Materials Science & Engineering A 661 (2016) 115–125

Contents lists available at ScienceDirect

Materials Science & Engineering A

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

Strain-controlled low cycle fatigue properties of a rare-earth

containing ME20 magnesium alloy
F.A. Mirza a,n, K. Wang a, S.D. Bhole a, J. Friedman a, D.L. Chen a, D.R. Ni b, B.L. Xiao b,
Z.Y. Ma b,n
a
Department of Mechanical and Industrial Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada
b
Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, China

art ic l e i nf o a b s t r a c t

Article history: The present study was aimed to evaluate the strain-controlled cyclic deformation characteristics and low
Received 10 January 2016 cycle fatigue (LCF) life of a low ( ! 0.3 wt%) Ce-containing ME20-H112 magnesium alloy. The alloy con-
Received in revised form tained equiaxed grains with ellipsoidal particles containing Mg and Ce (Mg12Ce), and exhibited a rela-
3 March 2016
tively weak basal texture. Unlike the high rare earth (RE)-containing magnesium alloy, the ME20M-H112
Accepted 3 March 2016
Available online 5 March 2016
alloy exhibited asymmetrical hysteresis loops somewhat similar to the RE-free extruded Mg alloys due to
the presence of twinning-detwinning activities during cyclic deformation. While cyclic stabilization was
Keywords: barely achieved even at the lower strain amplitudes, cyclic softening was the predominant characteristics
Magnesium alloy at most strain amplitudes. The ME20M-H112 alloy showed basically an equivalent fatigue life to that of
Rare-earth elements
the RE-free extruded Mg alloys, which could be described by the Coffin-Manson law and Basquin's
Cyclic deformation
equation. Fatigue crack was observed to initiate from the near-surface imperfections, and in contrast to
Twinning-detwinning
Texture weakening the typical fatigue striations, the present alloy showed some shallow dimples along with some fractions
of quasi-cleavage features in the crack propagation area.
& 2016 Elsevier B.V. All rights reserved.

1. Introduction There are currently intensive studies in the development of

Due to the increasing global energy demand and great cogni-
zance of human-caused pollution such as CO2 emissions in recent
years, lightweighting has turned into a crucial approach in the au-
tomotive and aerospace industries [1–10]. There has been major
interest and great deal of research activity for finding alternatives to
reduce the fuel consumption of passenger vehicles [10,11]. Being
the ultra-lightweight structural metallic material, magnesium (Mg)
alloys have been increasingly applied in the auto industry for ve-
hicles weight reduction [12]. The application of Mg alloys as a
structural material in automotive and aerospace industry would
require the evaluation of fatigue and cyclic deformation character-
istics, since structural components would unavoidably experience
dynamic loading in service, which leads to the occurrence of fatigue
failure [13–15]. Thus, to guarantee the structural integrity and
durability of such engineering components it is essential to un-
derstand the fatigue and cyclic deformation behavior of Mg alloys.
As there is a growing application of Mg alloys in powertrain, chassis
and body areas, developing wrought Mg components with im-
proved mechanical properties became a prominent necessity [16].
n
Corresponding authors.
E-mail addresses: f4mirza@ryerson.ca (F.A. Mirza), zyma@imr.ac.cn (Z.Y. Ma).
Mg alloys with high strength, good corrosion resistance and su-
perior formability for structural applications [6,12,14,16–28].
Nevertheless, several restraints are faced in the application of Mg
alloys, including limited ductility, tension-compression yield
asymmetry, pronounced directional anisotropy arising from the
presence of strong crystallographic texture related to their hex-
agonal close-packed (HCP) structure due to a limited number of
slip systems that could activated during extrusion or rolling
processes [29–31]. For the vehicle components subjected to dy-
namic cyclic loading, this mechanical anisotropy along with the
tension-compression yield asymmetry could lead to irreversi-
bility of cyclic deformation, which may have a serious influence
on the material performance [32]. Both the tension-compression
yield asymmetry and formability at room-temperature could be
effectively improved by the addition of alloying elements, espe-
cially rare-earth (RE) elements due to their affinity to induce
texture randomization during hot deformation processes (e.g.,
extrusion or rolling), which results in the decrease of texture
intensities and the activation of basal slip [29,33–38]. Though
these alterations in the tension-compression yield asymmetry
and mechanical anisotropy due to RE elements additions are re-
cently reported, the potential advantage of such wrought RE-Mg
alloys as structural components under dynamic cyclic loading
conditions has not yet been well appreciated.

http://dx.doi.org/10.1016/j.msea.2016.03.024
0921-5093/& 2016 Elsevier B.V. All rights reserved.
116 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125
Recently several studies on the fatigue of RE-Mg alloys have
been reported in the literature [29,39–47]. For instance, Yang et al.
[29] conducted very high cycle fatigue tests of a high RE-con-
taining extruded Mg-12Gd-3Y-0.5Zr alloy, and observed much
relieved tension-compression yield asymmetry and enhanced fa-
tigue failure resistance in comparison with RE-free AZ31 alloy. The
fatigue strength of extruded Mg-10Gd-1Nd and Mg-10Gd alloys in
the form of S-N curves was also evaluated via stress-controlled
high cycle fatigue tests [47]. A few strain-controlled low cycle fa-
tigue tests on the RE-Mg alloys have been performed as well [48–
55]. Although the high performance of the Mg–10Gd-3Y-0.5Zr
(GW103K) alloy was achieved [48–50], it was quite expensive due
to the addition of a fairly high amount (totally ! 13 wt%) of RE
elements. Since the cost is one of the major considerations in the
automotive applications, recently developed rolled ME20M-H112
alloy with a low amount of RE element ( ! 0.3 wt% Ce) would be a
promising candidate. To the authors’ knowledge, no studies have
been conducted on the cyclic deformation behavior of ME20M-
H112 Mg alloy under strain control in the open literature. It is
unclear to what extent the ME20M-H112 alloy would exhibit the
tension-compression asymmetry, whether cyclic hardening or
softening would occur, and what are the cyclic stress response and
fatigue life. Therefore, the objective of the present study was
aimed at identifying cyclic deformation behavior of a low RE-
containing rolled ME20M-H112 alloy, and evaluating the fatigue
life under varying strain amplitudes.
2. Material and experimental procedure
The material under investigation in the present study is a
newly developed Mg alloy ME20 sheet, which was processed by
hot rolling and H112 tempered, whose nominal composition is
provided in Table 1. Microstructural evolution was examined in
the samples by using an optical microscope (OM) equipped with
Clemex quantitative image analysis software, and scanning elec-
tron microscope (SEM) JSM-6380LV equipped with Oxford energy
dispersive X-ray spectroscopy (EDS) system. Standard metallo-
graphic sample preparation techniques were used to grind and
polish sample surfaces, and etching was done with an acetic picral
solution containing 4.2-g picric acid, 10-ml acetic acid, 10-ml H2O,
and 70-ml ethanol. Textures were obtained using a PANalytical
X-ray diffractometer (XRD) with Cu Kα radiation at 45 kV and
40 mA in a back reflection mode by measuring partial pole figures
(i.e., ranging between Ψ ¼0° and 75°). Texture data were after-
wards analyzed based on MTEX software [56]. It should be noted
that the defocusing stemming from the rotation of the XRD sample
holder was corrected using the experimental data obtained from
Mg powder diffraction. The sample was positioned in the machine
with the rolling direction parallel to the x-direction and defocusing
arising from the rotation of XRD sample holder was corrected
using experimentally determined data obtained from the diffrac-
tion of Mg powders received from Magnesium Elektron. Sub-sized
tensile and fatigue samples were machined with the loading axis
parallel to the rolling direction (RD) in accordance with ASTM: E8
standard. The samples had a gauge length of 25 mm (or a parallel
length of 32 mm), thickness of ! 7 mm, and a width of !6 mm.
The gauge section of tensile and fatigue samples was ground
progressively along the loading direction with emery papers up to
Table 1
Chemical composition of the rolled ME20M-H112 Mg alloy.
Element Al Zn Mn Ce
Content (wt%) o 0.2 o0.3 1.69–1.81 0.29–0.33
a grit number of 600 to remove the machining marks and to
achieve a smooth surface.
Tensile tests were performed in accordance with the ASTM: E8
standard by means of a computerized United tensile testing ma-
chine with a gauge length of 25 mm at a strain rate of 1 # 10 $ 3 s $ 1
at room temperature. Strain-controlled “pull-push” type fatigue
tests (in accordance with the ASTM: E606 standard) were con-
ducted in air at room temperature with a 25 mm extensometer
using a computerized Instron 8801 fatigue testing system that was
controlled by a Fast Track Low Cycle Fatigue (LCF) program. The
cyclic deformation test conditions consisted of a zero mean strain
(i.e., a strain ratio of Rε ¼ $ 1, completely reversed strain cycle) and a
constant strain rate of 1 # 10 $ 2 s $ 1 with triangular loading wave-
form, as noted in the ASTM: E606 standard for continuous cyclic
tests and generally for strain-rate sensitive materials, since the tri-
angular waveform results in a constant strain rate in the course of
cycling. Low cycle fatigue tests were performed at total strain am-
plitudes of 0.2%, 0.4%, 0.6%, 0.8%, 1.0%, and 1.2% (at least two samples
were tested at each level of the strain amplitudes). At lower strain
amplitudes (e.g., 0.1% and 0.2%), strain-controlled tests were sus-
tained for 10,000 cycles before being converted to load control, with
a sine cyclic waveform at a frequency of 50 Hz. Once tests were
completed, SEM was used to examine the fracture surfaces of fati-
gued samples, aiming to identify the various features involving fa-
tigue initiation and propagation mechanisms. In addition, a special
interest was given to the near fracture surface areas of the fatigued
samples, which were cut, mounted, ground, polished and etched to
examine the eventual appearance of residual twins.
3. Results and discussion
3.1. Initial microstructure and texture
Fig. 1 shows a typical optical micrograph and a backscattered
electron (SEM) image of a rolled ME20M-H112 alloy, where the
arrow indicates the rolling direction (RD). The microstructure of
the alloy was composed of equiaxed grains with an average grain
size ! 12 mm due to the dynamic recrystallization occurred in the
rolling process [48]. The grain size was fairly small in comparison
with the common extruded Mg alloys, such as AZ31 and AM30
[14,16–19]. This was due to the role of added RE elements [48–50]
and the grains can be prohibited from coarsening during the hot
deformation process as cerium (Ce) can form some dispersed and
thermally stable particles [57]. It was reported that a uniform re-
crystallized grain structure was observed with a grain size de-
creasing from ! 13 to ! 10 mm as Ce content increased from 0% to
1% for rolled and rolled/annealed ME alloys [58,59]. A similar type
of microstructure of as-cast ME alloys has been reported in refs.
[60–62]. The microstructure of the ME20 sample also revealed a
few dispersed, fine, ellipsoidal particles as seen from Fig. 1(b)
(indicated using arrow). An EDS line scan was performed as shown
in Fig. 2(a) and (b), which confirmed the presence of RE particle
(Ce) along with Mn. The very fine solid particles were identified as
pure Mn, where Li et al. [61] also reported the existence of similar
ellipsoidal particles containing Mg and Ce (Mg12Ce). Similar find-
ings have been reported for ME20 alloys in Refs. [62–64].
Fig.3 shows the crystallographic textures (basal (0001),
prismatic (1010¯ ), and pyramidal (1011 ¯ ) pole figures) of rolled
Si Fe Cu Ni Be Mg
o 0.10 o 0.05 o0.05 o 0.007 o 0.01 Balance
 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125
Fig. 1. Microstructure of ME20M-H112 alloy, (a) OM image and (b) SEM (backscattered electron) image, where RD stands for the rolling direction.
Fig. 2. SEM backscattered electron images indicating an EDS line scan position and the corresponding EDS line scan results of a ME20M-H112 alloy.
ME20M-H112 alloy evaluated using MTEX software, where RD
indicates the rolling direction, TD stands for transverse direction
and ND indicates the normal direction. A relatively weak con-
centration (with a maximum intensity of 3.8 multiples of random
distribution (MRD)) of basal (0001) pole mainly displays some
splitting towards the RD and TD, along with prismatic (1010¯ ) and
pyramidal (1011¯ ) poles towards the RD, in comparison with the
extruded AM30 [65,66] and rolled AZ31 [67]. This indicates that
the majority of hcp unit cells in the undeformed ME20 have their
c-axes oriented perpendicular to the rolling direction. The pre-
sence of such a weaker texture, in comparison with the extruded
AM30 [65,66] and rolled AZ31 [67], was another benefit of the RE
elements added into Mg alloys. As also reported by Stanford and
Barnett [68], a similar type of weaker texture was observed for an
extruded ME10 alloy prior to tensile deformation and revealed a
strong orientation peak which was termed as the RE texture
component [69]. Huppmann et al. [59] have also discovered two
additional RE texture components in an extruded ME21 Mg alloy
that were parallel to the extrusion direction: 1122¯ and 2021¯ .
The RE extrusion texture was associated with the presence of Ce in
dynamic recrystallization (DRX) nucleation at shear bands of de-
formed grains [69], which leads to randomize the texture and thus
enhance ductility [59]. Similar weaker textures in Mg-based alloys
with the addition of RE elements were reported in Refs. [29,32,48–
50,61,62,63].
117
RD
1600
Mg
1400
Intensity, Counts
1200
1000
800
600
400
200 Ce Mn
0
0 10 20 30 40 50 60 70 80 90
Distance , μm
3.2. Tensile properties
Typical tensile stress-strain curve of the rolled ME20M-H112
alloy at a strain rate of 1 # 10 $ 3 s $ 1 is shown in Fig. 4 and the
tensile properties obtained are listed in Table 2. The present ME20
alloy exhibited a flow curve with a gradual transition from elastic to
plastic deformation like most of the RE-free extruded Mg alloys, i.e.,
AZ31, AM30 [16,17]. However, as seen from Fig. 4 and Table 2, a
lower tensile yield strength (YS) was attained in the present ME20
alloy in comparison with the RE-free extruded Mg alloys [16,17], but
had an equivalent or slightly higher ductility. While one benefit of
the RE elements added into Mg alloys was to improve the tensile
properties, e.g., microalloying with RE elements exhibited a higher
tensile yield strength than the extruded AM30 alloy [48], however,
it seemed that the tensile properties of the rolled ME20 alloy were
not improved, in comparison with the reported results in other RE-
Mg alloys [29,48–50,67]. The fine grained ME20 should exhibit a
higher yield strength than coarser RE-free extruded Mg alloys
[16,48], which is probably related to the weaker textures with a
broad basal pole scatter from RD (Fig. 3). Such textures are generally
very easy to deform, for example, by basal slip at lower stresses
compared with sheet textures similar to standard sheet basal tex-
tures, where c-axis compression leads to a very high yield strength
[60]. Similar results have also been reported for finer grains of ME20
sheets by Li et al. [60], where the tensile yield strength was less
than 200 MPa as well.
118 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125

RD

TD
ND

Basal plane Prismatic plane Pyramidal plane

(0001) (10 1 0 ) (10 11)

¯ ) plane, and pyramidal (1011

Fig. 3. Pole figures of basal (0001) plane, prismatic (1010 ¯ ) plane of the rolled ME20M alloy, where RD indicates the rolling direction, TD stands
for the transverse direction and ND indicates the normal direction.

current ME20 alloy was observed to exhibit unsymmetrical hys-

300 teresis loops, which were somewhat similar to those of RE-free
Engineering stress, MPa

extruded Mg alloys [16–18]. However, the extent of the asymmetry

250 or skewness of the hysteresis loops is less than that of RE-free
extruded Mg alloys [16–18]. This was, however, in contrast to the
hysteresis loops in the RE extruded Mg alloys [48–50], where the
200 hysteresis loops were nearly symmetrical, which were somewhat
similar to those of face-centered cubic (fcc) metals as a con-
sequence of the dislocation slip-dominated deformation in most
150 materials [70]. This unsymmetrical tensile and compressive
yielding phenomenon was distinguished as the Bauschinger-like
100 effect [16,17], and was associated with twinning in the compres-
sive phase and detwinning in the tensile phase [16,17]. It was re-
ported that low yield strength and strain hardening plateau in
50 compression is a direct characteristic of materials deformed by
twinning (e.g., in the descending phase of the first hysteresis loop
in Fig. 5), and the deformation by detwinning is responsible for the
0 low yield strength and strain hardening rate phenomena in the
0 5 10 15 20 25 subsequent ascending phase [71,72]. Twinning-induced plastic
strain could be completely recovered during detwinning, however,
Engineering strain, % due to its lower critical resolved shear stress (CRSS) detwinning
Fig. 4. Tensile stress-strain curve of the rolled ME20M-H112 alloy tested at a strain finished at the early stage of tensile loading [73]. Therefore, acti-
rate of 1 # 10 $ 3 s $ 1. vation of slip system is needed to accommodate the strain applied
in the subsequent stages of tensile loading, and this would trigger
a significant increase in tensile stress curve [74], as shown in Fig. 5
3.3. Hysteresis loops (the ascending phase of the hysteresis loop). Furthermore, unlike
the fcc metals where the slope of hysteresis loops after the strain
Typical stress-strain hysteresis loops of the first, second, and reversal either at the maximum or minimum stress was basically
mid-life cycles at a total strain amplitude of 1.2% and strain ratio of equal to the value of Young’s modulus, the pseudoelastic or non-
Rε ¼ $ 1 for the as-rolled ME20M-H112 alloy is shown in Fig. 5. The linear elastic behavior in both descending and ascending phases in
 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125 119

Table 2
Tensile properties of the rolled ME20M-H112 alloy, in comparison with those reported in the literature [16,17,48,50].

Tensile properties Yield strength, MPa Ultimate tensile strength, MPa Elongation, % Young's modulus, GPa Strain hardening exponent (n) K, MPa

Rolled ME20M-H112 140 234 20.0 42 0.19 390

Extruded AZ31 [16] 201 264 15.2 45 0.13 –
Extruded AM30 [17] 189 242 13.4 44 0.15 –
Extruded GW103 K [48,50] 232 318 8.3 – 0.15 –

200 200 1.2%

Total strain amplitudes
175 1.0%
150

Stress amplitude, MPa

0.8%
150 0.6%
100 0.4%
Stress, MPa

125 0.2%
50 0.1%
100
0
75
-50
50
-100
1st cycle 25
-150 2nd cycle 0
Mid-life cycle 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
-200
-2 -1 0 1 2 Number of cycles, N
Total strain, % Fig. 6. Stress amplitude vs. the number of cycles of the rolled ME20M-H112 alloy
tested at different total strain amplitudes at a strain ratio of Rε ¼ $ 1.
Fig. 5. Typical stress-strain hysteresis loops of different cycles of the rolled
ME20M-H112 alloy tested at a given total strain amplitude of 1.2% and strain ratio
of Rε ¼ $ 1.
0.8 Total strain amplitudes 1.2%
Plastic strain amplitude, %

the current ME20 alloy was also visible. It seemed reasonable to 1.0%
0.7
consider that the pseudoelastic behavior would be an inherent 0.8%
characteristic of Mg alloys, as it originated from reversible move- 0.6%
ment of dislocations, twinning, and stress induced phase trans-
0.6
0.4%
formation [16–18,75].
0.5 0.2%
0.1%
3.4. Cyclic stress and strain responses
0.4
The evolution of stress amplitudes and plastic amplitudes with
respect to the number of cycles at different strain amplitudes from
0.3
0.2% to 1.2% is shown in Figs. 6 and 7, respectively, under a semi-
logarithmic scale along the X-axis. It is observed that both stress
0.2
amplitudes and plastic strain amplitudes augmented, whereas
fatigue life of the rolled ME20M-H112 alloy decreased with in- 0.1
creasing total strain amplitudes. Unlike the extruded GW103K al-
loy [48–50] where microalloying with RE elements led to the 0
cyclic stabilization at almost all the strain amplitudes, it seemed 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
that cyclic stabilization was barely achieved in the rolled ME20M-
H112 alloy as seen from Fig. 6, except for 0.1% strain amplitude. At
Number of cycles, N
strain amplitudes of 1.2% and 1.0%, only slight change, i.e., an initial Fig. 7. Plastic strain amplitude vs. the number of cycles of the rolled ME20M-H112
slight cyclic hardening within the first three cycles and then minor alloy tested at different total strain amplitudes at a strain ratio of Rε ¼ $ 1.
cyclic softening up to failure, could be seen from Fig. 6. More or
less for all other strain amplitudes (0.2%, 0.4%, 0.6%, and 0.8%), In the low cycle fatigue tests, plastic strain amplitude has been
cyclic softening continues up to failure. In general, cyclic response
considered as a physical quantity that results in several damaging
(strain hardening [76] and strain softening [77]) is governed by the
processes and influences the internal microstructure which is
cyclic stability of the microstructural features, dislocation glide
closely related to the strain resistance and eventually the fatigue
and multiplication, twinning, and twin-dislocation interactions
[78]. The cyclic strain softening behavior for ME20 may be ex- life [16,17,48]. The variation of the plastic strain amplitude (Δεp/2)
plained by the twin softening mechanism; softening is caused by during cyclic deformation is shown in Fig. 7, which corresponded
the reorientation of lattice in the twinned area to provide an easier well to the change of the stress amplitude during cyclic de-
orientation for slip to happen [77,79]. formation as shown in Fig. 6 at different total strain amplitudes. It
120 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125
1.4
AM30 [1]
1.2 NZ30K [9]
AM30 [14]
1.0 GW103K [29]
AZ31 [56]
∆εt /2, %
0.8 AZ31 [75]
AZ61-ED [80]
0.6 ME20-present study
0.4
0.2
0.0
1E+1 1E+3 1E+5 1E+7
Number of cycles to failure, Nf
Fig. 8. Total strain amplitude versus the number of cycles to failure for the rolled
ME20M-H112 alloy, in comparison with the data reported in the literature for
various wrought Mg alloys [1,9,14,29,56,75,80].
is seen from Fig. 7 that as the total strain amplitudes increased the
plastic strain amplitudes also increased however fatigue life de-
creased. Cyclic stabilization only occurred at the strain amplitude
of 0.1%. For the rest of the strain amplitudes, it showed a trend of
increasing with the number of cycles, which is another proof of
cyclic softening, that occurred in all the tested samples except the
one under strain amplitude of 0.1%.
3.5. Fatigue life and fatigue parameters
Fig.8 displays the total strain amplitude (Δεt /2) as a function of
the number of cycles to failure (Nf, i.e., fatigue life) for the rolled
ME20M-H112 alloy, in comparison with the data reported in the
literature for both RE-free and RE-containing wrought Mg alloys
[1,9,14,29,56,75,80]. The run-out data points were specified by
arrows pointing horizontally at or over 107 cycles. The present
rolled ME20M-H112 alloy showed a trend of increasing fatigue life
with decreasing strain amplitude and overall, the alloy basically
showed an improved fatigue life than the RE-free extruded Mg
alloys [18–22,69]. However, at lower stain amplitude of 0.2% the
ME20 alloy showed the least fatigue life. This phenomenon is
closely related with material strength and ductility. The present
rolled ME20-H112 alloys showed a higher ductility and lower
strength compared with other alloys [18–22,69] and this resulted
in better fatigue resistance at larger stain amplitudes. The total
strain amplitude could be expressed as two parts of elastic strain
amplitude and plastic strain amplitude according to the Basquin
equation and Coffin-Manson relation [1,6–9,14,17,19,23], i.e.,
Δεt Δε Δεp σ f′(2Nf )b
= e + = + εf′(2Nf )c ,
2 2 2 E
where E is the Young's modulus (for the present alloy the average
value obtained during fatigue testing was ! 42 GPa), Nf is the fa-
tigue life or the number of cycles to failure, σ′f is the fatigue
strength coefficient, b is the fatigue strength exponent, ε′f is the
fatigue ductility coefficient, and c is the fatigue ductility exponent.
In addition, cyclic deformation behavior is normally considered to
be related to the portion of the plastic strain amplitude and is
-1
Total
-1.5 Plastic
Log(∆εt/2) = -0.3276(2Nf) - 1.0335
R² = 0.8757
Elastic
-2
Log (∆ε/2)
-2.5 Log(∆εe/2) = -0.1512(2Nf) - 1.8008
R² = 0.7698
-3
-3.5 Log(∆εp/2) = -0.7636(2Nf) - 0.1269
R² = 0.8691
-4
1 2 3 4 5 6
Log(2Νf)
Fig. 9. Total, plastic, and elastic strain amplitude-fatigue life response of the rolled
ME20M-H112 alloy.
independent of the elastic strain amplitude, which could be ex-
pressed by the following equation [17],
Δσ ⎛ Δεp ⎞n ′
= K ′⎜ ⎟ ,
2 ⎝ 2 ⎠ (2)
where Δσ
is the mid-life stress amplitude,
Δεp
is the mid-life plastic
2 2
strain amplitude, n′ is the cyclic strain-hardening exponent and K’
is the cyclic strength coefficient. Fig.9 illustrates the elastic, plastic,
and total strain amplitudes as a function of the number of re-
versals to failure (2Nf). In order to make sure that cyclic stabili-
zation, also called cyclic saturation, has already occurred, the
stress and strain values of the mid-life cycles were used. Hence,
the fatigue life parameters obtained by means of (Eqs. (1) and 2)
were presented in Table 3, in comparison with the data reported in
the literature for various extruded Mg alloys [16,17,48,50]. It is
seen that while the obtained fatigue parameters were well within
the range in other Mg alloys reported in the literature
[16,17,48,50], the cyclic strain hardening exponent n’ of the
ME20M-H112 alloy was lower than that of RE-free Mg alloys
[16,17].
Fig. 10 illustrates a superposition of the monotonic tensile
stress-strain curve at a strain rate of 1 # 10 $ 2 s $ 1 and cyclic stress-
strain curve (CSSC) at mid-life cycles with the Y-axis stress in-
dicated as the normal stress amplitude ( Δσ ) for the ME20M-H112
2
Mg alloy tested at a strain ratio of Rε ¼ $ 1. It is seen that cyclic
stress-strain curve was positioned below the monotonic one of the
ME20M-H112 alloy, suggesting the presence of cyclic softening
and being consistent with the response of stress and plastic strain
amplitudes as shown in Figs. 6 and 7. The obtained cyclic yield
strength (s‘y) and cyclic strain hardening exponent (n0 ) of the
present ME20M-H112 alloy were also slightly lower than the
corresponding monotonic yield strength and strain hardening
exponent, as seen from Tables 2 and 3. This could be the reason
(1)
why this alloy exhibited a certain extent of cyclic softening even at
higher strain amplitude of 1.2%. In addition, as the monotonic
hardening exponent n was previously obtained to be about 0.19
(Table 2), whereas the cyclic strain hardening exponent (n0 ) was
about 0.11 which was lower than n. The fact that the normal CSSC
was located below the monotonic stress-strain curve (Fig.10) di-
rectly reflected a slight cyclic softening at total strain amplitudes
from 0.4% to 1.2%, as seen from Fig.10. This is in contrast to the
 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125 121

Table 3
Low cycle fatigue parameters obtained for the rolled ME20M-H112 alloy, in comparison with those reported in the literature [16,17,48,50].

Low cycle fatigue parameters Rolled ME20M-H112 Extruded AZ31 [16] Extruded AM30 [17] Extruded GW103 K [48,50]

Cyclic yield strength, s‘y,MPa 115 – – 215

Cyclic strain hardening exponent, n′ 0.11 0.34 0.33 0.19
Cyclic strength coefficient, K’, MPa 240 1976 1610 734
Fatigue strength coefficient, s‘f, MPa 665 616 678 518
Fatigue strength exponent, b $ 0.15 $ 0.15 $ 0.16 $ 0.11
Fatigue ductility coefficient, ε‘f 0.75 1.78 2.97 0.05
Fatigue ductility exponent, c $ 0.76 $ 0.40 $ 0.44 $ 0.44

seen from the low magnification images (Fig. 12) that fatigue crack
200 initiated basically from the specimen surface. It was observed
clearly that fatigue crack propagation area was larger at a lower
strain amplitude of 0.2% than at a higher strain amplitude of 1.0%
∆σ/2 at mid-life, MPa

150 (Fig. 11(a) and (b)). In contrast to typical fatigue striations

[14,16,17,19,34,48–51], the present alloy showed mainly some
shallow dimples along with some fractions of quasi-cleavage fea-
tures in the crack propagation area as shown in Fig.13(a) and (b) at
100 a higher magnification.

3.7. Comparison between different Mg alloys

50 Since the present rolled ME20M-H112 alloy exhibited a certain

Monotonic stress-strain curve
Cyclic stress-strain curve
0 winning behavior, it is thus necessary to identify the difference
0 0.2 0.4 0.6 0.8 1 1.2 1.4
∆εt /2 at mid-life, %
Fig. 10. Cyclic stress-strain curve for the rolled ME20M-H112 alloy, where the
corresponding monotonic stress-strain curve is also potted for comparison.
situation of a RE-free extruded AZ31 Mg alloy which exhibited
strong cyclic hardening, where the values of 0.13 and 0.34 for n
and n0 , respectively, were reported [16]. Therefore, it could be
concluded that the present ME20M-H112 alloy was susceptible to
cyclic softening under cyclic loading conditions rather than the
monotonic ones.
3.6. Fractography
Fig.11 showed the overall view of the fracture surfaces of the
rolled samples at a total strain amplitude of (a) 0.2% and (b) 1.0%,
respectively, containing fatigue crack initiation (as indicated by
dashed yellow-box), propagation, final fast fracture regions. It is
extent of asymmetrical stress-strain behavior in tension and
compression (Fig. 5), which are of major concern in relation to the
influence of RE element addition as well as the twinning-det-
from RE-Mg and RE-free Mg alloys based on the stress-strain
hysteresis loops. Fig. 14 illustrates the stress-strain hysteresis loops
of the mid-life cycle for a RE-free AM30 Mg alloy and a RE-con-
taining GW103K Mg alloy, along with that of the present RE-con-
taining rolled ME20M-H112 alloy. It is seen from Fig. 14 that the
hysteresis loop of ME20M-H112 alloy showed a strong Bauschin-
ger-like effect associated with a skewed asymmetrical shape si-
milar to the AM30 alloy, which was predominantly associated with
the twinning-detwinning process [14,16–19,71]. The similarity in
hysteresis loops between the ME20M-H112 alloy and AM30 alloy
is directly related to the presence of crystallographic texture which
dominates the orientation of slip or twinning planes and direc-
tions relative to the externally applied stress [81]. As mentioned
earlier in Section 3.1, the present ME20M-H112 alloy showed re-
latively weak textures (3.8 MRD) (Fig. 3), which lay in-between the
textures of both comparison alloys (e.g., AM30 and GW103K, ! 7,
and 2.1 MRD, respectively). In this context, Fig. 14 actually helped
emphasize the conclusions that the stress-strain behavior of the
present ME20M-H112 alloy exhibit a characteristic lying in-

Fig. 11. SEM images of overall fracture surfaces of the rolled ME20M-H112 specimens fatigued at a total strain amplitude of (a) 0.2% and (b) 1.0%.
122 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125

Fig. 12. SEM micrographs of the fracture surface near crack initiation of the rolled ME20M-H112 specimens fatigued at a total strain amplitude of (a) 0.2% and (b) 1.0%.

Fig.13. SEM micrographs of the fatigue crack propagation region of the rolled specimens fatigued at total strain amplitude of (a) 0.2% and (b) 1.0%.

400 (Fig.14), which was also in-between both comparison alloys (e.g.,
Mid-life cycle AM30 and GW103K, ! 0.55 and ! 1, respectively). This indicates
300 that yield asymmetry is not very prominent as compared with
other wrought Mg alloy, and this is probably due to the light ad-
200 dition of RE element addition. While there was almost no yield
asymmetry for GW103K during cyclic loading as reported in ref.
Stress, MPa

[48], the RE content is over 10 wt% for this alloy as compared with
100
merely 0.29–0.33 wt% for ME20. Fig.15 shows the distribution of
deformation features in the area near the fracture surface in the
0
fatigued samples of the rolled ME20M-H112 alloy, along with
AM30 alloy and GW103K alloy. It is seen that the grain size of the
-100
ME20M-H112 alloy was significantly smaller than that of AM30
alloy and even GW103K alloy (Fig.15). While the grain size was
-200 ME20
reported to have a significant effect on the tendency of twinning
GW103K [80], it is in contrast to the usual hypothesis in the case of ME20M-
-300
AM30 H112 alloy. The area near the fracture surface of ME20M-H112
alloy exhibited lots of small twins. Thus, one of the obvious factors
-400
responsible for the asymmetry of hysteresis loops (Figs. 5 and 14)
-1.5 -1 -0.5 0 0.5 1 1.5 is the presence of abundant deformation twins, even though grain
Strain, % refinement was achieved due to the RE-element and process
control. The combined role of the deformation twins and ellip-
Fig. 14. Typical stress-strain hysteresis loops of the mid-life cycle at a given total
soidal particles containing Mg and Ce (Mg12Ce) as reported by Li
strain amplitude of 1.2% and strain ratio of Rε ¼ $ 1 for the rolled ME20, GW103 K,
and AM30 alloys, respectively. et al. [61] would be responsible for the asymmetric hysteresis loop
(Figs. 5 and 14). The above results and analyses indicate that even
between both comparison alloys. However, this is in contrast with the presence of RE-particles and the grain refinement in the pre-
the typically cast Mg alloys, which have no such yield asymmetry sent ME20M-H112 alloy, the decisive role in suppressing the oc-
because of their weak or no texture [68]. currence of deformation twins was absent, which led to the ten-
In addition, the ratio of the compressive-to-tensile yield stress sion-compression asymmetry almost similar to AM30 alloy
was ! 0.67 for ME20 alloy in the mid-life cycle of hysteresis loop (Fig. 5 and Fig. 14).
 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125
Fig. 15. Optical micrographs in the area near the fracture surface at a total strain amplitude of 1.2%, showing the distribution of deformation features in the fatigued samples
of (a) rolled ME20M-H112 alloy, (b) GW103K alloy, and (c) AM30 alloy, respectively.
4. Conclusions
Strain-controlled low cycle fatigue tests were conducted on a
low rare-earth containing rolled ME20M-H112 alloy under varying
strain amplitudes. The following conclusions could be drawn from
this investigation:
1. The microstructure of the ME20M-H112 alloys consisted of
equiaxed grains with an average grain size ! 12 mm arising from
the dynamic recrystallization occurred in the rolling process.
Unlike the RE-free extruded magnesium alloys, the present alloy
exhibited a relatively weak basal texture with a maximum in-
tensity of 3.8 MRD.
2. Unlike the high RE-containing GW103K alloy, the rolled
ME20M-H112 alloy exhibited asymmetrical hysteresis loops si-
milar to the RE-free extruded Mg alloys. This was pre-
dominantly due to the twinning-detwinning activities during
cyclic deformation.
3. The cyclic stabilization was barely achieved even at the lower
strain amplitudes. At higher strain amplitude, an initial cyclic
hardening occurred within a few cycles, followed by cyclic
softening.
4. The fatigue life of the ME20M-H112 alloy was equivalent to that
of other RE-free extruded Mg alloys within the experimental
scatter, which could be well described by the Coffin-Manson
law and Basquin’s equation.
5. SEM examinations revealed that fatigue cracks initiated from
the specimen surface or near-surface defects. In contrast to the
typical fatigue striations, the present alloy showed some
123
shallow dimples along with some fractions of quasi-cleavage
features in the crack propagation area.
Acknowledgements
The authors would like to thank the Natural Sciences and En-
gineering Research Council of Canada (NSERC) and AUTO21 Net-
work of Centres of Excellence for providing financial support. D.R.
Ni would like to give thanks to National Natural Science Founda-
tion of China for the financial support (Grant no. 51371179). D.L.
Chen is grateful for the financial support by the Premier's Research
Excellence Award (PREA), NSERC-Discovery Accelerator Supple-
ment (DAS) Award, Canada Foundation for Innovation (CFI), and
Ryerson Research Chair (RRC) program. The authors would like to
thank Dr. R. Tandon and Dr. B. Davies (Magnesium Elektron) for
supplying magnesium powders for the defocusing calibration. The
authors would also like to thank Messrs. A. Machin, Q. Li, J.
Amankrah and R. Churaman for easy access to the laboratory
facilities of Ryerson University and their assistance in the
experiments.
References
[1] M. McNutt, The beyond-two-degree inferno, Science 349 (6243) (2015) 7.
[2] T.M. Pollock, Weight loss with magnesium alloys, Science 328 (2010) 986–987.
[3] J.F. Nie, Y.M. Zhu, J.Z. Liu, X.Y. Fang, Periodic segregation of solute atoms in fully
coherent twin boundaries, Science 340 (2013) 957–960.
[4] S. Chu, A. Majumdar, Opportunities and challenges for a sustainable energy
124 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125
future, Nature 488 (2012) 294–303.
[5] F. Creutzig, P. Jochem, O.Y. Edelenbosch, L. Mattauch, D.P. van Vuuren,
D. McCollum, J. Minx, Transport: a roadblock to climate change mitigation?
Science 350 (2015) 911–912.
[6] M. McNutt, Climate change impacts, Science 341 (2013) 435.
[7] C. Ash, E. Culotta, J. Fahrenkamp-Uppenbrink, D. Malakoff, J. Smith, A. Sugden,
S. Vignieri, Once and future climate change, Science 341 (2013) 473.
[8] S.J. Davis, K. Caldeira, H.D. Matthews, Future CO2 emissions and climate
change from existing energy infrastructure, Science 329 (2010) 1330–1333.
[9] J.D. Shakun, P.U. Clark, F. He, S.A. Marcott, A.C. Mix, Z.Y. Liu, B. Otto-Bliesner,
A. Schmittner, E. Bard, Global warming preceded by increasing carbon dioxide
concentrations during the last deglaciation, Nature 484 (2012) 49–54.
[10] T.A. Schaedler, A.J. Jacobsen, W.B. Carter, Toward lighter, stiffer material, Sci-
ence 341 (2013) 1181–1182.
[11] M. Anderson, M. Auffhammer, Pounds that kill: the external costs of vehicle
weight, Rev. Econ. Stud. 81 (2014) 535–571.
[12] A.A. Luo, Magnesium: current and potential automotive applications, JOM 54
(2002) 42–48.
[13] V.V. Ogarevic, R.I. Stephens, Fatigue of magnesium alloys, Annu. Rev. Mater.
Sci. 20 (1990) 141–177.
[14] X.Z. Lin, D.L. Chen, Strain controlled cyclic deformation behavior of an ex-
truded magnesium alloy, Mater. Sci. Eng. A 496 (2008) 106–113.
[15] F.A. Mirza, D.L. Chen, Fatigue of lightweight magnesium alloys, in: S. Zhang, D.
L. Zhao (Eds.), Aerospace Materials Handbook, CRC Press, Taylor & Francis, New
York, 2013, pp. 647–698.
[16] S. Begum, D.L. Chen, S. Xu, A.A. Luo, Low cycle fatigue properties of an ex-
truded AZ31 magnesium alloy, Int. J. Fatigue 31 (2009) 726–735.
[17] S. Begum, D.L. Chen, S. Xu, A.A. Luo, Strain-controlled low-cycle fatigue
properties of a newly developed extruded magnesium alloy, Metall. Mater.
Trans. A 39 (2008) 3014–3026.
[18] S. Begum, D.L. Chen, S. Xu, A.A. Luo, Effect of strain ratio and strain rate on low
cycle fatigue behavior of AZ31 wrought magnesium alloy, Mater. Sci. Eng. A
517 (2009) 334–343.
[19] C.L. Fan, D.L. Chen, A.A. Luo, Dependence of the distribution of deformation
twins on strain amplitudes in an extruded magnesium alloy after cyclic de-
formation, Mater. Sci. Eng. A 519 (2009) 38–45.
[20] X.B. Liu, R.S. Chen, E.H. Han, Effects of ageing treatment on microstructures
and properties of Mg-Gd-Y-Zr alloys with and without Zn additions, J. Alloys
Compd. 465 (2008) 232–238.
[21] K.U. Kainer, Magnesium - Alloys and Technology, Wiley-VCH, Cambridge,
2003.
[22] N. Tahreen, D.L. Chen, M. Nouri, D.Y. Li, Influence of aluminum content on
twinning and texture development of cast mg-al-Zn alloy during compression,
J. Alloys Compd. 623 (2015) 15–23.
[23] N. Tahreen, D.F. Zhang, F.S. Pan, X.Q. Jiang, C. Li, D.Y. Li, D.L. Chen, Character-
ization of hot deformation behavior of an extruded Mg‐Zn‐Mn‐Y alloy con-
taining LPSO phase, J. Alloys Compd. 644 (2015) 814–823.
[24] N. Tahreen, D.L. Chen, M. Nouri, D.Y. Li, Effects of aluminum content and strain
rate on strain hardening behavior of cast magnesium alloys during compres-
sion, Mater. Sci. Eng. A 594 (2014) 235–245.
[25] N. Tahreen, D.F. Zhang, F.S. Pan, X.Q. Jiang, C. Li, D.Y. Li, D.L. Chen, Influence of
yttrium content on phase formation and strain hardening behavior of Mg-Zn-
Mn magnesium alloy, J. Alloys Compd. 615 (2014) 424–432.
[26] W.Q. Xu, N. Birbilis, G. Sha, Y. Wang, J.E. Daniels, Y. Xiao, M. Ferry, A high-
specific-strength and corrosion-resistant magnesium alloy, Nature Mater 14
(2015) 1229–1235.
[27] G.S. Frankel, Magnesium alloys: ready for the road, Nature Mater 14 (2015)
1189–1190.
[28] L.Y. Chen, J.Q. Xu, H. Choi, M. Pozuelo, X.L. Ma, S. Bhowmick, J.M. Yang,
S. Mathaudhu, X.C. Li, Processing and properties of magnesium containing a
dense uniform dispersion of nanoparticles, Nature 528 (2015) 539–543.
[29] F. Yang, F. Lv, X.M. Yang, S.X. Li, Z.F. Zhang, Q.D. Wang, Enhanced very high
cycle fatigue performance of extruded Mg-12Gd-3Y-0.5Zr magnesium alloy,
Mater. Sci. Eng. A 528 (2011) 2231–2238.
[30] S.M. Yin, H.J. Yang, S.X. Li, S.D. Wu, F. Yang, Cyclic deformation behavior of as-
extruded Mg-3%Al-1%Zn, Scr. Mater. 58 (2008) 751–754.
[31] E.A. Ball, P.B. Prangnell, Tensile-compressive yield asymmetries in high
strength wrought magnesium alloys, Scr. Metall. Mater. 31 (1994) 111–116.
[32] J. Bohlen, M.R. Nurnberg, J.W. Senn, D. Letzig, S.R. Agnew, The texture and
anisotropy of magnesium-zinc-rare earth alloy sheets, Acta Mater. 55 (2007)
2101–2112.
[33] J.B. Jordon, J.B. Gibson, M.F. Horstemeyer, H. El-Kadiri, J.C. Baird, A.A. Luo, Ef-
fect of twinning, slip, and inclusions on the fatigue anisotropy of extrusion-
textured AZ61 magnesium alloy, Mater. Sci. Eng. A 528 (2011) 6860–6871.
[34] F.A. Mirza, D.L. Chen, Fatigue of rare-earth containing magnesium alloys: a
review, Fatigue Fract. Eng. Mater. 37 (2014) 831–853.
[35] T.A. Samman, X. Li, Sheet texture modification in magnesium-based alloys by
selective rare earth alloying, Mater. Sci. Eng. A 528 (2011) 3809–3822.
[36] J. Bohlen, S.B. Yi, D. Letzig, K.U. Kainer, Effect of rare earth elements on the
microstructure and texture development in magnesium-manganese alloys
during extrusion, Mater. Sci. Eng. A 527 (2010) 7092–7098.
[37] R. Cottam, J. Robson, G. Lorimer, B. Davis, Dynamic recrystallization of Mg and
Mg-Y alloys: crystallographic texture development, Mater. Sci. Eng. A 485
(2008) 375–382.
[38] J.P. Hadorn, K. Hantzsche, S.B. Yi, J. Bohlen, D. Letzig, J.A. Wollmershauser, S.
R. Agnew, Role of solute in the texture modification during hot deformation of
Mg‐rare earth alloys, Metall. Mater. Trans. A 43 (4) (2012) 1347–1362.
[39] J. Dong, W.C. Liu, X. Song, P. Zhang, W.J. Ding, A.M. Korsunsky, Influence of
heat treatment on fatigue behavior of high-strength Mg-10Gd-3Y alloy, Mater.
Sci. Eng. A 725 (2010) 6053–6063.
[40] Z.M. Li, Q.G. Wang, A.A. Luo, P.H. Fu, L.M. Peng, Y.G. Wang, High cycle fatigue of
cast Mg-3Nd-0.2Zn magnesium alloys, Metall. Mater. Trans. A 44A (2013)
5202–5215.
[41] Y. Yang, Y.B. Liu, S.Y. Qin, Y. Fang, High cycle fatigue properties of die-cast
magnesium alloy AZ91D with addition of different concentrations of cerium, J.
Rare Earths 24 (2006) 591–595.
[42] Y. Yang, Y.B. Liu, High cycle fatigue characterization of two die-cast magne-
sium alloys, Mater. Charact. 59 (2008) 567–570.
[43] H. Bayani, E. Saebnoori, Effect of rare earth elements addition on thermal fa-
tigue behaviors of AZ31 magnesium alloy, J. Rare Earths 27 (2009) 255–258.
[44] L. Nascimento, S.B. Yi, J. Bohlen, L. Fuskova, D. Letzig, K.U. Kainer, High cycle
fatigue behaviour of magnesium alloys, Procedia Eng. 2 (2010) 743–750.
[45] Y.L. Xu, K. Zhang, X.G. Li, J. Lei, Y.S. Yang, T.J. Luo, High cycle fatigue properties
of die-cast magnesium alloy AZ91D-1%MM, Trans, Nonfer. Met. Soc. China 18
(2008) 306–311.
[46] W.-C. Liu, J. Dong, P. Zhang, L. Jin, T. Peng, C.Q. Zhai, W.-J. Ding, Fatigue be-
havior of hot extruded Mg-10Gd-3Y magnesium alloy, J. Mater. Res. 25 (4)
(2010) 773–783.
[47] P. Maier, G. Tober, C.L. Mendis, S. Müller, N. Hort, Influence of Nd in extruded
Mg-10Gd base alloys on fatigue strength, Adv. Mater. Res. 783 (2014) 419–424.
[48] F.A. Mirza, D.L. Chen, D.J. Li, X.Q. Zeng, Low cycle fatigue of a rare-earth con-
taining extruded magnesium alloy, Mater. Sci. Eng. A 575 (2013) 65–73.
[49] F.A. Mirza, D.L. Chen, D.J. Li, X.Q. Zeng, Effect of strain ratio on cyclic de-
formation behavior of a rare-earth containing extruded magnesium alloy,
Mater. Sci. Eng. A 5 (2013) 250–259.
[50] F.A. Mirza, D.L. Chen, D.J. Li, X.Q. Zeng, Cyclic deformation behavior of a rare-
earth containing extruded magnesium alloy: effect of heat treatment, Metall.
Mater. Trans. A 46 (3) (2015) 1168–1187.
[51] F.A. Mirza, D.L. Chen, D.J. Li, X.Q. Zeng, Cyclic deformation of rare-earth con-
taining magnesium alloys, Adv. Mater. Res. 891 (2014) 391–396.
[52] F.H. Wang, J. Dong, Y.Y. Jiang, W.J. Ding, Cyclic deformation and fatigue of
extruded Mg-Gd-Y magnesium alloy, Mater. Sci. Eng. A 561 (2013) 403–410.
[53] F.H. Wang, J. Dong, M.L. Feng, J. Sun, W.J. Ding, Y.Y. Jiang, A study of fatigue
damage development in extruded Mg-Gd-Y magnesium alloy, Mater. Sci. Eng.
A 589 (2014) 209–216.
[54] F. Mokdad, D.L. Chen, Strain-controlled low cycle fatigue properties of a rare-
earth containing ZEK100 magnesium alloy, Mater. Des. 67 (2015) 436–447.
[55] F. Mokdad, D.L. Chen, Cyclic deformation and anelastic behavior of ZEK100
magnesium alloy: effect of strain ratio, Mater. Sci. Eng. A 640 (2015) 243–258.
[56] F. Bachmann, R. Hielscher, H. Schaeben, Texture analysis with MTEX - free and
open source software toolbox, Solid State Phen. 160 (2010) 63–68.
[57] K. Yu, S.T. Rui, J.M. Song, W.X. Li, L. Guo, Effects of grain refinement on me-
chanical properties and microstructures of AZ31 alloy, Trans, Nonfer. Met. Soc.
China 18 (2008) s39–s43.
[58] M. Masoumi, M. Hoseini, M. Pekguleryuz, The influence of Ce on the micro-
structure and rolling texture of Mg-1%Mn alloy, Mater. Sci. Eng. A 528 (2011)
3122–3129.
[59] M. Huppmanna, S. Gall, S. Müller, W. Reimers, Changes of the texture and the
mechanical properties of the extruded Mg alloy ME21 as a function of the
process parameters, Mater. Sci. Eng. A 528 (2010) 342–354.
[60] X. Li, T. Al-Samman, G. Gottstein, Microstructure development and texture
evolution of ME20 sheets processed by accumulative roll bonding, Mater. Lett.
65 (2011) 1907–1910.
[61] X. Li, T. Al-Samman, S. Mu, G. Gottstein, Texture and microstructure devel-
opment during hot deformation of ME20 magnesium alloy: experiments and
simulations, Mater. Sci. Eng. A 528 (2011) 7915–7925.
[62] X. Li, T. Al-Samman, G. Gottstein, Mechanical properties and anisotropy of
ME20 magnesium sheet produced by unidirectional and cross rolling, Mater.
Des. 32 (2011) 4385–4393.
[63] Q.S. Yang, B. Jiang, X. Li, H.W. Dong, W.J. Liu, F.S. Pan, Microstructure and
mechanical behavior of the Mg‐Mn‐Ce magnesium alloy sheets, J. Magnesium
Alloys 2 (2014) 8–12.
[64] X. Li, W. Qi, Effect of initial texture on texture and microstructure evolution of
ME20 Mg alloy subjected to hot rolling, Mater. Sci. Eng. A 560 (2013) 321–331.
[65] D. Sarker, D.L. Chen, Detwinning and strain hardening of an extruded mag-
nesium alloy during compression, Scr. Mater. 67 (2012) 165–168.
[66] D. Sarker, D.L. Chen, Texture transformation in an extruded magnesium alloy
under pressure, Mater. Sci. Eng. A 582 (2013) 63–67.
[67] F. Lv, F. Yang, Q.Q. Duan, Y.S. Yang, S.D. Wu, S.X. Li, Z.F. Zhang, Fatigue prop-
erties of rolled magnesium alloy (AZ31) sheet: influence of specimen or-
ientation, Int. J. Fatigue 33 (2011) 672–682.
[68] N. Stanford, M. Barnett, Effect of composition on the texture and deformation
behaviour of wrought Mg alloys, Scr. Mater. 58 (2008) 179–182.
[69] J. Hirsch, T. Al-Samman, Superior light metals by texture engineering: opti-
mized aluminum and magnesium alloys for automotive applications, Acta
Mater. 61 (2013) 818–843.
[70] S. Suresh, Fatigue of Materials, second ed, Cambridge University Press, Cam-
bridge, 1998.
[71] L. Wu, A. Jain, D.W. Brown, G.M. Stoica, S.R. Agnew, B. Clausen, D.E. Fielden, P.
K. Liaw, Twinning-detwinning behavior during the strain-controlled low-cycle
fatigue testing of a wrought magnesium alloy, Acta Mater. 56 (2008) 688–695.
[72] D. Sarker, J. Friedman, D.L. Chen, De-twinning and texture change in an
 F.A. Mirza et al. / Materials Science & Engineering A 661 (2016) 115–125
extruded AM30 magnesium alloy during compression along normal direction,
J. Mater. Sci. Tech. 31 (2015) 264–268.
[73] S.G. Hong, S.H. Park, C.S. Lee, Enhancing the fatigue property of rolled AZ31
magnesium alloy by controlling {10–12} twinning-detwinning characteristics,
J. Mater. Res. 25 (2010) 784–792.
[74] S.H. Park, S.-G. Hong, B.H. Lee, W.K. Bang, C.S. Lee, Low-cycle fatigue char-
acteristics of rolled Mg-3Al-1Zn alloy, Int. J. Fatigue 32 (2010) 1835–1842.
[75] C.H. Cáceres, T. Sumitomo, M. Veidt, Pseudoelastic behaviour of cast magne-
sium AZ91 alloy under cyclic loading-unloading, Acta Mater. 51 (20) (2003)
6211–6218.
[76] M.R. Barnett, Twinning and ductility of magnesium alloys: (part I) “tension”
twins, Mater. Sci. Eng. A 464 (2007) 1–7.
[77] S. Xu, V.Y. Gertsman, J. Li, J.P. Thompson, M. Sahoo, Role of mechanical twin-
ning in tensile compressive yield asymmetry of die cast Mg alloys, Can. Metall.
125
Quart. 44 (2) (2005) 155–166.
[78] F.A. Mirza, D.L. Chen, D.J. Li, X.Q. Zeng, Low cycle fatigue of extruded Mg-3Nd-
0.2Zn-0.5Zr (wt%) magnesium alloy, Mater. Des. 64 (2014) 63–73.
[79] S.R. Kalidindi, A.A. Salem, R.D. Doherty, Role of deformation twinning on
strain hardening in cubic and hexagonal polycrystalline metals, Adv. Eng.
Mater. 5 (2003) 229–232.
[80] J. Koike, T. Kobayashi, T. Mukai, H. Watanabe, M. Suzuki, K. Maruyama,
K. Higashi, The activity of non-basal slip systems and dynamic recovery at
room temperature in fine-grained AZ31B magnesium alloys, Acta Mater. 51
(2003) 2055–2065.
[81] J. Jain, W.J. Poole, C.W. Sinclair, M.A. Gharghouri, Reducing the tension-com-
pression yield asymmetry in a Mg-8Al-0.5Zn alloy via precipitation, Scr. Mater.
62 (2010) 301–304.
JOM, Vol. 68, No. 5, 2016
DOI: 10.1007/s11837-015-1796-7
Ó 2016 The Minerals, Metals & Materials Society

Microstructure and Fatigue Properties of Ultrasonic Spot

Welded Joints of Aluminum 5754 Alloy

F.A. MIRZA ,1,2 A. MACWAN,1 S.D. BHOLE,1 and D.L. CHEN1,3

1.—Department of Mechanical and Industrial Engineering, Ryerson University, 350 Victoria Street,
Toronto, ON M5B 2K3, Canada. 2.—e-mail: f4mirza@ryerson.ca. 3.—e-mail: dchen@ryerson.ca

The purpose of this investigation was to evaluate the microstructural change,

lap shear tensile load, and fatigue resistance of ultrasonic spot welded joints of
aluminum 5754 alloy for automotive applications. A unique ‘‘necklace’’-type
structure with very fine equiaxed grains was observed to form along the weld
line due to the mechanical interlocking coupled with the occurrence of dy-
namic recrystallization. The maximum lap shear tensile strength of 85 MPa
and the fatigue limit of about 0.5 kN (at 1 9 107 cycles) were achieved. The
tensile fracture occurred at the Al/Al interface in the case of lower energy
inputs, and at the edge of nugget zone in the case of higher energy inputs. The
maximum cyclic stress for the transition of fatigue fracture mode from the
transverse through-thickness crack growth to the interfacial failure increased
with increasing energy input. Fatigue crack propagation was mainly charac-
terized by the formation of fatigue striations, which usually appeared per-
pendicular to the fatigue crack propagation.

to expand the use of Al alloys in automotive body

INTRODUCTION
Light-weighting in ground vehicles is deemed as
one of the most effective strategies to improve fuel
economy and reduce anthropogenic climate-chang-
ing, environment-damaging, costly and human
death-causing* emissions.1,2 To manufacture
lighter-weight vehicles, advanced high-strength
steels, aluminum (Al) alloys, magnesium alloys,
and polymers are being used in the aerospace and
automotive sectors. The development and applica-
tion of Al alloys have been significantly increasing
in the transportation industry because of their high
strength-to-weight ratio, good machinability, envi-
ronmental friendliness and recyclability.3–5 In order
*According to Science News entitled ‘‘Air pollution kills 7 million
people a year’’ on March 25, 2014 at http://news.sciencemag.org/
signal–noise/2014/03/air-pollution-kills-7-million-people-year: ‘‘Air
pollution isn’t just harming Earth; it’s hurting us, too. Startling
new numbers released by the World Health Organization today
reveal that one in eight deaths are a result of exposure to air
pollution. The data reveal a strong link between the tiny particles
that we breathe into our lungs and the illnesses they can lead to,
including stroke, heart attack, lung cancer, and chronic obstruc-
tive pulmonary disease.’’
manufacturing, lower-cost joining methods are
important.5 One of the predominant processes of
welding Al alloys is resistance spot welding
(RSW),5,6 however it has a limited application to
Al welding due to high energy consumption.6 More-
over, the major metallurgical problem of joining
dissimilar metals is the formation of brittle inter-
metallic compounds (IMCs), which can seriously
degrade a joint’s mechanical properties.3 Nowadays,
two new solid state welding processes, i.e., friction
stir spot welding (FSSW)7–9 and ultrasonic spot
welding (USW),3,10,11 are being considered as poten-
tial replacement technologies for RSW.5,7 On the
other hand, USW uses localized high-frequency
vibratory energy and moderate clamping forces
during joining and demonstrated the ability to
achieve superior strengths than FSSW when com-
pared on the basis of the alloy thickness and
processing condition (e.g., same nugget
area).3–5,12–14 From an energy consumption point
of view, Al welding using the USW process con-
sumes only about 0.3 kWh per 1000 joints, com-
pared to 20 kWh with RSW, and 2 kWh with
FSSW.5 It has also been reported that ultrasonic
spot welded (USWed) joints have higher failure
energy than that of FSSWed joints.9,15

(Published online January 11, 2016) 1465

1466
In higher-power USW of thicker gauge sheet
compatible with automotive applications, to date,
only a few studies have reported on joining alu-
minum.9,13,16–19 It was reported that USW of opti-
mized Al 1 mm- to 2-mm-gauge sheet joints are
produced at weld energies in the range of 0.7–
1.5 kJ.9 At a lower level of weld energy, the
microbond density was too low, failure occurred
through the interface.10 Higher weld energies can
effectively promote joining in the entire weld, thus
improving weld strength. Jahn et al.13 evaluated
the effects of welding energy on the microstructure
and weld strength of ultrasonic spot welds of
AA6111-T4 Al alloy using a single transducer,
unidirectional wedge-reed welding machine. It was
observed that the weld strength increased with
increasing welding energy up to 500 J, and the
maximum failure load obtained was between 2.6 kN
and 3.1 kN. Bakavos and Prangnell16 reported the
Fig. 1. EBSD orientation maps of an USWed Al5754 similar joint at a welding energy of 2500 J: (a) BM and (b) near the nugget zone.
Mirza, Macwan, Bhole, and Chen
joining mechanisms and microstructure formation
in a high power ultrasonic spot-welded 6111 Al
sheet and sound welds were obtained with a shear
failure load up to 3.5 kN. Strong welds could be
achieved only after the welding energy exceeded a
threshold, then a failure mode of button pullout
occurred.10 Typically, the welding energy required
during USW is closely related to the clamping
pressure and welding time, as well as the vibration
frequency. However, with the exception of the
notable work of Jahn et al.4,13 and a few other
authors,10,20 the integrity and durability issues of
similar Al welds especially under cyclic loading
must be considered in the design of many struc-
tures. To the authors’ knowledge, there is no
systematic studies on fatigue behavior for such
USW similar thicker-sheet Al welds in relation to
the change in the failure mode that has been
reported in the open literature. This study was,
Microstructure and Fatigue Properties of Ultrasonic Spot Welded Joints
of Aluminum 5754 Alloy
Fig. 2. Grain size distribution of an USWed Al5754 similar joint at a welding energy of 2500 J: (a) BM and (b) near the nugget zone.
Fig. 3. Grain boundary mapping of an USWed Al5754 similar joint at a welding energy of 2500 J: (a) BM and (b) near the nugget zone.
therefore, aimed to gain a better understanding of
the lap shear tensile load and fatigue properties of
USWed-similar Al5754-O Al alloy joints, with a
particular emphasis on the weld microstructure
evolution.
MATERIAL AND EXPERIMENTAL
PROCEDURE
In the present study, commercial 1.5-mm-thick
sheet of Al alloy Al5754-O (3.42% Mg, 0.63% Mn,
0.23% Sc, 0.22% Zr, and balance Al) was selected for
USW. The specimens, 60 mm long and 15 mm wide,
were sheared and 120 emery papers were used to
ground the surfaces. The samples were cleaned and
dried before welding. The welding was performed
with a dual wedge-reed Sonobond-MH2016 HP-
1467
USW system. An energy input ranging from 500 J
to 3000 J, a constant power setting of 2000 W with
an impedance setting of 8 and a pressure of
0.414 MPa on the machine were used for welding
the samples. Five samples were welded in each
welding condition. One of them was used for
microstructural examination, specifically electron
back-scattered diffraction (EBSD), two were used
for the lap shear tensile tests and the remaining two
were used for fatigue tests. The metallographic
samples for the EBSD examination were cut from
the weld cross-section, then ground and polished
using diamond paste and MasterPrep. EBSD obser-
vations were performed on the as-polished material
using JEOL JSM-6380LV-HKL equipped with a
Nordlys detector for EBSD analysis. An acceleration
voltage of 20 keV and a working distance of 15 mm
1468
Fig. 4. Misorientation angle distribution of an USWed Al5754 similar joint at a welding energy of 2500 J: (a) BM and (b) near the nugget zone.
100
90
Lap shear strength, MPa
80
70
60
50
40
30
Interfacial
20 failure
10
0 sites and propagation mechanism of the fatigue
0 500 1000 1500 2000 2500 3000 3500
Weld energy, J
Fig. 5. Lap shear tensile strength of an USWed similar Al5754 alloy
at different energy inputs.
were used for the examination of sample
microstructural development. The sample was tilted
70° for the collection of data. Analysis of the EBSD
results was performed using MTEX software. To
evaluate the mechanical strength of the joints and
establish the optimum welding conditions, tensile
shear tests of the welds were conducted to measure
the lap shear failure load using a fully computerized
United testing machine with a constant crosshead
speed of 1 mm min 1 in air at room temperature. At
least two samples were tested at each energy level;
however, due to the experimental scatter at certain
energy levels, i.e., 1500 J and 2000 J, five samples
were tested. Load control fatigue tests were per-
formed in accordance with ASTM E466 on a fully
computerized Instron 8801 servo-hydraulic testing
Mirza, Macwan, Bhole, and Chen
system at different maximum loads. To avoid poten-
tial buckling of the test specimens, tension–tension
cyclic loading at a stress ratio of R (Pmin/Pmax) = 0.2
was applied at a frequency of 50 Hz and sinusoidal
waveform. Two samples were tested at each cyclic
load level. To prevent the rotation and bending
moment of the specimen during the tensile and
fatigue tests, two spacers with a thickness of
1.5 mm, width of 15 mm and length of 35 mm were
Nugget attached at both ends of the specimen. The fatigue
zone
failure fracture surfaces were examined via a scanning
electron microscope (SEM) to identify the initiation
crack.
RESULTS AND DISCUSSION
Microstructure Evolution
Figure 1a and b shows the EBSD orientation
image mapping of base metal (BM) and the center of
the weld nugget (WZ) of USWed-similar Al5754
alloy sample (2500 J energy input), respectively. It
can be seen from Fig. 1a that the BM exhibits
almost equiaxed grain structure. The average grain
size was approximately 8 lm (as shown in Fig. 2-
a). It can be seen from Fig. 1b that the material at
the center of the weld nugget exhibits slightly more
equiaxed and finer grain structure than the base
metal due to dynamic recrystallization.12,21–24 The
grain size distribution clearly indicates finer grain
size than base metal, i.e., approximately 7 lm as
shown in Fig. 2b. A similar type of fine grain
formation has been reported at the weld zone in
ultrasonic welding16 and friction stir welding25,26
for Al alloys. Dislocations would be introduced due
to severe plastic deformation during welding, fol-
lowed by quick recrystallization at higher frictional
Microstructure and Fatigue Properties of Ultrasonic Spot Welded Joints
of Aluminum 5754 Alloy
Fig. 6. Typical tensile failure locations (indicated by yellow arrows) of an USWed similar Al5754 alloy at different energy inputs: (a) 500 J, (b)
1000 J, (c) 1500 J, (d) 2000 J, (e) 2500 J, and (f) 3000 J.
temperature leading to the formation of fine grains
in and near the weld zone. It has been reported that
optimum weld strengths are obtained when exten-
sive plastic deformation takes place throughout the
entire sheet thickness between the sonotrode
tips.16,27 It has also been reported that discontinu-
ous dynamic recrystallization (DDRX) occurs in
metals of moderate to low stacking fault energy
during hot deformation.25 In DDRX, new, disloca-
tion-free grains form at sites such as prior grain
boundaries, deformation band interfaces or bound-
aries of newly recrystallized grains (Fig. 1b). The
driving force for the initiation of recrystallization is
the strain-induced build-up of stored strain energy
in the form of fine cells and a high free dislocation
density.25 During deformation, a low stacking fault
energy retards the climb and cross-slip processes,
thereby enabling the formation of stable grain
nuclei at various sites in the deformation
microstructure at the center of the nugget zone as
pointed out by the arrow (as shown in Fig. 1b). The
1469
newly formed grains deform as they grow, and the
build-up of stored strain energy in the new grains
may reduce the driving force for their continued
growth but increase the frequency of initiation of
new grains. A ‘‘necklace’’-type structure of newly
formed grains may result when prior grain bound-
aries are the initiation sites for DDRX. This is
reflected in almost equiaxed subgrains with nearly
dislocation-free interiors, constant subgrain size
and subgrain boundary misorientation of grains
near the weld zone (Fig. 3b) as compared with the
base metal (Fig. 3a). A condition for initiation of
DDRX may be given by,25
q3m 2cb
i ð1Þ
e_ KLMGb5
;
where qm is the mobile dislocation density, e_ is the
strain rate, cb is the grain boundary energy, K is a
constant fraction of the dislocation line energy that
is stored in the newly formed grains, L is mean slip
1470 Mirza, Macwan, Bhole, and Chen

Fig. 7. Typical SEM images of tensile fracture surface of an USWed Al5754 similar joint at a welding energy of 2500 J: (a) and (b) overall view
(with an arrow indicating the crack initiation), (c) and (d) at lower magnification, (e) and (f) at higher magnification.

distance of dislocations in these grains, M is the
boundary mobility, G is the shear modulus and b is
Burger’s vector. This inequality suggests a strong
dependence of the conditions for DDRX on the
mobile dislocation density: a low stacking fault
energy tends to suppress recovery, leading to high
values of qm .25 Generally, if the restricted number of
slip systems or solute drag inhibit the slip contri-
butions and boundary migration, lattice rotation
may develop progressively at grain boundaries as
dislocations accumulate during straining. This may
lead to dynamic recrystallization by lattice rotation
Microstructure and Fatigue Properties of Ultrasonic Spot Welded Joints
of Aluminum 5754 Alloy
3.5
(a)
3.0
Maximum load, kN 2.5
2.0
1.5
1.0
0.5
0.0
1E+0 1E+2 1E+4 1E+6
Number of cycles to failure, Nf
2.4
(b)
2 Interfacial failure mode
max
1.6
Log
Transverse through-thickness
crack growth mode
1.2
0.8
0 2 4 6
Log (2Nf)
Fig. 8. (a) S-N curves of USWed similar Al5754 alloy joints
tested at RT, R = 0.2, and 50 Hz, and (b) the maximum tensile
shear stress versus the number of reversals to failure (2Nf) in
the double-log scale for welded joints at a welding energy of
2000 J.
as new grains form at prior boundaries, and a
necklace structure similar to that often observed in
DDRX (Fig. 1b). In other words, the necklace-type
structure of the newly formed grains can be inter-
preted as the mechanical interlocking as reported
by Zhang et al.10 since the welding mechanisms of
the Al alloy involve solid-state deformation and
mechanical interlocking, as well as the formation
and progressive spreading of microwelds.16,28 This
is associated with a considerable rise in tempera-
ture in the weld zone, which softens the material
and allows the sonotrode tips to sink into the sheet
surfaces. In the process, the weld interface is
displaced into complex wave-like flow pat-
terns10,16,27,28 (Fig. 1b). Figure 3 shows the grain
boundary mapping of USWed-similar Al5754 alloy
of base metal and near the nugget zone. The
frequencies of misorientaion of grains are mostly in
the range of 35°–50° (as indicated by the black and
yellow lines) for both cases. The wave-like patterns
or ‘‘necklace’’-type grain structure is also obvious
from Fig. 3b and the misorientation angles in that
section are in the range of 55°–60°. The corre-
sponding misorientation angle distribution of
USWed-similar Al5754 alloy is also shown in
Fig. 4.
1471
Lap Shear Tensile Strength and Failure Mode
1000J
2000J The lap shear failure strength of USWed similar
Al5754 alloy as a function of welding energy is
shown in Fig. 5. The average maximum lap shear
strength was found to be 85 MPa at 2000 J energy
input. As shown in Fig. 5, the lap shear tensile
strength increased with increasing energy inputs
due to high temperatures and strain rate, which
leads to increased diffusion between the Al alloy
sheets. However, at high energy inputs, the lap
shear strength was decreased. At lower energy
inputs, temperature was not high enough to soften
the sample, thereby Al has not been fully deformed
to fill up the valleys of the knurl patterns of the
sonotrodes and exhibits very low strengths, i.e., 20–
40 MPa for 500 J and 1000 J. It was observed that,
in the cases of lower energy inputs, the lap shear
tensile fracture occurred from the Al/Al interface, as
shown in Fig. 6 (typical tensile failure locations of
USWed-similar Al5754 alloy at different energy
inputs). While on the other hand, at higher energy
inputs, the weld specimen was subjected to higher
temperatures under a larger vibration amplitude
for a longer time. As a result, more diffusion
occurred between the two Al alloy sheets, which
8 10 deteriorated the strength. This corresponded well to
the failure that occurred at the edge of the nugget
zone (Fig. 6f). Thus, it can be concluded that this
was a consequence of the competition between the
increasing diffusion bonding arising from higher
temperatures at the higher energy inputs and the
deterioration effect of the Al/Al interface.11 Similar
results have been reported by Bakavos and Prang-
nell16 and Zhang et al.10 For dissimilar Al alloy to
Mg alloy sheet welded joints, Patel et al.11 also
reported that the lap shear strength was increased
with increasing energy input and it was decreased
at very high energy inputs.
Lap Shear Tensile Fractography
Typical SEM images of tensile fracture surfaces of
USWed similar Al 5754 alloy at 2500 J energy
inputs are shown in Fig. 7. As seen from the overall
view of both sides of the joint (Fig. 7a and b, a bright
zone was observed at the outside periphery of the
weld. As identified by Zhang et al.,10 this zone is an
oxide layer formed during welding due to the direct
contact of air with the material at an elevated
temperature. It can also be seen from Fig. 7a and b
that there is a dark zone just inside the oxide zone
which represents the bonded zone (as indicated by
label ‘‘A’’). It can be seen from the low-magnification
images (Fig. 7a and b) that the crack initiated from
one end of the bonding zone as indicated by the
arrow, where a big crack can be seen from the
border area (Fig. 7b). The initiation site (bonded
zone) had obvious dimple-rupture failure features
as observed at higher magnifications (Fig. 7c and d).
At higher magnification (Fig. 7e and f), the elon-
gated dimples indicate the shear fracture via void
1472
Fig. 9. Typical fatigue failure locations (indicated by yellow arrows) at different Pmax values of an USWed similar Al5754 alloy at 2000 J energy
input: (a) 3 kN (interfacial failure mode), (b) 2.5 kN (interfacial failure mode), (c) 2 kN (interfacial failure mode, initiation of crack), (d) 1.5 kN
(interfacial failure mode, crack starting to propagating perpendicular to the loading direction), (e) 1 kN (transverse through-thickness crack growth
mode), and (f) 0.5 kN (transverse through-thickness crack growth mode).
formation/nucleation, growth, and coalescence.10
There are also several secondary cracks which
occurred during the tensile shear. This suggested
that this occurred where the base material locally
reached its fracture strength and the fracture
initiated in that area. Apart from some small,
isolated dimples and tearing ridges, the predomi-
nant feature in the center of the joint was the
scrubbing lines parallel to the loading direction
(Fig. 7e), which might be generated due to abrasion
or galling of asperities of the microscale roughness
of the ground sheet surfaces.29
Fatigue Behavior and Failure Mode
Fatigue tests of the similar USWed Al5754 Al
alloy joints were carried out at room temperature
at a stress ratio of R = 0.2, and a frequency of
50 Hz. The obtained S–N curves are plotted in
Fig. 8a. Like the tensile testing (Fig. 5), it can be
seen that at the higher Pmax loads of 2 kN, 2.5 kN,
and 3 kN, the 2000-J samples showed higher
fatigue resistance compared to the 1000-J welded
samples. However, lower fatigue life was obtained
for the for the 2000-J welded samples at lower
Pmax loads especially at 0.5 kN and 1 kN as
compared to the 1000-J welded samples. This
might due to the premature fatigue fracture
Mirza, Macwan, Bhole, and Chen
occurring at the edge of the nugget zone on the
base metal at lower Pmax loads as reported by Patel
et al.,3 which is referred to here as the transverse
through-thickness crack growth mode. Typically,
the transverse through-thickness crack growth
mode occurred (perpendicular to the loading direc-
tion) at the edge of the sonotrode in contact with
the sample due to the higher welding energies
(leading to greater temperatures and a greater
level of bending deformation in the weld zone).3 As
a result, a small micro-level crack tip at the notch
of the two sheets was produced and this crack tip
experienced a higher stress concentration effect
which allowed the cracks to grow in the transverse
direction of the Al sample. In the 2000-J USWed
sample, the results of the fatigue tests were
consistent with the tensile results, which showed
the highest fatigue life at the applied Pmax load of 2
kN or less. It can be noticed from Fig. 8a that the
fatigue limit was 0.5 kN for all welding conditions.
Figure 8b presents the failure mode in conjunction
with the logarithmic S–N plots in the form of the
maximum tensile shear cyclic stress versus the
number of reversals to failure (2Nf) in terms of a
Basquin-type relationship for the 2000-J USWed
samples. It should be noted that, since there were
no run-out data of non-failed samples at or over
1 9 107 cycles, all the data points were included in
Microstructure and Fatigue Properties of Ultrasonic Spot Welded Joints
of Aluminum 5754 Alloy
Fig. 10. Typical SEM images of fatigue fracture surface of an USWed Al5754 similar joint at a welding energy of 2000 J tested at a Pmax of 2.0
kN: (a) overall view, (b) crack initiation area, (c) magnified image of crack initiation area, and (d and e) fatigue striations in the crack propagation
zone at higher magnifications.
the curve fitting. For the 2000-J welded samples,
interfacial failure mode was observed at the higher
cyclic load levels where the slope was flatter, while
transverse through-thickness crack growth mode
was observed at the lower cyclic load levels where
the slope was steeper. Figure 9 shows a typical
fatigue failure location at the different Pmax load
values of USWed -imilar Al5754 alloy joint at
2000 J energy input. It can be seen that, when the
samples were tested at cyclic loads of 3 and 2.5 kN,
interfacial failure occurred. When samples were
tested at 2 kN and 1.5 kN, interfacial failure also
occurred; however, initiation of cracks was started
at the circumference around the nugget (Fig. 9c
and d). At 1 kN applied load, the crack was
propagated perpendicular to the loading direction.
1473
Now, shear stress acting on the area was such that
the remaining cross-section could not withstand
the shear overload, and the severely deformed
nugget edge in due course experienced transverse
through-thickness crack growth failure mode
(Fig. 9e and f).
On the other hand, in the 1000-J samples, most of
the failure occurred at the interface (except for the
0.5-kN applied fatigue load where the failure was
through-thickness). The 1000-J welded samples
experienced lower sonotrode penetration (i.e., no
micro-level cracks at the edge of the nugget zone).
The change of the failure mode from the transverse
through-thickness crack growth mode to the inter-
facial failure occurred at logrmax = 1.49 or rmax =
31 MPa in the 2000-J USWed joints (Fig. 8b).
1474
Likewise, such a transition of failure mode from the
deformed edge failure to the interfacial failure
occurred at logrmax = 1.32 or rmax = 21 MPa in the
1000-J samples (not shown in Fig. 8b). Therefore, it
can be concluded that the maximum cyclic stress for
the transition of fatigue fracture mode from the
transverse through-thickness crack growth to the
interfacial failure increased with increasing energy
input.
Fatigue Fractography
For a more detailed study of the fracture frac-
tures, SEM fracture surface images of the failed
2000-J welded samples tested at a minimum cyclic
load of 2 kN are shown in Fig. 10. It can be seen
from Fig. 10a that the fatigue crack developed from
the region ‘‘b’’ and propagated at right angles to the
loading direction along the Al sheet with increasing
number of cycles, and later experienced transverse
through-thickness crack growth mode (Fig. 9e).
Figure 10b shows the location of the crack initiation
and Fig. 10c shows the magnified image of region
‘‘c’’ in Fig. 10b. Fatigue crack propagation was
mainly characterized by the formation of fatigue
striations (Fig. 10d and e), which usually appeared
perpendicular to the fatigue crack propagation
direction. Fatigue striations normally occur by a
repeated plastic blunting–sharpening process in the
face-centered cubic materials, stemming from the
glide of dislocations on the slip plane along the slip
direction within the plastic zone ahead of the
fatigue crack tip.14
CONCLUSION
The following conclusions can be drawn from this
investigation:
(1) A special ‘‘necklace’’-type structure containing
newly formed very fine grains was observed
along the weld line after USW due to the
occurrence of dynamic crystallization, along
with the presence of mechanical interlocking.
(2) The lap shear tensile strength increased with
increasing energy inputs and reached the
maximum of 85 MPa at an energy input of
2000 J and then decreased. The lap shear
tensile fracture occurred from the Al/Al inter-
face in the cases of lower energy inputs,
whereas at higher energy inputs the failure
of the welded specimen occurred at the edge of
the nugget zone.
(3) In both welding conditions, the fatigue limit
was found to be 0.5 kN. The maximum cyclic
stress at which transition of the fatigue frac-
ture mode from the transverse through thick-
ness crack growth to the interfacial failure
increased with increasing energy input. Fati-
gue crack propagation was mainly character-
ized by the formation of fatigue striations,
Mirza, Macwan, Bhole, and Chen
which usually appeared perpendicular to the
fatigue crack propagation direction.
ACKNOWLEDGEMENTS
The authors would like to thank the Natural
Sciences and Engineering Research Council of Ca-
nada (NSERC) and AUTO21 Network of Centres of
Excellence for providing financial support. The au-
thors thank Dr. A.A. Luo, Ohio State University
(formerly General Motors Research and Develop-
ment Center) for the supply of test materials. One of
the authors (D.L. Chen) is grateful for the financial
support by the Premier’s Research Excellence
Award (PREA), NSERC-Discovery Accelerator
Supplement (DAS) Award, Canada Foundation for
Innovation (CFI), and Ryerson Research Chair
(RRC) program. The authors would also like to
thank Messrs. A. Machin, Q. Li, J. Amankrah and
R. Churaman for easy access to the laboratory
facilities of Ryerson University and their assistance
in the experiments.
REFERENCES
1. S. Chu and A. Majumdar, Nature 488, 294 (2012).
2. M. McNutt, Science 341, 435 (2013).
3. V.K. Patel, S.D. Bhole, and D.L. Chen, Metall. Mater. Trans.
A 45, 2055 (2014).
4. G. Wagner, F. Balle, and D. Eifler, JOM 64, 401 (2012).
5. P.B. Prangnell, F. Haddadi, and Y.C. Chen, Mater. Sci.
Technol. 27, 617 (2011).
6. L. Xiao, L. Liu, Y. Zhou, and S. Esmaeili, Metall. Mater.
Trans. A 41–6, 1511 (2010).
7. D. Bakavos and P.B. Prangnell, Sci. Technol. Weld. Join. 14,
443 (2009).
8. T. Liyanage, J. Kilbourne, A.P. Gerlich, and T.H. North, Sci.
Technol. Weld. Join. 14, 500 (2009).
9. P.B. Prangnell and D. Bakavos, Mat. Sci. Forum. 638–642,
1237 (2010).
10. C.Y. Zhang, D.L. Chen, and A.A. Luo, Weld. J. 93, 131s
(2014).
11. V.K. Patel, S.D. Bhole, and D.L. Chen, Sci. Technol. Weld.
Join. 17, 202 (2012).
12. L.E. Murr, G. Liu, and J.C. McClure, J. Mater. Sci. Lett. 16,
1801 (1997).
13. R. Jahn, R. Cooper, and D. Wilkosz, Metall. Mater. Trans. A
38, 570 (2007).
14. S.H. Chowdhury, D.L. Chen, S.D. Bhole, X. Cao, and P.
Wanjara, Mater. Sci. Eng. A 556, 500 (2012).
15. S. Elangovan, S. Semeer, and K. Prakasan, J. Mater. Pro-
cess. Technol. 209, 1143 (2009).
16. D. Bakavos and P.B. Prangnell, Mater. Sci. Eng. A 527, 6320
(2010).
17. A. Macwan, X.Q. Jiang, and D.L. Chen, JOM 67, 1468
(2015).
18. A. Macwan and D.L. Chen, Mater. Des. 84, 261 (2015).
19. J. Schneider and R. Radzilowski, JOM 66, 2123 (2014).
20. S.M. Allameh, C. Mercer, D. Popoola, and W.O. Soboyejo,
J. Eng. Mater. Technol. 127, 65 (2005).
21. N. Pathak, K. Bandyopadhyay, M. Sarangi, and S.P.
Kumar, JMEPEG 22, 131 (2013).
22. A. Panteli, Y.C. Chen, D. Strong, X. Zhang, and P.B.
Prangnell, JOM 64, 414 (2012).
23. F. Khodabakhshi, M. Haghshenas, S. Sahraeinejad, J.
Chen, B. Shalchi, J. Li, and A.P. Gerlich, Mater. Charact.
98, 73 (2014).
24. L. Ying, L.E. Murr, and J.C. McClure, Mater. Sci. Eng. A
271, 213 (1999).
Microstructure and Fatigue Properties of Ultrasonic Spot Welded Joints 1475
of Aluminum 5754 Alloy
25. Z. Zhang, B.L. Xiao, and Z.Y. Ma, Mater. Charact. 106, 255 27. Y.C. Chen, D. Bakavos, A. Gholinia, and P.B. Prangnel, Acta
(2015). Mater. 60, 2816 (2012).
26. H.E. Hu, L. Zhen, B.Y. Zhang, L. Yang, and J.Z. Chen, 28. F. Haddadi, D. Strong, and P.B. Pragnell, JOM 64, 407 (2012).
Mater. Charact. 59, 1185 (2008). 29. H.P.C. Daniels, Ultrasonics 3, 190 (1965).
Materials 2015, 8, 5138-5153; doi:10.3390/ma8085138
OPEN ACCESS

materials
ISSN 1996-1944
www.mdpi.com/journal/materials
Article

A Unified Model for the Prediction of Yield Strength in

Particulate-Reinforced Metal Matrix Nanocomposites
F. A. Mirza and D. L. Chen *

Department of Mechanical and Industrial Engineering, Ryerson University, 350 Victoria Street,
Toronto, ON M5B 2K3, Canada; E-Mail: f4mirza@ryerson.ca

* Author to whom correspondence should be addressed; E-Mail: dchen@ryerson.ca;

Tel.: +1-416-979-5000 (ext. 6487); Fax: +1-416-979-5265.

Academic Editor: Geminiano Mancusi

Received: 9 July 2015 / Accepted: 5 August 2015 / Published: 10 August 2015

Abstract: Lightweighting in the transportation industry is today recognized as one of the

most important strategies to improve fuel efficiency and reduce anthropogenic climate-changing,
environment-damaging, and human death-causing emissions. However, the structural
applications of lightweight alloys are often limited by some inherent deficiencies such as
low stiffness, high wear rate and inferior strength. These properties could be effectively
enhanced by the addition of stronger and stiffer reinforcements, especially nano-sized
particles, into metal matrix to form composites. In most cases three common strengthening
mechanisms (load-bearing effect, mismatch of coefficients of thermal expansion, and
Orowan strengthening) have been considered to predict the yield strength of metal matrix
nanocomposites (MMNCs). This study was aimed at developing a unified model by taking
into account the matrix grain size and porosity (which is unavoidable in the materials
processing such as casting and powder metallurgy) in the prediction of the yield strength of
MMNCs. The Zener pinning effect of grain boundaries by the nano-sized particles has also
been integrated. The model was validated using the experimental data of magnesium- and
titanium-based nanocomposites containing different types of nano-sized particles (namely,
Al2O3, Y2O3, and carbon nanotubes). The predicted results were observed to be in good
agreement with the experimental data reported in the literature.

Keywords: metal matrix nanocomposites; Orowan strengthening effect; Hall-Petch

relationship; Zener pinning effect
Materials 2015, 8 5139

1. Introduction

Lightweighting in ground vehicles is deemed as one of the most effective strategies to improve fuel
economy and reduce anthropogenic climate-changing, environment-damaging, costly, and human
death-causing emissions [1–7] due to the tremendous environmental concerns. To manufacture lightweight
vehicles, advanced high-strength steels, aluminum alloys, magnesium (Mg) alloys, and polymers are
being used in the automotive and aerospace sectors, but substantial weight reductions could be further
achieved by employing ultra-lightweight Mg alloys due to their low density, high strength-to-weight ratio,
and superior damping capacity [1,4,8,9]. However, the applications of Mg alloys are often restricted by
some inherent deficiencies such as low stiffness, high wear rate, and inferior creep resistance or
mechanical strength at elevated temperatures [10,11]. Reinforcement with a discontinuous phase
especially nano-sized particles has been considered to be the most favored choice by researchers in recent
years to improve physical, mechanical, and damping properties of Mg beyond the limits dominated by
traditional alloying [11–19].
In the past few years, several studies have been done to develop constitutive relationships that can be
used to predict the mechanical properties of metal matrix nanocomposites (MMNCs) as a function of
the reinforcement, matrix, and processing conditions [10,15,20–27]. Ramakrishnan [20] predicted the
yield strength of micro-sized particulate-reinforced metal matrix composites (MMCs), using a composite
sphere model for the intra-granular type of MMCs and incorporating two improvement parameters
associated with the dislocation strengthening of the matrix and the load-bearing effect of the reinforcement.
Recently, Zhang and Chen [10,21] predicted the yield strength of MMNCs via considering the Orowan
strengthening mechanism, enhanced dislocation density due to the mismatch of coefficients of thermal
expansion (CTE) between the reinforcement and matrix, and load-bearing effect which has been used
by many researchers, e.g., [15,23–27], to predict the yield strength of particulate-reinforced MMNCs.
However, in all of the above models no effect of porosity was taken into account, which could lead to
an overestimate of the yield strength of composites. In the composite processing (casting, powder
metallurgy, electrodeposition, plasma and cold spray, etc.), the complete elimination of porosity is either
difficult or impossible [28–37]. Porosity is among the dominant factors causing failure of discontinuous
reinforced metal matrix composites (DRMMCs) in the tensile tests [17]. The commonly encountered
spherical pores or gas porosity are observed to create stress concentrations and thus lead to failure.
The ductility and toughness of most metal based composites are influenced by the presence of voids and
the balance between reinforcing particles sharing the load [14]. The large amount of porosity associated
with oxides is often surrounded by either individual or clusters of reinforcement particles, which
significantly decreases the mechanical properties of DRMMCs [32]. Also, it was reported that the
presence of nanoparticles led to a higher level of porosity and more irregular pores with bigger pore size
in the nanocomposites [34]. Thus, it is necessary to consider the effect of porosity on the mechanical
properties of nanocomposites.
Furthermore, very limited investigations have been reported to account for the influence of matrix
grain size in the MMNCs, while the relation between the yield or flow stress and grain size in
polycrystalline alloys was well established by Hall and Petch [38,39]. Recrystallization studies in
DRMMCs have shown that the nucleation potency increases with increasing reinforcement size and
volume fraction, called particle stimulated nucleation (PSN) [40,41]. It would be expected that the
Materials 2015, 8 5140

nucleation density is high and the recrystallized grain size is smaller than the interparticle spacing, since
normal grain growth continues after complete recrystallization until the grain boundaries are pinned by
the reinforcement particles. As reported by Hassold et al. [42], the inert, second-phase particles can
inhibit grain growth and lead to a pinned microstructure where grain growth ceases (as shown in Figure 1).
Most particles are located on the grain boundaries, which are strongly pinned by the particles. The grain
boundaries are highly non-random in location and the particles are more closely spaced as the volume
fraction increases. Also, the pinned grain size decreases with increasing volume fraction of particles
(Figure 1b). Zener [40–45] first observed this effect, commonly called Zener pinning (or Zener drag) [41,45],
and predicted the dependence of the pinned grain size on the volume fraction and radius of the particles.
The presence of Zener pinning in the MMNCs will further decrease the matrix grain size, depending on
the second-phase particles size and volume fraction. The objective of this study was, therefore,
to develop a unified model of integrating all the above strengthening effects to achieve a more accurate
prediction of the yield strength of MMNCs, which was then validated using the experimental data of
magnesium-based nanocomposites available in the literature.

Figure 1. Pinned microstructures of nanocomposites for two volume fractions of particles

of (a) Vp = 0.01 and (b) Vp = 0.05 [42].

2. Model Development

To facilitate a better understanding of an analytical model for predicting the yield strength of
MMNCs, a brief review on the underlying factors affecting the yield strength of composites is given
as follows.

2.1. Load Transfer to the Reinforcement Particles

Due to the nano-size of the reinforcement particles and the sound synthesizing methods, there is a
strong cohesion at the atomic level between the matrix and nano-sized particles, i.e., the nano-sized
particles are directly bonded to the matrix [10,12,20–22,46]. Then the nano-sized particles would carry
a certain portion of load depending on the volume fraction, giving rise to some degree of load transfer
between matrix and reinforcement [10,21,22]. Provided that the porosity is present in the composites,
the porosity would affect the load-bearing term. Thus, the improvement factor associated with the
load-bearing effect of reinforcement in the presence of porosity can be derived as [46],
Materials 2015, 8 5141

Vp  P
1
fl (1)
2
where fl is an improvement factor associated with load-bearing effect, P is the volume fraction of porosity,
and Vp is the volume fraction of the reinforcement nano-particles.

2.2. Dislocation Density

In MMNCs, the increased interfacial area between the reinforcement and matrix contributes to the
enhanced mechanical properties due to the presence of nano-sized particles [10,21]. Also, thermal
mismatch dislocations in the matrix around the nano-sized particles would be generated to relieve the
thermal stresses occurred at the interface during cooling from the processing temperature. The thermal
stresses around the nanoparticles would be large enough to induce plastic deformation in the matrix near
the interface region [13,47,48]. The improvement factor related to the dislocation density in the matrix,
fd can be expressed as follows [10,21,46],
fd kGm b ȡ / ı ym (2)

12'Į'TVp
bd p (1  Vp )
ȡ (2a)

where ıym is the yield strength of the monolithic matrix without any porosity, Gm is the shear modulus
of the matrix, b is the Burgers vector of the matrix, k is a constant (approximately equal to 1.25, based
on the theoretical estimates in reference [10]), ȡ is the enhanced dislocation density due to the difference
in the coefficients of thermal expansion between the reinforcement phase and the matrix (ǻĮ), dp is the
particle size, and ǻT is the difference between the processing and test temperatures.

2.3. Orowan Strengthening

In the precipitation-hardened alloys the internally precipitated nano-sized particles are proven to be
either sheared (for smaller and softer particles) or by-passed by dislocations (i.e., Orowan strengthening
mechanism of dislocation bowing) for larger and harder particles [37]. In the case of MMNCs where the
hard and external nano-sized particles are added, it would be reasonable to assume that the Orowan
strengthening mechanism would occur [10,21]. The improvement factor related to the Orowan
strengthening of nanoparticles added, fOrowan can be expressed as follows [10,21,46],
0.13Gmb r
fOrowan ln (3)
Ȝı ym b

ª º
§ 1 ·3 »
Ȝ | d p ««¨
1

¨ 2V ¸¸  1»
«¬© p ¹ »¼
(3a)

where r is the particle radius (r = dp/2), and Ȝ is the interparticle spacing.

Materials 2015, 8 5142

2.4. Porosity

As mentioned above, some inherent characteristics such as porosity were inevitably present in
MMNCs which would have a significant effect on the yield strength of MMNCs [28–37]. For example,
the occurrence of porosity as discontinuities in cast MMC interrupted the balance between the
reinforcing particles carrying the load, generated a stress concentration and facilitated the crack initiation
and propagation, and thus reduced its mechanical properties [32]. Also, most fatigue cracks initiated
from the discontinuities in materials, mainly in the highly stressed regions of components. With
increasing volume fraction of reinforcement particles, the likelihood of forming the processing-induced
voids became higher, leading to a degradation of the yield strength [12]. Accordingly, it is necessary to
consider the effect of porosity in more realistically predicting the yield strength of MMNCs. Based on
our previous publication [46], fporosity is the deterioration factor associated with the presence of porosity
in MMNCs, which could be expressed as [29],
f porosity 1  e nP (4)

where n is an empirical constant depending on the porosity characteristics such as pore size, geometry,
and orientation [29]. If the average pore shape was assumed to be close to cylinder orientated between
45° and 90° with respect to the loading axis, n could be estimated as [29],

 0.318  1.22
dp
n 0.405 l (4a)
dp l

where l is the length of the particle, respectively.

Based on the above mechanisms, the concept of multiplication as reported by Zhang and Chen [10,21],
now referred to as the “Zhang and Chen method”, “Zhang and Chen approach”, “Zhang and Chen
model”, or “Zhang and Chen (ZC) summation method”, e.g., in [26,49–53], was used to account for both
additive and synergistic effects of the strengthening and weakening factors, and a further modified
analytical model for predicting the yield strength of MMNCs with consideration of the effect of porosity
was proposed as follows [46],
ı yc ı ym (1  f l )(1  f d )(1  f O ro w an )(1  f p o ro sity ) (5)
where ıyc is the yield strength of MMNCs. It should be noted that since the yield strength of MMNCs is
considered in this model, no strain hardening effect during the subsequent plastic deformation beyond
yielding is taken into account.

2.5. Effects of Grain Size on the Yield Strength of MMNCs

The influence of grain size on the mechanical properties is complex since the grain boundaries may
either act as obstacles to dislocation slip (strengthening effect) or provide a positive contribution to the
deformation of the material (softening effect). However, the following well-known Hall-Petch relation
between the yield stress and grain size was proposed [38,39,54],
ı YS V o  kd  1/ 2 (6)
where d is the grain size, k and ıo are constants. This equation was revisited Li et al. in terms of the
collective motion of interacting dislocations and rearranged as follows [55],
Materials 2015, 8 5143

ıYS ıo  KG b d (6a)

where K is constant, G is the shear modulus, and b is the burgers vector. The typical value of K for pure
fcc metals is 0.05 to 0.5 [54,56].
As mentioned earlier, the second-phase particles can impede grain growth and lead to a pinned
microstructure where grain growth ceases [40]. The dependence of the pinned grain size on the volume
fraction and radius of the particles could be expressed as follows [40,42–44],
4rp
R (7)
3Vp

where R is the matrix grain radius. By combining Equations (6a) and (7), a modified Hall-Petch equation
with consideration of Zener pinning is obtained as follows,
§ 4d ·
ı o  KG b ¨ p ¸¸
¨ 3Vp
ı YS
© ¹
(8)

and the improvement factor can be expressed as,

f H all-P etch -Z en er ' ı H all-P etch -Z en er / ı ym (9)
where ' ı H all-P etch-Zener is,

'ı Hall-Petch-Zener
3bVp
KGm (9a)
4d p

Based on this additional consideration of the grain growth retardation by dispersed-particle-pinning

of grain boundaries, the analytical model for predicting the yield strength of MMNCs becomes,
ı yc ı ym (1  f l )(1  f d )(1  f O ro w an )(1  f H all-P etch -Z en er )(1  f P o ro sity ) (10)
Now substituting Equations (1)–(4a) and (9) into Equation (10) and considering ' T T p ro c e ss  T te st
and ' Į Į m  Į p (the difference in the coefficients of thermal expansion between the reinforcement phase
and the matrix), the following equation for the yield strength of MMNCs can be derived,
§ AB ·§ C ·  nP
(1  0.5Vp  P) ¨ ı ym  A  B  ¸¨ 1 ¸¸ e
¨ ı ym ¸¨
ı yc
© ¹© ı ym ¹
(11)

where A, B, and C can be expressed as follows,

12(Tprocess  Ttest )(Įm  Įp )Vp
bdp (1  Vp )
A 1.25Gmb (11a)

0.13Gmb d
ª º 2b
B ln p
« § 1 ·3 »
1

dp «¨  1»
¨ 2V ¸¸
(11b)

«¬© p ¹
»¼
Materials 2015, 8 5144

3bVp
C KGm (11c)
4d p

3. Results and Model Validation

First of all, the individual contribution of all improvement factors (using Equations (1)–(3) [10,21,46],
and (9)) with respect to the size and volume fraction of nanoparticles in Mg/Al2O3 nanocomposites is
identified and plotted in Figures 2 and 3. Considering Vp + Vm + P = 1 and substituting Equation (4a)
into Equation (4), the following equation for fporosity can be derived,
§ ·

¨ 0.405 l  0.318 p 1.22 ¸ 1Vp Vm 
f porosity 1  e
d
© dp l ¹ (12)

where Vm is the volume fraction of matrix. Figure 2 shows the variation of the improvement factors with
the nanoparticle size with specific volume fractions of 0.03 and 0.01 for nanoparticles and porosity in
Mg/Al2O3 nanocomposites as an example. It is seen that the contributions of fl and fporosity are fairly small;
the contributions of fd, fOrowan, and fHall-Petch-Zener are relatively large and in the sequence of
fd > fOrowan > fHall-Petch-Zener which increase monotonically and strongly with decreasing size of
nanoparticles. It is also observed that fOrowan and fHall-Petch-Zener increase more rapidly when the
nanoparticle size becomes very small (e.g., <50 nm). As seen from Figure 3 the change of the
improvement factors with the volume fraction of nanoparticles illustrates similar findings. That is, fl and
fporosity do not exhibit a significant influence, while fd, fOrowan, and fHall-Petch-Zener monotonically increase
with increasing volume fraction of nanoparticles in the same order of fd > fOrowan > fHall-Petch-Zener.

1
fl Mg/Al2O3
Vp= 0.03
fd
0.8
Improvement factor

fOrowan

0.6

0.4

0.2

0
0 100 200 300 400 500 600
Size of nanoparticles, nm

Figure 2. The improvement factors as a function of nanoparticle size with a volume fraction
of 0.03 and 0.01 for nanoparticles and porosity, respectively, in Mg/Al2O3 nanocomposites.
Materials 2015, 8 5145

1.5
fl Mg/Al2O3
fd dp=50 nm

1.2 fOrowan

Improvement factor
fporosity

0.9

0.6

0.3

0
0 0.02 0.04 0.06 0.08 0.1
Volume fraction of nanoparticles

Figure 3. The improvement factors as a function of volume fraction of nanoparticles with a

particle size of 50 nm and a volume fraction of 0.01 for porosity in Mg/Al2O3 nanocomposites.

To further indicate the changes of the yield strength of MMNCs with the volume fraction of
nanoparticles (Vp), the same set of data for the Al2O3-reinforced Mg nanocomposites tested at room
temperature in reference [46] is selected here, which was originally from Gupta and co-workers [11,12]
together with the relevant values from Brassell et al. [29]: ıym = 97 MPa, Gm = 16.5 GPa, b = 0.32 nm,
Įm= 28.4 × 10í6 (°C)í1, Įp = 7.4 × 10í6 (°C)í1, Tprocess = 300 °C, Ttest = 20 °C, M = 3.06, k = 70 MPa¥ȝm,
dp = 50 nm, K = 0.05 [54,56], and n = 1.94 (according to Equation (4a) for equiaxed particles where l is
equal to dp). In view of the change of porosity amount present in different composite fabrication
processes, e.g., 0.07–1.04 vol% [12], 0.25–1.15 vol% [14], 3–12 vol% [29], 0–1.18 vol% [34], up to
12.25 vol% [32], about 6 vol% [57], up to 13 vol% [33], and up to 12.45 vol% [58], three typical porosity
values of 1 vol%, 3 vol%, and 5 vol% were selected in the present model calculation.
To verify the validity of the present model with consideration of grain size and porosity, comparisons
were made between the model prediction and experimental data of three types of Mg-based
nanocomposites containing Al2O3, carbon nanotube (CNT), and Y2O3, respectively. Prior to doing so,
a comparison between the present model prediction and those in the literature [10,15,20,59–63] is shown
in Figure 4. It is seen that the present model shows a similar trend to those of other models, and lies
indeed in-between the other models.
Materials 2015, 8 5146

300

250

Yield Strength, MPa 200

150

100
Present model (P=1%) Present model (P=3%)
Present model (P=5%) Zhang and Chen's model [10]
Zhong's model [15] Ramkrishnan's model [20]
50 Geranmayeh's model [59] Tohgo's model [60]
Goh's model [61] Asgharzadeh's model [62]
Yazdi's model [63]
0
0 0.005 0.01 0.015
Volume fraction of nanoparticles

Figure 4. Comparison of the present model with different models reported in

literature [10,15,20,59–63].

First, the yield strength predicted for carbon nanotube (CNT)-reinforced Mg nanocomposites is shown
in Figure 5, together with the experimental data reported in refs. [14,16–18]. In the calculation of yield
strength from Equation (11) the following data for the CNT-reinforced Mg nanocomposites tested at
room temperature were used [11,12,29]: ıym = 97 MPa, Gm= 16.5 GPa, b = 0.32 nm, Įm = 28.4 × 10í6 (°C)í1,
Įp = í1.52 × 10í6 (°C)í1 (thermal contraction of CNT occurred, leading to a negative Į value [64]),
Tprocess = 350 °C, Ttest = 20 °C, K = 0.05 [54,56], n = 1.94 (for equiaxed particles), dp = 30 nm, P = 1%,
3% and 5%. As shown in Figure 5, due to the considerations of porosity weakening effect and grain
growth retardation by dispersed-particle-pinning of grain boundaries, the predicted yield strength for
P = 1%, 3% and 5% in the present model lies almost in-between those of Zhang and Chen [10,21] and
Ramakrishnan’s [20] model.
However, the predicted yield strength in a lower porosity level shows a better agreement with Zhang
and Chen’s model [10,21]. The first and second experimental data were slightly lower and higher than
that predicted from all methods including the present model, since all nanoparticles were assumed to be
spherical like Zhang and Chen model [10,21]. The scatter of the experimental data might also be related
to the waviness (or curviness) and agglomeration of CNTs [17,18,64–66], since the uniform dispersion
of such tiny nano-sized particles in nanocomposites still poses a significant challenge [65]. Next, four
data points showed a close agreement with the present model. It follows that incorporating the effects of
porosity and dispersed-particle-pinning of grain growth in the analytical models could improve
the predictability.
Materials 2015, 8 5147

250

200

Yield strength, MPa

150

100
Present Model (P=1%) Present Model (P=3%)
Present model (P=5%) Zhang and Chen's model [10]
50 Ramkrishnan's model [20] Experimental data-1 [14]
Experimental data-2 [16] Experimental data-3 [17]
Experimental data-4 [18]
0
0 0.001 0.002 0.003
Volume fraction of nanoparticles

Figure 5. Comparison of the present model (solid curves) with the experimental data
reported in [14,16–18], along with Zhang and Chen’s model [10] and Ramakrishnan’s
model [20], for CNT-reinforced Mg nanocomposites.

To further examine the trend of the yield strength of MMNCs, another prediction of yield strength
for Al2O3-reinforced Mg nanocomposites is shown in Figure 6, together with the experimental data
reported in References [11,12,66–69]. The same set of data for the Al2O3-reinforced Mg nanocomposites
tested at room temperature in reference [46] is first selected here, which was originally from Gupta and
co-workers [11,12] together with the relevant values from Brassell et al. [29]: ıym = 97 MPa, Gm = 16.5 GPa,
b = 0.32 nm, Įm= 28.4 × 10í6 (°C)í1, Įp = 7.4 × 10í6 (°C)í1, Tprocess = 300 °C, Ttest = 20 °C, Į = 0.3, Į1 = 0.35,
ȕ = 1.25, ij = 6 for Mg, k = 70 MPa¥ȝm, dp= 50 nm, K = 0.05 [54,56], n = 1.94 (according to
Equation (4a) for equiaxed particles), and P = 1%, 3% and 5%. Again, the yield strength predicted via
the present model is in fairly good agreement with the experimental data reported in
references [11,12,66–69]. Furthermore, with a higher amount of nanoparticles the dependence of the
yield strength on the volume fraction of porosity became stronger. The scatter of the experimental data
shown in Figure 6 would be associated with the processing techniques and conditions/parameters during
the fabrication of Mg-Al2O3 nanocomposites [11,12,66,67].
A further comparison involves Y2O3 particulate-reinforced Ti nanocomposite, as shown in Figure 7,
in which the following parameters were used in the calculation [70–73]: ıym = 450 MPa, Gm = 44.8 GPa,
b = 0.29 nm, Įm= 11.9 × 10í6 (°C)í1, Įp = 9.3 × 10í6 (°C)í1, Tprocess = 827 °C for particle size dp = 10, 12,
20, and 22 nm and Tprocess = 900 °C for particle size dp = 10, 40, and 170 nm, Ttest = 20 °C, K = 0.05 [54,56],
n = 1.94 (for equiaxed particles), P = 1%.
On the basis of the values of the weight fraction given in [70], the following converted values of
volume fraction/percent Vp = 0.25%, 0.27%, 0.38%, 0.41%, 0.54%, and 0.59% were utilized. The
straight line with a slope of m = 1 was drawn to show the value deviation. It is seen from Figure 7 that
Materials 2015, 8 5148

the present model with the consideration of both dispersed-particle-pinning of grain boundaries and
porosity predicts the yield strength of the Ti-Y2O3 nanocomposites fairly nicely, where a combined effect
of varying volume fraction of nanoparticles, thermo-mechanical treatment, and microstructures has been
taken into consideration.

250

200
Yield Strength, MPa

150

100
Present Model (P=1%)
Present Model (P=3%)
Present model (P=5%)
Zhang and Chen's model [10]
50 Ramkrishnan's model [20]
Experimental data-1 [11]
Experimental data-2 [12]
Experimental data-3 [66]
Experimental data-4 [67]
0
0 0.005 0.01 0.015
Volume fraction of nanoparticles

Figure 6. Comparison of the present model (solid curves) with the experimental data
reported in references [11,12,66–69], in conjunction with Zhang and Chen’s model [10] and
Ramakrishnan’s model [20], for Al2O3-reinforced Mg nanocomposites tested at 20 °C.

900
Predicted yield strength, MPa

800

700

600

500

400
400 500 600 700 800 900
Experimental yield strength, MPa

Figure 7. A comparison of the prediction via the present model with the experimental data
of Y2O3-reinforced Ti nanocomposites tested at 20 °C.
Materials 2015, 8 5149

4. Conclusions

While it was confirmed that the Orowan strengthening plays a significant role in MMNCs, the above
comparisons between the present model prediction and the experimental data reported in the literature
corroborate that both the positive and fairly strong effect of grain size refinement (or Hall-Petch
equation) and the negative effect of porosity should also be taken into consideration in predicting the
yield strength of MMNCs, although the effect of a small amount of porosity is rather small. The Zener
pinning effect of grain boundaries by the nano-sized particles has been integrated into the present
modelling as well. The proposed model has been validated using the experimental data of a number of
MMNCs containing different types of nano-sized particles, and it shows fairly good agreement.
This suggests that the positive effects of the Orowan strengthening mechanism, enhanced dislocation
strengthening mechanism and load-bearing effect of the reinforcement, and grain size effect, as well as
the negative weakening effect of porosity all need to be considered.

Acknowledgments

The authors would like to thank the financial support provided by the Natural Sciences and
Engineering Research Council of Canada (NSERC), Premier’s Research Excellence Award (PREA),
NSERC-DAS Award, AUTO21 Network of Centers of Excellence, and Ryerson Research Chair
program for providing financial support. The helpful discussion with Sanjeev Bhole is gratefully
acknowledged. This paper is presented at the 20th International Conference on Composite Materials
(ICCM20), Copenhagen, Denmark, July 19–24, 2015.

Author Contributions

F.A.M. and D.L.C. conceived, analyzed the data, and wrote the paper.

Conflicts of Interest

The authors declare no conflict of interest.

References

1. Pollock, T.M. Weight loss with magnesium alloys. Science 2010, 328, 986–987.
2. Chu, S.; Majumdar, A. Opportunities and challenges for a sustainable energy future. Nature 2012,
488, 294–303.
3. Zheng, X.Y.; Lee, H.; Weisgraber, T.H.; Shusteff, M.; DeOtte, J.; Duoss, E.B.; Kuntz, J.D.;
Biener, M.M.; Ge, Q.; Jackson, J.A.; et al. Ultralight, ultrastiff mechanical metamaterials. Science
2014, 344, 1373–1377.
4. Nie, J.F.; Zhu, Y.M.; Liu, J.Z.; Fang, X.Y. Periodic segregation of solute atoms in fully coherent
twin boundaries. Science 2013, 340, 957–960.
5. Underwood, E. Models predict longer, deeper U.S. droughts: Future western “megadroughts” could
be worse than ever. Science 2015, 347, 707.
6. McNutt, M. Climate change impacts. Science 2013, 341, 435.
Materials 2015, 8 5150

7. Schmale, J.; Shindell, D.; von Schneidemesser, E.; Chabay, I.; Lawrence, M. Air pollution:
Clean up our skies. Nature 2014, 515, 335–337.
8. Luo, A.A. Magnesium: Current and potential automotive applications. JOM 2002, 54, 42–48.
9. Kainer, K.U. Magnesium—Alloys and Technology; Wiley-VCH Verlag GmbH & Co. KGaA:
Weinheim, Germany, 2003.
10. Zhang, Z.; Chen, D.L. Consideration of Orowan strengthening effect in particulate-reinforced metal
matrix nanocomposites: A model for predicting their yield strength. Scr. Mater. 2006, 54, 1321–1326.
11. Hassan, S.F.; Gupta, M. Development of high performance magnesium nanocomposites using
solidification processing route. Mater. Sci. Tech. 2004, 20, 1383–1388.
12. Wong, W.L.E.; Karthik, S.; Gupta, M. Development of hybrid Mg/Al2O3 composites with improved
properties using microwave assisted rapid sintering route. J. Mater. Sci. 2005, 40, 3395–3402.
13. Han, B.Q.; Dunand, D.C. Microstructure and mechanical properties of magnesium containing high
volume fractions of yttria dispersoids. Mater. Sci. Eng. A 2000, 277, 297–304.
14. Goh, C.S.; Wei, J.; Lee, L.C.; Gupta, M. Development of novel carbon nanotube reinforced
magnesium nanocomposites using the powder metallurgy technique. Nanotechnology 2006, 17, 7–12.
15. Zhong, X.L.; Wong, W.L.E.; Gupta, M. Enhancing strength and ductility of magnesium by
integrating it with aluminum nanoparticles. Acta Mater. 2007, 55, 6338–6344.
16. Goh, C.S.; Wei, J.; Lee, L.C.; Gupta, M. Effects of fabrication techniques on the properties of
carbon nanotube reinforced magnesium. Solid State Phenom. 2006, 111, 179–182.
17. Liu, S.Y.; Gao, F.P.; Zhang, Q.Y.; Zhu, X.; Li, W.Z. Fabrication of carbon nanotubes reinforced
AZ91D composites by ultrasonic processing. Trans. Nonferr. Met. Soc. China 2010, 20, 1222–1227.
18. Goh, C.S.; Wei, J.; Lee, L.C.; Gupta, M. Simultaneous enhancement in strength and ductility by
reinforcing magnesium with carbon nanotubes. Mater. Sci. Eng. A 2006, 423, 153–156.
19. Tun, K.S.; Gupta, M. Improving mechanical properties of magnesium using nano-yttria reinforcement
and microwave assisted powder metallurgy method. Comp. Sci. Technol. 2007, 67, 2657–2664.
20. Ramakrishnan, N. An analytical study on strengthening of particulate reinforced metal matrix
composites. Acta Mater. 1996, 44, 69–77.
21. Zhang, Z.; Chen, D.L. Contribution of Orowan strengthening effect in particulate-reinforced metal
matrix nanocomposites. Mater. Sci. Eng. A 2008, 483–484, 148–152.
22. Nardone, V.C.; Prewo, K.M. On the strength of discontinuous silicon carbide reinforced aluminum
composites. Scr. Metall. 1986, 20, 43–48.
23. Moghadam, A.D.; Schultz, B.F.; Ferguson, J.B.; Omrani, E.; Rohatgi, P.K.; Gupta, N. Functional
metal matrix composites-self-lubricating, self-healing, and nanocomposites-An outlook. JOM
2014, 66, 872–881.
24. Lurie, S.; Belov, P.; Volkov-Bogorodsky, D.; Tuchkova, N. Nanomechanical modeling of the
nanostructures and dispersed composites. Comp. Mater. Sci. 2003, 28, 529–539.
25. Lin, K.; Law, E.; Pang, S. Effects of interphase regions of particulate-reinforced metal matrix
nanocomposites using a discrete dislocation plasticity model. J. Nanomech. Micromech. 2014,
doi:10.1061/(ASCE)NM.2153-5477.0000098.
26. Meenashisundaram G.K.; Gupta, M. Synthesis and characterization of high performance low
volume fraction TiC reinforced Mg nanocomposites targeting biocompatible/structural applications.
Mater. Sci. Eng. A 2015, 627, 306–315.
Materials 2015, 8 5151

27. Park, M.S. An enhanced mean field material model incorporating dislocation strengthening for
particle reinforced metal matrix composites. J. Mech. Sci. Tech. 2014, 28, 2587–2594.
28. Bocchini, G.F. The influence of porosity on the characteristics of sintered materials. Int. J.
Pow. Metal. 1986, 22, 185–186.
29. Brassell, G.W.; Horak, J.A.; Butler, B.L. Effects of porosity on strength of carbon-carbon composites.
J. Comp. Mater. 1975, 9, 288–296.
30. Ray, S. Synthesis of cast metal matrix particulate composites. J. Mater. Sci. 1993, 28, 5397–5413.
31. Viswanathan, V.; Laha, T.; Balani, K.; Agarwal, A.; Seal, S. Challenges and advances in
nanocomposite processing techniques. Mater. Sci. Eng. R 2006, 54, 121–285.
32. Ahmad, S.N.; Hashim, J.; Ghazali, M.I. Effect of porosity on tensile properties of cast particle
reinforced MMC. J. Comp. Mater. 2005, 41, 575–589.
33. Molina, J.M.; Prieto, R.; Narciso, J.; Louis, E. The effect of porosity on the thermal conductivity of
Al-12 wt. % Si/SiC composites. Scr. Mater. 2009, 60, 582–585.
34. Zhong, X.L.; Gupta, M. Development of lead-free Sn-0.7Cu/Al2O3 nanocomposite solders with
superior strength. J. Phys. D Appl. Phys. 2008, 41, doi:10.1088/0022-3727/41/9/095403.
35. Thein, M.A.; Li, L.; On, L.M.; Liu, T.; Liu, Z.; Ringer, S.P. Creep behavior of in-situ formed aln
reinforced Mg–5Al nanocomposite at low temperatures and low stresses. Nanosci. Nanotechnol. Lett.
2009, 1, 204–207.
36. Paul, B.; Banerji, P. Grain structure induced thermoelectric properties in PbTe nanocomposite,
Nanosci. Nanotechnol. Lett. 2009, 1, 208–212.
37. Xiao, L.; Chen, D.L.; Chaturvedi, M.C. Shearing of gamma double prime precipitates and formation
of planar slip bands in Inconel 718 during cyclic deformation. Scr. Mater. 2005, 52, 603–607.
38. Hall, E.O. The deformation and ageing of mild steel: III discussion of results. Proc. Phys. Soc. B
1951, 64, 747–753.
39. Petch, N.J. The cleavage strength of polycrystals. J. Iron Steel Inst. 1953, 173, 25–28.
40. Van Aken, D.C.; Krajewski, P.E.; Vyletel, G.M.; Allison, J.E.; Jones, J.W. Recrystallization and
grain growth phenomena in a particle-reinforced aluminum composite. Metall. Mater. Trans. A
1995, 26, 1394–1405.
41. Doherty, R.D.; Hughes, D.A.; Humphreys, F.J.; Jonas, J.J.; Juul Jensen D.; Kassner, M.E.;
King, W.E.; McNelley, T.R.; McQueen, H.J.; Rollett, A.D. Current issues in recrystallization:
A review. Mater. Sci. Eng. A 1997, 238, 219–274.
42. Hassold, G.N.; Holm, E.A.; Srolovitz, D.J. Effects of particle size on inhibited grain growth.
Scr. Metall. Mater. 1990, 24, 101–106.
43. Stearns, L.C.; Harmer, M.P. Particle-inhibited grain growth in Al2O3-SiC: I, experimental results.
J. Am. Ceram. Soc. 1996, 79, 3013–3019.
44. Nishizawa, T.; Ohnuma, I.; Ishida, K. Examination of the Zener relationship between grain size and
particle dispersion. Mater. Trans. JIM 1997, 38, 950–956.
45. Nes, E.; Ryum, N.; Hunderi, O. On the Zener drag. Acta Metall. 1985, 33, 11–22.
46. Mirza, F.A.; Chen, D.L. An analytical model for predicting the yield strength of particulate-reinforced
metal matrix nanocomposites with consideration of porosity. Nanosci. Nanotechnol. Lett. 2012, 4,
794–800.
Materials 2015, 8 5152

47. Shao, J.C.; Xiao, B.L.; Wang, Q.Z.; Ma, Z.Y.; Yang, K. An enhanced FEM model for particle size
dependent flow strengthening and interface damage in particle reinforced metal matrix composite.
Comp. Sci. Technol. 2011, 71, 39–45.
48. Dunand, D.C.; Mortensen, A. On plastic relaxation of thermal stresses in reinforced metals.
Acta Metall. Mater. 1991, 39, 127–139.
49. Sankaranarayanan, S.; Nayak, U.P.; Sabat, R.K.; Suwas, S.; Almajid, A.; Gupta, M. Nano-ZnO
particle addition to monolithic magnesium for enhanced tensile and compressive response.
J. Alloys Compd. 2014, 615, 211–219.
50. Alizadeh, M.; Beni, H.A. Strength prediction of the ARBed Al/Al2O3/B4C nano-composites using
Orowan model. Mater. Res. Bull. 2014, 59, 290–294.
51. Akbarpour, M.R.; Salahi, E.; Hesari, F.A.; Kim, H.S.; Simchi, A. Effect of nanoparticle content on
the microstructural and mechanical properties of nano-SiC dispersed bulk ultrafine-grained Cu
matrix composites. Mat. Des. 2013, 52, 881–887.
52. Kim, C.-S.; Sohn, I.; Nezafati, M.; Ferguson, J.B.; Schultz, B.F.; Bajestani-Gohari, Z.;
Rohatgi, P.K.; Cho, K. Prediction models for the yield strength of particle-reinforced unimodal pure
magnesium (Mg) metal matrix nanocomposites (MMNCs). J. Mater. Sci. 2013, 48, 4191–4204.
53. Sanaty-Zadeh, A. Comparison between current models for the strength of particulate-reinforced
metal matrix nanocomposites with emphasis on consideration of Hall-Petch effect. Mater. Sci. Eng. A
2012, 531, 112–118.
54. Saada, G. Hall-Petch revisited. Mater. Sci. Eng. A 2005, 400–401, 146–149.
55. Li, J.C.M. Petch relation and grain boundary sources. Trans. Metall. Soc. AIME 1963, 227, 239–247.
56. Louchet, F.; Weiss, J.; Richeton, T. Hall-Petch law revisited in terms of collective dislocation
dynamics. Phys. Rev. Lett. 2006, 97, doi:10.1103/PhysRevLett.97.075504.
57. Liu, T.M.; Chao, C.G. Effect of magnesium on mechanical properties of alumina-fiber-reinforced
aluminum matrix composites formed by pressure infiltration casting. Mater. Sci. Eng. A 1993, 169,
79–84.
58. Ahmad, S.N.; Hashim, J.; Ghazali, M.I. The effects of porosity on mechanical properties of cast
discontinuous reinforced metal- matrix composite. J. Comp. Mater. 2005, 39, 451–466.
59. Geranmayeh, A.R.; Mahmudi, R.; Kangooie, M. High-temperature shear strength of lead-free
Sn-Sb-Ag/Al2O3 composite solder. Mater. Sci. Eng. A 2011, 528, 3967–3972.
60. Tohgo, K.; Itoh, Y.; Shimamura, Y. A constitutive model of particle-reinforced composites taking
account of particle size effects and damage evolution. Comp. Part A 2010, 41, 313–321.
61. Goh, C.S.; Gupta, M.; Wei, J.; Lee, L.C. Characterization of high performance Mg/MgO
nanocomposites. J. Comp. Mater. 2007, 41, 2325–2335.
62. Asgharzadeh, H.; Simchi, A.; Kim, H.S. Hot deformation of ultrafine-grained Al6063/Al2O3
nanocomposites. J. Mater. Sci. 2011, 46, 4994–5001.
63. Zehtab Yazdi, A.; Bagheri, R.; Zebarjad, S.M.; Razavi Hesabi, Z. Incorporating aspect ratio in a
new modeling approach for strengthening of MMCs and its extension from micro to nano scale.
Adv. Comp. Mater. 2010, 19, 299–316.
64. Kwon, Y.K.; Berber, S.; Tomanek, D. Thermal contraction of carbon fullerenes and nanotubes.
Phys. Rev. Lett. 2004, 92, 015901–015904.
Materials 2015, 8 5153

65. Morsi, K.; Esawi, A. Effect of mechanical alloying time and carbon nanotube (CNT) content on the
evolution of aluminum (Al) CNT composite powders. J. Mater. Sci. 2007, 42, 4954–4959.
66. Wong, W.L.E.; Gupta, M. Improving overall mechanical performance of magnesium using
nano-alumina reinforcement and energy efficient microwave assisted processing route. Adv. Eng.
Mater. 2007, 9, 902–909.
67. Hassan, S.F.; Tan, M.J.; Gupta, M. High-temperature tensile properties of Mg/Al2O3 nanocomposite.
Mater. Sci. Eng. A 2008, 486, 56–62.
68. Hassan, S.F.; Gupta, M. Effect of length of Al2O3 particulates on microstructural and tensile
properties of elemental Mg. Mater. Sci. Eng. A 2006, 425, 22–27.
69. Hassan, S.F.; Gupta, M. Development of high performance magnesium nano-composites using
nano- Al2O3 as reinforcement. Mater. Sci. Eng. A 2005, 392, 163–168.
70. De Castro, V.; Leguey, T.; Muñoz, A.; Monge, M.A.; Pareja, R. Relationship between hardness and
tensile tests in titanium reinforced with yttria nanoparticles. Mater. Sci. Eng. A 2005, 400–401,
345–348.
71. Callister, W.D.; Rethwisch, D.G. Materials Science and Engineering: An Introduction, 9th ed.;
John Wiley and Sons: New York, NY, USA, 2014.
72. Bever, M.B. Encyclopedia of Materials Science and Engineering; Pergamon Press: New York, NY,
USA, 1986.
73. Kingery, W.D. Introduction to Ceramics, 2nd ed.; John Wiley and Sons: New York, NY, USA,
1976.

© 2015 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article
distributed under the terms and conditions of the Creative Commons Attribution license
(http://creativecommons.org/licenses/by/4.0/).
 Materials Science & Engineering A 575 (2013) 65–73

Contents lists available at SciVerse ScienceDirect

Materials Science & Engineering A

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

Low cycle fatigue of a rare-earth containing extruded magnesium alloy

F.A. Mirza a, D.L. Chen a,n, D.J. Li b, X.Q. Zeng b,1
a
Department of Mechanical and Industrial Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario, Canada M5B 2K3
b
The State Key Laboratory of Metal Matrix Composites, School of Materials Science and Engineering, Shanghai Jiao Tong University, 800 Dongchuan Road,
Shanghai 200240, PR China

art ic l e i nf o a b s t r a c t

Article history: The application of ultra-lightweight magnesium alloys inevitably involves fatigue resistance under cyclic
Received 26 January 2013 loading. The present study was aimed at evaluating strain-controlled cyclic deformation behavior and the
Received in revised form relevant effect of microstructure in a rare-earth (RE) element containing extruded Mg–10Gd–3Y–0.5Zr
12 March 2013
(GW103K) alloy. The microstructure of this alloy consisted of fine equiaxed grains with an average grain
Accepted 14 March 2013
size of about 12 μm and a large number of RE-containing precipitates. Unlike the RE-free extruded
Available online 27 March 2013
magnesium alloys, this alloy exhibited essentially cyclic stabilization and symmetrical hysteresis loops
Keywords: without tension–compression asymmetry due to the presence of the relatively weaker texture and the
Magnesium alloy suppression of twinning activities arising from the fine grain size and especially RE-containing precipitates.
Cyclic deformation
A detailed analysis for understanding the obstructive role of the precipitate to twinning has been presented.
Low cycle fatigue
While this alloy had a lower cyclic strain hardening exponent than the RE-free extruded magnesium alloys,
Twinning–detwinning
Texture it had a longer fatigue life which can also be described by the Coffin–Manson law and Basquin's equation.
Tension–compression yield asymmetry Fatigue crack was observed to initiate from the specimen surface with some cleavage-like facets at the
initiation site. Crack propagation was basically characterized by fatigue striations in conjunction with
secondary cracks.
& 2013 Elsevier B.V. All rights reserved.

1. Introduction the availability of slip systems and leads to preferred orientations

Due to the increasing global energy demand and high con-
sciousness of environmental protection in recent years [1,2],
lightweighting has become a key strategy in the automotive and
aerospace industries [1,3,4]. It has been reported that the fuel
efficiency of passenger vehicles can be enhanced by 6–8% for each
10% reduction in weight [5]. Magnesium alloy, as the lightest
structural metallic material, has been increasingly used in the
transportation industry to reduce the weight of motor vehicles [6].
The structural application of magnesium alloys involves unavoid-
ably fatigue and cyclic deformation characteristics due to the fact
that structural components in the vehicles experience dynamic
loading, which results in the occurrence of fatigue failure [7–9].
Hence, an understanding of fatigue and cyclic deformation of
magnesium alloy is critical for the design and durability evaluation
of engineering components.
Wrought alloys exhibit superior fatigue properties than casting
counterparts and are appropriate for studying the intrinsic fatigue
mechanisms of magnesium alloys [10–13]. However, the hexago-
nal close-packed (hcp) crystal structure of magnesium alloys limits
n
Corresponding author. Tel.: +1 416 979 5000x6487; fax: +1 416 979 5265.
E-mail addresses: dchen@ryerson.ca (D.L. Chen),
xqzeng@sjtu.edu.cn (X.Q. Zeng).
1
Tel.: +021 5474 2911; fax: +021 3420 2794.
(or textures) in the deformation process [14–16]. The commonly
reported tension–compression yield asymmetry and mechanical
anisotropy in wrought magnesium alloys are the main result of
those preferred orientations [17–19]. For the structural compo-
nents subjected to alternating cyclic loading, such a mechanical
anisotropy could result in irreversibility of cyclic deformation
which may exert an unfavorable influence on the material perfor-
mance. This problem could be conquered through texture mod-
ification via an alloy composition adjustment. It has recently been
reported that magnesium alloys containing rare-earth (RE) ele-
ments can develop more random textures during hot extrusion
[20–22]. The addition of RE elements in magnesium alloys leads to
a decrease of the overall texture sharpness or intensity, allowing
the activation of easy basal slip to a higher extent, compared to the
RE-free wrought magnesium alloys [21,23,24]. For example, Stan-
ford and Barnett [24] reported that the anisotropy of yield strength
of a rolled Mg–Zn–RE alloy was reversed as compared with the
traditional Mg–Zn alloy. Despite the fact that the addition of RE
elements sheds some light on the alteration in the mechanical
anisotropy, the potential advantage of such RE–Mg alloys as
structural components under cyclic loading condition has not
been well appreciated. The earlier studies were conducted mainly
on the high cyclic fatigue properties of the RE-containing magne-
sium alloys [14,25–27], such as Yang et al. [14] evaluated very
high-cycle fatigue behavior of an extruded Mg–12Gd–3Y–0.5Zr

0921-5093/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.msea.2013.03.041
66 F.A. Mirza et al. / Materials Science & Engineering A 575 (2013) 65–73
magnesium alloy at an ultrasonic frequency of 20 kHz and a stress
ratio of R¼ −1 and Yang et al. [27] reported the effect of addition of
different amounts of RE elements on high-cycle fatigue behavior of
die-cast magnesium alloy AZ91D. Studies on the strain-controlled
low cycle fatigue behavior of RE-containing magnesium alloys are
very limited [28,29]. It is unclear how and to what extent the RE
elements can change the tensile–compressive yield asymmetry
which generally occurs in the RE-free extruded magnesium alloys,
whether RE–Mg alloys exhibit cyclic hardening or softening, and
via what mechanisms the RE elements affect the cyclic deforma-
tion behavior and fatigue life. Thus, an understanding about the
influence of grain size and texture on the cyclic deformation
characteristics and twin formation due to the addition of RE
elements is necessary. The objective of the present work was to
evaluate the cyclic deformation behavior of an extruded Mg–
10Gd–3Y–0.5Zr (GW103K) alloy, and determine the fatigue life
under varying strain amplitudes, with particular attention to the
effect of the RE element addition on the twin formation and twin-
particle interaction.
2. Material and experimental procedure
The material investigated was a recently developed extruded
magnesium alloy with the following composition (wt%): 10Gd, 3Y,
0.5Zr, and Mg (balance). The received alloy bars with a diameter
of 20 mm were extruded at 400 1C with an extrusion ratio of 9:1.
Microstructural examinations were performed using an optical
microscope (OM), and scanning electron microscope (SEM) JSM-
6380LV equipped with Oxford energy dispersive X-ray spectroscopy
(EDS) system. Standard metallographic sample preparation techni-
ques were used with an etchant based on an acetic picral solution
containing 4.2-g picric acid, 10-ml acetic acid, 10-ml H2O, and 70-ml
ethanol. The texture was determined by measuring incomplete pole
figures between Ψ¼01 to 751 in the back reflection mode using a
PANalytical X-ray diffractometer with Cu Kα radiation at 45 kV and
40 mA and analyzed using MTEX software, and defocusing due to
the rotation of sample holder was corrected using experimentally
determined data obtained from the diffraction of magnesium
powders received from Magnesium Electron. Sub-sized fatigue
samples in accordance with ASTM E8 standard were machined with
the loading axis parallel to the extrusion direction (ED). The samples
had a gauge length of 25 mm (or a parallel length of 32 mm) and a
width of 6 mm. The thickness of the samples was 6 mm as well. The
gage section of fatigue samples was ground progressively along the
loading direction with emery papers up to grit number of 600 to
remove the machining marks and to achieve a smooth surface.
Strain-controlled, pull–push type fatigue tests were conducted
using a computerized Instron 8801 fatigue testing system at a zero
mean strain (strain ratio Re ¼−1, completely reversed strain cycle),
a constant strain rate of 1 " 10−2 s−1 and room temperature of
25 1C. Triangular loading waveform was applied during the tests.
Low cycle fatigue tests were performed at total strain amplitudes
of 0.2%, 0.4%, 0.6%, 0.8%, 1.0%, and 1.2%, and at least two samples
were tested at each level of the strain amplitudes. The strain-
controlled tests at lower strain amplitude levels were continued
up to 10,000 cycles, then the tests were changed to load control at
a frequency of 50 Hz using sine cyclic waveform. For the sake of
comparison, some samples of extruded RE-free AM30 magnesium
alloy (with a composition of 3.4 wt% Al, 0.33 wt% Mn, 0.16 wt% Zn,
0.0026 wt% Fe, 0.0006 wt% Ni, 0.0008 wt% Cu and balance Mg)
were also tested. The fracture surfaces of fatigued specimens were
examined via scanning electron microscope (SEM) to identify
fatigue crack initiation sites and propagation characteristics.
The residual twins in the regions near the fracture surface were
observed as well.
3. Results
3.1. Microstructure and texture
Fig. 1(a) shows a typical microstructure of as-extruded GW103K
alloy where the extrusion direction (ED) is indicated. It is seen that
uniform equiaxed grains with an average grain size of about 12 μm
were obtained in this alloy due to the occurrence of dynamic
recrystallization (DRX) in the hot extrusion process at 400 1C. The
grain size was fairly small in comparison with the common
extruded magnesium alloys, such as AZ31 and AM30 [8,11–13,30].
This was due to the role of added RE elements and zirconium where
Zr mainly restricted the grain growth [31]. A similar role of grain
refinement by RE elements, e.g., gadolinium [32], yttrium [33],
cerium [34], neodymium [35], erbium [36], has been reported as
well. Besides, as seen from Fig. 1(a) no twins were present in the
un-deformed samples. Fig. 1(b) shows a typical SEM back-scattered
electron image of as-extruded sample where many rare-earth
containing particles can be seen, as indicated by arrows. To confirm
the presence of RE-rich particles in the as-extruded alloy, EDS line
scan was performed as shown in Fig. 1(c) and (d), where the middle
rectangle-like particle contained Gd and the right circular particle
contained Zr. Similar microstructure was presented in [37] and
three different phases were observed: α-Mg solid solution matrix
phase with supersaturated Gd+Y elements, (Gd+Y)-rich eutectic
compound which had a higher Gd+Y content than the matrix, and
intracrystalline Zr-rich cores. Similar microstructures were also
reported in as-extruded Mg–11.90Gd–0.81Y–0.44Zr alloy by Zhang
et al. [38], as-cast samples of Mg–7Gd–3Y–0.4Zr (GW73K) alloy by
Liang et al. [39], and cast Mg–10Gd–3Y–Zr alloy by Wang et al. [40].
Fig. 2 shows a (0002) pole figure of the extruded GW103K alloy
evaluated using MTEX software. A relatively weak basal texture (tilted
towards the radial direction (RD)) with a maximum intensity of 2.1 m.
r.d. (multiples of random distribution) after de-focusing correction was
observed, where the c-axes of most grains were aligned perpendicular
to the ED. The presence of such a weaker texture, in comparison with
the extruded AM30 [41] and rolled AZ31 [42], was another benefit of
the RE elements added into magnesium alloys, as also reported by
Stanford and Barnett [24] who observed that microalloying with RE
elements can weaken texture in the forming process. Similar weaker
texture in Mg-based alloys with the addition of RE elements was
reported [14,20,43,44]. In the current extruded GW103K alloy, grain
boundary nucleation, coalescence of low-angle boundaries and shear
band assisted nucleation are absent (as seen in Fig. 1(a)) due to the
presence of a large number of RE-containing particles (Fig. 1(b)).
Zhu et al. [43] considered that only particle-stimulated nucleation was
a factor influencing texture weakening. Since the fine particles cannot
deform compatibly with the matrix during extrusion, stresses could
concentrate easily near the precipitates and subsequent plastic
relaxation would occur to relieve the stress build-up around the
precipitates, resulting in local lattice rotations in the regions adjacent
to the precipitates [14,45]. Indeed, Humphreys [46] observed that the
high RE content with a mixed addition resulted in the precipitation of
a large number of the second-phase particles, which can effectively
hinder the movement of dislocations and build up a stress concentra-
tion near the precipitates, in turn promote the particle-stimulated
nucleation of dynamic recrystalliazation and diminish the basal
texture. In addition, the weakening of the texture was also associated
with the appearance of deformation bands containing twins and the
restriction of grain growth [47].
3.2. Hysteresis loops and tensile properties
Fig. 3 shows typical stress–strain hysteresis loops of the first,
second, and mid-life cycles at a total strain amplitude of 1.2% and
 F.A. Mirza et al. / Materials Science & Engineering A 575 (2013) 65–73 67

Precipitate
Precipitate

ED

Intensity, Counts x1000

Mg
3

1 Zr
Gd
0
0 5 10 15 20 25 30
Distance , m
Fig. 1. Microstructure of extruded GW103K alloy, (a) OM image, and (b) SEM back-scattered electron image, (c) SEM back-scattered electron image indicating an EDS line
scan position, and (d) the corresponding EDS line scan results.

360

270

180
ED
90
Stress, MPa

RD 0

-90

AM30-1st cycle
-180 AM30-2nd cycle
AM30-Mid-life cycle
-270 GW103K-1st cycle
GW103K-2nd cycle
GW103K-Mid-life cycle
Fig. 2. (0002) pole figure of extruded GW103K alloy, where ED stands for the -360
extrusion direction and RD indicates the radial direction. -1.5 -1 -0.5 0 0.5 1 1.5
Total strain, %
strain ratio of Re ¼−1 for the extruded GW103K and AM30 alloys, Fig. 3. Typical stress-strain hysteresis loops of different cycles at a total strain
respectively. The yield strength at a strain rate of 1 " 10−2 s−1 amplitude of 1.2% and strain ratio of Re ¼ −1 for the extruded AM30 and Mg–10Gd–
3Y–0.5Zr (GW103K) alloy, respectively.
could be easily determined from the initial quarter of the first-
cycle hysteresis loops. The present GW103K alloy exhibited a
higher tensile yield strength than the extruded AM30 alloy, i.e., similar to those of face-centered cubic (fcc) metals (e.g., Al, Cu, and
#240 MPa vs. # 200 MPa [41]. This reflected an obvious beneficial Ni) as a result of the dislocation slip-dominated deformation in
role of RE element addition in the magnesium alloy. The obtained most materials [49]. Furthermore, the maximum and minimum
tensile properties were in agreement with those reported in an peak stresses showed no significant change among the first,
extruded Mg–8Al–xRE alloy [48], where the improved tensile second, and mid-life cycle loops in the GW103K alloy. In contrast,
properties due to the addition of RE elements were presented as the situation in the extruded AM30 alloy was very different, where
well. In particular, it is seen from Fig. 3 that, unlike the extruded the hysteresis loops were skewed although the mid-life hysteresis
RE-free AM30 alloy, the current GW103K alloy exhibited nearly loop appeared slightly less distorted than the first and second
symmetrical hysteresis loops in shape, which were somewhat cycle hysteresis loops. This was predominantly due to the activity
68 F.A. Mirza et al. / Materials Science & Engineering A 575 (2013) 65–73

of twinning in compression in the descending phase and subse-
quent detwinning in the ascending phase [8,9,11–13,30,50–52].
3.3. Fatigue behavior
Fig. 4 shows the evolution of stress amplitudes with respect to
the number of cycles at different applied strain amplitudes on a
semi-log scale. As the total strain amplitude increased, the stress
amplitude increased and the fatigue life of the material decreased.
Unlike the RE-free extruded AM30 [11] and AZ31 [8,11–13] where
cyclic stabilization occurred only at lower strain amplitudes of about
0.1% and 0.2%, the GW103K alloy exhibited cyclic stabilization until
failure up to a strain amplitude of 1.0%. Even at a strain amplitude of
1.2%, only slight change, i.e., an initial slight cyclic hardening within
the first three cycles and then minor cyclic softening, could be seen
from Fig. 4. Such a change was indeed trivial, as also seen from the
evolution of hysteresis loops mentioned above (Fig. 3). In the low
cycle fatigue tests, plastic strain amplitude has been considered as
400
Total strain amplitudes
300
Stress amplitude, MPa
a physical quantity that results in several damaging processes and
influences the internal microstructure which is closely related to the
strain resistance and eventually the fatigue life [12]. The variation of
the plastic strain amplitude ðΔεp =2Þ during cyclic deformation is
shown in Fig. 5 for different total strain amplitudes. Again, it is seen
that cyclic stabilization basically occurred in the present RE-
containing GW103K alloy at all strain amplitudes applied up to as
high as 1.2%. As a result, the addition of RE elements was able to
change significantly the cyclic deformation behavior of the magne-
sium alloy.
The fatigue life (i.e., the number of cycles to failure, Nf) as a
function of the applied total strain amplitudes ðΔεt =2Þ of the
extruded GW103K alloy is plotted in Fig. 6, along with the
experimental data reported in the literature for various extruded
magnesium alloys [11,12,42,53–55] for comparison. Run-out data
points were indicated by arrows pointing horizontally at or more
than 107 cycles. The alloy basically showed an improved fatigue life
than the RE-free extruded magnesium alloys [8,11,12,42,53–55].
As described in refs. [8,9,11,12,56–59] the total strain amplitude
1.2% 1.4
GW103K-present study
1.0%
AM30 [11]
0.8% 1.2
AZ31 [12]

0.6%
AZ31 [42]
0.4% 1.0
ZK60 [53]
0.2% AZ61-ED [54]
200 0.8
εt/2, %

AZ61-ETD [54]
AM30 [55]
0.6
∇

100
0.4

0.2
0
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 0.0
Number of cycles, N 1E+1 1E+3 1E+5 1E+7
Number of cycles to failure, Nf
Fig. 4. Stress amplitude vs. the number of cycles at different total strain amplitudes
applied. Fig. 6. Total strain amplitude as a function of the number of cycles to failure for the
extruded GW103K alloy, in comparison with the data reported in the literature for
various extruded Mg alloys.

0.75
Total strain amplitudes 1.2% -1
1.0%
Total
0.8%
Plastic strain amplitude, %

0.55 -1.5 Plastic

0.6% YT = -0.1888X T - 1.5297
R² = 0.9762 Elastic
0.4%
-2
0.2%
0.35
-2.5
Log

YP = -0.4449XP - 1.3191
R² = 0.975
-3
0.15
Y E= -0.1087X E - 1.9342
-3.5 R² = 0.9555

-0.05 -4
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1 2 3 4 5 6
Number of cycles, N Log ( f )

Fig. 5. Plastic strain amplitude vs. the number of cycles at different total strain Fig. 7. Evaluation of fatigue parameters in the form of log elastic, plastic and total
amplitudes applied. strain amplitudes vs. log number of reversals to failure, respectively.
 F.A. Mirza et al. / Materials Science & Engineering A 575 (2013) 65–73 69

could be expressed as elastic strain amplitude and plastic strain fatigue life or the number of cycles to failure, s′f is the fatigue
amplitude, i.e., strength coefficient, b is the fatigue strength exponent, ε′f is the
fatigue ductility coefficient, and c is the fatigue ductility exponent.
Δεt Δεe Δεp s′f ð2Nf Þb The elastic strain component is referred to as Basquin's equation
¼ þ ¼ þ ε′f ð2N f Þc ; ð1Þ
2 2 2 E and the plastic strain component is known as the Coffin–Manson
where E is the Young's modulus (for the present alloy the average relation. Fig. 7 shows the elastic, plastic, and total strain amplitudes
value obtained during fatigue testing was # 44.5 GPa), Nf is the plotted as a function of the number of reversals to failure. The
values of the strain amplitudes were taken from the mid-life cycles,
Table 1 even though cyclic stabilization or saturation was present in this
Low cycle fatigue parameters obtained for the extruded GW103K alloy. alloy at all the strain amplitudes up to the applied strain amplitude
of 1.2% (Figs. 3–5). The fatigue life parameters obtained on the basis
Low cycle fatigue parameters Extruded GW103K
of Eq. (1) are summarized in Table 1. It is seen that while the
Cyclic strain hardening exponent, n′ 0.20 obtained fatigue parameters were well within the range in other
Cyclic strength coefficient, K′, MPa 754 fatigued magnesium alloys reported in the literature [8,11,12,50],
Fatigue strength coefficient, s′f, MPa 521 the cyclic strain hardening exponent n′ of GW103K alloy was lower
Fatigue strength exponent, b −0.11 than that of extruded AZ31 alloy [8,11,12]. This corresponded well
Fatigue ductility coefficient, ε′f 0.05
Fatigue ductility exponent, c −0.44
to the higher yield strength and cyclic stabilization characteristics of
this alloy (Figs. 4 and 5).

Fig. 8. SEM images of overall fracture surfaces of the extruded specimens fatigued at a total strain amplitude of (a) 0.4% and (b) 1.0%.

Fig. 9. SEM micrographs of the fracture surface near crack initiation of the extruded specimens fatigued at a total strain amplitude of 0.4% ((a) secondary electron image and
(b) back-scattered electron image) and (c) 1.0% (secondary electron image).
70 F.A. Mirza et al. / Materials Science & Engineering A 575 (2013) 65–73
Fig. 10. SEM micrographs of the fatigue crack propagation region of the extruded specimens fatigued at a total strain amplitude of (a) 0.4% and (b) 1.0%.
3.4. Fractography
Fig. 8 shows an overall view of typical fracture surfaces at a lower
magnification for the specimens tested at a strain amplitude of 0.4%
and 1.0%, respectively. It is seen that fatigue crack initiated basically
from the specimen surface. On a close examination at a higher
magnification (Fig. 9(a)), fatigue crack indeed initiated from near-
surface defects. From a back-scattered electron image (Fig. 9(b)),
which was taken at the same location as that of Fig. 9(a), it is seen
that the RE-containing precipitates were uniformly distributed on the
fracture surface. The initiation site at a strain amplitude of 1.0% had
more obvious features of cleavage-like facets (Fig. 9(c)) than that at a
strain amplitude of 0.4% (Fig. 9(a) and (b)). Fatigue crack propagation
was basically characterized by fatigue striations observed at higher
magnifications (Fig. 10(a) and (b)), in conjunction with some sec-
ondary cracks. In addition, the fatigue striations were perpendicular
to the crack propagation direction. These characteristics were more
obvious at higher strain amplitudes, as see in Fig. 10(b).
4. Discussion
4.1. Texture
In polycrystalline alloys, the yield asymmetry is directly related to
the presence of crystallographic texture which dominates the orienta-
tion of slip or twinning planes and directions relative to the externally
applied stress [60]. Normally the yield asymmetry is less obvious in
cast magnesium alloys which tend to have weak textures compared
with wrought magnesium alloys [24]. Yin et al. [61] also reported that
randomly textured casting samples exhibit almost no yielding asym-
metry. As mentioned earlier in Section 3.1, the present RE-containing
GW103K alloy showed a relatively weak texture (Fig. 2) due to the
presence of second-phase particles and finer grain size which were
responsible for the nearly symmetrical hysteresis loops in Fig. 3.
Generally, the initial deformation asymmetry behavior of metals is
associated with the Basuschinger effect [56]. From Fig. 3 it is clearly
seen that the hysteresis loops of AM30 alloy showed a strong
Basuschinger-like effect, which concurred with the results reported
by other investigators [11,12,56,62]. The ratio of the compressive-to-
tensile yield stress was # 0.55 for AM30 alloy, whereas it was almost
1 for GW103K alloy, indicating that there was almost no difference in
the tensile and compressive yielding due to the microalloying with RE
elements (Fig. 3). Thus, it could be concluded that the addition of RE
elements played an important role in eliminating or diminishing the
yield asymmetry of the material via texture randomization.
4.2. Twinning–detwinning during cyclic deformation
Twinning has an important effect on the plastic deformation
of magnesium alloys [8–13,15,30,41,59]. In general, the forming
process (rolling or extrusion) of magnesium alloys makes most
grains orientated in such a manner that their basal planes are
nearly parallel to the rolling/extrusion direction. When a load is
applied in the extrusion or rolling direction, the activation stress is
relatively low for the extension twinning (in compression) or basal
slip (in tension), and is relatively high for the prismatic and
pyramidal slip systems in magnesium alloys at room temperature
[63]. Then under cyclic loading a fairly large portion of the cyclic
strain would be undertaken by the twinning–detwinning activities
[11,12]. As seen from the near fracture surface area in Fig. 11(a) and (b),
dense residual twins appeared near the fracture surface of AM30 alloy.
The formation of these twins was attributed to an insufficient number
of slip systems in magnesium alloys deformed at room temperature
arising from the hcp structure with a low crystal symmetry, in
conjunction with the presence of strong texture as discussed above.
However, twins could barely be visible in the near fracture
surface area of GW103K alloy (Fig. 11(c) and (d)). The absence of
twins suggested that twinning was not a dominating deformation
mode in the cyclic deformation of GW103K alloy, and the basal slip
of dislocations would become a main deformation mechanism as
corroborated by the nearly symmetrical hysteresis loops (Fig. 3),
somewhat like the situation of the common fcc metals. This was
mainly due to the effect of grain size and precipitates in the
GW103K alloy, to be discussed in the following section.
4.3. Effect of grain size and precipitates
An important factor that influences the twinning in tension-
compression cyclic deformation is grain size. As seen from Figs. 1(a),
and 11(c), the grains in the GW103K alloy were equiaxed and fairly
small due to the presence of RE elements and zirconium, in
comparison with the extruded Mg alloys, such as AZ31 and AM30
[8,11–13,30]. It was reported that grain size had a significant effect
on the tendency of twinning since the energy required to form twin
interfaces was particularly high in the fine-grained magnesium
alloy [64]. In addition, there is a general agreement that a small
grain size inhibits twinning somewhat like the slip of dislocations
[41,51,52,57–61]. Generally, the grain size dependence of the
activation of deformation mechanisms in polycrystals could be
described in terms to the Hall–Petch relationship ðs ¼ so þ kd
−1=2
Þ
[56], where s is the yield strength, so is a friction stress for the
movement of dislocations, and d is the average grain size. The slope
k is called as the “Hall–Petch strength coefficient”. The effect of the
grain size or the content of alloying elements on twinning was
studied using Hall–Petch analysis by Bohlen et al. [65], who also
observed the distinctive tension/compression asymmetry in the
yield behavior of magnesium alloy, similar to the hysteresis loops of
AM30 alloy shown in Fig. 3. While the asymmetry in the tensile and
compressive yield strengths reflects a different contribution of
twinning and slip to the macroscopic strain, the difference between
the tensile and compressive yield strengths generally decreases
 F.A. Mirza et al. / Materials Science & Engineering A 575 (2013) 65–73 71

Fig. 11. Optical micrographs and SEM images in the areas near the fracture surface at a strain amplitude of 1.2%, showing the distribution of residual twins in the fatigued
samples of AM30 alloy ((a) and (b)) and GW103K alloy ((c) and (d)), respectively.

with decreasing grain size [65]. Thus, the grain size has a strong [0001]
influence on the yield stress associated with both the slip of (Matrix)
dislocations and twinning. The twinning stress was indeed reported {10 1 2}
twinning plane K1
to increase with decreasing grain size more rapidly than the stress
required to activate slip [61,66,67]. It follows that for the fine- Matrix 2 [10 1 1]
grained alloys dislocation slip would dominate the deformation no *[10 1 1] 7.4° Twin
2
matter whether it is in compression or in tension, and the yield
asymmetry would be absent or reduced in extent due to the 86.3° Shear
suppression of twinning activities, as also seen from the hysteresis
loops of GW103K alloy in Fig. 3.
In addition to the effect of grain size, it has also been reported Precipitate 86.3°
that the presence of precipitates modifies the process of twinning
in magnesium alloys [68]. The interaction between twinning and Twin basal plane
precipitates has been considered to arise primarily from the
difficulty experienced by migrating twin boundaries during the
twin propagation through a densely distributed precipitates [60].
43.15°
To better understand the interaction between a twin and a plate- Matrix [10 1 0]
shaped precipitate, a schematic illustration showing the obstruc- (Matrix)
tion of a precipitate/particle to twinning is plotted in Fig. 12. Let us Matrix basal plane
first establish a coordinate system with the Y-axis being parallel to Fig. 12. A schematic illustration of the interaction between a f1 0 1 2g twin and a
the c-axis [0001] of an hcp unit cell in the magnesium matrix and plate-shaped precipitate in the GW103K alloy. (For interpretation of the references
the X-axis being oriented in the direction of ½1 0 1 0( which is to color in this figure, the reader is referred to the web version of this article.)

positioned on the basal plane. In such a way the f0 0 0 2g basal

plane of the magnesium matrix becomes perpendicular to the sur-
face of “paper” (or “computer screen”) and parallel to the X-axis, as
indicated in Fig. 12. Then the precipitate can be plotted vertically
to the X-axis (the basal plane of the magnesium matrix), since it
was observed to form on the prismatic planes of the magnesium
matrix in a triangular arrangement [37]. In the magnesium alloys,
f1 0 1 2g extension twinning is normally activated when a com-
pressive stress is applied perpendicular to the [0001] c-axis or
parallel to the f0 0 0 2g basal plane [9,69]. Using the magnesium
lattice parameters of a ¼0.32 nm and c¼ 0.52 nm [70], one can
readily calculate the angle between the f0 0 0 2g basal plane and
the f1 0 1 2g twinning plane K1 (which is an invariant plane being
perpendicular to the surface of the “paper” or “computer screen”
as well) or the twinning direction η1 ½1 0 1 1( residing on the
twinning plane K1 to be 43.151, as indicated in Fig. 12. In response
to the twinning shear and the symmetry requirement to generate
a mirror image of the atoms inside the twin versus those in the
matrix with respect to the invariant twinning plane K1, an
operation equivalent to the rotation of the lattice atoms within
the twin (i.e., the portion in-between the two red-colored lines in
Fig. 12) has to take place under an external (or internal) driving
force. Such a change is similar to re-orienting the basal planes of
the matrix (i.e., the fine horizontal lines parallel to the X-axis) to
72 F.A. Mirza et al. / Materials Science & Engineering A 575 (2013) 65–73
those indicated by the fine parallel lines inside the twin, which are
inclined at an angle of 86.31 to the fine horizontal lines, as shown
in Fig. 12. Due to the unique nature of twin characterized by the
perfectly symmetrical (or mirror) image of atoms with respect to
the twinning plane, the inclined fine parallel lines within the twin
must have an atomic arrangement identical to that represented by
the fine horizontal lines. Therefore, the inclined fine parallel lines
inside the twin indeed stand for the f0 0 0 2g twin basal planes. It
is clear that the occurrence of twinning leads to the formation of a
large “kink” or interruption of the regular atomic arrangement on
the basal plane along the fine horizontal lines, with the kink size
depending on the twin width. At the same time, the conjugate
f1 0 1 2g extension twinning plane K2 (not shown in the schematic
diagram of Fig. 12) in an adjoining unit cell would rotate counter-
clockwise from η2 ½1 0 1 1( to η2n ½1 0 1 1( with a rotating angle of
7.41 relative to the red-colored twinning plane considered, as
plotted in Fig. 12, where η2 (the dashed arrow) is the twinning
direction in the adjoining unit cell before twinning, and η2n (the
solid arrow) is the twinning direction in the same adjoining unit
cell after twinning, with a fixed angle of 86.31 between η1 and η2.
Such twinning in the magnesium matrix grains has also been
reported, e.g., in [60,65,67,70]. If the particle/precipitate is not
sheared during twinning, as the twinning ledge approaches the
particle there exists a back stress stemming from the harder and
more rigid particle, which would impede the free shear of the
moving twin ledge and must be overcome by the applied stress
[68]. The portion of the matrix in the absence of twinning due to
the blockage of the precipitate is indicated by the dashed lines in
Fig. 12. Furthermore, when the twinning shear reaches the inter-
face between the matrix and the precipitate, the movement of
atoms towards the precipitate should be accommodated by the
surrounding material, which is likely accomplished by the basal
slip of the matrix near the particle/matrix interface. However, the
basal slip in the matrix along the fine horizontal lines would also
be inhibited by the precipitate, thereby providing an opportunity
of shear being accommodated by the f0 0 0 2g twin basal slip (i.e.,
along the fine parallel lines inside the twin), as indicated in Fig. 12.
Then the need for the additional basal slip on a local scale around
the precipitate is likely to increase the stress required to form the
twin [68]. As a result, the second-phase particles in the GW103K
alloy (Fig. 1(b)) would significantly modify the potential twin
nucleation sites and retard the rate of twin formation. Similar
results were reported by Jain et al. [60]. It should also be noted
that other precipitates, e.g., the common eutectic β-Mg17Al12
particles present in the cast magnesium alloys and semi-solid
processed magnesium alloys would also generate a strong resis-
tance to the twinning (and basal slip as well) [71–76]. More
detailed studies on the twin-particle interactions in these cast
magnesium alloys are needed.
The above results and analyses indicated that both the grain
refinement and the presence of precipitates in the present GW103K
alloy due to the addition of RE elements played a crucial role in
suppressing the occurrences of deformation twins which led to the
reduction or elimination of the tension–compression asymmetry, as
shown in Fig. 3. In brief, the addition of RE elements was able to
influence substantially the overall cyclic deformation behavior of the
extruded magnesium alloys via the refinement of grains, the pre-
sence of RE-containing precipitates, the relatively weaker crystal-
lographic texture, and the suppression of twinning activities.
5. Conclusions
Strain-controlled low cycle fatigue tests were conducted on an
extruded Mg–10Gd–3Y–0.5Zr (GW103K) alloy with different total
strain amplitudes at room temperature. The following conclusions
can be drawn from this investigation:
1. The microstructure of the RE-containing GW103K alloy con-
sisted of fine recrystallized and equiaxed grains with an
average grain size of about 12 μm and a large number of RE-
containing precipitates.
2. Unlike the RE-free extruded magnesium alloys, the present
GW103K alloy exhibited basically symmetrical hysteresis loops,
suggesting the absence of the tensile–compressive yield asym-
metry. This was predominantly due to the presence of the
relatively weaker crystallographic texture and the lack of
twinning–detwinning activities stemming from the fine grain
size and RE-containing precipitates.
3. Strain-controlled fatigue tests also showed that the present RE-
containing GW103K alloy exhibited essentially cyclic stabiliza-
tion until failure up to a strain amplitude of 1.0%. The cyclic
strain hardening exponent of the alloy was observed to be
lower than that of RE-free extruded magnesium alloys obtained
at the same strain rate of 1 " 10−2 s−1 also due to its finer grain
size and RE-containing precipitates.
4. The fatigue life of the present alloy was observed to be longer
than that of the RE-free extruded magnesium alloys, which can
be well described by the Coffin–Manson law and Basquin's
equation.
5. SEM examinations revealed that fatigue cracks initiated from
the near-surface defects and the initiation site contained some
cleavage-like facets. The propagation was basically character-
ized by fatigue striations perpendicular to the crack propaga-
tion direction observed at higher magnifications in conjunction
with secondary cracks.
Acknowledgments
The authors would like to thank the Natural Sciences and
Engineering Research Council of Canada (NSERC) and AUTO21
Network of Centres of Excellence for providing financial support.
This investigation involves part of Canada–China–USA Collabora-
tive Research Project on the Magnesium Front End Research and
Development (MFERD). The authors also thank Dr. A.A. Luo,
General Motors Research and Development Center for the supply of
extruded AM30 magnesium alloy. One of the authors (X.Q. Zeng)
would like to give thanks to Ministry of Science and Technology of
China (MOST) and Natural Science Foundation of China (NSFC) for
their financial support (Project nos. 2011BAE22B02, 2011DFA50907,
and 51171113). One of the authors (D.L. Chen) is also grateful for the
financial support by the Premier's Research Excellence Award (PREA),
NSERC-Discovery Accelerator Supplement (DAS) Award, Canada
Foundation for Innovation (CFI), and Ryerson Research Chair (RRC)
program. The authors would like to thank Dr. R. Tandon and
Dr. B. Davies (Magnesium Elektron) for supplying magnesium pow-
ders for the defocusing calibration. The authors would also like to
thank Messrs. A. Machin, Q. Li, J. Amankrah and R. Churaman for easy
access to the laboratory facilities of Ryerson University and their
assistance in the experiments.
References
[1] N.P. Gillett, V.K. Arora, K. Zickfeld, S.J. Marshall, W.J. Merryfield, Nature Geosci.
4 (2011) 83–87.
[2] J. Murray, D. King, Nature 481 (2012) 433–435.
[3] T.M. Pollock, Science 328 (2010) 986–987.
[4] M. Wise, K. Calvin, A. Thomson, L. Clarke, B.B. Lamberty, R. Sands, S.J. Smith,
A. Janetos, J. Edmonds, Science 324 (2009) 1183–1186.
[5] W.J. Joost, JOM 64 (2012) 1032–1038.
 F.A. Mirza et al. / Materials Science & Engineering A 575 (2013) 65–73 73

[6] A.Α. Luo, JOM 54 (2002) 42–48. [44] J. Bohlen, M.R. Nurnberg, J.W. Senn, D. Letzig, S.R. Agnew, Acta Mater. 55
[7] V.V. Ogarevic, R.I. Stephens, Annu. Rev. Mater. Sci. 20 (1990) 141–177. (2007) 2101–2112.
[8] X.Z. Lin, D.L. Chen, Mater. Sci. Eng. A 496 (2008) 106–113. [45] F.J. Humphreys, Acta Metall. 27 (1979) 1801–1814.
[9] F.A. Mirza, D.L. Chen, in: S. Zhang, D.L. Zhao (Eds.), Aerospace Materials [46] F.J. Humphreys, Acta Mater. 45 (1997) 5031–5039.
Handbook, CRC Press, Taylor & Francis, New York, 2013, pp. 647–698. [47] K. Hantzsche, J. Bohlen, J. Wendt, K.U. Kainer, S.B. Yi, D. Letzig, Scr. Mater. 63
[10] L. Chen, C. Wang, W. Wu, Z. Liu, G.M. Stoica, L. Wu, P.K. Liaw, Metall. Mater. (2010) 725–730.
Trans. A 38 (2007) 2235–2241. [48] W.G. Yang, C.H. Koo, Mater. Trans. 44 (2003) 1029–1035.
[11] S. Begum, D.L. Chen, S. Xu, A.A. Luo, Metall. Mater. Trans. A 39 (2008) [49] S. Suresh, Fatigue of Materials, second ed., Cambridge University Press,
3014–3026. Cambridge, 1998.
[12] S. Begum, D.L. Chen, S. Xu, A.A. Luo, Int. J. Fatigue 31 (2009) 726–735. [50] U. Noster, B. Scholtes, Zeitschrift fur Metallkunde 94 (2003) 559–563.
[13] S. Begum, D.L. Chen, S. Xu, A.A. Luo, Mater. Sci. Eng. A 517 (2009) 334–343. [51] S. Hasegawa, Y. Tsuchida, H. Yano, M. Matsui, Int. J. Fatigue 29 (2007)
[14] F. Yang, F. Lv, X.M. Yang, S.X. Li, Z.F. Zhang, Q.D. Wang, Mater. Sci. Eng. A 528 1839–1845.
(2011) 2231–2238. [52] L. Wu, A. Jain, D.W. Brown, G.M. Stoica, S.R. Agnew, B. Clausen, D.E. Fielden,
[15] E.A. Ball, P.B. Prangnell, Scr. Metall. Mater. 31 (1994) 111–116. P.K. Liaw, Acta Mater. 56 (2008) 688–695.
[16] Y.N. Wang, J.C. Huang, Mater. Chem. Phys. 81 (2003) 11–26. [53] Q. Yu, J. Zhang, Y. Jiang, Q. Li, Int. J. Fatigue 36 (2012) 47–58.
[17] S. Kleiner, P.J. Uggowitzer, Mater. Sci. Eng. A 379 (2004) 258–263. [54] J.B. Jordon, J.B. Gibson, M.F. Horstemeyer, H. El Kadiri, J.C. Baird, A.A. Luo,
[18] S.M. Yin, H.J. Yang, S.X. Li, S.D. Wu, F. Yang, Scr. Mater. 58 (2008) 751–754. Mater. Sci. Eng. A 528 (2011) 6860–6871.
[19] S.H. Safi-Naqvi, W.B. Hutchinson, M.R. Barnett, Mater. Sci. Technol. 24 (2008) [55] T.J. Luo, Y.S. Yang, W.H. Tong, Q.Q. Duan, X.G. Dong, Mater. Des 31 (2010)
1283–1292. 1617–1621.
[20] N. Stanford, M.R. Barnett, Mater. Sci. Eng. A 496 (2008) 399–408. [56] G.E. Dieter, Mechanical Metallurgy, SI metric ed., McGraw-Hill Inc., New York,
[21] N. Stanford, G. Sha, J.H. Xia, S.P. Ringer, M.R. Barnett, Scr. Mater. 65 (2011) 1986.
919–921. [57] Z. Liu, Y.Y. Xu, Z.G. Wang, Y. Wang, Z.Y. Liu, Acta Metall. Sin. (English letters) 13
[22] D. Wu, R.S. Chen, W.N. Tang, E.H. Han, Mater. Des. 41 (2012) 306–313. (2000) 961–966.
[23] N. Stanford, D. Atwell, A. Beer, C.H. Davies, M.R. Barnett, Scr. Mater. 59 (2008) [58] S.H. Chowdhury, D.L. Chen, S.D. Bhole, E. Powidajko, D.C. Weckman, Y. Zhou,
772–775. Metall. Mater. Trans. A 43 (2012) 2133–2147.
[24] N. Stanford, M.R. Barnett, Scr. Mater. 58 (2008) 179–182. [59] H.A. Patel, D.L. Chen, S.D. Bhole, K. Sadayappan, Mater. Sci. Eng. A 528 (2010)
[25] L. Wu, Z. Yang, W. Xia, Z. Chen, L. Yang, Mater. Des. 36 (2012) 47–53. 208–219.
[26] H. Bayani, E. Saebnoori, J. Rare Earths 27 (2009) 255–258. [60] J. Jain, W.J. Poole, C.W. Sinclair, M.A. Gharghouri, Scr. Mater. 62 (2010)
[27] Y. Yang, Y.B. Liu, S.Y. Qin, Y. Fang, J. Rare Earths 24 (2006) 591–595. 301–304.
[28] F.H. Wang, J. Dong, Y.Y. Jiang, W.J. Ding, Mater. Sci. Eng. A 561 (2013) 403–410. [61] S.M. Yin, C.H. Wang, Y.D. Diao, S.D. Wu, S.X. Li, J. Mater. Sci. Technol. 27 (2011)
[29] C.Y. Ma, X. Che, B. Zhao, L.J. Chen, J. Northeast. Univ. 28 (2007) 144–149. 29–34.
[30] C.L. Fan, D.L. Chen, A.A. Luo, Mater. Sci. Eng. A 519 (2009) 38–45. [62] J.P. Nobre, U. Noster, M. Kornmeier, A.M. Dias, B. Scholtes, Key Eng. Mater.
[31] Z.-K. Peng, X.-M. Zhang, J.-M. Chen, Y. Xiao, H. Jiang, Mater. Sci. Technol. 21 230–232 (2002) 267–270.
(2005) 722–726. [63] M.R. Barnett, Metall. Mater. Trans. A 34 (2003) 1799–1806.
[32] W.P. Li, H. Zhou, Z.F. Li, J. Alloys Compd. 475 (2009) 227–232. [64] J. Koike, T. Kobayashi, T. Mukai, H. Watanabe, M. Suzuki, K. Maruyama,
[33] X.Y. Fang, D.Q. Yi, B. Wang, W.H. Luo, W. Gu, Trans. Nonferrous Met. Soc. China K. Higashi, Acta Mater. 51 (2003) 2055–2065.
16 (2006) 1053–1058. [65] J. Bohlen, P. Dobron, J. Swiostek, D. Letzig, F. Chmelik, P. Lukac, K.U. Kainer,
[34] K. Yu, S.T. Rui, J.M. Song, W.X. Li, L. Guo, Trans. Nonferrous Met. Soc. China 18 Mater. Sci. Eng. A 462 (2007) 302–306.
(2008) s39–s43. [66] M.R. Barnett, Z. Keshavarz, A.G Beer, D. Atwell, Acta Mater. 52 (2004)
[35] M.Z. Li, Y.Q. Wang, C. Li, X.G. Liu, B.S. Xu, Mater. Sci. Technol. 27 (2011) 5093–5103.
1138–1142. [67] A. Jain, O. Duygulu, D.W. Brown, C.N. Tome, S.R. Agnew, Mater. Sci. Eng. A 486
[36] J. Zhang, X.F. Zhang, F.S. Pan, C.P. Liu, Q.B. He, Mater. Sci. Forum 610–613 (2008) 545–555.
(2009) 810–814. [68] N. Stanford, M.R. Barnett, Mater. Sci. Eng. A 516 (2009) 226–234.
[37] S.M. He, X.Q. Zeng, L.M. Peng, X. Gao, J.F. Nie, W.J. Ding, J. Alloys Comp. 427 [69] Y.N. Wang, J.C. Huang, Acta Mater. 55 (2007) 897–905.
(2007) 316–323. [70] P.G. Partridge, Int. Mater. Rev. 12 (1967) 169–194.
[38] F. Zhang, K.X. Zhang, C.W. Tan, X.D. Yu, H.L. Ma, F.C. Wang, H.N. Cai, Trans. [71] H.A. Patel, N. Rashidi, D.L. Chen, S.D. Bhole, A.A. Luo, Mat. Sci. Eng. A 546
Nonferrous Met. Soc. China 21 (2011) 2140–2146. (2012) 72–81.
[39] S.Q. Liang, D.K. Guan, L. Chen, Z.H. Gao, H.X. Tang, X.T. Tong, Mater. Des. 32 [72] H.A. Patel, D.L. Chen, S.D. Bhole, K. Sadayappan, Mater. Des. 49 (2013)
(2011) 361–364. 456–464.
[40] Q. Wang, G. Wu, H.Q. Zheng, B. Chen, Y. Zheng, W. Ding, China Found. J. 7 [73] J.B. Clarke, Acta Metall. 16 (1968) 141–152.
(2010) 6–12. [74] M.A. Gharghouri, G.C. Weatherly, J.D. Embury, Philos. Mag. A 78 (1998)
[41] D. Sarker, D.L. Chen, Scr. Mater. 67 (2012) 165–168. 1137–1149.
[42] F. Lv, F. Yang, Q.Q. Duan, Y.S. Yang, S.D. Wu, S.X. Li, Z.F. Zhang, Int. J. Fatigue 33 [75] L.M. Wang, C.C. Chen, Mater. Lett. 67 (2012) 158–161.
(2011) 672–682. [76] N. Stanford, J. Geng, Y.B. Chun, C.H.J. Davies, J.F. Nie, M.R. Barnett, Acta Mater.
[43] R. Zhu, Y.J. Wu, J.T. Wang, K.C. Lu, Adv. Mater. Res. 320 (2011) 222–227. 60 (2012) 218–228.
 Materials and Design 46 (2013) 411–418

Contents lists available at SciVerse ScienceDirect

Materials and Design

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

Effect of rare earth elements on deformation behavior of an extruded

Mg–10Gd–3Y–0.5Zr alloy during compression
F.A. Mirza a, D.L. Chen a,⇑, D.J. Li b, X.Q. Zeng b,⇑
a
Department of Mechanical and Industrial Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada
b
The State Key Laboratory of Metal Matrix Composites, School of Materials Science and Engineering, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai 200240, PR China

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

Article history: The aim of this study was to identify the influence of rare-earth (RE) elements on the strain hardening
Received 30 August 2012 behavior in an extruded Mg–10Gd–3Y–0.5Zr magnesium alloy via compression in the extrusion direction
Accepted 24 October 2012 at room temperature. The plastic deformation behavior of this RE-containing alloy was characterized by a
Available online 6 November 2012
rapidly decreasing strain hardening rate up to a strain level of about 4% (stage A), followed by a fairly flat
linear strain hardening rate over an extended strain range from 4% to 18% (stage B). Stage C was rep-
Keywords: resented by a decreasing strain hardening rate just before failure. The extent of twinning in this alloy was
Magnesium alloy
observed to be considerably less extensive than that in the RE-free extruded Mg alloys. The weaker crys-
Rare-earth element
Plastic deformation
tallographic texture, refined grain size, and second-phase particles arising from the addition of RE ele-
Strain hardening ments were responsible for the much higher strain hardening rate in stage A due to the increased
Twinning difficulty on the formation of twins and the slip of dislocations at lower strains, and for the occurrence
Slip of quite flat linear strain hardening in stage B at higher strains which was likely related to the dislocation
debris and twin debris (or residual twins) stemming from dislocation–twin interactions as well as the
interactions between dislocations/twins and second-phase particles and grain boundaries.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction
The extreme precipitation events and environmental problems
under global warming, which are today recognized to stem from
anthropogenic environment-damaging emissions, are increasing
severely in recent years [1]. To lower the greenhouse gas emissions
and promote the fuel efficiency in a time of environmental chal-
lenge and impending energy crisis [2], lightweight vehicle design
has become an important goal in the automotive and aerospace
industries [3]. Recently, the development and application of the ul-
tra-lightweight magnesium alloys have been significantly increas-
ing in the transportation sectors due to their low density, high
strength-to-weight ratio, and superior damping capacity [4–6].
In contrast to the cubic metals, the deformation of magnesium
is fairly complex because its hexagonal close-packed (hcp) crystal
structure has a limited number of active slip systems at room tem-
perature, with basal slip and twinning playing a key role in the
plasticity of magnesium alloys [5,6]. Both basal slip and twinning
mechanisms tend to align the c-axes of grains with the direction
of predominantly compressive strains during forming (rolling,
extrusion). In the magnesium sheet and plate, the c-axes align with
⇑ Corresponding authors. Tel.: +1 416 979 5000x6487; fax: +1 416 979 5265 (D.L.
Chen), tel.: +86 21 5474 2301; fax: +86 21 3420 3730 (X.Q. Zeng).
E-mail addresses: dchen@ryerson.ca (D.L. Chen), xqzeng@sjtu.edu.cn (X.Q. Zeng).
their normal direction; in the extrusion the c-axes tend to align
transversely to the extrusion axis. These forming processes would
result in the presence of a strong preferred orientation or texture in
Mg alloys. Other mechanisms such as discontinuous recrystalliza-
tion due to grain boundary sliding (GBS) would also produce some
reorientations [7].
It has recently been observed that magnesium alloys containing
rare earth (RE) elements can develop more random textures during
hot extrusion [8,9]. The addition of RE elements in magnesium al-
loys leads to a decrease of the overall texture sharpness allowing
the activation of easy basal slip to a higher extent, compared to
other RE-free wrought magnesium alloys [8,10,11]. Furthermore,
RE elements could result in the formation of a f1 2 1 1g texture
component parallel to the extrusion direction [8,10,11]. In the
presence of strong texture, wrought Mg alloys were prone to
over-concentrated twinning–detwinning activities which may
aggravate the cyclic deformation irreversibility and increase the
propensity of crack initiation [5,6,12]. During cyclic deformation,
the level of strain amplitudes applied at a strain ratio of R = ÿ1
was normally limited to a maximum value of about 1–2% in both
tensile and compressive phases [5]. This would potentially limit
the occurrence of twinning to its full capacity in the compressive
phase during cyclic deformation. Although the addition of RE ele-
ments offers the possibility of effectively improved mechanical
properties via texture weakening, it remains unclear what role

0261-3069/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.matdes.2012.10.040
412 F.A. Mirza et al. / Materials and Design 46 (2013) 411–418

the RE elements play in the formation of twins and how the twin- 3. Results and discussion
ning develops beyond the above strain level, whether the detwin-
ning would occur with increasing compressive strain, and how the 3.1. Microstructure and texture
twinning or detwinning affects the strain hardening behavior. This
study was aimed to identify the deformation behavior of an ex- Fig. 1a shows a typical optical image on the microstructure of
truded Mg–10Gd–3Y–0.5Zr alloy in compression along the extru- Mg–10Gd–3Y–0.5Zr alloy in the as-extruded condition where the
sion direction at different strain levels, with special attention to ED is indicated. For the sake of comparison, the microstructure of
the effect of the RE element addition on the twin formation. an extruded AM30 magnesium alloy is shown in Fig. 1b as well.
It is seen that uniform equiaxed grains with an average grain size
of about 12 lm were obtained in the Mg–10Gd–3Y–0.5Zr alloy
2. Experimental procedure
(Fig. 1a) due to the occurrence of dynamic recrystallization (DRX)
in the hot extrusion process at 400°C, which is fairly small in com-
The material selected was a recently developed extruded mag-
parison with AM30 alloy (Fig. 1b). This was due to the role of added
nesium alloy with the following composition (wt.%): 10Gd, 3Y,
RE elements and zirconium (Zr) where Zr mainly restricted the
0.5Zr, and Mg (balance). The received alloy bars were extruded at
grain growth [13]. A similar role of grain refinement by RE ele-
400 °C with a diameter of 20 mm and an extrusion ratio of 9:1.
ments, e.g., gadolinium [14], yttrium [15], cerium [16], neodymium
Microstructural examinations were performed using an optical
[17], erbium [18], has been reported as well. For example, an addi-
microscope (OM), scanning electron microscope (SEM, JEOL, JSM-
tion of Gd into AZ31 alloy resulted in the formation of Al2Gd
6380LV) equipped with Oxford energy dispersive X-ray spectros-
(which was still present after homogenization), and it played an
copy (EDS) system and three-dimensional (3D) surface/fracto-
important role in refining the grain size and improving the rolling
graphic analysis capacity, and transmission electron microscope
capability of the alloy [14]; the cerium exhibited a good grain
(TEM, JEM-2100) operated at 200 kV. Standard metallographic
refinement effect on the as-cast AZ31 alloy [16]; and remarkable
sample preparation techniques were used with an etchant based
grain size refinement of Mg–1.5Zn–0.1Zr alloy with an addition
on an acetic picral solution containing 4.2-g picric acid, 10-ml ace-
of erbium was observed through recrystallization during the roll-
tic acid, 10-ml H2O, and 70-ml ethanol. The average grain size was
ing process [18]. Besides, as seen from Fig. 1 no twins were visible
measured via a linear intercept method. Texture measurements
in the un-deformed samples. Fig. 2a shows a typical SEM image of
were performed using a PANalytical X-ray diffractometer with
as-extruded sample where it contained three different phases as
Cu Ka radiation at 45 kV and 40 mA. The texture were determined
identified by He et al. [19]: a-Mg solid solution matrix phase with
by measuring incomplete pole figures between w = 0 to 75° in a
supersaturated Gd + Y elements, (Gd + Y)-rich eutectic compound
back reflection mode, and analyzed using PANalytical X’Pert Tex-
which had a higher Gd + Y content than the matrix, and intracrys-
ture software. Cylindrical specimens with a diameter of 5 mm
talline Zr-rich cores. By means of EDS point analysis, the content at
and height of 8 mm were machined with the compression axis par-
points A, B, and C of Fig. 2a was estimated to be 81.7% Mg–4.9% Y–
allel to the extrusion direction (ED) of the as-received bars. Com-
13.4% Gd, 62.5% Mg–6.1% Y–31.1% Gd, and 32.4% Mg–67.6% Zr,
pression tests were conducted using a computerized Instron
respectively. Similar observations have been reported for as-cast
machine at a strain rate of 10ÿ3 sÿ1 and stopped at different
samples of Mg–7Gd–3Y–0.4Zr (GW73K) alloy by Liang et al. [20],
amounts of strain at room temperature. The deformed samples
cast Mg–10Gd–3Y–Zr alloy by Wang et al. [21], as-extruded Mg–
were cut along the compression axis diameter using a slow dia-
11.90Gd–0.81Y–0.44Zr alloy by Zhang et al. [22]. In addition, EDS
mond cutter, cold-mounted, ground, polished, and etched to exam-
line scan was performed as shown in Fig. 2b and c which also con-
ine the evolution of deformation twins during compression. A
firmed the presence of RE-rich particles in the as-extruded alloy,
point counting procedure according to ASTM: E562-11 was used
where the middle rectangle-like particle contained both Gd and
to evaluate the volume fraction of twins. Five digital micrographs
Y. Such RE-enrichment particles was also observed via TEM which
were taken from each sample at a magnification of 200, and then
was operated in a microdiffraction mode, as shown in Fig. 3a. Elec-
a grid of 31 points by 39 points was superimposed on the images.
tron microdiffration pattern in Fig. 3b shows the structure of the
The volume fraction was determined as a ratio of the number of
particle which was identified as cuboid-shaped (RE-rich) particles
points positioned within the twins to the total number of grid
(face-centered cubic, fcc with a = 0.56 nm). Similar results were re-
points. It should be noted that in evaluating the stress–strain
ported in a Mg–10Gd–2Y–0.5Zr alloy [19].
curves and strain hardening rates, the deformation amount of ma-
Fig. 4 shows a basal texture where the reflecting surface was
chine system itself was eliminated using a calibration curve so as
parallel to the extrusion direction (ED) and the radial direction
to obtain the actual or net deformation of test samples.

(a) (b)

ED ED

Fig. 1. Microstructures of (a) as-extruded Mg–10Gd–3Y–0.5Zr, and (b) as-extruded AM30 alloy.
 F.A. Mirza et al. / Materials and Design 46 (2013) 411–418 413

normal direction and some grains with about half of the maximum
(a) C basal intensity were oriented toward the RD. The presence of such
a weaker texture was another benefit of the RE elements added
into Mg alloys, as also reported by Stanford and Barnett [11] who
observed that microalloying with RE elements can weaken the
A extrusion texture. A similar weaker texture in Mg-based alloys
with the addition of RE elements was reported [23]. In the current
Mg–10Gd–3Y–0.5Zr alloy, grain boundary nucleation, coalescence
of low-angle boundaries and shear band assisted nucleation are ab-
sent (as seen in Fig. 1). Zhu et al. [23] reported that only particle-
stimulated nucleation was considered to be a factor influencing
B texture weakening. Indeed, Humphreys [24] observed that the high
RE content and mixed addition resulted in the precipitation of a
large number of the second-phase particles. Such precipitates can
hinder the movement of the dislocations so that the stress could
concentrate easily near the precipitates, which promoted the par-
ticle-stimulated nucleation of dynamic recrystalliazation, and
weaken the basal texture. In addition, the weakening of the texture
was also associated with the appearance of deformation bands
containing twins and the restriction of grain growth [25].
(b)
3.2. Uniaxial compression behavior

The stress–strain curves of the extruded Mg–10Gd–3Y–0.5Zr al-

loy during compression along the ED are shown in Fig. 5a, where
the stress–strain curves of several samples deformed at varying
amounts of strain are plotted together to show the reproducibility
of the compression tests. An average compressive yield strength of
232 ± 4 MPa was obtained from the present multiple tests, which
was higher than the tensile yield strength of 219 MPa for a Mg–
10Gd–2Y–0.5Zr alloy [19]. While the tensile percent elongation
of Mg–10Gd–2Y–0.5Zr alloy was reported to be 25.6% for the rod
also extruded at 400°C [19], which was higher than the compres-
sive percent shortening (18.7%) of the present Mg–10Gd–3Y–
0.5Zr alloy, the ultimate compressive strength of the present alloy
(493 MPa) was significantly higher than the ultimate tensile
strength (305 MPa) of Mg–10Gd–2Y–0.5Zr alloy [19]. This might
2000 be attributed to the favorable role of one more percent yttrium
added, while all other elements remained the same. Furthermore,
(c) in comparison with the compressive test results of an extruded
AM30 alloy [26], the present RE-containing alloy had both higher
(b) ultimate compressive strength and ductility (as indicated by the
1500 Mg compressive percent shortening). In particular, the current Mg–
Intensity, Counts

10Gd–3Y–0.5Zr alloy exhibited over threefold higher compressive

yield strength than the extruded AM30 alloy, i.e., 232 MPa vs.
71 MPa [26]. Again this reflected the significant beneficial effect
1000
of RE elements added into Mg alloys. Furthermore, unlike the sig-
moidal (S-shaped) hardening curves observed in the twinning-
dominated deformation, such as in the extruded AM30 Mg alloy
[26], the compressive stress–strain curves in Fig. 5a showed
500 roughly a parabolic-like hardening, indicating a likely change in
the deformation mechanisms from the twinning-dominated defor-
Gd Y mation to the slip-dominated deformation. To better understand
the deformation behavior of the present alloy, the strain hardening
0 rate dr/de (where r is true stress) with respect to true strain e from
0 5 10 15 20 25 the specimen tested until failure is presented in Fig. 5b along with
Distance, µm the RE-free AM30 extruded Mg alloy [26]. Three stages of strain
hardening in the extruded Mg–10Gd–3Y–0.5Zr alloy could be dis-
Fig. 2. (a) SEM backscattered electron image indicating EDS point analysis locations tinguished from Fig. 5b. Stage A was characterized by a rapidly
(A, B, and C), (b) SEM backscattered electron image indicating EDS line scan decreasing strain hardening rate from the onset of plastic deforma-
position, and (c) the corresponding EDS line scan results in the as-extruded Mg–
tion up to a strain of 4%, followed by stage B with a fairly flat and
10Gd–3Y–0.5Zr alloy.
almost linear strain hardening rate over a large range of strain up
to 18%, and then stage C with an decreasing strain hardening rate
(RD). A basal plane fiber texture was observed, where the c-axis of again until failure. Three stages of strain hardening rate have been
most grains was aligned perpendicular to the ED. The peak inten- reported by Sarker and Chen [26] where stage A and stage C had a
sity of the basal plane split to a few degrees around the surface falling trend, as observed in the present study as well. However, as
414 F.A. Mirza et al. / Materials and Design 46 (2013) 411–418
(a)
Fig. 3. Transmission electron micrographs of the as-extruded Mg–10Gd–3Y–0.5Zr alloy: (a) bright-field image taken along [0 0 1] zone axis of particle A, and (b) [0 0 1] zone
axis microdiffraction pattern recorded from particle A.
Fig. 4. (0 0 0 2) Pole figure of as-extruded Mg–10Gd–3Y–0.5Zr alloy, where ED stands for the extrusion direction and RD indicates the radial direction.
seen from Fig. 5b, the value of strain hardening rate in stage A was
much higher in the present extruded Mg–10Gd–3Y–0.5Zr alloy
than in the extruded AM30 alloy; stage B of the AM30 Mg alloy
was totally different from that of the present alloy. Similar trend
in the strain hardening rate to the Mg–10Gd–3Y–0.5Zr alloy has
also been observed in extruded Mg–Y–Nd alloy [27] and ZK60
Mg alloy [28], where the strain hardening rate started with a sim-
ilarly steep decrease at small strains and then approached a linear/
steady state with a very low value over a large range of strain. The
significant difference in the strain-hardening rate in stages A and B
between the present Mg–10Gd–3Y–0.5Zr and AM30 alloys could
be understood on the basis of the amount of twins formed during
compression and the subsequent interactions between twin and
twin, twin and dislocation, twin–precipitate, etc., which will be
discussed in the next section.
3.3. Twin volume fraction
To understand the strain hardening behavior, the deformation
characteristics of the samples compressed to different amounts
(b)
Intensity Color
1 0.231
26 1.194
51 2.158
76 3.122
100 4.048
ED
RD
of strain were examined. Fig. 6 shows the change of twin volume
fraction with the applied compressive strain, together with some
typical images showing the visual change of twins. It is seen from
image (a) in Fig. 6 that at a compressive strain of e = 0.9%, no twins
were visible, whereas at e = 3.2% a very small number of twins
were observed in some grains (image (b) in Fig. 6). The twins
formed during compression in the extruded Mg alloys along the
ED direction has been identified as f1 0 1  2g extension twins in a
number of publications [23,26]. At a strain of e = 7.7%, more twins
were observed in more grains (image (c) in Fig. 6). When the spec-
imen was deformed at a compressive strain of e = 12.3%, the twin
volume fraction reached its maximum of 0.04 (image (d) in
Fig. 6). Further compression beyond this strain value resulted in
a decrease in the twin volume fraction or the disappearance of
twins, as seen in images (e) and (f) in Fig. 6. In addition, Fig. 7
shows the typical SEM micrographs at strain levels of 3.2%, and
12.3%, respectively. It is obvious that the compressive deformation
at a strain of 12.3% resulted in more twins than that at a strain of
3.2%. The phenomenon of detwinning occurred during pure com-
pression in an extruded Mg alloy was recently reported in [26]. It
 F.A. Mirza et al. / Materials and Design 46 (2013) 411–418
600
(a)
500
Engineering stress, MPa
400
300
200
100
0
0 4 8 12
Engineering strain, %
7000
(b) This investigation
6000
Strain hardening rate, MPa
AM30 [26]
5000
4000 A ning. It was also reported that it is common for the twinning stress
3000
2000 B compared with AM30 alloy [26] due to the grain refinement role
1000
0
0 4 8 12
True strain, %
Fig. 5. (a) Compressive stress–strain curves and (b) strain hardening rate as a
function of true strain of as-extruded Mg–10Gd–3Y–0.5Zr alloy in the extrusion
direction tested at a strain rate of 10ÿ3 sÿ1.
was explained through a twin–dislocation interaction model,
where the moving dislocations during compression cut and de-
stroyed twins, leading to the disappearance of twins [26].
A comparison of the twin volume fraction as a function of the
applied compressive strain between the extruded Mg–10Gd–3Y–
0.5Zr alloy and extruded AM30 alloy is shown in Fig. 8. It is clear
that twinning occurred in the present Mg–10Gd–3Y–0.5Zr alloy
was much less extensive than that in the extruded AM30 alloy
[26]. The maximum volume fraction in the AM30 alloy (0.47), oc-
curred at a compressive strain of 2.5%, was considerably (or over
one order of magnitude) higher than that in the Mg–10Gd–3Y–
0.5Zr alloy (0.04) that appeared at a compressive strain of
12.3%. Other investigators also observed the occurrence of a sig-
nificant amount of twinning in the extruded RE-free Mg alloys
[5,11]. The presence of RE elements in the current Mg–10Gd–3Y–
0.5Zr alloy played a key role in suppressing the excessive twinning
and delaying the appearance of the maximum twinning. It is also
due to the addition of RE elements that resulted in a huge differ-
ence in the strain hardening characteristics between Mg–10Gd–
3Y–0.5Zr alloy and AM30 alloys, especially in stages A and B
(Fig. 5b). Explanations for this are given below.
Plastic deformation of Mg alloy is mainly determined by the
hexagonal crystal structure and strongly affected by the crystallo-
graphic texture. Thus, crystallographic texture would be one of the
key factors influencing the deformation behavior of Mg–10Gd–3Y–
0.5Zr alloy. As the orientation of each grain changes, the combina-
tion of active slip system changes [29,30]. When a change of active
slip systems occurs, the flow stress will rise which will change the
415
hardening [30]. Most grains re-orient to a stable orientation so that
the effect of flow stress would be expected to saturate and not lead
to any new strain hardening at very large strains. As the addition of
RE elements offers an avenue to control or change the crystallo-
graphic texture (as mentioned in Section 3.1), it would thus affect
the slip and twinning systems and change the deformation behav-
ior. It was observed that, in the case of a weak basal texture (Fig. 4),
it is expected that hai dislocation slip with a relatively low CRSS
would contribute more to accommodate the deformation and twin
up to failure
formation will be more difficult [31]. Moreover, Hantzsche et al.
12.3% [25] reported that the deformation texture changes at a higher Y
7.7% content, which is related to an increasing activity of hc + ai slip
3.2% and thus can also reflect in the twinning activity.
0.9% There is a general agreement that a smaller grain size inhibited
twinning [32] and grain size had a significant effect on the ten-
16 20
dency of twinning since the energy required to form twin inter-
faces is particularly high in Mg [33]. Finer grains, which have a
higher surface-to-volume ratio, increase the difficulty of twin
nucleation and thereby suppress twinning by limiting the sizes of
the twins [34]. The addition of RE elements and Zr in the present
Mg–10Gd–3Y–0.5Zr alloy led to a fairly fine grain size, as seen from
Fig. 1. Thus in the present alloy, the grain size would be another
important factor influencing the incidence of deformation twin-
to increase with decreasing grain size more rapidly than the stress
required to activate slip [35]. As a result, the formation of deforma-
tion twins was more difficult in the Mg–10Gd–3Y–0.5Zr alloy as
C of RE elements as mentioned in the above Section 3.1. In addition
to the effect of texture and grain size, the presence of precipitates
has also been reported to impede the process of twinning in mag-
16 20 nesium alloys [36]. During plastic deformation as the twinning
ledge approaches a particle there exists a back-stress arising from
the rigid particle inhibiting the free shear of the moving twin ledge.
This back-stress must be overcome by the applied stress. When the
shear takes place at the interface between the matrix and the pre-
cipitate, the movement of atoms toward the precipitate must be
accommodated by basal slip away from the particle/matrix inter-
face [36]. Then the need for additional basal slip on a local scale
around the precipitates is likely to increase the stress required to
form the twin. As a result, the second-phase particles in the Mg–
10Gd–3Y–0.5Zr alloy (Figs. 2 and 3) would modify potential twin
nucleation sites and hinder the rate of twin formation. Similar re-
sults were reported in Ref. [28].
It was reported that at large compressive strains the spacing be-
tween the second-phase decreased [29]. If the second-phase is
hard enough to block the dislocation flow or twinning in the ma-
trix, the mean free path for dislocation motion and twinning de-
creases strongly with strain and leads to rises of the flow stress
by the Hall–Petch effect, thus contributing to the strain hardening
[29,32]. In addition, strain hardening is in nature associated with
elastic and inelastic incompatibilities between matrix and particles
during plastic deformation. The direction of this resistance is oppo-
site to the applied stress, and therefore opposes the bulk slip and
twinning, contributing to the overall strain hardening as well. Fur-
thermore, the plastic relaxation of the matrix at the RE particle
interface could create both deformation twins and a high local dis-
location density, which are obstacles to the moving dislocations,
and would also affect the strain hardening [37]. Similar result
regarding the effect of second-phase particles on the strain harden-
ing has also been reported in Ref. [25]. Therefore, the presence of
RE-containing particles in conjunction with finer grain sizes and
weaker texture is responsible for the much higher strain hardening
rate in stage A in the Mg–10Gd–3Y–0.5Zr alloy (Fig. 5b), due to the
increased difficulty on the slip of dislocations and the formation of
twins.
416 F.A. Mirza et al. / Materials and Design 46 (2013) 411–418
0.08
0.07
(c)
0.06
Twin volume fraction
0.05
(b)
0.04
0.03 20 µm
0.02
(a)
0.01
0
0 5
Fig. 6. Variation of the volume fraction of twins with the applied strain during compression in the as-extruded Mg–10Gd–3Y–0.5Zr alloy, where some typical micrographs at
different strain levels are inserted to show the evolution of deformation twins.
Fig. 7. Typical SEM micrographs of as-extruded Mg–10Gd–3Y–0.5Zr alloy deformed in compression at a strain level of (a) 3.2%, and (b) 12.3%.
The concept of dislocation hardening could also be used to
explain the linear portion in stage B of the strain hardening rate
curve of Mg–10Gd–3Y–0.5Zr alloy, as shown in Fig. 5b. Several
models have been proposed to explain stage B (in some studies
which is referred to as stage IV) [30,38–40]. It was reported that
stage B might be associated with microscopic instabilities such
as microbands commonly observed at large strains in copper
[38]. The cell model proposed by Nix et al. [40] indicated that
plastic straining at low temperatures leads to the accumulation
of dislocation debris such as dipoles and loops. The accumulation
of dipolar debris may occur as part of the glide process and it
may also occur as part of the dislocation rearrangements due
to dynamic recovery [39]. However, the rate of dislocation
rearrangement inside the cell walls is expected to be affected
by the debris left and would have forward internal stresses with-
in the cell walls that lead to rearrangement or even annihilation
(d)
20 µm 20 µm
(e)
20 µm
(f)
20 µm
20 µm
10 15 20 25
Strain, %
of dislocation segments in this region. If the effect of the accu-
mulating is to raise the flow stress within the cell walls of the
dislocation substructure, then superposition of flow stress would
contributed in hardening and ultimately leads to a finite
hardening rate in stage B (fairly flat linear or sometimes con-
stant). Similar observations of stage IV or stage B in different ex-
truded Mg alloys were reported in several studies [27,28]. The
above discussion indicated that the addition of RE elements
was able to affect significantly the deformation behavior of the
extruded Mg alloys by predominantly suppressing the twinning
due to the finer grain size, the relatively more random crystallo-
graphic texture, and the presence of RE-containing precipitates.
However, more studies related to the dislocation–twin interac-
tions, twin–particle interactions, twin–grain boundary interac-
tions from both experimental and modeling/theoretical aspects
are needed.
 F.A. Mirza et al. / Materials and Design 46 (2013) 411–418
0.6
This investigation
AM30 [26]
Twin volume fraction
0.4
0.2
0 Scripta Mater 2008;59:772–5.
0 5 10 15 20
Strain, %
Fig. 8. A comparison of twin volume fraction as a function of the applied strain
during compression between the extruded Mg–10Gd–3Y–0.5Zr alloy and extruded
AM30 alloy.
4. Conclusions
The deformation behavior of as-extruded Mg–10Gd–3Y–0.5Zr
magnesium alloy in compression along the extrusion direction
was characterized by a rapidly decreasing strain hardening rate
up to a strain level of 4% (stage A), followed by a fairly flat linear
strain hardening rate over an extended strain range from 4% to
18% (stage B). Stage C was represented by a decreasing strain
hardening rate just prior to failure. The amount of twins observed
in the present alloy was much less extensive than that in the RE-
free extruded AM30 Mg alloy, with a higher strain value of
12.3% corresponding to the maximum twin volume fraction.
The weaker crystallographic texture, refined grain size, and RE-
containing second-phase particles were responsible for the much
higher strain hardening rate in stage A in the Mg–10Gd–3Y–0.5Zr
alloy due to the increased difficulty on the formation of twins
and the slip of dislocations at lower strains, and for the occurrence
of fairly flat linear strain hardening rate in stage B at higher strains
which was likely associated with the presence of dislocation debris
and twin debris (or residual twins) resulting mainly from disloca-
tion–twin interactions as well as the interactions between the dis-
locations/twins and second-phase particles and grain boundaries.
Acknowledgements
The authors would like to thank the Natural Sciences and Engi-
neering Research Council of Canada (NSERC), Premier’s Research
Excellence Award (PREA), NSERC-DAS Award, AUTO21 Network
of Centers of Excellence, and Ryerson Research Chair program for
providing financial support. This work was also co-funded by Tech-
nology Support Project Program of the Ministry of Science and
Technology of China (No. 2011BAE22B02 and No.
2011DFA50907), and the project (No. 51171113) received from
the National Natural Science Foundation of China. This investiga-
tion involves part of Canada–China–USA Collaborative Research
Project on the Magnesium Front End Research and Development
(MFERD). The authors also thank Q. Li, C. Ma, A. Machin, J. Amank-
rah, and R. Churaman for assisting in the experiments.
References
[1] Shakun JD, Clark PU, He F, Marcott SA, Mix AC, Liu ZY, et al. Global warming
preceded by increasing carbon dioxide concentrations during the last
deglaciation. Nature 2012;484:49–54.
417
[2] Murray J, King D. Oil’s tipping point has passed. Nature 2012;481:433–5.
[3] Pollock TM. Weight loss with magnesium alloys. Science 2010;328:986–7.
[4] Zhang Y, Zeng XQ, Lu C, Ding WJ. Deformation behavior and dynamic
recrystallization of a Mg–Zn–Y–Zr alloy. Mater Sci Eng A 2006;428(1–2):91–7.
[5] Begum S, Chen DL, Xu S, Luo AA. Effect of strain ratio and strain rate on low
cycle fatigue behavior of AZ31 wrought magnesium alloy. Mater Sci Eng A
2009;517:334–43.
[6] Lin XZ, Chen DL. Strain controlled cyclic deformation behavior of an extruded
magnesium alloy. Mater Sci Eng A 2008;496(1–2):106–13.
[7] Agnew SR, Duygulu O. Plastic anisotropy and the role of non-basal slip in
magnesium alloy AZ31B. Int J Plast 2005;21:1161–93.
[8] Stanford N, Sha G, Xia JH, Ringer SP, Barnett MR. Solute segregation and texture
modification in an extruded magnesium alloy containing gadolinium. Scripta
Mater 2011;65:919–21.
[9] Wu D, Chen RS, Tang WN, Han EH. Influence of texture and grain size on the
room-temperature ductility and tensile behavior in a Mg–Gd–Zn alloy
processed by rolling and forging. Mater Des 2012;41:306–13.
[10] Stanford N, Atwell D, Beer A, Davies CH, Barnett MR. Effect of microalloying
with rare-earth elements on the texture of extruded magnesium-based alloys.
[11] Stanford N, Barnett MR. Effect of composition on the texture and deformation
behaviour of wrought Mg alloys. Scripta Mater 2008;58:179–82.
[12] Zhang XP, Castagne S, Luo XF, Gu CF. Effects of extrusion ratio on the ratcheting
behavior of extruded AZ31B magnesium alloy under asymmetrical uniaxial
cyclic loading. Mater Sci Eng A 2011;528(3):838–45.
[13] Peng Z-K, Zhang X-M, Chen J-M, Xiao Y, Jiang H. Grain refining mechanism in
Mg–9Gd–4Y alloys by zirconium. Mater Sci Technol 2005;21(6):722–6.
[14] Li WP, Zhou H, Li ZF. Effect of gadolinium on microstructure and
rolling capability of AZ31 alloy. J Alloys Compd 2009;475(1–2):
227–32.
[15] Fang XY, Yi DQ, Wang B, Luo WH, Gu W. Effect of yttrium on microstructures
and mechanical properties of hot rolled AZ61 wrought magnesium alloy. Trans
Nonferrous Met Soc China 2006;16(5):1053–8.
[16] Yu K, Rui ST, Song JM, Li WX, Guo L. Effects of grain refinement on mechanical
properties and microstructures of AZ31 alloy. Trans Nonferrous Met Soc China
2008;18:s39–43.
[17] Jia SG, Liu P, Dong QM, Xia CQ, Tian BH, Ren FZ. Effect of rare earth neodymium
on microstructure and properties of AZ31B magnesium alloy. Trans Mater
Heat Treat 2010;31(7):99–103.
[18] Zhang J, Zhang XF, Pan FS, Liu CP, He QB. Microstructure and mechanical
properties of an Mg–Zn–Zr alloy modified by rare earth of Erbium. Mater Sci
Forum 2009;610–613:810–4.
[19] He SM, Zeng XQ, Peng LM, Gao X, Nie JF, Ding WJ. Microstructure and
strengthening mechanism of high strength Mg–10Gd–2Y–0.5Zr alloy. J Alloys
Compd 2007;427:316–23.
[20] Liang SQ, Guan DK, Chen L, Gao ZH, Tang HX, Tong XT. Precipitation and its
effect on age-hardening behavior of as-cast Mg–Gd–Y alloy. Mater Des
2011;32:361–4.
[21] Wang Q, Wu G, Zheng HQ, Chen B, Zheng Y, Ding W. A comparative study of
Mg–Gd–Y–Zr alloy cast by metal mould and sand. China Foundry J
2010;7(1):6–12.
[22] Zhang F, Zhang KX, Tan CW, Yu XD, Ma HL, Wang FC, et al. Microstructure and
mechanical properties of Mg–Gd–Y–Zr alloys processed by equal channel
angular pressing. Trans Nonferrous Met Soc China 2011;21:2140–6.
[23] Zhu R, Wu YJ, Wang JT, Lu KC. Mechanical anisotropy of extruded Mg–10Gd–
2Y–0. 5Zr alloy. Adv Mater Res 2011;320:222–7.
[24] Humphreys FJ. Unified theory of recovery, recrystallization and grain growth,
based on the stability and growth of cellular microstructures – II. The effect of
second-phase particles. Acta Mater 1997;45(12):5031–9.
[25] Hantzsche K, Bohlen J, Wendt J, Kainer KU, Yia SB, Letting D. Effect of rare earth
additions on microstructure and texture development of magnesium alloy
sheets. Scripta Mater 2010;63:725–30.
[26] Sarker D, Chen DL. Detwinning and strain hardening of an extruded
magnesium alloy during compression. Scripta Mater 2012;67(2):165–8.
[27] Xin RL, Song B, Zeng K, Huang GJ, Liu Q. Effect of aging precipitation on
mechanical anisotropy of an extruded Mg–Y–Nd alloy. Mater Des
2012;34:384–8.
[28] Chen XH, Pan FS, Mao JJ, Wang JF, Zhang DF, Tang AT, et al. Effect of heat
treatment on strain hardening of ZK60 Mg alloy. Mater Des 2011;32:1526–30.
[29] Rollett AD, Kocks UF, Doherty RD. Stage IV work hardening in cubic metals.
Symp on formability metall struct. The Metallurgical Society, AIME; 1987. p.
211–25.
[30] Haasen P. A cell theory for stage IV work hardening of metals and
semiconductors. J Phys France 1989;50:2445–54.
[31] Yi S, Bohlen J, Heinemann F, Letzig D. Mechanical anisotropy and deep drawing
behaviour of AZ31 and ZE10 magnesium alloy sheets. Acta Mater
2010;58:592–605.
[32] Partridge PG. Crystallography and deformation modes of hexagonal close
packed metals. Met Rev 1967;12:169–94.
[33] Koike J, Kobayashi T, Mukai T, Watanabe H, Suzuki M, Maruyama K, et al. The
activity of non-basal slip systems and dynamic recovery at room
temperature in fine-grained AZ31B magnesium alloys. Acta Mater
2003;51:2055–65.
[34] Dobron P, Chmelik F, Yi S, Parfenenko K, Letzig D, Bohlen J. Grain size effects on
deformation twinning in an extruded magnesium alloy tested in compression.
Scripta Mater 2011;65:424–7.
418 F.A. Mirza et al. / Materials and Design 46 (2013) 411–418

[35] Yin SM, Wang CH, Diao YD, Wu SD, Li SX. Influence of grain size and texture on
the yield asymmetry of Mg–3Al–1Zn Alloy. J Mater Sci Technol
2011;27(1):29–34.
[36] Stanford N, Barnett MR. Effect of particles on the formation of deformation
twins in a magnesium-based alloy. Mater Sci Eng A 2009;516:226–34.
[37] Yakubtsov IA, Diak BJ, Sager CA, Bhattacharya B, MacDonald WD, Niewczas M.
Effects of heat treatment on microstructure and tensile deformation of Mg
AZ80 alloy at room temperature. Mater Sci Eng A 2008;496:247–55.
[38] Kocks UF, Mecking H. Physics and phenomenology of strain hardening: the FCC
case. Prog Mater Sci 2003;48(3):171–273.
[39] Argon AS, Haasen P. A new mechanism of work hardening in the late stages of
large strain plastic flow in fcc and diamond cubic crystals. Acta Met Mater
1993;41(1):3289–306.
[40] Nix W, Gibeling J, Hughes D. Time-dependent deformation of metals. Met
Trans 1985;16A:2215–26.