Cement and Concrete Research 104 (2018) 68–79

Contents lists available at ScienceDirect

Cement and Concrete Research

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

Uncovering the role of micro silica in hydration of ultra-high performance T

concrete (UHPC)
⁎
Nam Kon Leea, K.T. Kohb, Min Ook Kimc, G.S. Ryub,
a
Department of Civil and Environmental Engineering, National University of Singapore, 1 Engineering Drive 2, 117576, Singapore
b
Structural Engineering Research Division, Korea Institute of Civil Engineering and Building Technology, 283 Goyangdae-Ro, Ilsanseo-Gu, Goyang-Si, Gyeonggi-Do,
Republic of Korea
c
Coastal Development Research Center, Korea Institute of Ocean Science and Technology, 787 Haean-no, Sangnok-gu, Ansan, Gyeonggi 15627, Republic of Korea

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

Keywords: This study aims to clarify the role of micro silica in hydration of UHPC. The behavior of two different types of
UHPC micro silica in UHPC was investigated with regard to their filler effect and pozzolanic reaction by using ana-
Micro silica lytical techniques; XRD, TG, 29Si and 27Al NMR spectroscopy, and MIP. The micro silica with high pozzolanic
Hydration product activity led to the higher level of Al-substitution for Si in C-S-H and the denser structure, thus increasing
Microstructure
compressive strength, whereas that facilitating filler effect limited the consumption of Ca(OH)2 even after high
Filler effect
temperature curing and led to the higher amount of AFm phase. In the UHPC showing high filler effect, addi-
tional hydration of C3S and C2S occurred at later ages, increasing the fraction of Q1 site in C-S-H and decreasing
the porosity in the pore diameter region below 10 nm. Consequently, a significant increase in the compressive
strength of this UHPC was achieved.

1. Introduction
Ultra-high performance concrete (UHPC) was recently defined as a
cement-based composite with a compressive strength exceeding
150 MPa [1]. UHPC is typically fabricated using cement, water, micro
silica, filler and fine aggregate as well as superplasticizer. The most
distinctive properties of UHPC are extremely low porosity and high
packing density, resulting from particles with widely distributed dia-
meters ranging from 0.1 to 1000 μm and a water-to-cement (w/c) ratio
as low as 0.20. The key feature in the material design of UHPC is the use
of amorphous silica which can be classified into silica fume, micro silica
and nano silica, etc., according to their particle size and physical
property. The distinct characteristics of UHPC such as workability and
mechanical strength are considerably dependent on physical properties
and types of amorphous silica used.
Amorphous silica, which consists solely or mostly of SiO2, plays a
role in improving the rheology and the mechanical performance of
UHPC due to their inherited properties (e.g. high purity, spherical
particle with very fine size, high pozzolanic activity [2,49,51] and high
specific surface areas) [3]. One of commonly used amorphous silica is
silica fume which is a by-product of industrial silicon production and it
has particle sizes in the sub-micrometer range. The generally known
behavior of amorphous silica in UHPC is twofold as follows. First is
pozzolanic reaction in which the incorporated micro silica reacts with
portlandite (Ca(OH)2, CH) resulted from the hydration of alite [4,5,50].
Second, the particle size of amorphous silica ranging in sub-micro-
meters is effective for filling the voids present among cement particles
and other constituent materials, termed as filler effect [4].
Cement matrix is significantly densified by the filler effect, since the
voids present in the matrix can be filled with undissolved silica particles
[6,7]. Another principle of filler effect is that nucleation of calcium
silicate hydrate (C-S-H) phases which are formed from alite hydration
occurs on the surface of filler materials [2,8]; this effect accelerates
cement hydration and is more precisely termed as the seeding effect
[9,10,11]. Incorporation of supplementary cementitious materials
(SCMs) such as amorphous silica in cement may act as nucleation sites
for the growth of C-S-H [17], and similar effect can be anticipated by
incorporating other mineral powders (e.g., quartz [13], rutile [14],
limestone [15,16], synthetic C–S–H [10,17]). In general, filler effect of
amorphous silica is accepted to accelerate the hydration of the clinker
phases (especially, C3A and C3S) at an early age, while pozzolanic re-
action of amorphous silica starts at a later stage and is enhanced with
pH and temperature [2].
It should be noted that pozzolanic reaction of amorphous silica with
CH to form C-S-H gel is limited due to the very low w/c ratio (< 0.25
by mass) of UHPC system, instead the role of amorphous silica

⁎
Corresponding author.
E-mail address: ryu0505@kict.re.kr (G.S. Ryu).

https://doi.org/10.1016/j.cemconres.2017.11.002
Received 1 June 2017; Received in revised form 20 October 2017; Accepted 6 November 2017
Available online 12 November 2017
0008-8846/ © 2017 Elsevier Ltd. All rights reserved.
N.K. Lee et al. Cement and Concrete Research 104 (2018) 68–79

predominantly facilitates the filler effect [18]. Even though filler effect Table 1
and pozzolanic reaction were found to be clearly inclusive of each other Chemical composition of binder materials used in this study.
[5], the filler effect is a more notable phenomenon in UHPC than in
(wt%) OPC Micro silica-I Micro silica-II
other concrete blended with SCMs. The incorporation of a large amount
of amorphous silica in UHPC mixture and its low w/c ratio can promote XRD
filler effect rather than pozzolanic reaction. Especially during early ages Rietveld analysis
(< 24 h), amorphous silica is anticipated to most dominantly behave as
SiO2 21.23 C3S 59.2 92.0 93.8
a filler in UHPC, which increases hydration rates by seeding effect Al2O3 5.64 C2S 21.9 0.53 0.38
[19,20], while pozzolanic reaction is probably negligible due to the Fe2O3 3.38 C4AF 9.2 0.97 0.25
relatively low alkalinity of the pore solution at this stage [2]. CaO 61.4 C3A 3.6 0.48 0.03
Some researchers experimentally demonstrated an enhanced hy- MgO 2.20 Gypsum 2.3 1.18 –
TiO2 0.23 Anhydrite 1.9 – –
dration degree of cement at early age by incorporating silica fume with
K2O 1.15 Others 1.8 0.70 0.02
small size particles and high surface area, which effectively increases SO3 2.25 0.23 0.07
the extent of C-S-H nucleation [11,21,22]. However, the surface area of Na2O 0.11 0.56 0.06
amorphous silica is not always correlated with the extent of the seeding Cl 0.06 0.14 –
ZnO 0.15 0.17 –
effect, and its reactivity is also a governing factor. For instance, if one is
ZrO2 – – 4.16
easily dissolved due to its high reactivity, its filler effect is less likely
and the pozzolanic reaction is to be dominant. On the other hand, an-
other one with relatively low reactivity but greater particle size can be 2. Experimental procedure
more favorable for the filler effect, since undissolved particles can play
a role of nucleation seeding. In a previous study, it was found that the 2.1. Materials
reactivity of three types of silica (Pyrogenic silica, silica fume and
Stoeber silica) had a close relation with the specific surface area, sur- Ordinary Portland cement (OPC, ASTM C150 Type I) used in this
face silanol group density, total content of silanol groups and solubility study was provided by Asia Cement Corp., South Korea. Two types of
in alkaline suspension [4]. It was revealed that the degree of pozzolanic amorphous silica (micro silica-I and -II) were supplied by Elkem Corp.,
reaction was higher for Stoeber particles (with higher amount of silanol South Korea and S. Corp., China, respectively. Silica powder with dia-
group) than for the less reactive pyrogenic silica and silica fume (with meters of 1–5 μm was used as a fine filler, and micro sand with dia-
lower surface silanol group density and lower total amount of silanol meters of 100–800 μm were used as a fine aggregate.
groups). In their later study, silica fume and pyrogenic silica was found The chemical compositions of these materials are listed in Table 1.
to accelerate the alite hydration by increasing the surface for nucleation Micro silica-I and -II showed very similar chemical compositions in
of C-S-H phases whereas Stoeber particles showed no accelerating effect terms of SiO2, Al2O3, Fe2O3 and CaO, while the micro silica-I contained
[5]. 1.18% of MgO and the micro silica-II had no MgO. It is noted that only
Filler effect is more likely to accelerate the hydration of aluminate the micro silica-II contained 4.16% of ZrO2. The specific gravity of
phases, which normally gives rise to a second or shoulder peak in a micro silica-I and -II were 2.20 and 2.50, respectively. The miner-
calorimetric curve after the main peak associated with the hydration of alogical composition of the anhydrous OPC determined by XRD Riet-
alite, rather than that of calcium silicate phases [2,23]. The additions of veld analysis is also listed in Table 1.
the fine materials, i.e., corundum and rutile, have much more sig- The maximum water adsorption/desorption contents of micro silica-
nificant impact on the aluminate reaction than on the silicate reaction, I were 56.8 (P/P0 = 0.9730) and 6.78 cm3/g (P/P0 = 0.0965), re-
as the hydration of the aluminate phases is often more sensitive to spectively, and those of micro silica-II were 10.8 (P/P0 = 0.9466) and
nucleation effects [23]. In other studies [11,24], it was found that the 2.03 cm3/g (P/P0 = 0.0857), as measured using a BELSORP-aqua3
high surface area of amorphous silica has a high capacity to adsorb (BEL JAPAN, INC.) instrument. This property may affect the level of
sulfate ions, and the decreased sulfate concentration accelerates the slump flow of UHPC by adsorbing some of the mix water in the fresh
formation of monosulfate. At the same time, the change in the sulfate state. The amount of superplasticizer required to obtain high slump
concentration also influenced the dissolution and formation of sulfate flow of UHPC (> 750 mm) can be varied depend on the type of micro
containing phases [11]. silica with different adsorption content. For instance, micro silica with a
To date, the filler effect and pozzolanic reaction of amorphous silica higher water adsorption content requires a higher superplasticizer
in UHPC have been investigated in a limited number of studies. In more amount to reach similar flow.
detail, the influence of the two different behaviors (filler effect and The surface area was measured by nitrogen adsorption using a
pozzolanic reaction) of micro silica on the hydration and microstructure Micromeritics ASAP2420 by degasifying at 300 °C and 1.333 Pa
of UHPC has been poorly understood. In this regard, this paper in- (10 μmHg) for 2 h. The surface areas of micro silica-I and -II were 23.5
vestigated the role of micro silica in hydration and strength of UHPC. and 9.42 m2/g, respectively. The micro silica-II used in this study
Two types of micro silica with different extent of pozzolanic reactivity showed the lower surface area than the minimum specific area (15 m2/
and filler effect were considered: one is silica fume and another one is g) for silica fume stated in ASTM C1240.
amorphous silica containing a minor quantity of zirconium. The water The particle size distribution of the raw materials was measured
adsorption/desorption rate, surface area, and particle size of the micro using a laser particle-size analyzer (CILAS 1090L, CILAS, FRANCE), and
silica were measured, and the correlation of the results with the hy- is shown in Fig. 1. The mean particle size of the micro silica-I (0.31 μm)
dration and strength development of UHPC was discussed. The role of was lower than that of the micro silica-II (0.93 μm).
micro silica in UHPC were investigated with regard to their filler effect
and pozzolanic reaction by using analytical techniques; X-ray diffrac-
tion (XRD), thermogravimetry (TG), nuclear magnetic resonance 2.2. Mixture proportions and samples preparation
(NMR) spectroscopy, and mercury intrusion porosimetry (MIP).
The mixture proportions of UHPC are provided in Table 2. A water-
to-binder (cement + micro silica) ratio was 0.187 and a shrinkage re-
ducing agent/cement ratio was 0.01. Superplasticizer-to-cement ratios
for UHPC incorporating micro silica-I and -II were 0.023 and 0.062,
respectively. The UHPC incorporating the micro silica-II required much

69
N.K. Lee et al.
Fig. 1. Particle size distribution of raw materials (micro silica-I and -II, silica powder
(SP), ordinary Portland cement (OPC), and silica sand (SS)).
Table 2
Mix proportions of UHPC with different types of micro silica (kg/m3).
Mixture Water OPC Micro silica Silica powder Micro sand
I II
M1U 187 800 200 240 880
(222)a
M2U 187 800 200 240 880
(200)a
a
The value indicates the total amount of water including the amount of water in the
superplasticizer.
lower amount of superplasticizer than that incorporating the micro si-
lica-I due to the lower water adsorption/desorption content of micro
silica-II. Steel fiber was excluded in the mix for the convenience of
sample preparation and for analysis of the hydration reaction and mi-
crostructure of the UHPC. Note that the UHPC samples for XRD, TG,
NMR and MIP analyses were synthesized without micro sand, while the
samples for compressive strength tests were fabricated as listed in
Table 2.
The procedure for sample preparation is as follows [52]. Anhydrous
OPC, silica powder, micro silica, and micro sand were dry-mixed in a
mixing bowl for 4 min. Water, superplasticizer, and SRA were added to
the mixing bowl slowly, and then the mixing was continued for another
3 min until a sufficient slump flow (above 750 mm) of UHPC was ob-
tained. The fresh UHPC was placed into a 50 × 50 × 50 mm cubic
molds slowly up to the upper face to prevent the occurrence of en-
trapped air. The mold was tapped sufficiently using a rubber hammer to
remove the air voids entrapped in the UHPC. The mold was im-
mediately covered with a poly-plastic vinyl sheet to prevent surface dry.
The UHPC samples were cured at a temperature of 20 °C and at
RH > 99% in a sealed condition for the initial 24 h (Stage I), and then
the 1 d compressive strength was measured. The samples were then
cured in a water bath at a temperature of 90 °C for 72 h (Stage II), and
then the 4 d compressive strength was measured. Thereafter, the sam-
ples were cured at a temperature of 20 °C in a water bath (Stage III),
and the 28 d compressive strength was measured. Descriptions of Stage
I to III are found in Table 6. The following notations are used to identify
samples throughout this paper: M1U and M2U indicate the UHPC
synthesized using the micro silica-I and the micro silica-II, respectively,
and the figure following the sample name indicates the curing time
(days). For example, M1U-4 means the UHPC incorporating the micro
silica-I cured for 4 d (i.e. ambient curing for first 1 day and then high
temperature curing for the following 3 d). M2U-28 indicates the UHPC
incorporating the micro silica-II cured for 28 d (i.e. ambient curing for
first 1 d, and then high temperature curing for the following 3 d, fol-
lowed by ambient curing for the following 24 d).
Cement and Concrete Research 104 (2018) 68–79
2.3. Test methods
The unconfined compressive strength test was conducted using a
300 kN universal testing machine in accordance with ASTM C39, and
the strength was averaged from three samples.
The samples for the XRD, TG, 29Si and 27Al MAS NMR, and MIP
analysis were mechanically grounded for the analysis at the designated
test day.
The XRD tests were performed using a SmartLab device (manu-
factured by Rigaku). A CuKα radiation at 45 kV and 200 mA, a step size
of 0.01° and 0.2 s per step were employed over a 2θ range of 5° to 70°.
The International Centre for Diffraction Data (ICDD) PDF database was
used for phase identification. The MIP tests were conducted using an
Autopore VI 9500 (manufactured by Micromeritics Instrument Corp.).
A pressure range of 0.2–413.7 MPa (30–60,000 psia) was employed
during the intrusion and extrusion of mercury. TG analyses were con-
ducted on a TA Instrument Q600 (pH 407, KBSI, Pusan Center) under a
N2 environment at a heating rate of 10 K/min up to 1000 °C.
Solid-state 29Si and 27Al MAS NMR spectra were acquired using an
INOVA 600 spectrometer (manufactured by Agilent Technologies, at
Super- KBSI Western Seoul Center) at 119.182 MHz and 156.320 MHz, re-
plasticizer spectively. The 29Si MAS NMR spectra were obtained using a 5 mm HX-
CPMAS probe and a 5 mm zirconia rotor by employing a spinning speed
50.5 of 10.0 kHz, a pulse width of 2.2 μs and a relaxation delay of 22 s. The
solid-state 27Al MAS NMR spectra were obtained using a 2.5 mm HX-
18.4 CPMAS probe and a 2.5 mm low-Al zirconia rotor by employing a
spinning speed of 22.0 kHz, a pulse width of 1.8 μs and a relaxation
delay of 2 s. The 29Si and 27Al chemical shifts were referenced to an
external sample of tetrakis (trimethylsiyl) silane at −135.5 ppm with
respect to TMS at 0 ppm and aqueous AlCl3 at 0 ppm, respectively.
3. Results and discussion
3.1. X-ray powder diffraction (XRD)
Fig. 2(a) shows the XRD patterns of the raw binder materials (i.e.,
OPC, micro silica-I & -II). Major crystalline phases present in the an-
hydrous OPC were C3S, C2S, C3A, C4AF, and gypsum. The XRD pattern
of the micro silica-I showed a diffuse band at 15–30° corresponding to
amorphous silica, while that of micro silica-II showed peak at 26.5° due
to the presence of quartz (SiO2, ISCD# 98-008-3849) and at 30.2°, 35.3°
and 50.2° corresponding to ZrO2 (ICSD# 98-006-6782), as well as a
diffuse band at 15–30°.
It was observed in Figs. 2(b) and (c) that there was no significant
difference between the XRD patterns of M1U-1 and M2U-1. The hy-
dration products of M1U-1 and M2U-1 were C-S-H, portlandite (Ca
(OH)2, ICSD# 98-005-1411), and ettringite (ICSD# 98-015-5395). The
presence of CH at 1 d indicates the low pozzolanic activity of both
micro silica-I and -II. The predominant reaction during an early age
(i.e., 1 d) was the hydration of C3S, C3A and gypsum.
Fig. 2(d) and (e) show the XRD patterns of M1U-4 and M2U-4. The
peaks associated with ettringite was absent in both patterns. The in-
tensity of peaks corresponding to CH was relatively lower in the XRD
pattern of M2U-4 in comparison with that of M2U-1, while those peaks
completely vanished in that of M1U-4. This implies that most of the CH
present in M1U-4 was consumed by the pozzolanic reaction of micro
silica-I during the 72 h of curing at a high temperature of 90 °C, while
the CH in M2U-4 was not fully reacted with the micro silica-II. This
clearly shows the different pozzolanic activity of the two micro silica in
the hydration of UHPC. The peaks corresponding to CH still persisted in
the XRD pattern of M2U-28, as shown in Fig. 2(f) and (g), meaning that
a significant amount of the micro silica-II remained unreacted in the
UHPC even after 28 d.
70
N.K. Lee et al. Cement and Concrete Research 104 (2018) 68–79

Fig. 2. XRD patterns of raw materials and UHPC (△: C3S, □: C2S, ○: CH, ☆: Ettingite, +: C2F, X: C3A, ◇: Quartz, G: Gypsum, C-S-H, Z: ZrO2).

71
N.K. Lee et al.
3.2. Thermogravimetry (TG)
Fig. 3(a) shows the derivative thermogravimetry (DTG) and TG
curves for M1U-1 and M2U-1 cured at 20 °C and RH > 99% for 24 h.
The DTG curves of both samples showed a weight loss at the
72
Cement and Concrete Research 104 (2018) 68–79
Fig. 3. DTG and TG results of (a) M1U-1 and M2U-1; (b) M1U-4 and
M2U-4; (c) M1U-28 and M2U-28.
temperature of around 100 °C due to the evaporation of free and phy-
sically bound water, the dehydration of C-S-H and ettringite. The
weight loss at this region can also result from the presence of third
aluminate hydrate (TAH) species which decomposes at 70–90 °C [25]
and will be discussed in Sec. 3.4 in detail. The weight loss at
N.K. Lee et al.
29
Fig. 4.
temperatures of around 150 °C was attributed to the dehydration of
AFm [26]. The peak centered at 420 °C was associated with the dehy-
droxylation of CH, and the peak between 620 and 710 °C was observed
due to the decarbonation of CaCO3. The presence of C-S-H, ettringite
and CH was observed in both M1U-1 and M2U-1, which is consistent
with the XRD results for M1U-1 and M2U-1.
The DTG and TG curves of M1U-4 and M2U-4 are shown in
Fig. 3(b). The sharp peak at 100 °C corresponding to the decomposition
of ettringite was reduced, due to the thermal decomposition which took
place during the high temperature curing, while the peak at around
150 °C associated with the interlayer water removal of AFm was ob-
served [26,27]. It should be noted that the weight loss at 420 °C cor-
responding to the dehydroxylation of CH was not identified in M1U-4.
The absence of CH in M1U-4 may be attributed to the occurrence of
Cement and Concrete Research 104 (2018) 68–79
Si MAS NMR spectra of UHPC.
carbonation or pozzolanic reaction during the high temperature curing.
However, the weight loss at the region between 620 and 710 °C asso-
ciated with the decarbonation of CaCO3 was not increased in M1U-4
compared to M1U-1, indicating that the carbonation of CH can be
considered negligible in M1U-4. Thus, it is likely that most of the CH in
M1U-4 was consumed by the pozzolanic reaction with micro silica-I to
produce C-S-H during the curing at a temperature of 90 °C. The peak
centered at 420 °C was observed in M2U-4 although the peak became
more diffuse than that in M2U-1, as seen in Fig. 3(b). The low pozzo-
lanic activity of the micro silica-II in the M2U samples might be due to
the poor dispersion and agglomeration of micro silica-II [28], however,
a measured slump flow (750 mm) of UHPC containing the micro silica-
II was as high as that of UHPC containing the micro silica-I despite the
fact that relatively lower amount of superplasticizer was used for M2U
73
N.K. Lee et al. Cement and Concrete Research 104 (2018) 68–79

Table 3
29
Deconvolution results for the Si MAS NMR spectra (%)a.

Sample C3S, C2S Q1 Q2(1Al) Q2b Q2p Q3(1Al) Q3 − 99 ppmb Q4 Al/Si ratioc MCLc,d
− 71 ppm −79 ppm − 81 ppm −83 ppm − 85 ppm − 90 ppm − 94 ppm − 110 ppm

M1U-1 36.7 9.6 6.7 2.0 8.6 1.9 0 0 34.5 0.12 10.1
M2U-1 36.2 9.3 0.4 3.1 6.3 0 0 0 44.8 0.01 6.2
M1U-4 31.2 11.2 12.3 3.5 21.9 5.1 2.4 1.5 13.1 0.09 16.0
M2U-4 29.6 9.7 5.5 2.2 13.0 2.8 1.5 1.1 34.7 0.08 11.6
M1U-28 28.6 10.2 11.5 3.4 22.1 5.6 2.2 1.4 14.9 0.10 17.9
M2U-28 29.7 15.3 5.1 2.1 9.9 2.6 0.7 1.5 33.1 0.07 7.5

3
2 ⎡Q1 + Q2 + Q2 (1Al) ⎤
MCLnc = ⎣ 2 ⎦
Q1 (1)
1 2
Q (1Al)
2
Al/Sinc =
Q1 + Q2 + Q2 (1Al) (2)
4[Q1 + Q2 + Q2 (1Al) + Q3 + 2Q3 (1Al)]
MCLc =
Q1 (3)
Q3 (1Al)
Al/Sic = 1
Q + Q2 + Q2 (1Al) + Q3 + Q3 (1Al) (4)
Q2 p − 2(Q3 + Q3 (1Al)) ≥0 (5)

The following was assumed to obtain the overall trend of the samples, which provided basis for the calculation of MCL.

Q2 p − 2(Q3 + Q3 (1Al)) = 0 (6)

Q2 (1Al) − 2Q3 (1Al) = 0 (7)

The hydration product of UHPC in this study is believed to be poorly ordered crystalline C-(A)-S-H, meaning that the use of C-A-S-H model for the calculation of MCL may not be adequate
to predict the absolute value. However, this assumption does not affect the overall trends observed in the calculated MCL values as long as the method applied is consistent.
a
Example of spectral decomposition for the UHPC sample is shown in Fig. 4(e).
b
The resonance at 99 ppm is attributed to silicon at Q3 or Q4 site in silica fume. The curing at the temperature of 90 °C led to the depolymerization of Q4 network of the silica fume by
hydrolysis [33], which is supported by the fact that M1U-1 and M2U-1 before curing at the high temperature showed no resonance at 99 ppm.
c
The MCL and Al/Si ratio for noncross-linked C-(A-)S-H (NC) and cross-linked C-(A-)S-H structures (C) containing Q3 and/or Q3(1Al) sites were calculated by Eqs. (1)–(2) [45] and Eqs.
(3)–(4) [46], respectively.
d
In general, the structural constraints related to Q2 and Q3 for the cross-linked C-(A-)S-H products are specified as Eq. (5) [46].

mixtures. Thus, it is reasonable to attribute the low pozzolanic reaction
of M2U samples to the low pozzolanic activity of micro silica-II, rather
than its poor dispersion. Meanwhile, the presence of CH in M2U-4 is in
close agreement with the XRD results for M2U-4, indicating that the
pozzolanic effect induced by the addition of micro silica-II was not
dominant in M2U samples.
Fig. 3(c) shows the DTG and TG curves for M1U-28 and M2U-28.
The peak associated with the decomposition of CH was still observed in
M2U-28. Despite the fact that all samples were cured in a water bath at
20 °C from the end of the high temperature curing to 28 days, the CH in
the M2U samples was not completely consumed by the pozzolanic re-
action with the micro silica-II, featuring a clear difference in the poz-
zolanic activity between micro silica-I and -II in the UHPC.
These two different behaviors of micro silica are possibly attributed
to its physical properties. Micro silica-II has a higher mean particle size
than micro silica-I, and has a lower surface area. This property of micro
silica-II may result in the lower level of pozzolanic reaction due to its
inherent lower reactivity than micro silica-I. It makes sense that, all
other things being equal, reactivity of micro silica increases as it is
finer. Meanwhile, micro silica-II contains 4.16% ZrO2 and micro silica-I
contains 1.18% MgO. The amount of ZrO2 and MgO in the two micro
silica are very few, and no hydration product related to these compo-
nents was identified through the XRD and 27Al NMR analysis. Thus, the
chemical composition difference between the two micro silica cannot
be a main reason for the different behaviors. Another reason could be
found in the literature [4]. The study showed that the reactivity of
micro silica was highly associated with surface silanol group density,
total content of silanol groups and solubility in alkaline suspension. The
higher amount of silanol group was, the higher degree of pozzolanic
reaction was. In summary, filler effect and pozzolanic effect of micro
silica are influenced both by its physical properties (i.e., particle size
and surface area) and the silanol group density, which can accordingly
affect the hydration of UHPC.
Note that the weight loss at a temperature of 300 °C was observed in
M2U-28, while not in M1U-28. This peak is assigned to dehydroxylation
of AFm phase [26,27,29]. The lower pozzolanic activity of micro silica-
II than that of micro silica-I may affect the presence of AFm. This me-
chanism will be elucidated in detail with aid of 29Si and 27Al NMR
results.
29
3.3. Si MAS NMR spectroscopy
Fig. 4(a) shows the 29Si NMR spectra of the raw materials (OPC,
micro silica-I and -II). The spectra of OPC showed narrow resonance at
− 71.7 ppm associated with Q0 site of belite (C2S) and broad resonance
between 66 and 78 ppm originating from alite (C3S) [30]. The spectra
of micro silica-I and -II showed a broad signal centered at around
− 110 ppm which is assigned to amorphous SiO2.
29
Si MAS NMR spectra of hydrated OPC are mostly dominated by
intense bands at −79.4 ppm, −83.5 ppm and − 85.3 ppm, which are
characteristic of silicate species in chain-end (Q1), bridging (Q2b) and
paired (Q2p) sites of C-S-H, respectively [32,48]. In addition, the re-
sonance at − 81 ppm corresponds to mid-groups, Q2(1Al) site, in which
Al species substitutes for Si in the bridging site of C-S-H.
The 29Si NMR spectra of M1U-1 and M2U-1 in Fig. 4(b) showed
resonance at − 72 and − 110 ppm, which correspond to the anhydrous
OPC and the unreacted micro silica, respectively. The resonance at
− 79 and −85 ppm in the spectra was observed in both samples, which

74
N.K. Lee et al.
Table 4
Degree of hydration of OPC and degree of reaction of micro silica calculated from the 29Si
NMR spectra (%).
Sample Degree of hydration of OPCa Degree of reaction of micro silicaa,b
M1U-1 43.2 6.1
M2U-1 44.0 0
M1U-4 51.7 100
M2U-4 54.1 5.0
M1U-28 55.6 100
M2U-28 54.0 15.0
a
It is calculated as follows [47,48]; Degree of hydration of OPC or micro silica = 1 − I
(t)/I(t0), where I(t) is the relative intensity for OPC or micro silica after the hydration time
(t) and I(t0) is the corresponding relative intensity for the anhydrous OPC or the raw
micro silica.
b
The reactivity of silica powder was assumed to be zero as it plays a role as a filler in
UHPC.
correspond to the Q1 and Q2 sites, respectively. The resonances at −81
and − 89 ppm assigned to Q2(1Al) and Q3(1Al), respectively, were also
observed with very low intensity. The intensity of signal at −110 ppm
corresponding to Q4(0Al) was lower in M1U-1 than in M2U-1, in-
dicating that the micro silica-I had higher solubility than micro silica-II.
However, despite the higher solubility of micro silica-I, it is worth re-
calling the fact that the extent to pozzolanic reaction with CH in both
samples at 1 d was found to be similar according to the DTG results in
Fig. 3(a).
After the curing at a high temperature of 90 °C, the intensity of
peaks at −85 and − 89 ppm associated with Q2 and Q3(1Al) sites, re-
spectively, in the hydration product C-(A)-S-H increased significantly in
both samples, as shown in Fig. 4(c). This observation may be attributed
to the relationship between elevated temperature and the silicate chain
of C-S-H [32,33]. Specifically, the frequency of cross-linking (additional
Q3 and Q3(1Al) sites) and aluminum fraction at Q2p or Q3 sites in the
silicate chains was found to increase with the temperature [33]. The
intensity at peaks assigned to Q2, Q2(1Al) and Q3(1Al) sites in the
spectrum of M1U-4 were much higher than that of M2U-4. It is rea-
sonable to assume that the addition of the micro silica-I with a higher
pozzolanic activity can lower a Ca/Si ratio of C-S-H in UHPC [2,34,35].
This promoted the substitution of Si by Al in the bridging site of C-S-H
as well as increased the MCL of C-(A)-S-H, as shown in Table 3, which is
consistent with previous observations [36,37,38]. Meanwhile, the
lower pozzolanic reaction of micro silica-II inherited a relatively higher
Ca/Si ratio of C-S-H compared to that of micro silica-I, decreasing the
extent of Al substitution for Si in C-S-H. Thus, the mean chain length
(MCL) was shorter than that in case of micro silica-II (see Table 3). The
correlation of increased chain length with high pozzolanic reaction in
the M1U samples showed good agreement with the 29Si NMR analysis
results of UHPC blended with 20% silica fume (by weight) in the lit-
erature [39]. The resonance at −110 ppm in the spectrum of M2U-4 in
Fig. 4(c) indicates that a significant amount of micro silica-II remained
unreacted due to its low pozzolanic activity, which is consistent to the
XRD and DTG results.
Fig. 4(d) illustrates the 29Si NMR spectra of M1U-28 and M2U-28.
Note that the intensity of signal at 79 ppm corresponding to Q1 site of C-
S-H was increased in M2U-28 compared to M2U-4. This may be due to
the further occurrence of pozzolanic reaction of micro silica-II and
hydration of C3S or C2S during the curing period between 4 d and 28 d.
In the DTG results (see Fig. 3(c)), there was no significant change in the
intensity of peak at 420 °C associated with the decomposition of CH
between M2U-4 and M2U-28, implying that the occurrence additional
pozzolanic reaction was unlikely. Thus, the increased intensity of signal
at 79 ppm is likely to be attributed to the further hydration of C3S and
C2S. The high pozzolanic reaction in M1U-28 led to a decrease in the
amount of CH and the Ca/Si ratio of C-S-H, which in turn can decrease
the pH of the pore solution [2,40]. The low pH may reduce the addi-
tional hydration of C3S and C2S, while the high pH resulting from the
75
Cement and Concrete Research 104 (2018) 68–79
presence of CH in M2U-28 is likely to induce the hydration of anhy-
drous OPC at later ages.
The signal at −81 ppm assigned to Q2(1Al) is an important para-
meter in distinguishing between filler effect and pozzolanic effect ex-
hibited by the two types of micro silica. The 29Si MAS NMR results
confirmed that the UHPC incorporating micro silica with high pozzo-
lanic activity (M1U-1, -4, and -28) showed a higher level of Al sub-
stitution for Si in C-S-H, in comparison to that with relatively lower
pozzolanic activity (M2U-1, -4, -28). It was found in a previous study
that the filler effect promotes the hydration reaction at early ages
(< 24 h) and results in a higher degree of hydration of aluminate
phases (C3A and C4AF) than that of calcium silicate phases (C3S and
C2S) [2,23]. Thus, it can be expected that Al3 + dissolved from alumi-
nate phases contribute to the formation of typical aluminum-bearing
hydrates (AFt, AFm and TAH) rather than the substitution for Si in C-S-
H, provided that the filler effect of micro silica takes place pre-
dominantly in UHPC. This is because higher filler effect, namely, less
pozzolanic effect of micro silica leads to a relatively higher Ca/Si ratio
of C-S-H and also the decrease in the level of Al substitution for Si in C-
S-H and the MCL. The apparent difference between the pozzolanic ac-
tivity of the micro silica-I and -II can lead to constituting different Al-
coordination environments, i.e., tetra-, penta- and hexa-coordinated,
noted as Al(IV), Al(V) and Al(VI), of aluminum species in hydration
products of UHPC, which will be discussed in Sec. 3.5 with Al NMR
results.
Table 4 presents the degree of hydration of OPC and the degree of
reaction of micro silica calculated from the 29Si NMR spectra. M1U and
M2U samples show a significant difference in the degree of reaction of
micro silica at the same days, which is consistent with the DTG results.
The degree of reaction of micro silica-I in the UHPC at 28 days was >
99% while that of micro silica-II was 15%.
3.4. 27
Al MAS NMR spectroscopy
Fig. 5(a) shows the 27Al NMR spectra of anhydrous OPC. The spectra
of anhydrous OPC showed very broad resonance between −10 and
90 ppm. The 27Al NMR spectra of M1U-1 and M2U-1 in Fig. 5(b)
showed a predominant peak at 12.5 ppm and a small shoulder at
10 ppm corresponding to Al(VI) in AFt and Al(VI) in AFm, respectively.
The peak at 5 ppm was observed with very weak intensity, which can be
attributed to the presence of amorphous aluminate hydrate in (asso-
ciated with) C-A-S-H (third aluminate hydrate, TAH) [41,42] or that
adsorbed on the C-S-H particles [25]. The spectra in Fig. 5(b) also
presented diffuse resonance between 50 and 75 ppm, which can be
assigned to Al(IV) adsorbed on C3S and C2S as well as to Al(IV) in C-S-H.
The broad resonance below 0 ppm in Z-1 may arise from possibly the Al
incorporation in ZrO2 [43].
The spectra of M1U-4 and M2U-4 in Fig. 5(c) showed the signals
centered at 68 ppm associated with the Al(IV) in C-S-H and at 10 ppm
and 5 ppm corresponding to Al(VI) in AFm and TAH, respectively. The
Al(VI) signal assigned to ettringite disappeared upon curing at a high
temperature of 90 °C in both SF-4 and Z-4 due to the thermal instability
of ettringite. The stability of ettringite can be affected by the two me-
chanisms. Firstly, its structural alteration occurs partially at 50 °C, and
completely at 80 °C [33], and secondly, pH of the pore solution of hy-
drated silica fume-blended OPC falls below 10 due to the consumption
of CH, which then makes ettringite unstable [2]. No resonance corre-
sponding to ettringite herein was observed in the 27Al NMR spectra of
M2U-4 and M2U-28, despite the presence of CH in M2U-4 and M2U-28
incorporating micro silica-II. Therefore, the decomposition of ettringite
was attributed to its thermal instability, rather than the reduced pH.
The Al species from the decomposed ettringite seemed to partially
substitute for Si in the bridging site of C-S-H and to contribute to the
formation of AFm and TAH. The signal related to Al(IV) substituting for
Si in C-S-H was higher in M1U-4 than M2U-4, while that corresponding
to Al(VI) in AFm and TAH was higher in M2U-4. This is attributed to the
N.K. Lee et al.
29
Fig. 5.
different role played by the micro silica-I and -II in UHPC. That is, the
micro silica-I in M1U-4 predominantly facilitated pozzolanic effect by
which most of the CH was consumed, as evidenced by the XRD, TG and
29
Si NMR results, while the filler effect was predominant in the micro
silica-II. The high pozzolanic activity of micro silica-I led to a relatively
low Ca/Si ratio of C-S-H, and thus the level of Al substitution for Si in C-
S-H increased, in particular, upon the decomposition of ettringite. The
relationship between a level of Al substitution for Si and a Ca/Si ratio of
C-S-H was also found in the 27Al NMR results from previous studies
[25,41,44]. In addition, the level of Al(IV) substitution in C-S-H af-
fected the Al(VI) in TAH; the intensity of signal corresponding to Al(VI)
in TAH was lower in M1U-4 accompanying with the higher level of Al
(IV) substitution in C-S-H, but the opposite was in M2U-4.
These results confirm that the Al binding sites, Al(IV) and Al(VI), in
the hydration products of UHPC are strongly dependent on the type of
micro silica used. The different behavior between micro silica-I (pre-
dominant with pozzolanic effect) and -II (predominant with filler ef-
fect), thus affected a Ca/Si ratio of C-S-H in UHPC. In case of micro
silica-I, a Ca/Si ratio of C-S-H was decreased by the high pozzolanic
effect of micro silica-I, thus allowing preferential substitution of Al(IV)
for Si in C-S-H. Meanwhile, a Ca/Si ratio of C-S-H in the M2U samples
was higher by the filler effect of micro silica-II, and the octahedrally
coordinated Al contributed to the formation of AFm and TAH [41,42].
This is consistent with the 29Si NMR results. As shown in Figs. 4(c) and
(d), the fraction of Q2, Q2(1Al) and Q3(1Al) sites was higher in M1U
samples containing the micro silica-I than in M2U samples containing
the micro silica-II. Similarly, it was also reported in a previous study
that Al mostly tends to be octahedral-coordinated (i.e., AFm, AFt, and
TAH [25,41,44]) when the Ca/Si ratio of C-S-H is higher than around
1.1, while it is likely to end up with tetrahedral coordination in C-A-S-H
at a lower Ca/Si ratio [42].
Cement and Concrete Research 104 (2018) 68–79
Al MAS NMR spectra of UHPC.
In Fig. 5(d), the spectra of M1U-28 and M2U-28 are compared to
those of M1U-4 and M2U-4, respectively. From 4 d (i.e. after high
temperature curing) to 28 d, the M1U and M2U samples were cured in a
water bath at an ambient temperature of 20 °C. During this period, the
microstructure of hydration products in the UHPC was not changed
significantly.
In summary of the Si and Al NMR results, the incorporation of micro
silica-I in M1U samples led to an increase in the MCL of SiO4 tetrahedra
in C-S-H and promoting the level of Al substitution for Si in the bridging
tetrahedral chains of C-S-H, compared to the M2U samples in-
corporating the micro silica-II. Effectively, the Al uptake in the C-S-H is
favored when higher pozzolanic reaction is achieved, as confirmed by
the Si & Al NMR results. In contrast, Al is mostly taken up by AFm and
TAH at the lower pozzolanic reaction where the chain length of C-S-H is
relatively shorter, accordingly a lower fraction of Q2(1Al) and Q3(1Al)
sites in C-S-H is observed.
3.5. MIP results
The structure of pores with diameters of < 1 um in the samples as
measured by MIP are strongly related to the dissolution of micro silica
and the formation of hydration products. Due to the micro silica-I and
-II possessing different pozzolanic effect (or filler effect), the dissolution
degree of the micro silica and the type and amount of hydration pro-
ducts varied accordingly; the 29Si and 27Al NMR results showed that
higher pozzolanic activity of micro silica led to the higher dissolution
degree of micro silica and a reduction in the amount of octahedrally
coordinated aluminum (Al(VI)) site in AFm phases. Thus, the property
of micro silica used can affect the overall pore size distribution of
UHPC.
Fig. 6(a) shows that the volume of pores with diameters of < 7 nm
76
N.K. Lee et al.
Fig. 6. Pore size distribution of UHPC as measured by MIP.
was higher in M1U-1, while that of pores with diameters between 7 and
100 nm was higher in M2U-1. The 29Si NMR results confirmed that the
amount of C-S-H was higher in SF-1, which led to the decrease in the
capillary pore volume and to the increase in the gel pore volume in
M1U-1.
After the high temperature curing (Fig. 6(b)), a similar pore size
distribution was observed in both samples. A noticeable difference
featured by these samples (M1U-4 and M2U-4) in comparison with
those (M1U-1 and M2U-1) at 1 d were that the volume of pores with
diameters of < 5 nm was increased, and that of pores with diameters
of > 5 nm was decreased significantly. As illustrated in the DTG curves,
the different amount of CH was present in the samples due to the dif-
ference in the pozzolanic activity, but the two samples showed similar
pore characteristics. The difference in the compressive strength be-
tween the two samples was not reflected by their pore characteristics
77
Cement and Concrete Research 104 (2018) 68–79
Fig. 7. Compressive strength results of UHPC.
(Fig. 7). On one hand, one may assume that pozzolanic reaction of
micro silica with CH would result in a denser cement matrix, and hence
reduced porosity. However, M2U-4 incorporating micro silica-II with
relatively lower pozzolanic activity showed the pore size distribution
analogous to M1U-4 which showed higher pozzolanic reaction. This can
attributed to the filler effect of micro silica-II, minimizing the voids
between particles in the UHPC.
Fig. 6(c) shows that the volume of pores with diameters between 4
and 20 nm was remarkably reduced in M2U-28 compared to M2U-4,
while this was not reduced in M1U-28. This suggests that the additional
hydration of UHPC occurred from 4 to 28 d to more extent in M2U-28.
This is likely to be due to the pore-filling effect of additionally formed
C-S-H with Q1 silicate units, as revealed in the 29Si NMR results
(Fig. 4(d)) where the intensity of Q1 site was increased from 4 to 28 d.
In addition, this hydration products seemed to contribute to an increase
in the compressive strength of M2U-28.
3.6. Different effect of two types of micro silica on compressive strength
development of UHPC
The compressive strength results are shown in Fig. 7. The com-
pressive strength of M2U sample at 1 d was 66.5 MPa and higher than
that of M1U (i.e., 59.7 MPa). This can be attributed to the seeding effect
of the micro silica incorporated in the M2U samples enhancing the
reactivity at an early age, whereas the higher dosage (6.2% by w.t of
cement) of superplasticizer used in the M1U samples may delay the
strength development. However, the 29Si NMR results showed that the
amount of C-S-H was lower in the M2U-1, meaning that the seeding
effect of silica fume was unclear at this stage. Accordingly, it is likely
that the higher dosage of superplasticizer in M1U had a significant ef-
fect on delaying the strength development. Based on the DTG result, the
amount of CH in both samples at 1 d was similar and it can be said that
pozzolanic reaction had no significant effects on the difference in
measured strengths at this age. On the other hand, the intensity at the
site corresponding to the raw micro silica was much lower in the 29Si
NMR spectra of M1U than M2U. It is attributed to the dissolution of the
silica fume in the pore solution rather than the occurrence of pozzolanic
reaction.
After curing at a high temperature, the compressive strength of
M1U-4 and M2U-4 was 156.2 and 131.6 MPa, respectively. This dif-
ference (i.e., 25 MPa) was noticeable considering the difference in the
strength at other ages, and can be attributed to the hydration and
pozzolan reaction during the high temperature curing. It is evident that
the silica fume in the M1U samples were effective in terms of pozzolanic
reaction, which facilitated higher strength, showing an agreement with
the 29Si NMR, TG and XRD results.
Note that higher compressive strength was observed in the M2U
N.K. Lee et al. Cement and Concrete Research 104 (2018) 68–79

Table 5 observed in its compressive strength, since high degree of hydration

Pore characteristics of UHPC as measured by MIP that can be attained with limited availability of water in the UHPC
mixture has already been reached during the second stage.
Sample Median pore Average pore Porosity (%) Total intrusion
diameter (nm) diameter (nm) volume (ml/g)
4. Concluding remarks
M1U-1 14.6 9.5 8.50 0.0389
M2U-1 21.1 10.8 9.32 0.0408
This study presented experimental results and discussions to un-
M1U-4 6.3 7.1 5.02 0.0229
M2U-4 5.2 6.4 5.80 0.0248 cover the role of micro silica in hydration of UHPC. The reaction
M1U-28 5 6.2 5.37 0.0246 characteristics of two different types of micro silica in UHPC were in-
M2U-28 4.6 6.5 3.73 0.0160 vestigated with regard to their filler effect and pozzolanic reaction by
using analytical techniques; X-ray diffraction (XRD), thermogravimetry
(TG), nuclear magnetic resonance (NMR) spectroscopy, and mercury
samples at 28 d. The M2U samples showed a strength increase of intrusion porosimetry (MIP). The following conclusions can be drawn
35 MPa from 4 to 28 d, while the strength increase of M1U corresponds from the results presented in this paper.
to 2 MPa only for the identical period. This is interesting results as
UHPC only shows the slight increase in compressive strength under the 1. The XRD and TG results showed that the micro silica-I mainly ex-
post-high-temperature curing process. The peak associated with dehy- hibited pozzolanic effect while filler effect was predominant in the
droxylation of CH in the DTG result even after 28 d indicates that the micro silica-II, clearly showing different roles being played by re-
pozzolanic activity of the micro silica-II was relatively lower. The spective micro silica.
strength increase in the M2U samples despite the lower pozzolanic 2. In the 29Si NMR results, the higher pozzolanic activity of micro si-
activity of the micro silica-II can be attributed to the formation of C-S-H lica-I in the UHPC led to the higher level of Al substitution for Si in
with Q1 coordination, which reduced the porosity, as shown in Fig. 4(d) the bridging site of C-S-H as well as increased the MCL of C-(A)-S-H.
and Table 5. The fraction of Q1 site of C-S-H was increased between 4 and 28 d in
The hydration behavior of UHPC can be divided into the three the UHPC incorporating the micro silica-II with higher filler effect
stages considering high-temperature curing typically applied to UHPC due to the further hydration of C3S or C2S during the period.
and the properties of micro silica used, as stated in Table 6. The first 3. The 27Al NMR results confirmed that the Al binding sites, Al(IV) and
stage is before high-temperature curing; the seeding effect and pozzo- Al(VI), in the hydration products of UHPC are strongly dependent on
lanic reaction of silica fume are relatively less apparent at this stage. the type of micro silica used. The Al substitution for Si in the C-S-H
The Al-containing hydrates are mostly AFt phases, and the compressive was favored when higher pozzolanic reaction is achieved, in con-
strength of UHPC is similar regardless of the micro silica used. The trast, the Al was mostly taken up by AFm and TAH at the lower
second stage is during the high-temperature curing. During the high- pozzolanic reaction where the chain length of C-S-H was relatively
temperature curing of the UHPC incorporating micro silica with rela- shorter.
tively higher pozzolanic activity, CH is mostly consumed through the 4. The volume of pores with diameters between 4 and 20 nm was re-
pozzolanic reaction of micro silica, and Al-substitution for Si in C-S-H is markably reduced from 4 to 28 d in the UHPC incorporating the
promoted. On the other hand, the UHPC incorporating micro silica with micro silica-II. It is due to the pore-filling effect of additionally
relatively lower pozzolanic activity inherits a higher amount of residual formed C-S-H with Q1 silicate chain, which resulted from the addi-
CH and lower compressive strength. In this case, the Al environment tional hydration of UHPC during the period. This hydration products
mostly consists of AFm and TAH. The third stage is during the room- seemed to contribute to an increase in the compressive strength of
temperature curing which takes place after the high-temperature curing M2U-28.
process. In the case of UHPC incorporating micro silica with relatively 5. The hydration behavior of UHPC can be divided into three stages:
lower pozzolanic activity, additional hydration of C3S and C2S occurs, before (I) and during (II) high-temperature curing, and post-high
increasing the amount of Q1 site in C-S-H and decreasing the porosity in temperature curing (III). At Stage I (during the initial 24 h), the
the pore diameter region below 10 nm. Consequently, a significant in- presence of AFt phase was predominant, and the compressive
crease in the compressive strength is achieved. In contrast, the UHPC strength of UHPC is similar regardless of the micro silica used. At
containing micro silica with higher pozzolanic activity undergoes lesser Stage II, the use of micro silica with relatively higher pozzolanic
extent of microstructural change, and hence no significant change is

Table 6
Hydration kinetics of UHPC according to three stages.

Stage Micro silica

Predominant Predominant
Pozzolanic effect Filler effect

I • No apparent pozzolanic reaction • Accelerated hydration by seeding effect a

(Before high-temperature curing) • Formation of AFt hydrate • Formation of AFt hydrate

II • Full pozzolanic reaction (Full consumption of CH) • Low pozzolanic reaction (Less consumption of CH)
(During high-temperature curing) • Complete decomposition of AFt due to thermal instability • Complete decomposition of AFt due to thermal instability
• Increase in the tetrahedrally coordinated Al substitution for Si in
bridging site of C-S-H
• Increase in the fraction of octaherally coordinated Al in
AFm and TAH
• Increase 3
in the fraction of Q site and the MCL of C-S-H
III • No further hydration • Further hydration of C S and C S
• Increase
3 2
1
(Post-high temperature curing) in the fraction of Q site of C-S-H
• Reduction in porosity and significant increase in
• No significant structural change and compressive strength
development
compressive strength

a
Seeding or pozzolanic effect was not evident at early ages (i.e. Stage I) in the present study. However, initiation of these effects at early ages is highly dependent on the properties of
amorphous silica used (i.e., reactivity, surface area, particle size) and the amount of superplasticizer used.

78
N.K. Lee et al.
reactivity led to the increase in Al-substitution for Si in C-S-H and
the denser structure, thus increasing the compressive strength. On
the other hand, the UHPC incorporating micro silica with relatively
lower pozzolanic activity showed that the Al environment mostly
consists of AFm and TAH. At Stage III, further hydration of C3S and
C2S occurred in the UHPC incorporating micro silica with relatively
lower pozzolanic activity, increasing the amount of Q1 site in C-S-H
and decreasing the porosity in the pore diameter region below
10 nm. Consequently, a significant increase in the compressive
strength was achieved.
Acknowledgments
This research was supported by a grant (13SCIPA02) from Smart
Civil Infrastructure Research Program funded by Ministry of Land,
Infrastructure and Transport (MOLIT) of Korea government and Korea
Agency for Infrastructure Technology Advancement (KAIA).
References
[1] A.E. Naaman, K. Wille, The path to ultra-high performance fiber reinforced concrete
(UHP-FRC): five decades of progress, in: M. Schmidt, E. Fehling, C. Glotzbach, et al.
(Eds.), Proceedings of 3rd Hipermat: International Symposium on UHPC and
Nanotechnology for High Performance Construction Materials, Kassel University
Press, Kassel, 2012, pp. 3–16.
[2] B. Lothenbach, K. Scrivener, R.D. Hooton, Supplementary cementitious materials,
Cem. Concr. Res. 41 (12) (2011) 1244–1256.
[3] T. Oey, A. Kumar, J.W. Bullard, N. Neithalath, G. Sant, The filler effect: the influ-
ence of filler content and surface area on cementitious reaction rates, J. Am. Ceram.
Soc. 96 (6) (2013) 1978–1990.
[4] T. Oertel, F. Hutter, U. Helbig, G. Sextl, Amorphous silica in ultra-high performance
concrete: first hour of hydration, Cem. Concr. Res. 58 (2014) 131–142.
[5] T. Oertel, U. Helbig, F. Hutter, H. Kletti, G. Sextl, Influence of amorphous silica on
the hydration in ultra-high performance concrete, Cem. Concr. Res. 58 (2014)
121–130.
[6] H.W. Krauss, H. Budelmann, Hydration kinetics of cement paste with very fine inert
mineral additives, Proceedings RILEM 79, Hongkong, 2011, pp. 58–65.
[7] A. Kumar, S. Bishnoi, K.L. Scrivener, Modelling early age hydration kinetics of alite,
Cem. Concr. Res. 42 (2012) 903–918.
[8] H.W. Krauss, H. Budelmann, Agglomeration und räumliche Anordnung von
Feinstoffpartikeln in Zementleim - physikalische Mechanismen und Einfluss auf die
Hydratation, in: F. Winnefeld, F. Deschner (Eds.), Proceedings of Tagung
Bauchemie, GDCH- Fachgruppe Bauchemie, Dübendorf, 2012, pp. 35–42.
[9] F. Deschner, F. Winnefeld, B. Lothenbach, et al., Hydration of Portland cement with
high replacement by siliceous fly ash, Cem. Concr. Res. 42 (2012) 1389–1400.
[10] J.J. Thomas, H.M. Jennings, J.J. Chen, Influence of nucleation seeding on the hy-
dration mechanisms of tricalcium silicate and cement, J. Phys. Chem. C 113 (2009)
4327–4334.
[11] G. Land, D. Stephan, The influence of nano-silica on the hydration of ordinary
Portland cement, J. Mater. Sci. 47 (2) (2012) 1011–1017.
[13] V. Rahhal, R. Talero, Early hydration of Portland cement with crystalline mineral
additions, Cem. Concr. Res. 35 (7) (2005) 1285–1291.
[14] W.A. Gutteridge, J.A. Dalziel, Filler cement: the effect of the secondary component
on the hydration of Portland cement: part 2: fine hydraulic binders, Cem. Concr.
Res. 20 (6) (1990) 853–861.
[15] B. Lothenbach, G. Le Saout, E. Gallucci, K. Scrivener, Influence of limestone on the
hydration of Portland cements, Cem. Concr. Res. 38 (6) (2008) 848–860.
[16] M. Antoni, J. Rossen, F. Martirena, K. Scrivener, Cement substitution by a combi-
nation of metakaolin and limestone, Cem. Concr. Res. 42 (12) (2012) 1579–1589.
[17] W.A. Gutteridge, J.A. Dalziel, Filler cement: the effect of the secondary component
on the hydration of Portland cement: part I. A fine non-hydraulic filler, Cem. Concr.
Res. 20 (5) (1990) 778–782.
[18] T. Oertel, F. Hutter, R. Tänzer, U. Helbig, G. Sextl, Primary particle size and ag-
glomerate size effects of amorphous silica in ultra-high performance concrete, Cem.
Concr. Compos. 37 (2013) 61–67.
[19] J. Hjorth, J. Skibsted, H.J. Jakobsen, 29Si MAs NMR studies of Portland cement
components and effects of microsilica on the hydration reaction, Cem. Concr. Res.
18 (1988) 789–798.
[20] H. Justness, P. Colombet, et al. (Ed.), Kinetics of Reaction in Cementitious Pastes
Containing Silica Fume as Studied by 29Si MAS NMR, Nuclear Magnetic Resonance
Spectroscopy of Cement-Based Materials, Springer, Berlin, 1998, pp. 245–267.
[21] J. Bjornstrom, A. Martinelli, A. Matic, L. Borjesson, I. Panas, Accelerating effects of
colloidal nano silica for beneficial calcium-silicate-hydrate formation in cement,
Chem. Phys. Lett. 392 (2004) 242–248.
[22] A.C.A. Muller, K.L. Scrivener, J. Skibsted, A.M. Gajewicz, P.J. McDonald, Influence
of silica fume on the microstructure of cement pastes: new insights from 1H NMR
relaxometry, Cem. Concr. Res. 74 (2015) 116–125.
79
Cement and Concrete Research 104 (2018) 68–79
[23] G. Le Saout, K. Scrivener, Early hydration of Portland cement with corundum ad-
dition, 16, Internationale Baustofftagung (ibausil). Weimar, Germany, 2006, pp.
409–416.
[24] H.J. Kuzel, Formation of AFm and Aft phases in hydrating Portland cements,
Proceedings of the 16th International Conference on Cement Microscopy,
Duncanville, Texas, 1994, pp. 125–136.
[25] X. Pardal, F. Brunet, T. Charpentier, I. Pochard, A. Nonat, 27Al and 29Si solid-state
NMR characterization of calcium-aluminosilicate-hydrate, Inorg. Chem. 51 (3)
(2012) 1827–1836.
[26] B.Z. Dilnesa, B. Lothenbach, G. Renaudin, A. Wichser, E. Wieland, Stability of
monosulfate in the presence of iron, J. Am. Ceram. Soc. 95 (10) (2012) 3305–3316.
[27] S.M. Leisinger, B. Lothenbach, G. Le Saout, C.A. Johnson, Thermodynamic mod-
eling of solid solutions between monosulfate and monochromate 3CaO∙ Al2O3 ∙ Ca
[(CrO 4)x(SO4)1-x]∙nH2O, Cem. Concr. Res. 42 (1) (2012) 158–165.
[28] M.C. Juenger, R. Siddique, Recent advances in understanding the role of supple-
mentary cementitious materials in concrete, Cem. Concr. Res. 78 (2015) 71–80.
[29] S. Medvešček, R. Gabrovšek, V. Kaučič, A. Meden, Hydration products in water
suspension of Portland cement containing carbonates of various solubility, Acta
Chim. Slov. 53 (2) (2006).
[30] J. Skibsted, H.J. Jakobsen, C. Hall, Quantification of calcium silicate phases in
Portland cements by 29Si MAS NMR spectroscopy, J. Chem. Soc. Faraday Trans. 91
(1995) 4423–4430.
[32] R.J. Myers, E. L'Hôpital, J.L. Provis, B. Lothenbach, Effect of temperature and
aluminium on calcium (alumino) silicate hydrate chemistry under equilibrium
conditions, Cem. Concr. Res. 68 (2015) 83–93.
[33] T.T.H. Bach, C.C.D. Coumes, I. Pochard, C. Mercier, B. Revel, A. Nonat, Influence of
temperature on the hydration products of low pH cements, Cem. Concr. Res. 42 (6)
(2012) 805–817.
[34] C.A. Love, I.G. Richardson, A.R. Brough, Composition and structure of C–S–H in
white Portland cement–20% metakaolin pastes hydrated at 25 °C, Cem. Concr. Res.
37 (2) (2007) 109–117.
[35] A.V. Girao, I.G. Richardson, R. Taylor, R.M.D. Brydson, Composition, morphology
and nanostructure of C–S–H in 70% white Portland cement—30% fly ash blends
hydrated at 55 °C, Cem. Concr. Res. 40 (2010) 1350–1359.
[36] E. L'Hôpital, B. Lothenbach, D.A. Kulik, K. Scrivener, Influence of calcium to silica
ratio on aluminium uptake in calcium silicate hydrate, Cem. Concr. Res. 85 (2016)
111–121.
[37] J.J. Chen, J.J. Thomas, H.F.W. Taylor, H.M. Jennings, Solubility and structure of
calcium silicate hydrate, Cem. Concr. Res. 34 (2004) 1499–1519.
[38] G.K. Sun, J.F. Young, R.J. Kirkpatrick, The role of Al in C–S–H: NMR, XRD, and
compositional results for precipitated samples, Cem. Concr. Res. 36 (2006) 18–29.
[39] I. Schachinger, H. Hilbig, T. Stengel, E. Fehling, Effect of curing temperature at an
early age on the long-term strength development of UHPC, 2nd International
Symposium on Ultra High Performance Concrete, 10 2008, March, pp. 205–213.
[40] J.E. Rossen, B. Lothenbach, K.L. Scrivener, Composition of C–S–H in pastes with
increasing levels of silica fume addition, Cem. Concr. Res. 75 (2015) 14–22.
[41] M.D. Andersen, H.J. Jakobsen, J. Skibsted, A new aluminum hydrate species in
hydrated Portland cements characterized by 27Al and 29Si MAS NMR spectroscopy,
Cem. Concr. Res. 36 (2006) 3–17.
[42] B. Lothenbach, A. Nonat, Calcium silicate hydrates: solid and liquid phase com-
position, Cem. Concr. Res. 78 (2015) 57–70.
[43] E. Zhao, S.E. Hardcastle, G. Pacheco, A. Garcia, A.L. Blumenfeld, J.J. Fripiat,
Aluminum-doped mesoporous zirconia obtained from anionic surfactants,
Microporous Mesoporous Mater. 31 (1) (1999) 9–21.
[44] G. Renaudin, J. Russias, F. Leroux, C. Cau-dit-Coumes, F. Frizon, Structural char-
acterization of C–S–H and C–A–S–H samples—part II: local environment in-
vestigated by spectroscopic analyses, J. Solid State Chem. 182 (2009) 3320–3329.
[45] I.G. Richardson, G.W. Groves, The incorporation of minor and trace elements into
calcium silicate hydrate (C-S-H) gel in hardened cement pastes, Cem. Concr. Res. 23
(1) (1993) 131–138.
[46] R.J. Myers, S.A. Bernal, R. San Nicolas, J.L. Provis, Generalized structural de-
scription of calcium–sodium aluminosilicate hydrate gels: the cross-linked sub-
stituted tobermorite model, Langmuir 29 (2013) 5294–5306.
[47] F. Brunet, T. Charpentier, C.N. Chao, H. Peycelon, A. Nonat, Characterization by
solid-state NMR and selective dissolution techniques of anhydrous and hydrated
CEM V cement pastes, Cem. Concr. Res. 40 (2) (2010) 208–219.
[48] J. Skibsted, M.D. Andersen, The effect of alkali ions on the incorporation of alu-
minum in the calcium silicate hydrate (C–S–H) phase resulting from Portland ce-
ment hydration studied by 29Si MAS NMR, J. Am. Ceram. Soc. 96 (2) (2013)
651–656.
[49] W. Huang, H. Kazemi-Kamyab, W. Sun, K. Scrivener, Effect of replacement of silica
fume with calcined clay on the hydration and microstructural development of eco-
UHPFRC, Mater. Des. 121 (2017) 36–46.
[50] A. Korpa, T. Kowald, R. Trettin, Phase development in normal and ultra high per-
formance cementitious systems by quantitative X-ray analysis and thermoanalytical
methods, Cem. Concr. Res. 39 (2) (2009) 69–76.
[51] H. Zanni, M. Cheyrezy, V. Maret, S. Philippot, P. Nieto, Investigation of hydration
and pozzolanic reaction in reactive powder concrete (RPC) using 29Si NMR, Cem.
Concr. Res. 26 (1) (1996) 93–100.
[52] N.K. Lee, K.T. Koh, S.H. Park, G.S. Ryu, Microstructural investigation of calcium
aluminate cement-based ultra-igh performance concrete (UHPC) exposed to high
temperatures, Cem. Concr. Res. 102 (2017) 109–118.