Microstructures and Mechanical Properties of Al-Base Composite Materials

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Journal of Alloys and Compounds 493 (2010) 453–460

Contents lists available at ScienceDirect

Journal of Alloys and Compounds


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

Microstructures and mechanical properties of Al-base composite materials


reinforced by Al–Cu–Fe particles
G. Laplanche a , A. Joulain a , J. Bonneville a,∗ , R. Schaller b , T. El Kabir a
a
Université de Poitiers, Laboratoire de Physique des Matériaux, UMR CNRS 6630, Bd Marie et Pierre Curie, Bât. SP2MI, 86962 Futuroscope Chasseneuil, France
b
Ecole Polytechnique Fédérale de Lausanne, Institut de Physique de la Matière Condensée, CH 1015 Lausanne, Switzerland

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

Article history: In this study, we produced four composite materials with Al-based matrix reinforced by Al–Cu–Fe par-
Received 13 July 2009 ticles initially of the quasicrystalline (QC) phase. The processing route was a gas-pressure infiltration of
Accepted 20 December 2009 QC particle preforms by molten commercial Al and Al alloys. The resulting composites were investigated
Available online 28 December 2009
by scanning electron microscopy (SEM) working in the energy dispersive spectroscopy (EDS) mode and
by X-ray diffraction (XRD). It is shown that such a synthesis technique leads to the formation of various
Keywords:
phases resulting from specific diffusion processes. Compression tests were performed at constant strain
Composite
rate in the temperature range 290–770 K. The stress–strain curves look similar to those of Al–Cu–Fe
Quasicrystal
Mechanical properties
poly-quasicrystals and show the yield point, the origin of which is however of very different nature.
Composite deformation is recognised to occur through the rupture of a hard phase skeleton and localised
plastic deformation in the matrix.
© 2009 Elsevier B.V. All rights reserved.

1. Introduction which was attributed to strong matrix/reinforcement interfacial


bonding. Because the matrix/particle interfaces were resulting
The continuous requirement of light and high-performing mate- from diffusion processes, further enhancements may be expected
rials for aerospace and automotive industries has led to the using the wetting properties of molten Al at higher temperatures.
development of discontinuous metal matrix composite (MMC) In this study, a gas-pressure infiltration technique was used
materials. A discontinuous MMC usually consists of a light matrix to produce composite materials with Al-based matrix reinforced
reinforced by particles or short fibres of a hard material. The most by Al–Cu–Fe particles. The Al–Cu–Fe particles are initially of the
commonly developed MMCs are Al and/or Mg based-alloy matrix icosahedral QC phase. Scanning electron microscopy, working in
with SiC and/or Al2 O3 reinforcement phases [1]. the energy-dispersive spectroscopy (EDS) mode, and X-ray diffrac-
In this context, quasicrystalline (QC) materials, which exhibit tion (XRD) were used to investigate the chemical composition and
remarkable mechanical properties at low and intermediate tem- the crystallographic structure of the fabricated composites, respec-
peratures such as high hardness and yield stress [2], seem very tively. Compression tests were performed at constant strain rate in
competitive materials compared with SiC and Al2 O3 . The use the temperature range 290–770 K to investigate the temperature
of Al–Cu–Fe quasicrystalline particles to produce MMC therefore dependence of the yield stress. These tests were complemented by
constitutes a promising challenge. A review on the microstruc- some load relaxation (LR) and strain-rate change experiments (SRC)
ture, fabrication and properties of quasicrystalline Al–Cu–Fe alloys for evaluating the strain-rate sensitivity of the stress. Local Vick-
can be found in [3]. In particular, because of the metallic nature ers hardness measurements were performed at room temperature
of Al–Cu–Fe better matrix/particle interface compatibilities and (RT) to obtain a more complete knowledge of the MMC mechanical
recycling facilities are expected in comparison to Al/ceramic com- properties.
posites. A few studies have already been devoted to produce and to
characterise Al/Al–Cu–Fe MMCs. The processing routes were essen- 2. Experimental procedures
tially hot-pressing consolidation methods [4–6], which utilise only
solid phases. The most complete study, reported by Tang et al. 2.1. Materials
[7–10], concluded on high load transfer efficiency for these MMCs,
Four Al-based MMCs were produced using the gas-pressure infiltration tech-
nique [11]. The MMCs were prepared by melting the matrix in a melting chamber,
under vacuum or protective atmospheres, which is afterwards injected under pres-
sure into a backed bed of reinforcements particles. The injection pressure is of the
∗ Corresponding author. Tel.: +33 549 496829; fax: +33 549 496692. order of 30 bar. The reinforcement particles were prepared from Al–Cu–Fe ingot
E-mail address: joel.bonneville@univ-poitiers.fr (J. Bonneville). pieces by ball milling in a planetary crucible. The ingots were basically of the

0925-8388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2009.12.124
454 G. Laplanche et al. / Journal of Alloys and Compounds 493 (2010) 453–460

Table 1
Initial chemical compositions of the matrices (in wt.%) used for producing the composite materials.

Matrix Element

Si Cu Mn Mg Cr Zn Ti + Zr Al

Al3 Mg2 – – – 37.5 – – – 62.5


Al–Cu–Mg <0.7 4.0 0.7 0.7 <0.10 0.25 0.25 Balance
Al–Mg–Si 1 – 0.7 0.8 – – – 97.5

Al–Cu–Fe icosahedral phase, but also contain a small amount of ␭ phase with the between RT and 720 K under argon atmosphere using an Instron testing machine
Al74 Cu5 Fe21 composition embedded in the QC particles [12]. Details concerning QC of the 1114 type. Mechanical tests were performed at a nominal strain-rate ε̇ of
alloy processing can be found elsewhere [12,13]. It was checked by XRD that the 2.10−4 s−1 . The samples were deformed up to fracture. Load relaxation experiments
ball milling process did not produce detectable phase transformation. Note that the and strain-rate change (SRC) tests were performed at 620 K and 720 K to measure the
small amount of initial ␭ phase is not detected by XRD prior to and after particle pro- strain-rate sensitivity of the stress. Both transient tests were carried out at strains
cessing. The final particle size was ranging between 200 ␮m and 500 ␮m. Different beyond the upper yield stress (UYS). LR duration was 30 s and the changes in strain
matrix compositions, listed in Table 1, were selected for the following reasons: rate for SRC were by a factor of 2, 4 or 10. Assuming that plastic kinetics can be
described by a power law:
• an aluminium of commercial purity because it is a simple system, which should
ε̇p = A n , (1)
allow for identification of diffusion processes between matrix and particles during
the synthesis, where ε̇p is the plastic strain-rate,  the applied stress and A a material dependent
• an Al3 Mg2 alloy because it has low melting temperature as compared to pure Al,
parameter, then the apparent stress exponent n can be determined from LR and SRC
which is of particular interest in consideration of the fabrication procedure, where experiments using the relation
the molten matrix is pushed into the backed bed of QC particles,
• an Al–Cu–Mg alloy because it exhibits high mechanical properties and, in partic- ∂ log ε̇p
n= . (2)
ular, high solid solution hardening [14], ∂log
• an Al–Cu–Si alloy because Si has been reported as a stabilising element of the
Because derivative computation of experimental data is always a tedious work, it
Al–Cu–Fe QC phase [15,16], which behaviour may be also of interest when Si is
can be avoided in the case of LR experiment as follows. Starting from the so-called
present in the surrounding matrix.
machine equation

For all composites, the QC particle volume in the backed bed was adjusted to 
ε̇ = ε̇e + ε̇p = + ε̇p , (3)
correspond to a volume fraction of about 50 vol.% of the total composite volume. M
The temperature used for homogenizing the molten matrixes was 970 K. The other ε̇ is the total applied strain rate, ε̇e the elastic strain-rate and M denotes the combined
temperatures used at the various fabrication stages for producing the composites sample-machine elastic modulus, one can write in the case of LR test, where ε̇ = 0,
are given in Table 2, together with the composite designations. After infiltration, the
composites were cooled at rates that are estimated of the order of several tens of 
ε̇p = − . (4)
degrees per second at the onset of cooling. The cooling time to RT is nearly 5 × 103 s. M
A combination of Eqs. (1) and (4) yields after integration
2.2. Composition and phase characterisations    
 1 t
log = log 1+ , (5)
The local composite composition was analysed by EDS of backscattered- o 1−n c
electrons (BSE) using a JEOL 5600LV scanning electron microscope (SEM). Relative
errors on the atomic percentages are estimated of the order of 2%. The crystallo-  o is the stress at the onset of LR and c integration constant. Then, by properly tuning
graphic structure was further characterised by XRD on a D5005 diffractometer using the integration constant c, a plot of log(/ o ) versus log(t) yields exponent n.
CuK␣ radiations.
3. Results
2.3. Mechanical test

2.3.1. Microindentation test 3.1. Composition and phase characterisations


Microindentations were performed at RT using a normalized ShimadzuTM HMV-
2000 Vickers indenter operating at 0.5 N load with a load sustain of 15 s. The imprints In Fig. 1 are presented SEM images obtained in the BSE mode
of the Vickers microindentations were analysed by optical microscopy. For each
for the four composites. Under these conditions, the contrast
phase, the micro-hardness has been determined from an average of at least 15 dif-
ferent indentations. The micro-hardness (H) is calculated according to the relation: corresponds to a chemical contrast, i.e., light elements, such as
aluminium, appear in dark contrast while heavier elements have
P a brighter appearance. Fig. 1 clearly reveals the complexity of the
H = 18, 544 ,
d2 composite materials in terms of both the number and the morphol-
where P is the applied load and d is the mean size value of the diagonals of the imprint ogy of the phases. Whatever the initial composition of the matrix
indentation. Some indentation imprints were also observed by SEM to allow for a at least three phases1 were identified in each composite, but the
more careful examination and to accurately visualise the indented phase. Imprints number of phases can be much higher. The chemical compositions
located at the vicinity of different phases were not considered. corresponding to the main contrasts observed by EDS are reported
2.3.2. Deformation test
in Table 3 for the four composites. After composite processing, the i-
Compression samples were spark-eroded with a parallelepiped shape of dimen- phase was preserved in the AM and AMS composites only. Note that
sions 7 mm × 4 mm × 4 mm for A and AMS composites and 6 mm × 3 mm × 3 mm the dendrite aspect of the i-phase observed in Fig. 1b and d already
for AM and AMC composites. The samples were deformed at temperatures ranging existed in the original QC particles [12]. The Al74 Cu4 Fe22 chemi-
cal composition observed in all composites well matches with the
Table 2 ␭-phase already present in the QC particles [12]. One must notice
Processing parameters used for composite fabrication by the gas-pressure infiltra- that for the AM composite a small amount of Mg is detected in
tion technique [11].
the ␭-phase. New phases are also identified such as, for instance,
Composite Matrix Preform Infiltration a phase with a chemical composition close to Al70 Cu20 Fe10 , which
designation composition preheat (K) temperature (K)

A Al 970 1000
AM Al3 Mg2 820 920 1
At this stage, only chemical compositions are determined, nevertheless for the
AMC Al–Cu–Mg 970 970
sake of clarity they are already referred to as ‘phases’. The phase structures were
AMS Al-Mg-Si 970 970
thereafter confirmed by XRD.
G. Laplanche et al. / Journal of Alloys and Compounds 493 (2010) 453–460 455

Table 3
Chemical composition of various phases identified by EDX analysis in the four composite materials. The (±) signs in the phase compositions designate the composition
deviations obtained between the different composite materials. (*) Some Mg content was found in the ␭ phase that surrounds the QC particles.

Composite materials Al64.5±0.5 Cu22.5±0.5 Fe13 i-phase Al71±2 Cu17±2 Fe10.5±0.5 ␻-phase Al73±1 Cu4.5±0.5 Fe21±1 ␭-phase Matrix (at.%)
√ √
A – Al100−x Cux (X ≤ 2)
√ √
AM – (*) Al3 Mg2
√ √
AMC – Al100−x Cux (X ≤ 2)
√ √ √
AMS Al100−x Cux (X ≤ 3)

is observed in the A, AMC and AMS composites (see Table 3). As an A sample was successively deformed prior to, at and beyond
shown below, this phase corresponds to the ␻ tetragonal phase. the upper yield stress,  UYS (Fig. 4a). The sample surfaces exam-
The morphology and spatial distribution of the various phases ined by SEM after each deformation step are presented in Fig. 4b–d.
depend of the MMC. In the AM and AMS composites the QC parti- Fig. 4b shows that at the end of the first deformation run the rein-
cles are surrounded by ␭- and ␻-phases, respectively. In the AMS forcement particles are crack-free. Small cracks are formed inside
composite, ␭-mono-phase particles are observed, that only concern the hard particles after the second deformation performed up to
small particles. The ␭-phase encircles the ␻-phase in the A com- nearly  UYS . These cracks are roughly aligned parallel to the com-
posite, while no obvious correlation appears between both phases pression axis. With increasing strain, at the end of the third test,
in the AMC composite. As a rule, Cu was detected in all compos- long cracks crossing the hard particles are now present (Fig. 4d).
ite matrixes2 with a content that varies from place to place for a These features are common at all temperatures for all compos-
given composite, but which does not exceed 3 at.% except for com- ites, at least as soon as limited ductility is observed. Note that
posite A. Indeed, composite A contains regions of 17.4 at.% Cu–Al the flow stress recorded for the third deformation test is lower
alloy that shows the lamellar nature of the eutectic mixture, result- than that of the second test. This may be due to some recovery
ing from the simultaneous growth of the ␣-Al and ␪-Al2 Cu phases. arising during the heating of the sample up to the temperature
The latter phase is unambiguously identified by XRD. For the AM deformation test, which takes approximately 2 h. At temperatures
composite, zones with complex compositions such as for instance higher than 520 K, cracks are still observed in the hard phases,
Al49 Cu16 Mg35 are found as well. The intermixing and the complex but the samples do not break anymore entirely except at very
geometry of the various phases do not allow neither for surface, i.e. large strains, i.e. at strains usually larger than 10%. At deforma-
volume, fraction estimates nor grain size measurements. tion temperature higher than 670 K, the cracks crossing the hard
The crystallographic structures of the various phases were phases are more randomly oriented. A high density of cracks is
examined using XRD. Likewise the numerous contrasts observed in located at the interface between matrix and particles. Fractures
EDS, XRD patterns exhibit a drastic number of diffraction peaks. All occur now at an angle of approximately 45◦ from the compression
peaks were not indexed, but for each major composition reported axis.
above the corresponding phase was coherently deduced. To con- To summarise, at all temperatures, cracks are observed at low
cisely illustrate the results, only the AMS composite is presented plastic strain in the hard phases. The crack occurrence is likely
thereafter, which is the more representative fabricated composite responsible for the yield point and the upper yield stress observed
in terms of phase diversity. Fig. 2 shows the XRD pattern corre- on the stress–strain curves (see below Section 3.2.2).
sponding to this composite. The i-Al65 Cu22 Fe13 phase and the Al
face-centred cubic (fcc) phase are clearly identified (see in Fig. 2 3.2.2. Stress–strain curves and strain-rate sensitivity
peaks labelled i and Al, respectively). Additional diffraction peaks Stress–strain curves obtained under identical deformation con-
are observed as well, which are identified as corresponding to the ditions, T = 670 K and ε̇ = 2 × 10−4 s−1 , are shown in Fig. 5 for the
␭-Al13 Fe4 monoclinic phase [17] and to the ␻-Al7 Cu2 Fe tetragonal four composites. The curves exhibit similar shape, but significantly
phase [18]. One can reasonably assume that the Al74 Cu4 Fe22 com- differ in stress amplitude. The elastic stage is followed by a plastic
position determined by EDS corresponds to the ␭-phase, where stage of rapidly decreasing strain hardening, ensues a yield point
Cu would substitute to Al atoms, i.e. (Al,Cu)13 Fe4 [17], while the and a plastic stage of strain softening. The AMC composite, which
Al70 Cu19 Fe11 composition can be attributed to the ␻-phase. Note shows the sharper yield point, is the only composite that re-hardens
that the ␭-phase, which was not previously detected by XRD in the after the softening stage.
initial QC particles, is now unambiguously identified in the four The stress–strain curves of the four composites have similar
composites. The four identified phases are visible in Fig. 1d: the pri- characteristics from RT up to 720 K. Typical stress–strain curves
mary i-phase surrounded by the ␻-phase, regions corresponding to are given in Fig. 3 for composite A. The yield stress  y , defined as
the ␭-phase and to the fcc matrix. the first deviation from elastic linear behaviour, is used as the rep-
resentative stress for plastic yielding.3 This excludes temperatures
below 470 K at which the samples break within the elastic stage.
3.2. Mechanical properties
Fig. 6 shows the temperature dependence of  y associated with the
four composite materials. It is observed that A and AMC composites
3.2.1. Fracture features
behave very similarly. Both decrease from about 430 MPa ± 40 MPa
When tested at RT, compression samples extracted from the
at 520 K to 200 MPa ± 55 MPa at 670 K. For the AM composite,  y
four composite materials all break before plastic yielding (see, for
starts higher, but decreases with temperature more rapidly than
instance, Fig. 3 for composite A). At this temperature, the samples
for the other composites. The AMS composite shows the lowest  y
either break off or exhibit surface cracks of millimetre sizes paral-
values, except at T = 670 K.
lel to the compression axis. Limited ductility starts to be observed
at 520 K, where a yield point (YP) is recorded. At this temperature,

3
The upper yield stress is commonly used as the critical stress for characterising
the flow stress behaviour of material exhibiting a yield point. However, because
2
The low Mg and Si contents initially present in the Al–Mg–Si matrix were not in this study upper yield stresses are accompanied by severe particle cracking, in
accurately detected by EDS. particular at low temperatures,  y is more adequate for characterising plastic flow.
456 G. Laplanche et al. / Journal of Alloys and Compounds 493 (2010) 453–460

Fig. 2. XRD patterns of the AMS composite evidencing the fcc-Al, i-Al65 Cu22 Fe13 ,
␻-Al7 Cu2 Fe and ␭ − Al13 Fe4 phases.

The stress exponent n was determined all along stress–strain


curves by SRC and LR tests (see for instance Fig. 4 at T = 620 K and
720 K) according to Eqs. (2) and (5), respectively. The tests were per-
formed on the A and AMC composites only. At temperatures lower
than 520 K, the sample brittleness does not allow us to carry out
SRC and LR tests. The n values are presented in Fig. 7 as a function
of temperature. It must be noticed that, at a given temperature, the
stress levels at which n values are evaluated can vary by a factor
of more than 2. n values of 5 ± 1 are obtained, whatever stress and
temperature for the AMC composite. On the contrary, n is temper-
ature dependent for the A composite and starts high, n = 9 ± 1, at RT
to achieve at 720 K a value similar to that of the AMC composite,
n = 5 ± 1.

3.2.3. Micro-hardness H
The H values measured for the i-, ␻- and ␭-phases are presented
in Table 4 where they are compared to the results obtained by
Köster et al. [19], Kang and Dubois [20] and by Giacometti et al. [21].
For the i-phase in the AM and AMS composites H = 8.6 ± 0.6 GPa,
which is little lower than that of the ␭-phase in the A and AMC
composites for which H = 10 ± 0.5 GPa. In comparison, H is found
drastically lower for the ␻-phase for which H = 4.2 ± 0.5 GPa in the
A composite and 5 ± 0.5 GPa in the AMC composite. The micro-
hardness was also measured for the matrix materials of A and AMC
composites. For both a value of H = 0.12 ± 0.02 GPa is measured, sug-
gesting that the composition difference between the two matrix
materials does not play a significant role in hardness.

Fig. 1. SEM micrographs in backscattering mode of the composite microstructures.


(a) A, (b) AM, (c) AMC, and (d) AMS. Light elements, such as aluminium, appear in
dark contrast while heavier elements have a brighter appearance. The letters ␻ and
i designate the tetragonal phase and icosahedral phase, respectively.

Fig. 3. Stress–strain curves of A composite as a function of temperature. LRE and


SRC indicate load relaxation experiments and strain-rate changes, respectively.
G. Laplanche et al. / Journal of Alloys and Compounds 493 (2010) 453–460 457

Fig. 5. Stress–strain curves of the four composites A, AM, AMC and AMS at 670 K
and ε̇ = 2 × 10−4 s−1 The curves exhibit similar shape: the elastic stage is followed
by a plastic stage of rapidly decreasing strain hardening, ensues a yield point and a
plastic stage of strain softening.

Fig. 6. Yield stress,  y , of the four composites as a function of temperature.  y is


defined as the first deviation from elastic linear behaviour.

Fig. 4. Stress–strain curves and related microstructures for composite A at T = 520 K.


An A sample was successively deformed prior to, at and beyond the upper yield
stress,  UYS (a). The sample surfaces examined by SEM after each deformation step
are presented in (b) (prior to the upper yield stress), (c) (at the upper yield stress)
and (d) (beyond the upper yield stress).

Fig. 7. Temperature dependence of the stress exponent n measured by SRC and LR


tests for A and ACM composites. It must be noticed that at a given temperature the
corresponding stress level may vary by a factor of 2.
458 G. Laplanche et al. / Journal of Alloys and Compounds 493 (2010) 453–460

Table 4 ␻-phase particles surrounded by ␭-phase, the QC particles must


Hardness values reported in the literature and measured in the present study by
transform by emitting Cu from a surface layer into the matrix,
Vickers indentations at room temperature.
while concomitantly Al diffuses from this surface layer towards the
Load (N) H (GPa) Reference inner part of the particles. This also causes Cu enrichment of the
i-Phase Al matrix, leading to the formation the ␪-Al2 Cu phase. According
0.25 9.95 ± 1.00 Köster et al. [19] to SEM observations, the latter phase is produced by solidification
0.5 7.85 ± 0.45 Giacometti et al. [21] from a liquid matrix (see Fig. 1a). In the AMC composite, the ␭-
0.5 8.5 ± 0.5 Present work—AM composite
phase and ␻-phase appear as whole particles within the matrix,
0.5 8.8 ± 0.5 Present work—AMS composite
2 7.15 ± 0.45 Kang et al. [20] which may result from the growth of the ␭-phase already present
in the QC particles and from a phase transformation of the QC par-
␭-Phase
0.25 10.70 ± 0.95 Köster et al. [19]
ticles into ␻ particles, respectively. The differences between the
0.5 10.0 ± 0.5 Present work—A composite A and the AMC composites may arise from the lower infiltration
0.5 10.0 ± 0.5 Present work—ACM composite temperature and the matrix composition [23].
␻-Phase For the AM and AMS composite materials, particles with the
0.25 9.35 ± 0.80 Köster et al. [19] i-phase were preserved. The infiltration and preform preheat tem-
0.5 4.2 ± 0.5 Present work—A composite peratures were the lowest for the AM composite, which certainly
0.5 5.0 ± 0.5 Present work—ACM composite favoured the stability of the i-phase. Indeed, it has been reported
that, during HIP processing, the i-phase was stable when the pro-
cessing temperature is not sufficiently high [4].5 Here, the ␭-phase
4. Discussion
is observed around the QC particles suggesting that, in presence
of a high Mg content and an at.% of Al slightly higher in the QC
4.1. Phase transformations
particles (64 at.%) than in the injected matrix (60 at.%), Cu and Al
both migrate from the QC particle surface into the AM matrix.
EDS and XRD both reveal that a variety of phases are produced
The transformation process leads again to a Cu-rich phase, i.e.
using the gas-pressure infiltration technique. It is not easy to make
Al49 Cu16 Mg35 , at some regions in the Al3 Mg2 matrix. Fine exami-
clear how the various phases were formed from the initial Al–Cu–Fe
nation of XRD patterns has not allowed for the identification of the
QC particles and matrix compositions, but general trends can be
crystallographic structure associated with the Al49 Cu16 Mg35 com-
proposed. First of all, it must be emphasised that:
position, even considering the possibility of an icosahedral phase
[24]. In the AMS, the QC particles have an ␻-phase shell that result
• Cu diffusion out of the QC particles is obvious from the present
from Al diffusion from the matrix into the particles. The i–␻-phase
experimental results, since this element is present everywhere transformation is not complete, like for the A composite, what can
in the composites. Of course, this is only experimentally verified be ascribed to the lower infiltration temperature and, again, to
for the A, AM and AMS composites, since for the AMC composite the difference in matrix chemical composition, in particular to Si
Cu is already present in the matrix. [25].
• Fe diffusion is not evidenced, in agreement with its low solubility
To summarise, all phase transformations are coherently
and diffusivity [3,15]. explained by (1) Cu depletion of the QC particles6 (2) almost con-
• XRD suggests that the volume fraction of the ␭-phase is higher in
stant Fe content in QC particles and (3) Al fluxes from the QC
all composites than in the initial QC particles. particles to the matrix and/or vice versa. As specified above, phase
• The QC particles are not expected to melt4 at the infiltration
diagrams do not exist for the present production process, but it is
temperatures of the molten matrixes, indicating that phase trans- satisfactory to note that such an explanation would be in agreement
formations take place by solid-state reaction with the initial QC with the formation and the coexistence of ␭ + i, ␻ + i, ␭ + ␻ + i and
particles. ␭ + ␻ phases according to the isothermal ternary phase diagram at
T = 950 K proposed in [26].
Taking into account the above considerations, the reactions
occurring during the fabrication process may be described by the
4.2. Mechanical properties
following scenarii.
For the A and AMC composites the QC particles are completely
The micro-hardness (H) values of the various Al–Cu–Fe phases
transformed and are not present in the final material. In the A
respect the sequence reported by Köster et al. [19]. However, as a
composite the QC particles are transformed into two-phase par-
rule, our values are lower than those of [19], which is particularly
ticles composed of a ␻-phase core surrounded by a surface layer
true for the ␻-phase (see Table 4). While the small differences for
of the ␭-phase. The formation of the ␭-phase from the i-phase
the i-and ␭-phases can be here attributed to the use of a higher load
particle requires diffusion of both Cu and Al out of the particles,
[27], there is no clear explanation to account for the factor of about
while the phase transformation from the i-phase to the ␻-phase
2 recorded for the ␻-phase. More global measurements, that do
only requires Al diffusion into the QC particles. Such a latter phase
not allow for extracting separate H value for each phase, were per-
transformation has already been reported in the literature dur-
formed by [4] and [28] on Al MMCs. They indicate that H is higher for
ing Al/QC-Al–Cu–Fe MMCs processing using high isostatic pressure
the i-phase reinforcement particles than for the ␻-phase reinforce-
(HIP) technique at 870 K [4,9,22]. At the infiltration temperature of
ment particles, which support that H(i-phase) > H(␻-phase). More
the molten Al matrix, which is the highest used infiltration temper-
recent measurements performed as a function of temperature on
ature, T = 1000 K, diffusion may arise from both matrix states, i.e.
molten or solid. The phase transformation from the i-phase to the
␻-phase may take place, likewise in the HIP technique, by solid-
state diffusion of Al into the QC particles. For accounting for the 5
In all these studies, the phase transformation temperature is not defined pre-
cisely. It is reported that the i-phase is stable for a processing temperature of 673 K,
while it transforms into the ␻-phase at 873 K.
6
QC particles are used here to designate the original phase structure of the
4
Equilibrium phase diagrams do not exist for the present molten and solid reinforcement particles, which of course evolves during processing when phase
matrixes in contact with the Al–Cu–Fe QC particles. transformation takes place.
G. Laplanche et al. / Journal of Alloys and Compounds 493 (2010) 453–460 459

␻-mono-phase samples give at RT H = 7 ± 1 GPa and a brittle-to- compression tests at ε̇ ≈ 6 × 10−4 s−1 [37] and above 750 K for the
ductile transition temperature at nearly 750 K [29]. ␻-Al–Cu–Fe phase for indentation tests [29]. Above 523 K, all  y
It seems therefore reasonable to assume that, all over the inves- rapidly decrease with increasing temperatures. This behaviour is
tigated temperature range (290–720 K), all the crystalline phases ascribed to the matrix material plasticity, because the hard phases
resulting from the phase transformation of the QC particles have are invariably brittle. As above-mentioned, dislocation climb is
high hardness, suggesting a brittle behaviour. The brittle behaviour certainly involved in matrix plastic deformation, but due to the
of the various Al–Cu–Fe phases is reflected by the numerous cracks complex structures of the present composites it is very difficult to
appearing during compression testing. It is also clearly supported establish the respective weight of various possible contributions to
by the interrupted deformation tests and systematic sample sur- their total flow strength.
face observations that show particle cracking at the onset of plastic
deformation. This leads us to interpret the yield point as result-
5. Conclusion
ing from the breaking of a continuous hard phase skeleton. That
is, from the mechanical point of view, the samples can be con-
Four MMCs with Al-based matrix reinforced by Al–Cu–Fe par-
sidered as two-phases materials composed by a hard continuous
ticles initially of the icosahedral phase were produced using a
phase, which would be produced by hard phase percolation during
gas-pressure infiltration technique, which consists to inject under
the fabrication process, embedded in a softer phase corresponding
pressure the matrix in the liquid state into a preform of reinforce-
to the matrix. At low temperatures, i.e. below nearly 573 K both
ment particles. This synthesis procedure leads to rather complex
phases are hard to deform plastically, so that the samples fracture
composite materials containing a variety of phases that are in the
by exhibiting large cracks running approximately parallel to the
vicinity of the icosahedral phase in the Al–Cu–Fe phase diagram.
compression axis. The samples can be considered as monolithic
As the rule, the new produced phases require Cu diffusion from
not plastically deformable samples. With increasing temperatures,
the QC particles into the matrix, Al diffusion from the QC parti-
the matrixes become more ductile and the samples show limited
cles into the matrix and/or vice versa, while the diffusivity of Fe
plastic behaviour. Matrix plastic deformation induces high shear
seems to be very low. Indentation experiments indicate that the
stresses at the matrix–particle interfaces, causing the break of the
dominant Al–Cu–Fe phases have high hardness, suggesting brit-
hard phase skeleton. The rupture now occurs at an angle of approx-
tle behaviour. Constant strain-rate compression tests performed in
imately 45◦ about the compression axis, which corresponds to the
the temperature range 290–770 K confirm the brittle behaviour of
highest resolved shear stress component. Rupture of the hard phase
the hard phases that fracture at the onset of plastic deformation.
skeleton is revealed on the stress–strain curves by the rapid stress
It is proposed that the hard phases form a continuous hard phase
decrease after the yield point, but in the contrary to low temper-
skeleton resulting from the synthesis process. Therefore, composite
ature deformation the samples do not break entirely. Fractures
deformation requires the deformation of this hard skeleton, which
of the hard skeletons most probably occur at the weaker parti-
explains their very early and systematic fractures. Above approx-
cle bonds. For preserving the sample wholeness, the multiple hard
imately 520 K, breaking of the hard phase skeleton do not lead to
phase breakages must be spatially correlated and localised in spe-
sample rupture anymore, because matrix plasticity can now blunt
cific sample volumes where the matrix accommodates the imposed
the hard phase cracks and accommodate some plastic strain. In
plastic shear-strain. Indeed, it is remarkable that during LR experi-
this temperature regime, matrix plasticity is ascribed to dynamic
ments performed beyond the yield point the load decrease is nicely
recovery of the dislocation microstructures by dislocation climb
fitted with Eq. (5) and provides, for the two investigated com-
processes.
posites, strain-rate sensitivity exponents in good agreement with
published values for Al-MMCs [30–32]. It is also noteworthy that LR
and SRC experiments lead to similar n values. These results must be Acknowledgments
cautiously interpreted due to the concomitant hard phase fractures.
They nevertheless suggest that, beyond the upper yield stress, the Thanks are due to V. Gauthier-Brunet and S. Dubois for valu-
matrix materials only accommodate plasticity. At present, the ele- able comments on the manuscript. Région ‘Poitou-Charentes’ is
mentary deformation mechanisms are not clearly identified but, at gratefully acknowledged for financial support.
least at T = 720 K, the n value of approximately 5 suggests a defor-
mation process that involves dislocation climb [32]. A value of n ∼ 5 References
is currently reported in the literature for Al [33] and Al alloys [34].
The  y stresses result therefore from a complex interplay between [1] D.J. Lloyd, Int. Mater. Rev. 39 (1994) 1–23.
[2] J. Bonneville, D. Caillard, P. Guyot, in: J.P. Hirth (Ed.), Dislocations in Solids, vol.
hard phase strength and matrix plasticity that also include inter- 14, 2008, p. 254.
face cohesion and grain size dependence. All MMCs exhibit brittle [3] E. Huttunen-Saarivirta, J. Alloys Compd. 363 (2004) 150–174.
behaviour below 523 K. Assuming a continuous hard phase skeleton [4] A.P. Tsaï, K. Aoki, I. Akihisa, T. Masumoto, J. Mater. Res. 8 (1993) 5–7.
[5] F. Schürack, J. Eckert, L. Schultz, Philos. Mag. 83 (2003) 1287–1305.
allows us a rough estimate of the stress supported at this temper- [6] F. Tang, PhD Thesis, Iowa State University, 2004.
ature by the soft and hard phases. In the present case, the two [7] F. Tang, I.E. Anderson, S.B. Biner, Mater. Sci. Eng. A 363 (2003) 20–29.
phases must accommodate the same elastic strain, so that at  y [8] F. Tang, I.E. Anderson, T. Gnaüpel-Herold, H. Prask, Mater. Sci. Eng. A 383 (2004)
362–373.
the stress on each phase is in the ratio of their Young’s modulus. [9] F. Tang, T. Gnaüpel-Herold, H. Prask, I.E. Anderson, Mater. Sci. Eng. A 399 (2005)
That is in the frame of iso-strain model using the elastic coefficients 99–106.
of [35,36], with an initial 50% volume fraction of hard phase and [10] F. Tang, H. Meeks, J.E. Spowart, T. Gnaüpel-Herold, H. Prask, I.E. Anderson,
Mater. Sci. Eng. A 386 (2004) 194–204.
a mean value of  y ≈ 440 MPa at T = 523 K, the calculated applied
[11] E. Carreno-Morelli, T. Cutard, R. Schaller, C. Bonjour, Mater. Sci. Eng. A 251
stress is about 600 MPa and 280 MPa on the hard phases and on (1998) 48–57.
the matrix materials, respectively. Although the present estimates [12] E. Giacometti, J. Fikar, N. Baluc, J. Bonneville, Philos. Mag. Lett. 82 (2002)
183–189.
are very crude, they nevertheless indicate that the applied stress
[13] E. Giacometti, PhD Thesis, Lausanne-Suisse, EPFL, 1999.
on the hard phases is much larger than  y , which supports their [14] M. Van Lancker, Metallurgy of Aluminium Alloys, Chapman and Hall, London,
early rupture. It must be noticed that the brittle-to-ductile tran- 1967.
sition temperatures of the hard phases are expected to be higher [15] K.-B. Kim, S.-H. Kim, W.T. Kim, D.H. Kim, K.-T. Hong, Mater. Sci. Eng. A 304
(2001) 822–829.
than the highest temperature used in this study. For instance, the [16] B.S. Murty, P. Barua, V. Srivinas, F. Schürack, J. Eckert, J. Non-Cryst. Sol. 334
transition temperature is above 800 K for the i-Al–Cu–Fe phase for (2004) 44–47.
460 G. Laplanche et al. / Journal of Alloys and Compounds 493 (2010) 453–460

[17] C. Freiburg, B. Grushko, J. Alloys Compd. (1994) 149–152. [28] S.D. Khaloshkin, V.V. Tcherdyntsev, A.I. Laptev, A.A. Stepashkin, E.A. Afonina,
[18] L.M. Brown, R.M. Ham, in: R.W. Cahn, R.B. Nicholson (Eds.), Strengthening A.L. Pomadchik, V.I. Bugakov, J. Mater. Sci. 39 (2004) 5399–5402.
Methods in Crystals, Elsevier, Amsterdam, 1971, p. 12. [29] G. Laplanche, A. Joulain, V. Gauthier-Brunet, J. Bonneville, S. Dubois, submitted
[19] U. Köster, W. Liu, H. Liebertz, M. Michel, J. Non-Cryst. Solids 153–154 (1993) for publication.
446–452. [30] A. Dlouhy, N. Merk, G. Eggeler, Acta Metall. Mater. 41 (1993) 3245–3526.
[20] S.S. Kang, J.M. Dubois, Philos. Mag. A 66 (1992) 151–163. [31] G. Kausträter, B. Skrotzki, G. Eggeler, Mater. Sci. Eng. A 319–321 (2001)
[21] E. Giacometti, N. Baluc, J. Bonneville, J. Rabier, Scripta Mater. 41 (1999) 989–994. 716–721.
[22] T. El Kabir, A. Joulain, V. Gauthier, J.M. Dubois, J. Bonneville, J. Mater. Res. 23 [32] S. Spigarelli, E. Evangelista, E. Cerri, T.G. Langdon, Mater. Sci. Eng. A 319–321
(2008) 904–910. (2001) 721–725.
[23] J. Gui, J.Y. Wang, R. Wang, D. Wang, J. Liu, F. Chan, J. Mater. Res. 16 (2001) [33] O.A. Ruano, O.D. Sherby, Revue de Physique Appliquée 23 (1988) 625–
1027–1046. 637.
[24] W.A. Cassada, Y. Shen, S.J. Poon, G.J. Shiflet, Phys. Rev. B 34 (1986) 7413–7416. [34] R. Kaibyshev, O. Sitdikov, I. Mazurina, D.R. Lesuer, Mater. Sci. Eng. A 334 (2002)
[25] A.P. Tsaï, A. Inoue, T. Masumoto, J. Mater. Sci. Lett. 8 (1989) 470–472. 104–113.
[26] D. Gratias, Y. Calvayrac, J. Devaud-Rzepski, F. Faudot H.M., A. Quivy, J. Non-Cryst. [35] G. Simmons, H. Wang, Handbook of Single Crystal Elastic Constants and Calcu-
Sol. 153–154 (1993) 482–488. lated properties, 1971.
[27] M. Wollgarten, H. Saka, New Horizons in Quasicrystals: Research and Applica- [36] K. Tanaka, Y. Mitarai, M. Koiwa, Philos. Mag. A 73 (1996) 1715–1723.
tions, vol. 620, World Sci. Publishing Ltd., 1996. [37] E. Giacometti, N. Baluc, J. Bonneville, Philos. Mag. Lett. 79 (1998) 1–7.

You might also like