You are on page 1of 12

Materials and Design 45 (2013) 253264

Contents lists available at SciVerse ScienceDirect

Materials and Design


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

Evaluation of fatigue life of AZ31 magnesium alloy fabricated by squeeze casting


Miroslava Horynov , Josef Zapletal, Pavel Dolezal, Pavel Gejdo
Institute of Material Science and Engineering, Faculty of Mechanical Engineering, Brno University of Technology, Technick 2896/2, 616 69 Brno, Czech Republic

a r t i c l e

i n f o

Article history:
Received 16 February 2012
Accepted 27 August 2012
Available online 12 September 2012
Keywords:
Magnesium alloy
Squeeze casting
Cyclic properties
Low cycle fatigue
High cycle fatigue

a b s t r a c t
Cyclic deformation behavior and fatigue life of squeeze-cast AZ31 magnesium alloy was studied under
stress amplitude-control at room temperature. Low and high cycle fatigue tests with engineering stress
amplitudes in the range from 40 to 110 MPa were conducted. Analysis of hysteresis curves was performed. Tensioncompression asymmetry of hysteresis loops was not observed; the alloy exhibited cyclic
hardening in tension and compression. The fatigue life in the low cycle fatigue region was expressed by
Whler and derived MansonCofn curves. Experimental data in both, the low and high cycle fatigue
regions were tted by means of regression functions. SN curves exhibited a smooth transition from
the low to the high cycle fatigue regions and signicant scattering of experimental points was observed.
Furthermore, metallographic and fractographic analyses were performed. Crack initiation occurred from
the specimen surface or on clusters of secondary particles; the region of nal fracture was characterized
by a transgranular ductile fracture.
It can be concluded that the fatigue properties of squeeze cast magnesium alloy AZ31 are signicantly
improved comparing to materials prepared by common methods of casting. Squeeze casting also enables
the cost-effective fabrication of complicatedly shaped parts.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
Magnesium alloys are increasingly used in the automotive and
aircraft industries. Reducing the weight of components while
maintaining good mechanical properties leads to reduced fuel consumption and more economical and environmentally-friendly
operation. To predict the behavior of the material under cyclic
loading, it is essential to describe the fatigue behavior of magnesium alloys in the low and high cycle regions.
The fatigue behavior of AZ31 magnesium alloy in the low and
high cycle regions has already been investigated in the literature.
However, there have only been limited studies on the fatigue
behavior of cast magnesium alloys. The low cycle fatigue behavior
of extruded AZ31 was studied using strain controlled pushpull
tests [1] and rotating bending fatigue tests [2]. Hasegawa et al.
[3] investigated the low cycle fatigue behavior of extruded AZ31
magnesium alloy using both, stress and strain controlled tests.
The authors examined the effect of mean stress and analyzed cyclic
stressstrain behavior. Stress and strain controlled fatigue tests of
different die cast magnesium alloys were alos performed by Sonsino and Dieterich [4]. The inuence of notches, mean stress, and
elevated temperatures was evaluated and compared to other structural materials. Zberov et al. [5] studied the tensile, low and high
cycle fatigue behavior of AZ31 magnesium alloy fabricated by
Corresponding author. Tel.: +420 54114 3147.
E-mail address: horynova@fme.vutbr.cz (M. Horynov).
0261-3069/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.matdes.2012.08.079

squeeze casting (SC), hot rolling (HR), and equal channel angular
pressing (ECAP). The average grain size for SC was 450 lm; the
grain size for HR ranged from 3 to 20 lm, and the grain size for
ECAP ranged 12 lm; yield strength was 50 MPa for SC, 175 for
HR, and 115 for ECAP. The strong inuence of production procedure on fatigue properties was found. The fatigue limit of SC material was 40 MPa; HR and ECAP materials exhibited about 95 MPa.
Chamos et al. [6] investigated the tensile and fatigue behavior of
hot rolled AZ31 and AZ61 magnesium alloys. Longitudinal and
transverse directions were evaluated and the fracture surfaces of
specimens were examined. Matsuzuki and Horibe [7] studied the
inuence of heat treatment on the fatigue behavior of extruded
AZ31 magnesium alloy. Plastic strain controlled tests were conducted on as-extruded and annealed specimens. The fatigue life
of as-extruded material, was slightly longer than that of heat treated material. No other signicant difference was found in the
fatigue behavior of the experimental materials.
Casting is the most economical way to transform material into
shaped components. The major drawback of casting techniques is
the formation of defects such as porosity or segregation defects
that can be potential crack initiation sites during the life of the
component. New casting techniques such as squeeze casting have
been developed to minimize casting defects and related problems
[8,9].
Squeeze casting is a technique in which the material solidies
under high pressure and has signicant advantages such as an
improvement in mechanical properties and the possibility to

254

M. Horynov et al. / Materials and Design 45 (2013) 253264

Nomenclature
a ()

extrapolated value of the function in the point of inection for N = 1


A (%)
elongation
b ()
fatigue strength exponent
c ()
fatigue ductility coefcient
C ()
parameter of KV function
E (GPa) Youngs modulus
m ()
exponent of the RambergOsgood function
K (MPa) strain hardening coefcient
K0 (MPa) cyclic hardening coefcient
n ()
strain hardening exponent
n0 ()
cyclic hardening exponent
N ()
number of cycles
Nf ()
number of cycles to failure
transition number of cycles
Nt ()
R ()
stress ratio
Rm (MPa) ultimate tensile strength

produce parts of complicated shape. This process is also capable of


producing components from difcult-to-process materials. Components fabricated by this method are ne grained with excellent
surface nish and have almost no porosity. The applied pressure
also leads to a remarkable decrease in the secondary dendrite
arm spacing [811].
Two methods of squeeze casting, direct and indirect, are known.
The direct method requires a precise amount of molten metal,
which is poured into the bottom part of a pre-heated die. Then the
die is closed by the upper part and pressure is applied so that the
molten metal solidies under pressure. High pressure eliminates
the formation of a gap between the die and the casting, leading to
more effective heat transfer and a higher rate of cooling. The fully
solidied casting is removed from the die by ejectors. A machine
for indirect squeeze casting has a dosing chamber which is placed
under the die and can be tilted during lling process. When the dosing chamber is lled, it is moved back to the working position and
attached to the die and molten metal is forced into the die. Due to
the low rate of lling, no turbulent ow or oxidation occurs. Once
the die is lled, the maximum value of applied pressure is achieves
and the casting solidies under pressure. And again, fully solidied
casting is removed from die by ejectors. The disadvantage of the
indirect squeeze casting method is the presence of a sprue and the
need for a more sophisticated die, which is more complicated to
produce than dies for the direct method. On the other hand, direct
method requires a precise amount of molten melt.
Chadwick and Yue [12] studied the mechanical properties of
sand-cast, gravity die-cast a squeeze-cast AZ91 magnesium alloy
in as cast state and after heat treatment. In both as-cast and
heat-treated state, the squeeze-cast material exhibited higher values of ultimate tensile strength, yield strength and elongation than
materials cast by other technologies. The effect of applied pressure
on the density, macrostructure and hardness of squeeze cast AlSi
alloy LM13 were investigated by Maleki et al. [13]. An applied pressure higher than 50 MPa resulted in an increase in density and a
pressure higher than 100 MPa led to the elimination of gas and
shrinkage porosities. An increase in pressure up to 100 MPa also
resulted in renement of the macrostructure due to better contact
between the metal and the die surface during solidication. The
hardness of the alloy increased with increasing pressure due to
renement of the microstructure and the modication of eutectic
particles.
In the present study, low and high cycle engineering stress controlled fatigue tests were conducted on AZ31 magnesium alloy
fabricated by squeeze casting. The microstructure, cyclic deforma-

Rp0.2 (MPa) proof stress


R0p0:2 (MPa) cyclic yield strength
R00p0:2 (MPa) cyclic yield strength established by the Ramberg
Osgood function
VH ()
loop shape parameter
W (MJ/m3) area of hysteresis loop
S ()
sum of squares
eap () plastic strain amplitude
eae () elastic strain amplitude
eat () total strain amplitude
em () mean strain
ef0 ()
fatigue ductility coefcient
ra (MPa) stress amplitude
rf0 (MPa) fatigue strength coefcient
rc (MPa) fatigue limit
r1 (MPa) permanent fatigue limit

tion behavior, and low and high cycle fatigue life were investigated
and compared to available data on AZ31 magnesium alloys prepared by different technologies.
2. Experimental details
2.1. Material and specimens
The experimental material used was AZ31 magnesium alloy
fabricated by squeeze casting at ZFW GmbH in Clausthal. The direct squeeze casting method was used: molten metal was
squeezed and solidied at a pressure of 150 MPa. The material
was supplied in the form of billet 200 mm in diameter and
40 mm in height. As squeeze casting is a method producing a ner
grain and more homogeneous structure than other casting methods, no additional heat treatment was performed. The chemical
composition measured by a Spectrumat GDS 750 optical emission
spectrometer with glow discharge is shown in Table 1. The basic
mechanical properties (Young0 s modulus, ultimate tensile strength,
proof stress and elongation), strain hardening coefcient and strain
hardening exponent established by means of tensile tests are given
in Table 2. Tensile tests were performed according to EN ISO 68921 [14] on a PC controlled testing device with a strain rate of
0.00025 s1 using cylindrical specimens with a diameter of 6 mm
and gauge length of 30 mm. Strain was measured by an axial
extensometer with a gauge length of 30 mm. A representative
engineering stressstrain curve is given in Fig. 1.
As shown in Fig. 2, cylindrical specimens of two different geometries were machined from experimental material and used for fatigue tests. The test specimens were machined so that the axes of
the samples were perpendicular to the axis of the billet. The surface of the gauge section was 0.4 Ra ground to remove the machining marks and to achieve a smooth surface.
2.2. Experimental procedure
Specimens for metallographic assessment were prepared in the
usual way and etched with a mixture of picric acid (5 ml acetic acid,
6 g picric acid, 10 ml water, and 100 ml ethanol). Microstructural
Table 1
Chemical composition of AZ31 alloy (wt.%).
Al

Zn

Mn

Si

Fe

Zr

Mg

2.86

1.07

0.35

0.005

0.004

0.01

Bal.

M. Horynov et al. / Materials and Design 45 (2013) 253264

255

identify fatigue crack initiation sites and mechanisms of nal


fracture.

Table 2
Tensile properties of AZ31 alloy.
E (GPa)

Rp0.2 (MPa)

Rm (MPa)

A (%)

K (MPa)

40.66

55

174

8.8

404.8

0.32

3. Experimental results
3.1. Microstructure of AZ31 magnesium alloy

Fig. 1. Tensile stressstrain curve of AZ31 magnesium alloy.

Fig. 3 shows the secondary electron micrograph of AZ31 magnesium alloy and elemental maps measured by energy-dispersive Xray spectroscopy. According to the results of local analysis of
chemical composition (Table 2), the structure of AZ31 alloy is
formed by the solid solution d phase, the intermetallic c phase
(Mg17Al12) and the intermetallic u phase (Mg21(Al, Zn)17), eutectic
(d + c), and AlMn-based particles.
Several differently shaped AlMn particles were analyzed.
According to the AlMn binary phase diagram [17], a Mn content
of between 34 and 39.5 at.% corresponds to the c2 phase (Al8Mn5). The calculated Al/Mn ratio of 1.7 0.2 also indicates that
these particles are most likely Al8Mn5.
The material is heterogeneous; it is evident from the analysis of
element distribution (Fig. 3) that there is an Al- and Zn-rich d phase
along the grain boundaries. The difference in aluminum content
between the d phase and enriched d phase was about 3 at.%. The
average grain size of the material was about 50 lm.
3.2. Cyclic plasticity and low-cyclic fatigue

Fig. 2. Shape and dimension of specimens for (a) high-resonance pulsator and (b)
servo-hydraulic testing system.

evaluation, local analysis of chemical composition, and the analysis


of elemental distribution were performed on PHILIPS XL30 scanning electron microscope with EDX analyzer. Grain size was estimated by the linear intercept method using an Olympus GX71
light microscope with Olympus Stream motion software.
Low-cycle fatigue behavior in the stress amplitude control mode
was determined on an Instron 8801 servo-hydraulic testing system.
Experiments were conducted with engineering stress amplitudes in
the range of from 45 to 110 MPa at constant frequencies of 3 Hz and
20 Hz. To assess fatigue behavior in the high-cycle regime, fatigue
tests with engineering stress amplitudes in the range from 40 to
60 MPa and at a frequency of 130 5 Hz were conducted on an
Amsler HPF 1478 high-resonance pulsator. The symmetrical sine
cycle (R = 1) was used and all tests were carried out under laboratory conditions at room temperature according to ASTM E46696(2002) [15] ASTM E606-92(2004) [16]. Strain was measured by
an axial extensometer with a gauge length of 12.5 mm. Hysteresis
loops were obtained using LCF INSTRON software; in addition maximum and minimum values of stress and strain and Young0 s modulus were obtained for each cycle.
The fracture surfaces of fatigue specimens were examined using
a PHILIPS XL30 scanning electron microscope with EDX analyzer to

As the shape of hysteresis curves indicates whether a material


undergoes cyclic hardening or softening or remains cyclically stable, hysteresis curves were analyzed.
The cyclic response of the material under cyclic loading with
engineering stress amplitudes of 65 and 110 MPa is shown in
Fig. 4. To make evident the changes in shape the of hysteresis
loops, only curves for some cycles are presented. Namely cycles
number 4, 40, 400, 4000, and 20000 and cycles number 2, 16,
160, and 800 are shown for the two stress amplitudes respectively.
Cycles No. 4 and 2 roughly correspond to Nf/1000; cycles No. 20000
and 800 roughly correspond to Nf/2.
The elastic parts of the hysteresis loops are rather short, compared to the plastic parts. In both tension and compression regions,
symmetric deformation response occurs except for the rst cycles
at higher stress amplitudes. For higher stress amplitudes, concave
regions can be observed in tension and compression (e.g. the second and sixteenths cycle in Fig. 4b).
At each stress amplitude, the width of the loops decreases with
increasing number of cycles suggesting that the material cyclically
hardens.
Loop shape parameters VH were obtained using Eq. (1). The area
of hysteresis loop W, representing the specic energy dissipated in
a material within one cycle was determined by numerical integration using the trapezoidal rule. Fig. 5 shows the dependence of hysteresis loop area W, plastic strain amplitude eap and loop shape
parameters VH on the number of cycles. As the plastic strain amplitude and area of the hysteresis loop decrease, the material hardens
without saturation. The parameter VH increase gradually after a
steeper increase during the rst cycles; no local minima were
observed.

VH

W
4eap  ra

Fig. 6 shows the evolution of plastic strain amplitude with number of cycles at different stress amplitudes; the evolutions of total
and mean strain are illustrated in Figs. 7 and 8, respectively. A
change of 5% in the effective modulus was used as a criterion for
determining Nf.

256

M. Horynov et al. / Materials and Design 45 (2013) 253264

Fig. 3. (a) Secondary electron micrograph of AZ31 magnesium alloy and elemental distribution maps measured by energy-dispersive X-ray spectroscopy: (b) Al, (c) Mn, (d)
Zn, (e) EDX spot analysis of AlMn based particle.

With an increasing number of cycles, the plastic and total strain


amplitudes gradually decrease until the rapid increase due to crack
opening. Though the experimental material was cast, curves show
the same trend for all stress amplitudes and no inconsistency was
observed. As seen from comparison of Figs. 6 and 7, plastic deformation is greater than elastic deformation during the rst cycles.
The mean strain is constant or very gradually decreases until the
rapid increase due to crack opening.
At the highest stress amplitudes, signicant plastic response
during the rst half cycle occurs. The material exhibits symmetrical behavior in both compression and tension with a subsequent
decrease in mean strain. As the material is cast, the values of mean
strain are not proportional to stress amplitude.

The cyclic deformation curve (CDC), expressed by the regression


function in loglog coordinates, is plotted in Fig. 9a. Parameters K
and n were obtained using Eq. (2) while Eq. (3) was used to establish the cyclic yield point R0P0:2 (Table 4). For comparison, the static
tensile curve and the cyclic stressstrain curve, together with the
cyclic yield strength R00P0:2 , are shown in Fig. 9b. The experimental
points of the CDC were tted to a modied RambergOsgood function (4); value of Young0 s modulus used in the regression was
determined from the tensile test. The obtained parameters are
given in Table 3. The comparison shows that the cyclic deformation
curve has a greater slope in the elasticplastic region than the
static tensile curve. Thus, the experimental material cyclically
hardens.

M. Horynov et al. / Materials and Design 45 (2013) 253264

257

Fig. 6. Evolution of plastic strain amplitude in engineering stress controlled low


cycle fatigue tests.

Fig. 4. Hysteresis loops for (a) 65 MPa and (b) 110 MPa. Cycle No. 20000 (65 MPa)
and 800 (110 MPa) roughly correspond to Nf/2.
Fig. 7. Evolution of total strain in engineering stress controlled low cycle fatigue
tests.

Fig. 5. Dependence of hysteresis loop area W, plastic strain amplitude eap and loop
shape parameters VH on number of cycles. Each set of symbols represents different
stress amplitude.

ra K 0  enap

2
0

R0p0:2 K 0  0:002n


r
r
er
r0
E

m
4

The Whler curve (Fig. 10) and derived MansonCofn curve


(Fig. 11) were tted by means of regression analysis according to
power laws (5) and (6), respectively [4,17]. Only specimens tested
at low frequencies were considered in this regression analysis. Eq.
(7) [4] was used to plot the dependence of total strain amplitude

Fig. 8. Evolution of mean strain in engineering stress controlled low cycle fatigue
tests.

eat, plastic strain amplitude eap, and elastic strain amplitude eae
on the number of cycles to failure (Fig. 11). The material parameters found by the tting procedure are summarized in Table 5. Furthermore, the transition number of cycles was also established
(Nt = 1025). The intersection of plastic and elastic strain amplitude
dependence on number of cycles is located in the region that is not
covered by experimental data. The amplitudes of elastic and plastic
strain were therefore extrapolated to Nf = 200. The value of the
plastic strain amplitude eap = 6  106 for 107 cycles was obtained
from Eq. (6).

ra r0f  2Nf b

258

M. Horynov et al. / Materials and Design 45 (2013) 253264

Fig. 10. Wohler curve in loglog coordinates tted by power low.

sum of squares of the Stromayer function is lower than the sum of


the squares of the KV function. This indicates that the Stromayer
function better describes the fatigue behavior of the experimental
material. On the other hand, comparing parameter b of the Basquine function (Table 5) with the corresponding parameter shows
that the KV function is the most suitable for describing low-cycle
fatigue behavior.

ra a  Nbf r1
ra r1 
Fig. 9. (a) Cyclic deformation curve in loglog coordinates (b), comparison of static
tensile curve and cyclic stressstrain curve tted to RambergOsgood function.

Table 3
Composition of constituent phases in squeeze cast AZ31 magnesium alloy (at.%).

d phase
c phase Mg17Al12
u phase Mg21(Al, Zn)17
AlMn particle

Mg

Al

Mn

Zn

98.72
77.11
75.22
63.46

1.14
21.42
19.24
24.02

0.04

12.52

0.10
1.47
5.54

Table 4
Summary of material parameters.
Power law
0

RambergOsgood function
0

K (MPa)

445.1

0.23

R0p0:2

(MPa)

106

r0 (MPa)

R00p0:2 (MPa)

469

4.03

101

eap e0f  2Nf c


eat

r0 f
E

 2Nf b e0f  2Nf c

6
7

3.3. Regression analysis of SN curves


Experimental data were tted to a smooth continuous function
using the LevenbergMarquardt nonlinear least squares algorithm.
Commonly used regression functions were applied, using 1000
iterations and 106 tolerance.
The dependence of stress amplitude ra on the number of cycles
to failure Nf is given in Fig. 12. Experimental data in the low and
high-cycle fatigue regions were tted by the Stromayer (8) and
the KohoutVechet (KV) functions (9) [18]. The regression parameters are summarized in Table 6. As can be seen from the table, the

Nf
Nf C

8
b
9

3.4. Fractographic analysis


The fracture surfaces were examined using SEM and energy-dispersive X-ray spectroscopy analysis.
The inuence of stress amplitude on the characteristics of the
fracture surface is given in Fig. 13. The number of initiation sites,
marked with arrows, increased with increasing stress amplitude
as well as the size of region of the nal fracture. Fatigue cracks initiated from the specimen surface except for specimens where crack
initiation took place at clusters of secondary particles.
Fig. 14 shows fracture surface with initiation on cluster of secondary particles. Maximum size of cluster is approximately
550 lm. The area containing increased number of secondary particles is clearly visible on back-scattered electron micrographs
(Fig. 14b and c). The EDX spot analysis (Fig. 14d) shows that those
particles, of approximately 3 lm in size, are AlMn based. Clusters
of AlMn particles were also observed in longitudinal section of fatigued specimens at various distances from the fracture surface.
There was no evidence of fatigue cracks near the clusters which
were further from the specimen surface.
The fracture surface of a sample loaded with a stress amplitude
of 65 MPa is shown in Fig. 15. As can be seen in Fig. 13, multiple
crack initiation occurred at this stress amplitude. Fatigue cracks
initiated from the specimen surface, initiation sites are characterized by cleavage-like facets. Initiation sites are very similar, differing only in the size of the facets. Detail from the initiation site
presented in Fig. 15a shows cleavage plane with river pattern.
Fig. 15b presents higher magnication image of the beginning of
the crack propagation area where ridges parallel to the crack propagation direction can be seen. Striations perpendicular to the crack
propagation direction were observed in the crack propagation area
(Fig. 15c). The region of nal fracture exhibits a transgranular ductile fracture with a pronounced dimple pattern (Fig. 15d).
Fig. 16 shows the fracture surface of the specimen tested at
110 MPa. As can be seen from Fig. 16a, the fracture surface is more
rugged than that of specimen tested at 65 MPa. Multiple crack

M. Horynov et al. / Materials and Design 45 (2013) 253264

Fig. 11. Dependence of total strain amplitude eat, plastic strain amplitude eap, and
elastic strain amplitude eae on the number of cycles in loglog coordinates.

Table 5
Low cycle fatigue parameters.

r0f (MPa)

e0f

410.52

0.169

1.45

0.729

initiation with pronounced ratchet marks between the initiation


sites was observed. Initiation sites were mostly similar to that observed for the specimen tested at 65 MPa. The beginning of the
crack propagation area with a secondary crack is shown in
Fig. 16b. As in the previous sample, ridges parallel to the crack
propagation direction were observed. Propagation area is characterized by striations perpendicular to the crack propagation direction (Fig. 16c). The region of nal fracture exhibits a transgranular
ductile fracture (Fig 16d) similar to the fracture observed for the
specimen tested at 65 MPa.
4. Discussion
The microstructure of AZ31magnesium alloy consists of the
solid solution d phase, the intermetallic c phase (Mg17Al12) and

Fig. 12. SN curves of AZ31 magnesium alloy at R = 1 tted to Stromayer and


KohoutVechet functions.
Table 6
Summary of regression parameters.

a
b
C

r1 (MPa)
S

Stromayer

Kohout-Vechet

1075.3
0.38

43.3
1473.5

0.17
2.03  105
45.8
1531.5

259

the intermetallic u phase (Mg21(Al, Zn)17), eutectic (d + c), and


AlMn-based secondary particles. According to EDX analysis, these
particles are supposed to be Al8Mn5, which is in agreement with
the kinetic model of the solidication of AZ31 in a previous study
[19]. However, another solidication model [20] suggests, that
Al8Mn5 transforms to Al11Mn4 and then, below 250 C, transforms to AlMn4.
Based on binary MgAl (Fig. 17) [21] and ternary MgAlZn
(Fig. 18) diagrams [22], the presence of eutectic and such amounts
of c phase are not supposed. The microstructure of wrought and
cast AZ31 magnesium alloy has been investigated in the literature.
According to the authors of articles [2325] the microstructure of
wrought alloy consists of solid solution and AlMn(Fe) based particles, Song and Xu [26], Cheng et al. [27] and Shahri et al. [28] also
found a small amount of intermetallic c phase. Uniformly distributed ne particles of intermetallic c phase with maximum a volume fraction of 5% were observed in cast AZ31 alloy [29]. The
presence of eutectic and a higher amount of c phase in the experimental material can be explained by a high cooling rate when
using the squeeze casting method. Due to the high cooling rate,
segregation and structural heterogeneity occurs and locally increased concentrations of aluminum result in the precipitation of
eutectic and a higher amount of c phase.
A minimum of casting defects was observed in the experimental
material. Clusters of AlMn based particles were locally found,
which may play the role of initiation sites during the fatigue process. Even so, the presence of manganese in AZ magnesium alloys
is desirable. It improves the corrosion resistance of AZ type magnesium alloys by lowering the corrosion potential of secondary phase
with iron [30,31].
The basic mechanical properties of the experimental material,
established via tensile tests, are worse than those of hot rolled
[3,25] and extruded [1] AZ31 alloys. Higher strength of wrought alloys is caused by ner grain, a higher dislocation density and a
more homogeneous structure compared to our experimental material. Because of the experimental materials structural heterogeneity, fracture occurred without the formation of neck which results
in lower ductility of the studied material.
From an analysis of the hysteresis curves it was established that
the material undergoes marked cyclic hardening at each stress
amplitude while cyclic softening was not observed. This nding
is in agreement with literature [1,3,5,7,32]. It is well known, that
material undergoes cyclic softening when containing high dislocation density (e.g. wrought materials) or being age hardened. Cyclic
loading leads to the formation of complicated dislocation structures such as persistent slip bands with different dislocation
arrangement compared to the matrix. Persistent slip bands are
softer than the matrix, which leads to the localization of plastic
strain and cyclic softening. Age hardened materials soften when
coherent precipitates are cut by dislocations and subsequently dissolve because the precipitate fragments are smaller than the critical size, thus thermodynamically unstable. However, extruded AZ
type magnesium alloys do not give an impression of cyclic softening [1,3,7]. Furthermore, as the c phase precipitates rstly in the
form of incoherent precipitates [33,34], AZ type magnesium alloys
do not cyclically soften by the dissolution of precipitates and for
AZ91-T6 [35] only cyclic hardening has been observed.
Also, there was a clear increase in cyclic yield strength
(101 MPa) versus static yield strength (55 MPa) for our experimental material.
The shapes of hysteresis curves were also studied for the purpose of explaining deformation mechanisms. The deformation
modes of magnesium and its alloys are dislocation slip in the basal,
prismatic, pyramidal I and pyramidal II slip systems and also
 0gh1 0 1
 1i
 and compressive
mechanical tensile twinning f1 1 2
 1h1
 0 12i.

twinning 1 0 1
Mechanical twinning occurs, when the

260

M. Horynov et al. / Materials and Design 45 (2013) 253264

Fig. 13. Inuence of stress amplitude on characteristics of fracture surface of AZ31 magnesium alloy. Initiation sites are marked by arrows, dashed line indicates the
transition between propagation area and area of nal fracture.

c-axis is parallel to the tensile axis or perpendicular to the compression axis [3640].
As only pyramidal slip, which is in magnesium alloys activated
at about 200 C [38,41,42], can fulll the Von Misses criterion without the activation of other deformation modes, various primary
and secondary slip and twinning mechanisms can and have to be
activated at the same time at room temperature [4345]. Evidence
of pyramidal slip at room temperature was found for MgLi single
crystal by Agnew et al. [36] and Ando and Tonda [46]. As single
crystal was strained along the direction, the Schmid factors for basal and pyramidal II slip were nearly identical and pyramidal slip
occurred at room temperature. For polycrystalline magnesium
and its alloys, combinations of other deformation modes were observed at room temperature. Gharghouri et al. [37] observed basal
slip and tensile and compressive twining for Mg7.7Al. Basal slip
with tensile twinning was reported for AM60 [42], AZ31 [47] and
also for pure Mg [40]. Basal slip and tensile twining were considered as major deformation modes for extruded AZ31 magnesium
alloy [39]. Only in the case that the c-axis was nearly perpendicular
to the loading direction, the Schmid factors of the three basal slip
systems is so small, that prismatic and pyramidal I slip system
could be additionally activated. Basal slip and twining can therefore be considered as the most important deformation modes at
room temperature for polycrystalline magnesium alloys with randomly oriented grains.
Symmetric hysteresis curves with respect to tension and compression are typical for deformation predominantly accomplished
by means of dislocation slip [48]. On the other hand, asymmetric
and sigmoidal-shaped curves are the result of mechanical twining
[49].
At all stress amplitudes, except for the rst cycles at higher
stress amplitudes, hysteresis loops were symmetric, which suggests slip-dominated plastic deformation. Hasegawa et al. [3] reported asymmetric hysteresis curves until half-life using the
stress control mode and asymmetry until the end of the test for
the strain control mode. Since stress control mode tests are not frequently used, the shape of hysteresis curves were further discussed
with strain control mode fatigue test results.

Symmetric hysteresis curves were reported for die cast [50,51]


and thixomolded [32] magnesium alloys (AZ91, AE42) subjected to
the strain control mode fatigue test while extruded [1,7,51,52]
magnesium alloys (AZ31, AZ80, AM30) exhibited asymmetric hysteresis curves. As the mechanical twining depends on crystallographic orientation, strong texture of extruded alloys leads to
tensile-compression asymmetry. Cast alloys have no preferentially
oriented grains and hysteresis curves are therefore symmetric.
For the rst cycles at higher stress amplitudes, the hysteresis
curves were observed to have a sigmoidal shape, which is in accordance the with ndings of Matsuzuki and Horibe [7] who reported
concave loop regions resulting from twining as the predominant
effect at higher strain amplitudes.
Gradual cyclic hardening is also evident from the decrease in
plastic strain with increasing number of cycles which corresponds
to a decrease in the area of hysteresis loop W. On the other hand,
the loop shape parameter VH is practically constant for different
stress amplitudes. This can be explained by the uniform deformation of the experimental material without signicantly localized
plastic deformation.
The value of parameter n0 of the experimental material was
higher than values reported for extruded AZ31 alloy [1]. This is
caused by the greater plastic deformation capacity of cast material.
For wrought materials, a portion of the plastic deformation capacity is used during fabrication.
The low-cycle fatigue parameters e0f , c, r0f and b were obtained
by means of the regression function. Parameters b and c were in
agreement with values reported for extruded AZ31 alloy [3] and
slightly higher than that of die-cast A91HP [35,53]. Parameter e0f
of our experimental material is higher than that of extruded
AZ31 alloy [3] and die-cast A91HP [35,53] due to the higher plastic
response.
According to the regression analysis of the whole fatigue life it
was established that the Stromayer function is the most suitable
for describing the fatigue behavior of the experimental material.
A comparison of parameter b with the Basquin function shows that
the KV function better describes the fatigue behavior in the lowcycle region.

M. Horynov et al. / Materials and Design 45 (2013) 253264

261

Fig. 14. (a) Secondary electron micrograph of fatigue fracture surface with initiation site on cluster of secondary particles, (b) back-scattered electron micrograph of initiation
site, (c) more detailed back-scattered electron micrograph of area with AlMn based particles, and (d) EDX spot analysis of area with AlMn based particles.

Fig. 15. Secondary electron micrographs of fatigue fracture surfaces of sample loaded with stress amplitude of 65 MPa. (a) Initiation site, (b) beginning of propagation area,
(c) striations, and (d) region of nal fracture.

262

M. Horynov et al. / Materials and Design 45 (2013) 253264

Fig. 16. Secondary electron micrographs of fatigue fracture surfaces of sample loaded with stress amplitude of 110 MPa (a) initiation site, (b) beginning of propagation area
with secondary crack, (c) striations, and (d) region of nal fracture.

The values of fatigue limits for 108 cycles were basically the
same when tted to each function. The value of the fatigue limit
(rc = 45 MPa) and the fatigue ratio (rc/Rm = 0.26) were slightly
higher than those reported for squeeze-cast AZ31 magnesium
alloy by Zberov et al. [5]. Comparing to AZ type magnesium
alloys prepared by common methods of casting, the value of
fatigue limit of our experimental material is signicantly higher

than that of as-cast AZ61 magnesium alloy [54] and comparable


with gravity-cast [55] and high-pressure die-cast [56,57] AZ91
magnesium alloy. It should be noted, that increasing aluminum
content in AZ type magnesium alloys results in higher fatigue
strength; Chamos et al. [6] reported, that fatigue limit for
AZ61 magnesium alloy is 23% higher compared to AZ31 magnesium alloys.

Fig. 17. MgAl phase diagram [20].

M. Horynov et al. / Materials and Design 45 (2013) 253264

263

Fig. 18. MgAlZn system computed isothermal section at 335 C [21].

The fracture surfaces of selected specimens were examined. Fatigue cracks initiated from the specimen surface or from the clusters of AlMn based particles. Those clusters were observed along
the longitudinal section of the fatigued specimen and because
there was no evidence of fatigue cracks near the non-surface clusters, we can conclude, that those clusters act as initiation sites only
when placed at specimen surface. This is in agreement with nding
of Patel et al. [32] who reported near surface AlMn particle in
AZ91 magnesium alloy which did not act as an initiation site. Documented fatigue crack initiated at a free surface and changed its
direction when reached AlMn particle.
Initiation sites of examined samples were mostly similar, characterized by cleavage-like facets and in some cases by ratchet
marks. As ratchet marks indicate relatively high stresses, its presence was observed only for specimen tested at 110 MPa.
Ridges parallel to the crack propagation direction which have
been observed at the beginning of the crack propagation areas
are most probably formed by local shears that merge plateaus corresponding to different fracture plains in individual grains [58].
5. Conclusions
Low and high cycle engineering stress controlled fatigue tests
were conducted on AZ31 magnesium alloy fabricated by squeeze
casting. The microstructure, cyclic deformation behavior, and low
and high cycle fatigue life were investigated and compared to
available data on AZ31 magnesium alloys prepared by different

technologies. The conclusions obtained can be summarized as


follows:
(1) The microstructure of squeeze-cast AZ31 alloy exhibits a
high level of microstructural and chemical heterogeneity. It
is formed by solid solution, c and u phase, eutectic and Al8Mn5 particles. The average grain size of the material is about
50 lm.
(2) The material cyclically hardens at all test conditions and
exhibits symmetrical behavior in compression and tension.
Sigmoidal shape of hysteresis curves, which suggests twining as a predominant mechanism of plastic deformation,
appears at rst cycles at higher stress amplitudes.
(3) The Stromayer function is optimal for describing the whole
fatigue life of squeeze-cast AZ31 alloy, while the KV function better describes the fatigue behavior in the low-cycle
region.
(4) Fatigue cracks initiates from specimen surface or from the
surface clusters of AlMn based particles.
(5) Even though the squeeze cast material exhibits a high level
of heterogeneity, no signicant scattering of experimental
points was observed.
(6) The squeeze casting method enables the fabrication of complicatedly shaped parts with mechanical properties worse
than those of wrought materials but signicantly better
than those of material produced by common methods of
casting.

264

M. Horynov et al. / Materials and Design 45 (2013) 253264

Acknowledgments
The work was nanced by the Czech Science Foundation, Project No. 101/09/P576 and was also carried out in the frame of the
NETME centre supported by the European Regional Development
Fund and CEITEC Central European Institute of Technology with
research infrastructure supported by the Project CZ.1.05/1.1.00/
02.0068 nanced from the European Regional Development Fund.
The authors would also like to thank Prof. Tom Kruml for
helpful discussion.
References
[1] Begum S, Chen DL, Xu S, Lou AA. Low cycle fatigue properties of an extruded
AZ31 magnesium alloy. Int J Fatigue 2009;31:72635.
[2] Ishihara S, Nan Z, Goshima T. Effect of microstructure on fatigue behavior of
AZ31 magnesium alloy. Mater Sci Eng A 2007;468470:21422.
[3] Hasegawa S, Tsuchida Y, Yano H, Matsiu M. Evaluation of low cycle fatigue life
in AZ31 magnesium alloy. Int J Fatigue 2007;29:183945.
[4] Sonsino CM, Dietrich K. Fatigue design with cast magnesium alloys under
constant and variable amplitude loading. Int J Fatigue 2006;28:18393.
[5] Zberov Z, Kunz L, Lamark TT, Estrin Y, Janecek M. Fatigue and tensile
behavior of cast, hot-rolled and severely plastically deformed AZ31
magnesium alloy. Metall Mater Trans A 2007;38:193440.
[6] Chamos AN, Pantelakis G, Haidemenopoulos GH, Kamoutsi E. Tensile and
fatigue behaviour of wrought magnesium alloys AZ31 and AZ61. Fatigue
Fracture Eng Mater Struct 2008;31:81221.
[7] Matsuzuki M, Horibe S. Analysis of fatigue damage process in magnesium alloy
AZ31. Mater Sci Eng A 2009;504:16974.
[8] Ghomashchi MR, Vikhrov A. Squeeze casting: an overview. J Mater Process
Technol 2000;101:19.
[9] Britnell DJ, Neailey K. Macrosegregation in thin walled castings produced via
the direct squeeze casting process. J Mater Process Technol 2003;138:30610.
[10] Zhang M, Zhang W, Zhao H, Zhang D, Li Y. Effect of pressure on microstructure
and mechanical properties of AlCu based alloy prepared by squeeze casting.
Trans Nonferr Metal Soc 2007;17:496501.
[11] Youn SW, Kang CG, Seo PK. Thermal uid/solidication analysis of automobile
part by horizontal squeeze casting process and experimental evaluation. J
Mater Process Technol 2004;146:294302.
[12] Chadwick GA, Yue TM. Principles and application of squeeze casting. Met
Mater 1989;5:612.
[13] Maleki A, Niroumand B, Shafyei A. Effect of squeeze casting parameters on
density, macrostructure and hardness of LM13 alloy. Mater Sci Eng A
2006;428:13540.
[14] EN ISO 6892-1:2009. Metallic materials tensile testing Part 1: methods of
testing at room temperature; 2009.
[15] ASTM E466-07. Standard practice for conducting force controlled constant
amplitude axial fatigue test of metallic materials; 2002.
[16] ASTM E606-92. Standard practice for strain-controlled fatigue testing; 2004.
[17] Murray JL, McAlister AJ, Schaefer RJ, Bendersky LA. Stable and metastable
phase equilibria in AlMn system. Metal Trans 1987;18A:38592.
[18] Kohout J, Vechet S. A new function for fatigue curves characterization and its
multiple merits. Int J Fatigue 2001;23:17583.
[19] Laser T, Nrnberg MR, Janz A, Hartig Ch, Letzig D, Schmid-Fetzer R, et al. The
inuence of manganese on the microstructure and mechanical properties of
AZ31 gravity die cast alloys. Acta Mater 2006;54:303341.
[20] Park JP, Kim MG, Yoon US, Kim WJ. Microstructures and mechanical properties
of MgAlZnCa alloys fabricated by high frequency electromagnetic casting
method. J Mater Sci 2009;44:4754.
[21] Okamoto H. AlMg (AluminumMagnesium). J Phase Equilib 1998;19:598.
[22] Liang P, Tarfa T, Robinson JA, Wagner S, Ochin P, Harmelin MG, et al.
Experimental investigation and thermodynamic calculation of the AlMgZn
system. Therm Acta 1998;314:87110.
[23] Merino MC, Pardo A, Arrabal R, Merino S, Casajs P, Mohamedano M. Inuence
of chloride ion concentration and temperature on the corrosion behaviour of
MgAl alloys in salt spray fog. Corros Sci 2010;31:1696704.
[24] Alvarez-Lopez M, Pereda MD, Valle JA, Fernandez-Lorenzo M, Garcia-Alonzo
MC, Ruano OA, et al. Corrosion behaviour of AZ31magnesium alloys with
different grain size in simulated biological uids. Acta Biomater
2010;6:176371.
[25] Chamos AS, Pantelakis G, Spiliadis V. Fatigue behaviour of bare and precorroded magnesium alloy AZ31. Mater Design 2010;31:41307.
[26] Song G-L, Xu ZQ. The surface, microstructure and corrosion of magnesium
alloy AZ31 sheet. Electrochimicia Acta 2010;55:414861.
[27] Cheng Y, Qin T, Wang H, Zhang Z. Comparison of corrosion behaviors of AZ31,
AZ61, AM60 and ZK magnesium alloys. Trans Nonferr Metal Soc
2009;19:51724.

[28] Shahri Z, Zarei-Hanzaki A, Abedi HR, Fatemi-Varzaneh SM. An investigation to


the hot deformation characteristics of AZ31 alloy through continuous cooling
compression testing method. Mater Des 2012;36:4706.
[29] Tian S, Wang L, Sohn KY, Kim KH, Xu Y, Hu Z. Microstructure evolution and
deformation of AZ31 Mg-alloy during creep. Mater Sci Eng A
2006;415:30916.
[30] Zeng R, Zhang J, Huang W, Dietzel W, Kainer KU, Blawert C, et al. Review of
studies on corrosion of magnesium alloys. Trans Nonferr Met Sco China
2006;16:76371.
[31] Lunder O, Nisancioglu K, Hanses RS. Corrosion of die cast magnesium
aluminium alloys. SAE Technical Paper; 1993. p. 930755.
[32] Patel HA, Chen DL, Bhole SD, Sadayappan K. Cyclic deformation and twining in
a semi-solid processed AZ91D magnesium alloy. Mater Sci Eng A
2010;528:20819.
[33] Wang Y, Liu G, Fan Z. Microstructural evolution of rheo-diecast AZ91D
magnesium alloy during heat treatment. Acta Mater 2006;54:68999.
[34] Xiuqing Z, Lihua L, Naiheng M, Haowei W. Effect of aging hardening on in situ
synthesis magnesium matrix composites. Mater Chem Phys 2006;96:915.
[35] Liu Z, Wang Z-G, Wang Y, Chen L-H, Zhao H-J, Klein F. Cyclic deformation
behaviour and fatigue crack propagation in AZ91HP and AM50HP. Mater Sci
Technol 2001;17:2648.
[36] Agnew SR, Yoo MH, Tom CN. Application of texture simulation to
understanding mechanical behavior of Mg and solid solution alloys
containing Li or Y. Acta Mater 2001;49:427789.
[37] Gharghouri MA, Weatherly GC, Embury JD, Root J. Study of the mechanical
properties of Mg-7.7 at.% Al by in-situ neutron diffraction. Philos Mag
1999;79:167195.
[38] Gottstein G, Al Samman T. Texture Development in pure Mg and Mg Alloy
AZ31. Mater Sci Forum 2005;495497:62332.
[39] Yi S-B, Davies CHJ, Brokmeier H-G, Bolmaro RE, Kainer KU, Homeyer J.
Deformation and texture evolution in AZ31 magnesium alloy during uniaxial
loading. Acta Mater 2006;54:54962.
[40] Klimanek P, Ptzsch A. Microstructure evolution under compressive plastic
deformation of magnesium at different temperatures and strain rates. Mater
Sci Eng A 2002;324:14550.
[41] Mthis K, Gubicza J, Nam NH. Microstructure and mechanical behavior of AZ91
Mg alloy processed by equal channel angular pressing. J Alloys Compd 2005;1
2:1949.
[42] Mthis K, Chmelk F, Janecek M, Hadzima B, Trojanov Z, Lukc P. Investigating
deformation processes in AM60 magnesium alloy using the acoustic emission
technique. Acta Mater 2006;54:53616.
[43] Graff S, Brocks W, Steglich D. Yielding of magnesium: form single crystal to
polycrystalline aggregates. Int J Plasticity 2007;23:195778.
[44] Lukc P, Trojanov Z. Hardening and softening in selected magnesium alloys.
Mater Sci Eng A 2007;462:238.
[45] Figueiredo RB, Szraz Z, Trojanov Z, Lukc P, Langdon T. Signicance of
twinning in the anisotropic behavior of magnesium alloy processed by equalchannel angular pressing. Scripta Mater 2010;63:5047.
[46] Ando S, Tonda H. Non-basal slips in magnesium and magnesiumlithium alloy
single crystals. Mater Sci Forum 2000;350351:438.
[47] Jiang J, Godfrey A, Liu W, Liu Q. Microtexture evolution via deformation
twinning and slip during compression of magnesium alloy AZ31. Mater Sci Eng
A 2008;483484:5769.
[48] Suresh S. Fatigue of materials. 2nd ed. Cambridge: Cambridge University Press;
1998.
[49] Brown WD, Jain A, Agnew SR, Clausen B. Twining and detwining during cyclic
deformation of Mg alloy AZ31B. Mater Sci Forum 2007;539543:340713.
[50] Liu Z, Xu YY, Wang ZG, Wang Y, Liu ZY. Low cycle fatigue behavior of AZ91HP
alloy in as high pressure die casting. Acta Metall Sin 2000;13:9616.
[51] Zenner H, Renner P. Cyclic material behaviour of magnesium die castings and
extrusion. Int J Fatigue 2002;24:125560.
[52] Begum S, Chen DL, Xu S, Luo A. Strain-controlled low-cycle fatigue properties
of a newly developed extruded magnesium alloy. Metall Mater Trans A
2008;39:301426.
[53] Eisenmeier G, Holzwarth B, Hlzwarth HW, Mughrabi H. Cyclic deformation
and fatigue behaviour of the magnesium alloy AZ91. Mater Sci Eng A
2001;319321:57882.
[54] Sajuri ZB, Miyshita Y, Hosokai Y, Mutoh Y. Effect of Mn content and texture on
fatigue properties of as-cast and extruded AZ61 magnesium alloys. Int J Mech
Sci 2006;48:198209.
[55] Murugan G, Raghukandan K, Pillai UTS, Pai BC, Mahadevan K. High cyclic
fatigue characteristics of gravity cast AZ91 magnesium alloy subjected to
transverse load. Mater Des 2009;30:263641.
[56] Mayer D, Papakyriacou M, Zettl B, Stanzl-Tschegg SE. Inuence of porosity on
the fatigue limit of die cast magnesium and aluminium alloy. Int J Fatigue
2003;25:24556.
[57] Mayer HR, Lipowsky Hj, Papakyriacou M, Rsch R, Stich A, Stanzl-Tschegg S.
Application of ultrasound for fatigue testing of lightweight alloys. Fatigue
Fracture Eng Mater 2001;22:5919.
[58] Pokluda K, andera P. Micromechanism of fracture and fatigue. 1st
ed. London: Springer; 2010.

You might also like