Composites: Part A 71 (2015) 84–94

Contents lists available at ScienceDirect

Composites: Part A
journal homepage: www.elsevier.com/locate/compositesa

Microstructure evolution during fabrication and microstructure–

property relationships in vapour-grown carbon nanofibre-reinforced
aluminium matrix composites fabricated via powder metallurgy
Fumio Ogawa ⇑, Chitoshi Masuda
Kagami Memorial Institute for Materials Science and Technology, Waseda University, 2-8-26, Nishi-waseda, Shinjuku-ku, Tokyo 169-0051, Japan

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

Article history: Microstructure evolution of vapour-grown carbon nanofibre (VGCF)-reinforced aluminium matrix com-
Received 18 September 2014 posites during fabrication and microstructure–property relationships of these materials were studied.
Received in revised form 3 December 2014 Composites were fabricated using powder metallurgy, i.e. by mixing VGCFs and aluminium powder via
Accepted 12 January 2015
ball-milling followed by sintering or hot extrusion. The mixing condition was selected to achieve small
Available online 20 January 2015
powder particle size and homogeneously dispersed VGCFs. Aluminium grains and VGCFs were elongated
along the longitudinal direction of aluminium particles in the mixed powder. Detailed observation of the
Keywords:
aluminium grains in the composites found grain size and morphology dominated by recrystallization.
A. Metal matrix composites (MMCs)
B. Mechanical properties
Apparently, grain growth was inhibited by VGCFs. Theoretical models considering strength increment
B. Microstructures due to grain refinement resulting from VGCF addition, load bearing of VGCFs, thermal mismatch of VGCFs
E. Powder processing and aluminium and Orowan effect were developed. Theoretical values coincided well with hardness,
yield strength, and ultimate tensile strength of the composites, and thus the models could precisely
explain the microstructure–property relationships.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction were achieved using smaller aluminium powder particles due to

Aluminium matrix composites reinforced by vapour-grown car-
bon nanofibres (VGCFs) are attractive materials due to their
expected high strength, ductility and thermal conductivity [1–
10]. These properties originate from the superior properties of
the VGCFs. In addition, VGCFs have larger diameters than multi-
walled carbon nanotubes (MWCNTs) [11] and can thus be rela-
tively easily dispersed in aluminium matrices. Recently, various
methods for the fabrication of aluminium matrix composites rein-
forced by VGCFs have been proposed. Powder metallurgy, i.e. mix-
ing the VGCFs and aluminium powder followed by sintering or
plastic forming, such as hot extrusion [8] or hot rolling [9], is an
effective method for obtaining composites with superior proper-
ties, because the fabrication process is not affected by the wettabil-
ity of the VGCFs and the aluminium [4]. Xu et al. [1] fabricated
VGCF-reinforced aluminium matrix composites by wet mixing of
VGCFs and aluminium powder of different sizes followed by spark
plasma sintering (SPS) for consolidation of the mixed powder. A
high Vickers hardness and a low coefficient of thermal expansion
⇑ Corresponding author. Tel.: +81 3 5286 3792.
E-mail address: f.ogawa@akane.waseda.jp (F. Ogawa).
the excellent dispersion of the VGCFs in the aluminium matrix.
However, poor bonding at the VGCF/aluminium matrix interface
resulted in composites with low strength. Choi et al. [8] fabricated
MWCNT-reinforced aluminium matrix composites via mechanical
mixing of MWCNTs and aluminium powder followed by hot
extrusion. The composite strength was dramatically improved by
mixing the aluminium and MWCNTs using a ball-mill due to
refinement of the aluminium grains. However, the reduction in
the length of the reinforcement and the variation of the aluminium
grains due to extrusion were not fully investigated. Past studies
have revealed that the properties of reinforced aluminium compos-
ites are strongly dominated by their microstructures, i.e. alumin-
ium grains; the length, distribution and orientation of the
reinforcing material; the characteristics of the interface between
the reinforcing material and the aluminium matrix [12–16]. How-
ever, variation of the microstructures in the aforementioned alu-
minium matrix composites (using VGCFs or MWCNTs as
reinforcements) during fabrication has not been sufficiently inves-
tigated, and the relationships between the microstructures and the
properties of these composites are also not yet fully understood. In
particular, while several investigations of MWCNT-reinforced
aluminium matrix composites have appeared in the literature

http://dx.doi.org/10.1016/j.compositesa.2015.01.005
1359-835X/Ó 2015 Elsevier Ltd. All rights reserved.
 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94
[17,18], few studies have been reported on microstructure–prop-
erty relationships for VGCF-reinforced aluminium matrix
composites.
Herein, we describe the results of a study of VGCF-reinforced
aluminium matrix composites prepared via the ball-milling of
VGCFs with aluminium powder followed by SPS or hot extrusion.
The evolution of the microstructure, i.e. the variations of the alu-
minium grains and the VGCF length, due to mixing, sintering and
hot extrusion were investigated. The microstructure–property
relationships and the mechanism for reinforcement of the compos-
ites are also discussed.
2. Experimental
2.1. Mixing of VGCFs and aluminium powder
First, the conditions for mixing VGCFs and aluminium powder
for the preparation of the composites were investigated. The VGCFs
(Showa Denko Co. Ltd, Japan; average diameter: 150 nm, average
length: 15 lm) and aluminium powder (Kojundo Chemical
laboratory; average diameter: 30 lm) were mixed using a plane-
tary ball-mill (Fritsch Pulverisette 5). Table 1 lists the mechanical
and thermal properties of the VGCFs and aluminium. The VGCFs
and aluminium powder (total weight: 40 g) were placed in a stain-
less-steel jar containing stainless-steel balls (800 g). Therefore, the
weight fraction of the stainless-steel balls to the mixture of VGCFs
and aluminium powder was 20:1. To prevent excessive cold weld-
ing of the aluminium powder, stearic acid (C17H35COOH, 0.6 g) was
added as a process control agent. The VGCF volume fraction in the
powder was 0% or 3%. The stainless-steel jar was filled with argon
gas and rotated at 200 rpm for 0.5, 1, 2, 3, or 10 h. Rotation for
20 min followed by suspension for 40 min was repeated when
the milling duration was greater than or equal to 1 h. During
mixing, an argon atmosphere inside the jar was maintained. The
mixed powders were recovered after milling, and their morphol-
ogy was observed via scanning electron microscopy (SEM) using
JSM-6060 (JEOL). Subsequently, each powder was embedded in a
commercial epoxy resin in order to observe its interior. The resins
were polished and their surfaces were etched using nitric acid in
order to study the distribution and the length of the VGCFs inside
the powders. The aluminium grain sizes in the powders were eval-
uated via X-ray diffraction (XRD) analysis using Rint-Ultima III
(Rigaku). The accelerating voltage and filament emission were
40 kV and 40 mA, respectively, and a copper (Cu Ka) radiation
source was used. The Williamson–Hall method [7] was employed
to estimate the grain sizes, which were calculated by fitting the
following equation to the experimental results:
k
B0 cos h ¼ þ g sin h ð1:1Þ
L
1 The aluminium grains in the composites were observed as fol-
B0 ¼ ðB2T  B2I Þ2 ð1:2Þ
Table 1
Properties of the aluminium powder and VGCFs.
Property Aluminium VGCF
Geometry Average diameter Average diameter
30 lm 150 nm
Average length 15 lm
Density 2.7 g/cm3 2.0 g/cm3
Young’s modulus 76.0 GPa 516.5 GPa [20]
Tensile strength 100 MPa 3100 MPa [20]
Coefficient of thermal 23.6 ⁄ 106 K1 [10] 4.0 ⁄ 106 K1 [10]
expansion
85
where BT is the broadening, i.e. the breadth at half maximum inten-
sity, of each peak in the X-ray diffraction patterns for the powders,
BI is the instrumental broadening of the peaks in the XRD pattern of
the standard sample (aluminium powder annealed at 500 °C for
10 h), B’ is the corrected broadening, k is the wavelength of the X-
rays, h is the Bragg’s angle, L is the grain size and g is the lattice
strain stored during the milling process. The optimum mixing con-
dition for the fabrication of the composites was determined from
the morphologies of the mixed powders, their sizes, the VGCF distri-
butions in the powders, and the lengths of the VGCFs. Using the
optimized condition, we then prepared composites with VGCF vol-
ume fractions of 0, 0.5 and 1 vol.%. After mixing, these powders
were embedded in epoxy resins (G1 epoxy; Gatan) in order to
directly observe the aluminium grains. The resins were also pol-
ished and their surfaces were finished using a cross-section polisher
(CP; JEOL). The finished surfaces of the resins containing the embed-
ded powders were then observed via field emission scanning elec-
tron microscopy (FESEM) using JSM-7001F (JEOL).
2.2. Fabrication of composites
Some of the mixed aluminium and VGCF powders were consol-
idated using SPS (Dr. Sinter; SPS Syntex) in a vacuum. A carbon die
and a punch were used for sintering. The powder was placed in the
die, pressed with the punch, heated to 600 °C at the rate of 20 °C/
min, and then, sintered under an applied pressure of 50 MPa for
0.5 h (total heating time = 1 h). The volume fraction of the VGCFs
in the sintered composites was 1 vol.%.
For the composites that were prepared using extrusion, the
mixtures of aluminium and VGCF powders were encapsulated in
aluminium A1050 containers (outer diameter: 40 mm, inner diam-
eter: 24 mm, height: 40 mm, depth: 37 mm) in order to avoid oxi-
dation. Each powder was first placed in the container and then
compressed using a steel punch, and subsequently, the container
was evacuated using the SPS system and an aluminium lid was
installed by applying a force. The container was then machined
to a diameter of 30 mm and served as an extrusion billet. A 60°
conical die was used for the extrusion step, which was performed
at 550 °C at an extrusion ratio of 9. It should be noted that the tem-
perature was raised to 500 °C over 0.5 h and then to 550 °C over
another 0.5 h in order to avoid overheating. The temperature was
then held constant at 550 °C for 0.5 h prior to initiation of the
extrusion process. The volume fraction of VGCFs in the extruded
composites was 0.5 vol.%. In order to evaluate the changes in the
microstructure and mechanical properties of the different samples,
aluminium-sintered products and aluminium-extruded rods were
also fabricated for comparison.
2.3. Characterization of composites
lows: The composites were cut into 1.5-mm-thick, half-moon-
shaped pieces, which were polished and finished using the CP.
The composites fabricated via SPS or hot extrusion were observed
via FESEM. Several analyses were also performed in order to eval-
uate the mechanical properties of the composites. The Vickers
micro-hardness was determined for the samples by using an
indentation load of 5 N and a loading time of 5 s. The tensile
strength of the hot extruded composites was also determined.
Samples were machined to create tensile specimens in accordance
with Japanese Industrial Standard (JIS) Z 2241, and tensile testing
was performed on a tensile testing machine (Shimadzu; AG-
100KNC) by using a 0.5 mm/min crosshead speed. The strain in
the composites was also measured using an ultra-high-elongation
foil strain gauge (Kyowa). After tensile testing, the fracture surfaces
of the tested samples were observed via SEM.
86 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94
3. Results and discussion
3.1. Observation and characterization of milled powders
Fig. 1 shows SEM images of the aluminium powder and VGCFs
prior to mixing. The size of the aluminium powder particles was
not homogeneous, and they were shaped with elongated struc-
tures as shown in dashed ellipsoids (Fig. 1a). The arrows in the fig-
ure indicate particles smaller than 30 lm in diameter. The VGCFs
had a fibrous shape (Fig. 1b).
Fig. 2 presents SEM images of mixed powders containing 3 vol.%
VGCFs that were milled for 1, 2, 3 and 10 h. It can be seen in the
figure that the aluminium particles were flattened during initial
ball-milling (1 h; Fig. 2a); then, they began to stick to one another
(2 h; Fig. 2b). After further milling, the agglomerates were frac-
tured into small, spherical particles (3 h; Fig. 2c) due to embrittle-
ment resulting from work hardening. At longer milling times, the
particles began to stick to one another again due to the heat gen-
erated during milling (10 h; Fig. 2d).
The surface morphologies of the mixed powders are shown in
Fig. 3 (1 h: Fig. 3a; 3 h: Fig. 3b). In the powder milled for 1 h, many
VGCFs were observed on the surface, and agglomerated VGCFs
appeared to be surrounded by ellipsoids. In contrast, after milling
for 3 h, only a few VGCFs were detected on the surface, suggesting
that the VGCFs were incorporated into the aluminium particles. No
VGCFs were observed on the surface of the powder milled for 10 h
(data not shown).
Fig. 4 summarizes the variation in the aluminium grain size as
determined via XRD analysis. In the figure, the grain sizes of the
aluminium powder and the composite powders are compared.
Mixing using a ball-mill clearly reduced the grain size in both
the pure aluminium and the composite samples. Notably, the grain
sizes in the aluminium powder and the composite powders were
nearly the same for milling duration greater than or equal to 3 h.
The lattice strain in the aluminium grains induced by ball-milling
is also shown in Fig. 4 for the milled aluminium and composite
powders. After milling for 1 h, the strain in the composite powder
was less than that in the aluminium powder, because the process
of crushing and fracturing the VGCF agglomerates consumed the
energy associated with ball-milling. When the milling duration
was longer than 3 h, however, the lattice strain in the composite
powders was greater than or equal to that in the aluminium pow-
der. This increase in the lattice strain indicates the strain hardening
of the powders due to ball-milling.
In Fig. 5, the interiors of mixed powders containing 3 vol.% VGCF
can be seen after ball-milling for 3 h and 10 h. Numerous VGCFs
were observed in the composite powder milled for 3 h (Fig. 5a).
They were relatively dispersed and maintained their original
shape. In contrast, only a few VGCFs were detected in the compos-
ite powder milled for 10 h (Fig. 5b). It should also be noted that in
both the 3 h- and 10 h-milled powders, the lengths of the VGCFs
Fig. 1. SEM images of the (a) aluminium powder and (b) VGCFs prior to mixing.
were shorter than those before milling. The average lengths of
the VGCFs released from the etched surfaces of the powders were
found to be 6 and 1 lm for the 3 h- and 10 h-milled powders,
respectively. The high energy introduced by ball-milling clearly
resulted in the shortening of the VGCFs. These results indicated
that the VGCFs were incorporated into the aluminium powder dur-
ing the milling process and were dispersed by being cleaved into
shorter pieces. In addition, while the VGCFs were relatively well-
dispersed after both 3 h and 10 h of milling in powders with up
to 3 vol.% of the additive, agglomeration of the VGCFs was observed
in composite powders with VGCF volume fractions greater than or
equal to 4 vol.% (data not shown).
The length of the VGCFs is important because it significantly
influences their reinforcing efficiency and thus affects their ability
to enhance the strength of the composites. This information is
therefore essential for determining the appropriate mixing dura-
tion. Table 2 lists the critical lengths of the VGCFs, which are given
as a function of the matrix yield strength and calculated using Eq.
(2) [19]:
Drf
lc ¼ ð2:1Þ
2s
r
s ¼ pmffiffiffi ð2:2Þ
3
where lc, rf, D, rm and s correspond to the critical length, strength
and diameter of the VGCFs, the matrix yield strength and the inter-
facial yield strength, respectively. It is well known that the compos-
ite strength cannot be significantly enhanced beyond that of the
aluminium matrix when the VGCF length is shorter than the critical
length, because the reinforcement efficiency of VGCFs is consider-
ably lower (3100 MPa for the present study [20]) than that of
MWCNTs (50 GPa). On the other hand, VGCFs are more readily dis-
persed in aluminium matrices than MWCNTs [1,21]. Importantly,
for nearly all of the evaluated regions, the length of the VGCFs in
the 3 h-milled powder (6 lm) was greater than the critical length,
while that in the 10 h-milled powder (<1 lm) was shorter.
Thus, for further experiments, a mixing time of 3 h was selected
for the fabrication of the composites for the following reasons:
(i) The size of the powder particles was the smallest after mix-
ing for 3 h (Fig. 2c). A smaller powder particle size is benefi-
cial for densifying the composites.
(ii) The length of the VGCFs in the powder mixed for 3 h was
longer than the critical length, while the length of the VGCFs
in the powder mixed for 10 h was shorter than the critical
length.
Next, the aluminium powder size, grain size and grain structure
in composite powders with different VGCF volume fractions milled
for 3 h were evaluated. First, the variation in the powder size as a
 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94 87

Fig. 2. SEM images of the 3 vol.% VGCF/aluminium powder after ball-milling at 200 rpm for (a) 1 h, (b) 2 h, (c) 3 h and (d) 10 h.

Fig. 3. Surface morphology of the 3 vol.% VGCF/aluminium powder after ball-milling at 200 rpm for (a) 1 h and (b) 3 h.

containing 1 vol.% VGCF can be seen in Fig. 6. While the particle

Fig. 4. Variations in the aluminium grain size and the lattice strain as functions of
ball-milling duration. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
function of the VGCF volume fraction was investigated, and the
SEM images of the aluminium powder and the mixed powder
size of the aluminium powder after milling was large and exhibited
considerable variation, the particle size of the composite powder
and its variation were reduced due to the addition of the VGCFs.
The grain sizes were then evaluated via XRD for composite pow-
ders with VGCF volume fractions of 0, 0.5, 1, 2, 3, 4 and 5 vol.%
(data not shown). Despite some experimental error, the aluminium
grain sizes were similar for the powders with up to 3 vol.% VGCFs
(average 85 nm). For the composite powders with 4 and 5 vol.%
VGCFs, however, the grain sizes were less refined because the mill-
ing energy was consumed to shorten and disperse the VGCFs.
To date, aluminium grain sizes in milled powders prepared for
the fabrication of composites have been frequently evaluated using
XRD. However, to the best of our knowledge, a direct observation
of aluminium grains has rarely been reported. Fig. 7 shows the alu-
minium grains in the powder containing 0.5 vol.% VGCF. The pow-
der was embedded in an epoxy resin and the aluminium particles
were relatively flattened (Fig. 7a). The red arrows in the figure indi-
cate the longitudinal directions of the relevant particles. Fig. 7b
shows the grain structure in the region indicated by the white
dashed ellipsoid in Fig. 7a. While the black arrows indicate that
the VGCFs were oriented along the longitudinal direction of the
88 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94
Fig. 5. SEM images of the interiors of 3 vol.% VGCF/aluminium powders after ball-milling at 200 rpm for (a) 3 h and (b) 10 h.
Table 2
Relationship between the yield strength and theoretical critical length.
Matrix yield strength (MPa) 50 100 200 300
Critical length (lm) 8.1 4.0 2.0 1.3
particle, the black dashed arrows point to VGCFs shortened by
milling or oriented perpendicular to the observed surface. It is
apparent from the figure that the VGCFs were incorporated
between the aluminium grains. The black ellipsoids surround alu-
minium grains that were elongated along the longitudinal direc-
tion of the particle, while the dashed circles enclose shortened
and refined grains. It is thought that the aluminium grains were
initially elongated during the early stage of milling and then short-
ened into finer grains. Fig. 7c shows the distribution of the minor-
axis sizes of the grains in the composite powder. Notably, the
minor-axis size varied from 25 to 225 nm with an average of
74.1 nm, which is slightly less than the aluminium grain size as
determined via XRD (88.8 nm). The difference in the values
obtained using the different measurement methods is likely to be
caused by the orientation of the aluminium grains.
3.2. Microstructure, mechanical properties and the microstructure–
property relationships of the composites
Fig. 8 shows the Vickers micro-hardness of the aluminium and
composites fabricated via sintering. For comparison, the hardness
of a composite specimen containing 3 vol.% VGCF is also shown.
The hardness of the sintered aluminium was dramatically
improved after 3 h of ball-milling due to grain refinement. In addi-
tion, as the VGCF content increased, the hardness of the sintered
composites improved. This increase in hardness is thought to result
because the VGCFs act as obstacles for dislocation motion [22].
Fig. 6. SEM images of 3 h-milled powders: (a) Aluminium and (b) 1 vol.% VGCF/aluminium powder.
Because the enhancement of hardness is strongly related to the
morphology and size of the aluminium grains and the VGCF distri-
400 bution [22,23], the aluminium grains in the sintered composites
1.0 were directly observed via FESEM. Fig. 9a shows the aluminium
grains in the pure aluminium specimen fabricated by sintering of
3 h-milled powder, while Fig. 9b presents an FESEM image of the
aluminium grains in the 1 vol.% VGCF-reinforced composite speci-
men fabricated via sintering of 3 h-milled composite powder. The
aluminium grains in an aluminium specimen and a composite
specimen fabricated by sintering of 1 h-milled powders are shown
in Fig. 9c and d for comparison. The images in Fig. 9 are coloured
for ease of observation, with the white arrows indicating the thick-
ness directions, i.e. the directions along which pressure for consol-
idation was applied.
The materials fabricated from the powders milled for just 1 h
exhibited a lamellar grain structure (Fig. 9c and d) and the alumin-
ium grains were elongated perpendicular to the thickness of the
specimens. In addition, the shorter axis length of the aluminium
grains varied considerably, suggesting insufficient grain refinement
due to the short milling duration. The black arrows in Fig. 9c and d
indicate pores generated between unbonded aluminium grains.
These pores were formed by aluminium oxide films that remained
on the surfaces of the aluminium particles. The white dashed arrows
in Fig. 9d indicate VGCF agglomerations that were not adequately
dispersed. On the other hand, the grains in the composite fabricated
from the powder that was milled for 3 h were finer than those in the
composite fabricated from the powder milled for 1 h. The white
dashed arrows shown in Fig. 9b indicate relatively homogeneously
dispersed VGCFs, which were also observed in the grain boundaries
of the aluminium grains. The shorter axis, longer axis and the effec-
tive grain sizes of the grains in the aluminium and composite spec-
imens determined from the FESEM analysis (Fig. 9) are summarized
in Table 3. The standard deviations (SDs) of the grain sizes are listed
in bracketed, bold numerals in the table.
 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94 89

Fig. 7. Analysis of the interior of the 0.5 vol.% VGCF/aluminium composite powder: (a) FESEM image of the powder embedded in an epoxy resin, (b) FESEM image of the
aluminium grains inside an aluminium particle and (c) distribution of the minor-axis size of the aluminium grains. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

fabricated from powders that were milled for 1 h. However, the

Fig. 8. Vickers micro-hardness values for the composites fabricated via spark
plasma sintering as a function of the VGCF volume fraction. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web
version of this article.)
aluminium grains in the composite fabricated from the powder
milled for 3 h were clearly refined due to the presence of
the VGCFs. The SDs also decreased. These results indicate that the
addition of 1 vol.% VGCF inhibited aluminium grain growth in the
composite fabricated from the powder milled for 3 h. In addition,
a comparison of the aluminium grains in the composite powder
before sintering (Fig. 7b) and those in the composite specimen
obtained after sintering (Fig. 9b) clearly showed that recrystalliza-
tion of the aluminium grains occurred during sintering and that
grain growth during cooling was suppressed by the VGCFs (Fig. 9b).
Next, using the calculated values for the effective grain sizes of
the different sintered composites, we calculated the micro-
hardness values for the samples and compared them to the exper-
imental data (Table 3). The hardness of composites can be
expressed by Eqs. (4.1)–(4.6), that consider the aluminium grain
refinement, thermal mismatch of the constituents and the Orowan
effect [24–29]. It should be noted that because the VGCFs are
oriented nearly perpendicular to the thickness direction, in this
model, the load bearing of the VGCFs is neglected.

The effective grain sizes were then calculated using the follow- HV ¼ C ry ð4:1Þ
ing equation:
1 0:83M lb lnðD=bÞ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ry ¼ raluminium þ Drgrain size þ alq2 b þ ð4:2Þ
~ ¼ 3 pd2 h=4;
d ð3Þ 2pð1  mÞ1=2 L
 1 
where d is the average of the shorter axis lengths of the aluminium Drgrain size ¼ k d ~12
~ 2d ð4:3Þ
c aluminium
~
grains, h is the average of the longer axis lengths of the grains and d
is the effective grain size.  
4V f e 1 1 1
Very little difference in the effective grain sizes was observed q¼ þ þ ð4:4Þ
ð1  V f Þb t 1 t 2 t3
for the aluminium and the 1 vol.% VGCF-reinforced composite
90 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94

Fig. 9. FESEM images of the aluminium grains in the sintered products: (a) 3 h-milled aluminium, (b) 3 h-milled 1 vol.% VGCF/aluminium composite, (c) 1 h-milled
aluminium and (d) 1 h-milled 1 vol.% VGCF/aluminium composite. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version
of this article.)

Table 3
Aluminium grain sizes in the sintered aluminium and 1 vol.% VGCF/aluminium composites.

Milling Specimen Shorter axis d (nm) Longer axis h (nm) ~

Effective grain size d Experimental micro- Theoretical micro-
duration (h) (SD) (SD) (nm) hardness (HV) hardness (HV)

1 Aluminium 431.4 (314.3) 1734 (1336) 632.8 92.31 92.31

1 vol.% VGCF 389.2 (221.5) 1978 (1445) 617.3 81.22 110.5
composites
3 Aluminium 316.3 (181.1) 602.7 (339.0) 361.8 103.6 103.6
1 vol.% VGCF 234.7 (87.70) 427.5 (214.0) 264.3 120.3 129.5
composites

e ¼ DaDT
sffiffiffiffiffiffiffiffiffi
pD2
L¼
2V f
where HV and ry are the micro-hardness and yield strength of the
composites, C is a constant (3.16), raluminium is the yield strength
of the aluminium-sintered product calculated from the experimen-
tal micro-hardness using Eq. (4.1), and Drgrain size is the increase
in the matrix yield strength due to grain refinement upon addition
of the VGCFs. The third term in Eq. (4.2) expresses the increase in
the yield strength due to thermal mismatch of the VGCFs and the
aluminium matrix, and the fourth term in Eq. (4.2) corresponds to
the Orowan strengthening effect. The variables l, q, b, M, m, D and
a represent the shear modulus of the aluminium matrix, the disloca-
tion density, the Burgers vector of aluminium (0.286 nm), the Taylor
factor (3), Poisson’s ratio for the aluminium matrix, the VGCF diam-
eter and a geometric constant (1.25), respectively. The parameters k,
~ ,d~
d c aluminium, Vf, e, ti (i = 1, 2, 3), Da, and DT are the Hall–Petch slope
for the aluminium (0.09 for cryomilled aluminium [26]), effective
grain size of the aluminium matrix in the composites, effective grain
size in the aluminium specimen, VGCF volume fraction, thermal
misfit strain, actual size in the ith dimension of the reinforcement,
ð4:5Þ difference in the coefficients of thermal expansion for the constitu-
ents and the difference in the fabrication and test temperatures,
respectively, while L expresses inter-VGCF distance.
ð4:6Þ The micro-hardness of the 1 vol.% VGCF-reinforced composite
fabricated from the powder milled for 1 h deteriorated due to poor
dispersion of the VGCFs and poor densification. When the milling
duration was 3 h, however, the micro-hardness was improved
due to the addition of the VGCFs. The theoretical micro-hardness
values calculated using Eqs. (4.1)–(4.6) can explain this improve-
ment in the experimental micro-hardness.
As already mentioned, fabrication of composites via hot extru-
sion was also investigated in order to determine the effect of the
processing method on the microstructure and properties of the
composites. Fig. 10 shows the load-stroke curves for pure alumin-
ium and a 0.5 vol.% VGCF composite during extrusion. The load
drop during extrusion was attributed to the presence of the
A1050 aluminium lids in the upper portion of the extrusion billets,
because they had a low Young’s modulus as compared to the sam-
ples in the billets. Notably, the extrusion load for the 0.5 vol.%
VGCF-reinforced aluminium composite was lower than that of
pure aluminium. This result was likely to be due to the smaller par-
ticle size of the 0.5 vol.% VGCF/aluminium powder as compared to
that of the aluminium powder.
 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94 91

suggests that considerable strain remained in the extruded alu-

Fig. 10. Extrusion load-stroke curves for pure aluminium and the 0.5 vol.% VGCF/
aluminium composite.
Fig. 11 shows the aluminium grains in a sample of extruded alu-
minium observed via FESEM (low- and high-magnification images:
Fig. 11a and b, respectively). The images in Fig. 11 are coloured for
ease of observation, with the extrusion directions indicated by
white arrows. The aluminium grain size varied considerably, and
the grains were elongated in the extrusion direction (Fig. 11a). In
addition, highly deformed regions were observed between the alu-
minium grains in the high-magnification image, as indicated by the
black arrows in Fig. 11b. These areas were likely to be caused by
the high extrusion load generated as a result of the large and scat-
tered aluminium particle size before extrusion, and their formation
minium product. Finally, the white dashed ellipsoid indicates a
region consisting of aluminium grains that are smaller than the
other grains in the image. The presence of this region suggests spa-
tial strength variation in the extruded aluminium rod.
Fig. 12 shows the aluminium grains in a 0.5 vol.% VGCF-
reinforced aluminium matrix composite (low- and high-magnification
images: Fig. 12a and b, respectively). The images in Fig. 12 are also
coloured for ease of observation, with the extrusion directions
indicated by white arrows (those with numbers indicate the
VGCFs). Notably, the aluminium grains in the composite
(Fig. 12a) were not elongated in the extrusion direction as signifi-
cantly as those in the aluminium-extruded rod (Fig. 11a). A VGCF
(numbered 1) can also be seen bridging multiple aluminium grains.
In addition, the number 3 VGCF had a diameter larger than the
average diameter of the VGCFs (150 nm) and was oriented perpen-
dicular to the extrusion direction (Fig. 12a). In the higher-magnifi-
cation image (Fig. 12b), it can be seen that the aluminium grains
were refined as compared to those in the aluminium-extruded
rod and they were smaller near the VGCFs (indicated by white
arrows). This result suggests that the VGCFs function as originators
of grain refinement and inhibitors of grain growth. In addition,
barely any highly deformed regions were observed between the
aluminium grains. Table 4 summarizes the aluminium grain sizes
obtained using the images in Figs. 11 and 12, along with the exper-
imental and theoretical (calculated using Eqs. (4.1)–(4.6)) hardness
values of the extruded aluminium and composite specimens.
As was observed for the sintered composite, the effective size of
the aluminium grains in the extruded composite decreased due to
the addition of the VGCFs. The micro-hardness was also improved
by the presence of the VGCFs and the theoretical values were in
good agreement with the experimental results; this confirms the

(a) (b)

Fig. 11. FESEM images of the aluminium grains in the extruded pure aluminium: low-magnification image and (b) high-magnification image. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 12. FESEM images of aluminium grains in the 0.5 vol.% VGCF/aluminium composite: (a) low-magnification image and (b) high-magnification image. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)
92 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94
Table 4
Aluminium grain sizes in the extruded aluminium rod and 0.5 vol.% VGCF/aluminium composites.
VGCF volume fraction Shorter axis d nm Longer axis h (nm)
% (SD) (SD)
0 266.4 (164.5) 562.1 (334.6)
0.5 243.2 (129.8) 401.3 (257.1)
Fig. 13. True stress-true strain curves for pure aluminium and the extruded
0.5 vol.% VGCF/aluminium composite.
contributions of grain refinement, the thermal mismatch of the
constituents and the Orowan effect.
Fig. 13 shows the true stress-true strain curves for the extruded
aluminium and 0.5 vol.% VGCF-reinforced aluminium composite
specimens. The true stress and the true strain were calculated from
the load-stroke curves using Segal’s equation, which expresses the
true stress-true strain curves of metallic materials after severe
plastic deformation [30]. The ultimate strength, yield strength
and fracture strain of the composite were improved as compared
to those of the aluminium. In this study, the 0.2% proof stress
was used as the yield strength. Notably, the fracture strain of the
composite was 30.22%, which was approximately twice that of
the aluminium-extruded rod (14.39%). This large fracture strain
was attributed to the refined and spherical aluminium grains
generated near the VGCFs (Fig. 12).
Fig. 14 shows the fracture surface of the aluminium extruded
rod. Fig. 14a and b shows the images for different positions in
the fracture surface. In Fig. 14a that is the image of the ductile
fracture surface, numerous elongated dimples can be seen on the
fracture surface. Dimples generated near fractured hard aluminium
particles are indicated by the white arrows in Fig. 14b. These
(a)
5µm
Fig. 14. SEM images of the fracture surface of the aluminium-extruded rod.
~
Effective grain size d Experimental micro-hardness Theoretical micro-hardness
(nm) (HV) (HV)
315.3 104.3 104.3
265.2 116.6 121.1
particles likely correspond to the region in the white dashed ellip-
soid in Fig. 11b. Based on this analysis, it is evident that the frac-
ture process and the strength of the extruded aluminium were
strongly dominated by the aluminium grain size and its variability.
Fig. 15 shows the fracture surface of the 0.5 vol.% VGCF-rein-
forced aluminium matrix composite. Fig. 15a and b shows the
images of different positions in the fracture surface. Numerous
fractured VGCFs and dimples generated near the fractured VGCF
can be observed in Fig. 15a, indicating that the ductile fracture
was accompanied by high fracture strain. It can also be seen that
the VGCFs were well-oriented along the longitudinal direction of
the composite specimen. However, VGCFs oriented perpendicular
to the longitudinal direction of the composite specimen were also
detected, as shown in Fig. 15b (indicated by white arrows). It can
be seen in the figure that these VGCFs maintained their original
shapes. Though, note that most of the VGCFs in the composite
specimen were oriented along the longitudinal direction. These
results therefore confirmed that hot extrusion is an effective
method for controlling the reinforcement orientation and the fab-
rication of composites with superior mechanical properties.
Observation of the fracture surface of the extruded 0.5 vol.%
VGCF-reinforced aluminium matrix composite (Fig. 15) also sug-
gested that the load was effectively transferred from the matrix
to the VGCFs, because numerous fine dimples were generated
around the fractured VGCFs, and pulled-out VGCFs were hardly
observed. XRD analysis of the extruded 0.5 vol.% VGCF-reinforced
aluminium matrix composite specimen revealed that a minor
amount of aluminium carbide (Al4C3) was formed (data not
shown). Kwon and Leparoux [31] also reported the formation of
Al4C3 in MWCNT-reinforced aluminium matrix composites fabri-
cated via hot extrusion. A detailed transmission electron micros-
copy (TEM) study revealed that Al4C3 formation occurred at the
MWCNT/aluminium interface [31]. They demonstrated that the
formation of Al4C3 improved the stress transfer at the interface
with only minimal degradation of the MWCNT strength. Similarly,
in the present study, it was found that the tensile stress was effec-
tively transferred from the aluminium matrix to the VGCFs
through the interface due to the strong interfacial bonding
between the VGCFs and the aluminium matrix. However, it has
been reported that the hydrolysis of Al4C3 leads to the degradation
of composite properties [32]. Therefore, further studies are
(b)
2µm
 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94 93

(a) (b)

2µm 2µm

Fig. 15. SEM images of the fracture surface of the 0.5 vol.% VGCF-aluminium-extruded rod.

Table 5
Mechanical properties of composites fabricated via hot extrusion.

VGCF volume Young’s modulus Theoretical modulus Yield strength Theoretical yield strength Ultimate Strength Theoretical strength
fraction % (GPa) (GPa) (MPa) (MPa) (MPa) (MPa)
0 69.40 69.40 239.3 239.3 378.3 378.3
0.5 71.84 71.48 287.5 296.9 427.1 432.9

required to elucidate the effect of Al4C3 on the long-term mechan- determined and calculated values for the Young’s modulus, yield
ical properties of composites. strength, and the ultimate strength of the extruded aluminium
Finally, modelling of the tensile properties of the extruded com- and 0.5 vol.% VGCF-reinforced aluminium matrix composite sam-
posite samples was performed. The Young’s moduli were calcu- ples are summarized in Table 5. The enhancement of strength due
lated using Cox’s well-known model [33], which is frequently to the addition of VGCFs was only slight due to the low ratio of
used for composites. The theoretical yield strength and tensile VGCF strength (3100 MPa) to that of the matrix (378.3 MPa). On
strength were modelled using the following equations: the other hand, the fracture strain, i.e. the ductility, was quite high
! (Fig. 13). Most importantly, by taking into account the microstruc-
1 0:83M lb lnðD=bÞ ture of the composite, it was possible to theoretically predict its
rc ¼ ð1  V p Þ rL þ alq2 b þ ð5:1Þ
2pð1  mÞ1=2 L performance properties, and the calculated values were in good
agreement with the experimental results. Therefore, the models
 
lc used in the present study can accurately describe the microstruc-
rL ¼ V f rf F þ ð1  V f Þrm refined ð5:2Þ
ture–property relationships and strengthening mechanisms in
l
VGCF-reinforced aluminium matrix composites, even when differ-
rm refined ¼ raluminium þ Drgrain size ð5:3Þ ences in the aluminium grain morphologies are neglected
(Figs. 11b and 12b). The effect of the variation in the spatial alumin-
 1 
~12
~ 2  d ium grain size on strengthening in aluminium matrix nanocompos-
Drgrain size ¼k d 0 ð5:4Þ
ites will be pursued in future studies.
  ( lc
lc 1  2l ðl > lc Þ
F ¼ ð5:5Þ 4. Conclusions
l l=ð2lc Þðl < lc Þ
The evolution of the microstructures of VGCF-reinforced alu-
D rf minium matrix composites during the fabrication process was
lc ¼ ð5:6Þ
2s investigated. The VGCFs were shortened and the aluminium grains
rm refined in the mixed powders were refined by ball-milling. In particular,
s¼ pffiffiffi ð5:7Þ the aluminium grains were elongated along the longitudinal direc-
3 tion of the aluminium particles in the mixed powders (Fig. 7b).
rc is the theoretical yield strength or tensile strength and Vp is the Recrystallization during sintering or hot extrusion followed by
porosity of the composite (2.22%). The load contribution for the cooling largely determined the aluminium grain sizes and the
VGCFs, i.e. rL, was expressed using a modified Kelly–Tyson equation microstructures of the composites. Preferential recrystallization
(Eq. (5.2)) considering the increase in the aluminium matrix of the grains occurred near the VGCFs, and the resulting smaller
strength and interfacial strength due to the refined grain size upon grains in these regions contributed to the high ductility of the com-
addition of the VGCFs. The parameter rm_refined corresponds to the posites. Theoretical models for the hardness and strength of the
yield or tensile strength of the aluminium matrix in the composite composites were also developed considering the increase in the
specimen, raluminium is the experimental yield strength or tensile strength of the aluminium matrix due to recrystallization and grain
strength of the aluminium-extruded rod, Drgrain size is the increase refinement, the load bearing of the VGCFs, the increase in strength
in the matrix strength due to grain refinement upon addition of the due to the thermal mismatch of the constituents and the Orowan
VGCFs and l is the length of the VGCFs. The second term in the strengthening effect. The calculated values were in good agree-
bracket in Eq. (5.1) expresses the effect of the thermal mismatch ment with the experimental values, thus confirming that all of
between the VGCFs and the aluminium matrix, while the third term the above mechanisms contribute to the strengthening of the com-
expresses the Orowan strengthening effect. The experimentally posites. The results of the present study should be useful not only
94 F. Ogawa, C. Masuda / Composites: Part A 71 (2015) 84–94
for elucidating the microstructure–property relationships and
strengthening mechanisms but also for establishing methods for
the fabrication of VGCF-reinforced aluminium matrix composites
with superior properties.
Acknowledgements
This work was partly supported by Grants-in-Aid for Scientific
Research from the Ministry of Education, Culture, Sports, Science
and Technology, Japan. The first author (F. O.) would like to express
his sincere gratitude to the Mitsubishi Material Corporation (MMC)
for financial support. Financial supports from the Light Metal Edu-
cational Foundation Inc. and the Japan Aluminium Association are
also greatly acknowledged.
References
[1] Xu ZF, Choi YB, Matsugi K, Li DC, Sasaki G. Mechanical and thermal Properties
of vapor-grown carbon fiber reinforced aluminum matrix composites by
plasma sintering. Mater Trans JIM 2010;51:510–5.
[2] Sasaki G, Kondo F, Matsugi K, Yanagisawa O. Electrical conductivity of VGCF/
aluminum composites fabricated by electro spark sintering. Mater Sci Forum
2007;561–5:729–32.
[3] Kwon H, Park DH, Silvain JF, Kawasaki A. Investigation of carbon nanotube
reinforced aluminum matrix composite materials. Compos Sci Technol
2010;70:546–50.
[4] Uozumi H, Kobayashi K, Masuda C, Yoshida M. Fabrication process of
carbonaceous fiber reinforced Al and/or Mg alloy(s) composite by squeeze
casting. Adv Mater Res 2006;15–17:209–14.
[5] Esawi AMK, Morsi K, Sayed A, Gawad AA, Borah P. Fabrication and properties of
dispersed carbon nanotube–aluminum composites. Mater Sci Eng A
2009;508:167–73.
[6] Esawi AMK, Morsi K, Sayed A, Taher M, Lanka S. Effect of carbon nanotubes
(CNT) content on the evolution of aluminum (Al)-CNT composite powders.
Compos Sci Technol 2010;70:2237–41.
[7] Poirier D, Gauvin R, Drew RAL. Structural characterization of a mechanically
milled carbon nanotube/aluminum mixture. Composites Part A
2009;40:1482–9.
[8] Choi HJ, Kwon GB, Lee GY, Bae DH. Reinforcement with carbon nanotubes in
aluminum matrix composites. Scr Mater 2008;59:360–3.
[9] Choi H, Shin J, Min B, Park J, Bae D. Reinforcing effects of carbon nanotubes in
structural aluminum matrix nanocomposites. J Mater Res 2009;24:2610–6.
[10] George R, Kashyap KT, Rahul R, Yamdagni S. Strengthening in carbon
nanotube/aluminum (CNT/Al) composites. Scr Mater 2005;53:1159–63.
[11] Advani SG. Processing and properties of nanocomposites. World Scientific;
2006.
[12] Christman T, Needleman A, Nutt S, Suresh S. On microstructural evolution and
micromechanical modeling of deformation of a whisker-reinforced metal-
matrix composite. Mater Sci Eng A 1989;107:49–61.
[13] Christman T, Suresh S. Microstructural development in an aluminum alloy –
SiC whisker composite. Acta Metall 1988;36:1691–704.
[14] Hall JN, Jones JW, Sachdev A. Particle size, volume fraction and matrix strength
effects on fatigue behaviour and particle fracture in 2124 aluminium–SiCp
composites. Mater Sci Eng A 1994;183:69–80.
[15] Spowart JE, Maruyama B, Miracle DB. Multi-scale characterization of spatially
heterogeneous systems: implications for discontinuously-reinforced metal
matrix composite microstructures. Mater Sci Eng A 2001;307:51–66.
[16] Watanabe H, Ohori K, Takeuchi Y. Effects of hot extrusion and rolling on the
tensile strength of SiC whisker reinforced aluminum alloy composites. J Jpn
Inst Light Met 1988;38:633–8.
[17] Choi HJ, Bae DH. The effect of milling conditions on microstructures and
mechanical properties of Al/MWCNT composites. Composites Part A
2012;43:1061–72.
[18] Liu ZY, Xiao BL, Wang WG, Ma ZY. Singly dispersed carbon nanotube/
aluminum composites fabricated by powder metallurgy combined with
friction stir processing. Carbon 2012;50:1843–52.
[19] Kelly A, Tyson WR. Tensile properties of fiber-reinforced metals: copper/
tungsten and copper/molybdenum. J Mech Phys Solid 1965;13:329–50.
[20] Hu N, Li Y, Nakamura T, Katsumata T, Koshikawa T, Arai M. Reinforcement
effects of MWCNT and VGCF in bulk composites and interlayer of CFRP
laminates. Composites Part B 2012;43:3–9.
[21] Wu Y, Kim GY, Russel AM. Effects of mechanical alloying on an Al6061-CNT
composite fabricated by semi-solid powder processing. Mater Sci Eng A
2012;538:164–72.
[22] Casati R, Vedani M. Metal matrix composites reinforced by nano-particle–a
review. Metals 2014;4:65–83.
[23] Basariya MR, Srivastava VC, Mukhopadhyay NK. Microstructural
characteristics and mechanical properties of carbon nanotube reinforced
aluminum alloy composites produced by ball milling. Mater Des
2014;64:542–9.
[24] Qiao XG, Gao N, Starink MJ. A model of grain refinement and strengthening of
Al alloys due to cold severe plastic deformation. Philos Mag 2012;92:446–70.
[25] Arsenault RJ, Shi N. Dislocation generation due to differences between the
coefficients of thermal expansion. Mater Sci Eng A 1986;81:175–87.
[26] Hayes RW, Witkin D, Zhou F, Lavernia EJ. Deformation and activation volumes
of cryomilled ultrafine-grained aluminum. Acta Mater 2004;52:4259–71.
[27] Khan AS, Suh YS, Chen X, Takacs L, Zhang H. Nanocrystalline aluminum and
iron: mechanical behavior at quasi-static and high strain rates, and
constitutive modeling. Int J Plasticity 2006;22:195–209.
[28] Li Q, Viereckl A, Rottmair CA, Singer RF. Improved processing of carbon
nanotube/magnesium alloy composites. Compos Sci Technol 2009;69:1193–9.
[29] Queyreau S, Monnet G, Devincre C. Orowan strengthening and forest
hardening superposition examined by dislocation dynamics simulations.
Acta Mater 2010;58:5586–95.
[30] Segal VM, Ferrasse S, Alford F. Tensile testing of ultra fine grained metals.
Mater Sci Eng A 2006;422:321–6.
[31] Kwon H, Leparoux M. Hot extruded carbon nanotube reinforced aluminum
matrix composite materials. Nanotechnology 2012;23:415701.
[32] Rodriguez-Reyes M, Pech-Canul MI, Parga-Torres JR, Acevedo-Dávila JL,
Sánchez-Araiza M, López HF. Development of aluminum hydroxides in Al–
Mg–Si/SiCp infiltrated composites exposed to moist air. Ceram Int
2011;37:2719–22.
[33] Cox HL. The elasticity and strength of paper and other fibrous materials. Brit J
Appl Phys 1952;3:72–9.