Strengthening behavior of few-layered graphene/ aluminum composites

C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1

.sc ienced i rec t .com

ScienceDirect

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

Strengthening behavior of few-layered graphene/aluminum composites

http://dx.doi.org/10.1016/j.carbon.2014.10.0440008-6223/� 2014 Elsevier Ltd. All rights reserved.

* Corresponding author.E-mail address: [email protected] (D.H. Bae).

S.E. Shin a, H.J. Choi b, J.H. Shin a, D.H. Bae a,*

a Department of Materials Science and Engineering, Yonsei University, Seoul 120-749, Republic of Koreab School of Advanced Materials Engineering, Kookmin University, Seoul 136-702, Republic of Korea

A R T I C L E I N F O

Article history:

Received 26 February 2014

Accepted 17 October 2014

Available online 25 October 2014

A B S T R A C T

Strengthening behavior of composite containing discontinuous reinforcement is strongly

related with load transfer at the reinforcement–matrix interface. We selected multi-walled

carbon nanotube (MWCNT) and few-layer graphene (FLG) as a reinforcing agent. By varying

a volume fraction of the reinforcement, aluminum (Al) matrix composites were produced

by a powder metallurgy method. Uniform dispersion and uniaxial alignment of MWCNT

and FLG in the Al matrix are evidenced by high-resolution transmission electron micro-

scope analysis. Although the reinforcements have a similar molecular structure, FLG has

a 12.8 times larger specific surface area per volume more than MWCNT due to geometric

difference. Therefore an increment of a yield stress versus a reinforcement volume fraction

for FLG shows 3.5 times higher than that of MWCNT Consequently, for both reinforce-

ments, the composite strength proportionally increases with the specific surface area on

the composite, and the composites containing 0.7 vol% FLG exhibit 440 MPa of tensile

strength.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Graphene has attracted interest as a reinforcing agent for

metal matrix composites due to excellent mechanical proper-

ties based on the strong sp2 CAC bonds, which are similar to

fullerene and carbon nanotube [1,2]. Furthermore, it has mer-

its over other carbon-based nano materials, which originate

from its inherent two-dimensional (2-D) morphology; the pla-

nar structure is more favorable to load transfer as well as to

impeding atomic diffusion at high temperatures, as

compared to its 0-D and 1-D counterparts. Consequently, it

provides superior strength for composites at both room

temperature and high temperatures.

In order to transmit the excellent properties of graphene to

composites, uniform dispersion of an individually-exfoliated

graphene is one key factor. Several processes have been intro-

duced to exfoliate graphite to single-layer or few-layer graph-

ene by mechanical and/or chemical means [3–5]. Mechanical

exfoliation using a tape dispenser [2] or atomic force micros-

copy (AFM) [6] has exhibited inadequate productivity for

large-scale industrial applications. Although large-scale syn-

thesis of graphene by gas phase techniques (e.g., thermal

chemical vapor deposition) [7–9] is actively ongoing, this pro-

cess is still costly and has restrictions in terms of the selec-

tion of the substrate materials. Chemical exfoliation by a

solution process has been suggested as a relatively cheap pro-

cess [10–12], and yet presents difficulties for scalable synthe-

sis due to complex synthesis steps and the requirement for a

large amount of chemicals and acid. Recently, solid phase

techniques, combined with ball-milling processes, have

144 C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1

enabled production of scalable quantities of carbon-base

nano-materials by applying shear forces on pristine agglom-

erated particles [13–15].

Even though one has developed a scalable process to pro-

duce few-layer graphene (FLG), an additional technical hurdle

is in the uniform dispersion of the graphene in the metal

matrix, which has closely packed atomic structures. Hence,

graphene/metal composites have been seldom investigated

compared to polymer matrix composites, although the great

strength and light weight features of graphene/metal com-

posites are expected to lead to applications of such compos-

ites in the automotive and aerospace industries.

Aluminum (Al) matrix composites reinforced with FLG

have recently been produced using different synthesis tech-

niques with powder metallurgy (PM) routes. Table 1 summa-

rizes the fabrication processes and resulting properties of

recently developed Al/FLG composites [16–21]. Similar to

other nanostructured composites, dispersion of nanoscale

reinforcement is a critical issue in Al/FLG composites. Poorly

dispersed FLG may act as defect sites, significantly deteriorat-

ing the performance of the final composites [16,17]. Consoli-

dation processes are also important for Al/FLG composites

because a selection of high-processing temperatures to avoid

insufficient powder consolidation leads to unfavorable reac-

tions (e.g., transformation of FLG to carbides) [17,18–20]. Even

though some composites have been produced via cost-inef-

fective routes or using expensive materials (e.g., graphene

oxide), they did not exhibit tensile elongation or they exhib-

ited very limited tensile strength (<300 MPa). Furthermore,

the microstructure of Al/FLG composites with an atomic-scale

resolution has not been well reported; atomic-scale resolution

images may provide detailed information on the morphology

of FLG and the interface between Al and FLG [17,20,21]. Rein-

forcement in composites may enhance the strength of

the matrix by (i) carrying a great amount of load instead

of the matrix and by (ii) interrupting the plastic deformation

of the matrix. Hence, two important factors to determine

the strengthening efficiency of reinforcement, other than its

intrinsic mechanical properties and volume fraction, are (i)

how much load can be effectively transmitted from the

matrix to the reinforcement and (ii) how much the stress dis-

tribution can be altered by the reinforcement. Both are

Table 1 – Experimental tensile strength and hardness of FLG-re

Graphene content Fabrication techniques

0.1 wt% Ball-millingHot isostatic pressing: 550 �C, 4 hHot extrusion: 550 �C, 4:1 ratio

0.3 wt% Ball-millingSintering: 580 �C, 2 hHot extrusion: 440 �C, 20:1 ratio

0.3 wt% Ball-millingSintering: 600 �C, 6 hHot extrusion: 470 �C, 2:1 ratio

0.3 wt% Ball-millingSintering: 500 �C, 5 h

5 wt.% Ball-millingSintering: 600 �C, 5 h

affected by bonding strength between the matrix and rein-

forcement, morphology/surface area of the reinforcement,

and spatial distribution of the reinforcement [22,23]. Although

experimental work on the effect of volume fraction of the

reinforcement on the strength of composites [24,25] and the-

oretical work on the bonding strength between metal and car-

bon-based nano materials [26–29] has been conducted, the

role of morphology and surface area on the reinforcement

has not been extensively investigated. In particular, a com-

parison study on the strengthening behavior of reinforce-

ments with a variety of shapes (e.g., fullerene–sphere/0-D,

carbon nanotube–tube/1-D, and graphene–planar/2-D), but

with similar intrinsic mechanical properties, would be highly

interesting.

The primary objective of this work is to produce Al matrix

composites with well-dispersed FLG using a scalable powder

metallurgy approach. Here, we introduce a favorable route

to produce centimeter-scale Al/FLG composites with excellent

performances. This work also aims to examine tensile proper-

ties of the Al/FLG composite as a function of volume fraction

of FLG, so as to compare the strengthening behavior of FLG

with that of multi-walled carbon nanotube (MWCNT). The

present study also carefully investigates the microstructure

of the composites at an atomic scale to provide detailed infor-

mation on the structure of nano-carbon materials and nano-

carbon/Al interface structure. Further, this work quantifies

the effect of shape factor (i.e., tube versus sheet) and specific

surface area of reinforcement on the composite strength.

2. Experimental

2.1. Sample preparation

Aluminum (Al)-based composites containing FLG were fabri-

cated by hot-rolling of ball-milled powder. At first, graphite

flakes (6–8 nm thickness and 120–150 m2/g typical specific

surface area) were mechanically exfoliated by planetary ball

milling in the presence of isopropyl alcohol ((CH3)2CHOH). A

stainless steel bowl (500 mL) was charged with graphite flakes

(2 g) and stainless steel balls (�5 mm diameter, 30 g) at a ball-

to-powder weight ratio of 15:1, together with 50 mL of isopro-

pyl alcohol. Planetary milling was performed at a rotation

inforced composites.

Mechanical properties Research group

Tensile strength:262 MPa

Bartolucci et al. [16]

Tensile strength:250 MPa

Wang et al. [17]

Vickers hardness:85 HvTensile strength:280 MPa

Rashad et al. [18]

Vickers micro-hardness:90 lHv

Perez et al. [19]

Vickers hardness: 70 HvCompressive strength: 180 MPa

Latief et al. [20]

C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1 145

speed of 200 RPM for 1 h; it was paused for 75 min after every

15-min milling to maintain ambient processing temperature

without any process control agent. Afterwards, the isopropyl

alcohol was evaporated and dried at 150 �C for 3 h. Graphite

flakes are supposed to be exfoliated by shear forces on con-

tact between powder and balls during milling. Due to the

weak van der Waals-like coupling between graphite layers,

the graphene sheets in graphite can slide easily with respect

to one another. To address this, additional milling was con-

ducted for further exfoliation of graphite nanosheets into

FLG and for dispersion of the FLG in Al powder. The exfoliated

graphite flakes were exfoliated further into FLG and were also

mixed with Al powder (99.5% purity and <150 lm diameter)

using a planetary mill at a rotation speed of 100 RPM for 3 h,

with a process control agent of 1 wt.% stearic acid (CH3

(CH2)16CO2H). Milling was paused for 15 min after every

15 min of milling. Finally, for further dispersion of FLG in Al

powder, the planetary-milled mixture was high energy ball-

milled in an attritor at 500 RPM for 6 h in a purified argon

atmosphere. FLG, attached on the Al powder surface, is grad-

ually embedded and is dispersed inside Al powder.

A variety of ball-milled powders were fabricated by varying

the volume fraction of FLG (i.e. 0.3, 0.5 and 0.7 vol%). The ball-

milled powder was containerized in a copper tube (60 mm in

diameter, 150 mm in height, and 1.5 mm in thickness), com-

pacted, and was then hot-rolled. The sample was heated to

a pre-determined temperature of 500 �C at a heating rate of

15 �C/min, and rolling was conducted with a 12% reduction

per pass; the final thickness of samples was 1 mm. After roll-

ing, the copper container was mechanically removed. Since

FLG was embedded inside the Al powder, it did not signifi-

cantly interrupt consolidation of Al powder during hot rolling,

providing a fully dense composite sheet.

Fig. 1 – SEM images of (a) FLG attached to Al powder using a pla

embedded and dispersed in Al powder using an attrition mill at

(c), respectively.

2.2. Characterization

Morphology of the ball-milled powder varied according to

milling condition, as was observed using scanning electron

microscopy (SEM). Microstructure of the Al/FLG composite

was observed using a high resolution transmission electron

microscope (HRTEM, JEOL 2000). Thin foil specimens from

the sheets were carefully prepared by an ion beam 6 milling

method (Gatan, Model 600, Oxford, UK). The structure of

FLG was also investigated by Raman spectroscopy using a

Jobin–Yvon microspectrometer (LabRam HR, Jobin–Yvon Co.

Ltd., France). Spectra were collected under ambient condi-

tions using the 532-nm line of an argon-ion laser. Uniaxial

tension tests were conducted for the annealed specimens

under the constant cross-head speed condition of an initial

strain rate of 1 · 10�4 s�1. The apparent yield properties were

determined using the 0.2% offset method. Tensile specimens

(thickness: 1 mm, gauge length: 12.5 mm, gauge width: 6 mm,

and grip length: 25 mm for both upper and lower) were pre-

pared from the hot-rolled Al composite sheet according to

the standard ASTM B-209 and loaded in the rolling direction.

The tests were repeated five times, and each test was

conducted from different samples that were processed in a

different batch using the same route.

3. Results and discussion

3.1. Microstructures

Morphologies of ball-milled Al/FLG composite powders are

shown in Fig. 1. Exfoliated FLG sheets with sizes below

10 lm are observed on the surface of the Al powder, where

FLG is marked by an arrow in Fig. 1a. Fig. 1b displays magnified

netary mill at 100 RPM (FLG is marked by an arrow). (c) FLG

500 RPM. (b) and (d) display the magnified images of (a) and

1400 1600 1800 2000 2200 2400 2600 2800

Exfoliated graphite powder

Al/0.7 vol.% FLG

Al/0.5 vol.% FLG

Al/0.3 vol.% FLG

Initial graphite powder

D band 2D band

Inte

nsity

(a.u

.)

Raman shift (cm-1)

532 nmG band

Fig. 2 – Raman spectrum of initial graphite powder,

exfoliated graphite powder and hot-rolled Al/FLG

composites. (A colour version of this figure can be viewed

online.)

146 C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1

images of rectangles in Fig. 1a, where FLG is thin enough to

transmit the electron beam so the Al powder underneath the

FLG can be seen. Fig. 1c exhibits the powder morphology after

attrition milling. FLG is not observed on the powder surface

even in the magnified images of Fig. 1d. Hence, the FLG is sup-

posed to be embedded and dispersed inside the Al powder.

Fig. 2 provides Raman spectra of the initial graphite pow-

der, exfoliated graphite powder and Al/FLG composites with

various volume fractions of FLG. Typical D-band (from defect

and amorphous carbon), G-band (from graphite), and 2D band

Fig. 3 – TEM images of the graphenes in the hot-rolled Al/0.3 vol

plane (graphene is marked by white arrows). In (c) and (d), FLGs

(marked by white arrows), highly deformed regions (marked by

of this figure can be viewed online.)

(shape of the second-order Raman bands) of graphitic carbons

are detected at 1359, 1603, and 2727 cm�1 respectively [30,31],

for the initial graphite powder. These bands shift to lower val-

ues (red-shift) as exfoliation proceeds, presumably because

the surface of graphite powder might strongly interact with

isopropyl alcohol, and the functionalized surface may gener-

ate a certain amount of in-plane internal stresses and weaker

CAC bonds [32]. The peaks shift back toward their expected

position after attrition and planetary milling with Al powder.

As FLG is dispersed inside Al, it may not interact with other

material any longer, and hence the residual stresses could

be released. The ratio between the intensities of the D and

G bands, ID/IG, is considered to be the ratio of structural

defects and domain size in graphitic materials [33]. The ID/IGratio increases significantly after planetary milling with iso-

propyl alcohol and remains largely the same after attrition

milling and hot-rolling processes. Therefore, FLG is thought

to be damaged most during milling with isopropyl alcohol.

The down-shifted 2D band, which is related to the crystalline

graphitic structure, from 2727 to 2678 cm�1, arises from a

reduction of the number of graphite layers during the ball

milling processes [34]. For Al/FLG composites, the 2D band

can be fitted with a sharp and symmetric peak, while that

of graphite can be fitted with two peaks. It can be seen in

Fig. 2 that the 2D band becomes sharper and shifted when

the graphene thickness decreases [35].

TEM images in the RD–TD and the ND–RD planes of the

hot-rolled sheets are shown in Fig. 3; here, RD is the rolling

direction, TD is the transverse direction, and ND is the normal

direction. FLG and deformation bands are marked by a white

arrow and circles, respectively. The mechanically exfoliated

% composite observed on the (a) RD–TD plane and (b) ND–RD

(marked by red lines) are given. (e) Between the graphenes

circles) are observed after 6% deformation. (A colour version

0.00 0.05 0.10 0.15 0.20 0.250

100

200

300

400

500

True strain

True

stre

ss (M

Pa) Al/0.3 vol. % FLG

Al/0.7 vol. % FLG

Al/0.5 vol. % FLG

Pure Al (unmilled)

Pure Al (milled)

strain rate - 1x10-4s-1

0.00 0.01 0.02 0.03 0.04 0.05250

300

350

400

450

500

550

600

650

Experimental data (Al/FLG) Experimental data (Al/MWCNT [36]) Calculated strength

Reinforcement volume fraction

Yie

ld st

ress

(MPa

)

dσc /dVFLG= 26.5 GPa dσc /dVMWCNT= 7.5 GPa

(a)

(b)

Fig. 4 – (a) True stress-true strain curves for the hot-rolled Al/

FLG composites for various volume fractions of FLG. (b)

Variation of yield stress as a function of the volume fraction

of reinforcements (FLG and MWCNT [36]) using the exper-

imental and calculated data given in Table 2.

C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1 147

FLG, with sizes ranging from 50 to 200 nm, are observed to be

dispersed piece by piece in the Al matrix, as shown in Fig. 3a

and b. FLGs, aligned along the rolling direction (Fig. 3b) are

slightly wrinkled and folded with 100–200 nm in the long axis.

The folded part is clearly found on the edge of the graphene.

Folding of graphene commonly occurs during mechanical

exfoliation because intrasheet van der Waals attraction is siz-

able, and doubling over within a sheet sometimes provides an

energetic minimum. However, from the rolling direction in

the ND–RD plane, the FLG keep their typical two dimensional

structures well. The average thickness and layer number of

graphene in the composite are statistically measured to be

�5 nm and �5 layers, respectively. The average thickness of

the composite is �5 nm (standard deviation �2 nm), while

the number of layers of the graphene is �5 layers (standard

deviation �2 nm). Measurement of the average thickness is

based on hundreds of FLGs from more than 50 TEM images.

Fig. 3c and d show few-layered thick graphene and relatively

thin graphene, respectively.

As the composite is loaded, Al grains with an average size

of �200 nm [36] may experience plastic deformation via con-

ventional manners; macroscopic yielding occurs when dislo-

cation pile-ups within a deforming grain produce a stress

concentration sufficient to activate slip in the adjacent grains

that have an unfavorably oriented slip system. This process

involves the accommodation of a strain gradient, which pro-

vides compatible displacements between adjacent grains.

When FLG separates these two adjacent grains, however,

the plastic flow would be constrained by FLG. The extent of

shear localization in the matrix leads to a highly localized

deformation band. Owing to the high specific surface of

FLG, FLG may restrict the mass flow in the Al matrix, resulting

in a significant decrease of plastic strain-to-failure of the Al/

FLG composites. As shown in Fig. 3e, highly deformed regions

are found in the Al/0.3 vol% composite after 6% deformation.

That is, between FLGs, development of nanoscale shear bands

could accommodate the constrained shear strain.

3.2. Mechanical properties

To examine strengthening efficiency of FLG, tensile tests were

carried out for the hot-rolled Al/FLG composites having differ-

ent contents of FLG, as plotted in Fig. 4a. Yield strength of

monolithic Al is first enhanced by grain refinement; it

increases �2.6 times after grain refinement by ball-milling.

It is also significantly enhanced further with an increase in

volume fraction of FLG; specifically, a �71.8% enhancement

from 262 to 440 MPa was observed when only 0.7 vol% FLG

was added to the monolithic Al. On the other hand, a general

trend of ductility reduction with strength enhancement is

observed as the content of FLG is increased. The strain-to-fail-

ure of Al/FLG composites is lower by 3–5.5% when compared

with the milled pure Al (�13%). The results also indicate that

the strain field around the FLGs significantly alters the plastic

flow, inducing highly localized deformation behavior, as evi-

denced by the formation of the bands.

Fig. 4b compares strengthening efficiency of FLG (solid

squares, based on experimental data in the present study)

and MWCNT (open squares, obtained from [36]) in Al-based

composites. An increase in strength with volume fraction of

the reinforcements, calculated as drc/dVr, is closely related

to load transfer from the matrix to the reinforcement. We

may ignore other microstructural effects such as grain size

differences between two composites because they both are

produced by the same processing routes (hot-rolling at

500 �C subsequent to 6-h-ball-milling at 500 RPM). Interest-

ingly, the linear fit of the slope, drc/dVgraphene, is �25.7 GPa

while drc/dVCNT is �7.5 GPa. Despite the similar molecular

structure and mechanical properties, FLG is found to be

�3.5 times more effective in strengthening aluminum as

compared to MWCNT.

When a composite is loaded, the matrix is strained and

then the strained matrix may transfer the load to reinforce-

ment by means of shear stresses that develop along the rein-

forcement–matrix interface. The shear forces on the interface

parallel to the load direction are balanced with the normal

forces on the fiber cross-sections normal to the load direction.

Hence, the interfacial area (S) and the fiber cross-sectional

area (A) are key factors that control the amount shear stresses

generated at the interface, and these eventually determine

the load transfer efficiency of the reinforcement. Table 2

provides calculations of the interfacial and cross-sectional

areas per unit volume for MWCNT and FLG, and also

Table 2 – Schematic depiction of effective volume difference for various reinforcements.

CNT Graphene

Depiction

Critical length (lc, lm) 2 0.381S/1 ea pdfl 2(w + t)l

A/1 eapdf

4

2wt

Volume/1 ea (·10�22 m3) 9.620.75 0.75Number of reinforcement/Volume(·1019/m3)

1.01 13.3

Vol% Experimentaldata

Calculateddata

Vol% Experimentaldata

Calculateddata

Yield stress (rc, MPa) 1.0 352 347 0.3 360 3381.5 386 389 0.5 405 3893.0 483 5174.5 610 645 0.7 440 440

rcVr;GPa

� �7.5 8.5 26.5 24.2

*Values are based on statistical estimates for a given measurement of reinforcement using TEM images in Fig. 3.

148 C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1

compares theoretical and experimental yield strength of com-

posites with both reinforcements.

Our calculation begins with the aforementioned force bal-

ance between the interfacial shear stress and reinforcement

tensile stress for the case of composites containing MWCNT

as [37]:

smðpdf Þdx ¼ pdf

4

2 !

drf ð1Þ

where sm is the shear strength of the matrix (�0.5 rm, rm is the

tensile yield strength of the matrix), which is supposed to be

the shear stress that the matrix transmits to the MWCNT, x is

the axial distance from the tips of the MWCNT, df is the aver-

age diameter of MWCNT, and rf is the (position specific) ten-

sile stress of the MWCNT. This expression can be modified for

composites containing FLG as:

smf2ðwþ tÞgdx ¼ ðwtÞdrp ð2Þ

where w is the width of the FLG, t is the thickness of FLG, x is the

distance from the end of the FLG, and rp is the (position spe-

cific) tensile stress of FLG. As denoted in Table 2, the cross-

section areas (A) for MWCNT and FLG arepdf

4

2and wt, and the

interfacial areas (S) are pdfl and 2(w + t)l, respectively; substi-

tuting both Eqs. (1) and (2), we obtain the following equation:

smS

ldx ¼ A drf ð3Þ

This formula implies that the tensile stress of the rein-

forcement is maximum at its mid-point and zero at its ends.

When the maximum stress reaches the tensile strength of the

reinforcement, the reinforcement length is a critical length (lc)

of the reinforcement. If the reinforcement length exceeds the

critical length (l > lc), the reinforcement carries stress equal to

its tensile strength throughout a certain length (l � lc). The

critical length of the reinforcements, lc, can be calculated as:

lc ¼ rrAlsmS

ð4Þ

where rr is the yield strength of the reinforcements, namely,

30 GPa for MWCNT [38] and FLG. lc values calculated for

MWCNT and FLG are 1 and 0.381 lm, which are much larger

than the length of MWCNT and FLG with measurements from

statistically estimation from TEM images, respectively. Hence,

stress that the reinforcement carries may increase propor-

tional to the distance from the end of the reinforcement.

The theoretical composite strength can be expressed as:

rc ¼ rrVrl

2lc

��þ rmVm ð5Þ

where Vr and Vm are the volume fraction of the reinforce-

ments and the matrix, respectively. Applying this simplifica-

tion to the reinforcements, the expression proposed for

composite strength can be obtained from Eq. (5):

rc ¼ VrSA

� �sm

2

��þ rmVm ð6Þ

Consequently, the calculated data using Eq. (6), which is a

good correlation containing a quantitative estimate of the

reinforcements and contact area based on force balance,

exhibit good agreement with the experimental values as

given in Table 2 (marked by dotted lines in Fig. 4b). Stress dis-

tribution along the reinforcements occurs when the length of

0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014250

300

350

400

450

500

550

600

650

700

Reinforcement volume fraction

Present study

Short fiber composites [37]

Piggott [43]

Halpin-Tsai [42]Shear-lag [40]

Yie

ld st

ress

(MPa

)

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14

150

200

250

300

350

400

450

500 [17] [18] [19] [20] Present study

Reinforcement volume fraction

Yie

ld st

ress

(MPa

)

(a)

(b)

Fig. 5 – (a) Plots of composite strengths theoretically

predicted [37,40,42,43] and experimentally obtained in this

study. (b) Yield stresses of several Al-based composites

containing graphene with varying the volume fraction of

reinforcement (data taken from [17–20]) in which calculated

data using Eq. (6) are marked by dotted lines. (A colour

version of this figure can be viewed online.)

C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1 149

the reinforcements is less than some critical length. Calcula-

tions for the specific surface area-composite strength rela-

tionship based on statistical estimates in a given

measurement of reinforcements using TEM images are dis-

played in Table 2. The effective volume, which means equiv-

alent quantity of reinforcements per arbitrary region, takes

into consideration the length as well as quantity of the rein-

forcements. For MWCNT, the volume per reinforcement is

9.62 · 10�22 m3, while FLG shows a value of 0.75 · 10�22 m3.

Although the ability to carry the load of MWCNT is excellent

compared to FLG, nearly 12.8 times more FLG can occupy

the unit volume than MWCNT, with a surface area that is

2.6 times as large. In comparison with MWCNT, a FLG con-

taining such two dimensional sheets can provide substantial

contact with the matrix due to their upper and lower sur-

faces. Therefore, a sheet having an efficient interface with

sufficient stress transfer between the reinforcement and the

matrix is very powerful and effective. To achieve such an

increase in strength, the FLG contact area is estimated to be

on the order of 3.5 times that of MWCNT.

Table 3 summarizes commonly used models suggested for

tensile strength for MMCs, and Fig. 5a shows a comparison of

the theoretical predictions and experimental results as a

function of the FLG volume fraction. The short fiber and

shear-lag models describe the strengthening by general dis-

continuous fibers. The shear-lag model [39] describes

strengthening of reinforcement with a high-aspect ratio,

and the modified shear-lag model [40] is complemented by

considering the effect of fiber orientation. The Halpin–Tsai

equation and Piggott model are specific for CNTs or platelet-

reinforced composites, respectively [43]. The Halpin–Tsai

model [42] gives a semi-empirical description of short-fiber

reinforced composites using the rule of mixtures for discon-

tinuous reinforcement, and the modified Halpin–Tsai equa-

tion [41] is specified for MMCs. The Piggott model modifies

the discontinuous fiber model by considering two-dimen-

sional geometric characteristics for platelet reinforcements.

Experimental data and theoretically calculated values as a

function of graphene content are provided in Fig. 5a. Overall,

the experimental data are well-matched only with our model.

Since both FLG and CNT have high volume-to-surface area

ratios, the matrix/reinforcement interface plays a significant

role in strengthening. However, only our model takes into

account the features of matrix/reinforcement interface (e.g.,

Table 3 – Depiction of theoretical models for prediction ofcomposite strength.

Model Equation

Short-fiber [37] rc ¼ rf Vfl

2lc

� �þ rmð1� Vf Þ l < lc

Developed Shear-lag [40] rc ¼Vf rml

2d

Developed Halpin–Tsai [42] rc ¼1þngVf

1�gVfrm

*

Piggott [43] rc ¼ rm2Lt Vp

4 þ Vmrm**

* The value of n has to be optimized in order to take care of disper-

sion of the reinforcement (2ðl=dÞeð�40Vf�1:0Þ), and g depends on (rf/rm).** Vp and L are the volume fraction and the long axis of the platelet,

respectively.

effective interfacial area); the present study modifies the dis-

continuous fiber model with a new term (S/A), which enables

the equation to be matched with the experimental data.

Moreover, the model developed in the present study is

adapted to compared with experimental data that have been

reported in the literature, as shown in Fig. 5b [17–20]. Since

the size of graphene varies reported in the article, the calcu-

lated values vary as well. The size of graphene was measured

from the TEM or SEM images when information was not avail-

able. The present model shows good agreement with the

experimental data except for Latief et al.’s study [20]. Insuffi-

cient consolidation, evident by a low density of composite,

may cause this deviation in the experimental data from the

model. Overall, our calculation has a good fit for the strength

of nanoscale-reinforced MMCs. As mentioned previously, an

efficient interface with sufficient load transfer between the

reinforcement and the matrix is very influential in the deter-

mination of the strength in nanocomposites.

Fig. 6 shows variation in yield stress as a function of

surface area per composite volume, where strengthening

0 1 2 3 4 5 6250

300

350

400

450

500

550

600

650

700

FLG MWCNT

Surface area per unit volume (106 μ μm-1)

Yie

ld st

ress

(MPa

)

Fig. 6 – Variation of yield stress as a function of surface area

per unit volume with various reinforcements.

150 C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1

efficiency for two reinforcements is compared. The composite

strength proportionally increases with the surface area per

composite volume, and the increase is very similar for both

reinforcements.

4. Conclusion

Al matrix composite containing FLG was successfully pro-

duced through a novel fabrication approach that combines

mechanical milling and hot rolling. FLG was mechanically

exfoliated using wet ball milling, and was then uniformly dis-

persed in the Al matrix using high energy ball milling. The

strengths of the Al/FLG composites are significantly enhanced

with the volume fraction of FLG. With an addition of only

0.7 vol% FLG, the composite exhibits �440 MPa of tensile

strength, about two times higher than that of monolithic Al.

Furthermore, the composites produced via favorable indus-

trial routes using inexpensive graphite will increase a poten-

tial market opportunity of Al/FLG composites. On the other

hand, with respect to the yield stress of MWCNT-reinforced

Al matrix composites, FLG is a much more effective reinforce-

ment because it has a larger surface area per unit volume. Our

study highlighted that the specific surface area of the rein-

forcement can determine the strength of the composites.

Acknowledgements

This research was supported by the National Research Foun-

dation of Korea (NRF) funded by the Ministry of Education

(2013R1A2A2A01068931). H.J. Choi acknowledges the support

of the NRF funded by the Science, ICT & Future Planning

(MSIP) (2013R1A1A3005759 and 2013K1A4A3055679).

R E F E R E N C E S

[1] Park S, Ruoff RS. Chemical methods for the production ofgraphenes. Nat Nanotechnol 2009;4(4):217–24.

[2] Geim AK, Novoselov KS. The rise of graphene. Nat Mater2007;6(3):183–91.

[3] Martinez A, Fuse K, Yamashita S. Mechanical exfoliation ofgraphene for the passive mode-locking of fiber lasers. ApplPhys Lett 2011;99(12):121107.

[4] Hernandez Y, Nicolosi V, Lotya M, Blighe FM, Sun ZY, De S,et al. High-yield production of graphene by liquid-phaseexfoliation of graphite. Nat Nanotechnol 2008;3:563–8.

[5] Sidorov AN, Bansal T, Ouseph PJ, Sumanasekera G. Graphenenanoribbons exfoliated from graphite surface dislocationbands by electrostatic force. Nanotechnology2010;21(19):195704.

[6] Lu XK, Yu MF, Huang H, Ruoff RS. Tailoring graphite with thegoal of achieving single sheets. Nanotechnology1999;10(3):269–72.

[7] Kim K, Zhao Y, Jang H, Lee SY, Kim JM, Kim KW, et al. Large-scale pattern growth of graphene films for stretchabletransparent electrodes. Nature 2009;457:706–10.

[8] Wang XB, You HJ, Liu FM, Li MJ, Wan L, Li SQ, et al. Large-scale synthesis of few-layered graphene using CVD. ChemVapor Depos 2009;15(1–3):53–6.

[9] Bae SK, Kim HK, Lee YB, Xu XF, Park JS, Zheng Y, et al. Roll-to-roll production of 30-inch graphene films for transparentelectrodes. Nat Nanotech 2010;5(8):574–8.

[10] Stankovich S, Dikin DA, Dommett GHB, Kohlhaas KM,Zimney EJ, et al. Graphene-based composite materials.Nature 2006;442:282–6.

[11] Pinera RD, Nguyenb ST, Ruoffa RS. Synthesis and exfoliationof isocyanate-treated graphene oxide nanoplatelets. Carbon2007;44(15):3342–7.

[12] Xu Y, Bai H, Lu G, Li C, Shi GQ. Flexible graphene films via thefiltration of water-soluble noncovalent functionalizedgraphene sheets. J Am Chem Soc 2008;130(18):5856–7.

[13] Zhao W, Fang M, Wu F, Wu H, Wang L, Chen G. Preparation ofgraphene by exfoliation of graphite using wet ball milling. JMater Chem 2010;20:5817–9.

[14] Zhao W, Wu F, Wu H, Chen G. Preparation of colloidaldispersions of graphene sheets in organic solvents by usingball milling. J Nanomater 2010:528235.

[15] Antisari MV, Montone A, Jovic N, Piscopiello E, Alvani C,Pilloni L. Low energy pure shear milling: a method for thepreparation of graphite nano-sheets. Script Mater2006;55(11):1047–50.

[16] Wang J, Li Z, Fan G, Pan H, Chen Z, Zhang D. Reinforcementwith graphene nanosheets in aluminum matrix composites.Script Mater 2012;66(8):594–7.

[17] Bartolucci SF, Paras J, Rafiee MA, Rafiee J, Lee S, Kapoor D,et al. Graphene–aluminum nanocomposites. Mater Sci Eng A2011;528(27):7933–7.

[18] Rashad M, Pan F, Tang A, Asif M. Effect of graphenenanoplatelets addition on mechanical properties of purealuminum using a semi-powder method. Prog Nat Sci Mat Int2014;24:101–8.

[19] Perez-Bustamante R, Bolanos-Morales D, Bonilla-Martınez J,Estrada-Guel I, Martınez-Sanchez R. Microstructural andhardness behavior of graphene-nanoplatelets/aluminumcomposites synthesized by mechanical alloying. J AlloysComp 2014. http://dx.doi.org/10.1016/j.jallcom.2014.01.225.

[20] Latief FH, Sherifa EM, Almajid AA, Junaedi H. Fabrication ofexfoliated graphite nanoplatelets-reinforced aluminumcomposites and evaluating their mechanical properties andcorrosion behavior. J Anal Appl Pyrol 2011;92:485–92.

[21] Latief FH, Sherif EM. Effects of sintering temperature andgraphite addition on the mechanical properties of aluminum.J Ind Eng Chem 2012;18(6):2129–34.

[22] Chawla N, Shen Y. Mechanical behavior of particle reinforcedmetal matrix composites. Adv Eng Mater 2001;3(6):357–70.

C A R B O N 8 2 ( 2 0 1 5 ) 1 4 3 –1 5 1 151

[23] Song SG, Shin N, Gray GT, Roberts JA. Reinforcement shapeeffects on the fracture behavior and ductility of particulate-reinforced 6061-Al matrix composites. Metall Mater Trans A1996;27A:3739–46.

[24] George R, Kashyap KT, Rahul R, Yamdagni S. Strengthening incarbon nanotube/aluminium (CNT/Al) composites. ScriptMater 2005;53(10):1159–63.

[25] Barrera EV, Sims J, Callahan DL. Development of fullerene-reinforced aluminum. J Mater Res 1995;10(2):366–71.

[26] Song HY, Zha XW. Mechanical properties of Ni-coated singlegraphene sheet and their embedded aluminum matrixcomposites. Commun Theor Phys 2010;54(1):143–7.

[27] Xu ZP, Buehler MJ. Interface structure and mechanicsbetween graphene and metal substrates: a first-principlesstudy. J Phys Condens Mat 2010;22(48):485301.

[28] Banhart F. Interactions between metals and carbonnanotubes: at the interface between old and new materials.Nanoscale 2009;1:201–13.

[29] He Y, Zhang J, Wang Y, Yu Z. Coating geometries of metals onsingle-walled carbon nanotubes. Appl Phys Lett 2010;96.063108-1-3.

[30] Castiglioni C, Tommasini M, Zerbi G. Raman spectroscopy ofpolyconjugated molecules and materials: confinement effectin one and two dimensions. Phil Trans R Soc Lond A2004;362:2425–59.

[31] Kudin KN, Ozbas B, Schniepp HC, Prud’homme RK, Aksay IA,Car R. Raman spectra of graphite oxide and functionalizedgraphene sheets. Nano Lett 2008;8:36–41.

[32] Wang Z, Ciselli P, Peijs T. The extraordinary reinforcingefficiency of single-walled carbon nanotubes in orientedpoly(vinyl alcohol) tapes. Nanotechnology 2007;18:455709.

[33] Antunes EF, Lobo AO, Corat EJ, Trava-Airoldi VJ, Martin AA,Verıssimo C. Comparative study of first- and second-order

Raman spectra of MWCNT at visible and infrared laserexcitation. Carbon 2006;44(11):2202–11.

[34] Ferrari AC, Meyer JC, Scardaci V, Casiraghi C, Lazzeri M, MauriF, et al. Raman spectrum of graphene and graphene layers.Phys Rev Lett 2006;97(18):187401.

[35] Li D, Zhan D, Yan J, Sun CL, Li ZW, Ni ZH, et al. Thickness andstacking geometry effects on high frequency overtone andcombination Raman modes of graphene. J Raman Spectrosc2013;44:86–91.

[36] Choi HJ, Shin JH, Bae DH. Grain size effect on thestrengthening behavior of aluminum-based compositescontaining multi-walled carbon nanotubes. Compos SciTechnol 2011;71(15):1699–705.

[37] Courtney TH. Mechanical behavior of materials. 2nded. Singapore: McGraw-Hill Book Co; 2000.

[38] Zhong R, Cong H, Hou P. Fabrication of nano-Al basedcomposites reinforced by single-walled carbon nanotubes.Carbon 2003;41:848–51.

[39] Hedgepeth JM. Stress concentrations in filamentarystructures. Technical report. Washington, D.C.: NationalAeronautics and Space Administration (US), LangleyResearch Center; 1961 May. Report No.: NASA-TN-D-882.

[40] Ryu HJ, Cha SI, Hong SH. Generalized shear-lag model forload transfer in SiC/Al metal-matrix composites. J Mater Res2003;18:2851–8.

[41] Halpin JC, Kardos JL. The Halpin–Tsai equations: a review.Polym Eng Sci 1976;16(5):344–52.

[42] Yeh MK, Tai NH, Liu JH. Mechanical behavior of phenolic-based composites reinforced with multi-walled carbonnanotubes. Carbon 2006;44:1–9.

[43] Piggott MR. Loading-bearing fibrecomposites. Oxford: Pergamon; 1980.