Construction and Building Materials 264 (2020) 120241

Contents lists available at ScienceDirect

Construction and Building Materials

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

Evaluating the effects of nanosilica on the material properties

of lightweight and ultra-lightweight concrete using image-based
approaches
Pawel Sikora a,b, Teresa Rucinska b, Dietmar Stephan a, Sang-Yeop Chung c,⇑, Mohamed Abd Elrahman a,d,⇑
a
Building Materials and Construction Chemistry, Technische Universität Berlin, Germany
b
Faculty of Civil Engineering and Architecture, West Pomeranian University of Technology Szczecin, Szczecin, Poland
c
Department of Civil and Environmental Engineering, Sejong University, South Korea
d
Structural Engineering Department, Mansoura University, Mansoura, Egypt

h i g h l i g h t s

3,

Lightweight aggregate concretes (LWACs) with a targeted density of 850 kg/m and 450 kg/m3 were developed.

Cement was replaced with 1, 2, 5 and 10% nanosilica (by mass of cement).

Shrinkage, transport, mechanical and microstructural (2D and 3D) properties of concretes were evaluated.

Nanosilica has a beneficial effect in improving the mechanical and transport properties of concretes.

Nanosilica significantly improves the pore characteristics of lightweight concretes.

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

Article history: This work is aimed at characterizing the effects of nanosilica (NS) on the properties of lightweight aggre-
Received 27 April 2020 gate concretes with different densities. Lightweight aggregate concrete (LWAC) and ultra-lightweight
Received in revised form 28 June 2020 aggregate concrete (ULWAC) with targeted oven-dry densities of 850 kg/m3 and 450 kg/m3, respectively,
Accepted 8 July 2020
were produced. The mixtures were modified by replacing cement with nanosilica, in concentrations of 1,
Available online 30 July 2020
2, 5 and 10 wt-%. For comparison purposes, control specimens containing either cement alone or cement
with silica fume (SF) were also produced. Their mechanical properties, including flexural and compres-
Keywords:
sive strengths and transport characteristics, were evaluated by measuring the water accessible porosity
Light-weight concrete
Ultra-lightweight concrete
and water absorption coefficients of the concretes. In addition, the thermal conductivity and drying
Mechanical performance shrinkage, being important parameters of lightweight concrete, were characterized. The pore structure
Drying shrinkage characteristics of the concretes were assessed using 2D and 3D image analysis techniques; namely, using
Porosity an automated air void analyser and micro-computed tomography (micro-CT), respectively. The experi-
Nanosilica mental results show that NS has a significant effect on improving the mechanical and transport proper-
micro-CT ties of lightweight concretes and that the efficiency of NS is much higher than that of SF. Moreover,
RapidAir depending on dosage, NS was found to have a negligible or decreasing influence on the drying shrinkage
Thermal conductivity
of concrete, after 28 days of curing. Microstructural studies confirmed that NS significantly affects the
Thermal insulation
pore characteristics of concretes, thus resulting in concretes with denser and stronger microstructures.
Ó 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://
creativecommons.org/licenses/by/4.0/).

1. Introduction

Lightweight concrete (LWC) is a widely used structural and

masonry element in the construction field. It attributes its popular-
⇑ Corresponding authors at: Building Materials and Construction Chemistry, ity to low density, as well as to substantial acoustic and thermal
Technische Universität Berlin, Gustav-Meyer-Allee 25, 13355, Berlin, Germany (M. insulation properties. The use of LWC decreases the total weight
Abd Elrahman); Department of Civil and Environmental Engineering, Sejong of a structure, where the mass of concrete is often considered to
University, Seoul 05006, Republic of Korea (S.-Y. Chung).
be responsible for more than half of the dead load of a construc-
E-mail addresses: [email protected] (S.-Y. Chung), mohamedattia@mans.
edu.eg (M. Abd Elrahman).
tion. According to EN 206-1 [1], the oven-dry density of

https://doi.org/10.1016/j.conbuildmat.2020.120241
0950-0618/Ó 2020 The Authors. Published by Elsevier Ltd.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
2 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241

lightweight concrete should be in the range of 800–2000 kg/m3. high pozzolanic activity [20–23]. Various authors have shown that
However, developments in recent decades have already made it NS has a much higher reactivity than conventional SF, with a lower
possible to produce concretes with a density lower than 800 kg/ dosage necessary to achieve comparable or better concrete proper-
m3, with this material often referred to as ultra-lightweight concrete ties [11,12,24–25]. While knowledge regarding the effects of NS on
(ULWC). Despite the beneficial properties of LWCs mentioned above, the properties of normal-weight cementitious composites is
this material has certain disadvantages, such as low mechanical widely established, only a few works are available on this topic,
strength and high water absorption, which are effects of noticeably indicating that NS can be useful in the production of LWCs with
increased porosity, induced by the porous nature of the aggregates advanced mechanical and durability-related properties.
used and by air entrainment. Moreover, LWC starts to shrink when As far as the authors of this paper are aware, only limited work
dried, leading to cracks from tensile stresses which are generated has been done in this field, with nanosilica dosages ranging from a
[2,3], as the concrete volume decreases significantly due to water in very low 0.1 wt-%, up to a very high 10 wt-%. The available studies
the small capillary pores being lost. The presence of aggregates hin- related to the production of LWAC containing NS are summarized
ders this contraction and restrains paste shrinkage, thus making the in Table 1.
concrete more stable. Accordingly, the use of weak and less rigid In a study by Zhang et al. [26], a low dosage of NS (from 0.1 to
aggregates with low elastic moduli reduces the effect, leading to a sig- 1 wt-%) was incorporated into LWAC. A significant improvement in
nificant increase in the volume of shrinkage, as in the case of LWCs both early (7 d) and standard (28 d) flexural and compressive
[4,5]. Moreover, the high amount of cement paste used in LWCs, either strength improvement was reported, with a small dosage of NS
for the purposes of workability or strength, increases shrinkage. (0.1–0.2 wt-%). The authors reported that exceeding the optimal
However, the water absorbed by lightweight aggregates (LWAs) dosage lead to a neutralization of the effects of NS, with specimens
reduces concrete shrinkage [6], with the LWAs thus acting as reser- exhibiting strengths comparable to that of the control specimen. A
voirs which compensate for the loss in moisture due to drying. slightly higher dosage of NS was evaluated in the work of Du et al.
Such so-called internal curing can reduce lightweight aggregate [27]. They evaluated structural LWACs which were incorporated
concrete (LWAC) shrinkage significantly [5]. In addition, the strong with 1 or 2 wt-% of NS. The authors reported early flexural and
bond between LWA and the cement matrix reduces length changes compressive strength improvements (after 1 d and 7 d) with cur-
caused by moisture movements or thermal effects [7]. A combina- ing, when NS was present in the mixture, although 28 d strength
tion of these parameters, as well as the wide variety of LWAs, make improvements were not reported. A decrement in water accessible
the shrinkage behavior of LWAC different from that of normal- porosity, water penetration depth, water sorptivity, as well as
weight concrete [2,7]. Also, mix composition has a strong influence higher resistance against chloride ion penetration, were also
on shrinkage cracking and the use of different types of aggregates reported for NS-modified specimens. In another work [13], the
results in very different concrete shrinkage behavior. It has been effects of 1, 2 and 3 wt-% of NS on the compressive strength devel-
reported that, as the modulus of elasticity of an aggregate opment, shrinkage and cracking sensitivity of LWACs were evalu-
decreases, shrinkage values increase. ated. Compressive strength improvement was found to have
In most lightweight concrete mixes a high content of cementi- occurred at every testing date, with the largest effect observed in
tious materials is used to achieve the required mechanical proper- early strength improvement. Moreover, no significant influence
ties; consequently a high hydration heat and subsequent high on the long-term shrinkage of LWAC was found, while evaluations
shrinkage occur [5]. Several supplementary cementitious materials of early cracking sensitivity showed that, with increasing NS
(SCMs), particularly ground granulated blast-furnace slag (GGBFS), dosage, the total LWAC cracking area decreased continuously.
fly ash (FA) or silica fume (SF), can be included to reduce these Bogas et al. [6] studied the influence of fine materials on the
harmful effects and to achieve the required strength, with mini- shrinkage of LWC and compared the results to normal-weight con-
mum negative influence on other properties. SF has proved to be crete (NWC). They found that the shrinkage of LWC is much higher
the most effective SCM for improving the performance of LWACs than that of NWC. In addition, LWC shrinkage is proportioned neg-
but it has a negative effect on drying shrinkage [4]. atively to aggregate volume. Their experimental results revealed
Although the effects of the commonly used SCMs have been rel- that the long term shrinkage of nanosilica is lower than that of sil-
atively well established regarding LWC performance, significant ica fume. A comparative study undertaken by Atmaca et al. [28]
developments in the field of concrete technology have recently showed that 3 wt-% of NS in LWAC is beneficial in improving com-
lead to the production of a whole variety of new admixtures and pressive strength, split tensile strength, water sorptivity and gas
additives [8–10]. permeability of high-strength LWAC. The effect of NS was more
Nanomaterials have garnered substantial interest from research pronounced in normal concrete than in concrete containing LWA.
and industry around the world in the last two decades. As nano- An interesting evaluation regarding the fresh and hardened prop-
sized admixtures possess spectacular surface reactivity, attributed erties of self-compacting structural LWAC, containing 1, 3, and
to their ultra-fine size and high specific surface area, they exhibit 5 wt-.% of nanosilica, has been presented in the work of Afzali
substantially higher reactivity than conventional SCMs and thus Naniz and Mazloom [29]. For comparison, they also produced a ref-
increase the performance of cement-based composites which uti- erence specimen with 10 wt-% of SF. By undertaking an extensive
lize them, at lower amounts [10]. Among the nanomaterials used evaluation of the fresh properties of concrete, they showed that
for the formulation of building materials, nanosilica (NS) has the replacement of cement with either NS or SF decreased the flu-
gained substantial interest, with a few commercial products idity of self-compacting lightweight concrete, thus necessitating a
already available on the market. The effects of NS on concrete are high dosage of superplasticizer (SP). Moreover, they reported that
comparable to that of silica fume, but due to its ultrafine size the presence of either SF or NS reduced bleeding in concrete. A
and high chemical reactivity, its performance is much better with higher NS dosage improved the mechanical performance of the
a lower amount of admixture required [11,12]. In general, the concretes. However, the best performance was obtained when a
improvement can be attributed to the acceleration of cement ternary mixture, containing both SF and NS, was produced. Similar
hydration (seeding-effect), the fast pozzolanic reaction, as well as conclusions, regarding the fresh properties of much lighter
to the optimized packing of the particles in the cement matrix (density < 1000 kg/m3) self-compacting LWAC, were recently
[13–15]. Beside, beneficial effect of NS on the the early age proper- reported by Abd Elrahman et al. [30]. Their study on the effects
ties of cementitious composites [16–19], it also makes a substan- of 1, 2, and 4 wt-% of NS on LWAC containing expanded glass
tial contribution to time dependant properties as a result of its aggregate showed that an increment in NS dosage leads to a
Table 1
Summary of available studies related to the incorporation of nanosilica in LWCs.

Ref. Dosage of Oven-dry density Tested properties Optimal Effects Type of LWA and w/c or w/b ratio
nanosilica [kg/m3] dosage [wt-%]
[wt.-%]
[26] 0, 0.05, 0.1, ~1000* Flexural strength (7, 28 d), compressive strength (7, 28 d) 0.1–0.2 Improved early and 28 days flexural and Expanded shale ceramsite

P. Sikora et al. / Construction and Building Materials 264 (2020) 120241

0.2, 0.5, 1.0 compressive strength w/c = 0.45
[27] 1, 2 1845–1900 Compressive strength (1, 7 and 28 d), water accessible 1 Early compressive strength improvement (1 d and Expanded clay spheres coarse aggregate
porosity, water sorptivity, water penetration depth, rapid 7 d), 28 d strength not affected. w/c = 0.42
chloride penetration test, chloride diffusion test, rapid Decreased water accessible porosity, water
chloride migration test penetration depth, water sorptivity and higher
resistance against chloride ion penetration
[13] 1, 2, 3 1800–1900* Compressive strength (3, 7, 28 d), long-term shrinkage, early 3 Early and final strength improvement, no Fly-ash clay ceramsite or shale ceramiste
cracking sensitivity significant influence on the long-term w/b = 0.35
shrinkage, total cracking area decreased
[29] 1, 3, 5 1800–1900* Compressive strength (28 d), split tensile 3–5 Strength improvements, improvement of ultrasonic Expanded clay aggregate
strength (28 d) and flexural strength (28 d), Ultrasonic pulse pulse velocity and electrical resistivity w/b = 0.35 and w/b = 0.45
velocity, electrical resistivity
[30] 1, 2, 4 915–955 Flexural strength (7, 28, 90 d) and compressive strength (7, 4 Improved of early and long term flexural and Expanded glass aggregate w/b = 0.40
28, 90 d), water accessible porosity, water sorptivity, compressive strength, decreased water accessible
microstructural properties, thermal conductivity porosity and water sorptivity, improved
microstructure and decreased total porosity of
cement paste
[31] 1,2,3 1075–1240 Compressive strength (1, 2,7, 28 and 91 d), water sorptivity, 2 Improved early and long term compressive Containing cenospheres and expanded
rapid chloride permeability test, water penetration, chloride strength, improved water absorptivity, water glass aggregates (up to 2 mm).
diffusion test, water accessible porosity accessible porosity and chloride diffusion, w/b = 0.32 and w/b = 0.24
[28] 3 2000–2100* Compressive strength (3, 7, 28 and 90 days), split-tensile 3 Compressive strength and split-tensile strength Cold-bonded aggregate (90% fly ash + 10%
strength (28 and 90 days). water sorptivity and gas improved in all ages tested. Decrement in water cement)
permeability sorptivity and gas permeability w/b = 0.35
[32] 5, 10 650–700 Compressive strength (7, 28 d), thermal conductivity 10 Improvement of 7 and 28 d compressive strength, Expanded recycled glass lightweight
no effect on thermal conductivity
[33] 10 900–1000* Compressive strength (7, 28 d), water absorption, sulfate 10 Improvement of 7 and 28 d compressive strength, Perlite, thermally
resistance decreased water absorption, decreased expansion expanded clay aliven, w/b = 0.35
due to sulfate attack

*Estimated values. Exact oven-dry density values were not provided in the manuscript and density range was therefore calculated indirectly, based on the available mixture formulas or measurements.

3
4 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241

significantly increased demand for superplasticizer. However, the was also used to improve concrete the strength development and
mixture produced in this case was more cohesive, thus necessitat- to compare its performance with nanosilica. Fine quartz sand
ing a lower amount of stabilizer. In the case of hardened properties, (0.1–0.3 mm) was used in the lightweight concrete mixes to
specimens containing > 1 wt-% of NS exhibited flexural and com- achieve the targeted density (Sand-Schulz GmbH, Berlin, Germany)
pressive strength improvements from the early days of curing. In , without using a high amount of active powder. Tables 2 and 3,
the work of Du [31], the author evaluated the effects of 1, 2 and respectively, show the physical properties of the fine materials
3 wt-% of NS on the properties of LWAC in the density range of and lightweight aggregates used.
1075–1240 kg/m3, showing that when the optimal amount of NS A commercially available colloidal silica suspension, Levasil CB8
is incorporated (2 wt-%), a substantial improvement in strength (Nouryon, Sweden), was used in this study. The liquid phase in the
development from, the first day hydration, can be observed, as well NS suspension was also considered to be a part of the mixing water
as lower water adsorption and chloride penetration. Higher used to prepare concrete (see Section 2.2). Details of the NS used in
dosages of NS on the properties of LWAC have been evaluated in this study are presented in Table 4 and have been described com-
the work of Yu et al. [32]. Namely, they examined the effects of 5 prehensively elsewhere [37].
and 10 wt-% of NS on the properties of ultra-lightweight aggregate
concretes (ULWACs) containing different amounts of cement (350, 2.2. Mixture design
400, and 450 kg/m3) and expanded recycled glass lightweight
aggregate. They found that specimens incorporated with 10 wt-% Two series of lightweight aggregate concretes, with two tar-
of NS exhibited approximately 20% higher compressive strength geted density classes of 450 kg/m3 and 850 kg/m3, were investi-
values than the control. Vargas et al.’s [33] work on the effects of gated in this study. The concrete mixture proportioning was
replacing cement with 10 wt-% of NS, on the compressive strength based on the packing density concept, with aggregate grading
of two types of LWACs, showed that the NS-incorporated speci- being an important parameter that can affect the properties of con-
mens exhibited similar 28-day strengths to that of plain cement. crete in both fresh and hardened states [35]. Optimization of the
They concluded that the mechanical performance of LWAC is ingredient grading improves the packing of the whole skeleton,
dependent only on the type of aggregate used. Moreover, due to enhances workability, reduces the amount of cement paste
refinement of the cement paste microstructure by nanosilica, the required to fill the voids between aggregate particles and conse-
NS-incorporated samples exhibited a lower volume of voids and quently improves compressive strength. Several models and for-
a decreased water absorption rate. mulas have been developed to optimize the granulation of solid
The literature review above shows that research in the field has materials [38]. To calculate mixture composition in this investiga-
focused on the evaluation of mechanical performance, along with tion, the model developed by Andresean and Andersen was applied
certain durability related issues. In contrast, less attention has been (Eq. (1)):
paid to drying shrinkage or the microstructural characterization of
lightweight concretes, with certain factors, such as NS content (low
Dq  Dqmin
PðDÞ ¼ ð1Þ
dosage vs. high dosage), yet to be comprehensively evaluated. In Dqmax  Dqmin
addition, knowledge in this field is mainly limited to the potential
where P(D) is the total fraction passing through a sieve with an
structural applications of higher density LWCs. In contrast, this
opening size D, Dmax is the maximum particle size, Dmin is the min-
study aims at shedding light on the design and performance of
imum particle size and q is the distribution factor [39,40].
lightweight and ultralightweight aggregate concretes modified
The distribution factor is an important parameter in regard to
with nanosilica, which might find further utilization in non-
grading characteristics, with a small q representing a high amount
structural applications, such as insulating concrete.
of fines and a high q pointing to a coarser mixture with a low fines
The study examines the influence of nanosilica addition, with
varied content ranging from 1 wt-% to 10 wt-% of cement replace-
ment, on the properties of lightweight aggregate concretes Table 2
(LWACs) with two target densities of 450 kg/m3 and 850 kg/m3. Properties of the fine materials used.
For comparison purposes, a specimen with SF was produced as a Material CEM III/A 42.5 N Silica fume Quartz sand
control mix. The paper also evaluates mechanical performance,
Specific density [g/cm3] 3.03 2.20 2.67
transport properties and drying shrinkage. Microstructural evalua- Surface area (Blaine) [cm2/g] 4180 200,000 780
tions, including 2D and 3D imagining techniques, are also applied
so as to characterize the effects of NS on the pore structures of
the ULWCs produced.
Table 3
Properties of the lightweight aggregates used.

2. Materials and methods Material LiaverÒ LiaverÒ LiaverÒ

0.5–1 mm 1–2 mm 2–4 mm

2.1. Raw materials Bulk density [kg/m3] 250 220 190

Particle density [kg/m3] 450 350 320
Crushing resistance [N/mm2] >2.6 >2.4 >2.1
Lightweight concrete is characterized by its low thermal con- Water absorption [wt. %] 15.4 15.8 14.4
ductivity. As such, cement with slow hydration should be used in
a mix design, to reduce thermal micro-cracking due to the evolu-
tion of heat from cement hydration. Slag cement CEM III/A
42.5 N, in accordance with EN 197-1 [34] and provided by Heidel-
bergCement AG (Heidelberg, Germany), was therefore used. Table 4
Expanded glass (Liaver) with different fractions (0.5–1, 1–2, and Colloidal silica properties.
2–4 mm) was used as a lightweight aggregate (Liaver GmbH &
Particle size* Solid content Density Viscosity pH
Co.KG, Ilmenau, Germany), due to its high strength/density ratio
10–140 nm 50 wt-% 1.4 g/cm3 8 cP 9.5
compared to other types [35]. Condensed silica fume, in accor-
dance with EN 13263-1 [36] (Sika GmbH, Stuttgart, Germany), *Based on TEM analysis.
 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241 5

content. A small q is not ideal in achieving ultra-lightweight con- density class. A concrete mixer (Zyklose mixer, Pemat, Germany)
crete with a density of 450 kg/m3, because of the high density of with a capacity of 50 L was used to manufacture the concrete. In
fine lightweight aggregates (0.5–1 mm), as compared to the den- the mixing process, aggregates, cement and silica fume were mixed
sity of coarser lightweight aggregates (2–4), as can be seen in in a dry state for 60 s. Water and superplasticizer were then added,
Table 2. What is more, fine Liaver (0.5–1 mm and 1–2 mm) has a after which mixing continued for an additional 60 s. In the case of
higher crushing strength than coarser Liaver (2.6, and 2.1 MPa, NS addition, the suspension was mixed together with mixing water
respectively). It is therefore important to select the q value in such and superplasticizer. The viscosity-enhancing admixture was
a way that a low target density, with the appropriate strength, can finally added, with everything being mixed for an additional
be achieved. For lightweight concrete mixes with a density of 2 min. Due to problems associated with vibrating fresh lightweight
850 kg/m3, a lower distribution factor (q) was used, in order to concrete, it was decided not to apply any vibration, and that the
increase the content of fine Liaver with higher strength and density concrete produced should be self-leveling. In order to reach this
and to reduce the coarse, low density and low crushing strength. goal the concrete mixture was designed in consistency class of
Moreover, the particle density of coarse Liaver aggregate is very F4/F5 (according to EN 206-1), corresponding to a slump flow
low and to achieve the targeted density of 850 kg/m3, a high diameter of about 500–600 mm. After mixing, concrete consistency
amount of high density cement, compared to the lightweight was checked by measuring the slump flow diameter, according to
aggregates, must be used. A small distribution factor (q) was there- EN 12350-5. The flow test was performed without dropping or
fore selected, so as to increase the content of higher density fine raising (jolting) as it might affect the stability of the mixture. More-
aggregate. Quartz sand was also added to the mixture to achieve over, after measuring the flow of the mixture, visual inspection of
the desired density. To improve the strength of the cement paste, each mix has been performed in order to satisfy that each concrete
as well as the bond between the aggregate and the cement matrix, mixture does not exhibit potential to bleeding or segregation. After
silica fume and nanosilica was used, as can be seen in Table 5. The achieving the required consistency class, 150  150  150 mm3
maximum lightweight aggregate size was set at 4 mm, since the and 100  100  100 mm3 cubical steel molds, as well as standard
crushing strength of larger aggregate is significantly lower. 40  40  160 mm3 prisms, were filled with fresh concrete without
compaction. All the samples were stored in a curing chamber with
2.3. Mix design controlled temperature and relative humidity of 21 ± 1 °C and 99%,
respectively, till demolding after 24 h.
Two groups of LWCs, namely lightweight aggregate concrete
(LWAC) with an initial density of 850 kg/m3 and ultra- 2.5. Testing methods
lightweight aggregate concrete (ULWAC), with an initial density
of 450 kg/m3, were produced. Cement was replaced with NS in a 2.5.1. Oven-dry density and thermal conductivity
proportion of 1, 2, 5 and 10 wt-%. Specimens containing nanosilica Oven-dry density and thermal conductivity were determined
were designated as NS1, NS2, NS5 and NS10, where the digit after curing the samples for 28 d. Oven-dry density was measured
describes the amount of NS. Control concretes (C) and concretes following the specifications of EN 12390-7. For this purpose, three
containing 10 wt-% of silica fume (SF), as a cement replacement, 15  15  15 cm3 samples of each type of concrete were oven-
were prepared for the purposes of comparison. To distinguish dried at 105 °C until reaching a constant mass. Afterwards, three
between different concrete densities, specimens were additionally mass measurements were performed with the mean value being
labelled with the letters L and U, meaning lightweight or ultra- selected as the dry density.
lightweight, respectively. Accordingly, UNS10 means ultra- For thermal conductivity measurements conforming to ISO
lightweight concrete (U) containing 10 wt-% of NS as a cement 22007-2, oven-dried specimens were tested with the transient
replacement. The mix proportions have been summarized in plane source method (TPS), using a Hot Disk measuring device
Table 1. (Göteborg, Sweden). The TPS method allows rapid, accurate and
non-destructive testing of the thermal parameters of various mate-
2.4. Sample preparation rials, including cement-based composites. Thermal conductivity
was measured using a Kapton-insulated sensor placed between
In the experimental part, 12 mixes were prepared and tested: 6 two identical 10  10  10 cm3 concrete specimens. Each mix
mixes of both LWAC and ULWAC. Cementitious material content, design was tested on three groups of specimens, with the mean
as well as aggregate grading and content, were constant for each value taken into consideration in each case.

Table 5
Concrete mixture compositions.

Sample Targeted density Mix proportions [kg/m3]

designation [kg/m3]
Cement Silica Nanosilica Quartz sand LiaverÒ LiaverÒ LiaverÒ Water SP* ST**
fume suspension 0.1–0.3 mm 0.5–1 mm 1–2 mm 2–4 mm
LC 850 480 – – 34 79.3 54.4 56.4 192 7.2 0.6
LSF 432 48 – 34 79.3 54.4 56.4 192 8.4 0.6
LNS1 474 – 12 34 79.3 54.4 56.4 186 7.5 0.6
LNS2 468 – 24 34 79.3 54.4 56.4 180 9 0.3
LNS5 456 – 48 34 79.3 54.4 56.4 168 9.5 –
LNS10 432 – 96 34 79.3 54.4 56.4 144 10.8 –
UC 450 200 – – – 51.4 60.6 98 140 5.0 0.7
USF 180 20 – – 51.4 60.6 98 140 6.0 0.7
UNS1 197.5 – 5 – 51.4 60.6 98 138 6.0 0.6
UNS2 195 – 10 – 51.4 60.6 98 135 6.5 0.5
UNS5 190 – 20 – 51.4 60.6 98 130 7.2 0.5
UNS10 180 – 40 – 51.4 60.6 98 120 7.5 0.5

*SP: superplasticizer, **ST: stabilizer.

6 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241

2.5.1.1. Water accessible porosity and water sorptivity. Water acces- to produce 15  15  1 cm3 slices. The samples were then thor-
sible porosity was measured using the water displacement oughly polished with polishing powders of different finenesses
method. Firstly, three 10  10  10 cm3 saturated concrete sam- and then dried. The polished samples were coated with a broad-
ples were surface dried, with their mass then determined. Next, tipped black marker. After the ink had completely dried, the sam-
the samples were dried at 105 °C till reaching a constant mass, ples were placed into a 55 °C oven and then removed and coated
after which their dry mass was measured. Porosity was calculated with white zinc paste (white pigment). After cooling, the excessive
from the difference between the dry and saturated masses, with paste was removed from the aggregate and cement mortar areas
the value expressed as a percentage. by dragging an angled razor blade across the surface. This pro-
For calculation of the water absorption coefficient, the water duced a polished concrete cross-section which was stained with
sorptivity test was performed on 40  40  160 mm3 sized speci- black ink and in which the voids were filled with a zinc paste,
mens. The test was performed using the partial immersion method allowing the Rapid Air device to distinguish between air voids
described in EN ISO 15148. Before the test, the specimens were and the concrete matrix. The RapidAir device automatically
kept in ambient conditions to stabilize their mass, with the sides scanned the sample surfaces and provided the air void parameters.
sealed with hot paraffin wax. Afterwards, specimens were placed Two specimens were tested for each type of concrete, with each
in a tank with a constant water level to allow a one directional sample tested twice. The surface in question had an area of
water uptake process. Three prisms of each type of concrete were 120  80 mm2, with the traverse from where the reading was
dedicated to this part of the experiment, with the weight of spec- taken having a length of 2413.5 mm.
imens determined at specified times up to 24 h. Afterwards, the
water absorption coefficient, which describes the ratio between 2.5.1.4. X-ray micro-computed tomography (micro-CT). A non-
the water quantity absorbed by the specimens per unit area and destructive approach – X-ray micro-computed tomography
the square root of time, was calculated. (micro-CT) – was utilized to obtain a detailed investigation of
microstructural characteristics. Micro-CT is a non-invasive method
2.5.1.2. Drying shrinkage. Drying shrinkage represents volumetric that can enable visualization and examination of the internal
changes due to the loss of water from concrete. In this investiga- microstructure of a target material, without damaging it [41,42].
tion, a standard Graf-Kaufmann method, following DIN 52450, The data obtained from micro-CT can be utilized for investigating
was used to determine the shrinkage behavior of lightweight con- pore and solid characteristics.
crete. Measurement started after demolding at 1 d and continued Fig. 1 presents the micro-CT imaging procedure used for classi-
till 28 d, to get an initial indication about volumetric changes in fying a specific target component from the original micro-CT
lightweight concrete. 40  40  160 mm3 concrete prisms were image. As shown in the 1st image of Fig. 1, a set of resulting images
used to determine changes in sample lengths. After casting, the of the specimens are generated using micro-CT; this two-
samples were stored in a controlled curing chamber with a relative dimensional (2D) image is composed of 500  500 pixels with a
humidity and temperature of 65% and 21 °C, respectively. Shrink- 29.8 lm pixel size. As can be seen in the figure, the 8-bit original
age was measured at 1, 3, 7 and 28 days. micro-CT image is expressed in greyscale with a number between
0 and 255 (i.e. a total of 256), which is determined according to the
2.5.1.3. RapidAir measurement – 2D imagining technique. A stan- relative material density. For example, the black represents pores,
dardized, 2D image-based technique on 15  15 cm3 cubic speci- while the light and dark grey represent the cement matrix and
mens, using a RapidAir 457 Automated-Air-Void-Analyzer aggregate solids, respectively.
(Concrete Experts International, Sweden) and following EN 480- For more effective characterization, target components such as
11:2005, was performed to evaluate the air void characteristics aggregates and pores were segmented using the multi-
of the hardened concretes. The samples were cut vertically from thresholding approach [43,44], as well as a modified watershed
the middle of the cube, using a water-cooled diamond blade saw, [35]. Using these methods, aggregate solids (2nd image) and pores

Fig. 1. Micro-CT imaging procedure for classifying target components from the original image, using multi-level thresholds and a modified watershed algorithm (Note: in the
images of the aggregate and pore parts, the white represents the target component, while the black is the background.)
 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241 7

(3rd image) were successively classified, making it possible to consequently higher dosages of superplasticizer were needed. In
effectively visualize the 3D microstructures of the aggregates and the preliminary experiments, all the concretes suffered segregation
pores, as shown in Fig. 1. In this study, pore characteristics such due to density differences between the cement paste and the light-
as porosity and pore size distribution were examined based on weight aggregate. The addition of stabilizer (viscosity-enhancing
the 3D microstructures obtained from the micro-CT imaging, as admixture) solved the problem of segregation efficiently, with
shown in the lower part of Fig. 1. In general, the matrix part in the mixture thus becoming more cohesive and homogeneous.
the 8-bit micro-CT image is represented in light grey, due to its Mixes with NS exhibit better stability than other mixes and there-
high solid density. The lightweight aggregates, however, are fore lower stabilizer dosages are needed, as mentioned in Table 5.
depicted in dark grey, because of their relatively sparse material Due to its very fine particle size, NS addition improves the stability
density, with the pores represented in black. The use of a multi- and viscosity of a fresh mix. However, consistency is thus nega-
threshold method [43,44] and a modified watershed algorithm tively affected and consequently higher dosages of superplasticizer
[35], allowed effective visualization of detailed pore shapes and were added to achieve the required consistency. It is clear from the
solid structures. The whole image can be classified into four cate- dosages added that LWAC needed more superplasticizer than
gories: pores within the matrix and the aggregates, as well as the ULWAC, due to the high cement content and the high viscosity of
solid parts of the matrix and the lightweight aggregates. The pores the LWAC mixture. Conversely, ULWAC needed more stabilizer
and solid characteristics related to pore structures were of main due to the low binder content in the mix resulting in a high possi-
concern in this study, because these are dominantant in determin- bility of segregation.
ing material properties.
The wall-thicknesses of the specimens were also computed
using their microstructures, so as to evaluate the effects of solid 3.2. Oven-dry density and thermal conductivity
characteristics on material properties. Wall thickness denotes the
thickness of the solid structure between pores, particularly Fig. 4 presents the experimental results of the oven-dry density
between the closest pores. In order to calculate this value, the cen- tests of different concrete mixes. The targeted densities in both
troid and radius information of all the pores were examined first. concrete series were achieved to a certain extent. For LWAC, it is
Based on the information obtained, the closest pore from another, clear that all the mixes had a density > 850 kg/m3, except for
selected pore was identified, with the wall thickness between the mix LNS10. The first three mixes (LC, LNS1, LNS2) had similar dry
pores computed by subtracting the radius of each pore from the densities, but when the NS dose was increased to 5 wt-% (LNS5),
distance between their centroids. This process was repeated for the density increased significantly > 900 kg/m3, due to the filling
all the pores, making it possible to obtain the wall thickness and better packing effects of the very fine NS particles. When the
distribution. dosage was increased further, to 10 wt-%, the density decreased
significantly to 792 kg/m3. This can be attributed to the high
agglomeration potential of NS. Furthermore, mixes with higher
3. Results and discussion NS contents exhibit higher cohesivity and viscosity and conse-
quently the air voids created cannot escape, mainly because com-
3.1. Mixture consistency paction is not applied. These factors combined resulted in a
significant decrease in the density of mixes with a high NS content.
The consistency of the fresh concrete was determined by mea- A similar trend was observed for the second group of ultra-
suring the slump flow diameter. As mentioned in Section 2.4., in lightweight concretes, although this was not the case for concrete
order to produce a self-leveling, homogeneous mix with a high fill- containing silica fume (LSF and USF), despite a high SF dosage also
ing ability, the targeted consistency class was set at F4/F5, which being used. This can be attributed to the lower specific surface area
corresponds to a slump flow diameter of about 500–600 mm. of SF, as compared to NS [25], which was also reflected in the
Exemplary images of ULWAC and LWAC consistency are presented amount of plasticizer needed (Table 3).
in Figs. 2 and 3, respectively. The water content of LWAC mixes The experimental thermal conductivity results show that con-
with a targeted density class of 850 kg/m3 is higher than what is crete density was mainly correlated to a material property: as
needed for ULWAC mixes with a lower density class (450 kg/m3), can be seen in Fig. 5, it increased with thermal conductivity.
due to differences in cement content. Superplasticizer was adopted ULWAC mixes therefore exhibited substantially lower thermal
to achieve the required consistency, as can be seen in Table 5. A conductivity when compared to the corresponding LWAC mixes.
consistency drop was observed with increasing NS content and Compared to dry density, mix composition and fine material prop-

Fig. 2. Images of the ULWAC in the fresh state (mix UC).

8 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241

Fig. 3. Images of the LWAC in the fresh state (mix LC).

3.3. Flexural strength and compressive strength after 28 days of curing

Fig. 6 presents the flexural strengths of two types of concretes

after 28 d of curing. It can be seen that ULWAC specimens
(Fig. 6a) exhibited lower flexural strengths than the corresponding
LWAC (Fig. 6b) mixtures, which is attributable to the increased
porosity and lower density of ULWAC. The flexural strength of
UC was less than a third of that of LC. It can be seen that the incor-
poration of silica fume (LSF, USF) in the mixture did not noticeably
affect the flexural strength of ULWCs, with values being compara-
ble. However, the incorporation of NS had a beneficial effect in the
development of specimens’ flexural strengths. Specimen UNS1
exhibited a flexural strength improvement of over 28% when com-
pared to pristine and SF modified specimens (UC and USF, respec-
tively). In the case of LWAC, the effect of NS on flexural strength
was more limited, with the flexural strength of LNS2 being only
10% higher than that of LC. Increasing the NS dosage to 5 wt-%
led to a gradual strength decrement. However, specimens still
exhibited strengths comparable to that of the control specimen.
Fig. 4. Experimental results of the dry density tests of the lightweight concrete In contrast, specimens with 10 wt-% of NS (UNS10 and LNS10) pos-
series. sessed significantly lower flexural strengths, which is probably cor-
related with the creation of NS agglomerates and general concrete
compaction problems. These results are in line with previous find-
ings: that NS can positively affect the flexural strength of compos-
ites by improving the bond between the cement matrix and
aggregates to a certain degree, although this effect is not as pro-
nounced as in the case of compressive strength [26,45].
Fig. 7 presents the compressive strengths of concretes after 28 d
of curing. Just as in the case of flexural strength, compressive
strength decreases significantly with density. The compressive
strength value of UC was less than one-fifth of that of the LC spec-
imens. Specimen UC exhibited a compressive strength of almost
3 MPa, while the compressive strength of pristine LC was around
15 MPa. It can be seen that SF-incorporated concretes exhibited a
noticeable effect on compressive strength development, which is
attributable to SF’s filler and pozzolanic activity. The compressive
strength of USF increased by 49%, as compared to the control
(UC). Strength improvement was also found in the case of LSF,
but only by 10%. It can therefore be seen that SF contributed signif-
icantly to the strength of both concretes, with the effect being
more pronounced in the case of ULWAC. This is most probably
attributable to the ability of SF to fill the voids within the cement
Fig. 5. Experimental results of the lightweight concrete series thermal conductivity
matrix, which is very loose and porous and more imperfect, as
tests. compared to concrete with a higher density. However, it can be
seen that the incorporation of NS in much lower dosages than that
erties have only a marginal influence on thermal conductivity; this of SF, led to higher strength improvements. In the case of ULWC
being a matter of the ratio between the pore volume and the solid (Fig. 7a), while a small amount (1 wt-%) did not contribute to
structure volume of concrete. However, pore size distribution and strength improvement, increasing the dosage (up to 5 wt-%) led
pore shape have only a secondary influence in this regard [43]. to noticeable strength increases. Accordingly, the strength of
 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241 9

Fig. 6. Flexural strength of ultra-lightweight (left) and lightweight concretes (right) after 28 d of curing.

Fig. 7. Compressive strengths of ultra-lightweight (left) and lightweight concretes (right) after 28 d of curing.

UNS5 increased by 61% and 9% compared to UC and USF, respec- evaluating transport properties, which are often correlated with
tively. Exceeding the NS dosage (UNS10) resulted in a decrement durability related issues. The results of water porosity, as well as
of strength, leading to the production of concrete with similar the calculated water absorption coefficient from water sorptivity
strength to that of the control specimen (UC). In the case of LWAC tests, are presented in Fig. 8. It can be seen that concrete samples
(Fig. 7b), a lower NS dosage was required in order to see beneficial with lower densities had significantly increased transport proper-
compressive strength effects. The best performance was seen in the ties as a result of highly increased porosity. The water absorption
case of specimen LNS2, followed by LNS1. In this type of concrete, coefficient and water accessible porosity values of ULWAC
the strength increased by 17% and 8%, as compared to LC and LSF, (Fig. 8a) were therefore much higher than in the corresponding
respectively. Further addition of NS resulted in a gradual decre- LWAC (Fig. 8b).
ment in strength, as a result of inappropriate compaction and prob- It can be seen that the replacement of cement with 10% of SF
lems with workability, which affected the internal composite significantly improved the microstructure of concretes by decreas-
microstructure of the composite. ing the water accessible porosity. However, in both types of con-
These results are in line with the findings of other researchers crete incorporation of NS resulted in the production of a more
[13,29–31]; namely, that there is an optimal amount of NS which impermeable microstructure, with NS-modified concretes thus
is beneficial in improving the mechanical properties of cementi- presenting superior transport properties, in comparison to the con-
tious composites. However, the improvement rate depends on trol and SF-modified specimens. Although exceeding the optimal
the density of the concrete. The present study also confirms the dosage of NS resulted in a decrement in the beneficial effect of this
fact that, due to the superior activity of NS as compared SF, a much admixture on water accessible porosity, the water absorption coef-
lower dosage of NS is required to produce a lightweight concrete ficient remained improved. This confirms that NS has a significant
with comparable or better mechanical performance than one role not only in filling the pores of concrete (thus decreasing the
incorporating SF. total porosity) but that it also refines the pore structure. As
reported by other researchers [46–48], the following phenomena
3.4. Water accessible porosity and water absorption coefficient occur when NS is present in a cementitious composite: disconnec-
tion of continuous pores from discontinuous ones, a subdivision of
The positive effect of NS on microstructural improvement in larger pores into smaller ones and modification of the shape and
lightweight and ultra-lightweight concrete can also be proven by tortuosity of pores. These result in an improvement in transport
10 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241

Fig. 8. Water accessible porosity and water absorption coefficient of ultra-lightweight (left) and lightweight concretes (right) after 28 d of curing.

properties, even if NS dosage is excessive, which can lead to a the aggregate and cement paste; its addition therefore hinders
decrement in some properties. the movement of water in concrete to a certain extent, thus reduc-
ing drying shrinkage [50]. This effect was observed in mixes with
the appropriate NS contents, but the influence clearly vanished
3.5. Drying shrinkage when the dose was increased to 10 wt-%. This was due to the prob-
ability of the agglomeration of very fine particles, which affects the
The investigation focused on drying shrinkage at an early age, efficiency of nanosilica negatively. Moreover, when compared to
due to the high probability of cracking associated with low tensile SF-modified specimens, concretes with NS exhibited lower shrink-
strength in the first weeks of hydration. The experimental results age drying values. The results of our study confirmed the findings
of drying shrinkage tests are presented in Fig. 9. The presented val- of Wang et al. [13], who reported this phenomenon in the case of
ues are the means of 3 measurements of each mix. All samples LWCs, albeit, with much higher concrete densities (Table 1).
were stored in a curing chamber under the same humidity and
temperature conditions, since shrinkage measurements are highly
affected by these two parameters. It is clear from the results that 3.6. Microstructural analysis
all the lightweight concrete mixes had drying shrinkages higher
than that of traditional concrete [49]. In the two series of light- 2D and 3D imagining techniques were utilized to evaluate the
weight concretes with different densities, several parameters such effects of NS on the microstructural characteristics of lightweight
as aggregate type and content, as well as cement content, influ- concrete. Our previous investigations [51,52], as well as the find-
enced the drying shrinkage significantly. ULWAC specimens ings of other researchers [53], have confirmed that both investiga-
showed lower shrinkage values than LWAC, which can be attribu- tion techniques show good correlation and accuracy and can thus
ted to the high paste/aggregate ratio of LWAC, as compared to be applied together to improve evaluation precision and to extend
ULWAC, in spite of the high w/b ratio of the second series (0.7), the measurement range. Although the standardized 2D image-
compared to the first (0.4). based method enables relatively easy measurement, especially
In both series, drying shrinkage increased gradually with time, when using the automated air void analyser, it requires relatively
as can be seen in Fig. 9. However, the addition of fine materials thin specimens which need to be polished afterwards. As such, por-
had a significant influence on drying shrinkage. It is clear that dry- ous specimens with low strength cannot withstand the preparation
ing shrinkage decreased with increases in NS content. NS improves process and therefore only LWAC specimens were measured with
transition zone characteristics and strengthens the bond between the RapidAir device in this study.
As described in Section 3.5.5, both pore and solid characteristics
were analysed using the segmented images obtained with micro-
CT. Figs. 10 and 11 show the inner pore structures of the specimens
used. The general distributions of the pores inside each specimen
can be identified in these images. In comparing the LWACs
(Fig. 10) and ULWACs (Fig. 11), it can be seen that, in general, more
clustered pores exist in the ULWAC specimens; this indicates that
lightweight concrete with a lower density tends to have more clus-
tered pores, which affect the mechanical properties of the materi-
als. For a quantitative measurement, porosity values and pore size
distributions of the specimens were computed based on the voxel
information from the 3D pore images. Since the voxel (pixel) reso-
lution in this study was 29.8 lm, only pores larger than this were
taken into consideration.
Fig. 12 shows the porosity measured from the pore structure
images in Figs. 10 and 11. Here, each pore was secured by the
watershed algorithm and assumed to be a sphere, for more effec-
Fig. 9. Experimental results of drying shrinkage tests of lightweight concrete series. tive pore size measurement. The results in this figure confirm that
 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241 11

Fig. 10. 3D pore images of ultra-lightweight aggregate concretes (ULWACs).

Fig. 11. 3D pore images of lightweight aggregate concretes (LWACs).

ULWAC specimens show larger porosity than LWAC specimens. with the RapidAir device (Table 6), with a wider measuring range
The porosity of the binder (matrix) was <8% in all cases, with the of up to 1 mm, showed a slightly different trend, with the lowest
aggregate porosity being dominant in determining the pore char- total porosity obtained by LNS2, followed by LSF and then LNS1.
acteristics of the lightweight aggregate concrete. As such, in the However, although the porosities in the two different measuring
case of lightweight aggregate concrete, most of the material char- ranges were slightly different, the overall differences between
acteristics and the properties related to pore structures correlate porosities obtained with both techniques were rather minor.
strongly with the pores in the aggregates. Micro-CT evaluation Although it is widely known that porosity is one of the critical fac-
showed that for both density levels, the specimens with silica fume tors affecting compressive strength [49,52,54], a direct relationship
(LSF and USF) had the lowest porosity in each category within the between the total porosity of specimens and compressive strength
measured range. Total porosity values obtained for only LWAC was not established in this study. As such, to get a more detailed
12 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241

Fig. 12. Measured porosity results of ultra-lightweight (left) and lightweight concretes (right), using micro-CT images.

Table 6
Selected parameters of air-voids measured with RapidAir.

Specimen LC LSF LNS1 LNS2 LNS5 LNS10

Porosity from RapidAir* [%] 32.59 31.16 31.86 29.51 35.33 35.75
Number of voids < 0.5 mm 6027 4906 5075 4518 5443 6727
Number of voids < 1.0 mm 6258 5229 5437 4826 5809 6928
Percent of total number of voids < 0.5 mm 95.7 93.0 92.0 92.4 92.6 96.7
Spacing factor [mm] 0.030 0.034 0.035 0.039 0.030 0.031

*Total porosity up to 1 mm.

Fig. 13. Pore size distributions of the ultra-lightweight (left) and lightweight concretes (right) obtained with micro-CT.

analysis of pore characteristics, the pore size distributions of the the specimens obtained with micro-CT evaluations. In this image,
specimens were also investigated, as shown in Fig. 13. However, the general distributions of the ULWAC are similar (Fig. 14a). The
as the figure shows, there were no clear differences between the wall thicknesses of the LWAC specimens (Fig. 14b) are relatively
pore size distributions of the specimens, regardless of material larger than those of the ULWAC specimens, which supports the
density or porosity. As a result, another description that can sup- notion that the strength of the LWAC was higher. In particular,
port the pore structures was needed to achieve a better under- the LNS2 specimen showed a larger wall thickness distribution;
standing of the material characteristics. this could explain its larger compressive strength, despite the lar-
Therefore, in addition to the pore structures, the solid charac- ger porosity. The wall thickness values tended to increase in spec-
teristics of the specimens were also examined. Since pore informa- imens containing SF and NS, with dosages above 2%; indicating
tion such as porosity and pore radius can be obtained from that proper use of NS can contribute to thicker wall thickness,
segmented CT images, generalized wall thickness, according to which affects the mechanical properties of cement-based
the number of voxels, can be computed using pore characteristics. materials.
Wall thickness was measured by calculating the distances between The micro-CT evaluations can also be confirmed by RapidAir
the pores’ centroids, while taking into account the radius of each measurements, by analysing the amount of voids counted by the
associated pore. Fig. 14 shows the wall thickness distribution of instrument as well as by the spacing factor value. Spacing factor
 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241 13

Fig. 14. Wall thickness distributions of the ultra-lightweight (left) and lightweight concretes (right) obtained with micro-CT.

can be simply defined as the distance from the center of a unit cell to the significant increment of viscosity associated with the
to the nearest air void surface [55]. RapidAir evaluations (Table 6) addition of NS to the mix, a lower amount of stabilizer (or even
revealed that incorporation of both SF and NS in the LWAC mix- none) is required to prevent the mixture from bleeding and
ture, resulted in concrete with less voids, despite total porosities segregation.
remaining relatively comparable; thus indicating that modification
A noticeable improvement in flexural and compressive
of the specimen skeleton (solid fracture) occurred to a certain strengths after 28 d of curing were reported in both types of
extent. The lowest amount of voids counted by the device was in concretes, when cement was replaced by NS. Moreover, these
specimen LNS2, followed by LSF and LNS1; this is in line with com- specimens exhibited better mechanical performance than SF-
pressive strength values (Fig. 7). The number of voids increased incorporated concretes. However, the NS dosage should be care-
again with NS dosage, which was reflected in a decrement in the fully adapted, in order to improve the selected properties of the
mechanical performance of these specimens. A similar pattern LWAC and ULWAC.
can be observed when analysing the spacing factor value, which
LWAC concretes exhibited higher shrinkage than ULWACs.
confirms that the distance between pores increased with NS Depending on the dosage, nanosilica was found to have a negli-
dosage up to 2 wt-%. It can thus be assumed that the solid content gible or positive effect on decreasing the drying shrinkage of
(thickness of walls) between the pores increased. As discussed ear- concretes with its effect more beneficial than the addition of
lier, although the total void amount decreased, the total porosity of SF to the mixture.
specimens remained comparable, thus implying that the size of
Both SF and NS have a significant effect on improving the trans-
pores should have increased, although this was not proven by port properties of concretes by refining the pore structure in the
micro-CT evaluations (Fig. 13). In analysing the amount of pores capillary range, thus producing a more impermeable cement
counted by RapidAir, it can clearly be seen that the percentage of matrix. Even when the ideal NS dosage is exceeded, selected
total number of voids higher than 0.5 mm (above the detection rate transport properties are still improved due to microstructural
of micro-CT) increased noticeably in the specimens, resulting in alteration.
improved mechanical performance. This could explain why, in
2D and 3D imagining techniques confirm that improvements in
the case of ULWAC, no particular trends were observed despite mechanical performance are attributable to the improvement of
improvements in compressive strength. In addition, due to the very void structures in concrete, as a result of increasing concrete
high porosity of ULWAC; differences in specimens’ pore character- wall thickness. Both SF and NS have a substantial effect in the
istics (including alterations in wall thickness) are either very subtle production of a more robust void structure in lightweight con-
or cannot be tracked, due to the limited resolution of the micro-CT crete. However, the effect of NS is more pronounced even when
device. a lower dosage of the admixture is used.
The investigations proved that, in comparison to SF, the incor-
The study shows that both 2D and 3D imagining techniques,
poration of NS in smaller dosages produces lightweight concretes when applied together, can provide significant insight into the
with a much more robust microstructure and improved mechani- microstructure of lightweight concretes, making it possible to
cal and transport properties. In addition, the study has shown that understand the phenomena underlying the mechanisms
both imaging techniques can be successfully applied to evaluate responsible for improving the mechanical and transport proper-
the pore characteristics of lightweight concretes and that they ties of LWCs.
show strong a correlation with other results, thus providing good
insight into the material characteristics of both ULWAC and LWAC. Funding

4. Conclusions This project received funding from the European Union’s Hori-
zon 2020 research and innovation program, under the Marie
The following conclusions can be drawn from this work: Skłodowska-Curie grant agreement No. 841592.

The incorporation of nanosilica (NS) in lightweight aggregate Declaration of Competing Interest

concretes has a significant effect on their fresh properties. Due
to the much higher specific surface area of NS, as compared to The authors declare that they have no known competing finan-
SF, NS-modified concrete requires a higher dosage of superplas- cial interests or personal relationships that could have appeared
ticizer to meet consistency requirements. At the same time, due to influence the work reported in this paper.
14 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241

Acknowledgement [21] T. Kropyvnytska, M. Sanytsky, T. Rucinska, O. Rykhlitska, Development of

nanomodified rapid hardening clinker-efficient concretes based on composite
portland cements, Eastern-Eur. J. Enterprise Technol. 102 (6/6) (2019) 39–48,
P. S. is Supported by the Polish Foundation for Science. The https://doi.org/10.15587/1729-4061.2019.185111.
authors would like to thank the German Academy of Exchange Ser- [22] K. Skoczylas, T. Rucińska, The effects of low curing temperature on the
properties of cement mortars containing nanosilica, Nanobuild 11 (5) (2019)
vices (DAAD), as well as the Egyptian Science and Technology
536–544, https://doi.org/10.15828/2075-8545-2019-11-5-536-544.
Development Fund (STDF) through the GE-Seed funding program. [23] S.PH. Jonbi, A. Andreas, A. Nugraha, The long-term effects of nano-silica on
This work was supported by the Korea Agency for Infrastructure concrete, IOP Conf. Ser. Mater. Sci. Eng. 508 (2019) 12038, https://doi.org/
10.1088/1757-899X/508/1/012038.
Technology Advancement (KAIA) grant, funded by the Ministry of
[24] M. Bolhassani, M. Samani, Effect of type, size, and dosage of nanosilica and
Land, Infrastructure and Transport (Grant 20NANO-B156177-01). microsilica on properties of cement paste and mortar, ACI Mater. J. 112 (2)
In addition, the authors would like to thank Paul H. Kamm (Helm- (2015).
holtz Centre Berlin) for his assistance in X-ray CT imaging. [25] H. Biricik, N. Sarier, Comparative study of the characteristics of nano silica -
silica fume - and fly ash - incorporated cement mortars, Mat. Res. 17 (3) (2014)
570–582, https://doi.org/10.1590/S1516-14392014005000054.
[26] P. Zhang N. Xie X. Cheng L. Feng P. Hou Y. Wu Low dosage nano-silica
References modification on lightweight aggregate concrete Nanomaterials and
Nanotechnology 8 2018 184798041876128 10.1177/1847980418761283
[27] H. Du, S. Du, X. Liu, Effect of nano-silica on the mechanical and transport
[1] EN 206-1, Concrete - Part 1: Specification, performance, production and
properties of lightweight concrete, Constr. Build. Mater. 82 (2015) 114–122,
conformity, Aug. 2000.
https://doi.org/10.1016/j.conbuildmat.2015.02.026.
[2] M. Aslam, P. Shafigh, M.Z. Jumaat, Drying shrinkage behaviour of structural
[28] N. Atmaca, M.L. Abbas, A. Atmaca, Effects of nano-silica on the gas
lightweight aggregate concrete containing blended oil palm bio-products, J.
permeability, durability and mechanical properties of high-strength
Cleaner Prod. 127 (2016) 183–194, https://doi.org/10.1016/j.
lightweight concrete, Constr. Build. Mater. 147 (2017) 17–26, https://doi.org/
jclepro.2016.03.165.
10.1016/j.conbuildmat.2017.04.156.
[3] S. Choi, Internal relative humidity and drying shrinkage of hardening concrete
[29] O. Afzali Naniz, M. Mazloom, Effects of colloidal nano-silica on fresh and
containing lightweight and normal-weight coarse aggregates: a comparative
hardened properties of self-compacting lightweight concrete, J. Build. Eng. 20
experimental study and modeling, Constr. Build. Mater. 148 (2017) 288–296,
(2018) 400–410, https://doi.org/10.1016/j.jobe.2018.08.014.
https://doi.org/10.1016/j.conbuildmat.2017.05.057.
[30] M. Abd Elrahman, S.-Y. Chung, P. Sikora, T. Rucinska, D. Stephan, Influence of
[4] M. Gesoğlu, T. Özturan, E. Güneyisi, Shrinkage cracking of lightweight concrete
nanosilica on mechanical properties, sorptivity, and microstructure of
made with cold-bonded fly ash aggregates, Cem. Concr. Res. 34 (7) (2004)
lightweight concrete, Materials (Basel) 12 (19) (2019), https://doi.org/
1121–1130, https://doi.org/10.1016/j.cemconres.2003.11.024.
10.3390/ma12193078.
[5] K.-H. Lee, K.-H. Yang, H.-S. Yoon, Shrinkage strains of lightweight aggregate
[31] H. Du, Properties of ultra-lightweight cement composites with nano-silica,
concrete using expanded bottom ash and dredged soil granules, Constr. Build.
Constr. Build. Mater. 199 (2019) 696–704, https://doi.org/10.1016/
Mater. 188 (2018) 934–945, https://doi.org/10.1016/
j.conbuildmat.2018.11.225.
j.conbuildmat.2018.08.168.
[32] Q.L. Yu, P. Spiesz, H.J.H. Brouwers, Ultra-lightweight concrete: Conceptual
[6] J.A. Bogas, R. Nogueira, N.G. Almeida, Influence of mineral additions and
design and performance evaluation, Cem. Concr. Compos. 61 (2015) 18–28,
different compositional parameters on the shrinkage of structural expanded
https://doi.org/10.1016/j.cemconcomp.2015.04.012.
clay lightweight concrete, Mater. Des. 1980–2015 (56) (2014) 1039–1048,
[33] P. Vargas, N.A. Marín, J.I. Tobón, Performance and microstructural analysis of
https://doi.org/10.1016/j.matdes.2013.12.013.
lightweight concrete blended with nanosilica under sulfate attack, Adv. Civil
[7] O. Kayali, M.N. Haque, B. Zhu, Drying shrinkage of fibre-reinforced lightweight
Eng. 2018 (2018) 1–11, https://doi.org/10.1155/2018/2715474.
aggregate concrete containing fly ash, Cem. Concr. Res. 29 (1999) 1835–1840,
[34] EN 197-1, Cement. Composition, specifications and conformity criteria for
https://doi.org/10.1016/S0008-8846(99)00179-9.
common cements, Oct. 2015.
[8] J. Silvestre, N. Silvestre, J. de Brito, Review on concrete nanotechnology, Eur. J.
[35] S.-Y. Chung, M. Abd Elrahman, D. Stephan, Effect of different gradings of
Environ. Civ. Eng. 20 (4) (2016) 455–485, https://doi.org/10.1080/
lightweight aggregates on the properties of concrete, Appl. Sci. 7 (6) (2017)
19648189.2015.1042070.
585, https://doi.org/10.3390/app7060585.
[9] L.A. Ivanov, K.E. Razumeev, E.S. Bokova, S.R. Muminova, The inventions in
[36] EN 13263-1, Silica fume for concrete. Definitions, requirements and
nanotechnologies as practical solutions, Part V. Nanobuild 11 (6) (2020) 719–
conformity criteria, Sep. 2005.
729, https://doi.org/10.15828/2075-8545-2019-11-6-719-729.
[37] P. Sikora, K. Cendrowski, M. Abd Elrahman, S.-Y. Chung, E. Mijowska, D.
[10] P. Sikora, M. Abd Elrahman, D. Stephan, The influence of nanomaterials on the
Stephan, The effects of seawater on the hydration, microstructure and strength
thermal resistance of cement-based composites-A review, Nanomaterials
development of Portland cement pastes incorporating colloidal silica, Appl.
(Basel) 8 (7) (2018), https://doi.org/10.3390/nano8070465.
Nanosci. (2019), https://doi.org/10.1007/s13204-019-00993-8.
[11] L. Varghese, V.V.L.K. Rao, L. Parameswaran, Nanosilica-added concrete:
[38] M. Abd Elrahman, B. Hillemeier, Combined effect of fine fly ash and packing
strength and its correlation with time-dependent properties, Proc. Inst. Civ.
density on the properties of high performance concrete: an experimental
Eng. Construct. Mater. 172 (2) (2019) 85–94, https://doi.org/10.1680/
approach, Constr. Build. Mater. 58 (2014) 225–233, https://doi.org/10.1016/
jcoma.17.00031.
j.conbuildmat.2014.02.024.
[12] V. Lincy, V.V.L.K. Rao, P. Lakshmy, A study on nanosilica- and microsilica-
[39] R. Yu, D.V. van Onna, P. Spiesz, Q.L. Yu, H.J.H. Brouwers, Development of ultra-
added concretes under different transport mechanisms, Mag. Concr. Res. 70
lightweight fibre reinforced concrete applying expanded waste glass, J. Cleaner
(23) (2018) 1205–1216, https://doi.org/10.1680/jmacr.16.00504.
Prod. 20 (2016) 690–701, https://doi.org/10.1016/j.jclepro.2015.07.082.
[13] X.F. Wang, Y.J. Huang, G.Y. Wu, C. Fang, D.W. Li, N.X. Han, et al., Effect of nano-
[40] P.K. Mehta, P.J.M. Monteiro, Concrete: Microstructure, Properties and
SiO2 on strength, shrinkage and cracking sensitivity of lightweight aggregate
Materials, 3rd ed., McGraw-Hill, New York, 2005.
concrete, Constr. Build. Mater. 175 (2018) 115–125, https://doi.org/10.1016/
[41] N. Bossa, P. Chaurand, J. Vicente, D. Borschneck, C. Levard, O. Aguerre-Chariol,
j.conbuildmat.2018.04.113.
et al., Micro- and nano-X-ray computed-tomography: a step forward in the
[14] G. Kotsay, Peculiarities of hydration of Portland cement with synthetic nano-
characterization of the pore network of a leached cement paste, Cem. Concr. Res.
silica, Select. Sci. Pap. J. Civ. Eng. 12 (2) (2017) 101–106, https://doi.org/
67 (2015) 138–147, https://doi.org/10.1016/j.cemconres. 2014.08.007.
10.1515/sspjce-2017-0025.
[42] S.-Y. Chung, J.-S. Kim, D. Stephan, T.-S. Han, Overview of the use of micro-
[15] M. Samy El-Feky, P. Youssef, A. Maher El-Tair, S. Ibrahim, M. Serag, Effect of
computed tomography (micro-CT) to investigate the relation between the
nano silica addition on enhancing the performance of cement composites
material characteristics and properties of cement-based materials, Constr.
reinforced with nano cellulose fibers, AIMS Mater. Sci. 6 (6) (2019) 864–883,
Build. Mater. 229 (2019), https://doi.org/10.1016/j.conbuildmat.2019.116843
https://doi.org/10.3934/matersci.2019.6.864.
116843.
[16] P. Sikora, D. Lootens, M. Liard, D. Stephan, The effects of seawater and
[43] S.-Y. Chung, M. Abd Elrahman, J.-S. Kim, T.-S. Han, D. Stephan, P. Sikora,
nanosilica on the performance of blended cements and composites, Appl.
Comparison of lightweight aggregate and foamed concrete with the same
Nanosci. (2020), https://doi.org/10.1007/s13204-020-01328-8.
density level using image-based characterizations, Constr. Build. Mater. 211
[17] C. Liu, X. Huang, Y.-Y. Wu, X. Deng, J. Liu, Z. Zheng, et al., Review on the
(2019) 988–999, https://doi.org/10.1016/j.conbuildmat.2019.03.270.
research progress of cement-based and geopolymer materials modified by
[44] Z. Yang, W. Ren, R. Sharma, S. McDonald, M. Mostafavi, Y. Vertyagina, et al., In-
graphene and graphene oxide, Nanotechnol. Rev. 9 (1) (2020) 155–169,
situ X-ray computed tomography characterisation of 3D fracture evolution
https://doi.org/10.1515/ntrev-2020-0014.
and image-based numerical homogenisation of concrete, Cem. Concr. Compos.
[18] P. Aggarwal, R.P. Singh, Y. Aggarwal, Use of nano-silica in cement based
75 (2017) 74–83, https://doi.org/10.1016/j.cemconcomp.2016.10.001.
materials—A review, Cogent Eng. 2 (1) (2015), https://doi.org/10.1080/
[45] J.-W. Han, J.-H. Jeon, C.-G. Park, Bond characteristics of macro polypropylene
23311916.2015.1078018.
fiber in cementitious composites containing nanosilica and styrene butadiene
[19] G. Land, D. Stephan, The influence of nano-silica on the hydration of ordinary
latex polymer, Int. J. Polym. Sci. 2015 (2015) 1–9, https://doi.org/10.1155/
Portland cement, J. Mater. Sci. 47 (2) (2012) 1011–1017, https://doi.org/
2015/207456.
10.1007/s10853-011-5881-1.
[46] J.I. Tobón, O. Mendoza Reales, O.J. Restrepo, M.V. Borrachero, J. Payá, Effect of
[20] Krivenko PV, Sanytsky M, Kropyvnytska T. The Effect of Nanosilica on the Early
pyrogenic silica and nanosilica on portland cement matrices, J. Mater. Civ. Eng. 30
Strength of Alkali-Activated Portland Composite Cements. SSP 2019;296:21–6.
(10) (2018) 4018266, https://doi.org/10.1061/(ASCE)MT.1943-5533.0002482.
https://doi.org/10.4028/www.scientific.net/SSP.296.21.
 P. Sikora et al. / Construction and Building Materials 264 (2020) 120241 15

[47] G. Quercia, P. Spiesz, G. Hüsken, H.J.H. Brouwers, SCC modification by use of [52] S.-Y. Chung, P. Sikora, T. Rucinska, D. Stephan, E.M. Abd, Comparison of the
amorphous nano-silica, Cem. Concr. Compos. 45 (2014) 69–81, https://doi.org/ pore size distributions of concretes with different air-entraining admixture
10.1016/j.cemconcomp.2013.09.001. dosages using 2D and 3D imaging approaches, Mater. Charact. 162 (2020),
[48] T. Meng, H. Yu, S. Lian, R. Meng, Effect of nano-SiO 2 on properties and https://doi.org/10.1016/j.matchar.2020.110182 110182.
microstructure of polymer modified cementitious materials at different [53] K.Y. Kim, T.S. Yun, J. Choo, D.H. Kang, H.S. Shin, Determination of air-void
temperatures, Struct. Concr. (2019), https://doi.org/10.1002/suco.201900170. parameters of hardened cement-based materials using X-ray computed
[49] A.M. Neville Properties of concrete 5th ed. 2012 Prentice Hall tomography, Constr. Build. Mater. 37 (2012) 93–101, https://doi.org/
[50] H. Du, S.D. Pang, High performance cement composites with colloidal nano- 10.1016/j.conbuildmat.2012.07.012.
silica, Constr. Build. Mater. 224 (2019) 317–325, https://doi.org/10.1016/ [54] H. Garbalińska, J. Strzałkowski, Thermal and strength properties of lightweight
j.conbuildmat.2019.07.045. concretes with variable porosity structures, J. Mater. Civ. Eng. 30 (12) (2018)
[51] S.-Y. Chung, M. Abd Elrahman, P. Sikora, T. Rucinska, E. Horszczaruk, D. 4018326, https://doi.org/10.1061/(ASCE)MT.1943-5533.0002549.
Stephan, Evaluation of the effects of crushed and expanded waste glass [55] Garboczi EJ, Bentz DP, Snyder KA, Martys NS, Stutzman PE, Ferraris CF et al. An
aggregates on the material properties of lightweight concrete using image- electronic monograph: Modeling and measuring the structure and properties
based approaches, Materials (Basel) 10 (12) (2017), https://doi.org/ of cement-based materials. Materials and Structural Systems Division,
10.3390/ma10121354. National Institute of Standards and Technology 2014.