Академический Документы
Профессиональный Документы
Культура Документы
Laboratory of Advanced Materials Processing, Empa Swiss Federal Laboratories for Materials Science and Technology, Feuerwerkerstrasse 39, 3602 Thun, Switzerland
Department of Materials Engineering, Tallinn University of Technology, Ehitajate tee 5, 19086 Tallinn, Estonia
c
Department of Materials System Engineering, Pukyong National University, 365 Sinseon-ro, Nam-gu, Busan 608-739, Korea
b
a r t i c l e
i n f o
Article history:
Received 19 March 2013
Received in revised form 13 June 2013
Accepted 21 September 2013
Available online 30 September 2013
Keywords:
Aluminium matrix composite
X-ray diffraction
Hardness test
Carbon nanotubes
HallPetch effect
a b s t r a c t
The macro hardness (HV20) was measured for aluminium and 19 wt% Multi Wall Carbon Nanotubes
(MW-CNTs) composites that were milled and hot compressed. The hardness increased with increasing
fraction of MW-CNTs up to 6 wt% (HV20 = 151) and then remained constant. The content of MW-CNTs
was signicantly higher than reported for similar materials and measurements. The composites were
analysed by Raman, TEM and XRD. The Raman and TEM showed MW-CNTs were still present after milling
and hot pressing. The XRD was used to determine the Al crystallite size which was used to determine
HallPetch contribution to the composite hardness.
2013 Elsevier B.V. All rights reserved.
1. Introduction
The need to pursue new materials which are lighter and stronger is still ever present. This is evident in elds such as Aerospace,
Renewable Energy, Electronics, Architectural Structure and
Alternative Transport where Metal Matrix Composites (MMC)
materials have found increasing use [1]. Since the unique properties of Carbon Nanotubes (CNTs) were reported by Iijima in 1991
[2], they have been increasingly used as a reinforcement material
in polymers [3], ceramics [4] and metals [5] because of their low
density and very high theoretical strength [6]. At present, for
applications where high strength is needed, the CNTs are combined
with a matrix to utilise these unique properties. The light metals,
such as Mg and Al, are ideal matrices because there is a large scope
for increasing the strength, while not increasing the weight [1]. In
this study, pure Al was chosen because of its simple chemistry,
material properties and structure.
The use of CNTs in matrices other than organic matrices (i.e.
polymer) has been less developed. The initial research involving
CNTs and Metal Matrix Composites (MMC) has been slower to
develop from the initial work of Doome et al. and Kuzumaki
et al. in 1998 [7,8]. A number of publications have appeared recently regarding the fabrication of CNTMMC in different metals
matrices, such as Al, Cu, Mg, Ti, Fe, Fe3Al, Ni, Sb, Ag, Co and Metallic
C.R. Bradbury et al. / Journal of Alloys and Compounds 585 (2014) 362367
2. Experimental
The Al powder used in all experiments was supplied by Ecka Granules (AS 011
purity P99.5%) with a particle size of less than 65 lm and the MW-CNTs were from
Bayer (Baytubes C150P, purity >99.5%). A detailed analysis of the MW-CNTs has
previously been conducted by Tessonnier et al. [24]. Samples of MW-CNT Al
MMC were produced with varying amount of MW-CNTs, 1 wt%, 3 wt%, 6 wt% and
9 wt% added to the Al matrix. The MW-CNT and Al were mixed by a laboratory
planetary ball mill (PM400, Retsch GmbH) under an inert atmosphere of Argon.
Milling was conducted using 10 mm steel balls with a Ball to Powder Ration
(BPR) of 10:1 at a speed of 360 rpm. 20 wt% Heptane was added as a Process Control
Agent (PCA). Milling was conducted for a total milling time of 20 h with a repeated
sequence of 20 min milling, followed by a 10 min pause interval. Further details
regarding the milling process can be found elsewhere [25]. After milling, the milling
balls showed negligible wear. The milled powders were consolidated by hot pressing to form bulk samples of 30 mm in diameter and with a 3 mm thickness. The
powders were placed into a die made of high temperature steel, heated in an oven
up to a temperature of 350 C for 60 min and rapidly transported to the uniaxial
press (less than 5 s). Compaction was done by applying a pressure of 570 MPa for
10 s. For the extruded samples, the blend was put into an extrusion mould, heated
to 500 C for 1.5 h and then transported rapidly to the press. The extrusion was performed with an extrusion ration of 14:1, a ram speed of 1 mm s1 for the sample
with a diameter of 8 mm. The length was typically around 1520 cm allowing the
preparation of two tensile specimens per extrusion.
The density of all samples was measured by Archimedes method according to
ISO 3369:1975 and all had a theoretical density between 96% and 99%. The macro
Vickers Hardness (HV20) was measured according to EN ISO 6507-1 with a load
of 20 kg for 15 s (220, GNEHM Hrteprfer AG). Raman spectra were measured
on a Renishaw 2000 spectrometer equipped with holographic notch lters for elastic scattering and a CCD array detector. The samples were excited with the helium
neon laser (632.8 nm). The laser was focused onto the sample using the 50 objective that is mounted to the microscope. Each sample was measured in twelve
different places and each spectrum was corrected for the baseline. The instrument
363
was calibrated with Si single crystal. The electron microscope images were taken
with a Jeol JEM-2200FS high resolution transmission electron microscope. The fracture surface of the extruded samples was observed after tensile tests under a high
resolution scanning electron microscope (Hitachi S-4800). The XRD was measured
with XPert Pro diffractometer (PANAlytical) with a Cu Ka (k = 1.5148 , 35 kV and
40 mA) in the 2h range 25140 using a linear detector (XCelerator). A step size of
0.0167 and a scan rate of 0.021/s was used. The instrumental line broadening was
determined with an annealed sample of Y2O3 (Alfa Aesar, purity 99.999%). The Rietveld renement was done using a program called Maud developed by Prof. Lutterotti at the University of Torento [26].
Extruded samples were machined to achieve a standard tensile test specimen
with a gauge diameter of 4 mm according to DIN 50125. The tensile tests were performed on a UTS machine according to EN ISO 6852-1:2009.
Fig. 1. TEM images of the Al/MW-CNT composite. Image A and B show the presence of MW-CNTs on the surface of the milled Al particles. Image C shows the presence of
nanosized Al grains in the matrix (highlighted by white arrows) and the formations of some Al4C3 during the milling and hot pressing. Image D shows that MW-CNTs are
present in the milled and hot pressed Al matrix.
364
C.R. Bradbury et al. / Journal of Alloys and Compounds 585 (2014) 362367
Fig. 2. The Raman analysis of Al/MW-CNT with increasing wt% of MW-CNT. Typical Raman spectra of the D and G bands for 1 wt% to 9 wt% MW-CNTs (A). A Raman spectrum
of the MW-CNTs is given as a comparison. The area of the IG band with increasing wt% MW-CNTs and the ID/IG ratio is given in (B).
Fig. 3. The effect on hardness (HV20) for increasing wt% MW-CNTs. The error bars
are a standard deviation of a number of different samples. Also included is the
percentage of the theoretical density of the samples assuming a density of
Aluminium of 2.70 g cm3 and the density of the MW-CNTs is 2.0 g cm3.
C.R. Bradbury et al. / Journal of Alloys and Compounds 585 (2014) 362367
365
Fig. 4. The XRD analysis of Al/MW-CNT with increasing wt% of MW-CNT. A typical example of the XRD pattern is given for pure Al and 6 wt% MW-CNT for the milled powder
and the resulting hot compressed sample. The assignment of the Al peaks and two small peaks assigned to Al4C3 are given (A). Al crystallite size (lled symbols) versus the
wt% of MW-CNTs determined by Rietveld renement for both the milled powder (squares) and the hot compressed samples (circles) (B). The error bars are an estimation of
the error.
Fig. 5. The HallPetch effect for pure Al (dashed line), compared to the experimental values for the crystallite size and hardness presented in Figs. 3 and 4
(circles).
typical representative example. The XRD pattern of a pure Al sample and 6 wt% Al/MW-CNTs of the milled powder and a hot compressed sample are given in Fig. 4A. The broadening of the XRD
peaks of the milled Al/MW-CNT powders indicates a decrease in
crystallite size as described already by other authors using high energy milling for aluminium based nanocomposites. Two small
broad peaks at 31.2 and 54.9 observed in the hot compressed
sample, however not in the milled powder, are assigned to Al4C3.
This is in agreement with the TEM analyses performed on the compacted samples that revealed that whatever the CNT content, even
at the low pressing temperature of 350 C, some CNT are reacting
with aluminium to form an aluminium carbide phase (Fig. 1). It
is interesting to note that no aluminium carbide, at least over the
detection limit of the XRD measurements around 1%, is formed
during high energy milling [29,30]. No diffraction peak was observed for the MW-CNTs. The XRD patterns were further analysed
by Rietveld renement to determine the crystallite size. As shown
in Fig. 4B, the crystallite size for both the milled powder and the
hot compressed sample decreases to 50 nm with 6 wt% MW-CNTs
and remains almost constant to 9 wt% MW-CNTs. The crystallite
size measured here were slightly smaller than those observed by
Choi et al. [20]. However, the distribution of the crystallite size,
analysed by the Rietveld renement, showed no signicant different between the milled powder and the compressed composite.
Also separate annealing experiments (8 h at 350 C and then more
than 800 h at 450 C) showed no signicant increase in the crystallite size which could results from the hot rolling process. The
crystallite size of the pure aluminium sample could not be measured using this Rietveld renement technique. Actually, no peak
broadening could be measured for crystallite sizes over 200 nm.
Metals show an increase in strength with decreasing crystallite
size due to the pile up of dislocations at the crystal boundaries (referred as the HallPetch effect). Assuming the hardness is 1/3 of
the yield strength (Tabor relationship) where no work hardening
is observed [31], the Hardness increases (HVcomp) with decreasing
crystallite size by the following equation:
HVcomp
ro Kd0:5
3
where ro is the intrinsic strength of the material (ro = 15.7 MPa for
Al), K is the constant of individual metals (K = 0.068 MPa/m3/2 for
Al) and d is the diameter of the crystallite (m) [32].
Fig. 5 gives the predicted values for the HallPetch effect for Al
(dashed line) and the corresponding hardness and crystallite size
for the composites prepared in this work. The hardness of the composites containing 39 wt% MW-CNTs gives hardnesss 2045 HV
higher than predicted by the HallPetch effect. The reason for the
1 wt% MW-CNTs being below the predicted values could relate to
the lower density of the sample, but was not investigated any further. The crystallite size does not alone contribute to the hardness
of the MW-CNT composites, but adds to about 7080% of the reinforcement. This level of reinforcement contributed by the crystallite size is supported by other work considering ne grain metals
with particulate reinforcement [33]. This work supports the comparison performed by Bakshi et al. [9] that the distribution of the
366
C.R. Bradbury et al. / Journal of Alloys and Compounds 585 (2014) 362367
Table 1
Mechanical properties of the pure aluminium material and the composite reinforced with 3 wt% MW-CNT.
Al
Al3 wt% CNT
Hardness HV20
Elongation A5 (%)
45 5
117 8
118 2
428 8
72 1
385 6
25 1
3.7 1
68 1
96 3
>250
62 6
Fig. 7. A SEM image of the fractured surface after the tensile test presented in Fig. 6. The MW-CNTs are observed in the cracks highlighted by the arrows.
CNTs is vital parameter regarding the performance of AlCNT composites, with the further addition of the matrix hardening due to
grain renement.
Work has begun to extend this study to correlate the tensile
properties with the hardness values present here. The stress stain
curve of Al 3 wt% MW-CNT is given in Fig. 6. A comparison with
pure aluminium is given to illustrate the signicant increase in
strength and also the decrease in elongation. Mean characteristics
for these two materials are summarized in Table 1. The yield
strength ts with the Tabor relationship relating to the hardness
values given in Figs. 3 and 5. As observed by others, there is a subsequent decrease in the elongation to 24%. With the hydraulic
press used in these experiments, the Al6 wt% MW-CNTs powders
were too hard to extrude. Further work is continuing to optimise
the method and geometry to extrude higher concentration of
MW-CNTs. This sample illustrates the point that the application
of the Tabor relationship is valid. The fracture surface of the tensile
specimens was observed under high resolution electron microscope. Some carbon nanotubes could be observed in the cracks as
seen in Fig. 7. Some of them are even binding the crack surfaces together. The nanotubes seem also to be covered with aluminium
that would indicate a good interfacial cohesion between the matrix
and the CNT. The MW-CNTs are also aligned with the applied force.
The simple model of Al nano-crystallites obtained from the
HallPetch effect can account for a signicant amount of the hardness properties that were measured in this work. This highlights
the importance of determining the crystallite size when analysing
composite strengthening mechanisms. There still remains a significant difference between the experimental data and the hardness
predicted by the HallPetch effect. This discrepancy in hardness
would relate to other mechanisms such as Orowan strengthening
[34], Thermal mismatch [35], Shear lag [36] and rule of mixture
[37]. Some authors have proposed to couple a number of these
effects together [38,39]. Indeed, the CNTs found in the cracks of
the fracture area seem to conrm that a certain load transfer
occurs from the matrix to the CNTs. This also means that there is
potential to further increase the mechanical properties by focusing
on these other mechanisms.
4. Conclusions
The results presented here show that up to 9 wt% of MW-CNTs
can be incorporated into an Al matrix by planetary milling. Raman
and TEM analysis shows that the MW-CNTs are still present after
milling. No large CNTs agglomerates could be identied by electron
microscopy indicating that the milling process effectively dispersed the nanotubes within the aluminium matrix. The hardness
reaches a maximum at HV20 = 151 for 6 wt% MW-CNTs. As far as
the authors know, this is among the highest wt% of MW-CNTs
and hardness values for pure Al matrix using milling to disperse
the MW-CNTs that can be found in literature. A simple model of
the HallPetch effect accounts for a signicant improvement of
properties observed in our samples. Work is in progress to correlating the hardness values measured here and that of the yield
strength of extruded samples. Initial results with extruded samples
of 3 wt% MW-CNTs show no work hardening, meaning that the Tabor relationship is valid [31]. Some carbon nanotubes are observed
linking the crack surfaces together, indicating a load transfer
mechanism and a good interfacial bonding between the matrix
and the reinforcement material (shear lag effect). The other
strengthening mechanisms, such as Orowan and thermal mismatch need to be further investigated and quantied. This study
shows that there is a large scope to improve the mechanical properties further by optimising other strengthening mechanisms.
Acknowledgements
The authors would like to acknowledge M. Dvorak (TBS) for the
supply of Al powders and Prof. H. Adams (Bayer) for supplying the
MW-CNTs. The authors would also like to thank Prof. A. Weidenkaff (Empa) for access to their XRD and Raman facilities and
H.-B. Mosimann (Empa) and B. von Gunten (Empa) for their assistance in sample preparation and testing. The support of Empas
Electron Microscope Centre was gratefully appreciated.
References
[1] G. Walter, Metal matrix composites, in: G. Ibe (Ed.), Ullmanns Encyclopedia of
Industrial Chemistry, Wiley-VCH, Weinheim, 2003.
[2] S. Iijima, Nature 354 (1991) 5658.
[3] J.N. Coleman, U. Khan, W.J. Blau, Y.K. Gunko, Carbon 44 (2006) 16241652.
[4] E.T. Thostenson, Z.F. Ren, T.W. Chou, Compos. Sci. Technol. 61 (2001) 1899
1912.
[5] S.R. Bakshi, D. Lahiri, A. Agarwal, Int. Mater. Rev. 55 (2010) 4164.
[6] M.F. Yu, O. Lourie, M.J. Dyer, K. Moloni, T.F. Kelly, R.S. Ruoff, Science 287 (2000)
637640.
[7] R.J. Doome, A. Fonseca, J.B. Nagy, New metallic alloys incorporating fullerenes
and carbon nanotubes, in: H. Kuzmany, J. Fink, M. Mehring, S. Roth (Eds.), 12th
C.R. Bradbury et al. / Journal of Alloys and Compounds 585 (2014) 362367
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
367
[22] A.S. Khan, B. Farrokh, L. Takacs, Mater. Sci. Eng., A 489 (2008) 7784.
[23] H.J. Choi, S.W. Lee, J.S. Park, D.H. Bae, Scr. Mater. 59 (2008) 11231126.
[24] J.P. Tessonnier, D. Rosenthal, T.W. Hansen, C. Hess, M.E. Schuster, R. Blume, F.
Girgsdies, N. Pfander, O. Timpe, D.S. Su, R. Schlogl, Carbon 47 (2009) 1779
1798.
[25] L. Kollo, M. Leparoux, C.R. Bradbury, C. Jaggi, E. Carreno-Morelli, M. RodriguezArbaizar, J. Alloys Comp. 489 (2010) 394400.
[26] L. Lutterotti, Nucl. Instr. Meth. B 268 (2010) 334340.
[27] I. Sridhar, K.R. Narayanan, J. Mater. Sci. 44 (2009) 17501756.
[28] S.R. Bakshi, R.G. Batista, A. Agarwal, Composites: Part A 40 (2009) 13111318.
[29] F. Zhou, J. Lee, S. Dallek, E.J. Lavernia, J. Mater. Res. 16 (12) (2001) 34513458.
[30] K. Maung, R.K. Mishra, I. Roy, L.-C. Lai, F.A. Mohamed, J.C. Earthman, J. Mater.
Sci. 46 (2011) 69326940.
[31] D. Tabor, J. Inst. Metals 79 (1951) 118.
[32] M.A. Meyers, K.K. Chawla, Mechanical Metallurgy, Prentice-Hall, Englewood
Cliffs, NJ, 1984.
[33] U. Martin, M. Heilmaier, Adv. Eng. Mater. 6 (2004) 515520.
[34] E. Orowan, Zeitschrift fur Kristallograhie 89 (1934) 327343.
[35] R.J. Arsenault, N. Shi, Mater. Sci. Eng. 81 (1986) 175187.
[36] T.W. Clyne, P.J. Withers, An Introduction to Metal Matrix Composites,
Cambridge University Press, Cambridge, 1995.
[37] F. Tang, I.E. Anderson, T. Gnaupel-Herold, H. Prask, Mater. Sci. Eng., A 383
(2004) 362373.
[38] Z. Zhang, D.L. Chen, Scr. Mater. 54 (2006) 13211326.
[39] X. Zhou, D. Su, C. Wu, L. Liu, J. Nanomater. (2012) 851832.