Basal slip dominant fatigue damage behavior in a cast Mg-8Gd-3Y-Zr alloy

Accepted Manuscript Basal slip dominant fatigue damage behavior in a cast Mg-8Gd-3Y-Zr alloy Jipeng Pan, Penghuai Fu, Liming Peng, Bin Hu, Haiming Zhang, Alan A. Luo PII: DOI: Reference:

S0142-1123(18)30342-6 https://doi.org/10.1016/j.ijfatigue.2018.08.001 JIJF 4797

To appear in:

International Journal of Fatigue

Received Date: Revised Date: Accepted Date:

7 May 2018 23 July 2018 1 August 2018

Please cite this article as: Pan, J., Fu, P., Peng, L., Hu, B., Zhang, H., Luo, A.A., Basal slip dominant fatigue damage behavior in a cast Mg-8Gd-3Y-Zr alloy, International Journal of Fatigue (2018), doi: https://doi.org/10.1016/ j.ijfatigue.2018.08.001

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 proof before it is published in its final 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.

Basal slip dominant fatigue damage behavior in a cast Mg-8Gd-3Y-Zr alloy Jipeng Pana,d, Penghuai Fua, Liming Penga*, Bin Hub, Haiming Zhangc, Alan A. Luod a

National Engineering Research Center of Light Alloy Net Forming and State Key Laboratory of Metal

Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, P.R. China b

General Motors China Science Laboratory, Shanghai 201206, P.R. China

c

Institute of Forming Technology & Equipment, Shanghai Jiao Tong University, Shanghai 200240, P.R.

China d

Department of Materials Science and Engineering, Department of Integrated Systems Engineering, The

Ohio State University, Columbus, OH 43210, USA

* Corresponding author. Tel.: +86 18601603818.

E-mail address: [email protected] (Liming Peng).

ABSTRACT

Low fatigue strength has been a major barrier for structural applications of cast magnesium alloys. It is important to understand the fatigue mechanisms and improve the fatigue lives of these alloys. Stress-controlled high-cycle fatigue behaviors of cast Mg-8Gd-3Y-Zr (wt.%) alloy in as-cast, solution treated (T4) and aged (T6) conditions were studied at room temperature, and the fatigue damage morphologies were carefully characterized. During high-cycle fatigue, only basal slips were observed on the surface of fatigue samples under different stress amplitudes, which suggests that basal slip is the dominant 1

fatigue damage mechanism. The T4 and T6 alloys show distinguishing different damage morphologies at grain interiors: there are serried PSMs in the T4 alloy and only sparse PSMs in the T6 alloy. Basal slip deformation can be transferred among basal planes in the “soft” T4 alloy, while only limited on several PSBs in the “hard” T6 alloy. In the T4 alloy, a great number of basal slip planes in a single grain and most of the grains participate in the fatigue deformation, resulting higher fatigue strength (80 MPa, for 107 cycles). For T6 alloy, the fatigue deformation only happens in several PSBs and most of the grains have no plastic deformation, resulting lower fatigue strength (70 MPa). Keywords: Magnesium alloy; high cycle fatigue; fatigue morphology; basal slip; deformation transfer.

Highlights

1. Basal slip is the dominant fatigue damage mechanism during the high cycle fatigue. 2. Distinguishing different damage morphologies were observed in T4 and T6 alloys. 3. There are serried PSMs in the T4 alloy and only sparse PSMs in the T6 alloy. 4. The fatigue damage patterns significantly influences the fatigue strength.

1. Introduction

In recent years, light weighting has become a key strategy for energy saving and environmental protection, which renders magnesium (Mg) alloys quite attractive in automobile, aerospace and electronics industries due to their low density, high strength to weight ratio and excellent recyclability [1, 2]. However, low fatigue strength has been a major barrier for structural applications of cast Mg alloys. In the scenario of structural applications in vehicle and aircraft industries, Mg alloys components, e.g., engine blocks, gear housing and bogie structures, usually bear cyclic loading which results in cyclic deformation damage and

2

fatigue failure. According to the report, at least half of the components failure can be attributed to the fatigue [3]. For design and durability evaluation of structural components, it is important to understand the fatigue mechanisms and improve the fatigue lives of these alloys. For the high cycle fatigue (HCF) at lower loading amplitude, it has been proved that the crack initiation stage generally occupies a major fraction of the fatigue life (about 90%) and is hence life controlling [3, 4]. Hence, the initiation of fatigue cracks can be regarded as the most critical part to understand the HCF performance of Mg alloys. For Mg alloys free of casting defects, persistent slip bands (PSBs) [5, 6], twinning bands [5, 7, 8] and grain boundaries (GBs) [8, 9] were frequently reported as the initiation sites of fatigue cracks, but their microscopic mechanism and especially the overall evolution process of fatigue damage during crack initiation were not analyzed in detail. In other words, we haven’t got a clear idea of how the process of crack initiation influences the fatigue performance of Mg alloys. What is more, the characteristic and process of fatigue damage should always go hand in hand with the variation of microstructure. However, the effect of microstructure on the fatigue damage and fatigue properties were not fully revealed. Most of the earlier arts focused exclusively on the variation of the low-cycle fatigue properties and did not connect the macro-manifests with the intrinsic micro-mechanism [6, 10-14]. We know that the heat treatments such as solid solution and artificial aging in Mg alloys are commonly used means to adjust the microstructure and the thermal condition strongly influences the fatigue properties [12, 15]. Therefore, in this case, we took Mg alloys under different thermal condition as the object of research to further explain the process of fatigue damage during crack initiation in the HCF. The physical origin of the initiation of fatigue damage in ductile crystalline materials intimately relates to persistent slip bands (PSBs) and some kind of slip irreversibility in the bulk, which leads to unreversed slip steps in the form of intrusions and extrusions at the surface [4, 16, 17]. These distinct surface markings, namely persistent slip markings (PSMs), are usually regarded as the sign of fatigue damage. PSMs therefore are believed to be a critical precursor to the nucleation of fatigue cracks [5, 6, 18]. The process was first depicted by J.A. Ewing and J.C.W. Humfrey [19], representing how the fine sharp slip traces grow into 3

small cracks under cyclic loading. Both Z.M. Li et al. [20] and F.H. Wang et al. [6] found that the fatigue crack nucleates at slip bands in Magnesium-Rare Earth (Mg-RE) alloys and they suggested that there is a threshold stress that is necessary for the formation of slip bands and thus governs the resistance to the fatigue crack nucleation of these alloys under HCF. Thereby, they thought the fatigue strength of the studied Mg alloys is determined by the threshold stress, which is generally considered to be affected by the critical resolved shear stress (CRSS) of the slip system. Recent works [5, 21, 22] demonstrated that mechanical twinning, an alternative deformation mechanism, also contributes to the fatigue damage in both wrought Mg alloys and cast ones, especially at high stress/strain amplitudes. Besides, it was proved that grain orientation also significantly affects the fatigue damage mechanism in an as-cast Mg-3Nd-0.2Zn-Zr (NZ30K) alloy [5]. Generally, the increase of the basal Schmid Factor (SF) could enhance the percentage of slipped grains and the slip deformation intensity in a single grain both, and the increase of deformation intensity decreases the percentage of twinned grains [5]. Therefore, to reveal the damage mechanisms in polycrystalline Mg alloys needs further research on the evolution of fatigue damage, and it is crucial for comprehending the effect of microstructures on fatigue properties. For its simple microstructure and relatively good mechanical properties after optimized heat treatments, the cast Mg-8Gd-3Y-0.5Zr (GW83K) alloy is chosen for studying the intrinsic fatigue mechanism of magnesium alloys. The additive RE elements in the Mg-Gd-Y magnesium alloys effectively refine grains and cause significant precipitation hardening. The precipitation sequence in the Mg-Gd-Y alloy was extensively

investigated

and

clarified

as:

super-saturated

solid

solution

(S.S.S.S.)→β’’(D019)→β’(cbco)→β1(fcc)→β(fcc) [23]. At the peak-aged condition (T6), a huge amount of convex lens-shaped metastable β’ precipitates lie on the {11-20} planes and serve as the dominant strengthening phase, which can effectively hinder the {0001} <11-20> basal slip and improve its CRSS under uniaxial loading [24, 25]. The improvement of fatigue strength in the extruded and aged GW103K alloys comparing with the extruded alloys or the extruded and solution treated alloys was simply ascribed to 4

the enhanced CRSS of the basal slip, which is led by the combined effect of solid solution strengthening and precipitation strengthening [12]. The problem is that whether or not the CRSS under uniaxial loading could represent the resistance to the fatigue crack initiation under cyclic loading. To solve this problem, comprehensive studies on the effect of microstructure upon the high-cycle fatigue performance of the casting defects-free GW83K alloy under different heat treatments still need to be done. In the present research, the high-cyclic damage feature and fatigue properties of the cast GW83K Mg alloy under various heat treatment conditions were studied, with particular attention to the initiation and propagation process of fatigue cracks.

2. Material and experiments

2.1. Material fabrication and preparation A magnesium alloy with a nominal composition of Mg-8Gd-3Y-0.5Zr (wt. %, GW83K) was prepared by semi-continuous casting. Compared with conventional casting processes, the semi-continuous casting process is a superior casting technique capable of producing Mg ingots free of defects, such as shrinkages and oxide films. Furthermore, the prepared GW83K alloy shows a fine equiaxed grain microstructure. The fabricated ingot is 70 mm in diameter and 2500 mm in length. An inductively coupled plasma optical emission spectrometer (ICP-AES) analyzer (Thermo, iCAP6300) was employed and the actual chemical composition is Mg-7.7Gd-3.7Y-0.4Zr (wt. %). Samples segmented from the ingot were solution treated (T4) at 500 °C for 8 h, quenched in warm water, and then peak-aged (T6) at 200 °C for 80 h in an oil bath. After the T4 solution treatment, the secondary phase Mg24(Gd, Y)5 dissolves into the a-Mg matrix and leads to a super-saturated solid solution (S.S.S.S.). During the aging treatment, fine dispersed precipitates grow not only inside the grains but also along GBs. The dog-bone shape specimens for tensile and fatigue tests were machined fixing the loading axis parallel to the casting direction. The gauge length of the specimens is 10 mm, the cross-sectional size is 5 mm in length and 4 mm in width, and the radius between the gauge length 5

and the grip end is 15 mm. To get rid of the influence of machining marks on fatigue results, the specimens were milled with emery papers up to grit number of 7000 and then electrochemically polished.

2.2. The uniaxial tensile and fatigue tests Monotonic tensile experiments were conducted at a loading speed of 1 mm/min to obtain the quasi-static mechanical property. For fully reversible stress-controlled fatigue tests, pull-push cyclic loading at a frequency of 20 Hz was applied on the specimens at room temperature using a BOSE Electron force 3550 fatigue test machine. The stress amplitude range is 90-140 MPa for the as-cast alloy, 80-120 MPa for the T4 alloy and 70-180 MPa for the T6 alloys, respectively. At least two samples were tested at each stress level. The evolution of maximum displacement amplitude with respect to the number of loading cycles was recorded. For the convenience of quantifying the deformation ability under cyclic loading, the displacement amplitude was transformed into the calculated strain amplitude, which refers to the approximate strain corresponding to the same static loading stress in the monotonic tensile test curve. The value of the calculated strain was verified using the extensometer at first few cycles for each stress level in the stress-controlled fatigue tests. With the aim to identify the characteristics of the initiation and propagation of fatigue cracks, the material microstructure in the gauge area and the fracture surface of fatigued specimens were examined by using an optical microscope (OM, Olupus XJL-30) and a Zeiss Auriga scanning electron microscope (SEM), respectively. In consideration of the plastic deformation feature at the final stage of fatigue, not only the typical cracking caused by cyclic accumulative damage but also the final tearing contributes to the damage morphologies near the fatigue fracture surface (~400 μm). On this account, regions far away from the fracture surfaces in the gauge area (>1000 μm) were investigated in detail.

6

3. Results

3.1. Microstructure and mechanical properties Fig. 1 shows the OM images of the GW83K alloy after various heat treatments. The microstructure of the as-cast GW83K alloy is composed of α-Mg and Mg24(Gd, Y)5 eutectic compounds dispersed at GBs, as reported in previous study [26], with an average grain size about 80 μm. Fig. 1(b) and (c) show that the secondary phase Mg24(Gd, Y)5 was completely dissolved in the a-Mg matrix after the solution treatment and significant grain coarsening occurred during the solution treatment, and the grain size reaches about 150 μm.

Fig. 1. Microstructure of GW83K alloys under different conditions: (a) As-cast, (b) T4 and (c) T6.

Table 1 presents a comparison of the tensile properties of the GW83K alloy under different conditions. At room temperature, the yield strength (YS), ultimate tensile strength (UTS) and elongation of the as-cast alloy is 143 MPa, 211 MPa and 3.7%, respectively. After T4 treatment, the UTS and elongation of the alloy increases by 17 MPa and 14.4%, respectively, while the YS decreases by 17 MPa. After aging, the UTS and YS was improved up to 343 MPa and 220 MPa, respectively, while the elongation was greatly decreased to 2.7%.

Table 1 Tensile properties of the studied GW83K alloys under different heat treatment conditions.

Properties

Heat treatment As-cast

T4

T6

YS (MPa)

143

126

220

UTS (MPa)

211

228

343

Elongation (%)

3.7

14.4

2.7

7

3.2. High-cycle fatigue properties The relationship between the fatigue life, viz., the number (Nf) of cycles to failure and the applied stress amplitude of the as-cast, T4 and T6 GW83K alloys is plotted as S-N curves and presented in Fig. 2. The fatigue strength (run out at 107 cycles) and the ratios (σf/σb) of the fatigue strength to the UTS are listed in Table 2. At the stress amplitude above 90 MPa, the fatigue property (fatigue life) of the T6 alloy is apparently superior to those of the as-cast and T4 alloys, and the T4 alloy rank last. However, when the applied stress amplitude goes down, the as-cast and T4 alloys gradually show a better fatigue resistance than the T6 alloy. For 107 cycles fatigue life, the as-cast alloy exhibits the highest fatigue strength of 90 MPa, which is much higher than that of the T6 alloy (70 MPa) and the T4 alloy (80 MPa). In addition, the ratio (σf/σb) is about 0.43 for the as-cast alloy, 0.35 for the T4 alloy and 0.21 for the T6 alloy, respectively. The value of the T6 alloy is much lower than that reported for magnesium of 0.25-0.5 in literature [27].

Fig. 2. S-N curves of the GW83K alloy in various thermal conditions under cyclic tensile-compressive loading in air (R=-1).

Table 2 Fatigue strength of the studied GW83K alloys under different heat treatment conditions. Heat Treatment Properties As-cast

T4

T6

σf (MPa)

90

80

70

σf/σb

0.43

0.35

0.21

8

Fig. 3. Variation of maximum displacement amplitude/calculated strain amplitude with the increasing loading cycles for the GW83K alloys under different thermal conditions: (a) As-cast, (b) T4 and (c) T6 (The calculated strain refers to the approximate strain corresponding to the same static loading stress in the monotonic tensile test curve. The value of the calculated strain was verified using the extensometer at first few cycles for each stress level in the stress-controlled fatigue tests).

Fig. 3 shows the relationship of the maximum displacement amplitude/calculated strain amplitude with respect to the number of loading cycles at different applied stress amplitudes. Hereinafter the deformation 9

degree is measured quantitatively by the variety of the strain amplitude. It can be found that the feedback strain amplitude increases and the fatigue life of the alloy decreases with the increase of the imposed stress amplitude. Under the same loading stress amplitude, the strain amplitude of the T6 alloy is much lower than that of the as-cast and T4 alloys. In particular, the T4 alloy deforms most easily. At a representative loading stress of 120 MPa, the feedback strain amplitude at the initial cycles is 0.32%, 0.45% and 0.25% for the as-cast, T4 and T6 alloys respectively. For the as-cast and T4 alloys, the curves exhibit a considerable decline followed by an ascent before the final fracture or run out, and the same trend was also reported in an as-cast NZ30K alloy [5]. In the case of stress-controlled tests, cyclic hardening causes the decrease of strain amplitudes against fatigue cycles, while cyclic softening accounts for the increase of strain amplitudes. These features can be considered as a physical index that characterizes the fatigue damage process, which is closely associated with the deformation mechanism and influenced by the internal microstructure in Ref. [28, 29]. For the GW83K alloys under different heat treatment conditions, the cyclic hardening ability can be described by Δε at a certain stress amplitude and it is calculated as: Δε = ε0 - ε1

(1)

where ε0 is the initial strain amplitude at the beginning of fatigue tests and ε1 is the minimum strain amplitude of the whole curve (ε1 is at the turning point of cyclic hardening and cyclic softening). The statistical results, listed in Table 3, show that the hardening ability increases with the improvement of fatigue load for both the as-cast and T4 alloys. Furthermore, the T4 alloy exhibits a much higher capability of cyclic hardening than the as-cast alloy. Different from them, the T6 alloy shows cyclic stabilization or weak cyclic hardening ability as Δε is always below 0.01%. For a representative loading stress of 120 MPa, Δε is measured as 0.035%, 0.088% and 0.008% for the as-cast, T4 and T6 alloys, respectively. Especially at the elevated loading stress above 140 MPa for the T6 alloy, an initial slight cyclic hardening takes place only in the first few counting cycles and then the dominating stage of cyclic softening. Table 3 The level of cyclic hardening (Δε) at various stress amplitudes (σa) for the GW83K alloys under different heat treatment 10

conditions. Heat treatment

Cyclic stress amplitude (MPa)

As-cast (%)

T4 (%)

T6 (%)

180

--

--

0.007

160

--

--

0.007

140

0.040

--

0.007

120

0.035

0.088

0.008

100

0.020

0.062

0.010

90

0.008

0.042

0.009

80

--

0.023

0.007

70

--

--

0.005

“--”: Untested stress amplitudes in this research.

3.3.Fatigue damage morphologies Fig. 4 shows the EBSD information and fatigue damage morphology of the as-cast GW83K alloy under the stress amplitude of 120 MPa after 14000 loading cycles before failure (about 60% fatigue life, when the alloy began to show obvious cyclic softening, as shown in Fig. 3(a)). The EBSD information characterizes not only the grain orientation but also the surface conditions. Fig. 4(a) exhibits an overall view of the surface damage pattern by the EBSD Kikuchi band contrast map. Fig. 4(b) and (c) show the Inverse Pole Figure (IPF) with black lines indicating the basal plane slip traces on it and the Schmid Factor (SF) map of the basal plane for the considered grains, respectively. Representative grains with typical deformation characteristics after cyclic deformation were magnified and shown in Fig. 4(d)-(g). As shown in Fig. 4, serried parallel striations appear within some grains, namely PSBs, which are caused by the accumulation of the back and forth dislocation slips on crystal planes. A part of extrusion bands are significantly squeezed out of the sample surface like thin films. Furthermore, all there PSBs striations are parallel with their basal plane slip traces (Fig. 4(b)) and no non-basal slip or twinning is observed in the as-cast GW83K alloy under imposed stress amplitudes. It indicates that the cyclic plastic deformation is mainly dominated by the basal slip.

11

Fig. 4. SEM images of the fatigue damage morphology and EBSD result of the as-cast GW83K alloy, σa = 120 MPa, Nf=14000 cycles: (a) EBSD Kikuchi band contrast map of the surface damage morphology, (b) Inverse Pole Figure (IPF) mapping, (c) Schmid Factor mapping for (0001) plane, (d) and (e) The magnified images of selected grains in image (a).

Here, in this study, the fatigue damage morphology refers to the pattern of damage on the sample 12

surface after cyclic deformation. They are substantially permanent PSBs traces on the surface and caused by irreversible shear strain. In the EBSD Kikuchi band contrast map, such as Fig. 4(a), the variation of grey scales indicates the degree of lattice perfection i.e. the degree of fatigue damage in different grains [30]. The more black the color is, the more serious plastic deformation accumulates, and vice-versa, there is less plastic deformation and cyclic damage in the grains with white color. Moreover, grey scale varying among grains indicates the fatigue damage is different: different grains have different fatigue damage and the fatigue damage of different areas in a single grain also varies much. Representative grains are selected and marked by G1-G8. With regard to the Schmid Factor (SF), it is 0.49 for Grain 1 (G1), 0.44 for Grain 2 (G2), 0.41 for Grain 3 (G3), 0.36 for Grain 4 (G4), 0.32 for Grain 5 (G5), 0.17 for Grain 6 (G6), 0.05 for Grain 7 (G7) and 0.01 for Grain 8 (G8), respectively. It is noted that the level of fatigue damage in these grains is basically in proportion to the basal SF values and basal plane cracks are preferentially initiated in high SF grains. The fact indicates that the grains (Grain 1-5) with higher SFs are more susceptible to cyclic deformation, and more distinct PSMs are usually left on the sample surface. However, there are exceptions, such as G4. The basal SF of G4 & G5 is 0.36 and 0.32. Though G4 has a higher value of basal SF, it shows fewer and slighter PSMs compared with G5 (Fig. 4(d)). Besides, G2 and G3, who have similar values of basal SF, also show different configuration of serried PSMs and sparse PSMs, respectively (Fig. 4(f) and (g)). Therefore, consistent with the results in as-cast NZ30K Mg alloy [5], the basal SF of grains is not the exclusive factor which determines the degree and morphology of the fatigue damage. The ambient grains should also play an important role [31, 32].

13

Fig. 5. OM images of surface relief of the as-cast GW83K alloy indicating the evolution of fatigue damage with different number of loading cycles, σa = 120 MPa: (a) Nf=1000 cycles, (b) Nf=4000 cycles, (c) Nf=8000 cycles, (d) Nf=11000 cycles, (e) Nf=14000 cycles and (f) Nf=18000 cycles.

Fig. 5 presents a series of OM images of the selected area (more than 3000 grains were taken into account), which shows the evolution of fatigue damage of the as-cast alloy at the stress amplitude of 120 MPa with the increasing cycles from 1000 to 18000. As seen in Fig. 5(a) at 1000 cycles, grains with higher basal SFs are almost covered with serried PSBs. With the increase of fatigue cycles, the persistent slip markings (PSMs) get thicker and thicker due to the accumulation of dislocation slips. Up to 4000 cycles, as seen in Fig. 5(b), small cracks initiate along some specific PSBs in parts of the grains, but most of the cracks are impeded by GBs and restricted in individual grains. Whereafter, parts of the inner-grain cracks transpierce the GBs and propagate into the neighboring grains at about 8000 cycles, as shown in Fig. 5(c). 14

According to the descending trend of the strain amplitude in Fig. 3(a), cyclic hardening still plays a key role at this stage. Up to 11000 cycles, more inner-grain cracks intersect the whole grains and penetrate into neighboring grains. At this stage, most of the cracks propagate themselves via high-SF grains and only a few cracks in adjacent grains begin to coalesce through lower-SF grains (Fig. 5(d)). From then on, the effective bearing area of the sample cross-section constantly reduced, which resulted in a severer stress concentration at the crack front. This definitely causes the alloy to be subjected to a cyclic softening. Combined with analysis of the microstructure change, it can be proved that the turning point around 11000 cycles of cyclic softening and cyclic hardening (see Fig. 3(a)) indicates the ending of crack initiation stage and the beginning of crack propagation stage. More and more cracks can be seen to coalesce with each other via low-SF grains in the stage of crack propagation. Here we regard the fatigue life at the turning point as the crack initiation life. As shown in Fig. 5(e) for 14000 cycles, the crack front breaks through the GBs and coalesces in a rapid way of cleavage cracking in the low-SF (0.16) Grain 9 along basal plane. When the fatigue cycle increased to 18000 cycles, PSBs-induced cracks in adjacent grains coalesce with each other and turn into longer cracks that passing through several grains, as shown in Fig. 5(f). In the final stage, the cyclic softening rate rapidly accelerated till the final failure at 22370 cycles (Fig. 3(a)). The coalescence paths of cracks among grains and the orientation information of the relevant grains of the as-cast alloy are shown in Fig. 6. As presented in Fig. 6(a) and (c), enlarged SEM images of fatigue damage on the sample surface illustrate different ways of crack growth, including trans-granular and inter-granular fracture. Besides, the IPFs of the corresponding areas are shown in Fig. 6(b) and (d). The grain orientations are indicated by the diagrams of hexagonal lattice and basal plane traces. The path of Crack I is straightly crossing over a whole grain and parallel with the basal plane. This cracking mode is proved to be attributed to cleavage fracture along the PSBs, as shown in Fig. 11. It is remarkable that most of the coalescence cracks tend to propagate in this way (a trans-granular way). For Crack II, the crack paths show a zigzag characteristic, which represents a mixed crack growth mode, composed of both the fracture 15

along basal plane and the fracture linking them (another trans-granular way). The last one of Crack III shows cracking along the GBs (an inter-granular way). The grains in the two sides of the GB usually exist large difference of crystal orientation. Furthermore, it will make this kind of fracture easier to occur when the basal plane of one grain is orientated approximately perpendicularly to GB and the basal plane of the other grain, whose SF is usually low, is orientated approximately parallel to GB.

Fig. 6. Illustration of the crack propagation modes including inter-granular fracture (Grain boundary crack, Crack III) and trans-granular fracture (Cleavage fracture along the PSBs, Crack I and cleavage fracture zigzag steps, Crack II) on the sample surface, and EBSD result of as-cast GW83K alloy, σa = 120 MPa, Nf=14000 cycles.

Fig. 7 shows the OM pictures of the fatigue damage morphology on the specimen surface far away from the fracture of the as-cast GW83K alloy, which were cyclically loaded under the stress amplitude from 90 MPa to 140 MPa. For the case of 90 MPa, there are only few PSMs on the sample surface and the fatigue life exceeds 107 cycles, as shown in Fig. 7(a). With the increase of loading stress, more PSMs appear in grains, get distinct, and more of which turn into small cracks. Under higher stress, the number of crack 16

initiation site increases. Meanwhile, the PSBs-induced cracks in grains prefer to propagate into neighboring grains and coalesce with each other either trans-granularly or inter-granularly. As a result, long but fewer coalesced cracks and short but denser coalesced cracks can be seen under 100 MPa and 140 MPa in Fig. 7(b) and (d), respectively. Under 120 MPa, the crack pattern lays between them. Principally, all these cracks propagate from the surface to the interior of samples, with a direction perpendicular to the loading direction. Magnified images of the PSMs and cracks within several grains are located at the top right corner of the corresponding picture and show severer fatigue damage with the increase of loading stress.

Fig. 7. OM examination of the fatigue damage morphology on the sample surface, which is far away from the fracture surface of the as-cast alloy fatigued till ran-out or final failure: (a) σa = 90 MPa, Nf=107 cycles, (b) σa = 100 MPa, Nf=63297 cycles, (c) σa = 120 MPa, Nf=17389 cycles and (d) σa = 140 MPa, Nf=6349 cycles.

Fig. 8 shows the OM images of the surface damage morphologies in conjunction with the magnified SEM images of the corresponding extrusion topographies of both the T4 GW83K alloy after 11907 loading 17

cycles at the stress amplitude of 120 MPa and the T6 GW83K alloy after 1237 loading cycles at the stress amplitude of 180MPa. Most of fatigue damage feature of the T4 and T6 alloys, such as the basal slip, basal slip cracking and the evolution characteristics of fatigue cracks, are consistent with those of the as-cast alloy. As seen from Fig. 8(a) and (b), similar to the as-cast alloy, the crack growth way of T4 and T6 GW83K alloys also shows the co-occurrence of the trans-granular and inter-granular fractures. However, there is a difference between that of T4 and T6 alloys, i.e., the T4 alloy shows inter-granular crack propagation a lot besides the trans-granular way while the fatigue crack propagation way of the T6 alloy are mainly trans-granular, which reflects the different crack growth resistance of matrix and GBs in T4 and T6 alloys. Moreover, only limited PSBs traces can be seen on the sample surface of T6 alloys and they mostly act as the crack initiation sites or crack growth paths soon afterwards, which accordingly lead to the absence of the cyclic hardening stage, as displayed in Fig. 3(c). By contrast, the great mass of the grains in T4 alloys were filled with dense PSBs, which commonly transverse a whole grain. This kind of phenomenon reveals the occurrence of strong plastic deformation in T4 alloys under cyclic loading and accounts for the notable cyclic hardening shown in Fig. 3(b). Compared with the T6 alloy, only the specific sites of the PSBs in a grain develop into small cracks in T4 alloy. The magnification images of PSMs and extrusions are presented in Fig. 8(c) and (e) for T4 alloys and Fig. 8(d) and (f) for T6 alloys, respectively. For both alloys, the PSBs traces orientate parallel along the basal plane slip trace, which testifies that the plastic deformation of the T4 and T6 alloys is also dominated by the basal slip. Nevertheless, the PSBs traces arrange closely in T4 alloys while quite sparsely in T6 alloys. This can be attributed, on the one hand, to the limited plastic deformation of T6 alloys comparing with T4 alloys under the equivalent cyclic strain and, on the other hand, to the blocking of precipitates on the dislocation movement in aged alloys. Since considerable irreversible slips occur at some specific PSBs, a number of extrusion bands were squeezed out of the sample surfaces, which brings about further surface damage and crack initiation. For T4 alloys, the extrusion bands are thin and continuous in the direction of PSMs on the sample surface. However, for T6 alloys, they are much thicker 18

and mostly fragmented the PSMs. It can be attributed to that

the dislocation slips not only on different

basal planes but also on the same one basal plane are blocked in various degree by the precipitates for the T6 alloy. The characteristics of fatigued microstructure could be a good revelation of the strain amplitude versus loading cycle curves (as shown in Fig. 3(b) and (c)). Severer cyclic hardening corresponds well with the dense PSMs in grains of T4 alloys due to large plastic deformation, while for T6 alloy, cyclic stabilization and the sparse PSBs are well conformed.

19

Fig. 8. OM images of the fatigue damage morphology and the corresponding magnification SEM images of the extrusions topography of T4 and T6 alloys: (a), (c) and (e) for T4 alloy at σa = 120 MPa for 11907 cycles; (b), (d) and (f) for T6 alloy at σa = 180 MPa for 1237 cycles. Black lines indicate the basal plane slip traces characterized by EBSD.

3.4. Fractography SEM micrographs of overall fracture surfaces and the crack initiation areas of the studied as-cast, T4 and T6 GW83K alloys are shown in Fig. 9. As highlighted by dot lines and annotations in Fig. 9(a), (c) and (e), the overall fracture surfaces of fatigued GW83K alloys are characterized by three typical regions: the crack initiation region (Region 1), the crack propagation region (Region 2) and the unstable fracture region (Region 3). As fatigue cracks basically initiate from the sample surfaces, so multiple crack initiation sites (Region 1) are usually found around the edges of fracture. For the sake of convenience, they are marked by rectangles and magnified to show details in Fig. 9(b), (d) and (f). The split facets near the specimen surface indicate where the cracks nucleated and the cleavage-like facets in the vicinity of the crack initiation sites show the initial propagation of cracks among grains. We note that the cleavage facets of both as-cast and T4 alloys show a predominant river pattern. The reason may be that the grains of Region 1 are strongly influenced by the serried PSBs lying in them, which caused the interlayer cracking among different basal planes and formed the river pattern. Because the crack tips are impeded on different slip planes alternately in one grain, this way of cracking is believed to prolong the cracking paths, release the stress concentration and hinder the cracks to propagate into neighboring grains. However, the cleavage facets of the T6 alloy are smooth with no sign of interlayer cracking, which implies the rapid cutting through the grains. Besides, the 20

trans-granular step-like fractures were formed via the coalescence among cleavage planes. For the case of larger grain size, the T4 and T6 alloys show larger proportion of crack initiation region than the as-cast alloy. Based on Fig. 3, the ratio of the crack initiation life with respect to the entire fatigue life were calculated to be 52.5% for the as-cast alloys, 42.6% for the T4 alloys and 32.1% for the T6 alloys, respectively. Compared with T4 alloys, bigger crack initiation area was developed in T6 alloys after relatively fewer loading cycles, which also illustrates that the cracking way of T6 alloy is detrimental to the fatigue strength. Fig. 10 shows the fracture morphology of the crack propagation regions (Region 2) characterized by serrated fatigue striations, in conjunction with some secondary cracks. In this region, the trans-granular cracking plays the key role in all the alloys under different conditions. The crack tips extend through the grains and the serrated fatigue striations are left behind under cyclic deformation. Besides, to a certain degree, the cracks prefer to propagate along some crystallographic planes. So the fracture surface presents the topography of rough facets. From here we see that the persistent dislocation slip has no obvious effect on these regions.

21

Fig. 9. SEM micrographs of the overall fracture surfaces and the areas near the crack initiation site of the studied GW83K alloys under different thermal conditions: (a) and (b) As-cast, σa = 140 MPa, Nf=6349 cycles; (c) and (d) T4, σa = 120 MPa, Nf=11907 cycles; (e) and (f) T6, σa = 180 MPa, Nf=2307 cycles.

Fig. 10. Magnification images of the Region 2 in Fig. 9 of the fatigue crack propagation stage for the (a) As-cast, (b) T4 and (c) T6 alloys.

The characteristics of fatigue damage of the T4 and T6 GW83K alloys are summarized and compared in Table 4. As can be seen in the table, the deformation mechanism and the macroscopic morphology of the fatigue damage of the T4 and T6 alloys are qualitatively the same as the as-cast alloys. However, there is a big difference in the evolution of fatigue microstructure (PSMs, extrusions morphology and the way of crack 22

initiation/propagation) between the two alloys, and it results in their respective fracture topography of the crack initiation region. It is obvious that the microstructural characteristics of fatigue damage relates directly to the varied mechanical performance of the GW83K alloy under different thermal conditions. The process of fatigue damage greatly influences the cyclic curve and the ultimate of fatigue life. The relationship and impact among them will be discussed in detail in the next section.

23

Table 4 The characteristics of fatigue damage of the GW83K alloys under T4 & T6 conditions.

Thermal condition

T4

T6

Deformation mechanism

Mainly basal slip, twinning only plays a part at final fracture stage and can be found in the vicinity of fracture surface

Cyclic curve

Cyclic hardening and cyclic softening

Cyclic stabilization and cyclic softening

PSMs characteristics

Serried PSMs activated in a great number of high-SF grains

Sparse PSMs in some specific grains

Extrusions morphology

Crack initiation way

Crack propagation way

Thin and continuous

The specific sites of PSBs develop into small cracks

A lot inter-granular crack growth besides the trans-granular way

Much thicker and mostly fragmented

Most of PSBs act as crack initiation sites or crack growth paths

Mainly trans-granular crack growth and partly inter-granular way

24

Fatigue damage morphology (Macroscopic) Extension trend of cracks generally perpendicular to loading direction, from external to internal of the sample. With the increase of loading stress, the coalesced crack pattern varies from long and fewer ones to short but denser ones

Fracture topography (Crack initiation region)

Predominant river pattern due to interlayer cracking among basal planes

Smooth cleavage facets due to rapid cutting and coalescence feature

4. Discussion

4.1. Basal slip dominant fatigue damage behavior Magnesium has several families of potential slip systems for the movement of dislocations, including three {0001} <11-20> basal slip systems, three {10-10} <11-20> prismatic slip systems, six {10-11} <11-20> pyramidal slip systems and twelve {10-11} <11-23> pyramidal slip systems. At room temperature, the basal slip systems are most easily activated comparing with the non-basal slip systems because their initial CRSS are far below that of non-basal slip systems (about 100 times) [12, 33]. It is clearly seen from Fig. 4 and Fig. 8 that the basal slip of dislocations is the dominating plastic deformation mode during the cyclic deformation of cast GW83K alloy. Fig. 11 shows the SEM micrographs of surface morphologies close to the crack initiation region, with crystal orientation information from EBSD laid on them. For all the heat treatment conditions, the crystal orientation of the grains indicates that the basal plane of the hexagonal lattices is exactly parallel to the cleavage plane, which provides a direct evidence of the basal plane cracking. What’s more, the PSBs induced damage where cracks always initiate is usually accompanied by some remarkable extrusions, as shown in the magnified image of Fig. 11(d).

25

Fig. 11. SEM micrographs of the surface morphology close to the facture surface with crystal orientation information from EBSD laid on them. (a) As-cast, σa = 140 MPa, Nf=6349 cycles, (b) T4, σa = 120 MPa, Nf=11907 cycles, (c) T6, σa = 180 MPa, Nf=2307 cycles and (d) Magnification image of the selected area in Fig. 11(c).

In addition to dislocation slip, the {10-12} <10-11> tensile twinning is usually regarded as an alternative deformation mode for magnesium alloys, which can reorient the grains to activate the basal slip systems [34]. However, it was reported that there exists a kink point in the strain-life curves for AZ31B [35], ZK60 [36] and GW83K alloys [13], which represents a demarcation line above which twinning process is a significant plastic deformation mechanism in cyclic deformation. When the strain amplitude is below the kink point, the slip of dislocations dominates the cyclic plastic deformation. In the studied high-cycle fatigue (HCF) tests, the loading stress under various heat treatments were quite low and the calculated strain amplitudes are much lower than the kink point of reported alloys (0.75%-0.8%) [13]. Therefore, no obvious twinning could be observed on the sample surface of fatigued GW83K alloy. Twinning just plays a part at the final fracture stage and can be found it in the vicinity of some fracture surface. In the actual application environment, the stress or strain amplitude loaded during fatigue is relatively low. Then, for the Mg alloys, the fatigue deformation is usually provided by none but the basal slip. So we suggest that the main emphasis of fatigue research on the Mg alloys should be placed on the deformation mode of basal slip and their high-cycle properties.

26

Fig. 12. Schematic diagrams of the fatigue damage morphology for the T4 & T6 alloys.

4.2. The difference of fatigue damage mechanism between T4 and T6 heated alloys In general, the process of fatigue damage contains the following sequence of events: Under cyclic loading, dislocations prefer to slip back and forth in some low energy pathways in a specific grain. PSBs appear in the approach to cyclic saturation. As the cyclic strain becomes localized in the PSBs, surface relief develops at the traces of emerging PSBs. Fatigue cracks then initiate at the sites of stress concentration in the surface relief [16]. As described in the above section, the GW83K alloy under different thermal conditions follow the typical sequence of fatigue damage. Heat treatments make a great difference to the microstructure, which can change the strengthening mechanism of the alloys and influence the process of fatigue damage a lot [3]. The heat treatments stated in Section 2.1 imply the T4 and T6 GW83K alloys are mainly strengthened by solid solution atoms and precipitates, respectively. While the as-cast alloy is strengthened by both, so we take the T4 and T6 GW83K alloys as the main object of discussion here. Normally, precipitates perform as much more effective obstacle to pin dislocations than solute atoms in Mg alloys, though both of which can increase the CRSS of the basal slip [25]. However, high CRSS does not imply better resistance to fatigue damage all the time. As shown in Table 4, the most distinguishing feature of fatigue damage between the T4 and T6 alloys is the different patterns of PSMs in a specific grain: serried for the T4 alloy and sparse for the T6 alloy, as shown in the schematic diagrams of Fig. 12(a) and (b), respectively. 27

For the T4 alloy, the activation of basal slip systems is quite easy because of its low CRSS. It helps dislocation slip to transmit among different basal slip planes in the T4 alloy. Due to the increased dislocation density within slip bands and the interaction effect between dislocations and obstacles such as GBs and solute atoms, the “soft” T4 alloy exhibits obvious work hardening behavior in the approach to cyclic saturation. In this process, the notable thing is whenever the dislocation slip in some PSBs is blocked, the basal slip in the neighboring PSBs will be activated. Besides, this deformation pattern also allows the cyclic strain to be transmitted to neighboring grains easily. So in the corresponding performance of microstructure, the PSMs expand transversely to neighboring slip planes till the whole grain is occupied by serried PSBs. As a result, a great number of slip planes in a single grain and most of the grains in the bulk material participate in the plastic deformation during fatigue in the T4 alloy. In this way, the accumulative irreversible shear deformation can be relieved effectively and the localized cyclic strain can be avoided to some extent, which suppresses the stress concentration apparently and thereby postpones the process of crack initiation. Till the cyclic hardening is nearly saturated in some grains, the PSBs continue to develop into new micro-cracks, when the existed cracks evidently propagate into grains nearby, as mentioned in Section 3.4. For the scenario of the T6 alloy, to activate the basal slip is much harder because the dislocation slips are strongly suppressed by precipitates. Therefore, the basal slip only appears along some discrete low energy paths. The cyclic strain will get localized soon in the PSBs and be restricted to the limited slip planes. We can’t see obvious working hardening behavior on the fatigue curve in the “hard” T6 alloy. Accordingly, the microstructure is characterized by sparse PSMs and shows no sign of the PSMs extending to the adjacent slip planes. In this case, significant stress concentration appears in the vicinity of these PSBs, which pushes forward the initiation of fatigue cracks and promotes the micro-cracks to extend into the neighboring grains. At higher stress amplitudes, the fatigue damage process can proceed rapidly corresponding to the strain versus loading cycle curves, as shown in Fig. 3(c).

28

4.3. Effect of heat treatment on the fatigue strength of the GW83K alloy Heat treatments make a great impact on the fatigue strength of the cast GW83K alloy by changing the microstructure, the CRSS of basal slip and the corresponding process of fatigue damage. The strong dependence of the fatigue strength (σf) on the average grain size (d) can be explained by the Hall-Petch formula σf =σ0+Kfd-1/2 [20, 37]. The factor Kf was confirmed to be influenced by the CRSS of the activated deformation modes as well as the orientation relation between the interacting grains. Considering the CRSS of basal slip and the pattern of fatigue damage of the as-cast alloy just fall in between those of the T4 and T6 alloys, grain refinement contributes the most to the fatigue strength of the as-cast GW83K alloy. The solution treatment reduces or even eliminates the fine grain strengthening effect as well as the original precipitation strengthening effect of the as-cast alloy. Comparing the as-cast alloy with the T4 alloy, the average grain size grows from 80 μm to 150 μm and the fatigue strength at 107 cycles declines from 90 MPa to 80 MPa. However, with the same grain size, the contrast between the T4 and T6 alloys is more significant. It has been proved that the CRSS alone of the Mg alloys can’t represent the resistance to the fatigue crack initiation under cyclic loading. As mentioned above, the CRSS changes the transmitting pattern of plastic deformation and fatigue damage mechanism between the T4 and T6 alloys, which decides their final fatigue strength. Just for this reason, though with higher CRSS and tensile strength, the T6 alloy shows no obvious improvement and even descent in fatigue strength under HCF at the lower loading stress amplitude (70 MPa). The fatigue properties of some high strength magnesium alloys reported are listed in Table 5 for comparison. It is obvious that the T6 heat treatment significantly enhances the tensile strength (σb), while less improves the fatigue strength (σf) comparing with the T4 heat treatment. Except for NZ30K [20] alloys (from 68 MPa to 77 MPa) and GZ142K [38] alloys (from 112 MPa to 130 MPa), other alloys including GW103K [12] show no marked improvement of fatigue strength from T4 to T6 state. For AZ91C [1] and present studied GW83K alloy, the values go down even more from 75 MPa to 65 MPa and from 80 MPa to 29

70 MPa, respectively. Consequently, the fatigue ratio (σf/σb) of T6 alloys is always lower than that of T4 alloys. It coincides with the experimental results and implies simply enhancing the CRSS of the basal slip cannot effectively improve the fatigue strength. Table 5 Fatigue property of the studied GW83K alloy, in comparison with the data reported in the literature for various Mg alloys (R=-1, at room temperature). Thermal condition

Alloy (wt.%)

Casting process

GW83K-pres ent study

Semi-continuous casting

T4

GW103K [12]

Gravity permanent mold (PM) casting & extruding

T4

Gravity permanent mold (PM) casting

T4

NZ30K [20]

Semi-continuous casting

T4

GZ142K [38]

Average grain size (μm)

σf (MPa)

Fatigue ratio (σf/σb)

228

80

0.35

343

70

0.21

286

110

0.38

388

110

0.28

174

68

0.39

240

77

0.32

272

112

0.41

367

130

0.35

208

75

0.36

271

65

0.24

150 T6

50 T6

172 T6

20 T6 T4

AZ91C [1]

σb (MPa)

Sand casting

-T6

Test condition

Tension–compre ssion loading Rotating bend loading Rotating bend loading Tension–compre ssion loading Krouse bending loading

“--”: No such information in the literature.

5. Conclusions

Stress-controlled high-cycle fatigue behaviors of cast Mg-8Gd-3Y-Zr (wt.%) alloy under various thermal conditions were studied at room temperature, and the fatigue damage morphologies on the samples' surfaces were carefully characterized. Conclusions can be drawn as follows: (1) The plastic deformation of cast GW83K alloy during high-cycle fatigue is dominated by basal slip, with no twinning observed. It is, thus, suggested that basal slip related mechanisms are critical to the fatigue lives of Mg alloys. (2) The fatigue damage morphologies of the T4 and T6 alloys show distinguishing different features at

30

grain interiors (Fig.12): there are serried PSMs in the T4 alloy and only sparse PSMs in the T6 alloy. The main reason is that basal slip deformation can be transferred among basal planes in the “soft” T4 alloy, while only limited on several PSBs in the “hard” T6 alloy. (3) In the T4 alloy, a great number of basal slip planes in a single grain and most of the grains participate the fatigue deformation. In this way, the plastic deformation in the bulk material is uniform and serious stress concentration is suppressed. Therefore, the process of crack initiation is postponed, resulting higher fatigue strength (80 MPa, for the fatigue life at 107 cycles). (4) For T6 alloy, the fatigue deformation only happens in several PSBs and most of the grains have no plastic deformation. In such case, significant stress concentration is generated in these PSBs, which pushes forward the initiation of fatigue cracks. Since there is high stress concentration in the crack tip and nearly no plastic deformation takes place to release the stress, micro-cracks are more easily to extend into the neighboring grains or combine with other micro-cracks, resulting lower fatigue strength (70 MPa).

Acknowledgment

This work was supported by the National Key Research and Development Program of China (2016YFB0301000 & 2016YFB0701204), Shanghai Rising-Star Program (15QB1402700), Special Fund of Jiangsu Province for the Transformation of Scientific and Technological Achievements (BA2016039), Shanghai Sailing Program (17YF1408900), and Scientist Research Award from Shanghai Jiao Tong University (16X100040025). The first author would also like to express his gratitude to China Scholarship Council for supporting his stay at OSU as a visiting scholar.

31

References [1] Avedesian M, Baker H. Magnesium and magnesium alloys. ASM International 1999; Metals Park, OH, USA. [2] Taub AI, Krajewski PE, Luo AA, Owens JN. The evolution of technology for materials processing over the last 50 years: The automotive example. Jom-Us 2007;59:48-57. [3] Sangid MD. The physics of fatigue crack initiation. Int J Fatigue 2013;57:58-72. [4] Mughrabi H. Microstructural mechanisms of cyclic deformation, fatigue crack initiation and early crack growth. Philos T Roy Soc A 2015;373. [5] Yue HY, Fu PH, Peng LM, Li ZM, Pan JP, Ding WJ. Damage morphology study of high cycle fatigued as-cast Mg-3.0Nd-0.2Zn-Zr (wt.%) alloy. Mater Charact 2016;111:93-105. [6] Wang FH, Dong J, Feng ML, Sun J, Ding WJ, Jiang YY. A study of fatigue damage development in extruded Mg-Gd-Y magnesium alloy. Mat Sci Eng a-Struct 2014;589:209-216. [7] Hazeli K, Askari H, Cuadra J, Streller F, Carpick RW, Zbib HM, Kontsos A. Microstructure-sensitive investigation of magnesium alloy fatigue. Int J Plasticity 2015;68:55-76. [8] Uematsu Y, Kakiuchi T, Tamada K, Kamiya Y. EBSD analysis of fatigue crack initiation behavior in coarse-grained AZ31 magnesium alloy. Int J Fatigue 2016;84:1-8. [9] Wu LY, Yang Z, Xia WJ, Chen ZH, Yang L. The cyclic softening and evolution of microstructures for Mg-10Gd-2.0Y-0.46Zr alloy under low cycle fatigue at 573 K. Mater Design 2012;36:47-53. [10] Mirza FA, Chen DL. Fatigue of rare-earth containing magnesium alloys: a review. Fatigue Fract Eng M 2014;37:831-853. [11] Mirza FA, Chen DL, Li DJ, Zeng XQ. Low cycle fatigue of a rare-earth containing extruded magnesium alloy. Mat Sci Eng a-Struct 2013;575:65-73. [12] Dong J, Liu WC, Song X, Zhang P, Ding WJ, Korsunsky AM. Influence of heat treatment on fatigue behaviour of high-strength Mg-10Gd-3Y alloy. Mat Sci Eng a-Struct 2010;527:6053-6063. [13] Dong S, Wang FH, Wan Q, Dong J, Ding WJ, Jiang YY. Aging effects on cyclic deformation and fatigue of extruded Mg-Gd-Y-Zr alloy. Mat Sci Eng a-Struct 2015;641:1-9. [14] Mirza FA, Chen DL, Li DJ, Zeng XQ. Cyclic Deformation Behavior of a Rare-Earth Containing Extruded Magnesium Alloy: Effect of Heat Treatment. Metall Mater Trans A 2015;46a:1168-1187. [15] Peng LM, Fu PH, Li ZM, Yue HY, Li DY, Wang YX. High cycle fatigue behaviors of low pressure cast Mg-3Nd-0.2Zn-2Zr alloys. Mat Sci Eng a-Struct 2014;611:170-176. [16] Mughrabi H. Cyclic Slip Irreversibilities and the Evolution of Fatigue Damage. Metall Mater Trans A 2009;40a:1257-1279. [17] Polak J, Man J. Fatigue crack initiation - The role of point defects. Int J Fatigue 2014;65:18-27. [18] Mott NF. A theory of the origin of fatigue cracks. Acta Mater 1958;6:195-197. [19] Ewing JA, Humfrey JCW. The fracture of metals under repeated alternations of stress. Phil. Trans. R. Soc. A 1903;200:241-253. [20] Li ZM, Wang QG, Luo AA, Fu PH, Peng LM, Wang YX, Wu GH. High Cycle Fatigue of Cast Mg-3Nd-0.2Zn Magnesium Alloys. Metall Mater Trans A 2013;44a:5202-5215. [21] Wang SD, Xu DK, Wang BJ, Han EH, Dong C. Effect of solution treatment on the fatigue behavior of an as-forged Mg-Zn-Y-Zr alloy. Sci Rep-Uk 2016;6. [22] Park SH, Hong SG, Lee CS. Activation mode dependent {10-12} twinning characteristics in a polycrystalline magnesium alloy. Scripta Mater 2010;62:202-205. [23] He SM, Zeng XQ, Peng LM, Gao X, Nie JF, Ding WJ. Precipitation in a Mg-10Gd-3Y-0.4Zr (wt.%) alloy during isothermal ageing at 250 degrees C. J Alloy Compd 2006;421:309-313. [24] Nie JF, Gao X, Zhu SM. Enhanced age hardening response and creep resistance of Mg-Gd alloys containing Zn. Scripta Mater 2005;53:1049-1053. [25] He SM, Zeng XQ, Peng LM, Gao X, Nie JF, Ding WJ. Microstructure and strengthening mechanism of high strength Mg-10Gd-2Y-0.5Zr alloy. J Alloy Compd 2007;427:316-323. [26] Yang J, Wang QD, Wang H. Effects of heat treatment on microstructure and mechanical properties of Mg-8Gd-3Y-0.5Zr (wt.%) 32

alloy fabricated by semi-continuous casting. China Foundry 2015;12. [27] Ogarevic VV, Stephens RI. Fatigue of magnesium alloy. Annual Review of Materials Science 1990;20:141-177. [28] Hornqvist M, Karlsson B. Influence of heat treatment on the cyclic deformation properties of aluminium alloy AA7030. Mat Sci Eng a-Struct 2008;479:345-355. [29] Yu Q, Zhang JX, Jiang YY. Fatigue damage development in pure polycrystalline magnesium under cyclic tension-compression loading. Mat Sci Eng a-Struct 2011;528:7816-7826. [30] Barnett MR, Nave MD, Bettles CJ. Deformation microstructures and textures of some cold rolled Mg alloys. Mat Sci Eng a-Struct 2004;386:205-211. [31] Cheong KS, Busso EP. Effects of lattice misorientations on strain heterogeneities in FCC polycrystals. J Mech Phys Solids 2006;54:671-689. [32] Cheong KS, Smillie MJ, Knowles DM. Predicting fatigue crack initiation through image-based micromechanical modeling. Acta Mater 2007;55:1757-1768. [33] Gharghouri MA, Weatherly GC, Embury JD, Root J. Study of the mechanical properties of Mg-7.7at.% Al by in-situ neutron diffraction. Philos Mag A 1999;79:1671-1695. [34] Park SH, Hong SG, Lee CS. Role of initial {10-12} twin in the fatigue behavior of rolled Mg-3Al-1Zn alloy. Scripta Mater 2010;62:666-669. [35] Xiong Y, Yu Q, Jiang YY. Multiaxial fatigue of extruded AZ31B magnesium alloy. Mat Sci Eng a-Struct 2012;546:119-128. [36] Xiong Y, Jiang YY. Fatigue of ZK60 magnesium alloy under uniaxial loading. Int J Fatigue 2014;64:74-83. [37] Davis JR. Aluminum and aluminum alloys. ASM International 1993; Metals Park, OH, USA. [38] He ZL, Peng LM, Pe PH, Wang YX, Hu XY, Ding WJ. High cycle fatigue improvement by heat-treatment for semi-continuous casting Mg96.34Gd2.5Zn1Zr0.16 alloy. Mat Sci Eng a-Struct 2014;604:78-85.

33

1. Basal slip is the dominant fatigue damage mechanism during the high cycle fatigue. 2. Distinguishing different damage morphologies were observed in T4 and T6 alloys. 3. There are serried PSMs in the T4 alloy and only sparse PSMs in the T6 alloy. 4. The fatigue damage patterns significantly influences the fatigue strength.

34