Author’s Accepted Manuscript

Failure behavior of high pressure die casting

AZ91D magnesium alloy

X. Li, S.M. Xiong, Z. Guo

www.elsevier.com/locate/msea

PII: S0921-5093(16)30762-6
DOI: http://dx.doi.org/10.1016/j.msea.2016.07.009
Reference: MSA33840
To appear in: Materials Science & Engineering A
Received date: 7 January 2016
Revised date: 26 April 2016
Accepted date: 3 July 2016
Cite this article as: X. Li, S.M. Xiong and Z. Guo, Failure behavior of high
pressure die casting AZ91D magnesium alloy, Materials Science & Engineering
A, http://dx.doi.org/10.1016/j.msea.2016.07.009
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
 Failure behavior of high pressure die casting AZ91D magnesium alloy

X. Li a,b, S.M. Xiong a,b, Z. Guo a,b*

a
School of Materials Science and Engineering, Tsinghua University, Beijing 100084, China

b
Key Laboratory for Advanced Materials Processing Technology, Ministry of Education,

Tsinghua University, Beijing 100084, China

*
Corresponding author: zhipeng_guo@mail.tsinghua.edu.cn

Abstract

The failure behavior of high pressure die casting AZ91D magnesium alloy during both tensile

and fatigue tests was studied in situ by using scanning electron microscope. Attention was

focused on the role of microstructure played in crack initiation and propagation. Results

showed that the defects in castings, including gas pore, shrinkage pore and defect band, were

the crack initiation sources. In tensile test, the crack propagated in a combination of

intergranular and transgranular modes, and the specimen fractured by connecting defects at

the section with minimum effective force bearing area. In fatigue test, the crack propagated in

a transgranular mode at specific crystalline planes. When the crack was in contact with the

β-phase, the crack would pass through, and fracture the network β-phase, whereas bypass the

island β-phase by detaching it from the surrounding α-Mg grains. Besides, defects in front of

the crack would act as the secondary crack initiation sources, from which new cracks would

initiate and propagate. With the propagation of the fatigue crack, the actual maximum cyclic

1
stress would increase to the fracture stress of the left cross section and lead to the final

fracture of the specimen.

Keywords: High pressure die casting, AZ91D magnesium alloy, Failure behavior, Crack

1 Introduction

Magnesium alloys are very attractive for industrial applications because of their light

weighting, energy saving, and environment-friendliness [1, 2]. As one of the most popular

casting methods for processing magnesium alloys, the high pressure die casting (HPDC)

process has various advantages including faster prototyping, better casting dimensional

accuracy, and fewer requirements for post-processing.

Because porosity is one of the main defects in die castings, studies have been conducted to

reveal the correlation between porosity and mechanical properties. Weiler et al [3]

investigated the relationship between internal porosity and fracture strength of die-cast

AM60B magnesium alloy by using the X-ray tomography, and concluded that the local area

fraction of porosity was the primary factor in determining the tensile properties of specimens.

The variability in tensile ductility of HPDC AM50 and AE44 magnesium alloy was studied

by Lee et al [4, 5], and the results showed that the ductility was correlated to the area fraction

of porosity measured in the fracture surface, not the average volume fraction of porosity in

the 3D microstructure. Song et al [6, 7] conducted the in situ observation of tensile

deformation of HPDC AM50 alloy to study the changes of micro-voids and β phase during

tensile deformation, and suggested that the fracture tended to occur at larger micro-voids or in

2
the cluster micro-voids area. On the other hand, Weiler and Wood [8] used an analytical

tensile failure model to predict the tensile ductility of die cast magnesium alloy. By

employing the size and location of the porosity in fracture surfaces of die-cast AZ91 and

AE44 specimens, the tensile elongation was determined with an average error of 5.3%. The

effect of porosity in HPDC magnesium alloy on the ductility was studied by Sun et al [9] via

a two-dimensional microstructure-based finite element modeling method, and results showed

that for the regions with lower pore size and lower volume fraction, the ductility generally

decreased as the pore size and pore volume fraction increase, whereas, for the regions with

larger pore size and larger pore volume fraction, other factors such as the mean distance

between the pores had substantial influence on the ductility.

Though porosity was proved to be the key factor in determining the mechanical properties in

many studies, there were still studies showed that the mechanical properties were influenced

by the whole microstructure not only the porosity. Leo Prakash et al [10] studied the effect of

position and section thickness on the tensile properties in HPDC magnesium alloy, and

suggested that the tensile properties were influenced by the average size, area fraction and

clustering tendency of pores and β phase particles as well as average grain size. Mechanism

for fatigue crack growth in HPDC magnesium alloys was studied by El Kadiri et al [11], and

results showed that these fatigue micromechanics were manifested by the concomitant effects

of casting pores, interdendritic Al-rich solid solution layer, β-phase particles, Mn-rich

inclusions, rare earth-rich intermetallics, dendrite cell size, and surface segregation

phenomena. A skin layer in super vacuum die casting which contained 6%-12% more eutectic

phase and was ~14% harder than those of the HPDC material was considered to be

3
responsible for the improved bending fatigue and corrosion properties of a Mg-Al-Mn

alloy[12]. To improve the mechanical properties of die castings, some new developed HPDC

processes or alternative design of runner geometry were investigated [13-21]. Results

suggested that the improved mechanical properties could be attributed to the fine and uniform

microstructure and the elimination of the large pores.

It is clear that the mechanical properties of die castings are influenced by the microstructure,

including porosity, β phase, grain size et al, and current attention has been focused on the

correlation between the primary factor (e.g. porosity) and mechanical properties, and the

methods to improve the mechanical properties. However, very limited studies have been

performed to study the changes of the microstructure during failure process in HPDC

magnesium alloy, and the related roles of typical structures played in determining the

mechanical properties of the components.

In this study, the changes of different microstructure features during the crack initiation and

propagation in HPDC AZ91D magnesium alloy were studied under unidirectional and cyclic

loads, based on which roles of these structures played in determination of the mechanical

properties were discussed.

2 Experimental

During the experiment, a specific casting (see Fig. 1a) was produced by a TOYO BD–350V5

cold chamber die casting machine with AZ91D magnesium alloy. Processing parameters

adopted during HPDC are listed in Table 1. Samples for further analysis were extracted at

location A, as shown in Fig. 1a. To observe the crack initiation and propagation during the

4
tensile and fatigue tests, in situ observation was conducted in the scanning electron

microscope (SEM) chamber on the tensile specimens (Fig. 1b) and axial fatigue specimens

(Fig. 1c). There were many stages with different stress levels and numbers of cycling, at

which the tensile test and fatigue test was paused for the SEM observation of the changes of

different structures. Both the fatigue and tensile tests were conducted under ambient

temperature (20 oC). Tensile tests were carried out with displacement control at a deformation

speed of 0.1 mm/min. The axial fatigue tests were performed with a strain ratio of R = 0.1, a

maximum cyclic stress of Smax = 170MPa, and a constant frequency of f = 10 Hz. Three

tensile specimens (Fig. 1b) and three fatigue specimens (Fig. 1c) were tested. Properties

including ultimate tensile strength (UTS), yield strength (YS), elongation (EI), fatigue life (N)

and fatigue crack growth rate (FCGR) were evaluated. The FCGR was calculated by the

following equation (1).

L2  L1
FCGR  (1)
N 2  N1

Where L1 and L2 are the crack length at the stages of the crack initiation and the end of crack

propagation, respectively. Accordingly, the N1 and N2 are the numbers of cycling during the

fatigue test corresponded to stages of L1 and L2, respectively.

For metallography observation of the microstructure, the specimens were sectioned, mounted,

and polished directly from the in situ specimen, then etched with a diluted acetic acid solution

of 50 ml distilled water, 150 ml anhydrous ethyl alcohol and 1 ml glacial acetic acid to reveal

the microstructure. The microstructure of the in situ test specimens was observed with a

ZEISS scope A1 optical microscope (OM) and a Hitachi S-4500 SEM. The fracture surfaces

of the in situ test specimens were examined using a Hitachi S-4500 SEM to identify the crack
5
initiation sites and fractography. To observe the size and morphology of the different

microstructure features, the diameter (d) and roundness (R) were analyzed by employing a

software named MIAPS (Micro-image Analysis&Process system), which were defined by the

following equations (2) and (3), respectively.

d  4 A (2)

R  P 2 4 A (3)

Where A and P are the actual area and perimeter of the target, respectively.

3 Results and discussion

3.1 Microstructure

Fig. 2 shows the typical microstructure of the HPDC AZ91D magnesium alloy, which was

characterized by externally solidified crystals (ESCs), α-Mg grains, porosity (gas pore and

shrinkage pore), defect band and β-phase (island and network). The quantitative

characterization of these microstructure features is listed in Table 2. The HPDC process was

composed of pouring, slow shot stage, fast shot stage, and intensification pressurization. Due

to the fast heat transfer [22], the solidification started immediately when the melt was poured

into the shot sleeve. Crystals nucleated, grew and subsequently the so-called ESCs were

created during the following slow shot stage. In the fast shot stage, the melt with ESCs was

injected into the die cavity. Some ESCs were broken or remelted, and the rest ESCs

continued to grow with the solidification of melt in the die cavity, forming large (~16.30 μm)

and dendritic (R=~5.43) ESCs (Figs. 2a and 2c). Due to much higher cooling rate in die

cavity, large number of α-Mg grains nucleated in die cavity, and had very little time to grow

6
before the end of solidification, leading to smaller (~5.32 μm) and rounder (R=~3.05) α-Mg

grains (Fig. 2e). Similar to [23], there were two types of β-phase, including island β-phase

and network β-phase (Fig. 2e). The island β-phase was small in size (~1.96 μm) and round in

morphology (R=~3.57), while the network β-phase was large in size (~4.7 μm) and complex

in morphology (R=~10.21).

Because of the fast filling speed and high cooling rate in HPDC process, defects including

gas pores (gas entrapment) and shrinkage pores (solidification contraction) were inevitable

(Figs. 2a, 2b and 2c). On the other hand, the air in gas pores was an efficient heat-insulating

medium, the local solidification rate could be lowered, subsequently the shrinkage formed

and usually connected to the gas pore [24]. Thus, there were always shrinkage pores around

the gas pores (Fig. 2b). Similar to the study conducted by Gourlay et al [25], the defect band

of positive porosity commonly following the surface contour of castings could be observed in

Figs. 2a and 2d. By comparing the gas pores, shrinkage pores and defect band in Fig. 2 and

Table 2, it can be concluded that the shrinkage pores were the most irregular (R=~3.59), and

had the largest size (28.64 μm). As for the gas pores and porosity in defect band, the average

diameter of the gas pores was the largest (13.01 μm), and the average roundness of the

aggregated porosity in defect band (Fig. 2d) was the smallest (R=~1.80).

3.2 In situ observation during tensile deformation

Results of the tensile tests are listed in Table 3, and the stress-strain curves of the three tensile

tests are shown in Fig. 3. Fig. 4 shows the changes of typical gas pore, shrinkage pore, defect

band and β-phase in specimens during tensile deformation. It can be observed that the crack

7
initiated at the gas pore, shrinkage pore, and defect band, and propagated along the grain

boundaries. As shown in Figs. 4c, 4g and 4k, the fracture morphology of the gas pore,

shrinkage pore and defect band was rough. A further observation of the zoom-in area at these

crack initiation sites showed that the local fracture was torn because of the complex

morphology of these structures.

In a finite plate with a center crack, the stress intensity factor K I depends on the plate width

2W, crack length 2a, and applied tensile stress  , and can be calculated by the empirical

formula (4) [26]:

KI  Y  a (4)

whereby

a a a
Y  1  0.256( )  1.152( ) 2  12.2( )3 (5)
W W W

Assuming that the gas pores, shrinkage pores and defect band are the crack sources during

tensile deformation, the diameters of different porosities are the crack length, and the

specimen is simplified as a finite plate with some cracks. The stress intensity factors of the

different porosities were calculated and listed in Table 4. It can be observed that the largest

and average K I of different porosities were very close, thus, the gas pores, shrinkage pores

and defect band all would act as the crack initiation sources to initiate during the tensile

deformation.

Because the gas pore, shrinkage pore and defect band all had complex morphology, there

were many tip regions on the porosities, which were mostly not in a plane. Under the tensile

stress, the tip regions of these defects would cause large stress concentration. If the local

stress exceeded the strength limit of the material around, crack would initiate and propagate
8
in different directions at different planes (Figs. 4b, 4f, 4j and 4c, 4g, 4k), leading to the rough

fracture morphology[27]. On the other hand, the β-phase hardly changed during the tensile

deformation, and the crack would propagate in a combination of intergranular and

transgranular modes along the direction roughly perpendicular to the tensile stress. Thus, the

crack would propagate roughly in a plane (Fig. 4o), leading to much flatter fracture

morphology. A further observation of the zoom-in area at the β-phase showed that there were

dimples on the fracture surface. The elastic modulus of the β-phase and ESCs or α-Mg grains

was much different, and the different strain levels would be induced by the same stress at

grain boundaries[7], leading to the formation of the plastic void between β-phase and ESCs

(or α-Mg grains). The plastic void would get expanded when in contact with the crack,

leading to the formation of dimples.

The crack propagation in specimen during tensile deformation is shown in Fig. 5. In Fig. 5a,

it can be observed that there was some porosity on the crack before fracture, and the crack

propagated roughly along the direction perpendicular to the tensile direction. As shown in Fig.

5b, further investigation revealed that the crack propagated by connecting the porosity in

front. At locations without porosity, the crack propagated in the transgranular mode, leading

to a flat fracture morphology. However, when the crack encountered with the shrinkage pore,

the crack would propagate along the grain boundaries in a intergranular mode, leading to

rough fracture morphologies.

According to the study of the tensile test, it can be concluded that the specimen would

fracture by connecting the porosity at the section with minimum effective force bearing area

during the tensile deformation. The defects in die castings, including gas pore, shrinkage pore,

9
and defect band would act as the crack initiation sources to initiate. During crack propagation,

the intergranular mode and transgranular mode would occur at the locations with and without

porosity, respectively.

3.3 In situ observation during fatigue tests

Fig. 6 shows the typical crack initiation sites of specimens during the fatigue test. Similar to

the tensile test, the crack initiated at the gas pore, shrinkage pore and defect band. In fatigue

test, the stress applied to the specimen was not uniformly distributed in the cross section, and

there would be the stress concentration phenomenon around the defect tip. Besides the

maximum cyclic stress Smax (170MPa) was larger than the yield stress  y (~ 160MPa [21]),

during continuous cycling, plastic strain would accumulate at the defect tip, leading to the

permanent damage with the persistent slip bands in α-Mg grains, and consequently the crack

would initiate.

The crack propagation during fatigue test is shown in Fig. 7. It can be observed from Fig. 7a

that the crack propagated roughly along the direction perpendicular to the applied stress. A

further observation of the zoom-in area of the crack showed that the crack propagated in a

transgranular mode, i.e. through the specific crystalline plane, as shown in Figs. 7b and 7c.

According to [7, 28], the main deformation mechanisms in hexagonal Mg were deformation

twinning and dislocation slip. At room temperature, there were only the primary basal slip of

{1000} , deformation twinning of {1012} and double twinning of {1011}  {1012} in tension

fatigue test. Prismatic slip was activated, and substantial cyclic hardening was also observed

above the fatigue limit, which was considered to be caused by cross-slipping of prismatic

10
dislocations as well as interactions between prismatic dislocations and {1012} twins. As the

cyclic hardening progressed, double twinning of the {1011}  {1012} type occurred.

Localized deformation in the double twins and the associated strain incompatibility would

lead to the crack initiation and propagation. In this fatigue test, the actual stress at the crack

tip was larger than the yield stress of ESCs and α-Mg grains due to the stress concentration.

Under the cyclic stress, the strains induced by the actual stress would accumulate at the crack

tip, which would cause the formation of slip band and twinning at the specific crystalline

plane in ESCs or α-Mg grains, leading to the transgranular propagation of crack.

Fig. 7d shows the changes of β-phase during crack propagation. During crack propagation,

the crack would pass through and fracture the network β-phase, whereas it would bypass the

island β-phase by detaching it from the surrounding α-Mg grains. This was because the

β-phase was hard and brittle, and difficult to be plastically deformed [29]. When the crack

encountered the island β-phase, because the island β-phase was small and round in

morphology, the strains induced by the stress would accumulate at the boudaries between

island β-phase and ESCs or α-Mg grains [7], not in the island β-phase, leading to the

detachment of island β-phase from ESCs or α-Mg grains. On the other hand, when the crack

was in contact with the network β-phase, the crack tip was encircled by network β-phase, the

strains induced by the stress would accumulate in the network β-phase, then the network

β-phase was plastically deformed, leading to brittle fracture of network β-phase at the crack

tip. Afterwards, when the propagation of crack was hindered by the large network β-phase,

the defects in front of the crack would act as the secondary crack initiation source, from

which new cracks would initiate. Detail of this behavior is shown in Figs 7e and 7f. With

11
further cycling, the propagation of the secondary crack would play a dominate role, and the

actual maximum cyclic stress would increase due to the propagation of crack. When the

actual maximum cyclic stress reached the fracture stress of the specimen, the specimen would

fracture instantaneously.

Fig. 8 shows typical fracture morphology of a fatigue specimen. The fatigue fracture

comprised three typical zones, including the crack initiation site (Fig. 8b, I-zone), crack

propagation zone (Fig. 8c, II-zone) and instantaneous fracture zone (Fig. 8d, III-zone).

Results of the fatigue tests are listed in Table 5, which showed that the crack normally

initiated at the gas pore or shrinkage pore, which was similar to [30]. The fatigue life (N) and

fatigue crack growth rate (FCGR) were shown to be inversely correlated to the area fraction

of the I-zone and II-zone (SI+II/ SI+II+III). Fig. 8c shows the striation existed in the crack

propagation zone. In the fatigue test, the crack propagation would be hindered if the local

stress was lower than certain magnitudes (yeild strength of the microstructure at crack tip),

e.g. during the relaxation stage of the fatigue cycle. Whereas, the plastic deformation would

occur at the crack tip and the crack would continue to grow if the local cyclic stress was

maximized. The cyclic hindering and growth of the crack subsequently disturbed the fracture

surface, and led to the formation of striation in the crack propagation zone. In the

instantaneous fracture zone as shown in Fig. 8d, the fracture morphology comprised porosity

and dimples, which was similar to that of the tensile specimen. When the local stress

exceeded certain magnitude (strength limit of the rest cross section), the fracture would occur,

i.e. reaching the last stage of the fatigue failure.

12
3.4 Comparison of the tensile test and fatigue test

Fig. 9 shows the comparison of the crack initiations in tensile test and fatigue test. In tensile

test, the crack initiated at the grain boundaries around porosity in an intergranular mode,

whereas in fatigue test, the crack initiated at the specific crystalline plane around porosity in a

transgranular mode. The primary difference between the tensile test and fatigue test was the

loading mode. During tensile deformation, the stress gradually increased until the final

fracture. The porosity with complex morphology, possibly located at the ESC boundaries,

would cause the stress concentration. With the tensile stress increased, the concentrated stress

would reach or exceed the bonding force between ESCs and other phases, leading to crack

initiation at the grain boundaries around the porosity. However, during the fatigue test, the

maximum stress was lower than the bonding force between ESCs and other phases, and the

crack would not initiate at the grain boundaries. On the other hand, the stress around or/and at

the tip of the defects would be much larger than the yield stress due to the stress

concentration, leading to the plastic deformation of ESCs or α-Mg grains. Under cyclic

plastic deformation, persistent slip bands and twins would form at specific crystalline planes

of both ESCs and α-Mg grains, and subsequently lead to crack initiation.

The fracture morphology of the tensile specimen and fatigue specimen is shown in Fig. 10. It

can be observed that the fracture morphology of the tensile specimen was rough. While in the

fatigue specimen, the fracture morphology in the crack propagation zone was very flat,

whereas it was rough in the instantaneous fracture zone. the difference of the fracture

morphology can be attributed to the different crack propagation modes during the tensile test

13
and fatigue test. In tensile test, the crack would propagate in a combination of intergranular

and transgranular modes, and the tensile specimen would fracture at the cross section with

minimum force bearing area [27]. Because of the non-uniform distribution of the defects in

specimen, the rough fracture morphology was created. In fatigue test, the crack propagated in

a transgranular mode along the direction roughly perpendicular to the stress in the crack

propagation zone, leading to the flat fracture morphology. While in the instantaneous fracture

zone, similar to the tensile test, the crack propagated in a combination of intergranular and

transgranular modes by connecting the defects in the rest cross section with minimum

effective force bearing area, leading to the rough fracture morphology.

4 Conclusions

In this paper, the failure behavior during the tensile test and fatigue test of the HPDC AZ91D

magnesium alloy was investigated. According to the related results, the following conclusions

can be drawn:

(1) During the tensile test and fatigue test, the defects in specimen, including gas pore,

shrinkage pore and defect band, were shown to be the crack initiation sources.

(2) In tensile test, the specimen fractured by connecting the defects at the section with

minimum effective force bearing area. The crack propagated in a combination of

intergranular and transgranular modes. At locations where defects were present, the crack

would propagate along the grain boundaries of ESCs, whereas the crack would propagate

along the direction roughly perpendicular to the tensile stress if no defects were present.

(3) In fatigue test, the crack propagated in a transgranular mode at the specific crystalline

14
planes in crack propagation zone. In early stage, when the crack was in contact with the

β-phase, the crack would pass through, and fracture the network β-phase, whereas bypass the

island β-phase by detaching it from the surrounding α-Mg grains. In the later stage, when the

propagation of the main crack was hindered by large network β-phase, the defects in front of

the main crack would act as the secondary crack initiation sources to initiate. With the

propagation of the fatigue crack, the actual maximum cyclic stress would increase to the

fracture stress of the left cross section, and lead to the final fracture of the specimen.

Acknowledgements

The authors would like to thank the National Natural Science Foundation of China (No.

51275269), the Tsinghua University Initiative Scientific Research Program (No.

20121087918), and the National Science and Technology Major Project of the Ministry of

Science and Technology of China (No. 2012ZX04012011) for financial support.

References

[1] H. Friedrich, S.J. Schumann. Research for a new age of magnesium in the automotive

industry. J. Mater. Process. Technol. 2001; 117 (3): 276-81.

[2] B.L. Mordike, T. Ebert. Magnesium properties-application-potential. Mater. Sci. Eng. A.

2001; 302(1): 37-45.

[3] J. P. Weiler, J. T. Wood, R. Klassen, E. Maire, R. Berkmortel and G. Wang. Relationship

between internal porosity and fracture strength of die-cast magnesium AM60B alloy.

Mater. Sci. Eng. A. 2005; 395: 315-322.

15
[4] S. G. Lee, G. R. Patel, A. M. Gokhale, A. Sreeranganathan, M. F. Horstemeyer. Variability

in the tensile ductility of high-pressure die-cast AM50 Mg-alloy. Scr. Mater. 2005; 53:

851-856.

[5] S. G. Lee, G. R. Patel, A. M. Gokhale, A. Sreeranganathan, M. F. Horstemeyer.

Quantitative fractographic analysis of variability in the tensile ductility of high-pressure

die-cast AE44 Mg-alloy. Mater. Sci. Eng. A. 2006; 427: 255-262.

[6] J. Song, S. M. Xiong, M. Li, J. Allison. The correlation between microstructure and

mechanical properties of high-pressure die-cast AM50 alloy. J. Alloys Compd. 2009; 477:

863-869.

[7] J. Song, S. M. Xiong, M. Li, J. Allison. In situ observation of tensile deformation of

high-pressure die-cast specimens of AM50 alloy. Mater. Sci. Eng. A. Microstruct.

Process. 2009; 520: 197-201.

[8] J. P. Weiler, J. T. Wood. Modeling the tensile failure of cast magnesium alloys. J. Alloys

Compd. 2012; 537: 133-140.

[9] X. Sun, K. S. Choi, D. S. Li. Predicting the influence of pore characteristics on ductility

of thin-walled high pressure die casting magnesium. Mater. Sci. Eng. A. 2013; 572:

45-55.

[10] D. G. Leo Prakash, D. Regener. Micro-macro interactions and effect of section thickness

of hpdc AZ91 Mg alloy. J. Alloys Compd. 2008; 464: 133-137.

[11] H. El Kadiri, M. F. Horstemeyer, J. B. Jordon, Y. B. Xue. Fatigue crack growth

mechanisms in high-pressure die-cast magnesium alloys. Metall. Mater. Trans. A. 2008;

39: 190-205.

16
[12] W. Wei, A. A. Luo, Z. Tongguang, J. Yan, C. Yang-Tse, I. Hoffmann. Improved bending

fatigue and corrosion properties of a Mg-Al-Mn alloy by super vacuum die casting. Scr.

Mater. 2012; 67: 879-882.

[13] S. Ji, Z. Zhen, Z. Fan. Effects of rheo-die casting process on the microstructure and

mechanical properties of AM50 magnesium alloy. Materials Science and Technology.

2005; 21: 1019-1024.

[14] L. Yuqin, Q. Ma, Z. Fan. Microstructure and mechanical properties of a rheo-diecast

Mg-10Zn-4.5Al alloy. Mater. Trans. 2005; 46: 2221-2228.

[15] S. Tzamtzis, H. Zhang, N. H. Babu, Z. Fan. Microstructural refinement of AZ91D

die-cast alloy by intensive shearing. Mater. Sci. Eng. A. 2010; 527: 2929-2934.

[16] J. Jufu, W. Ying, C. Gang, L. Jun, L. Yuanfa, L. Shoujing. Comparison of mechanical

properties and microstructure of AZ91D alloy motorcycle wheels formed by die casting

and double control forming. Mater. Des. 2012; 40: 541-549.

[17] H. A. Patel, N. Rashidi, D. L. Chen, S. D. Bhole, A. A. Luo; Cyclic deformation

behavior of a super-vacuum die cast magnesium alloy. Mater. Sci. Eng. A. 2012; 546:

72-81.

[18] D. R. Gunasegaram, M. Givord, R. G. O'Donnell, B. R. Finnin, Improvements

engineered in UTS and elongation of aluminum alloy high pressure die castings through

the alteration of runner geometry and plunger velocity. Mater. Sci. Eng. A. 2013; 559:

276-286.

[19] S. Ji, W. Yang, B. Jiang, J. B. Patel, Z. Fan. Weibull statistical analysis of the effect of

melt conditioning on the mechanical properties of AM60 alloy. Mater. Sci. Eng. A. 2013;

17
 566: 119-125.

[20] W. Li, H. Zuqi, W. Shusen, L. Xueqiang. Mechanical properties and fatigue behavior of

vacuum-assist die cast AlMgSiMn alloy. Mater. Sci. Eng., A. 2013; 576: 252-258.

[21] X. Li, S.M. Xiong, Z. Guo. Improved mechanical properties in vacuum-assist

high-pressure diecasting of AZ91D alloy. J. Mater. Process. Technol. 2016; 231:1-7.

[22] R. Helenius, O. Lohne, L. Arnberg, H.I. Laukli. The heat transfer during filling of a

high-pressure die-casting shot sleeve. Mater. Sci. Eng. A. 2005; 413-414: 52-55.

[23] M. Wu, S. Xiong. Microstructure characteristics of the eutectics of die cast AM60B

magnesium alloy. J. Mater. Sci. Technol. 2011; 27: 1150-1156.

[24] S. G. Lee, A. M. Gokhale. Formation of gas induced shrinkage porosity in Mg-alloy

high-pressure die-castings. Scr. Mater. 2006; 55: 387-390.

[25] C. M. Gourlay, H. I. Laukli, A. K. Dahle. Defect band characteristics in Mg-Al and Al-Si

high-pressure die castings. Metall. Mater. Trans. A. 2007; 38A: 1833-1844.

[26] W. Brown，J. Srawley, Plane strain crack toughness testing of high strength metallic

materials, American Society of Testing and Materials, Philadelphia, 1969.

[27] X. Li, S.M. Xiong, Z. Guo. Correlation between Porosity and Fracture Mechanism in

High Pressure Die Casting of AM60B Alloy. J. Mater. Sci. Technol. 2016; 32:54-61.

[28] J. Koike, N. Fujiyama, D. Ando, Y. Sutou. Roles of deformation twinning and

dislocation slip in the fatigue failure mechanism of AZ31 Mg alloys. Scr. Mater. 2010; 63:

747-750.

[29] H. E. Kadiri, X. Ybin, M. F. Horstemeyer, J. B. Jordon, P. T. Wang. Identification and

modeling of fatigue crack growth mechanisms in a die-cast AM50 magnesium alloy. Acta

18
 Mater. 2006; 54: 5061-5076.

[30] L. H. Rettberg, J. B. Jordon, M. F. Horstemeyer, J. W. Jones. Low-Cycle Fatigue

Behavior of Die-Cast Mg Alloys AZ91 and AM60. Metall. Mater. Trans. A. 2012; 43A:

2260-2274.

Figure 1 Configuration of (a) the specific casting including three tensile test bars (diameter at

the center was 6.4mm) and one plate sample (thickness was 2.5mm), (b) tensile specimen, (c)

fatigue specimen. All the specimens were extracted at location A.

Figure 2 Typical microstructure of HPDC AZ91D alloy, showing (a) overview of

microstructure in OM and (b) gas pore, (c) shrinkage pore and ESCs, (d) defect band, (e)

β-phase and α-Mg grains in SEM.

Figure 3 Stress-strain curves of the specimens during in-situ tensile test.

Figure 4 Morphology of the crack and fracture during tensile deformation. From top to

bottom, the four rows were related to gas pore, shrinkage pore, defect band and β-phase,

respectively. The first and second column corresponded to an applied stress level of 0 MPa

and 180 MPa, respectively. The third and fourth column shows the overall and zoom–in

morphology of the fracture surface.

Figure 5 Propagation of crack during tensile deformation, showing (a) before fracture, (b)

zoom-in area of (a), (c) after fracture and (d) zoom-in area of (c).

Figure 6 Crack initiation sites in fatigue tests. (a) and (d) gas pore with shrinkage pore, (b)

and (e) shrinkage pore, (c) and (f) defect band.

Figure 7 Propagation of crack during fatigue test in specimen, showing (a) overall, (b)

19
zoom-in area of (a), (c) zoom-in area of the crack front, (d) β-phase in the crack, (e)

secondary cracks in the crack front and (f) the next step of (e).

Figure 8 Fracture morphology of a fatigue specimen, (a) overall, (b) crack initiation site, (c)

crack propagation zone with striation and (d) instantaneous fracture zone with porosity and

dimple.

Figure 9 Comparison of the crack initiation in (a) and (b) tensile test and (c) and (d) fatigue

test.

Figure 10 Comparison of the fracture morphology in (a) and (b) tensile specimen and (c) and

(d) fatigue specimen.

Table 1 Processing parameters adopted during die casting

Intensification
Melting Initial mold Slow shot Fast shot
casting pressure
temperature (C) temperature (C) speed (m/s) speed (m/s)
(MPa)

680 150 79 0.2 2.5

Table 2 Quantitative characterization of the different microstructure features.

Microstructure Range of Average Range of Average
features diameter (μm) diameter (μm) roundness roundness
α-Mg grain 2.24-9.71 5.32 1.01-14.59 3.05
ESCs 10.01-36.07 16.30 1.46-16.95 5.43
Island β-phase 0.99-4.25 1.96 1.02-9.23 3.57
Network β-phase 2.59-10.95 4.70 3.50-28.82 10.21
Gas pore 7.76-24.99 13.01 1.24-8.31 2.46
Shrinkage pore 5.00-28.64 9.76 1.03-14.21 3.59
20
 Defect band 6.82-25.99 12.18 1.09-4.32 1.80

Table 3 Results of the in situ tensile test.

Sample UTS (MPa) YS (MPa) Elongation (%)

1 195 143 1.6

2 207 151 2.6
3 211 155 3.9

Table 4 Stress intensity factors of the different porosities.

Porosity Gas pore Shrinkage pore Defect band

Range of K I 3.49  -6.28  2.80  -6.73  3.28  -6.41 

Average K I 4.53  3.92  4.38 

Table 5 Results of the in situ fatigue test.

Sample Initiation source SI+II/ SI+II+III (%) N FCGR (μm/C)

1 Shrinkage pore 14.67 10250 0.14

2 Gas pore 14.78 25486 0.06
3 Gas pore 17.53 41120 0.02

Figures 1, 2, 4-10 were created in Microsoft office Visio 2007, Figure 3 was plotted in OriginPro9.0.

21
(a) (b)

A
(c)

100mm

Figure 1

(a) (b) (c)

40μm 40μm

(d) (e)

500μm 200μm 40μm

Figure 2

22
 225
Sample-1
200 Sample-2
175
Sample-3

150
Stress (MPa)

125

100

75

50

25

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Strain (%)

Figure 3

(a) (b) (c) (d)

50μm 50μm 1mm 25μm

(e) (f) (g) (h)

20μm 20μm 250μm 25μm

(i) (j) (k) (l)

20μm 20μm 800μm 50μm

(m) (n) (o) (p)

20μm 20μm 250μm 25μm

Figure 4

23
 (a) (b)

100μm 25μm

(c) (d)

50μm 25μm
Figure 5

(a) (b) (c)

40μm 40μm 20μm

(d) (e) (f)

50μm 50μm 50μm

Figure 6

24
(a) (b)

100μm 50μm

(c) (d)

20μm 25μm
(e) (f)

40μm 40μm
Figure 7

25
(a) (b)

500μm 100μm
(c) (d)

50μm 50μm
Figure 8

(a) (b)

50μm 100μm
(c) (d)

25μm 25μm
Figure 9

26
(a) (b)

500μm 500μm

(c) (d)

500μm 500μm
Figure 10

27