Journal of Alloys and Compounds 787 (2019) 519e526
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
Improvement in bending formability of rolled magnesium alloy through precompression and subsequent annealing Jong Un Lee, Sang-Hoon Kim, Ye Jin Kim, Sung Hyuk Park* School of Materials Science and Engineering, Kyungpook National University, Daegu 41566, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 23 November 2018 Received in revised form 26 January 2019 Accepted 6 February 2019 Available online 7 February 2019
A remarkable improvement in the bending formability of a rolled magnesium alloy at room temperature is achieved through application of precompression and subsequent annealing. This combined treatment results in the formation of a new grain structure with two texture components whose basal poles are oriented nearly along the rolling direction and normal direction. The tensile deformation occurring in the outer region of the samples during bending is effectively accommodated by the activated {10e12} twinning and promoted basal slip, which consequently results in a significant improvement in the bending formability of the alloy. © 2019 Elsevier B.V. All rights reserved.
Keywords: Magnesium Bending Formability Twinning Texture
1. Introduction The application range of rolled Mg alloys in the transportation industry has expanded recently because of the low weight and high specific strength of Mg. Although rolled sheets need to have excellent formability for the manufacture of final components, rolled Mg alloys have low formability at room temperature (RT) because of their strong basal texture and hexagonal close-packed (HCP) structure. Various methods for texture weakening aimed at improving the stretch formability of Mg alloy sheets have been employed, e.g., the addition of rare-earth (RE) elements, the application of severe plastic deformation (SPD) processes, and the application of cold rolling and subsequent annealing [1e5]. For example, Song et al. [4] showed that the RT stretch formability of a rolled AZ31 alloy sheet improved by 66% through cold rolling with a 10% reduction and subsequent recrystallization annealing at 400 C. During sheet forming processes, such as stamping and hemming, materials undergo considerable bending deformation. Unlike in stretch forming, where the in-plane tensile stress is dominant, in bending forming, tension and compression occur simultaneously in a material: compressive stress in the inner region touching the plunger and tensile stress in the outer region [6].
* Corresponding author. E-mail address:
[email protected] (S.H. Park). https://doi.org/10.1016/j.jallcom.2019.02.080 0925-8388/© 2019 Elsevier B.V. All rights reserved.
Studies have recently investigated the bending properties of Mg alloy sheets to analyze the microstructural evolutions during bending deformation [6e10]; however, few studies have attempted to improve their low bendability. Kohzu et al. [11] reported that the application of high-temperature annealing at 500 C before and after the final rolling pass effectively weakened the texture of a rolled AZ31 alloy sheet, which resulted in a significant improvement in the RT V-bending formability of the sheet. During the bending of rolled Mg alloys with an intense basal texture, deformation is dominated mainly by slip-assisted mechanisms and {10e12} extension twinning in the outer tension zone and inner compression zone, respectively [6e10]. {10e12} Twinning can accommodate large plastic deformation of up to 6.5% in Mg because of the easy formation and growth of {10e12} twins under favorable stress conditions [12]. Additionally, {10e12} twinning can relax local stresses near crack tips and impede crack propagation in pure Mg [13]. Therefore, in the compression zone of bending samples, {10e12} twinning easily accommodates the imposed strain and suppresses cracking. However, deformation accommodation in the tension zone is extremely limited because the crystal grains of Mg alloy sheets are unfavorably oriented for both {10e12} twinning and basal slip under in-plane tensile deformation. Moreover, in the tension zone, which has a c-axis contraction stress state, {10e11} contraction twins and {10e11}-{10e12} double twinsdwhich can act as crack sources [14,15]dare formed in the late stage of deformation. Consequently, in highly textured rolled Mg alloys, cracks
520
J.U. Lee et al. / Journal of Alloys and Compounds 787 (2019) 519e526
appear only in the tension zone [10], implying that the degree of accommodation of tensile deformation in the outer region during bending governs the bending properties. Herein, to enhance the ability of the tension zone to accommodate deformation, the texture of a rolled Mg alloy is modified using the lattice rotation feature of {10e12} twinning. This approach can significantly improve the bending formability of the material without the addition of expensive RE elements or use of noncommercial SPD processes. 2. Experimental procedure A commercial hot-rolled AZ31 Mg alloy, homogenized at 420 C for 24 h, was used; this material is hereafter denoted as AR material. For precompression, rectangular bars with dimensions of 61 mm (length) 31 mm (width) 21 mm (thickness)dwhich correspond to the rolling, transverse, and normal directions (RD, TD, and ND), respectivelydwere machined from the AR material and then compressed to a plastic strain of 6.0% along the RD to introduce {10e12} twins in the material. The precompressed (PC) bars were further annealed at 250 C for 1 h to obtain a stress-relieved stable microstructure through removal of dislocations and mobile twin boundaries formed by precompression; these subsequently annealed samples are hereafter denoted as PCA material. The AR, PC, and PCA materials were machined to 50 mm (length) 25 mm (width) 3 mm (thickness) rectangular samples for bending tests; the length, width, and thickness of these samples correspond to the RD, TD, and ND, respectively. A three-point bending test was performed on Instron 5985 at a constant crosshead speed of 10 mm/ min at RT in accordance with the ASTM E290 standard [16]. The diameter of the upper and lower rolls of the bending machine was 10 mm each, and the distance between the supports was 20 mm. The bending test was repeated thrice for each material to ensure repeatability and to verify the consistency of the results; however, for the sake of simplicity, a representative curve for each material was displayed. The microstructural characteristics of the AR, PC, and PCA materials were analyzed through electron backscatter diffraction (EBSD) measurements on the RDeTD plane. The microstructures on the RDeND plane at the mid-width of 2-mm-bent
samples and fracture bending samples of the AR and PCA materials were observed by optical microscopy (OM) and EBSD. 3. Results and discussion 3.1. Microstructural characteristics of AR, PC, and PCA materials Fig. 1 shows the microstructural characteristics of the AR material. This material exhibits a twin-free equiaxed grain structure with an average grain size of 35.8 mm (Fig. 1(a)e(c)). The material has a strong basal texture with a maximum intensity of 10.3, where the basal planes of most grains are aligned nearly parallel to the rolling plane (Fig. 1(d)). However, there is no preferred orientation for the (10-10) prismatic plane, indicating that the a-axes of the HCP lattice are randomly oriented in the rolling plane (Fig. 1(e)). The microstructural features obtained by EBSD analysis of the PC and PCA materials are depicted in Fig. 2. As the AR material is favorably oriented for {10e12} twinning under precompression, numerous {10e12} twins are formed in the PC material (Fig. 2(a) and (e)). The residual matrix region and twinned region have area fractions of 44.5% and 55.5%, respectively (Fig. 2(b) and (c)). Since {10e12} twinning causes a lattice reorientation of 86.3 toward the applied compression direction, the twinned region has an RDoriented basal texture (Fig. 2(c)). Furthermore, this RD twin texture is spread by ±30 toward the TD because the prismatic poles of the AR material are randomly distributed on the RDeTD plane [17,18]. Accordingly, the PC material has two texture components, ND matrix texture and RD twin texture, which correspond to the residual matrix region and twinned region, respectively. During the annealing treatment subsequent to precompression, the imposed thermal energy causes movement of the twin boundaries and grain boundaries of the PC material. It is known that the static recrystallization behavior within deformation twins varies depending on the twin type [19]. In the case of the {10e11} contraction and {10e11}-{10e12} double twins in Mg, basal slip is highly activated within the twins and dislocations accumulate near the boundaries of the twins during plastic deformation; this results in storage of a high amount of energy in the twins. As a result, the nucleation of new grains occurs easily in the {10e11} and {10e11}-
Fig. 1. Microstructural characteristics of AR material: (a) inverse pole figure map, (b) point-to-point misorientation angle distribution for boundaries, (c) grain size distribution, (d) (0001) pole figure, and (e) (10e10) pole figure. The value in parentheses in (c) denotes the standard deviation.
J.U. Lee et al. / Journal of Alloys and Compounds 787 (2019) 519e526
521
Fig. 2. Microstructural characteristics of (aee) PC and (fej) PCA materials: inverse pole figure map and (0001) and (10-10) pole figures of (a, f) total region, (b) residual matrix region, (c) {10e12} twinned region, (g) matrix-originated region (MOR), and (h) twin-originated region (TOR); (d, i) grain orientation spread (GOS) maps; and (e, j) high-angle grain boundary (HAGB) and twin boundary maps. deff and the values in parentheses in (a) and (f) denote the effective grain size and its standard deviation, respectively. fmatrix, ftwin, fMOR, and fTOR denote the area fractions of the residual matrix region, {10e12} twinned region, MOR, and TOR, respectively. GOSavg and ltwin denote the average GOS value and the length of {10e12} twin boundaries per unit area, respectively.
{10e12} twins during subsequent heat treatment [20e22]. On the other hand, since the {10e12} twin in Mg acts as an efficient sink for lattice dislocations, the dislocations are easily absorbed into the twin boundaries [23]. Accordingly, the energy stored in {10e12} twins is generally insufficient for the formation of new grains, which makes the nucleation of recrystallized grains at the twins during annealing difficult; instead, thermally activated twin boundary migration (TATBM) dominantly occurs in the {10e12} twins [24]. Xin et al. [24] reported that during high-temperature annealing of a wrought AZ31 alloy containing {10e12} twins, the narrow twin bands tended to be consumed by the large residual matrix around the twins, causing the twin texture to transform to the matrix texture. In contrast, when the twin bands were much larger than the residual matrix, they preferentially consumed this matrix; the texture change also co-occurred in reverse. In our study, the area fractions of the residual matrix region and twinned region of the PC material are nearly identical. Therefore, in the PCA material, the matrix-originated region (MOR) and twinoriginated region (TOR)dwhich are formed by the growth of the residual matrix and twin bands, respectivelydalso have highly similar area fractions (47.9% and 52.1% for the MOR and TOR, respectively; see Fig. 2(g) and (h)). Therefore, like the PC material, the PCA material also has two texture componentsdND texture and RD texture; however, their intensities are lower in the latter material (Fig. 2). In addition, because of the difference in the amounts of stored internal strain energy in adjacent grains, grain boundaries tend to grow toward the area with higher stored energy
during annealing in order to reduce the total internal energy of the material; this phenomenon is called the strain-induced boundary migration (SIBM) mechanism [25e28]. Therefore, during annealing of the PC material, the twin boundaries and grain boundaries move via the TATBM and SIBM mechanisms, respectively, which consequently results in grain coarsening (from 25.8 mm to 51.3 mm), texture intensity weakening (from 10.9 to 7.3), internal strain energy reduction, and twin boundary annihilation after annealing (Fig. 2). The reduction in the internal strain energy after annealing can be confirmed from the decrease in the average value of the grain orientation spread (GOS)dwhich represents the local misorientation in each graindfrom 1.29 in the PC material to 0.75 in the PCA material (Fig. 2(d) and (i)). Moreover, the length of {10e12} twin boundaries per unit area decreases considerably, from 0.188 mm1 in the PC material to 0.034 mm1 in the PCA material (Fig. 2(e) and (j)), which indicates the considerable annihilation of the twin boundaries during annealing.
3.2. Bending properties of AR, PC, and PCA materials Fig. 3(a) and (b) show the bending loadedisplacement curves and converted stressestrain curves, respectively, of the AR, PC, and PCA materials; Table 1 summarizes the corresponding bending properties. The maximum bending displacement before cracking (i.e., limiting bending depth) of the PC material (3.81 mm) is smaller than that of the AR material (4.98 mm), which can be attributed to the deterioration in the ductility of the PC material
522
J.U. Lee et al. / Journal of Alloys and Compounds 787 (2019) 519e526
Fig. 3. (a) Bending loadedisplacement curves and (b) bending stressestrain curves obtained by three-point bending tests of AR, PC, and PCA materials. Images showing fracture bending samples and optical micrographs taken at mid-width of fracture bending samples of (c) AR and (d) PCA materials. The intermediate regions marked by the yellow dotted lines in (c) and (d) denote the areas where few twins are formed during bending deformation. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
Table 1 Three-point bending properties of AR, PC, and PCA materials. Material
Limiting bending depth (mm)
Maximum bending load (kN)
Bending yield strength (MPa)
Ultimate bending strength (MPa)
Failure bending strain (%)
AR PC PCA
4.98 (±0.10) 3.81 (±0.35) 7.04 (±0.10)
1.97 1.91 1.82
109 112 88
394 383 364
10.0 7.6 14.1
due to the increase in the dislocation density by precompression. Meanwhile, the PCA material shows a lower bending flow stress and a much larger limiting bending depth (7.04 mm) than the AR material. This indicates that the combined application of precompression and subsequent annealing significantly improves the bending formability of the material (by ~41%). The optical micrographs of the fracture bending samples (Fig. 3(c) and (d)) show that in both the AR material and the PCA material, numerous deformation twins are present in the compression and tension zones but few twins are present in the intermediate region between the tension and compression zones. Additionally, cracking occurs at ~40 on the surface of the tension zone in both materials; however,
the area of the tension zone, where the twins are formed, is much larger in the PCA material.
3.3. Twinning behavior of AR and PCA materials during bending deformation To investigate the twinning behavior and textural evolution during bending in the tension zone, where macrocracking occurs and bending formability is decided, the microstructural characteristics in the tension zone of the 2-mm-bent and fracture bending samples of the AR and PCA materials were analyzed by EBSD measurements (Figs. 4 and 5). In the 2-mm-bent sample of the AR
J.U. Lee et al. / Journal of Alloys and Compounds 787 (2019) 519e526
523
Fig. 4. Inverse pole figure maps, boundary maps, and (0001) pole figure in tension zone of (a, c) 2-mm-bent samples and (b, d) fracture bending samples of (a, b) AR and (c, d) PCA materials. ltwin denotes the length of {10e12} twin boundaries per unit area.
Fig. 5. Analysis of twinning behavior in tension zone of (a, c) 2-mm-bent samples and (b, d) fracture bending samples of (a, b) AR and (c, d) PCA materials. TT, CT, and DT denote {10e12} tension twinning, {10e11} contraction twinning, and {10e11}-{10e12} double twinning, respectively.
524
J.U. Lee et al. / Journal of Alloys and Compounds 787 (2019) 519e526
material, some {10e12} twins are observed; these twins are formed in 14 of the 405 grains measured (Fig. 4(a)). Out of these 14 grains, 4 grains are favorably oriented for {10e12} twinning because their basal poles deviate considerably from the ND. However, the remaining 10 grains undergo c-axis contraction deformation under tension along the RD because their basal poles are oriented nearly parallel to the ND (Fig. 5(a)). Namely, the orientations of these 10 grains are completely unfavorable for {10e12} twinning and their Schmid factor (SF) value for {10e12} twinning is 0. This nonSchmid behavior of {10e12} twinning in polycrystalline Mg alloys has been studied recently; this behavior can be attributed mainly to the fluctuation of local stress induced by localized inhomogeneous straining or to the strain compatibility between neighboring grains. For example, Suh et al. [29] demonstrated that {10e12} twinning could be activated under tension along the RD in a rolled Mg alloy with a strong basal texture owing to the compression stress induced along the TD by the constraints of neighboring grains. McClelland et al. [30] also showed that during three-point bending of a rolled AZ31 alloy sheet, {10e12} twins were activated in the tension zone to accommodate the compressive strain applied along the ND during bending. However, in our study, since very few twins are formed in the tension zone of the AR material, they have virtually no contribution to the deformation accommodation. In contrast, numerous {10e12} twins are formed in the tension zone of the PCA material during the initial 2 mm bending; the twin boundary length per unit area increases from 0.034 to 0.103 mm1 (Figs. 2(j) and 4(c)). Since the TOR is favorable for {10e12} twinning under tension along the RD owing to its RD-oriented texture, {10e12} twins form easily during bending. Indeed, the area fraction of grains with the c-axis extension mode under tension along the RD is only 3.1% in the AR material, whereas it is very high, 56.9%, in the PCA material (Table 2). Furthermore, the average SF for {10e12} twinning of these grains in the PCA material is also high, 0.40 (Table 2). The propagation stress of a {10e12} twin is much lower than its nucleation stress; therefore, {10e12} twins formed by precompression easily disappear through detwinning under reversible tensile loading [31]. However, the twinned region of the PC material transforms to new RD-oriented grains (i.e., TOR) by the migration of twin and grain boundaries during annealing;
accordingly, {10e12} twinningdand not detwinningdoccurs in the TOR of the PCA material. In the fracture bending samples, as shown in Fig. 4(b) and (d), numerous narrow {10e11} contraction twins and {10e11}-{10e12} double twinsdrepresented by black lines in the boundary maps owing to the high strains accumulated in themdare observed in both materials. These contraction and double twins rarely accommodate plastic deformation; however, they can cause void formation and cracking [14,15]. It is worth noting that before bending, the textures of the AR and PCA materials are distinctly different, whereas in the fracture bending samples, the textures of the two materials are similar (Fig. 4(b) and (d)). This is because as the deformation progresses, the ND-oriented {10e12} twins formed in the TOR grow gradually, causing texture transformation from the RD to the ND (Fig. 5(c)). Therefore, in the AR material, contraction and double twins are formed in the initial matrix (Fig. 5(b)), whereas in the PCA material, they are formed in the fully {10e12}twinned TOR and also in the MOR (Fig. 5(d)). The microstructural evolution during the precompression and subsequent annealing and the twinning behavior in the tension zone of the AR and PCA materials during bending deformation are schematically illustrated in Fig. 6. This figure visually demonstrates that the TOR with the RD texture is formed via the combined application of precompression and annealing treatments and that the strain imposed during bending is accommodated by {10e12} twinning in the TOR, which consequently results in an improvement in the bending formability. 3.4. Improvement in bending formability of PCA material The twinning shear values for {10e12} extension twinning and {10e11} contraction twinning are similar: 0.129 and 0.138, respectively [32]. However, the dislocation width for {10e12} twinning is ~6 times that for {10e11} twinning, whereas the Burgers vector is much smaller for {10e12} twinning (0.049 and 0.135 nm for {10e12} twinning and {10e11} twinning, respectively) [33]. Therefore, {10e12} twinning can be easily activated at low stress levels owing to its low Peierls stress, and relatively large plastic deformation can be accommodated through the growth of {10e12} twins because the twin boundaries have high mobility. For
Table 2 Average Schmid factors for {10e12} twinning and basal slip under tension along RD for AR and PCA materials. Material
ftension (%)a
SF{10e12},
AR PCA
3.1 56.9
0.32 0.40
tension
b
ε{10e12} (%)c
SFbasal
0.13 2.94
e 0.34
d slip, MOR
SFbasal e 0.32
slip, TOR
d
SFbasal
d slip
0.21 0.33
a
ftension denotes the area fraction of grains with the c-axis tension mode under tension along the RD. SF{10-12}, tension denotes the average Schmid factor value for {10e12} twinning under tension along the RD of grains with the c-axis tension mode. ε{10-12} denotes the strain accommodated by {10e12} twinning. d SFbasal slip, MOR, SFbasal slip, TOR, and SFbasal slip denote the average Schmid factor values for basal slip under tension along the RD of the matrix-originated region, twinoriginated region, and total region, respectively. b c
Fig. 6. Schematic illustration describing microstructural evolution during precompression and subsequent annealing and twinning behavior in tension zone during bending deformation.
J.U. Lee et al. / Journal of Alloys and Compounds 787 (2019) 519e526
these reasons, activation of {10e12} twinning in the tension zone of the PCA material causes a reduction in the bending flow stress and an improvement in the bending formability. The strain accommodated by twinning (i.e., twinning strain) is calculated as εtwin ¼ ftwin $m$gtwin , where εtwin , ftwin , m, and gtwin denote the twinning strain, area fraction of the twinned region, average SF for {10e12} twinning under deformation conditions, and characteristic twinning shear, respectively [12]. In the tension zone of the PCA material, all regions with the c-axis extension mode under tension along the RD experience twinning during bending, which is confirmed from the missing RD texture in the fracture bending sample; therefore, the calculated strain accommodated by {10e12} twinning in the tension zone of the PCA material during bending is 2.94% (¼56.9% 0.40 0.129) (Table 2). This strain accommodation through {10e12} twinning consequently causes a 4.1% increase in the failure bending strain: from 10.0% for the AR material to 14.1% for the PCA material (Table 1). Additionally, the SF for basal slip under tension along the RD is also relatively high in both the MOR (0.34) and the TOR (0.32) of the PCA material (Table 2). Therefore, the average SF for basal slip in the PCA material (0.33) is 57% higher than that in the AR material (0.21) (Table 2), which also contributes to the improvement in bending formability via promotion of activation of basal slip in the tension zone. At the early stage of bending deformation, {10e12} twins are formed in the TOR of the PCA material, and then, the TOR gradually transforms to the {10e12} twinned region through the growth of the formed twins (Fig. 4). As {10e12} twinning leads to a lattice reorientation of 86.3 , the degree of activation of dislocation slips in the material changes with the nucleation and growth of the twins in the TOR during bending. Fig. 7(a) shows the basal pole distributions of the grains of the 2-mm-bent AR sample and those of the {10e12} twin bands of the 2-mm-bent PCA sample. Since the activation of {10e12} twinning in the TOR leads to texture transformation from the RD to the ND, the crystallographic orientation distribution of the {10e12} twin bands formed in the PCA sample becomes similar to that of the grains of the AR sample (Fig. 7(a)). As a result, the SFs for basal slip in the twin bands of the 2-mm-bent PCA sample and in the grains of the 2-mm-bent AR sample under tension along the RD are nearly identical (0.20 and 0.19 for the AR
525
and PCA samples, respectively), and similarly, those for prismatic slip in the twin bands of the 2-mm-bent PCA sample and in the grains of the 2-mm-bent AR sample under tension along the RD are also nearly identical (0.43 and 0.44 for the AR and PCA samples, respectively) (see Fig. 7(b)e(e)). This means that the deformation behavior within the {10e12} twins formed during bending in the PCA material is almost the same as that within the initial grains of the AR material. In the PC material, the area fraction of the residual matrix region and twinned region is ~50% each. If a rolled Mg alloy is fully twinned by application of large precompression, the precompressed alloy may have only a strong RD twin texture and lack the ND matrix texture. In this textural structure, accommodation of plastic deformation is difficult in the compression zone, but not in the tension zone, because the twinned region is unfavorably oriented for {10e12} twinning and basal slip under compression along the RD. Therefore, in this study, the amount of precompression is intentionally adjusted to achieve the formation of a twinned region with an area fraction of ~50% for accommodating the plastic deformation in both the tension zone and the compression zone through {10e12} twinning. Since the area fraction of the twinned region varies with the amount of precompression, the amount of strain that can be accommodated through twinning in the tension and compression zones of the PCA material during bending deformation also changes with the amount of applied precompression. Moreover, microstructural characteristics such as grain size, texture, and internal strain energy can vary with the conditions of annealing performed after precompression. Therefore, to maximize the improvement in the bending formability of rolled Mg alloys via the proposed method, the effects of the precompression amount (i.e., twin area fraction) and annealing conditions (e.g., temperature, time, and cooling rate) on the microstructure, texture, and bending properties of rolled Mg alloys should be studied further. 4. Conclusions The bending formability of a rolled AZ31 alloy with a strong basal texture at RT is significantly improved through combined
Fig. 7. (a) Comparison of crystallographic orientation distributions of grains of 2-mm-bent AR sample and twin bands of 2-mm-bent PCA sample as a function of deviation angle of basal pole from ND. (bee) Schmid factor (SF) maps for (b, d) basal slip and (c, e) prismatic slip under tension along RD: (b, c) grains of 2-mm-bent AR sample and (d, e) twin bands of 2-mm-bent PCA sample. SFbasal and SFprism denote the average SFs for basal slip and prismatic slip, respectively.
526
J.U. Lee et al. / Journal of Alloys and Compounds 787 (2019) 519e526
application of precompression and subsequent annealing. The introduction of {10e12} twins via precompression induces the formation of an RD twin texture, and subsequent annealing causes migration of grain boundaries and twin boundaries; this consequently yields a stable grain structure with two texture components: a weakened ND texture and a TD-spread RD texture. This tailored texture effectively accommodates the tensile deformation induced in the sample's outer region during bending through activation of {10e12} twinning and promotion of basal slip, thereby delaying cracking and improving the bending formability by ~41%. Acknowledgements This work was supported by National Research Foundation of Korea (NRF) grants funded by the Korean government (MSIP, South Korea) (Nos. 2016R1C1B2012140 and 2017R1A4A1015628). References [1] J. Bohlen, M.R. Nürnberg, J.W. Senn, D. Letzig, S.R. Agnew, Acta Mater. 55 (2007) 2101e2112. [2] K.K. Alaneme, E.A. Okotete, J. Mag. Alloys 5 (2017) 460e475. [3] D. Song, T. Zhou, J. Tu, L. Shi, B. Song, L. Hu, M. Yang, Q. Chen, L. Lu, J. Mater. Process. Technol. 259 (2018) 380e386. [4] B. Song, R. Xin, A. Liao, W. Yu, Q. Liu, Mater. Sci. Eng. A 627 (2015) 369e373. [5] H. Zhang, W. Cheng, J. Fan, B. Xu, H. Dong, Mater. Sci. Eng. A 637 (2015) 243e250. [6] J. Singh, M.S. Kim, S.H. Choi, Int. J. Plast. (2018). http://doi.org/10.1016/j.ijplas. 2018.01008. [7] J.C. Baird, B. Li, S.Y. Parast, S.J. Horstemeyer, L.G. Hector Jr., P.T. Wang, M.F. Horstemeyer, Scripta Mater. 67 (2012) 471e474. [8] L. Jin, J. Dong, J. Sun, A.A. Luo, Int. J. Plast. 72 (2015) 218e232. [9] L. Wang, G. Huang, T. Han, E. Mostaed, F. Pan, M. Vedani, Mater. Des. 68 (2015)
80e87. [10] I. Aslam, B. Li, Z. McClelland, S.J. Horstemeyer, Q. Ma, P.T. Wang, M.F. Horstemeyer, Mater. Sci. Eng. A 590 (2014) 168e173. [11] M. Kohzu, K. Kii, Y. Nagata, H. Nishio, K. Higashi, H. Inoue, Mater. Trans. 51 (2010) 749e755. [12] C.S. Roberts, Magnesium and its Alloys, John Wiley & Sons, New York, 1960. [13] B. Li, Q. Ma, Z. McClelland, S.J. Horstemeyer, W.R. Whittington, S. Brauer, P.G. Allison, Scripta Mater. 69 (2013) 493e496. [14] M.R. Barnett, Mater. Sci. Eng. A 464 (2007) 8e16. [15] S. Niknejad, S. Esmaeili, N.Y. Zhou, Acta Mater. 102 (2016) 1e16. [16] ASTM International, Standard Test Methods for Bend Testing of Material for Ductility, ASTM International, West Conshohocken, PA, 2004. [17] S.G. Hong, S.H. Park, C.S. Lee, Acta Mater. 58 (2010) 5873e5885. [18] S.H. Park, S.-G. Hong, C.S. Lee, Scripta Mater. 62 (2010) 202e205. [19] D. Guan, W.M. Rainforth, L. Ma, B. Wynne, J. Gao, Acta Mater. 126 (2017) 132e144. [20] I. Basu, T. Al-Samman, Acta Mater. 96 (2015) 111e132. [21] M.H. Yoo, Metall. Trans. A 12 (1981) 409e418. [22] K.D. Molodov, T. Al-Samman, D.A. Molodov, G. Gottstein, Acta Mater. 76 (2014) 314e330. , Acta Mater. 85 (2015) 354e361. [23] H.E. Kadiri, C.D. Barrett, J. Wang, C.N. Tome [24] Y. Xin, H. Zhou, G. Wu, H. Yu, A. Chapuis, Q. Liu, Mater. Sci. Eng. A 639 (2015) 534e539. [25] P.A. Beck, P.R. Sperry, J. Appl. Phys. 21 (1950) 150e152. [26] G. Gottstein, L.S. Shvindlerman, Grain Boundary Migration in Metals, second ed., Taylor & Francis, Boca Raton, FL, 2010. [27] R.R. Keller, R.H. Geiss, N. Barbosa III, A.J. Slifka, D.T. Read, Metall. Mater. Trans. A 38A (2007) 2263e2272. [28] A. Paggi, G. Angella, R. Donnini, Mater. Char. 107 (2015) 174e181. [29] B.C. Suh, M.S. Shim, D.W. Kim, N.J. Kim, Scripta Mater. 69 (2013) 465e468. [30] Z. McClelland, B. Li, S.J. Horstemeyer, S. Brauer, A.A. Adedoyin, L.G. Hector Jr., M.F. Horstemeyer, Mater. Sci. Eng. A 645 (2015) 298e305. [31] L. Wu, A. Jain, D.W. Brown, G.M. Stoica, S.R. Agnew, B. Clausen, D.E. Fielden, P.K. Liaw, Acta Mater. 56 (2008) 688e695. [32] P.G. Partridge, Metall. Rev. 12 (1967) 169e194. [33] R.E. Reed-Hill, R. Abbaschian, Physical Metallurgy Principles, third ed., PWSKent, Boston, 1994.