Materials Science & Engineering A 592 (2014) 189–200

Contents lists available at ScienceDirect

Materials Science & Engineering A

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

Influence of microstructural characteristics on mechanical properties

of ADC12 aluminum alloy
M. Okayasu a,n, K. Ota a, S. Takeuchi a, H. Ohfuji b, T. Shiraishi a
a
Department of Materials Science and Engineering, Ehime University, 3 Bunkyo-cho, Matsuyama, Ehime 790-8577, Japan
b
Geodynamics Research Center, Ehime University, 2-5 Bunkyo-cho, Matsuyama, Ehime 790-8577, Japan

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

Article history: The effects of microstructural characteristics on the mechanical properties of an aluminum alloy (ADC12:
Received 30 August 2013 Al–Si11.3–Cu1.9–Zn0.8–Fe0.8–Mg0.2–Mn0.2) produced by various casting technologies are studied experi-
Received in revised form mentally and numerically. Six different casting processes are employed: gravity casting, cold-chamber
28 October 2013
die-casting, hot-chamber die-casting, squeeze casting, twin-rolled continuous casting and heated-mold
Accepted 30 October 2013
continuous casting. Microstructural characteristics, dislocation density and defect density vary depend-
Available online 7 November 2013
ing on the casting method, owing to differences in solidification rate, casting pressure and injection
Keywords: speed. The material characteristics of the samples affect their mechanical properties. Multiple regression
Aluminum alloy analysis is carried out to find equations to predict tensile strength using five independent factors:
Casting process
secondary dendrite arm spacing, microporosity rate, diameter of eutectic structures, aspect ratio of
Mechanical property
eutectic structures and dislocation density. All these factors influence the tensile properties, although to
Microstructure
Regression analysis different degrees. The estimated values of tensile strength are in good agreement with experimental
results.
& 2013 Elsevier B.V. All rights reserved.

1. Introduction to produce high-quality components, although at a high

economic cost.
In recent years, a number of cast aluminum alloys have The material properties of aluminum alloys produced by
received particular attention because of their high productivity, various casting technologies have been the subject of a number
high strength and low density. Cast aluminum alloys have been of investigations. Aluminum alloys produced by pore-free casting
employed in various engineering applications, especially automo- have good tensile properties because of the presence of an α-Al
tive parts. The number of cast aluminum alloy components has phase with a fine structure [2]. The tear toughness of rheocast and
generally been increasing in Japan, although there was a slight squeeze-cast aluminum alloys was investigated by Kumai and
decrease due to the financial crisis and the great disaster (the 2011 Kobayashi [3], who found that the discontinuous distribution of
Tohoku earthquake and tsunami) in Japan. eutectic structures in squeeze-cast alloys provided greater resis-
More than 70% of cast aluminum alloys are produced by tance to crack growth than did the continuous network-like
pressure-casting processes. Several pressure-casting technologies arrangement of the eutectic region in rheocast alloys. Bai and
are used, including die-casting, squeeze casting, pore-free casting Zhao [4] examined the tensile properties and fracture behavior of
and rheocasting. In engineering production, casting processes are aluminum alloys produced by the addition of partial squeeze to
generally selected on the basis of benefits in terms of factors such slow-shot die-casting. They found that poor tensile properties
as cost, quality and strength. When large numbers of cast compo- resulted from heterogeneity of α-Al cells with fragmentary,
nents are required, die-casting processes are frequently chosen, rosette, angular and globular shapes, while better tensile proper-
because of their high production rate. Hot-chamber die-casting ties were associated with finer dendrites with smaller secondary
can produce high-quality components at a high production rate, dendrite arm spacing (SDAS) and more rounded silicon particles.
but has been rarely been used for the production of aluminum In a study by Ceschini et al. [5], relationships between ultimate
alloy components because of technical difficulties [1]. On the other tensile strength and SDAS were examined with the aim of
hand, squeeze casting, rheocasting and pore-free casting are able predicting the tensile strength of cast A357 aluminum alloy.
Kimura et al. [6] investigated the effect of abnormal structure
(defects) on the reliability of squeeze castings, with microstruc-
tural defects being intentionally created by changing the shot-time
n
Corresponding author. Tel./fax: þ 81 89 927 9811. lag. They found that fracture elongation of cast aluminum alloys
E-mail address: [email protected] (M. Okayasu). decreased dramatically with increasing number of defects.

0921-5093/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.msea.2013.10.098
190 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200

Several continuous casting technologies have been proposed components vary depending on the casting technology employed
with the aim of improving the mechanical properties of cast and that this variation is related to the presence of cast defects and
aluminum alloys, including twin-rolled continuous casting (TRC) the microstructural characteristics of the alloys.
[7] and heated-mold continuous casting (HMC) [8]. It has been Although several papers have discussed the mechanical proper-
reported that both TRC and HMC samples exhibited better tensile ties of cast aluminum alloys, there is an apparent lack of suitable
properties and higher fatigue strength than conventional die-cast experimental data. These properties have been explained in terms
and gravity-cast samples [8]. This has been attributed to the of material characteristics (defects and microstructure), with
presence of a finely structured α-Al phase and a low defect density either SDAS [5] or defect density [9,11] being taken as the
[9]. For the TRC samples, typical stress–elongation curves for the significant factor. However, we believe that there are more
rolling and transverse directions were also investigated, with a important factors determining the mechanical properties of these
high yield strength being found in the rolling direction. For the alloys, including, among others, eutectic structure, dislocation
HMC samples, a high elongation (about 20%) was obtained for density and internal stress, as well as interactions among these
Al–Si1.0–Cu0.1–Mg3.0–Mn0.5 (ADC6), because of its uniform crystal factors. Therefore, in the work reported here, an attempt was
orientation [10]. made to examine the influences of a number of material char-
Defects are among the factors that are significant in determin- acteristics, namely, microstructural characteristics, dislocation
ing the mechanical properties of cast alloys [9]. Several types of density and cast defects, on the mechanical properties of cast
internal defects in cast aluminum alloys have been reported [6,9], aluminum alloy components produced by various casting
including microporosity, shrinkage porosity and abnormal struc- processes.
tures (a coarse α-Al phase). These defects appear in the die cavity
during the casting process as a result of high-speed flow and high
pressure. Lee [11] has investigated the effect of microporosity on 2. Experimental procedures
the tensile properties of A356 aluminum alloy with a constitutive
prediction that takes into account the strain-rate sensitivity and 2.1. Materials
the strain-hardening exponent. The ultimate tensile strength and
elongation exhibit strong linear and inverse parabolic depen- The material investigated in this study was the cast aluminum
dences, respectively, on the microporosity. alloy ADC12, which is used widely in various automotive parts,
From the above literature survey, it is clear that the tensile such as transmission cases, converter housings and cylinder
strength and failure characteristics of cast aluminum alloy blocks. Its chemical composition, measured by an optical emission

Cast samples
Gravity casting (GC) Cold-chamber die-casting (CD)

100mm 50mm

Hot-chamber die-casting (HD) Squeeze casting (SQ)

10mm 10mm

Twin-rolled continuous casting (TRC) Heated-mold continuous casting (HMC)

Casting direction
Casting direction

50mm 20mm

Fig. 1. Photographs of cast components produced by various casting technologies.

 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200 191

spectrometer, is (in mass %) Al–Si11.3–Cu1.9–Zn0.8–Fe0.8–Mg0.2– The HMC system consists of a melting furnace, a cylindrical
Mn0.2. Test samples were produced by six different casting graphite displacer block for molten metal level control, a graphite
processes: (i) gravity casting (GC), (ii) cold-chamber die-casting mold, a graphite crucible, a cooling device and pinch rolls for
(CD), (iii) hot-chamber die-casting (HD), (iv) squeeze casting (SQ), withdrawal of the cast sample. The Al alloy ingot of about 0.2 kg
(v) twin-rolled continuous casting (TRC) and (vi) heated-mold was melted into the graphite crucible, and then fed into the
continuous casting (HMC). Fig. 1 shows photographs of the cast graphite mold continuously through a runner continuously at a
samples. The cast samples for CD, HD and SQ were actual speed of 110 mm/min via a dummy rod through the heated die.
automotive part components. The TRC and HMC samples were The molten aluminum alloy was solidified directly by a water
formed as rectangular plates (130 mm  7 mm  1300 mm) and droplet at approximately 80 mL/min [8]. For the GC process, the
round rods (5 mm diameter  1000 mm), respectively. The GC melt was simply poured into the die using a ladle.
sample was a metal ingot in the form of a rectangular block
(600 mm  90 mm  40 mm). 2.2. Mechanical tests
The temperature of the molten aluminum alloy for all the cast
samples was set at 933–973 K. The casting processes for GC, CD, Tensile and fatigue tests were conducted at room temperature
HD and SQ used dies made of hot-worked tool steel (e.g., SKD61). using an electro-servo-hydraulic system with 50 kN capacity. The load
The schematic illustrations of those casting methods are indicated and elongation values were measured by a commercial load cell and
in Fig. 2. The high pressure and the higher injection speed casting strain gauge, respectively. The tensile test was conducted with a
process were conducted for CD and HD. The casting speed at the loading speed of 1 mm/min to fracture. The fatigue strength of the
gates for the die-casting processes (CD and HD) was set at 40– cast samples was examined using the S–N approach; that is, in terms
50 m/s and that for SQ was approximately 0.2 m/s. The casting of the relationship between the applied stress and the cycle number to
pressure for HD and SQ was 25 MPa and that for CD was 40 MPa. final failure. Cyclic loading with load control was performed with a
For the TRC process, a pair of copper rollers of diameter sinusoidal waveform at a frequency of 30 Hz and a stress ratio of
400 mm and width 300 mm was used [7], with roll supporting 0.1 up to 107 cycles.
force and roll temperature 30 t and 298 K, respectively. The
melting weight 30 kg was installed in a high frequency induction 2.3. Microstructural analysis
furnace. The casting speed of the TRC was about 0.4 rpm. The
copper rollers were cooled during the casting process to control Microstructural characteristics were investigated using various
the solidification speed of the aluminum alloy. methods, including energy-dispersive X-ray spectroscopy (EDX),

Casting methods
Gravity casting (GC) Cold-chamber die-casting (CD)
Ejector pin Fixed die
Molten ADC12

Metal mold
Air Molten ADC12

Ejector die Sleeve Plunger tip

Hot-chamber die-casting (HD) Squeeze casting (SQ)

Ejector pin Fixed die

Sleeve Plunger tip

Metal die
Molten ADC12
Plunger tip
Pot
Ejector die
Molten ADC12

Twin-rolled continuous casting (TRC) Heated-mold continuous casting (HMC)

Copper roller Displacer block
Molten ADC12 Molten ADC12
Cast plate Cooling device (water)
Nozzle
Tundish Cast rod
Casting
direction Casting
direction

Heater Graphite mold

Lubricant and Pinch roll
cooling spray Graphite crucible

Fig. 2. Schematic illustration for the casting processes.

192 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200

X-ray diffraction (XRD), electron backscatter diffraction (EBSD) and mechanical thinning followed by electrolytic polishing to less than
transmission electron microscopy (TEM). 300 nm thickness at low temperature (about 243 K).
To clarify the microstructural characteristics of our samples,
EDX and XRD analyses were carried out. For EDX, a JSM-6510
scanning electron microscope (SEM) (JEOL Ltd) was used, and the 3. Results
analysis was conducted with an acceleration voltage of 20 kV. For
XRD analysis, a X’Pert Power (PANalytical) with a software X’Pert 3.1. Microstructural characteristics
High Score was employed; and this analysis was executed with Cu
Kα in which incident radiation was used. Fig. 3 shows the microstructures of the cast aluminum alloy
For EBSD, a high-resolution JSM-7000F SEM (JEOL Ltd) with samples (GC, CD, HD, SQ, TRC and HMC). The microstructures in all
HKL Channel 5 software was used, and the analysis was conducted the samples consist basically of an α-Al phase and eutectic
with an acceleration voltage of 15 kV, beam current of 5 nA and structures, although their microstructural characteristics (size
step size of 0.5–20 μm. Samples were sectioned to less than 5 mm and shape) are different. From the XRD and EDX analysis shown
thickness and the sample surfaces for the observation were in Fig. 3(c), several eutectic structures are found between the α-Al
polished to mirror flatness in a vibropolisher using colloidal silica grains, based on Si, Al–Cu (CuAl2), and Al–Fe–Si (Al8SiFe2) [12,13].
for less than 2 h. During analysis, the sample was tilted to an angle The α-Al grain size differs between the samples, see Fig. 3. The
of 701 to the electron beam. SDAS values of the samples are summarized in Table 1. Fine α-Al
The locations and densities of dislocations introduced by the phases with SDAS ¼3.7 μm are seen in the HD sample, and
casting process were examined by TEM using a JEM-100CXII relatively small SDAS values (6.7–9.2 μm) are found in the HMC,
microscope (JEOL Ltd) with an acceleration voltage of 100 kV. SQ, TRC and CD samples. In contrast, large α-Al grains with
The TEM samples were prepared by conventional methods, such as SDAS ¼33.3 μm occur in the GC sample. The different SDAS values

GC CD

Al8 SiFe2
Si
CuAl 2

HD SQ

TRC HMC

30μm

Fig. 3. (a) Optical micrographs, (b) SEM images of GC, CD, HD, SQ, TRC and HMC samples and (c) EDX and XRD analyses for GC and HMC samples.
 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200 193

GC CD
Si

Al8SiFe2
Si
CuAl2 Al8SiFe2

HD SQ

CuAl2
Si

Al8SiFe2 TRC HMC

Al8SiFe2

Si

5μm

Fig. 3. (continued)

are due to different solidification rates (SR); for example, a high SR needle-shaped structures are created. In the GC sample, larger
gives rise to a small SDAS. In a previous work [8], solidification eutectic structures were observed. These variations in the size of
rates of cast aluminum alloys were predicted using an experimen- the eutectic structures show a similar trend to that for SDAS. Fig. 4
tally obtained formula shows the size of the eutectic structure plotted versus the
solidification rate, and it can be seen that there is an almost-
SR ¼ 2  104 SDAS  2:67 :
linear relationship between them, with a high correlation (coeffi-
cient of determination R2 ¼0.85).
The solidification rates for our samples were estimated using this The microstructural characteristics were investigated further
formula, with the results shown in Table 1. Because of the high by EBSD. Fig. 5 depicts the inverse pole figure (IPF) and pole figure
casting speed at the gates in the die-casting processes, the solidifica- maps for the cast samples. The color levels of each pixel in the IPF
tion rates of the samples produced by these processes should be maps are determined according to the deviation of the measured
high, and indeed SR¼612.4 K/s for HD. However, the solidification orientation: see the stereographic projections in the right-hand
rate for CD was only 53.4 K/s, which is less than a tenth of that for column. The thin black solid lines in the IPF maps show misor-
HD. This is assumed to be due to the large volume of the CD sample ientation angles of more than 51, whereas the thin white lines
(see Fig. 1). Somewhat higher cooling rates were found for TRC and show angles of less than 51. The black areas in the IPF maps,
HMC compared with CD, and are attributed to cooling by copper especially noticeable in the CD and HD samples, are associated
roller-press and the water droplets, respectively. For GC, a very low with regions for which clear data were not available because of the
solidification rate of 1.7 K/s was obtained, which is about 360 times presence of complicated lattice structures and defects such as
lower than that for HD. It should be pointed out that SDAS is directly microporosity. As can be seen in Fig. 5, the crystal orientation
attributed to its mechanical properties, e.g., the Hall–Petch relation- characteristics vary between the samples. For the GC and SQ
ship, in which grain boundaries act as barriers to dislocation motion. samples, a relatively uniform lattice orientation can be seen over
The greater SDAS makes the lower mechanical strength. a large area of a few hundred micrometers. As the SDAS values for
It can also be seen from Fig. 3 that the size of the eutectic GC and SQ are 33.3 and 7.0 μm, respectively, areas of uniformly
structures varies between the samples. Fine structures consisting oriented lattice structures could be united across several grains to
of Si and Fe are obvious in the SQ and HD samples, and relatively form a large colony. In contrast, in both the die-casting samples
small structures are seen in the TRC and HMC samples. In contrast, (CD and HD), the crystal orientation depends on the grain size. For
slightly larger eutectic phases occur in the CD sample, where the HMC sample, a uniformly orientated crystal orientation (i.e., a
194 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200

GC
SEM Si Kα : Al : CuAl2
: Si : Al8SiFe2

200

150

Intensity
Cu Kα Fe Kα
100

50

0
30 40 50 60 70 80
20μm 2θ, degree

HMC
SEM Si Kα : Al : CuAl2
: Si : Al8SiFe2

1000
800

Intensity
Cu Kα Fe Kα 600
400
200
0
30 40 50 60 70 80
10μm 2θ, degree

Fig. 3. (continued)

Table 1 sample (ϕ5 mm). From this, it appeared that the crystal orienta-
Secondary dendrite arm spacing (SDAS) and solidification rate for GC, CD, HD, SQ, tion is almost uniformly orientated as shown in Fig. 5.
TRC and HMC samples. On the other hand, in the TRC sample, striped shapes of lattice
orientation are formed side by side in the phases. This is a result of
GC CD HD SQ TRC HMC
slip deformation and/or crystal rotation, arising from the rolling
SDAS, μm 33.3 9.2 3.7 7.0 8.1 6.7 press [7], which can be seen from the associated pole figures (see
Solidification rate, K/s 1.7 53.4 612.4 110.8 75.3 124.6 the dashed arrows).
Fig. 6 shows TEM images of the cast samples (note that the test
samples for TEM were cut from the same samples used for EBSD).
100 Dislocations of cell boundaries are distributed across all the cast
samples, but are of lower density in the GC and HMC samples. The
Size of eutectic structure, μm

dislocation density for the pressure-cast (TRC) sample is higher

GC than that for the die-cast (CD and HD) samples. On the other hand,
R² = 0.85
10 the dislocation density for the SQ sample is intermediate between
those of the GC and die-cast samples. This variation in dislocation
density may be due to the casting conditions: for example, the
CD higher the pressure and the higher the casting speed, the higher
HMC
1 will be the dislocation density, and conversely the low dislocation
TRC HD
densities of the samples produced by GC and HMC are due to the
low pressures and low casting speeds used in these processes.
SQ
On the basis of the TEM micrographs, the dislocation density
0.1 was estimated by measuring the total length L (m) of the disloca-
1 10 100 1000
tion line in a cube of side l (m) and then dividing this by the
Solidification rate, K/s
volume of the cube, l3 (m3), to give the dislocation density DD ¼L/
Fig. 4. Relationship between size of eutectic structure and solidification rate. l3 (m  2). The dislocation density levels thus obtained are shown in
Table 2. The lowest dislocation density is 2.50  1013 m  2, for the
single-crystal-like structure) can be seen, in which the crystal GC sample. The dislocation densities for the die-cast (CD and HD)
orientation is formed with 〈001〉 on the sample surface perpendi- samples are similar, at about 10  1013 m  2, which is close to the
cular to the casting direction; see the associated pole figures. This values found for similar aluminum alloys investigated previously
is caused by the unidirectional solidification in the cast sample. To [14]. The highest dislocation density is that for the TRC sample,
substantiate the crystal orientation characteristic in HMC, this 22.6  1013 m  2. In this case, the higher the dislocation density
examination was further conducted in the entire area of this could make high material hardness, because of the interruption of
 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200 195

EBSD analysis
IPF maps Pole figures

<100> <110> <111>

GC

200μm

CD

10μm

HD

10μm

SQ

200μm

TRC

100μm

200μm
HMC
111

2mm 001 101

Fig. 5. Microstructural characteristics of GC, CD, HD, SQ, TRC and HMC samples examined by EBSD.
196 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200

TEM images
GC CD

HD SQ

TRC HMC

200nm

Fig. 6. TEM images of GC, CD, HD, SQ, TRC and HMC samples, showing dislocation.

Table 2 1.4
Dislocation density for GC, CD, HD, SQ, TRC and HMC samples.
Vickers hardness, GPa

GC CD HD SQ TRC HMC
1.2
13 2
Dislocation density,  10 m 2.5 10.8 9.6 6.8 22.6 3.5

dislocation movement. Influence of the dislocation density on the 1.0

mechanical properties will be discussed in Section 3.2. It should be
noted that, even in the same sample, the dislocation density varies
slightly, depending on the measurement area. 0.8

3.2. Mechanical properties

0.6
GC CD HD SQ TRC HMC
The Vickers hardness (HV) was measured for a number of
samples of cast aluminum alloy produced by each casting method, Fig. 7. Vickers hardness of GC, CD, HD, SQ, TRC and HMC samples.
 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200 197

400
TRC
HMC
400
TRC
Tensile stress, MPa HD
300

UTS (σUTS), MPa

SQ
350
R² = 0.46 HMC

200 CD 300
HD
GC
SQ
GC CD CD
100 250
HD SQ GC
TRC HMC
200
0 0.6 0.8 1 1.2
0 2 4 6 8 10 Vickers hardness, GPa
Elongation, %

Fig. 8. Relationship between tensile stress and elongation for GC, CD, HD, SQ, TRC 300

0.2% proof strength (σ0.2), MPa

and HMC samples.

250 TRC
0.2% proof strength ( σ0.2), MPa
400 400 R² = 0.49
200
UTS (σUTS), MPa

300 300 HD HMC

150 SQ
CD GC
200 200

100
100 100 0.6 0.8 1 1.2
: UTS Vickers hardness, GPa
: 0.2% proof strength
0 0 10
GC CD HD SQ TRC HMC

SQ
Elongation ( εf), %

8 HMC
12

10 6 R² = -0.11
Elongation (εf), %

TRC
8 4

6 CD
2 GC
HD
4
0
0.6 0.8 1 1.2
2
Vickers hardness, GPa
0 Fig. 10. Relationship between tensile properties and Vickers hardness (HV):
GC CD HD SQ TRC HMC
(a) sUTS versus HV, (b) s0.2 versus HV and (c) εf versus HV.
Fig. 9. Tensile properties of GC, CD, HD, SQ, TRC and HMC samples: (a) ultimate
tensile strength and 0.2% proof strength and (b) elongation.
eutectic structures in cast aluminum alloys using a dynamic
ultra-microhardness tester (DUH-211, Shimadzu), finding values
and the results are shown in Fig. 7. The HD, TRC and HMC samples of 4.11 and 6.36 GPa for the hardness of Si-based and Al–Fe-based
have average HV values of about 1.1 GPa, which is more than eutectic structures, respectively [15], which are much higher than
1.3 times higher than the average value for the SQ samples, while the hardness of the α-Al phase. On the other hand, the hardness of
the CD and GC samples have intermediate average HV values of Al–Cu-based structures is as low as that of the α-Al matrix.
0.9 and 1.0 GPa, respectively. The high values of the hardness of Fig. 8 shows the representative tensile stress-versus-elongation
the HD, HMC and TRC samples are due to the presence of the fine curves for the cast samples, and Fig. 9 shows their tensile proper-
α-Al phases as seen in Fig. 3. Although the GC samples have an ties, namely, ultimate tensile strength sUTS, 0.2% proof strength s0.2
SDAS value considerably larger than those of the other samples and elongation εf. The average sUTS values for the TRC and HMC
(see Table 1), the hardness of the GC samples is only slightly samples are the highest, at about 360 MPa, followed by that for the
higher than those of the CD and SQ samples. It is possible that the HD samples at about 300 MPa, which is more than 15% higher than
hardness of cast aluminum alloys produced by the GC process may the averages for the SQ and CD samples. The GC samples have the
be affected by the enhanced growth of hard eutectic structures lowest average sUTS value, at about 220 MPa, which is approxi-
[15]. On the other hand, the low hardness of the SQ samples may mately 40% lower than the HMC value. Although the average
be caused by the spherical α-Al phases and very small eutectic hardness of the GC samples is greater than those of the CD and SQ
structures distributed between the α-Al grains. The effects of samples (see Fig. 7), the average sUTS value is lower for the GC than
microstructure on mechanical properties will be discussed below. for the CD and SQ samples. The reason for this is that the enhanced
One of the present authors has investigated the hardness of growth of hard eutectic structures in GC alloys impairs their
198 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200

ultimate tensile strength as a result of the generation of high stress GC, CD and HMC seem to be scattered. These results are unex-
concentrations. The 0.2% proof strength also differs among sam- pected, since materials with higher strength should have lower
ples produced by the different casting methods, and generally ductility. In this case, the tensile properties of GC, CD and HMC
follows a similar trend to the ultimate tensile strength. The higher alloys may be influenced by their microstructural characteristics,
elongations εf are found for the SQ, HMC and TRC samples, and are which will be discussed below.
more than double of those of the others, which are as low as about Fig. 12 shows the relationship between stress amplitude and
2%. The high elongation level for HMC can be attributed to the fatigue life (S–N curves). Note that the arrows at 107 cycles indicate
uniformly orientated crystal orientation, and that for SQ is due to test specimens that did not fail within 107 cycles, i.e., the
the presence of small spherical α-Al grains and very small eutectic endurance limit. It can be seen that the HMC and TRC samples
structures. It should be pointed out that the tensile properties have the highest fatigue strengths, and the GC, CD and HD samples
examined in the present work are slightly lower than those the lowest, with the SQ samples being intermediate. The S–N
investigated previously [16]. This may be attributed to the differ- curves for the GC, CD and HD samples almost coincide, in spite of
ent specimen shapes. their different tensile properties. To quantify the fatigue strength
The relationship between the results obtained for material clearly, the S–N curves are represented by a power-law depen-
hardness and tensile properties is shown in Fig. 10. Note that the dence on the applied cyclic stress and the number of cycles to final
data points for the hardness and tensile properties are the mean fracture:
values of all the measurement data in Figs. 7 and 9. In Fig. 10
(a) and (b), there is a linear relationship between tensile strength sa ¼ sf N bf ; ð1Þ
(sUTS and s0.2) and HV, with a coefficient of determination R2 of
where sa is the stress amplitude, sf is the fatigue strength
more than 0.4. In contrast, no clear correlation between elongation
coefficient, Nf is the number of cycles to final fracture and b is
and hardness can be detected in Fig. 10(c). Fig. 11 is a plot of
the fatigue exponent. In this case, a high fatigue strength is
ultimate tensile strength versus elongation, and, as in Fig. 10(c),
expected for a high fatigue strength coefficient sf. Values of sf
weak correlation can be seen (R2 ¼0.22), where the data points for
for the cast alloys, obtained by least squares analysis, are shown in
Table 3, from which it can be seen that the coefficients for the
HMC, SQ and TRC samples are much higher than those for the GC,
400
CD and HD samples, which reflects the behavior of their respective
TRC S–N curves. It should be pointed out that despite the fairly high
HMC tensile strength of the HD cast alloy, its fatigue strength is low.
350
UTS (σUTS), MPa

This trend is dissimilar to that for the related hot-chamber die-

casting samples examined previously [1].
HD
300 To analyze the influence of tensile properties on fatigue
strength, three different factors are plotted versus the fatigue
SQ strength coefficient sf in Fig. 13: (a) ultimate tensile strength sUTS,
250 CD R² = 0.22 (b) elongation εf and (c) strain energy εen. The strain energy for the
GC
cast aluminum alloys was measured from the stress–elongation
curves shown in Fig. 8. It can be seen that there is a clear
200 correlation between all the tensile properties and the fatigue
0 2 4 6 8 10 strength coefficients.
Elongation (εf), %
Fig. 11. Relationship between ultimate tensile strength and elongation.
4. Discussion
200
GC [8] To provide further insight into the effects of material properties
Stress amplitude (σa), MPa

CD on mechanical strength, the following additional microstructural

HD
150 HMC SQ characteristics of each cast aluminum sample were investigated:
SQ TRC [7] (i) size of eutectic structure (SE), (ii) aspect ratio of eutectic
HMC [8] structure (AE) and (iii) area fraction of microporosity (the defect
100 HD density, DR). In this case, Si-based and Al–Fe-based eutectic
TRC structures were selected, since both have much greater hardness
than the Al matrix and the Al–Cu-based eutectic structures, as
GC mentioned in Section 3. The reasons for the three factors selected
50
are based upon the following concepts. Due to the high hardness
CD
of the eutectic structures as described in Section 3.2 [15], those
structures could make brittleness. Moreover the defects make a
0 3
10 104 105 106 107 108 change in their mechanical properties. The results are shown in
Table 4. It should be noted first here that the results of Table 4 are
Number of cycles to failure (Nf)
average values of data from more than a hundred measurements.
Fig. 12. S–N curves for GC, CD, HD, SQ, TRC and HMC samples. Moreover, the HD samples have a high defect density, which is

Table 3
Fatigue strength coefficient sf for GC, CD, HD, SQ, TRC and HMC samples.

GC CD HD SQ TRC HMC

Fatigue strength coefficient sf, MPa 113.7 111.2 153.0 268.5 323.8 281.8
 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200 199

400 Table 5
Predictions of ultimate tensile strength using Eq. (2).
TRC

350 R² = 0.62 GC CD HD SQ TRC HMC

UTS (σUTS), MPa
HMC
Experimental UTS, MPa 224 246 305 263 367 353
HD Estimated UTS by Eq. (2), MPa 183 287 307 308 341 298
300
SQ
CD strength were carried out as a first approach. In order to predict
250
the tensile strength accurately, four more different cast aluminum
alloy samples made by gravity and pressure castings with different
GC
200 cooling rates were employed in addition to the above six samples.
0 100 200 300 400 The correlation coefficient R and the coefficient of determination
Fatigue strength coefficient (σf), MPa R2 give the proportion of the variation in the tensile properties
that is explained by the independent variable in the models. The
regression analysis was found to be highly reliable for both the
10 tensile strength, with a high multiple correlation coefficient of
SQ
more than 0.94. The five variables were employed as predictor
8 variables in a subsequent regression analysis, and the following
Elongation ( εf), %

HMC
models were derived for the prediction of tensile strength:
6 R² = 0.73
sUTS ¼ 330:2  3:35 SDAS  4:71 DR 0:1SE  7:66 AE þ2:88DD;
TRC
ð2Þ
4

CD HD
2 From the t-value of each factor obtained by this analysis, the
degree of influence on the tensile properties can be determined. It
GC
0 turns out that all the factors affect the tensile properties, but to
0 100 200 300 400 different degrees. However, almost all factors have a strong effect
Fatigue strength coefficient (σf), MPa on the tensile strength from their high t-values. The coefficient of
the dislocation density in Eq. (2) is a positive number (2.88), which
means that high dislocation density contributes the high tensile
25 strength. On the other hand, the other factors have a negative
Strain energy (εen), MJ/m3

HMC coefficient, thus, leading to the negative strength fact. Table 5 gives
20 R² = 0.91 a comparison of the tensile properties obtained experimentally
R² = 0.91 SQ
and those estimated numerically using Eq. (2). It is clear that the
TRC
15 estimated tensile strength values are relatively in good agreement
with the experimental data. It can be concluded from these results
10 that the tensile properties of the cast ADC12 aluminum alloy can
be predicted satisfactorily using the proposed equation together
5 HD with the five selected factors.
CD
GC
0
0 100 200 300 400 5. Conclusions
Fatigue strength coefficient (σf), MPa
The effects of microstructural characteristics on the mechanical
Fig. 13. Relationship between tensile properties and fatigue strength coefficient: properties of ADC12 aluminum alloy, produced by the GC, CD, HD,
(a) sUTS versus sf, (b) εf versus sf and (c) εen versus sf.
SQ, TRC and HMC processes, have been studied. Five independent
factors were considered: the secondary dendrite arm spacing, the
Table 4 microporosity rate, the diameter of eutectic structures, the aspect
Microstructural characteristics and defect density for GC, CD, HD, SQ, TRC and HMC
ratio of eutectic structures and the dislocation density. On the
samples.
basis of the results obtained, the following conclusions can be
GC CD HD SQ TRC HMC drawn.

Diameter of eutectic structure, μm 15.8 1.5 0.48 0.30 0.91 0.92 1) The microstructure of ADC12 consists mainly of an α-Al phase
Aspect ratio of eutectic structure 5.2 5.3 3.2 2.4 3.3 2.6
Defect density (porosity), % 0.36 0.50 2.91 0.01 0.34 0.03
and eutectic structures based on Si, Al–Cu and Al–Fe. The α-Al
grain size and the sizes of the eutectic structures vary between
samples depending on the casting method, and this variation
caused by the molten aluminum alloy splashing, scattering or can be attributed to differences in solidification rate and casting
spraying into the mold cavity during the injection process. conditions among the methods.
Based on the experimental data shown in Tables 1, 2 and 4, a 2) The crystal orientation characteristics also vary among the
statistical analysis was carried out using a mathematical software, samples. A relatively uniform lattice orientation is observed
with multiple regression analysis being conducted to find the over a large area for the GC and SQ samples, and a single
predictive equation for determination of the ultimate tensile crystal-like lattice formation is found for the HMC samples. On
strength using the independent variables SDAS, DR, SE, AE and the other hand, the crystal orientation is randomly orientated
DD. Regression analyses for determination of the ultimate tensile in the CD and HD samples.
200 M. Okayasu et al. / Materials Science & Engineering A 592 (2014) 189–200

3) The dislocation density is lowest for the samples produced support. This work was supported by grants from the Light Metal
using the GC and HMC methods, It is highest for those Educational Foundation Inc. and the Japanese Government (Min-
produced using pressure casting (CD, HD and TRC), owing to istry of Education, Science, Sports and Culture).
the high casting speed and casting pressure used in these
methods. The dislocation density for samples produced using References
the SQ method has an intermediate value.
4) The ultimate tensile strength is highest for the TRC and HMC [1] M. Okayasu, S. Yoshifuji, M. Mizuno, M. Hitomi, H. Yamazaki, Int. J. Cast Met.
samples and lowest for the GC samples, with the CD, HD and SQ Res. 22 (2009) 374–381.
samples having an intermediate strength. The elongation is the [2] K. Kanazawa, K. Shibayama, Trans. Jpn. Soc. Mech. Eng. 63 (1997) 478–486.
[3] S. Kumai, K. Kobayashi, J. Jpn. Inst. Light Met. 56 (2006) 21–27.
greater level for the SQ, HMC and TRC samples, and is more [4] Y. Bai, H. Zhao, Mater. Des. 31 (2010) 4237–4243.
than twice that for the other samples. The fatigue properties [5] L. Ceschini, A. Morri, A. Morri, A. Gamberini, S. Messieri, Mater. Des. 30 (2009)
are correlated with the tensile properties. 4525–4531.
[6] R. Kimura, M. Yoshida, G. Sasaki, J. Pan, H. Fukunaga, J. Mater. Process. Technol.
5) Multiple regression analysis leads to equations that allow 130–131 (2002) 299–303.
determination of the tensile properties. With five independent [7] M. Okayasu, R. Sato, S. Takasu, A. Niikura, T. Shiraishi, Mater. Sci. Eng. A 534
variables (secondary dendrite arm spacing, microporosity rate, (2012) 614–623.
[8] M. Okayasu, Y. Ohkura, S. Takeuchi, S. Takasu, H. Ohfuji, T. Shiraishi, Mater. Sci.
diameter of eutectic structures, aspect ratio of eutectic struc- Eng. A 543 (2012) 185–192.
tures and dislocation density), the ultimate tensile strength can [9] M. Okayasu, N. Nishi, K. Kanazawa, J. Jpn. Foundry Eng. Soc. 70 (1998). (799–
be determined accurately. 785).
[10] M. Okayasu, S. Takeuchi, T. Shiraishi, Corrosion and mechanical properties of
cast aluminum alloys, Int. J. Cast Met. Res. 26 (2013) 319–329.
[11] C.D. Lee, Mater. Sci. Eng. A 464 (2007) 249–254.
[12] H. Xin, Y. Hong, J. Wuhan Univ. Tech. Mater. Sci. Ed. 28 (2013) 202–205.
Acknowledgments [13] C.-L. Chen, R.C. Thomson, Intermetallics 18 (2010) 1750–1757.
[14] Y. Lin, Y. Zhang, B. Xiong, E.J. Lavernia, Mater. Lett. 82 (2012) 233–236.
[15] M. Okayasu, S. Takasu, M. Mizuno, J. Mater. Sci. 47 (2012) 241–250.
The authors would like to express their appreciation to Mr. H. [16] M. Okayasu, K. Ota, S. Takeuchi, T. Shiraishi, Mater. Sci. Forum 765 (2013)
Totsukawa of Hiroshima Aluminum Industry Co. Ltd for technical 241–244.