Construction and Building Materials 123 (2016) 327–335

Contents lists available at ScienceDirect

Construction and Building Materials

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

Incorporation of graphene oxide and silica fume into cement paste: A

study of dispersion and compressive strength
Xiangyu Li a, Asghar Habibnejad Korayem b, Chenyang Li c, Yanming Liu c, Hongsen He c, Jay G. Sanjayan d,
Wen Hui Duan c,⇑
a
Department of Civil and Environmental Engineering, The Hong Kong University of Science and Technology, Hong Kong, China
b
Department of Civil Engineering, Iran University of Science and Technology, Tehran, Iran
c
Department of Civil Engineering, Monash University, Clayton, VIC 3800, Australia
d
Centre for Sustainable Infrastructure, Faculty of Science, Engineering and Technology, Swinburne University of Technology, Hawthorn, VIC 3122, Australia

h i g h l i g h t s

 Dispersions of graphene oxide in pore solution and cement paste was studied.
 Mechanism of GO aggregation in cement paste is cross-linking of calcium cations.
 Silica fume is effective to improve GO dispersion in cement paste.
 Appropriate amount of silica fume improves GO dispersion and compressive strength.

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

Article history: As a 2D nanomaterial, graphene oxide (GO) has attracted considerable attentions for reinforcing cemen-
Received 4 January 2016 titious materials. However, special attention must be paid to the dispersion of GO in the matrix, as the
Received in revised form 7 July 2016 quality of nanomaterial dispersion correlates directly with its effectiveness for improving mechanical
Accepted 8 July 2016
and other properties. In this study, the dispersion of graphene oxide (GO) in simulated pore solution
Available online 15 July 2016
and cement paste was investigated. It was found that severe GO aggregation occurred in presence of
divalent calcium ions in both pore solution and cement paste. However, the GO aggregates were not
Keywords:
stable under shear mixing. After vigorous mixing, the massive GO aggregates split into medium-sized
Graphene oxide
Silica fume
particles, ranging from few to several 100 lm. To improve the GO dispersion in cement paste, silica fume
Dispersion was used to mechanically separate individual GO nanosheets. The dispersion was then investigated using
Cement paste microstructure analysis and mechanical properties. The results showed that, with the addition of silica
Pore solution fume, the dispersion of GO nanosheets was greatly improved.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction edges [1]. Due to the presence of these oxygen-containing

functional groups, GO can readily yield stable dispersions in water,
Cement is the principal binder material in the production of consisting mostly of 1-nm-thick sheets [2]. In terms of mechanical
concrete, which is the most widely used construction material in properties, the elastic modulus and tensile strength of GO is
the world. However, the major limitation of concrete and cement around 32 GPa and 130 MPa [3,4], respectively. Most importantly,
based materials is their quasi-brittle nature, which is blamed for it can be synthesized in large quantities from inexpensive graphite
its poor resistance to crack formation, low tensile strength and powder. As such, GO has attracted much attention in various
strain capacities. applications [3,5,6]. One of those applications is to reinforce
Recently discovered graphene oxide (GO) is a layered nanoma- cement based materials. Successful incorporation of GO into
terial consisting of hydrophilic oxygenated graphene sheets, bear- cement based materials can address the issues of cement based
ing hydroxyl and epoxide functional groups on their basal planes materials by virtue of its excellent mechanical properties, large
and having carbonyl and carboxyl groups located at the sheet specific surface area, large aspect ratio, and functional groups
favorable for cement hydration products.
⇑ Corresponding author. A few studies have investigated effects of GO nanosheets on
E-mail address: [email protected] (W.H. Duan). mechanical and transport properties of cement based materials,

http://dx.doi.org/10.1016/j.conbuildmat.2016.07.022
0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.
328 X. Li et al. / Construction and Building Materials 123 (2016) 327–335

including cement paste and cement mortar. Gong et al. [7] 4 mg/ml. A SEM image of GO is shown in Fig. 1. The mean size of
reported that the introduction of 0.03% by weight GO nanosheets the GO nanosheets was around 1 lm. The major functional groups
into cement paste increased the compressive strength and tensile on the GO surface were found to be –OH and –COOH. Elemental
strength by more than 40%. Zhu et al. [8] reported that the intro- analysis of the GO is shown in Table 1.
duction of 0.05% by weight of GO into cement paste increased com-
pressive strength by 15–33% and flexural strength by 41–59%. Lv 2.2. Cement powder
et al. [9,10] showed that mechanical properties were improved
by the incorporation of GO in cement paste and mortar, respec- General purpose ordinary Portland cement (OPC) conforming to
tively. To take advantage of the high surface area of GO, the trans- the requirements of Type I – Normal Cement, as defined by ASTM C
port properties of GO incorporated cement mortar were studied 150, was used throughout the study. The chemical composition of
[11]. The findings revealed enhancements in water sorptivity and the cement powder as determined by X-ray fluorescence is shown
chloride penetration. It should be noted that these studies paid lit- in Table 2.
tle attention to dispersion of GO nanosheets in the cement based
materials. 2.3. Silica fume
Although the incorporation of GO improves the properties of
cement based materials, its dispersion in cement based materials The chemical composition of the silica fume as determined by
still remains unknown. Park et al. [12] reported that GO X-ray fluorescence is shown in Table 3. The particle size distribu-
nanosheets can be chemically cross-linked by divalent cations to tion of silica fume as measured by a laser particle size analyzer
produce GO paper with much larger dimensions than that of (Cilas 1190) is shown in Fig. 2. The mean particle diameter is
original GO nanosheets. It is foreseeable that dispersion of GO 0.37 lm, the surface area of the silica fume particles was around
would not be good due to abundance of calcium cations in fresh 20 m2/g.
cement paste. Therefore, it is imperative to study the nature of
GO dispersion and how to improve that dispersion, as the quality 2.4. Cement paste preparation
of nanomaterial dispersion in the matrix correlates directly with
its effectiveness for improving mechanical and other properties. A high-speed shear mixer (CTE Model 7000) was employed to
In this study, the dispersion of GO in both simulated pore solu- prepare cement paste with GO and silica fume. Mixing procedures
tion (saturated Ca(OH)2 solution) and cement paste was first inves- similar to those in ASTM C1738-11a were adopted:
tigated. It was found that the stability of GO was greatly influenced
by divalent calcium ions. Severe aggregation of GO occurred when
it was introduced into the simulated pore solution (saturated Ca
(OH)2 solution) and cement paste. Therefore, GO dispersion must
be improved to more effectively enhance both mechanical and
other properties of cement based materials.
Silica fume is very fine amorphous silica produced by electric
arc furnace as a by-product of the production of metallic silicon
or ferrosilicon alloys [13]. It is a powder with particle size 100
times finer than that of Portland cement particles. It is also poz-
zolanic. Silica fume has been used in cement mortar and concrete
and significant effects on the properties of the resulting materials
have been identified [14,15]. These effects pertain to properties
such as strength [16], permeability [17], and bonding strength with
aggregates [18].
A few attempts have been made to improve the dispersion of
carbon nanofibers (CNFs) and carbon nanotubes (CNTs) in cement
paste by the incorporation of silica fume. Sanchez et al. [19]
reported that silica fume facilitated the dispersion of CNFs by act-
Fig. 1. SEM image of GO.
ing as wedges to mechanically separate the carbon nanofibers. Kim
et al. [20] found that the dispersion of CNTs was improved by the
incorporation of silica fume in cement paste, leading to enhance- Table 1
ments in mechanical and electrical properties. Yazdanbakhsh Elemental analysis of GO.

et al. [21] reported that silica fume, if used in sufficient propor- Element Carbon Hydrogen Nitrogen Sulphur Oxygen
tions, largely prevented the reagglomeration of CNFs in fresh % 49–56 0–1 0–1 0–2 41–50
cement paste.
In this study, a new mixing procedure was designed that used
silica fume to pre-disperse GO nanosheets before mixing with
cement. The GO dispersion was then investigated by both Table 2
Chemical composition of cement powder.
microstructure analysis and mechanical testing using scanning
electron microscopy (SEM) and compressive strength, respectively. Al2O3 SiO2 CaO Fe2O3 K2O MgO Na2O SO3 LOI
4.7 19.9 63.9 3.4 0.5 1.3 0.2 2.6 3.0

2. Materials and sample preparations

Table 3
2.1. GO Chemical composition of silica fume.

Al2O3 SiO2 CaO Fe2O3 K2O MgO Na2O SO3

GO solution was purchased from GrapheneaÒ. In the solution, – 96.43 0.78 0.073 1.28 0.70 – 0.38
GO nanosheets were dispersed in water at the concentration of
 X. Li et al. / Construction and Building Materials 123 (2016) 327–335 329

25 100 Table 4
Mix proportions of cement paste for SEM study.

Mix Water to binder⁄ ratio Silica fume (%) GO (%)

20 80

Cumulative volume (%)

CG0 1 0 0.4
CGS20 1 20% 0.4
15 60 CGS50 2 50% 0.8
Volume (%)

Note: Binder is combination of cement and silica fume.

10 40

Table 5
5 20
Mix proportion of cement paste for compressive strength study.

Mix Water to binder⁄ ratio Silica fume (%) GO (%)

0 0
0 1 2 3 4 CS0, CS2, CS5, CS10 0.4 0, 2, 5, 10 0
CGS0, CGS2, CGS5, CGS10 0.4 0, 2, 5, 10 0.02
Diameter (µm)
Note: Binder is combination of cement and silica fume.
Fig. 2. Particle size distribution of silica fume.

– Conduct sonication on diluted GO solution using a horn

sonicator (VCX 500 W) for 5 min to ensure that the GO is well
dispersed. To avoid raising the temperature, a water-ice bath
is set up during the sonication.
– Place the diluted GO solution in the bowl. If silica fume is used,
add silica fume at this stage; then start the mixer and mix at
slow speed (4000 rpm) for 30 s to disperse GO and silica fume.
– Add the entire quantity of cement slowly into the solution over
a 30 s period, while mixing at slow speed (4000 rpm).
– Continue mixing for another 30 s at slow speed (4000 rpm).
– Switch the mixing speed to high speed (12,000 rpm) and mix for
30 s.
– Stop the mixer and let the paste stand for 150 s. During this
interval, quickly scrape down into the batch any material that
may have collected on the side of the bowl; then for the remain-
der of this interval, close the mixer enclosure or cover the bowl
with the lid.
– Finish by mixing for 30 s at high speed (12,000 rpm).
Fig. 3. GO water solution (left) and GO aggregation in Ca(OH)2 solution (right).

After mixing of the cement paste, the fresh mixture was cast
represents cement paste incorporating 0.4% GO, CGS20 represents
into molds and vibrated on a vibration table to ensure good com-
the cement paste incorporating pre-dispersed 0.4% GO and 20% sil-
paction. The molds were then sealed with polyethylene sheets to
ica fume by weight of cement, and CGS50 represents the cement
prevent moisture escape. After 24 h, the samples were demolded
paste incorporating pre-dispersed 0.8% GO and 50% silica fume
and cured in a saturated lime water bath at 20 °C until testing.
by weight of cement.
The microstructural observations were made using a field emis-
3. Experiments sion scanning electron microscope (JEOL 7001F FEG). To make the
samples conductive, a thin layer of gold–palladium was deposited
3.1. Stability of GO in Ca(OH)2 test onto them using sputter deposition. To identify the GO and also to
demonstrate the dispersion of GO nanosheets, energy-dispersive
A certain amount of GO solution was added to saturated Ca X-ray spectroscopy (EDS) spectra were collected using an Oxford
(OH)2 solution. After gentle shaking, the stability of GO in the Ca instruments Aztec system on the surface of the samples.
(OH)2 solution was visually observed. The GO-Ca(OH)2 solution
was then mixed continuously using a magnetic mixer for 10 h. It 3.3. Compressive strength
is believed that 10-h mixing is enough for the GO-Ca(OH)2 becom-
ing stable. The GO aggregates were observed by using XJP300 To examine the influence of incorporating GO and of GO disper-
microscope (ProSciTech Pty Ltd). After 10-h mixing, a small sion on the mechanical properties of cement paste, compression
amount of the solution was removed and diluted. GO dispersion tests were conducted on cylindrical specimens (23.5  47 mm).
in the diluted solution was characterized by UV–vis spectroscopy Relatively small specimens were used due to the availability of
(Shimadzu UV 4800) with a wavelength range of 190–1100 nm. GO for this study.
In compression testing, the specimens were tested at the ages of
3.2. SEM analysis 3, 7, and 28 days, following ASTM C109. The loading rate was set at
0.2 mm/min, corresponding to approximately 0.3 MPa/s. The tests
GO dispersion in the pore solution and cement paste was inves- were performed using a Shimadzu testing machine with a capacity
tigated by SEM. A certain amount of GO was dispersed in saturated of 300 kN. At least eight samples were repeated for each test.
Ca(OH)2 solution. To facilitate observation of the GO in the cement The mix proportions of cement paste for the compressive
paste, a high dosage of GO was employed, because it was difficult strength study are shown in Table 5. Two series of samples were
to identify GO when the dosage was lower. The mix proportions of prepared, where CS represents the cement paste with water/cement
cement paste for the SEM study are shown in Table 4, where CG0 ratio 0.4 and silica fume replacement ratios 0%, 2%, 5%, 10%; CGS
330 X. Li et al. / Construction and Building Materials 123 (2016) 327–335

Fig. 4. SEM image (a) and elemental analysis (b) of GO aggregates.

represents the GO incorporated cement paste with 0.02% GO reported in the work of Hua et al. [25] in which, as the shear rate
by weight of cement and silica fume replacement ratios 0%, 2%, increased, the massive GO aggregates split into pieces of
5%, 10%. Thus, in total 8 samples were designed for compressive medium-sized particles, indicating fragmentation of large-scale
strength study. aggregates. In the current study, the lateral size of GO aggregates
were measured using an optical microscope, as shown in Fig. 5c
4. Results and discussion and Fig. 5d. The aggregate size was found to range from a few
micrometers to 100 s of micrometers.
4.1. Aggregation of GO in Ca(OH)2 solution The dispersion of GO-Ca(OH)2 and GO water solution was char-
acterized using UV–vis spectroscopy. For these two solutions, two
The stability of GO in Ca(OH)2 was examined by introducing a references, saturated Ca(OH)2 solution and water, were also mea-
small amount of GO solution into saturated Ca(OH)2 solution. In sured by UV–vis spectroscopy. The concentrations of solid GO
this case, the concentration of solid GO nanosheets in the solution nanosheets in these two solutions were both set at 0.04 mg/mL.
was set at 0.04 mg/mL. As can be seen in Fig. 3, aggregation of GO Fig. 6 shows the UV–vis curves of GO water solution and GO-Ca
occurred immediately when it was introduced into Ca(OH)2 solu- (OH)2 solution after mixing. The UV curves of GO and GO-Ca
tion. Compared with the GO water solution (Fig. 3 left), the aggre- (OH)2 solution were obtained after subtraction of the UV results
gation of GO made the solution cloudy. Without being disturbed, of water and saturated Ca(OH)2 solutions, respectively.
the GO aggregation settled down. For GO dispersed in water, as can be seen in Fig. 6, two charac-
The GO aggregates were observed under SEM. As can be seen in teristic features were observed. The band located at 230 nm is
Fig. 4, GO aggregates with wrinkled and folded surface were found. assigned to the p-p⁄ transitions of C@C bonds, and the shoulder
The size of the aggregates was much greater that of original GO band situated at about 300 nm is ascribed to the n-p⁄ transitions
nanosheets, which was around 1 lm. It has been reported that, of C@O bonds. For GO dispersed in Ca(OH)2 solution, the UV curve
in the presence of divalent ions, the mechanisms of GO aggregation exhibited two shoulders, one located at 230 nm and the other at
kinetics can be more complex because divalent ions can not only 300 nm. There was no single peak observed after the wavelength
screen surface charges but also interact with surface functional of 200 nm. A recent study of Chen et al. [26] reported similar
groups of GO sheets to cross-link them [12]. There are three types UV–vis results for divalent cations modified GO, which proved
of cross-linking interaction: bridging the edges of the GO sheets, existence of chemical cross-linking between divalent cations and
intercalating between the basal planes, and cross-linking of the functional groups on GO sheets.
hydrogen bonds formed among the oxygen functional groups on Lai et al. [27] recently reported that dispersions containing few
GO surfaces and the interlamellar water molecules [12,22,23]. layers (1–3 layers) of GO could be distinguished from those con-
Therefore, both lateral size and thickness increased after aggrega- taining multiple layers (4–10 layers) and thick layers (>10 layers)
tion, as observed in Fig. 3a. EDS results, as shown in Fig. 3b, of GO by the intense peak in their UV–vis spectra at 230 nm [27].
revealed large amounts of carbon and calcium in the GO aggregate, The few-layer GO showed a single peak at 230 nm, whereas the
confirming that GO nanosheets were cross-linked by calcium ions. multi-layer GO exhibited shoulder only. For the thick-layer GO dis-
Similar results have been found for carbon nanotubes (CNTs). persion, no peak or shoulder was observed in the UV results [27].
Mendoza et al. [24] studied the effect of superplasticizer and Ca The GO solution they prepared exhibited features similar to those
(OH)2 on the stability of OH- functionalized multiwalled CNT of the GO-Ca(OH)2 in the current study. The GO solution was
(MWCNT) dispersions. They found that Ca(OH)2 affects the stability mainly a mixture of multi-layer GO, with 4–10 layers accounting
of MWCNT dispersion due to its interaction with negative charges for 81% of the sample. Therefore, it is believed that in the current
of the OH– functional groups. Therefore, for GO and CNTs incorporated study, the GO aggregations in the Ca(OH)2 solution were mainly
in cement based materials, instability of these surface functional- multi-layer (4–10 layers) GO cross-linked by calcium ions.
ized nanomaterials caused by calcium ions must be considered. As shown in the results, GO aggregated in presence of calcium
To study GO aggregation after shear mixing, the GO-Ca(OH)2 ions. The GO aggregates were described in the study of Park et al.
solution was then stirred using a magnetic mixer at 300 rpm. [12] as GO paper modified by divalent ions. In current study, it
Fig. 5a shows GO aggregates immediately after introduction and was found that the aggregates were not stable under shear mixing,
Fig. 5b shows GO aggregates after mixing for 10 h. It is evident although severe aggregation occurred when GO was introduced
from Fig. 5 that the size of GO aggregates was reduced significantly into Ca(OH)2 solution. Large GO aggregates could be broken into
after mixing. The results indicate that breakage occurred of large small aggregates after shear mixing. As we detailed, three cross-
GO aggregates into small GO aggregates. Similar results were linking modes were evident between GO nanosheets. The addition
 X. Li et al. / Construction and Building Materials 123 (2016) 327–335 331

Fig. 5. GO aggregates immediately after introduction into Ca(OH)2 solution (a) and after 10 h mixing (b) optical observations of GO aggregates after mixing (c, d).

of divalent ions led to aggregation of the GO nanosheets due to is the dominant and stronger interaction mode of GO aggregation
strong cross-linking of GO sheets by bridging the edges of the GO in the presence of Ca2+ ions [22]. Therefore, during shear mixing,
sheets and intercalating between the basal planes. Edge-to-edge the GO aggregates tend to break through basal planes, not edge-
to-edge. This characteristic is favorable for maintaining the high
aspect ratio of GO aggregates. However, the mechanical properties
0.6
GO water solution of GO aggregates have been found to be orders of magnitude lower
GO-Ca(OH)2 solution after mixing than that of monolayer GO nanosheets [12,28]. Typically, the ten-
sile strength of GO paper is 100 s MPa [12], compared to
24.7 GPa of monolayer GO [28]. Strength and stiffness were found
0.4
to be strongly correlated to the effective volume and thickness of
Absorbance

the GO samples, respectively. Larger samples led to a decrease in

effective stiffness [29]. Also, brittle failure has been observed in a
highly localized region through cross-section of the GO nanosheets
0.2 without interlayer pull-out [29]. Therefore, it is desirable to avoid
GO aggregation in order to take full advantage of excellent
mechanical properties of monolayer GO nanosheets.

0.0
200 300 400 500 600 700 800 900 1000 1100 4.2. GO aggregation in cement paste without silica fume
Wave length (nm)
Fig. 7 shows GO aggregates observed in the cement paste sam-
0.18 ple CGS0, in which 0.4% GO and no silica fume was used. The GO
0.16 aggregate had wrinkled and folded morphology as shown in
GO-Ca(OH)2 solution after mixing Fig. 7a and b. The length of the aggregate was roughly up to
0.14
10 lm. However, the thickness was less than 1 lm. The EDS results
0.12 show that the aggregate contained large amount of carbon and cal-
cium, confirming the cross-linking interactions between GO
Absorbance

0.10
nanosheets and calcium ions.
0.08

0.06 4.3. Dispersion of GO in cement paste with silica fume

0.04
Fig. 8 shows SEM images of GO solution and silica fume mixture.
0.02 It can be seen that semi-transparent GO layers were covered by
silica fume particles located on the surface and edges. There are
0.00
200 300 400 500 600 700 800 900 1000 1100
two possible mechanisms for the prevention of GO aggregation by
Wave length (nm) silica fume. First, these silica fume particles could prevent aggrega-
tion of GO by mechanically separating GO nanosheets from calcium
Fig. 6. UV–vis curves of GO, Ca(OH)2, and GO-Ca(OH)2 solution after mixing. ions. Secondly, once silica fume and GO solution mixture are added
332 X. Li et al. / Construction and Building Materials 123 (2016) 327–335

Fig. 7. GO aggregates in cement paste without silica fume (a, b) EDS analysis (c, d).

into the cement paste, the silica fume reacts with Ca(OH)2 to pro- rating silica fume. The dosages of silica fume in CGS20 and CGS50
duce C-S-H and reduce the concentration of calcium ions around were calculated to ensure that the surface area of silica fume was
the GO nanosheets. It has been shown that such a chemical (poz- equal to that of the GO nanosheets. It is believed that sufficient sil-
zolanic) reaction can begin at early hydration times [18]. ica fume should be used to mechanically separate GO nanosheets
Fig. 9 shows SEM images of GO and GO-silica fume mixture. from calcium ions.
Fig. 9b is a backscattered electron image of a GO-silica fume mix- It should be noted that the hydration of cement paste samples
ture, in which the darker parts are GO nanosheets. From the size of (CGS20 and CGS50) was stopped at 1 day by immersion in ethanol
GO nanosheets it can be deduced that mixing silica fume with GO in order to observe the GO nanosheets. Otherwise, the hydration
did not cause any aggregation. products would cover the GO surface, making observation very dif-
Figs. 10 and 11 show GO nanosheets observed in cement paste ficult. This is why almost no hydration products are observed in
samples (CGS20 and CGS50) with silica fume. It can be seen that current SEM images.
semi-transparent GO layers were covered by silica fume particles,
on both the surfaces and edges. The GO nanosheets also exhibited 4.4. Compressive strength of GO-cement paste with silica fume
wrinkled and folded morphology, a feature that is favorable for
better mechanical interlocking with cement hydration products. Fig. 12 shows the effects of the addition of GO on the
No GO aggregations were observed in these two samples incorpo- compressive strength of the cement paste samples with different
 X. Li et al. / Construction and Building Materials 123 (2016) 327–335 333

Fig. 8. Silica fume and GO solution mixture.

Fig. 9. SEM images of GO (a) and GO-silica fume mixture (b).

Fig. 10. GO nanosheets in cement paste sample (CGS20) with silica fume.

Fig. 11. GO nanosheets in cement paste sample (CGS50) with silica fume.

dosages of silica fume. CS was a series of cement paste samples At 7 days, CGS0 and CS0 had similar strength, whereas CGS2,
incorporating 0%, 2%, 5%, and 10% of silica fume. The only differ- CGS5, and CGS10 had higher strength than their counterparts
ence between CGS and CS was that CGS had 0.02% added GO. At CS2, CS5, and CS10. At 28 days, only CGS2 had higher strength than
3 days, it can be seen that CGS had higher compressive strength CS2, whereas the other CGS samples had lower strength than their
than that of CS except for the sample (CGS0) without silica fume. counterparts.
334 X. Li et al. / Construction and Building Materials 123 (2016) 327–335

For CGS0, the addition of GO produced no improvement in early hydration stages (3 days), CGS5 and CGS10 exhibited
compressive strength. At 28 days, the compressive strength of strength similar to that of CS5 and CS10. At 7 and 28 days, slight
CGS0 was lower than that of CS0. The possible reason is that the decreases of strength were observed in CGS5 and CGS10. In
GO aggregated in CGS0 and these GO aggregations might have acted CGS2, the reinforcing effect of the addition of GO might be due
as pores and cracks, resulting in a decrease in compressive strength. to the improved mechanical interlocking between GO nanosheets
It was demonstrated that GO dispersion improved with the and hydration products, crack arresting by GO nanosheets, promo-
addition of silica fume. No GO aggregation was found in cement tion of hydration processes, or interfacial forces due to chemical
paste incorporating sufficient silica fume. Therefore, in CGS2, reactions between functional groups and hydration products
CGS5, and CGS10, GO dispersion was assumed to be similar, with- [7,8]. The absence of reinforcing effects from the addition of GO
out any aggregation. However, only CGS2 showed reinforcing in CGS5 and CGS10 could be due to an overdose of silica fume cov-
effects on compressive strength, regardless of hydration age. At ering the surface of the GO, preventing interactions between GO
nanosheets and hydration products by mechanically separating
the GO nanosheets from the hydration products. An appropriate
56
amount of silica fume should therefore be employed to ensure both
3d compressive strength
better GO dispersion and enhanced reinforcing effect of the GO on
54 mechanical properties. The appropriate amount of silica fume can
Compressive strength (MPa)

be determined by ensuring that the silica fume has a similar sur-

52 face area to that of the GO nanosheets. In this case, CGS2 had 2%
silica fume and 0.02% GO. The specific surface areas of the silica
50 fume and GO nanosheets were 20 m2/g and 2600 m2/g, respec-
tively. Therefore, the silica fume and the incorporated GO in
CS
CGS2 had roughly equivalent surface areas.
48 CGS
Fig. 13 shows schematic plot of GO nanosheets covered with
appropriate and over-dosed amount of silica fume particles. As it
46
can be seen, there is no exposed surface area for GO to be interact-
ing with cement hydration products if over-dosed silica fume was
44 employed. However, for GO covered with appropriate amount of
0 2 5 10 silica fume, there is still exposed surface area to interact with
Silica fume (wt. % of cement) cement hydration products. In this study, it was found the appro-
priate amount of silica fume could be determined by making its
70 surface area equal to that of GO sheets. Theoretically, such amount
68
7d compressive strength of silica fume would not completely cover surface area of GO which
makes interactions between cement hydration products with GO
Compressive strength (MPa)

66 sheets possible. Silica fume is pozzolanic. However, the hydrated

silica fume is not the bulk hydration phase in cement paste. There-
64
fore, in this study, the lower compressive strength of CGS5 and
62 CGS10 was ascribed to less chances of interactions between GO
and cement hydration products.
60 CS Similar results were reported for CNT reinforced cement paste
CGS
[20]. Appropriate amounts of silica fume were needed to disperse
58
the CNTs effectively and enhance the mechanical properties. The
56 addition of further silica fume did not correlate with greater
improvement in mechanical properties.
54 Fig. 14 shows the compressive strength development of cement
0 2 5 10
paste samples CS and CGS. The compressive strength of all eight
mix proportions increased with hydration age. The compressive
Silica fume (wt. % of cement)
strength also increased with the increase of silica fume content
in the sample. It can be seen, however, that the effect of the addi-
100
28d compressive strength tion of silica fume on compressive strength was not significant at
an early stage (i.e., 3 days). It became obvious at later ages (i.e., 7
Compressive strength (MPa)

95 and 28 days). Similar results were reported in a previous study

[30] of blended cement paste with silica fume.

90

85 CS
CGS

80

75

0 2 5 10
Silica fume (wt. % of cement)

Fig. 12. Compressive strength of cement paste samples (CS and CGS) at 3, 7, and
28 days. Fig. 13. Schematic plot of GO nanosheets covered with silica fume particles.
 X. Li et al. / Construction and Building Materials 123 (2016) 327–335 335

100 References
CS
[1] S. Stankovich et al., Graphene-based composite materials, Nature 442 (7100)
Compressive strength (MPa)

90 (2006) 282–286.
[2] S. Stankovich et al., Stable aqueous dispersions of graphitic nanoplatelets via
3d the reduction of exfoliated graphite oxide in the presence of poly (sodium 4-
80 7d styrenesulfonate), J. Mater. Chem. 16 (2) (2006) 155–158.
[3] Y. Zhu et al., Graphene and graphene oxide: synthesis, properties, and
28d
applications, Adv. Mater. 22 (35) (2010) 3906–3924.
70 [4] D.A. Dikin et al., Preparation and characterization of graphene oxide paper,
Nature 448 (7152) (2007) 457–460.
[5] Y. Wang et al., Graphene and graphene oxide: biofunctionalization and
60 applications in biotechnology, Trends Biotechnol. 29 (5) (2011) 205–212.
[6] D. Chen, H. Feng, J. Li, Graphene oxide: preparation, functionalization, and
electrochemical applications, Chem. Rev. 112 (11) (2012) 6027–6053.
50 [7] K. Gong et al., Reinforcing effects of graphene oxide on portland cement paste,
J. Mater. Civ. Eng. (2014).
[8] Z. Pan et al., Mechanical properties and microstructure of a graphene oxide–
0 2 5 10 cement composite, Cement Concr. Compos. 58 (2015) 140–147.
[9] S. Lv et al., Effect of GO nanosheets on shapes of cement hydration crystals and
Silica fume (wt. % of cement)
their formation process, Constr. Build. Mater. 64 (2014) 231–239.
[10] S. Lv et al., Effect of graphene oxide nanosheets of microstructure and
mechanical properties of cement composites, Constr. Build. Mater. 49 (2013)
100 121–127.
CGS [11] A. Mohammed et al., Incorporating graphene oxide in cement composites: a
study of transport properties, Constr. Build. Mater. 84 (2015) 341–347.
Compressive strength (MPa)

90 [12] S. Park et al., Graphene oxide papers modified by divalent ions—enhancing

mechanical properties via chemical cross-linking, ACS Nano 2 (3) (2008) 572–
3d 578.
80 7d [13] H. Ji et al., Synthesis of Si 3 N 4 powder with tunable a/b-Si 3 N 4 content from
28d waste silica fume using carbothermal reduction nitridation, Powder Technol.
252 (2014) 51–55.
70 [14] D.D. Chung, Multifunctional Cement-based Materials, CRC Press, 2003.
[15] Y. Senhadji et al., Influence of natural pozzolan, silica fume and limestone fine
on strength, acid resistance and microstructure of mortar, Powder Technol.
60 254 (2014) 314–323.
[16] H.A. Toutanji, T. El-Korchi, The influence of silica fume on the compressive
strength of cement paste and mortar, Cem. Concr. Res. 25 (7) (1995) 1591–
50 1602.
[17] H.-W. Song et al., Estimation of the permeability of silica fume cement
concrete, Constr. Build. Mater. 24 (3) (2010) 315–321.
0 2 5 10 [18] X. Ping, J.J. Beaudoin, Modification of transition zone microstructure—Silica
Silica fume (wt. % of cement) fume coating of aggregate surfaces, Cem. Concr. Res. 22 (4) (1992) 597–604.
[19] F. Sanchez, C. Ince, Microstructure and macroscopic properties of hybrid
Fig. 14. Compressive strength of cement pastes CS and CGS. carbon nanofiber/silica fume cement composites, Compos. Sci. Technol. 69 (7)
(2009) 1310–1318.
[20] H. Kim, I. Nam, H. Lee, Enhanced effect of carbon nanotube on mechanical and
electrical properties of cement composites by incorporation of silica fume,
5. Conclusions Compos. Struct. 107 (2014) 60–69.
[21] A. Yazdanbakhsh, Z. Grasley, Utilization of silica fume to stabilize the
dispersion of carbon nanofilaments in cement paste, J. Mater. Civ. Eng. (2014).
In this study, it was found that GO nanosheets aggregated in [22] L. Wu et al., Aggregation kinetics of graphene oxides in aqueous solutions:
simulated pore solution if no prevention was employed. The aggre- experiments, mechanisms, and modeling, Langmuir 29 (49) (2013) 15174–
gates formed a paper-like structure by cross-linking of calcium 15181.
[23] N.V. Medhekar et al., Hydrogen bond networks in graphene oxide composite
ions in the pore solution. Without prevention, GO aggregates were
paper: structure and mechanical properties, ACS Nano 4 (4) (2010) 2300–
also found in cement paste. The size of GO aggregates could be 2306.
reduced by shear mixing. If a sufficient amount of silica fume [24] O. Mendoza, G. Sierra, J.I. Tobón, Influence of super plasticizer and Ca (OH) 2 on
the stability of functionalized multi-walled carbon nanotubes dispersions for
was used, GO dispersion in the cement paste was highly improved,
cement composites applications, Constr. Build. Mater. 47 (2013) 771–778.
with the silica fume mechanically separating the GO sheets and [25] Z. Hua et al., Aggregation and resuspension of graphene oxide in simulated
preventing aggregation. However, excess silica fume had a nega- natural surface aquatic environments, Environ. Pollut. 205 (2015) 161–169.
tive effect on the compressive strength of the GO reinforced [26] K. Chen et al., Strong and tough layered nanocomposites with buried
interfaces, ACS Nano (2016).
cement paste by preventing interactions between the GO and [27] Q. Lai et al., Ultraviolet-visible spectroscopy of graphene oxides, AIP Adv. 2 (3)
cement hydration products. The results of this study may be used (2012) 032146.
to establish guidelines for mix design and practical applications of [28] C. Cao et al., High strength measurement of monolayer graphene oxide, Carbon
81 (2015) 497–504.
GO reinforced cement composites which exhibit both improved GO [29] C. Cao et al., Strengthening in graphene oxide nanosheets: Bridging the gap
dispersion and enhanced mechanical properties. between interplanar and intraplanar fracture, Nano Lett. 15 (10) (2015) 6528–
6534.
[30] J. Yajun, J.H. Cahyadi, Effects of densified silica fume on microstructure and
Acknowledgements compressive strength of blended cement pastes, Cem. Concr. Res. 33 (10)
(2003) 1543–1548.
The authors are grateful for the financial support of the
Australian Research Council in conducting this study.