Cement and Concrete Research 131 (2020) 106017

Contents lists available at ScienceDirect

Cement and Concrete Research

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

Carbon nanotubes (CNTs) in ultra-high performance concrete (UHPC): T

Dispersion, mechanical properties, and electromagnetic interference (EMI)
shielding effectiveness (SE)
⁎ ⁎
Myungjun Junga, Young-soon Leeb, Sung-Gul Honga, , Juhyuk Moonc,d,
a
Department of Architecture & Architectural Engineering, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea
b
Department of Electronic Engineering, Kumoh National Institute of Technology, 61 Daehak-ro, Gumi, Gyeongbuk 39177, Republic of Korea
c
Department of Civil and Environmental Engineering, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea
d
Institute of Construction and Environmental Engineering, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea

A R T I C LE I N FO A B S T R A C T

Keywords: The incorporation of carbon nanotubes (CNTs) in Ultra-High Performance Concrete (UHPC) was investigated
Carbon nanotubes (CNTs) with regard to the dispersion, mechanical properties, and electromagnetic shielding effectiveness (SE). Dispersed
Dispersion CNT solution (0–2.0 wt%) was prepared by sonication and subsequent shear mixing with superplasticizer to
Mechanical properties achieve the high flowability for UHPC slurry. The dispersion method was effective to manufacture CNT-UHPC
Shielding effectiveness (SE)
composite up to the critical incorporation concentration (CIC). Conducted experiments revealed that CNTs below
Ultra-high performance concrete (UHPC)
the CIC improve the mechanical properties of UHPC through pore filling, bridging, and a denser C–S–H structure.
CNTs substantially improve the electrical conductivity and resulting SE up to the percolation threshold. Two
methods of ASTM D4935-18 and IEEE-STD-299 with electromagnetic compatibility (EMC) theory verified the
result and suggested that larger specimen with sufficient incident area should be fabricated to accurately
evaluate the SE of actual structures. Otherwise, the ability of SE of CNT-UHPC composites could be biased.

1. Introduction important infrastructure such as dams, power plants, and the military.
These facilities require not only radio wave shielding capability but also
The development of carbon nanotubes (CNTs) has given various excellent physical protection to withstand vehicle or aircraft crashes,
engineering fields a new avenue for realizing innovative materials be- explosions, etc.
cause of their excellent mechanical, thermal, and electrical properties Thus, various works are in progress for developing cementitious
[1–3]. Among their many applications, CNTs can be used in construc- materials with not only improved mechanical properties but also EMI
tion engineering for nanoscale reinforcement to improve cementitious shielding effectiveness (SE). Most previous studies improved the EMI SE
materials at a significantly low dosage. Multiple studies have reported of cementitious materials by using conductive fillers such as carbon
on the improved mechanical properties and durability of construction materials, steel fibers, nickel powder, and steel slag. In particular,
materials reinforced with CNT [4–7]. CNTs also have excellent elec- carbon materials such as CNTs, graphene, and carbon black have been
trical conductivity, which has not been fully utilized in construction reported to be effective conductive fillers for EMI shielding because of
materials. Lately, the importance of electromagnetic interference (EMI) their extraordinary electrical conductivity. For example, CNTs exhibits
shielding is becoming highly emphasized in the construction field be- an electrical conductivity > 1000 times greater than that of copper
cause of the remarkable growth in applications of radio communica- [13]. Previous research on incorporating carbon materials in cement-
tions and electronic devices. EMI shielding mainly refers to the reflec- based binders and the resulting EMI SE properties is summarized below.
tion or absorption of electromagnetic (EM) waves by a conductive Nam et al. reported SE results for cementitious composites with various
material [8]. Long-term exposure to EM waves can directly affect the amounts of multi-wall CNTs (MWCNT) (0–1.5 wt%) at frequencies of
human body and promote the growth of tumor [9–12]. In addition, 45 MHz to 18 GHz. The SE of a composite with 1.5 wt% MWCNT was
invasive powerful EMI such as an electromagnetic pulse (EMP) can 2.6 dB at a frequency of 2 GHz [14]. They also analyzed the influence of
cause devices to malfunction or jam and paralyze security facilities and the CNT dispersion on the SE of CNT-cementitious composites and

⁎
Corresponding authors.
E-mail addresses: [email protected] (S.-G. Hong), [email protected] (J. Moon).

https://doi.org/10.1016/j.cemconres.2020.106017
Received 22 August 2019; Received in revised form 4 February 2020; Accepted 25 February 2020
0008-8846/ © 2020 Elsevier Ltd. All rights reserved.
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

maximized the enhancement at 0.6 wt% CNT and 20 wt% silica fume Table 1
[15]. A cementitious composite with carbon fibers (0.5 wt%) and MWCNT properties.
carbon black (0.5 wt%) was found to have a SE of 13 dB at a frequency Avg. diameter Density (g/m3) Length (μm) Purity Specific surface
of 1GHz [16]. Singh et al. found that a cement matrix with 15 wt% (nm) (wt%) area (m2/g)
MWCNT had a SE > 27 dB at frequencies of 8.2–12.4 GHz [17]. Pratap
10–20 100,000–150,000 10–70 > 95 165–205
et al. incorporated graphene oxide (30 wt%) along with an appropriate
amount of ferrofluid in cement paste and achieved a SE of 46 dB at
frequencies of 8.2–12.4 GHz [18]. Wang et al. investigated the EM wave
mechanical properties of composites were evaluated to determine how
absorption properties of a MWCNT–mortar composite with varying CNT
the dispersed CNTs affect the various characteristics of UHPC. Then, the
contents at frequencies of 2–8 GHz and 8–18 GHz and reported that a
SE of the composites was investigated through measurements of the
mortar with 0.6 wt% MWCNT had a SE of 28 dB at a frequency of
electrical conductivity and SE testing. Two different methods of SE
2.9 GHz, while a mortar with 0.9 wt% MWCNT had a SE of 10 dB at
testing were considered: ASTM D4935-18 (coaxial TEM cell method),
7.1 GHz [19]. Experiments and simulations showed that the EMI SE of a
and IEEE-STD-299 (i.e., representative method for large specimen
15 cm thick mortar with 3 wt% MWCNT was > 50 dB at frequencies of
testing). The results were interpreted based on electromagnetic com-
0.75–1.12 GHz [20]. Several other studies have reported similar results
patibility (EMC) theory.
on the application of carbon materials to the development of con-
ductive cementitious materials for EMI shielding [21–25].
Although there have been some successful applications of carbon- 2. Experimental details
based materials to cementitious binders, achieving both effective con-
ductivity and superior mechanical properties remains an engineering 2.1. Materials and mix proportion
challenge. This is because effectively dispersing conductive filler ma-
terials in a structural binder requires a high solution ratio, which is Specimens of UHPC with CNTs were made from MWCNT (Kumho
detrimental to the development of mechanical strength in typical con- Inc.), Type I Portland cement (Union Cement Co., Ltd.), silica fume
struction materials. Therefore, previous studies on carbon-based ce- (Grade 940U, Elkem), silica powder (S-SIL 10, SAC) and silica sand.
mentitious composites for EMI SE reports have reported limited com- MWCNT was fabricated through the chemical vapor deposition (CVD)
pressive strength of 62 MPa with a water-to-cement ratio (w/c) of 0.35 growth method. The type I Portland cement and silica fume were used
[16]. However, superior mechanical performance is especially im- as binders. Silica sand with a particle diameter of 0.2–0.3 mm and silica
portant for the application of EMI shielding to critical infrastructures. powder with a median particle size of 4.2 μm were used as the fine
Ultra-High Performance Concrete (UHPC) exhibits excellent durability aggregate and physical filler, respectively. Tables 1 and 2 list the
and crack resistance with a compressive strength of 150 MPa or higher properties of the MWCNT and chemical composition of the cementi-
and can be used to construct such facilities with sufficient physical tious materials, respectively. The mineralogical compositions of the
protection. However, UHPC cannot be obviously applied alone for EMI used cement and other silica-based materials were presented in a pre-
shielding because it has very low electrical conductivity like most ce- vious paper by the authors on UHPC [31].
mentitious materials, which is in the range of 0.001–1 μS/m [26]. Table 3 presents the mix proportion. CNTs were incorporated into
In addition, most previously published dispersion methods are not the base UHPC mixture at six different dosages: (0, 0.2, 0.5, 0.8, 1.0 and
directly applicable to making UHPC more conductive because UHPC 2.0 wt% relative to the cement). The weight ratios of the water and
generally has a low w/c of as low as 0.2 to achieve a high strength of polycarboxylate-ether (PCE)-type superplasticizer (SPPL) were kept
over 150 MPa. Therefore, the engineering challenge of incorporating constant at 23 and 4 wt%, respectively, except for the specimen with
CNTs in UHPC is that the high specific surface area of CNTs leads to 2.0 wt% CNT. For specimen CNT2.0, w/c of 0.28 was applied so that
high water absorption, which decreases workability. Moreover, UHPC the flow value was equal to that of specimen CNT1.0. This was to
is composed of not only cement and fine aggregate but also various prevent a false setting from the rapid decrease of flow with the high
materials such as silica fume and silica powder. Therefore, even if there CNT dosage.
is a way to uniformly disperse CNTs in a UHPC binder, it is not clear
how the CNTs work within the complex matrix and what their impact is 2.2. Specimen preparation
on the resulting electrical and mechanical properties.
Finally, most previous research tested the SE based on coaxial Fig. 1 presents the specimen preparation process, which consisted of
transversal electromagnetic (TEM) cell method such as ASTM D4935-18 four steps: (a) fabrication of the dispersed CNT solution, (b) mixing of
[27]. However, this method is not appropriate for testing cement based the UHPC dry pre-mixture with the solution, (c) casting, and (d) curing.
composites [20] because fabricating a homogeneously small less than First, the dispersed CNT solution was prepared by sonication and
diameter of 13.3 cm and thin specimen with a thickness of 25 μm (i.e., shear mixing with the SPPL (Fig. 1a). The dispersion of CNTs in ce-
the required dimension for ASTM D4935-18) is impossible. In addition, mentitious material has been an obstacle towards taking advantage of
if SE testing is conducted on a specimen that is several millimeters their attractive physical properties because of their hydrophobic nature,
thick, the SE could be overestimated because the gap between the high aspect ratio and strong van der Waals forces [32]. In addition, it
coaxial TEM cells increases with the specimen thickness. Therefore, has been found that CNTs partially adsorb water molecules on its sur-
even if accurate calibration is confirmed, the total transmission signal faces [33,34]. Sonication with PCE-type SPPL in water is one of the
cannot be directly received through the specimen, and some signals most commonly used techniques for achieving a desirable level of dis-
may leak through the gap. Moreover, it is difficult to assume that the SE persion within both an aqueous solution and cementitious composite
obtained from small specimen can be representative of that in actual [35–37]. However, this method could be inapplicable to UHPC with a
structure level. Thus, coaxial TEM cell method should only be used to low w/c. This is because the specific surface area of CNTs is large; as the
estimate a rough trend of the SE. Large specimen testing such as IEEE- dispersion progresses, some of the water and SPPL adsorb on the CNTs.
STD-299 [28], ASTM E1851-15 [29], and MIL-STD-188-125-1 [30], This results in a rapid decrease in the flow [37], which leads to a false
which provide SE that is sufficiently representative for the actual ap- setting. Thus, in this study, CNTs were dispersed by sonication first then
plication of developed materials in structures, should be used. shear mixing with SPPL. Specifically, CNTs were sonicated in distilled
Hence, in this study, CNTs were incorporated in UHPC and in- water without incorporation of SPPL through the use of an ultrasonic
vestigated with regard to the dispersion method, mechanical properties, processor (VCX 750, Sonics Inc.) with a 25 mm diameter cylindrical
and EMI SE. In the beginning, the dispersion of CNTs and fundamental boost probe. The sonicator was operated at an output power of 600 W

2
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

Table 2
Chemical composition of cementitious materials.
Chemical composition SiO2 Al2O3 TiO2 Fe2O3 MgO CaO Na2O K2O MnO P2O5 LOI Total

Cement 21.18 5.88 0.23 2.46 3.89 53.07 0.15 0.90 0.06 0.40 1.28 89.51
Silica fume 96.90 0.29 0.01 0.15 0.18 1.54 0.16 0.64 0.03 0.05 0.02 99.97
Silica powder 97.70 0.49 0.08 0.05 0.21 1.37 0.02 0.02 0.01 0.004 0.02 99.99

(amplitude of 80%) to deliver energy at a rate of 35,000–36,000 J/min. was completely removed by oven drying at 60 °C for 2 days prior to
During sonication, the temperature of the vessel was maintained at 7 °C testing. The purpose of this procedure was to identify the sole effect of
to prevent potential overloading of the equipment and deterioration of electronic conduction by CNTs on the results.
dispersion efficiency due to heat generation. In addition, the solution
was circulated with a pump for uniform dispersion. After sonication
3. Dispersion of CNTs
was completed, SPPL was added into the solution and shear mixed for
5 min with L5M-A (Silverson Inc.). With this method, the flow was
3.1. Dispersion of CNTs in an aqueous solution
increased compared to previously reported methods [38]. Because the
surface of the CNTs had already adsorbed sufficient water, the amount
The CNT particle size in the aqueous solution was analyzed through
of later adsorption of the SPPL on the CNTs was negligible. The dis-
a laser diffraction equipment (Bluewave, Microtrac) to evaluate the
persion progressed until the average particle size of the CNTs reached
dispersion of different amounts of CNTs. Fig. 3 shows the particle size
about 10–30 μm in which particle size analysis performed immediately
distributions of CNTs in the aqueous solution before and after disper-
after finishing dispersion.
sion (i.e., sonication and shear mixing procedures). The results are
After the dispersed CNT solution was prepared, a UHPC dry pre-
averaged from three identical measurements for each specimen. The
mixture was fabricated through the homogeneously blending of
mix proportions containing CNTs after dispersion had similar particle
Portland cement, silica fume, silica powder, and silica sand. The pre-
size distributions, and their mean volume diameter (MVD) was in the
mixture was then added into the dispersed CNT solution and mixed for
range of 13.93–29.01 μm (Fig. 3a–b). The average MVD value after
5 min with a Hobart mixer (Fig. 1b). For UHPC (Ref.), water and SPPL
dispersion of 22.75 μm was 38 times smaller than that before dispersion
were used instead of the dispersed CNT solution. Then, fresh mixture
(i.e., MVD of 506.87 μm) (Fig. 3c–d). Considering the size properties
was poured into molds for casting samples (Fig. 1c). Fig. 2 present the
listed in Table 1, it can be safely assumed that CNT bundles in the
geometries of the prepared specimens.
aqueous solution were dispersed well by the proposed method. Thus, a
Cubic specimens with a volume of 50 × 50 × 50 mm3 were fab-
dispersed CNT solution with a similar dispersion degree (i.e., similar
ricated for the measurements of the electrical conductivity and com-
MVD value) was used to manufacture the listed specimens.
pressive strength. For the electrical conductivity measurement, two
copper plates with dimensions of 20 × 20 × 0.4 mm3 were embedded
into the specimens as electrodes. The contact area of each copper plate 3.2. Dispersion of CNTs in a UHPC matrix
to the composite was coated with silver paint to minimize the surface
resistivity and avoid electrode corrosion during the curing process Scanning electron microscope (SEM) analysis was carried out with a
[39–41]. The specimens for SE testing based on ASTM D4935-18 were field emission scanning electron microscopy (FE-SEM, JSM-6700, JEOL
manufactured according to the required dimensions except for the Ltd.) to analyze the morphology of the dispersed CNTs in the matrix.
thickness. These included a reference specimen and load specimen with During the SEM analysis, energy dispersive spectroscopy (EDS) spectra
a thickness of 3 mm. The reference specimen for calibration was disc- were examined to distinguish CNTs from needle-shaped cement hy-
shaped; it comprised a circle with a diameter of 33 mm inside a ring dration products such as ettringite based on the chemical composition.
with an outer diameter of 133.1 mm and inner diameter of 76.2 mm. Thin sections of specimens with hydration stopped at 28 days were
The load specimen for SE measurement was in the shape of a circle with fabricated for the analysis.
a diameter of 133.1 mm. The specimens for EMI SE testing according to Fig. 4 shows the morphology of dispersed CNTs in the UHPC matrix.
IEEE-STD-299 were fabricated as square panels with two sizes: CNTs distributed as individual fibers with a length of < 1 μm were
350 × 350 mm2 with a thickness of 50 mm and 1250 × 1250 mm2 observed to be abundant in the matrix. This analysis qualitatively de-
with a thickness of 200 mm. monstrates that the proposed dispersion method used in this study ef-
Once a fresh mixture casted, the surface was immediately covered fectively disperses CNTs in both the aqueous solution and cementitious
with a plastic sheet to prevent sudden moisture evaporation, and the composite while a low w/c ratio was maintained for UHPC production.
specimens were cured for 24 h at 20 °C. Then, the molds were removed, The average CNT length was shorter than that based on particle size
and the specimens were steam-cured at 90 °C for 48 h. Subsequently, analysis because the CNTs were additionally dispersed by the silica
the specimens were kept at 20 °C for 25 days, for a total curing period of fume during mixing. Incorporating an appropriate amount of silica
28 days (Fig. 1d). For the electrical conductivity and SE testing speci- fume has been shown to effectively disperse CNTs in cementitious
mens, the moisture in the matrix, which can provide ionic conduction, composites [42]. In particular, in a SEM image taken of a fragment after
the compressive strength test, a large amount of CNT fibers was

Table 3
Mix proportions.
Sample Cement (C) CNTs Silica fume (SF) Silica powder (SP) Silica sand (SS) Superplasticizer (SPPL) Water/cement (w/c)

UHPC (Ref.) 1 0 0.25 0.35 1.1 0.04 0.23

CNT0.2 0.002
CNT0.5 0.005
CNT0.8 0.008
CNT1.0 0.01
CNT2.0 0.02 0.28

3
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

%06 522.
&KURGTUGF
5QNWVKQP
9CVGT
5QPKECVKQP 5JGCT OKZKPI

C (CDTKECVKPIVJGFKURGTUGF%06UQNWVKQP
7*2%
2TG/KZ

D /KZKPI E %CUVKPI F &GOQNFKPIEWTKPI

Fig. 1. Specimen preparation process.

observed between the crack surfaces (Fig. 4c). However, as the CNT between the CNT content of 0.8 and 1.0 wt% (i.e., mean value of 0.9 wt
content was increased to 0.8 wt% or more, agglomerated CNT bundles %) in this study.
were often observed on the surface of unhydrated cement or in the In summary, it was verified that shear mixing with SPPL after so-
hydration product (Fig. 4d–e). This may be because the additional nication can be an effective way to disperse CNTs within both an
amount of CNT exceeded the critical incorporation concentration (CIC). aqueous solution and UHPC matrix. However, if the incorporated CNT
Excessive addition of CNTs in a given volume of UHPC shortens the content in a given volume of the UHPC matrix is close to the CIC, CNTs
relative distance between fibers; this strengthens the van der Waals could be partly agglomerated.
forces, which leads to agglomeration of CNTs in the matrix. In addition,
the rapid decrease in the flow with an increase in the CNT content can
disturb the uniform dispersion in the matrix. Based on the morpholo-
gical study of the UHPC–CNT composites, the CIC could be proposed

C 5RGEKOGPHQTOGCUWTGOGPVQHGNGEVTKECN D 5RGEKOGPHQT'/+5'VGUVKPIDCUGFQP#56/&
EQPFWEVKXKV[CPFEQORTGUUKXGUVTGPIVJVGUV

E 5RGEKOGPHQT'/+5'VGUVKPIDCUGFQP+'''56&  F 5RGEKOGPHQT'/+5'VGUVKPIDCUGFQP+'''56& 
Fig. 2. Geometry of the specimens.

4
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

 
%06  C %06 D

 %06   %06

%06  %06
%JCPPGN   %06   %06

2CUUKPI 
%06  %06
 #XG  
#XG

 

 

 
          

2CTVKENGUK\GFKUVTKDWVKQP ƏO 2CTVKENGUK\GFKUVTKDWVKQP ƏO

 
$GHQTGFKURGTUKQP  E $GHQTGFKURGTUKQP F
 #HVGTFKURGTUKQP #XG   #HVGTFKURGTUKQP #XG



%JCPPGN 

2CUUKPI 






 

 
          

#XGTCIGRCTVKENGUK\GFKUVTKDWVKQP ƏO #XGTCIGRCTVKENGUK\GFKUVTKDWVKQP ƏO
Fig. 3. Particle size distribution of CNTs in aqueous solution (red number next to legend indicates the measured mean volume diameter of CNTs): (a, b) channel
(distribution ratio) and passing (cumulative ratio) data after dispersion; (c, d) comparison before and after dispersion. (For interpretation of the references to colour
in this figure legend, the reader is referred to the web version of this article.)

4. Mechanical properties was 177.1 MPa, which is similar to that of UHPC (Ref.), but the elastic
modulus was 6.4%–7% higher.
4.1. Compressive strength and elastic modulus The effect of dispersed CNTs on the mechanical properties of UHPC
can be explained based on the morphology of the composites (Section
The compressive strength of the CNT-UHPC composites was mea- 4.1). First, well-dispersed CNTs provide a massive surface area for the
sured in accordance with ASTM C109. The mechanical test was not precipitation of cement hydrates. This could conceivably occur in a
conducted for specimen CNT2.0 because of the different w/c, which did well-packed format that would contribute to the formation of denser
not allow a direct comparison of the mechanical performance. A dis- C–S–H agglomerates with high stiffness [4], as shown in Fig. 4. In ad-
placement-controlled universal testing machine (UTM) was used with a dition, embedded CNTs in the hydration product phase bridge the ad-
load cell capacity of 2000 tons. A loading rate of 0.02 mm/s was ap- jacent hydrated particles, which is called the bridging effect (Fig. 4c)
plied. Six specimens for each mix proportion were tested and the [32]. Therefore, the compressive strength and elastic modulus improve.
average value of four specimens excluding the largest and smallest However, when the CNT content close to the CIC (i.e., 0.8–1.0 wt%),
values was taken as the representative compressive strength. The elastic the CNTs become clumped (Fig. 4d–e). They, act as pores or cracks and
modulus was calculated according to ASTM C 469 and BS-1881 part covers the surface of the component, which hinders the development of
121. Fig. 5 summarizes the compressive stress–average strain curves of uniform hydration [42]. This decreases the mechanical properties.
the UHPC-CNT composites and their compressive strength and elastic
modulus. 4.2. Poromechanical properties
All specimens satisfied the strength requirement for UHPC (i.e.,
compressive strength > 150 MPa). The UHPC (Ref.) had a compressive Mercury intrusion porosimetry (MIP) testing was performed using
strength of 178.6 MPa, and its elastic modulus was 49.1 GPa by ASTM AutoPore IV 9500 (Micromeritics) to investigate the porosity and pore
and 52.1 GPa by BS code, respectively. The mechanical properties of size distribution of the composites. The composites cured for 28 days
specimen CNT0.2 did not significantly differ from those of the reference were cut into small pieces (approximately 5 × 5 × 5 mm3 cubes), and
specimen, whereas specimen CNT0.5 showed a 5.5% improvement in the hydration reaction was stopped. The MIP testing parameters were
the compressive strength and approximately 12.0% improvement in the set to a surface tension of 485 dynes/cm and mercury contact angle of
elastic modulus. However, as the CNT content approached or went 130°.
above the CIC, both the compressive strength and elastic modulus de- Figs. 6 and 7 show the pore size distribution and porosity, respec-
creased, although, the elastic modulus remained higher than that of the tively, divided into four ranges based on the pore diameter of the
reference specimen. For specimen of CNT0.8, the compressive strength specimens. The pore structure of UHPC changed completely as the CNT

5
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

C D %06U
GODGFFGFKPVJG
J[FTCVKQPRTQFWEVU

%06U
GODGFFGFKPVJG
J[FTCVKQPRTQFWEVU

E %06U
DTKFIGFVJGETCEMKPI
UWTHCEGU


F G
%06U
GODGFFGFKPVJG
J[FTCVKQPRTQFWEVU

%06U
CIINQOGTCVGFQPVJG %06U
5WTHCEGQHEGOGPV CIINQOGTCVGFKPVJGJ[FTCVKQPRTQFWEVU

Fig. 4. SEM images of the UHPC-CNT composites: (a) CNT0.2, (b, c) CNT0.5, (d) CNT0.8, and (e) CNT1.0.

Fig. 5. Compressive strength and elastic modulus of the UHPC-CNT composites: (a) compressive stress– average strain curves and (b) summary of the compressive
strength and elastic modulus.

6
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

 
C D
7*2% 4GH

.QI FKHHGTGPVKCN KPVTWUKQP O.I

7*2% 4GH

%WOWNCVKXGRQTG XQN O.I

 %06  %06
%06 %06
 %06  %06
%06 %06
 

 

 

 

 
       

2QTGUK\GFKCOGVGT ƏO 2QTGUK\GFKCOGVGT ƏO
Fig. 6. Pore size distribution of the UHPC-CNT composites at 28 days: (a) log differential intrusion and (b) cumulative pore volume.

/GEJCPKECN RTQRGTV[ QH 7* 2%

GPVTCKPGFGPVTCRRGFCKTXQKFU  ƏO +ORTQXG 9GCMGP
OCETQRQTG ƏO
 OGUQRQTG ƏO
OKETQRQTG  ƏO
 5VKHHGT%5* 5WRRTGUUKPIJ[FTCVKQP
2QTQUKV[ 

  

 
  $TKFIKPI #IINQOGTCVGF%06U
 

 2QTGHKNNKPI %TKVKECN #KTXQKFUKPETGCUKPI
  
+PEQTRQTCVKQP
   %QPEGPVTCVKQP
 %+%
  


     %06EQPVGPV


Fig. 8. Effect of CNTs on the mechanical properties of UHPC.
7*2% 4GH %06 %06 %06 %06
Fig. 7. Pore volume distribution of the UHPC-CNT composite at 28 days (red and poromechanical properties.
number on the bars indicate the total porosity). (For interpretation of the re- In summary (Fig. 8), dispersing CNTs in UHPC can improve the
ferences to colour in this figure legend, the reader is referred to the web version
mechanical properties up to the CIC because of the pore filling effect,
of this article.)
bridging effect, and formation of a hydration product with higher
stiffness. However, excessive incorporation beyond the CIC is rather
content increased. The total porosity of specimens containing CNTs ineffective for improving the mechanical properties because of the ag-
decreased regardless of the CNT content compared to the reference glomeration of CNTs that act as crack and pores, suppression of hy-
specimen. In particular, the mesopores (4.5–25 nm) were remarkably dration, and increase of air voids in the matrix.
reduced by the pore filling effect. However, entrained or entrapped air
bubbles formed at 100–1000 μm increased as the CNT content in-
5. EMI shielding effectiveness (SE)
creased because of the lower flowability. A low fluidity for a fresh
concrete mixture leads to more air voids being included in the hardened
5.1. Electrical conductivity
matrix [43].
The poromechanical properties can be correlated with the com-
The electrical resistance of the specimens was measured through the
pressive strength. As shown in Figs. 6a and 7, the total porosity and
four-wire method with an LCR meter (LCR-6100, GW INSTEK). For each
proportion of mesopores for specimen CNT0.5 decreased remarkably
mix proportion, six specimens were tested at a frequency of 10 kHz, and
because of the pore filling effect, which improved the compressive
the average value of four specimens excluding the largest and smallest
strength. On the other hand, the pore volume distribution of specimen
values was taken as the representative electrical resistance. The elec-
CNT0.2 was similar to that of the reference specimen, so the strength
trical conductivity, which is the reciprocal of the electrical resistivity,
was not improved significantly. As the CNT content approached the
was then calculated as follows:
CIC, although the total porosity and proportion of mesopores decreased,
the proportion of air voids became much larger than that of the re- 1 L
σ (S/m) = = ,
ference specimen, which decreased the compressive strength. In the ρ RA (1)
case of specimens CNT0.8 and CNT1.0, the relative pore volume in the
matrix increased because the hydration reaction was suppressed by an where σ is the electrical conductivity (S/m), ρ is the electrical resistivity
excessive amount of CNTs, which increased the cumulative mercury (Ω·m), R is the electrical resistance (Ω), A is the cross-sectional area of
intrusion (Fig. 6b). This demonstrates the correlation between the CIC the composite between the electrodes (m2), and L is the distance be-
tween the electrodes (m).

7
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

 respectively.
'NGEVTKECN EQPFWEVKXKV[ O5O

According to the law of conservation of energy, when incident en-

ergy is passing through a material, some of the energy is reflected off
 the surface of the material and is absorbed into the interior; this is
2GTEQNCVKQPVJTGUJQNF converted to heat energy. The remaining energy is transmitted through
(QTOCVKQP the material, some of which is reflected again on the opposite surface of
QH the material. Therefore, the SE that generated when EM waves pass

EQPFWEVKXG through a material is theoretically the sum of the absorption loss (SEA),
RCVJGPF reflection loss (SER), and multiple refection loss (SEM). Thus, Eq. (2) can
 be also expressed as follows:
(QTOCVKQPQH SE (dB) = SEA + SER + SEM. (3)
EQPFWEVKXGRCVJ
According to EMC theory, each SE component for a material without
 UVCTV
an opening can be calculated as follows depending on the frequency
region [48]:

 SEA (dB) = 131.4 t fμr σr , (4)

     
%06EQPVGPV ⎧ ⎛ μr ⎞ H⎛ λ
⎪ 14.6 − 10 log ⎜ fr 2σ ⎟ for magnetic (H) field, SER ⎜r < 2π
⎪ ⎝ r⎠ ⎝
Fig. 9. Electrical conductivity of the UHPC-CNT composites. ⎪ c ⎞
⎪ = ⎟, (a)
⎪ 2πf ⎠
Fig. 9 shows the electrical conductivity of the UHPC-CNT compo- SER (dB) =
⎨ fμr
sites with respect to the CNT content. The electrical conductivity im- λ
⎪ 168 − 20 log for plane wave, SEPR ⎛⎜r >
proved with the CNT content. When the CNT content was < 0.8 wt%, ⎪ σr ⎝ 2π
the conductivity improved slightly, whereas a dramatic improvement in ⎪
⎪ c ⎞
the conductivity was observed between 0.8 wt% and 1.0 wt%. The ⎪ = ⎟, (b)
2πf ⎠
percolation threshold is the ratio at which a conductive path is obtained ⎩
through the connection of adjacent conductive fibers and strongly de- (5)
pends on the degree of CNT dispersion [44]. The obtained electrical
K − 1 ⎞2 −2rt ⎤
conductivity at a CNT content of 0 wt% (i.e., reference specimen) was SEM (dB) = 20 log ⎡1 − ⎛ e ,
⎢
⎣ ⎝ K + 1⎠ ⎥
⎦ (6)
0.008 mS/m. However, it sharply improved to 209.8 mS/m with 1.0 wt
% of CNT. This also demonstrates that most CNTs were uniformly dis- where K is the impedance ratio, c is the velocity of light (3.0 × 108 m/
tributed throughout the composites [45]. If the CNT dispersion is poor, s), r is the distance from the antenna to the specimen (m), f is the fre-
the conductivity would gradually improve with increasing CNT content, quency (Hz), t is the thickness of the specimen (m), λ is the wavelength
and exhibit a modest increment. In this study, the mean CNT content (m), μr is the relative permeability (ref.: free space = 4π × 10−7H/m),
when the conductivity sharply improved was 0.9 wt%. Therefore, the and σr is the relative conductivity (ref.: copper = 5.82 × 107 S/m).
CNT content of 0.9 wt% can be regarded as the percolation threshold; SEM can be neglected when SEA is > 10 dB or in the case of a thick
this is the same as the previously obtained CIC value. At a CNT content material [35]. In opposite case, the SE of a material with an opening
of 2.0 wt%, despite being double the content of 1.0 wt%, the con- (SEo) (i.e., the opening effect) can be calculated as follows [49]:
ductivity did not improve significantly because the conductive path was
λ
already constructed at the percolation threshold [46,47]. Namely, the SEo (dB) = 20 log ⎛ ⎞,
formation of the conductive path started at a CNT content of 0.5 wt% ⎝ 2l ⎠ (7)
and ended at 1.0 wt% (i.e., past the percolation threshold). This de- where l designates the maximum linear dimension of the opening (m).
monstrates the correlation between the CIC and percolation threshold. Eq. (7) can only be used when l is less than half of the wavelength (λ) or
in the magnetic field (H-field).
5.2. EMI SE based on ASTM D4935-18 (coaxial TEM cell method) Fig. 11 shows a general trend of the SE for materials without and
with an opening. For materials without an opening, SEA increases with
EMI SE testing based on ASTM D4935-18 was performed in the the frequency regardless of the frequency region. SER increases with the
frequency range of 30 MHz–1.5 GHz [27]. The detailed procedure is frequency in the H-field but decreases with increasing frequency in the
presented in Fig. 10. A coaxial TEM cell with a discontinuous inner plane wave. For materials with an opening, SEo decreases with in-
conductor and flanged outer conductor, and a network analyzer creasing frequency.
(KEYSIGHT PNA series, E8364B) were prepared (Fig. 10a). Before the Fig. 12 presents the EMI SE testing results of the UHPC-CNT com-
testing, silver paint was applied to both the inner and outer edges of the posites based on ASTM-D4935-18. The SE of UHPC (Ref.) was close to
specimens to ensure electrical contact and prevent signal leak. Then, 0 dB in all frequency bands. However, as the CNT content increased, the
the specimens were placed on the TEM cell (Fig. 10b and d). EM waves SE not only improved at the same frequency but also increased with
were transmitted through the specimens placed between the flanges of increasing frequency. In particular, the SE noticeably improved from
the coaxial TEM cell, and the power levels of the signals for the re- the CNT content of 0.5 wt%, at which the conductivity developed and
ference and load specimen were measured with a spectrum analyzer indicated the beginning of the formation of a conductive path. How-
(Fig. 10c and e). The SE (in decibels, dB) was calculated from the fol- ever, the SE of specimens CNT1.0 and CNT2.0 was similar for the entire
lowing ratio: measured frequency range. This demonstrates that electrical con-
P ductivity is the key factor for the SE and that there is the direct cor-
SE (dB) = 10 log ⎛ i ⎞, ⎜ ⎟
relation between them.
⎝ Pt ⎠ (2)
The results can be validated through the EMC theory. First, the
where Pi and Pt are the amounts of electromagnetic power measured specimens used in this experiment were relatively thick (3 mm) com-
from the reference (free space in IEEE-STD-299) and load specimen, pared to the required dimension (25 μm). Like most cementitious

8
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

0GVYQTM %QCZKCN %QCZKCN 6'/EGNN

CPCN[\GT 6'/EGNN &KUEQPVKPWGFEQPFWEVQTU

(a) Experimental preparaon

4GHGTGPEG
URGEKOGP
UGVWR

(b) Reference specimen setup (c) Calibraon

.QCFURGEKOGP
UGVWR

(d) Load specimen setup (e) SE test

Fig. 10. EMI SE testing procedure based on ASTM D-4935-18.


*HKGNF 2NCPGYCXG

YQQRGPKPI 
YQRGPKPI

SE (dB)
5' F$


5'4

5'#



5'Q
     
(TGSWGPE[ *\ Frequency (MHz)
Fig. 11. General trend of SE for a material without or with an opening.
7*2% 4GH %06 %06
%06 %06 %06
material, UHPC is a dielectric that exhibits very little loss compared to Fig. 12. EMI SE testing results for UHPC-CNT composites based on ASTM-
metals. Thus, the SEM term can be ignored. In addition, the experi- D4935-18
mental frequency bands were distinguished in terms of the H-field be-
cause the distance from the signal to the specimen (r; assumed to be Eqs. (4) and (5a) and Fig. 11. Moreover, the relative permeability (μr) of
0.003 m, i.e., the thickness of the specimen) was much less than λ/2π. the composites can be assumed to be constant (i.e., close to 1) because
In the H-field, both SEA and SER increase as the electrical conductivity both the cementitious material and CNTs are not magnetic [23]. Hence,
and frequency increase, but the SER term is dominant as indicated in the SE of the reference specimen was close to zero because it had poor

9
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

%JCODGT 5KIPCN
5RGEVTWO IGPGTCVQT
O
CPCN[\GT

O O
.QQR $KEQPKECN .2
#PV #PV #PV
C HTGGURCEGECNKDTCVKQP  .QQR#PV  $KEQPKECN #PV  .2#PV

5RGEKOGP
5RGEKOGP 5GVWR
1RGPKPI 5GVWR

5OCNNKPEKFGPV CTGC
 OO .CTIG KPEKFGPV CTGC
 OO
D 5RGEKOGPUGVWR QRGPKPI CHVGTUGVURGEKOGPWR

6TCPUOKVVKPI
#PV

E #PVGPPCUGVWR F 5'VGUV
Fig. 13. EMI SE testing procedure based on IEEE-STD-299.

electrical conductivity. In other word, the SE of the specimens con- Table 4

taining CNTs at the same frequency improved because the electrical Distances between antennas and the specimen.
conductivity developed with increasing CNT content. In addition, the Ant. type Loop LP/bi-conical
SE increased with the frequency because the measured frequency bands
were in the H-field. However, the magnitude of the SE for specimen Distance (m) Calibration (between Antennas) 0.6 2
CNT2.0 was not significantly improved compared to that of CNT1.0 Testing (from specimen) Transmitting 0.3 1.7
Receiving 0.3 0.3
because they had similar conductivities.

5.3. EMI SE based on IEEE-STD-299 (large specimen testing method) the chamber, and the receiving antenna was placed inside the chamber
(Fig. 13c). Both antennas were installed perpendicular to the geometric
The EMI SE testing based on IEEE-STD-299 was performed in the center of the specimen. The distances between the antennas and spe-
frequency range 10 kHz–1.5 GHz [28]; the detailed procedure is pre- cimen are listed in Table 4.
sented in Fig. 13 [50]. A shielded chamber made of steel and zinc with a After the experimental setup was completed, a signal was generated
volume of 2400 × 2400 × 2400 mm3 with an opening, a signal gen- at the frequency of interest and transmitted with the transmitting an-
erator (SMB100A, ROHDE&SCHWARZ), a spectrum analyzer (8563E, tenna. Then, the power level of the signal detected by the receiving
HP), and antenna kits were used in the testing. The procedure began antenna (Pt) that passed through the specimen was measured with a
with a free-space calibration (Fig. 13a). The calibration values at the spectrum analyzer (Fig. 13d). Thereafter, the SE was determined ac-
frequency of interest (Pi) were taken as the zero point to set for the cording to Eq. (3). The SE was measured as the frequency was increased
receiver, which showed the difference from the nominal value in the 10 times within the frequency range of 10 kHz–1 GHz. The SE was
readout. Once the calibration was completed, a test specimen was additionally measured at 30 MHz and 1.5 GHz to compare the results
mounted on the opening (Fig. 13b). Conductive gaskets were fixed to with those based on ASTM D4935–18. To measure the SE, loop an-
the gap between the chamber and an edge of the specimen to prevent tennas were used in the low-frequency range (10 kHz–10 MHz), bi-
the leaking of the signal. The incident area of the specimens was smaller conical antennas were used in the transition and resonant ranges
than the size of the samples owing to the experimental setting: (10 MHz–100 MHz), and log periodic (LP) antennas were used in the
300 × 300 mm2 (small incident area) for the 3502 mm2 specimen and high-frequency range (100 MHz–1.5 GHz).
1200 × 1200 mm2 (large incident area) for the 12502 mm2 specimen. Fig. 14 shows the EMI SE testing results for UHPC-CNT composites
After mount specimen on, the transmitting antenna was set up outside

10
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

 SEeff (dB) = SE − SEopening = SEA + SER + SEM (8)

7*2% 4GH
%06 2NCPG Alternatively, a better solution would be to fabricate the specimen
 %06 YCXG to have a sufficiently large incident area. Otherwise, the SE could be
%06
 overestimated because of the small size of incident area.
This approach was applied to the SE test results (given in Fig. 14),
and the results are presented in Fig. 16. UHPC (Ref.) had a low SE value
5' F$


below 3 dB in all frequency bands, and the SE of specimens containing
CNTs slightly increased with increasing frequency at 10 MHz–1.5GHz.

These tendencies are similar to those of the results based on coaxial
TEM cell testing. The temporary decrease in the SE at frequencies of
 6TCPUKVKQP 30–100 MHz was because the frequencies were in the transition and
*HKGNF  resonant range (10 MHz–100 MHz). Specifically, the rate of increase in
 4GUQPCPV SEA was less than the rate of decrease in SER as the frequency region
converted from the H-field to a plane wave, so SEeff partially decreased.
 In the plane wave, SER decreases with increasing frequency, whereas
M M / / / / ) ) SEA increases.
Another observed difference was that SEeff in the H-field decreased
(TGSWGPE[ *\ with increasing frequency. This phenomenon is contrary to not only the
Fig. 14. EMI SE testing results for UHPC-CNT composites with a small incident results based on coaxial TEM cell testing but also the trend according to
area based on IEEE-STD-299 (symbols and lines represent the experimental EMC theory. This was also presumably due to the dimensions of the
values and their mean values, respectively). opening. Fig. 17 shows the effect of a shield discontinuity by the type of
material. The incident EM field induces a current into the material, and
this current then generates an additional field. This new field cancels or
with a small incident area (i.e., 350 mm2 specimen) based on IEEE-STD-
reflects the original field in some regions of space, specifically the re-
299. According to EMC theory and the experimental conditions, the
gion on the opposite side of the material from the incident field. For this
experimental frequencies were divided into three frequency regions: the
cancellation to occur, the induced shield current must be allowed to
H-field (below 10 MHz), transition and resonant region (from 10 MHz
flow undisturbed in the manner in which it is induced. If an opening
to 100 MHz), and plane wave (above 100 MHz). The SE of the com-
forces the induced current to flow in a different path, then the gener-
posites improved at the same frequency with increasing CNT content
ated field will not completely cancel the original field, and the SE will
compared to that of the reference specimen regardless of the frequency
be reduced. Forcing the current to detour more cause the SE to further
region. In addition, the SE of specimens CNT1.0 and CNT2.0 was almost
decrease. Like most cementitious materials, UHPC has pores of various
identical for the entire measured frequency range. These were the same
sizes acting as micro-openings. Therefore, it is more appropriate to
trends observed for the coaxial TEM-cell testing results. However, two
consider UHPC as a porous material with micro-openings rather than a
interesting differences were observed. Firstly, the magnitude of the SE
uniform material. In addition, it has very low electrical conductivity,
was significantly different. In particular, the SE of UHPC (Ref.), which
which results in a low SE close to 0 dB. Thus, the UHPC matrix can be
had no SE (close to 0 dB) in the coaxial TEM-cell testing results,
regarded as a single opening that can cause a shield discontinuity from
was > 25 dB at frequencies below 1 GHz. In addition, the SE of the
the viewpoint of electromagnetics. In the case of the UHPC-CNT com-
specimens decreased as the frequency increased above 30 MHz. Hence,
posites, only the UHPC matrix reinforced with CNTs acts as a shield,
the SE of the cementitious material seems to vary depending on the
and the other parts (i.e., the UHPC matrix without CNTs and pores) acts
testing method. The result is discussed in detail in the following section.
as an opening. Hence, Eq. (7), which allows the SE for a material with
an opening to be determined, should be considered with Eqs. (4)–(6) for
the SE of a material without an opening.
5.4. Discussion on EMI SE based on IEEE-STD-299
Based on this assumption, the new SEeff in an H-field (SEeffH∗)
considering the opening effect can be expressed as follows:
Considering that the electrical conductivity is the dominant factor
for the SE and was very low for UHPC (Ref.), it would be reasonable for
SEH ∗ H H
e ff (dB) = SEe ff + SEo = SEA + SER + S E M + SEo , (9)
the SE to be around 0 dB as observed in the coaxial TEM-cell testing
results. Thus, the reason why UHPC (Ref.) had a SE of > 25 dB was
SEM can be ignored as discussed previously in Section 5.2.
assumed to be that the electrical dimension of the opening, which had
Let SEeffH∗ at a given frequency (f1) be SEeff_1H∗. Then, that at a
the same size as the small incident area, was smaller than the trans-
certain frequency (f2), SEeff_2H∗, can be written as
mitted λ. Specifically, the transmitted EM waves were not completely
incident to the specimen, but some was shielded by a region of the SEH ∗ H∗ H∗
eff _ 2 = SEeff _ 1 + d SEeff , (10)
chamber in the vicinity of the opening. Therefore, the receiving antenna
received only a small part of the signal that should have been received. where dSEeffH∗ is the differential increment and can be calculated as
Thus, the SE was measured to be high as if UHPC (Ref.) had a SE. This is
described in Fig. 15. d
dSEH ∗
eff = (SEA + SERH + SEo ) df ,
To validate this assumption, the SE of the opening (i.e., without a df (11)
specimen being mounted) was tested. The resulting SE was nearly the
same as that of UHPC (Ref.) for the entire measured frequency range By substituting Eqs. (4), (5a), and (7) into Eq. (11) and differ-
(black dotted line in Fig. 14). This means that UHPC (Ref.) had no entiating with respect to the frequency (f), dSEeffH∗ can be derived as
shielding capability, and most SE was caused by a region of the follows:
chamber. Thus, to measure the SE of the specimen itself (SEeffective,
SEeff), a more accurate approach would be to subtract the SE of the
opening (SEopening) from the SE measured by the conventional method
(i.e., IEEE-STD-299) despite the calibration as follows:

11
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

5JKGNFGFEJCODGT
%QNNCVGTCNUJKGNFKPIQEEWT /CKPUJKGNFKPI
D[CRCTVQH 2TQEGUUQH
5JKGNFGFEJCODGT URGEKOGP

5RGEKOGP
%QPFWEVKXG
ICUMGV
(KZVWTG
6TCPUOKVVKPI#PV 4GEGKXKPI#PV

Fig. 15. Collateral shielding by a region of the shielded chamber because of the insufficient size of the incident area.

 d ⎡131.4t fμ σ + 14.6 − 10 log ⎛⎜ μr ⎞⎟ + 20 log ⎛⎜ c ⎞⎟ ⎤ df

dSE∗eff = r r
6TCPUKVKQP df ⎢ 2
⎝ fr σr ⎠
⎥
⎝ 2fl ⎠ ⎦
2NCPG ⎣
 *HKGNF 
YCXG
4GUQPCPV ⎛ 65.7t μr σr 1 fr 2σr μr 20 2fl c ⎞
=⎜ + 10 − df
 7*2% 4GH %06 f ln 10 μr f 2 r 2σr ln 10 c 2f 2 l ⎟
⎝ ⎠
%06 %06
5'GHH F$

%06 %06
 ⎛ 65.7t μr σr 10 1 20 1 ⎞
=⎜ + − df
f f ln 10 f ln 10 ⎟
⎝ ⎠

10 1 ⎞ 65.7t μr σr
≈ ⎛⎜− ⎟ df , where ≈0
 ⎝ f ln 10 ⎠ f
(12)

Given that the contribution of SEA in the H-field is small and the
 calculated dSEA is close to 0 dB based on the experimental conditions, it
can be considered as negligible.
M M / / / / ) )
Again, as the frequency increases in the H-field, both SEA and SER
(TGSWGPE[ *\ increase whereas SEo decreases. However, a negative increment has
been derived; this is noteworthy because it means that the rate of in-
Fig. 16. EMI SE results with collateral shielding by a region of the shielded crease in SER is smaller than the rate of decrease in SEo. Thus, SEeffH∗
chamber removed (SEeff) for UHPC-CNT composites with a small incident area decreases gradually instead of increasing. dSEeffH∗ can be calculated to
based on IEEE-STD-299.
be about −3.9 dB regardless of frequency based on Eq. (12). Because
the SE was measured at 10 frequency intervals from 10 kHz to 1 GHz, df
corresponded to 9f1 (f2 – f1 = 10 f1 – f1 = 9 f1). Thus, the SE in the H-
field decreased with the corresponding increment as the frequency by a

C 5GEVKQPQH5JKGNF D
+PFWEGF5JKGNF
%WTTGPVU
2CTVQH5JKGNF
%06TGKPHQTEGF
/KETQQRGPKPI
2QTGU7*2% OCVTKZ
Fig. 17. Effect of shield discontinuity on a magnetically induced shield current: (a) uniform material and (b) porous material.

12
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

 incident on the specimen with no collateral shielding by a region of the

6GUV 0WOGTKECN chamber. The opening effect disappeared when the dimension of the
TGUWNVU ECNEWNCVKQP
incident area was increased because the effect decreases as the max-
%06
%06 imum dimension of the opening (l) increases as indicated by Eq. (7).

%06 Therefore, the SE obtained with the large incident area can be regarded
as SEeff of UHPC-CNT composite with a thickness of 200 mm. The SE in
5' F$

the plane wave was higher than that with the small incident area be-
 cause the contribution of SEA increased; the specimen was four times
thicker. The SE of the specimens decreased temporarily in the frequency
range of 30–100 MHz. This is because the frequency was in the tran-
 sition and resonant region, as discussed in the results for the small in-
cident area. In addition, resonance that occurred in this range reduced
the SE. Resonance occurs when l is half of λ. Given that l of the large
incident area was 1700 mm ( 2 × 1200 mm), the resonance frequency
 corresponding to λ/2 was calculated to be 88.2 MHz, which is in the
M M / / transition and resonant region. For specimen CNT1.0, the SE was
17.16 dB at 10 kHz (H-field) and 47.14 dB at 1 GHz (plane wave),
(TGSWGPE[ *\ which are increase of 8.58 times and 23.57 times, respectively, com-
Fig. 18. Comparison between the SE reduction rates for the experimental re-
pared to UHPC (Ref.). This result is completely different from that of
sults and numerical calculation. the coaxial TEM cell testing. Thus, large specimen testing should be
required to evaluate the SE of cementitious materials for actual appli-
cations.
factor of 10. Good agreement between the SE reduction rates of the
In summary (Fig. 20), the SE of the UHPC-CNT composites improves
experimental results and derived result of −3.9 dB was observed, as
with the CNT content up to the percolation threshold, regardless of the
shown in Fig. 18 (the results for specimens CNT0–CNT0.5 are not
testing method because of the improved electrical conductivity. In
presented because they had low SE of < 2 dB within the error range).
particular, it significantly improved when the CNT content was be-
To validate the assumption, additional SE testing was performed
tween the beginning and end of the conductive path formation in the
with the 1250 × 1250 × 200 mm3 specimens, which had an incident
matrix that developed the conductivity (Fig. 20a). However, large
area of 1200 × 1200 mm2. The results are presented in Fig. 19.
specimen testing should be required to evaluate the SE for actual ap-
Similar to the SE results obtained with the small incident area, the
plications. For a reasonable measurement of the SE according to IEEE-
SE of the UHPC-CNT composites improved with the CNT content in all
STD-299, the specimen should be fabricated to have an incident area as
frequency bands. In particular, the SE noticeably improved above a
large as at least 1200 × 1200 mm2 to eliminate the opening effect,
CNT content of 0.5 wt%, at which the conductivity developed and was
which can cause a shield discontinuity. Otherwise, there is the possi-
regarded as the beginning of the formation of the conductive path. In
bility that the SE could be overestimated and distorted (Fig. 20b).
addition, the SE of the specimens CNT1.0 and CNT2.0 was similar for
the entire measured frequency range. However, unlike the results with
6. Conclusion
the small incident area, the SE of UHPC (Ref.) was below 3 dB in all
frequency bands, and the SE of the composites increased with the fre-
In this study, CNT-UHPC composites were considered from various
quency in the H-field. This is possibly because the effect related with
perspectives with regard to the dispersion method, mechanical prop-
the dimension of the opening was removed by the increased dimension
erties, and EMI SE. UHPC composites with six different CNT amounts
of the incident area. Specifically, given that the transmitted λ was
(0, 0.2, 0.5, 0.8, 1.0, and 2.0 wt%) were designed. First, the chosen
smaller than the electrical dimension of the opening, which was equal
method to disperse the CNTs in both an aqueous solution and UHPC
in size to the large incident area, the transmitted EM waves were almost
matrix were evaluated through particle size analysis and SEM analysis.
In addition, the fundamental mechanical properties were investigated
 through a compressive strength test, measurement of the elastic mod-
7*2% 4GH 6TCPUKVKQP ulus, and MIP test to determine the effect of CNTs on UHPC. Then, the
%06 *HKGNF  EMI SE was examined through measurement of the electrical con-
 %06 4GUQPCPV ductivity and SE. Two types of SE testing methods were considered, and
%06 the results were analyzed according to EMC theory. The following
 %06 conclusions were drawn:
%06
5' F$

2NCPGYCXG
 (i) Sonication and subsequent shear mixing with SPPL can be an ef-
fective way to disperse CNTs both in an aqueous solution and to
produce CNT-UHPC composites. However, if the incorporation of
 dispersed CNTs is increased to the CIC, CNTs could be partially
agglomerated.
 (ii) Dispersed CNTs can improve the mechanical properties of UHPC
below the CIC by the pore filling effect, bridging effect, and for-
 mation of a denser C–S–H structure with higher stiffness. However,
when an excessive amount of CNTs close to the CIC is in-
M M / / / / ) )
corporated, agglomerated CNTs act as pores or cracks and suppress
(TGSWGPE[ *\ the hydration, which weakens the properties. There is the direct
correlation among the CNT dispersion, CIC, and mechanical
Fig. 19. EMI SE testing results for UHPC-CNT composites obtained with a large properties.
incident area based on IEEE-STD-299 (symbols and lines represent the experi- (iii) Dispersed CNTs can remarkably improve the electrical con-
mental values and their mean values, respectively). ductivity of UHPC up to the percolation threshold. However,

13
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

C D 5K\G QHKPEKFGPV CTGC

5OCNN
.CTIG

(QTOCVKQPQH

5' F$
5' F$
EQPFWEVKXG
(QTOCVKQPQH RCVJGPF
EQPFWEVKXG 1XGTGUVKOCVKQP
RCVJUVCTV &KUVQTVKQP
2GTEQNCVKQP
VJTGUJQNF

%06EQPVGPV (TGSWGPE[ *\
Fig. 20. Effect of the CNTs and size of the incident area on EMI SE.

incorporation of CNTs beyond the percolation threshold does not [6] S. Parveen, S. Rana, R. Fangueiro, M.C. Paiva, Microstructure and mechanical
develop the conductivity because the conductive path is already properties of carbon nanotube reinforced cementitious composites developed using
a novel dispersion technique, Cem. Concr. Res. 73 (2015) 215–227.
constructed at the percolation threshold. [7] A.J.N. MacLeod, F.G. Collins, W. Duan, W.P. Gates, Quantitative microstructural
(iv) The EMI SE of UHPC improved with the CNT content up to the characterisation of Portland cement-carbon nanotube composites using electron
percolation threshold regardless of the testing method because of and x-ray microscopy, Cem. Concr. Res. 123 (2019) 105767.
[8] D.D.L. Chung, Electromagnetic interference shielding effectiveness of carbon ma-
the improved electrical conductivity. However, large specimen terials, Carbon 39 (2001) 279–285.
testing should be required to evaluate the SE for actual structures. [9] C. Beall, E. Delzell, P. Cole, I. Brill, Brain tumors among electronics industry
(v) To obtain reliable SE results for UHPC based on IEEE-STD-299, workers, Epidemiology 7 (1996) 125–130.
[10] J.K. Grayson, Radiation exposure, socioeconomic status, and brain tumor risk in the
which is a representative large specimen testing method, speci-
US air force: a nested case-control study, Am. J. Epidemiol. 143 (1996) 480–486.
mens should be fabricated to have a sufficiently large incident [11] S. Szmigielski, Cancer morbidity in subjects occupationally exposed to high fre-
area. Otherwise, the SE results could be overestimated and biased quency (radiofrequency and microwave) electromagnetic radiation, Sci. Total
Environ. 180 (1996) 9–17.
because of the opening effect.
[12] T.L. Thomas, P.D. Stolley, A. Stemhagen, E.T.H. Fontham, M.L. Bleecker,
P.A. Stewart, R.N. Hoover, Brain tumor mortality risk among men with electrical
CRediT authorship contribution statement and electronics jobs: a case-control study, J. Natl. Cancer Inst. 79 (1987) 233–238.
[13] D.D.L. Chung, Carbon Composites, Butterworth-Heinemann, Cambridge, MA, 2017,
pp. 1–87.
Myungjun Jung:Conceptualization, Investigation, Writing - ori- [14] I.W. Nam, H.K. Lee, J.B. Sim, S.M. Choi, Electromagnetic characteristics of cement
ginal draft.Young-soon Lee:Methodology, Data curation.Sung-Gul matrix materials with carbon nanotubes, ACI Mater. J. 109 (2012) 363–370.
Hong:Writing - review & editing, Supervision.Juhyuk [15] I.W. Nam, H.K. Kim, H.K. Lee, Influence of silica fume additions on electromagnetic
interference shielding effectiveness of multi-walled carbon nanotube/cement
Moon:Writing - review & editing, Supervision. composites, Constr. Build. Mater. 30 (2012) 480–487.
[16] S. Wen, D.D.L. Chung, Partial replacement of carbon fiber by carbon black in
Declaration of competing interest multifunctional cement–matrix composites, Carbon 45 (2007) 505–513.
[17] A.P. Singh, B.K. Gupta, M. Mishra, Govind, A. Chandra, R.B. Mathur, S.K. Dhawan,
Multiwalled carbon nanotube/cement composites with exceptional electromagnetic
The authors declare that they have no known competing financial interference shielding properties, Carbon 56 (2013) 86–96.
interests or personal relationships that could have appeared to influ- [18] S. Avanish Pratap, M. Monika, C. Amita, S.K. Dhawan, Graphene oxide/ferrofluid/
cement composites for electromagnetic interference shielding application,
ence the work reported in this paper.
Nanotechnology 22 (2011) 465701.
[19] B. Wang, Z. Guo, Y. Han, T. Zhang, Electromagnetic wave absorbing properties of
Acknowledgements multi-walled carbon nanotube/cement composites, Constr. Build. Mater. 46 (2013)
98–103.
[20] D. Micheli, R. Pastore, A. Vricella, R.B. Morles, M. Marchetti, A. Delfini, F. Moglie,
This work is supported by the Korea Agency for Infrastructure V.M. Primiani, Electromagnetic characterization and shielding effectiveness of
Technology Advancement (KAIA) grant funded by the Ministry of Land, concrete composite reinforced with carbon nanotubes in the mobile phones fre-
Infrastructure and Transport (Grant 19AUDP-B121595-04). The quency band, Mater. Sci. Eng. B 188 (2014) 119–129.
[21] J. Chen, D. Zhao, H. Ge, J. Wang, Graphene oxide-deposited carbon fiber/cement
Institute of Engineering Research in Seoul National University provided composites for electromagnetic interference shielding application, Constr. Build.
research facilities for this work. Mater. 84 (2015) 66–72.
[22] J.M. Chiou, Q. Zheng, D.D.L. Chung, Electromagnetic interference shielding by
carbon fibre reinforced cement, Composites 20 (1989) 379–381.
References [23] Y. Dai, M. Sun, C. Liu, Z. Li, Electromagnetic wave absorbing characteristics of
carbon black cement-based composites, Cem. Concr. Compos. 32 (2010) 508–513.
[1] Y. Li, H. Li, Z. Wang, C. Jin, Effect and mechanism analysis of functionalized multi- [24] X. Fu, D.D.L. Chung, Submicron carbon filament cement-matrix composites for
walled carbon nanotubes (MWCNTs) on C-S-H gel, Cem. Concr. Res. 128 (2020) electromagnetic interference shielding, Cem. Concr. Res. 26 (1996) 1467–1472.
105955. [25] H. Guan, S. Liu, Y. Duan, J. Cheng, Cement based electromagnetic shielding and
[2] I.W. Nam, S.M. Park, H.K. Lee, L. Zheng, Mechanical properties and piezoresistive absorbing building materials, Cem. Concr. Compos. 28 (2006) 468–474.
sensing capabilities of FRP composites incorporating CNT fibers, Compos. Struct. [26] B. Han, L. Zhang, J. Ou, Smart and Multifunctional Concrete Toward Sustainable
178 (2017) 1–8. Infrastructures, Springer, Singapore, 2017, pp. 247–400.
[3] B. Han, K. Zhang, X. Yu, Enhance the thermal storage of cement-based composites [27] ASTM D4935-18 Standard Test Method for Measuring the Electromagnetic
with phase change materials and carbon nanotubes, J. Sol. Energy Eng. 135 Shielding Effectiveness of Planar Materials, (2018).
(2013) 1–5. [28] IEEE-STD-299 IEEE Standard Method for Measuring the Effectiveness of
[4] S.P. Shah, M.S. Konsta-Gdoutos, Uncoupling modulus of elasticity and strength, Electromagnetic Shielding Enclosures, (2007).
Concr. Int. 39 (2017) 37–42. [29] ASTM E1851-15 Standard Test Method for Electromagnetic Shielding Effectiveness
[5] M.S. Konsta-Gdoutos, Z.S. Metaxa, S.P. Shah, Highly dispersed carbon nanotube of Durable Rigid Wall Relocatable Structures.
reinforced cement based materials, Cem. Concr. Res. 40 (2010) 1052–1059. [30] MIL-STD-188-125-1 High-Altitude EMP Protection for Fixed Ground-Based
Facilities, (2005).

14
M. Jung, et al. Cement and Concrete Research 131 (2020) 106017

[31] S.H. Kang, S.G. Hong, J.H. Moon, The use of rice husk ash as reactive filler in ultra- tension specimens of carbon fiber cement composites, Cem. Concr. Compos. 26
high performance concrete, Cem. Concr. Res. 115 (2019) 389–400. (2004) 873–881.
[32] Z. Chen, J.L.G. Lim, E. Yang, Ultra high performance cement-based composites [41] S. Vaidya, E.N. Allouche, Strain sensing of carbon fiber reinforced geopolymer
incorporating low dosage of plasma synthesized carbon nanotubes, Mater. Des. 108 concrete, Mater. Struct. 44 (2011) 1467–1475.
(2016) 479–487. [42] H.K. Kim, I.W. Nam, H.K. Lee, Enhanced effect of carbon nanotube on mechanical
[33] E. Dujardin, T.W. Ebbesen, H. Hiura, K. Tanigaki, Capillarity and wetting of carbon and electrical properties of cement composites by incorporation of silica fume,
nanotubes, Science 265 (1994) 1850–1852. Compos. Struct. 107 (2014) 60–69.
[34] Y. Homma, S. Chiashi, T. Yamamoto, K. Kono, D. Matsumoto, J. Shitaba, S. Sato, [43] S.H. Lee, S.H. Kim, D.Y. Yoo, Hybrid effects of steel fiber and carbon nanotube on
Photoluminescence measurements and molecular dynamics simulations of water self-sensing capability of ultra-high-performance concrete, Constr. Build. Mater.
adsorption on the hydrophobic surface of a carbon nanotube in water vapor, Phys. 185 (2018) 530–544.
Rev. Lett. 110 (2013) 157402. [44] D.D.L. Chung, Self-heating structural materials, Smart Mater. Struct. 13 (2004)
[35] B. Zou, S.J. Chen, A.H. Korayem, F. Collins, C.M. Wang, W.H. Duan, Effect of ul- 562–565.
trasonication energy on engineering properties of carbon nanotube reinforced ce- [45] F.J. Baeza, O. Galao, E. Zornoza, P. Garcés, Multifunctional cement composites
ment pastes, Carbon 85 (2015) 212–220. strain and damage sensors applied on reinforced concrete (RC) structural elements,
[36] A.H. Korayem, N. Tourani, M. Zakertabrizi, A.M. Sabziparvar, W.H. Duan, A review Materials 6 (2013) 841–855.
of dispersion of nanoparticles in cementitious matrices: nanoparticle geometry [46] M. Saafi, Wireless and embedded carbon nanotube networks for damage detection
perspective, Constr. Build. Mater. 153 (2017) 346–357. in concrete structures, Nanotechnology 20 (2009) 395502.
[37] B. Han, K. Zhang, X. Yu, E. Kwon, J. Ou, Fabrication of piezoresistive CNT/CNF [47] J.-L. Le, H. Du, S.D. Pang, Use of 2D Graphene Nanoplatelets (GNP) in cement
cementitious composites with superplasticizer as dispersant, J. Mater. Civ. Eng. 24 composites for structural health evaluation, Compos. Part B 67 (2014) 555–563.
(2012) 658–665. [48] R. Perez, Handbook of Electromagnetic Compatibility, Academic Press, San Diego,
[38] M.J. Jung, Y.S. Lee, S.G. Hong, Study on improvement in electromagnetic inter- 1995, pp. 401–443.
ference shielding effectiveness of ultra-high performance concrete (UHPC)/carbon [49] J.D. Kraus, R.J. Marhefka, Antennas for all Applications, McGraw-Hill, New York,
nanotube (CNT) composites, J. Korea Concr. Inst. 31 (2019) 69–77. 2002, pp. 17–85.
[39] F. Naeem, H.K. Lee, H.K. Kim, I.W. Nam, Flexural stress and crack sensing cap- [50] M.J. Jung, Y.S. Lee, S.G. Hong, Effect of incident area size on estimation of EMI
abilities of MWNT/cement composites, Compos. Struct. 175 (2017) 86–100. shielding effectiveness for ultra-high performance concrete with carbon nanotubes,
[40] F. Reza, J.A. Yamamuro, G.B. Batson, Electrical resistance change in compact IEEE Access 7 (2019) 183105–183117.

15