CARBON 4 8 ( 2 0 1 0 ) 1 7 4 3 –1 7 4 9

available at www.sciencedirect.com

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

Preparation and electrical properties of graphene

nanosheet/Al2O3 composites

Yuchi Fan a, Lianjun Wang c, Jianlin Li b,*, Jiaqi Li a, Shikuan Sun a,

Feng Chen a, Lidong Chen a, Wan Jiang a,c,*
a
State Key Laboratory of High Performance Ceramics and Superfine Microstructures, Shanghai Institute of Ceramics,
Chinese Academy of Sciences, Shanghai 200050, China
b
School of Materials Science and Engineering, Shanghai University, 149 Yanchang Road, Shanghai 200044, China
c
State Key Laboratory for Modification of Chemical Fibers and Polymer Materials, Donghua University, Shanghai 201620, China

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

Article history:
Received 15 October 2009
Accepted 12 January 2010
Available online 18 January 2010
Fully dense graphene nanosheet(GNS)/Al2O3 composites with homogeneously distributed
GNSs of thicknesses ranging from 2.5 to 20 nm have been fabricated from ball milled
expanded graphite and Al2O3 by spark plasma sintering. The percolation threshold of elec-
trical conductivity of the as-prepared GNS/Al2O3 composites is around 3 vol.%, and this
new composite outperforms most of carbon nanotube/Al2O3 composites in electrical con-
ductivity. The temperature dependence of electrical conductivity indicated that the as-pre-
pared composites behaved as a semimetal in a temperature range from 2 to 300 K.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction content of single-walled carbon nanotubes (SWCNTs) up to

The successful fabrication of graphene in 2004 has led to the
demonstration of plenty of fascinating physical properties in
this monolayered 2 dimensional (2D) material. Electrons were
found to move ballistically in the graphene lattice with a
mobility exceeding 200,000 cm2 V1 s1 [1]. Graphene nano-
ribbons were theoretically predicted and experimentally con-
firmed to exhibit semiconductivity if they had narrow widths
and smooth edges [2,3]. Additionally, graphene nanosheets
(GNSs) composed of one or more layers of graphene are an
alternative choice as conductive fillers comparable to carbon
nanotubes (CNTs) in polymers [4,5]. Although conductive
ceramics are always desirable, there are no reports on the
use of GNSs in electrically functional ceramics to the best of
our knowledge. Generally, research on composites incorpo-
rated with carbon-based fillers focused on CNTs. For instance,
Rul et al. [6] synthesized CNT/MgAl2O4 composites with the
24.48 vol% by hot-pressing. After Rul et al.’s work, the electri-
cal conductivities of a variety of other inorganic systems con-
taining multi-walled carbon nanotubes (MWCNTs) have been
reported, such as SiC [7], TiN [8], Fe2N [8], borosilicate glass [9],
SiO2 [10], ZrO2 [11] and Si3N4 [12] systems. More recently, we
have further fabricated bulk MWCNTs by spark plasma sinter-
ing (SPS) and investigated their interesting properties [13].
One of the main obstacles in processing CNT-reinforced
composites lies in the difficulty of dispersing nanotubes
homogeneously throughout the matrix since agglomerate
would deteriorate the electrical properties of composite se-
verely. CNTs have a tendency to agglomerate due to their un-
ique geometric structure and typically weak interactions with
solvents or matrix components. Moreover, CNTs always tend
to twist each other forming ‘‘ropes’’ and ‘‘bundles’’ in a ma-
trix. Graphene or GNSs, compared to CNTs, are 2D materials.
Albeit they have high specific surface area even larger than

* Corresponding authors: Address: School of Materials Science and Engineering, Shanghai University, 149 Yanchang Road, Shanghai
200044, China (J. Li), State Key Laboratory of High Performance Ceramics and Superfine Microstructures, Shanghai Institute of Ceramics,
Chinese Academy of Sciences, Shanghai 200050, China (W. Jiang). Fax: +86 21 52413122.
E-mail addresses: jlli@shu.edu.cn (J. Li), wanjiang@mail.sic.ac.cn (W. Jiang).
0008-6223/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2010.01.017
1744 CARBON 4 8 ( 2 0 1 0 ) 1 7 4 3 –1 7 4 9

that of CNTs, nano-particles of the matrix have a good chance
to attach to the large surfaces(lateral dimension above 1 lm
in average) and thus to prevent the agglomerates of graphene
and GNSs.
In terms of preparing graphene, the main problem is how
to obtain a large amount of GNSs as thin as possible (for in-
stance, less than 10 layers). Typically there are three methods
to prepare graphene sheets: micromechanical exfoliation,
epitaxial growth of graphene films and chemical processing
including oxidation, exfoliation and reduction. Among them,
the last one seems to be the most promising as it has the po-
tential to enable massive production of GNSs. However, cer-
tain sorts of functional groups are difficult to remove by
reduction, leaving a significant number of defects which im-
pair the electronic properties of GNSs severely.
In this work, we used expanded graphite as a starting
material to fabricate GNSs. To ensure the obtained GNSs have
a good electronic structure, we employed ball milling rather
than re-intercalation by oxidation to further exfoliated ex-
panded graphite, and the as-prepared GNSs were dispersed
homogeneously in powder mixtures at the same time. Then
dense bulk composites were prepared from the powder mix-
tures by SPS. The electrical conductivities of as-prepared
composites are far higher than CNT/Al2O3 composites with
the same conductive phase content.
The morphology and microstructure of samples were char-
acterized by field emission scanning electron microscopy
(FESEM) (JEOL JSM-6700F), transmission electron microscopy
(TEM) and high-resolution transmission electron microscopy
(HRTEM) (JEOL 200CX).
3. Results and discussion
3.1. Microstructure of the as-prepared samples
The SEM image of expanded graphite presents a loose struc-
ture containing multi-pores (Fig. 1a). Magnified image shows
the thickness of an individual flake is about 100 nm (Fig. 1b),
which indicates expanded graphite is a good start for fabricat-
ing GNSs compared with natural flake graphite.
The SEM observations of the composites and monolithic
ceramic reveal no obvious defects in the specimens and the
micrographs were featureless, as shown in Fig. 2, confirming
their fully dense structure. To evaluate the distribution of
GNSs in the ceramic matrix, fracture surface of the composite
containing 1 vol.% GNSs was investigated (Fig. 2b). By compar-
ing with the pure Al2O3 ceramic (Fig. 2a), the growth of the
grain are found to be restrained remarkably with the presence
of GNSs. Moreover, it is shown that the grain sizes of Al2O3 are

2. Experimental details

In the current work, we started by exfoliating commercial

expandable graphite (160–50 N, Grafguard, USA) by heating
to 1000 °C in 60 s in nitrogen atmosphere. Then the expanded
graphite was ground with a-Al2O3 (100 nm, Taimei Chemical
Co. Ltd., Japan) using a planetary mill for 30 h. Si3N4 balls
and nylon vials were used and the N-methyl-pyrrolidone
(NMP) was chosen as the dispersal media. After ball milling
the composite powders were dried by the rotary evaporator.
Finally the residual solvent was removed in forming gas at
600 °C.
Bulk composites were prepared using SPS apparatus (Dr.
Sinter 2040, Sumitomo Coal Co. Ltd., Japan). Samples of the
GNS/Al2O3 powder were loaded into a 10 or 20 mm inner
diameter graphite die. A sheet of graphitic paper was placed
between the punch and the powder as well as between the
die and the powder for easy removal. The powders were sin-
tered in a vacuum (residual cell pressure <6 Pa). The soaking
time was 3 min and the heating rate was 140 K/min. An uni-
axial pressure of 60 MPa was applied from 1000 °C upwards
and maintained during the dwell at 1300 °C.
Density measurements were conducted using Archime-
des’s method. Two methods were chosen for conductivity
testing because of the large differences in conductivity be-
tween samples. For the samples with GNS content of 0, 0.8
and 2 vol.%, high resistance meter (HP 4329A) were used to
measure the conductivity on a cylinder with a diameter of
20 mm and a thickness of 3 mm; for the samples with GNS
content of 3, 3.5, 4, 5, 10 and 15 vol.%, a four probes method
(Quantum Design, PPMS, Model 6000) was used to measure Fig. 1 – Low (a) and high (b) magnification images of
the conductivity on a cubic bar with a length of 8 mm and a expanded graphite; (b) the thickness of single graphite flake
cross-section of 2 by 1 mm. is around 100 nm.
 CARBON 4 8 ( 20 1 0 ) 1 7 4 3–17 4 9
Fig. 2 – SEM images of fractured surfaces of (a) a pure Al2O3 ceramic and (b–d) a GNS/Al2O3 composite containing 1 vol.% GNSs;
(c) and (d) are magnified parts of (b).
homogeneous, which verify the good dispersion of GNSs in
the matrix (Fig. 2c). Compared with the pure Al2O3 ceramic,
in GNS/Al2O3 composite the phenomena of transgranular
fractures increase, suggesting the high strength of GNSs.
The SEM images show no thin GNSs (below 20 nm) but a
few pulled-out sheets dispersed in the matrix(marked by yel-
low rings in Fig. 2b), but most of these pulled-out sheets are as
thick as 50 nm (Fig. 2d). This phenomenon can be explained
as follows: when the sheets are thin (less than 20 nm) and
vertical to the fracture surface, they are easy to rupture under
the shearing stress. Besides, when the thin sheets are located
on the fracture surface, they cannot be identified in a SEM im-
age due to the small thickness of GNSs. On the contrary, when
we get thick GNSs in the grain boundary, they are strong en-
ough to bear the shearing stress and expose themselves when
fractures occur.
Further characterization of the microstructure of the com-
posites was carried out by TEM. Fig. 3 are representative TEM
and HRTEM images of a composite containing 5 vol.% GNSs.
The image reveals that the GNSs semi-wrap around the
Al2O3 grains and connect with each other to form a network
structure (Fig. 3a). It is observed that more than half of the
GNSs are below 10 nm in thickness and sheets with the thick-
ness of more than 20 nm have rarely been observed. Some of
the GNSs are as thin as 2.5 nm, which means it comprises of
only seven layers of graphene, as shown in Fig. 3c. It can also
1745
be seen that only very small fraction of overlap exist in the
Al2O3 grain boundary (Fig. 3d), indicating the effectiveness
and success of our strategy in preparing and dispersing of
GNSs.
There are two key factors of preparing and dispersing
GNSs well in our method. The first one is the usage of plane-
tary ball mill. The possibility of preparing tailored carbon
nano-materials by mechanical deformation has been widely
investigated before [14–16], and the mostly used experimental
method is mechanical milling in a ball milling device. The two
most common types of ball milling equipments, planetary
and vibratory mills, have been both experimentally tested.
In this study we exploit planetary ball mill as the main imple-
ment to exfoliate graphite into GNSs because the planetary
mill favors the cleavage of the particles [17]. One of the effec-
tive methods for preparing graphene is called laborious draw-
ing or micromechanical exfoliation [18]. In a sense the
planetary ball mill method is another type of mechanical
exfoliation because the shear component of the applied stress
is dominant [16], as illustrated in Fig. 4.
The other important factor is the selection of dispersal
media. It is well-known that the presence of a liquid in the
milling process leads to lubrication, moderates the violent
shocks and avoids large agglomerates of the particles. For in-
stance, Inam et al. have reported dimethylformamide as an
effective dispersant for making CNT/ceramic composites
1746 CARBON 4 8 ( 2 0 1 0 ) 1 7 4 3 –1 7 4 9

Fig. 3 – TEM and HRTEM images of a GNS/Al2O3 composite containing 5 vol.% GNSs: (a) GNSs surrounding Al2O3 nano-
particles; (b) a magnified image of GNSs with a thickness of about 10 nm in (a); (c) GNSs with a thickness of 2.5 nm; (d) overlap
of GNSs between Al2O3 nano-particles.

[19]. Furthermore, it has been reported recently that graphite

could be exfoliated in certain solvents to achieve defect-free
monolayer graphene [20]. This phenomenon relies on using
particular solvents, such as NMP, whose surface energy is so
well matched to that of graphene that exfoliation occurs eas-
ily. Hence we chose NMP as the dispersal media and we as-
sumed that the nano-Al2O3 particles could prohibit the
agglomerates of exfoliated GNSs to a large extent.
In order to obtain fully dense composites and retain nano-
crystalline Al2O3 grain size without damaging the GNSs dur-
ing sintering, SPS was employed. The relative density of
SPSed samples is all above 97%.

3.2. Electrical properties of the as-prepared samples

It has been the most common way to increase the conductiv-

Fig. 4 – Schematic of exfoliation of GNSs from graphite ity of an insulating material by the addition of a conductive
pallets; the arrows indicate the shear effect in a ball milling material. Especially, as the increasing concentration of con-
process. ducting component is above a critical value, which is also
 CARBON 4 8 ( 20 1 0 ) 1 7 4 3–17 4 9
called the percolation threshold, the dc conductivity of the
composites exhibits a dramatic rise of eight to ten orders in
magnitude. The experimental results have shown that pure
Al2O3 is an insulator with extremely low electrical conductiv-
ity (1013 S/m), but the electrical conductivity increased
sharply as the content of GNSs was close to percolation
threshold around 3 vol.% (Fig. 5).
This behavior can be described by the percolation theory.
According to the theory [21–23], as the conducting loading is
above the percolation threshold, the dc conductivity of the
composites follows the power law as described by
rdc ¼ rc ðPGNSs  Pc Þt ; for PGNSs > Pc ð1Þ
where rdc and rc are the dc conductivities of the composite
and conducting component, respectively. PGNSs is the volume
fraction of the GNSs, Pc is the critical volume fraction or perco-
lation threshold, and the exponent t reflects the dimensional-
ity of the system and has been calculated to be 1.3 and 2.0
corresponding to two and three dimensions, respectively [24].
The fitting curve to Eq. (1) is shown in Fig. 5, and we got
Pc = 3 ± 0.405 vol.% and t = 1.54 ± 0.22. The percolation thresh-
old value confirms the experimental result and is much lower
than that of the graphene/silica composite film fabricated by
spin-coating graphene on glass and SiOx/silicon substrates
[25], while higher than the value of CNT-reinforced compos-
ites reported by Rul et al. [6] (0.64 vol.%, SWCNT/MgAl2O4
composites) and Ahmad et al. [26] (0.79 vol.%, MWCNT/
Al2O3 composites).
The low value of t = 1.54 indicates that charge seems not to
transport through a random three dimensional network of
GNSs. Previous works [6,26,27] show that a low value of t in
a bulk material is inductive to thermally induced hopping
transport between disconnecting or weakly connecting parts
of the network, also known as fluctuation induced tunneling
mechanism, which is accompanied by a relatively high value
of t (2) at low temperatures [27,28]. However, in our research,
the fitting values of t were found to be 1.38 and 1.44 at 2 and
Fig. 5 – Electrical conductivities of GNS/Al2O3 composites as a function of filler volume fraction. Inset: log rdc plotted against
log (PGNSs  Pc), where Pc is percolation threshold (see text).
1747
50 K, respectively, which suggests that the conductivity is not
mainly controlled by fluctuation induced tunneling mecha-
nism in the GNS/Al2O3 composite.
We presume that the low value of t arises from the unique
structure of the GNS network. The paper-like GNSs unques-
tionably form a three dimensional network, the shape of
GNS, however, is far from spherical. Furthermore, during
SPS, the applied pressure leads to some preferable in-plane
arrangements of GNSs. As a result of such a combination of
two dimensional and three dimensional configurations, the
composite does not present as a random three dimensional
system. The t = 1.54 reveals the complicate GNS network of
a fractal dimensions.
As shown Fig. 5, the samples’ conductivities increase with
the growth of volume fraction of GNSs. The increase in the
GNS content results in the increase of internanosheet con-
nections, and consequently numerous conductive paths be-
come available. The conductivity increases due to
availability of a large number of charge carriers traveling
through the entire network. It is worthy to mention that
the electrical conductivity still increases quickly when GNS
content is above percolation threshold, the highest value
we obtained is 5709 S/m in 15 vol.% GNSs/Al2O3 composite,
which is 170% higher than the best result of SWCNT/Al2O3
composite with the same conductive phase content [29]. This
phenomenon is very important and different from most
CNT-reinforced ceramic composites, in which electrical con-
ductivity levels off when the conductive phase content is
above percolation threshold [10,11,26], mainly because of
intrinsic property of CNTs. Firstly, the CNTs tend to agglom-
erate with the content increasing in the matrix. When the
content of the CNTs exceeds the percolation threshold, most
of the added CNTs form ‘‘bundles’’ and ‘‘ropes’’ which have
little contribution to the electrical conductivity. Besides, gen-
erally the connection between CNTs is of a point–point touch
form that leads to a high resistance. In contrast with CNTs,
GNSs are connected by an area–area touch form, which re-
1748 CARBON 4 8 ( 2 0 1 0 ) 1 7 4 3 –1 7 4 9
Fig. 6 – Temperature dependence of dc conductivity for 3.5
and 4 vol.% of GNS/Al2O3 composites.
sults in a lower resistance and conduces to increasing the
possibility for GNSs to contact each other in a randomly
mixed system.
The temperature dependent dc electrical conductivities of
the composites have been investigated on samples incorpo-
rated with 3.5 and 4 vol.% GNSs in the temperature range
from 2 to 300 K. The variation of conductivities as a function
of temperature is plotted in Fig. 6. The conductivity of the
composites also increases linearly with temperature which
indicates that the system is semimetallic. Obviously, the
change in electrical conductivity is unlikely to stem from
structural change of the composites due to structural stability
of the as-prepared samples. Thus the increase of conductivity
with temperature mainly arises from the increase in carrier
concentration, just like a graphite, indicating the good elec-
tronic structure of GNSs produced in this work.
4. Conclusions
Fully dense GNS/Al2O3 composites have been fabricated from
ball milled expanded graphite and Al2O3 by SPS. The GNSs
after ball milling are 2.5–20 nm in thickness and homoge-
neously dispersed in the ceramic matrix. The percolation
threshold of the as-prepared GNS/Al2O3 composites is around
3 vol.%. The conductivity achieves 5709S/m when composite
has 15 vol.% GNSs, which was 170% higher compared to the
best result previously reported in CNT/Al2O3 composites.
The as-prepared composites behaved as semimetal as indi-
cated by the temperature dependence of electrical conductiv-
ity in a temperature range from 2 to 300 K. The present work
paves a new way to fabricate GNS/ceramic composites with
much improved electrical properties.
Acknowledgements
This work was funded by Natural Science Foundation of Chi-
na (Nos. 50625414, 50821004), Program of Shanghai Subject
Chief Scientist (No. 08XD14046), Shanghai Rising-Star Pro-
gram (No. 09QA1406600) and Shanghai Nano Science and
Technology Special Project (No. 0852nm03300).
R E F E R E N C E S
[1] Bolotin KI, Sikes KJ, Jiang Z, Klima M, Fudenberg G, Hone J,
et al. Ultrahigh electron mobility in suspended graphene.
Solid State Commun 2008;146(9–10):351–5.
[2] Yan QM, Huang B, Yu J, Zheng FW, Zang J, Wu J, et al. Intrinsic
current–voltage characteristics of graphene nanoribbon
transistors and effect of edge doping. Nano Lett
2007;7(6):1469–73.
[3] Li XL, Wang XR, Zhang L, Lee SW, Dai HJ. Chemically derived,
ultrasmooth graphene nanoribbon semiconductors. Science
2008;319(5867):1229–32.
[4] Stankovich S, Dikin DA, Dommett GHB, Kohlhaas KM, Zimney
EJ, Stach EA, et al. Graphene-based composite materials.
Nature 2006;442(7100):282–6.
[5] Wei T, Luo GL, Fan ZJ, Zheng C, Yan J, Yao CZ, et al.
Preparation of graphene nanosheet/polymer composites
using in situ reduction-extractive dispersion. Carbon
2009;47(9):2296–9.
[6] Rul S, Lefevre-schlick F, Capria E, Laurent C, Peigney A.
Percolation of single-walled carbon nanotubes in ceramic
matrix nanocomposites. Acta Mater 2004;52(4):1061–7.
[7] Morisada Y, Miyamoto Y, Takaura Y, Hirota K, Tamari N.
Mechanical properties of SiC composites incorporating SiC-
coated multi-walled carbon nanotubes. Int J Refract Met Hard
Mater 2007;25(4):322–7.
[8] Jiang LQ, Gao L. Carbon nanotubes-metal nitride composites:
a new class of nanocomposites with enhanced electrical
properties. J Mater Chem 2005;15(2):260–6.
[9] Thomas BJC, Shaffer MSP, Boccaccini AR. Sol–gel route to
carbon nanotube borosilicate glass composites. Compos Pt A
– Appl Sci Manuf 2009;40(6–7):837–45.
[10] Guo SQ, Sivakumar R, Kitazawa H, Kagawa Y. Electrical
properties of silica-based nanocomposites with multiwall
carbon nanotubes. J Am Ceram Soc 2007;90(5):1667–70.
[11] Shi SL, Liang J. Electronic transport properties of multiwall
carbon nanotubes/yttria-stabilized zirconia composites. J
Appl Phys 2007;101(2):023708.
[12] Tatami J, Katashima T, Komeya K, Meguro T, Wakihara T.
Electrically conductive CNT-dispersed silicon nitride
ceramics. J Am Ceram Soc 2005;88(10):2889–93.
[13] Li JL, Wang LJ, He T, Jiang W. Transport properties of hot-
pressed bulk carbon nanotubes compacted by spark plasma
sintering. Carbon 2009;47(4):1135–40.
[14] Li JL, Peng QS, Bai GZ, Jiang W. Carbon scrolls produced by
high energy ball milling of graphite. Carbon
2005;43(13):2830–3.
[15] Li JL, Wang LJ, Bai GZ, Jiang W. Microstructure and
mechanical properties of in situ produced TiC/C
nanocomposite by spark plasma sintering. Scr Mater
2005;52(9):867–71.
[16] Antisari MV, Montone A, Jovic N, Piscopiello E, Alvani C,
Pilloni L. Low energy pure shear milling: a method for the
preparation of graphite nano-sheets. Scr Mater
2006;55(11):1047–50.
[17] Janot R, Guerard D. Ball-milling: the behavior of graphite as a
function of the dispersal media. Carbon 2002;40(15):2887–96.
[18] Novoselov KS, Jiang D, Schedin F, Booth TJ, Khotkevich VV,
Morozov SV, et al. Two-dimensional atomic crystals. Proc
Natl Acad Sci USA 2005;102(30):10451–3.
[19] Inam F, Yan H, Reece MJ, Peijs T. Dimethylformamide: an
effective dispersant for making ceramic–carbon nanotube
composites. Nanotechnology 2008;19(19):195710.
[20] Hernandez Y, Nicolosi V, Lotya M, Blighe FM, Sun ZY, De S,
et al. High-yield production of graphene by liquid-phase
exfoliation of graphite. Nat Nanotechnol 2008;3(9):563–8.
 CARBON 4 8 ( 20 1 0 ) 1 7 4 3–17 4 9
[21] Scher H, Zallen R. Critical density in percolation processes. J
Chem Phys 1970;53(9):3759–61.
[22] Kirkpatr S. Percolation and conduction. Rev Mod Phys
1973;45(4):574–88.
[23] Stauffer D. Scaling theory of percolation clusters. Phys Rep-
Rev Sec Phys Lett 1979;54(1):1–74.
[24] Stauffer D, Aharony A. Introduction to percolation theory.
2nd ed. London: Tayler and Francis; 1994.
[25] Watcharotone S, Dikin DA, Stankovich S, Piner R, Jung I,
Dommett GHB, et al. Graphene–silica composite thin films as
transparent conductors. Nano Lett 2007;7(7):1888–92.
[26] Ahmad K, Pan W, Shi SL. Electrical conductivity and dielectric
properties of multiwalled carbon nanotube and alumina
composites. Appl Phys Lett 2006;89(13):133122.
1749
[27] Kilbride BE, Coleman JN, Fraysse J, Fournet P, Cadek M, Drury
A, et al. Experimental observation of scaling laws for
alternating current and direct current conductivity in
polymer–carbon nanotube composite thin films. J Appl Phys
2002;92(7):4024–30.
[28] Sheng P. Fluctuation-induced tunneling conduction in
disordered materials. Phys Rev B 1980;21(6):2180–95.
[29] Zhan GD, Kuntz JD, Garay JE, Mukherjee AK. Electrical
properties of nanoceramics reinforced with ropes of single-
walled carbon nanotubes. Appl Phys Lett 2003;83(6):1228–30.