusion
rate and hence had less probability to collide with nearby NBO
compared with calcium atoms. Additionally, despite different reac-
 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426
tion pathways, the formation of both SiAOASi and CaAOASi bonds
was accompanied with water reduction, resulting in more con-
densed structures. The newly produced SiAOASi/Ca structures
were hard to de-polymerize again because they had to overcome
higher energy barrier.
4. Conclusions
In summary, the properties and structure studies of C-S-H were
conducted in combinatorial approach of experiment and simula-
tion at atomic, nano and micro level. The following conclusions
were drawn from the investigations.
(1) Based on the insights of hydration from C3S at high water/-
solid ratio, hydration evolution of C3S was traced to produce
C-S-H gradually from small localized wrinkles to large foil-
like wrinkled clusters with fibrillar-like wrinkles distributed
randomly.
(2) The diffraction patterns with obscure ring attained by TEM
analysis and broaden peaks achieved from X-ray diffraction
for C-S-H gel implied that the wrinkled product formed pos-
sessed poor crystallined or nearly amorphous feature. Such
was also evidenced by the fact that no layered crystal phase
was observed in the simulated calcium silicate gel cluster.
Besides, the simulated pair distribution function of C-S-H
gel demonstrating the short range order and long range dis-
order, confirmed glassy structural nature for C-S-H gel as
well.
(3) With respect to bond connection, hydroxyl group were pre-
sent and silicate polymerization occurred during the forma-
tion of C-S-H, which was exemplified by the simulated
polymerization process at nanoscale emphasizing the bond
transformation between SiAOH, CaAOH, SiAOASi and
CaAOASi.
(4) The presence of Q0–Q2 silicate connection conducted by sil-
icate polymerization evolution were evidenced under simu-
lation, exemplifying the feature of short and linear silicate
chains present in C-S-H. At high calcium concentration, the
polymerization of silicate monomers was significantly
restricted and silicate chains were greatly shortened. Never-
theless, low calcium content and large silicon amount could
induce longer silicate chains as well as more disordered
structure. Calcium ions and water molecules could influence
the formation of bond type as well as their corresponding
amount in C-S-H.
Acknowledgements
Financial supports from the National Natural Science Founda-
tion of China (NSFC51502112, NSFC51508292, NSFC 51508293),
the National High Technology Research and Development Program
of China (2015AA034701) and the Natural Science Foundation of
Shandong Province, China (ZR2014EMP010) are gratefully
acknowledged.
References
[1] J.F. Banfield, W.W. Barker, S.A. Welch, A. Taunton, Biological impact on mineral
dissolution: application of the lichen model to understanding mineral
weathering in the rhizosphere, Proc. Nat. Acad. Sci. 96 (7) (1999) 3404–3411.
[2] U.R. Berner, Modelling the incongruent dissolution of hydrated cement
minerals, Radiochimca Acta 44 (2) (1988) 387–394.
[3] M.W. Kanan, D.G. Nocera, In situ formation of an oxygen evolving catalyst in
neutral water containing phosphate and Co2+, Science 321 (2008) 1072–1075.
[4] H. Ma, Z. Li, Realistic pore structure of Portland cementpaste: experimental
study and numerical simulation, Comp. Concr. 11 (4) (2013) 317–336.
425
[5] J.W. Bullard, H.M. Jennings, R.A. Livingston, A. Nonat, G.W. Scherer, J.S.
Schweitzer, K.L. Scrivener, J.J. Thomas, Mechanisms of cement hydration, Cem.
Concr. Res. 41 (12) (2011) 1208–1223.
[6] I. Amato, Green cement: concrete solutions, Nature 494 (2013) 300–301.
[7] I.G. Richardson, The calcium silicate hydrates, Cem. Concr. Res. 38 (2) (2008)
137–158.
[8] A. Nonat, The structure and stoichiometry of C-S-H, Cem. Concr. Res. 34 (9)
(2004) 1521–1528.
[9] P.D. Jennings, A model for two types of calcium silicate hydrate in the
microstructure of Portland cement pastes, Cem. Concr. Res. 30 (6) (2000) 855–
863.
[10] R. Ylmén, U. Jäglid, B.M. Steenari, I. Panas, Early hydration and setting of
Portland cement monitored by IR, SEM and Vicat techniques, Cem. Concr. Res.
39 (5) (2009) 433–439.
[11] F. Bellmann, D. Damidot, B. Möser, J. Skibsted, Improved evidence for the
existence of an intermediate phase during hydration of tricalcium silicate,
Cem. Concr. Res. 40 (6) (2010) 875–884.
[12] D.M. Kirby, J.J. Biernacki, The effect of water-to-cement ratio on the hydration
kinetics of tricalcium silicate cements: testing the two-step hydration
hypothesis, Cem. Concr. Res. 42 (8) (2012) 1147–1156.
[13] M. Dovál, M. Palou, S.C. Mojumdar, Hydration behavior of C2S and C2AS
nanomaterials, synthesized by Sol-gel method, J. Therm. Anal. Calorim. 863 (3)
(2006) 595–599.
[14] H.F.W. Taylor, Cement Chemistry, 2 ed., Academic Press, London, 1997.
[15] P. Juilland, E. Gallucci, R. Flatt, K. Scrivener, Dissolution theory applied to the
induction period in alite hydration, Cem. Concr. Res. 40 (6) (2010) 831–844.
[16] M.M. Costoya Fernández, Effect of particle size on the hydration kinetics and
microstructural development of tricalcium silicate, 2008.
[17] K.O. Kjellsen, B. Lagerblad, Microstructure of tricalcium silicate and Portland
cement systems at middle periods of hydration-development of Hadley grains,
Cem. Concr. Res. 37 (2007) 13–20.
[18] R. Alizadeh, L. Raki, J.M. Makar, J.J. Beaudoin, I. Moudrakovski, Hydration of
tricalcium silicate in the presence of synthetic calcium-silicate-hydrate, J.
Mater. Chem. 19 (42) (2009) 7937–7946.
[19] S. Diamond, K.O. Kjellsen, Resolution of fine fibrous C-S-H in backscatter SEM
examination, Cem. Concr. Compos. 28 (2) (2006) 130–132.
[20] P.C. Fonseca, H.M. Jennings, The effect of drying on early-age morphology of C-
S-H as observed in environmental SEM, Cem. Concr. Res. 40 (12) (2010) 1673–
1680.
[21] S. Goñi, F. Puertas, M.S. Hernández, M. Palacios, A. Guerrero, J.S. Dolado, B.
Zanga, F. Baroni, Quantitative study of hydration of C3S and C2S by thermal
analysis, J. Therm. Anal. Calorim. 102 (3) (2010) 965–973.
[22] Q.Y. Chen, C.D. Hills, M. Tyrer, I. Slipper, H.G. Shen, A. Brough, Characterisation
of products of tricalcium silicate hydration in the presence of heavy metals, J.
Hazard. Mater. 147 (3) (2007) 817–825.
[23] J.J. Thomas, A.J. Allen, H.M. Jennings, Hydration kinetics and microstructure
development of normal and cacl2-accelerated tricalcium silicate pastes, J.
Phys. Chem. C 113 (2009) 19836–19844.
[24] B.Y. Lee, K.E. Kurtis, Influence of TiO2 nanoparticles on early C3S hydration, J.
Am. Ceram. Soc. 93 (10) (2010) 3399–3405.
[25] B.P. Feuston, S.H. Garofalini, Oligomerization in silica sols, J. Phys. Chem. 94
(13) (1990) 5351–5356.
[26] J.S. Dolado, M. Griebel, J. Hamaekers, A molecular dynamic study of
cementitious calcium silicate hydrate (C-S-H) gels, J. Am. Ceram. Soc. 90 (12)
(2007) 3938–3942.
[27] J.S. Dolado, M. Griebel, J. Hamaekers, F. Heberc, The nano-branched structure
of cementitious calcium-silicate-hydrate gel, J. Mater. Chem. 21 (12) (2011)
4445–4449.
[28] K. Haga, M. Shitaba, M. Hironaga, S. Tanaka, S. Nagasaki, Silicate anion
structural change in calcium silicate hydrate gel on dissolution of hydrated
cement, J. Nucl. Sci. Technol. 39 (5) (2002) 540–547.
[29] J.J. Chen, L. Sorelli, M. Vandamme, F. Ulm, G. Chanvillard, A coupled
nanoindentation/sem-eds study on low water/cement ratio portland cement
paste: evidence for c-s-h/ca(oh)2 nanocomposites, J. Am. Ceram. Soc. 93 (5)
(2010) 1484–1493.
[30] P. Faucon, J.M. Delaye, J. Virlet, Molecular dynamics simulation of the structure
of calcium silicate hydrates I. Ca41xSi6O1412x(OH)422x(H2O)2 (0 < x< 1), J.
solid state chem. 127 (1) (1996) 92–97.
[31] S. Garofalini, G. Martin, Molecular simulations of the polymerization of silicic
acid molecules and network formation, J. Phys. Chem. 98 (4) (1994) 1311–
1316.
[32] H. Hu, H. Hou, Z. He, B. Wang, Theoretical characterizations of the mechanism
for the dimerization of monosilicic acid in basic solution, Phys. Chem. Chem.
Phys. 15 (36) (2013) 15027–15032.
[33] J. Zhang, G.W. Scherer, Comparison of methods for arresting hydration of
cement, Cem. Concr. Res. 41 (10) (2011) 1024–1036.
[34] S. Wang, X. Peng, Z. Tao, L. Tang, L. Zeng, Influence of drying conditions on the
contact-hardening behaviours of calcium silicate hydrate powder, Constr.
Build. Mater. 136 (2017) 465–473.
[35] Z. Zhang, G.W. Scherer, Supercritical drying of cementitious materials, Cem.
Concr. Res. 99 (2017) 137–154.
[36] K. Johansson, C. Larsson, O.N. Antzutkin, W. Forsling, H.R. Kota, V. Ronin,
Kinetics of the hydration reactions in the cement paste with
mechanochemically modified cement 29Si magic-angle-spinning NMR study,
Cem. Concr. Res. 29 (10) (1999) 1575–1581.
426 L. Zhang et al. / Construction and Building Materials 155 (2017) 413–426
[37] I.G. Richardson, G.W. Groves, Microstructure and microanalysis of hardened
ordinary Portland cement pastes, J. Mater. Sci. 28 (1) (1993) 265–277.
[38] H.M. Jennings, Refinements to colloid model of C-S-H in cement: CM II, Cem.
Concr. Res. 38 (3) (2008) 275–289.
[39] J.A. Janika, W. Kurdowsk, R. Podsiadey, J. Samset, Fraxtal structure of CSH and
tobermorite phases, Acta Phys. Pol. Series A 100 (4) (2001) 529–537.
[40] S. Soyer-Uzun, S.R. Chae, C. Benmore, H. Wenk, P.J.M. Monterio, Compositional
evolution of calcium silicate hydrate (c-s-h) structures by total x-ray
scattering, J. Am. Ceram. Soc. 95 (2) (2012) 793–798.
[41] R.J. Pellenq, A. Kushima, R. Shahsavari, K.J. Van Vliet, M.J. Buehler, S. Yip, A
realistic molecular model of cement hydrates, Proc. Nat. Acad. Sci. 106 (38)
(2009) 16102–16107.
[42] J.J. Thomas, J. Chen, H.M. Jennings, D.A. Neumann, Ca-OH bonding in the C-S-H
gel of tricalcium silicate and white Portland cement pastes measured by
inelastic neutron scattering, Chem. Mater. 15 (20) (2003) 3813–3817.
[43] X. Cong, J.R. Kirkpatrick, 29Si MAS NMR Study of the Structure of calcium
silicate hydrate, Adv. Cem. Based Mater. 3 (3–4) (1996) 144–156.
[44] S. Merlino, E. Bonnacorsi, T. Armbruster, The real structure of tobermorite 11
A: normal and anomalous forms, OD character and polyptic modifications, Eur.
J. Mineral. 13 (3) (2001) 577–590.
[45] R.N. Mead, G. Mountjoy, A molecular dynamics study of the atomic structure
of (cao)x(sio2)1-x glasses, J. Phys. Chem. B 110 (29) (2006) 273–278.
[46] J.J. Chen, J.J. Thomas, H.F.W. Taylor, H.M. Jennings, Solubility and structure of
calcium silicate hydrate, Cem. Concr. Res. 34 (9) (2004) 1499–1519.