Modification effects of colloidal nanoSiO
2
on cement hydration and its gel property
Pengkun Hou
a,b,
⇑
, Shiho Kawashima
b
, Deyu Kong
b,c
, David J. Corr
b
, Jueshi Qian
a
, Surendra P. Shah
b
a
College of Materials Science and Engineering, Chongqing University, Chongqing 400045, China
b
Department of Civil and Environmental Engineering, Northwestern University, Evanston, IL 60208, USA
c
College of Architecture and Civil Engineering, Zhejiang University of Technology, Hangzhou 310014, China
article info
Article history:
Received 14 February 2012
Accepted 29 May 2012
Available online 21 June 2012
Keywords:
A. Nano-structures
B. Mechanical properties
B. Hydration property
D. Electron microscopy
abstract
To understand the effects of colloidal nanoSiO
2
(CNS) on cement hydration and gel properties in the early
and later age, hydration heat, calcium morphology, hydroxide content, non-evaporable water (NEW)
content and nanoscale mechanical properties were measured. Some comparison studies were conducted
on silica fume (SF) paste, as well. Results revealed that the accelerating effect of CNS on hydration in the
early age is achieved by the acceleration of cement dissolution and hydrate nucleation on reacted nano-
SiO
2
particles. Although cement hydration can be greatly accelerated by CNS in the early age, its later age
hydration is hindered. The NEW content of CNS-added paste experiences a higher rate of increase
initially, but gradually becomes smaller than that of the control paste due to changes in the gel structure,
making NEW content an unsuitable method for monitoring the hydration of CNS-added paste. However,
nanoindentation results revealed that CNS modifies the gel structure to increase the high-stiffness C–S–H
gel content.
Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Nanotechnology has been introduced to cement and concrete
research because it can achieve a stronger and more durable con-
crete [1,2]. As the most widely used nanomaterial for cement/con-
crete-engineering, nanoSiO
2
has been studied intensively [3–13].It
has been widely reported that nanoSiO
2
addition can greatly
improve properties of cementitious materials [3–10]. Jo et al. [6]
reported that 6% nanoSiO
2
can improve the compressive strength
of concrete by 152% and 142% at 7 and 28 days, respectively; Li
[7] found that 5% nanoSiO
2
can improve compressive strength by
17.5% at 28 day; Gaitero et al. [9] revealed reduced calcium leach-
ing of nanoSiO
2
-added cement pastes. They ascribed it to densifica-
tion of the paste, transforming of Portlandite into C–S–H gel by
means of pozzolanic reaction and modification of the internal
structure of C–S–H gel, all of which make the cement paste more
stable and more strongly bonded [10].
Property evolution of nanoSiO
2
-added cementitious materials
can be reflected by the effects of nanoSiO
2
on cement hydration,
gel property modification, and pore structure refinement
[3,13–16]. Results have shown that the strength enhancing effect
of nanoSiO
2
in the early age is related to its hydration acceleration
effect, but its effects in the later age have rarely been reported. For
the compressive strength enhancing effect of nanoSiO
2
, it was
found to be more pronounced in the early age, while rate of
strength gain can be lower than the control in the later ages [8].
The difference in the strength gain evolution at the early and later
ages requires that the effect of nanoSiO
2
on cement hydration in
both ages be investigated.
Thus, in this study, cement hydration characteristics in the
presence of nanoSiO
2
in the early and later ages, including hydra-
tion acceleration rate, hydration acceleration mechanisms, cement
hydration degree, and the resulting gel properties were investi-
gated. All of these results help to provide a comprehensive expla-
nation for the modification effects of nanoSiO
2
on cementitious
materials. To make a comparison, effects of silica fume on cement
hydration were also studied.
2. Experimental
2.1. Materials and mix proportions
A type I Portland cement with a Blaine fineness of 385 m
2
/kg
was used in this study. To facilitate a homogeneous distribution
of nanoSiO
2
in cement paste, colloidal nanoSiO
2
(CNS), instead of
nanoSiO
2
powder, was used. Sodium stabilized CNS with an
average particle size of 10 nm (CNS-10) produced by the sol–gel
technique was used. The basic properties were provided by the
manufacturer [17]. A silica fume was used for a comparison study
with CNS. The physiochemical properties of the raw materials are
given in Tables 1 and 2.
1359-8368/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compositesb.2012.05.056
⇑
Corresponding author at: College of Materials Science and Engineering,
Chongqing University, Chongqing 400045, China.
E-mail address: pkhou@163.com (P. Hou).
Composites: Part B 45 (2013) 440–448
Contents lists available at SciVerse ScienceDirect
Composites: Part B
journal homepage: www.elsevier.com/locate/compositesb
The transmission electron microscopy (TEM) images of CNS-10
in Fig. 1 shows that most of the CNS particles are well-dispersed
but some agglomeration can still occur. The near-perfect sphericity
of the SF particles is evident (Fig. 1) and the spheres are quite
smooth, with no obvious surface morphology. It is also apparent
that SF is composed of various sized particles ranging from several
to dozens of nanometers. XRD spectra shown in Fig. 2 indicate the
amorphous feature of both pozzolans. A similar full width at half
maximum indicates that the two materials have a comparable
crystallinity.
2.2. Sample preparation
Unless stated otherwise, cement pastes mixed with and without
5% CNS/SF at a w/b ratio of 0.4 were prepared and investigated
throughout this study. Samples were demolded 1 day after casting
and were cured in saturated lime solution at room temperature un-
til testing.
For the pozzolanic activity study, 20 g of chemical grade cal-
cium hydroxide (CH) were mixed with 5 g of CNS/SF at a w/b of
2 to simulate a cement-CNS/SF system. It was assumed that 20 g
of CH can be generated by 100 g of cement [18]. After mixing, sam-
ples were sealed in plastic vials. The CH content at different ages
were determined by a TGA technique.
2.3. Test methods
2.3.1. Hydration heat
The hydration temperature of each paste was measured by a
semi–adiabatic calorimeter to assess the effect of CNS/SF on the
hydration heat of cement. Samples were prepared at a w/b ratio
of 0.4, with 100 g of cement and mixing water at a temperature
of 27 °C. Mixes were cast in £5.08 cm 10.16 cm plastic cylinders
within 3 min after initial cement and water contact. The sample
was then covered, placed in the calorimeter, and the temperature
of the sample was recorded every 3 min for 20 h.
2.3.2. Morphology
Hitachi S-4800 FE-SEM and Hitachi S-3400 (equipped with
backscatter electron detector) were used to analyze the morphol-
ogy of the cement paste. Small fractured samples or powder sam-
ples at very early hydration ages were soaked in acetone to stop
hydration and dried at 80 °C for 2 h. Then the sample was coated
with 20 nm of gold to make it conductive.
2.3.3. CH content
Thermogravimetric analysis (TGA, TGA/sDTA 851) was carried
out to measure CH content. The weight loss between 440 °C and
510 °C was considered to be the decomposition of CH. Before mea-
suring, samples were oven – dried at 105 °C for 4 h. CH contents
were calculated on the ignited basis at 950 °C for 30 min.
2.3.4. Non-evaporable water (NEW) content
At the end of curing, the core of the sample was crushed into
small pieces and immediately immersed in acetone to stop hydra-
tion, as well as to minimize carbonization. Before measuring the
NEW content, samples were oven – dried at 105 °C for 4 h and then
ground to pass the 100 mesh size (ca. 150
l
m) sieve. The NEW con-
tent was measured as the weight loss of the sample between
105 °C and 950 °C. After firing at 950 °C for half an hour, the sample
weight loss was measured and NEW was calculated on the ignited
basis at 950 °C for 30 min of the sample using Eq. (1). For each mix,
three samples (ca. 2 g of each) were measured and the average va-
lue was taken as the representative value.
NEW% ¼ 100
w
105
C
w
950
C
w
950
C
f
cem
loss
cem
þ f
FA
loss
FA
ð
þ f
CNS=SF
loss
CNS=SF
ð1Þ
where w is the sample weight, f is the weight percentage of
material.
2.3.5. Nanoscale mechanical property
To determine the effect of CNS on the nanoscale mechanical
properties of cement paste, a statistical nanoindentation technique
was applied, through which the intrinsic gel modifying effect of
CNS can be shown. During this test, a load of 1000
l
N was applied
with a triangle diamond Berkovich indenter with a total included
angle of 142.3° to make an indent on the surface of the sample.
Table 1
Properties of colloidal nanosilica.
CNS Average particle size (nm) SiO
2
content (wt.%) pH
CNS-10 10 >99 10.5
Table 2
Physiochemical properties of raw materials.
Materials Type I cement Silica fume
SiO
2
20.2 90.1
Al
2
O
3
4.7 0.6
Fe
2
O
3
3.3 2.0
SO
3
3.3 –
CaO 62.9 0.5
MgO 2.7 5.1
LOI 1.1 1.0
Total 98.2 99.3
Fineness as surface area (m
2
/kg) 380 21000
CNS-10 (TEM)
SF(SEM)
Fig. 1. Morphology images of CNS and SF.
P. Hou et al. / Composites: Part B 45 (2013) 440–448
441
The elastic modulus of the sample can be calculated from the
known properties of the indenter and the unloading part of the
load – displacement curve, shown in Fig. 3. Irregular nanoindenta-
tion curves due to the presence of voids and cracking of the sample
were discarded [19,20]. Statistical nanoindentation tests were car-
ried out over three different areas of the gel phase on each sample.
Over each area, 64 indents (10
l
m grid point distance) were
performed.
For a reliable measure of the local mechanical properties, the
sample must have a flat surface [19]. In this study, samples were
cast in 2 cm 2cm 8 cm molds. After 1 day of casting, samples
were demolded and cured in saturated lime water at room temper-
ature. Before testing, thin sections of approximately 5 mm were
cut out of the specimens and mounted on a metal sample holder
for polishing. The surface of the samples were polished using sili-
con carbide paper of gradation 22
l
m, 14.5
l
m, and 6.5
l
m, and
diamond lapping film of gradation 3
l
m and 1
l
m. Water was used
for the duration of the polishing process. The polishing time was
2 min for the first 3 polishing steps and 2 h for each of the diamond
polishing steps. In the final step, the polished samples were ultra-
sonically cleaned in water for 1 min using a bath sonicator to re-
move polishing debris from the sample surface. The smoothness
of the sample surface was checked by SEM, as shown in Fig. 4.
The root mean-squared (RMS) average roughness, obtained by
Atomic Force Microscopy (AFM), for an area of 40
l
m 40
l
m
was 113 nm for the 5 month old control sample, which is compa-
rable to the published data [19].
3. Results and discussions
3.1. Pozzolanic activity
To evaluate the pozzolanic activity of CNS, the CH adsorption
capacity of CNS was investigated and a comparison study was
made with silica fume.
It can be seen in Fig. 5 that after 7 days of hydration, the pozzo-
lanic reaction of CNS is almost complete. However, for silica fume,
it takes as long as 1 month for the completion of this reaction. Sim-
ilar results can be seen in Ref. [21]. The difference in activity be-
tween CNS and SF can be due to the variation in their chemical
structure: the unsaturated Si–O bonds on the CNS particle surface
make the pozzolanic reaction happen quickly. However, for SF, a
prior step of breaking the saturated Si–O bonds on its surface,
which is relatively slow, makes the reaction slower [22]. The differ-
Fig. 3. Schematic diagram of nanoindentation and the typical load–displacement curve and the indent image on cement paste.
10 15 20 25 30 35 40
2 theta
CNS 10nm
silica fume
Fig. 2. XRD spectra of CNS and SF.
442 P. Hou et al. / Composites: Part B 45 (2013) 440–448
ence in final CH consumption can be due to the difference in hydra-
tion degree of these two pozzolans. When assuming all CNS takes
part in the pozzolanic reaction, a Ca/Si ratio of hydrates of 1.7 can
be calculated, which is the same as plain cement paste [23].
3.2. Hydration modification
3.2.1. Strength evolution
Modification caused by the cement hydration of small particles
can be reflected by their effect on the mechanical properties of
cementitious materials. The strength enhancing effects of CNS
and SF were measured in 40% fly ash replaced cement mortars. It
is shown in Fig. 6 that CNS has a more pronounced enhancing ef-
fect in the early age: for the 5% CNS mix, compressive strength
can be improved by as much as 16% and 45% at 3 and 7 days,
respectively. Meanwhile, the strengths of the 5% SF mix at these
ages were lower than 10%. However, the strength gain of both
pozzolans in the later age are comparable. The lower compressive
strength of 5% CNS-added mortar compared to the control at
84 days can be due to the hindrance of cement hydration as
described in Section 3.2.4.
3.2.2. Hydration heat
The semi-adiabatic calorimetry results of the control and sam-
ples with various additions of CNS-10 and SF are shown in Fig. 7.
It is clearly demonstrated that the addition of CNS increases both
the hydration peak temperature and the reaction rate, the latter
of which is shown by the 1st derivation of the hydration tempera-
ture curve. Similar effects can also be seen in SF-added cement
pastes. It is well known that cement hydration is a dissolution-pre-
cipitation process [24] and the acceleration of this process can be
monitored by the evolution of pH value and electrical conductivity
(revealing the ion concentration) of the paste solution. It is shown
in Fig. 8 that the addition of CNS introduces a higher rate of in-
crease in pH and electrical conductivity in the early age (effects
of CNS on pH and electrical conductivity were negligible), meaning
a quicker dissolution of cement particles. The decrease in electrical
conductivity is due to the adsorption of ions by the C–S–H gel and
the sharper decrease exhibited by the CNS-added cement solution
indicates a greater gel formation [25]. When small particles are
evenly distributed in cementitious materials, they act as nucleation
sites, which will benefit the hydration process [11,12,26]. The
accelerated dissolving process can also be observed in SEM images,
shown in Fig. 10.
A comparison of the effect of CNS and SF on the rate of cement
hydration is shown in Fig. 9. It demonstrates that CNS accelerates
cement hydration at a higher rate and this can be due to the higher
amount of nucleation sites, which results in a greater amount of
nucleus formation [27]. However, it is also demonstrated in Fig. 9
that the hydration peak temperature of 5% SF-added paste is higher
than that of CNS paste, implying a higher degree of cement hydra-
tion in SF-added paste. Similar results were shown by Thomas et al.
[11], as well as some contradicting results elsewhere [6].
The degree of acceleration in hydration of small particles can be
due to the effect of nucleus size on nucleation sites [27]. According
to the heterogeneous nucleation theory, the nucleation free energy
can be related to the volume of the solid nucleus, which can be
illustrated in the following equation
D
G
¼
1
2
V
D
G
V
ð2Þ
where
D
G
⁄
is the critical nucleation free energy, which depends on
the crystalline structure and cell parameter of the nucleus and
nucleation sites (assumed to be equal for CNS and SF);
D
G
V
is the
volume energy of solid nucleus, which depends on the particle
0 7 14 21 28 35 42 49 56 63
30
40
50
60
70
80
90
100
110
Calcium hydroxide content/%
Time/d
W/B=2
20g CH+5g CNS
20g CH+5g silica fume
Fig. 5. CH adsorption capability comparison of CNS and SF (w/b = 2).
Fig. 6. Strength improving effects of CNS and SF, Compressive strength ratio ¼
100%
strength of mix with CNS=SF
Strength of control mix
, w/b = 0.5, the fluidity of all mixes were adjusted by
water reducer to achieve 210 mm.
Fig. 4. Polished control sample of 5 months old.
P. Hou et al. / Composites: Part B 45 (2013) 440–448
443
composition (assumed to be the same for CNS and SF); V
⁄
is the crit-
ical volume of nucleus.
For the heterogeneous nucleation of C–S–H gel on CNS/SF sur-
faces, the volume of the nucleus should reach the critical volume.
Although there are more nucleation sites in CNS-added paste, the
probability of reaching the critical volume is lower, and thus the
hydration heat can be smaller.
3.2.3. Morphology
The effect of CNS on hydration in the early age were investi-
gated by the SEM technique (FE-SEM 4800). Noticeable changes
in the appearance of the paste morphology were observed in
CNS-added cement pastes.
The SEM images, Fig. 10, show the typical morphological
features of cement or cement paste at various ages. Cement grains
exhibit irregular shapes with flat surfaces covered by small debris.
After 1 h of hydration, for the plain cement paste, there are many
pits on the cement particle surface – these are due to the dissolu-
tion of the cement particle after meeting water [28]. Meanwhile,
some needle-like hydrates appear on the cement particle surface.
For the CNS-added paste, the surface is more seriously eroded,
indicating a greater extent of dissolution. However, the character-
istic needle-like hydrates are more difficult to see. A distinct
feature of CNS-added paste is that the cement particles are covered
with small particles that are ca. 50 nm. These can be the reacted
CNS particles, which are larger than their original size of 10 nm,
25
30
35
40
45
50
Temperature /
o
C
Hydration time / h
0% CNS
0.5% CNS
1.0% CNS
5.0% CNS
048121620
-4
0
4
8
Hydration time / h
1st. Dev.
048121620
048121620
25
30
35
40
45
50
Temperature /
o
C
0% SF
2.5% SF
5% SF
048121620
-4
-2
0
2
4
6
8
Hydration time/h
1st. Dev.
Hydration time/h
Fig. 7. Effects of CNS and SF on cement hydration heat.
0 40 80 120 160 200 240 280
12.60
12.65
12.70
12.75
12.80
pH
Time / min
0% CNS
5% CNS
w/b=2
17
18
19
Electrical conductivity
/ mS/cm
Fig. 8. Effect of CNS on pH value and electrical conductivity evolution of cement
paste.
048121620
25
30
35
40
45
50
5% CNS
Temperature /
o
C
Hydration time / h
Control
5% SF
048121620
-4
0
4
8
Hydration time / h
1st. Dev.
Fig. 9. Comparison of hydration acceleration effects of CNS and SF.
444 P. Hou et al. / Composites: Part B 45 (2013) 440–448