Influence of Volume Fraction of Long-Period Stacking Ordered Structure Phase on the Deformation Processes during Cyclic Deformation of Mg-Y-Zn Alloys

Abstract

Deformation mechanisms in extruded Mg-Y-Zn alloys with different volume fractions of the long-period stacking ordered (LPSO) structure have been investigated during cyclic loading, i.e., compression followed by unloading and reverse tensile loading. Electron backscattered diffraction (EBSD) and in situ neutron diffraction (ND) techniques are used to determine strain path dependence of the deformation mechanisms. The twinning-detwinning mechanism operated in the α-Mg phase is of key importance for the subsequent hardening behavior of alloys with complex microstructures, consisting of α-Mg and LPSO phases. Besides the detailed analysis of the lattice strain development as a function of the applied stress, the dislocation density evolution in particular alloys is determined.

1. Introduction

The development of magnesium (Mg) alloys with long-period stacking ordered (LPSO) structures belong to the most important metallic lightweight alloy innovations of the last two decades. The mechanical properties of such materials surpass those of conventional cast or wrought Mg alloys [1,2,3,4].

The LPSO phase-α-Mg matrix interface is reported as a site for dynamic recrystallization, usually leading to the formation of a bimodal grain structure in wrought Mg-Y-Zn alloys [5]. The complex microstructure of extruded Mg-LPSO alloys, including the fine dynamically recrystallized (DRX) and coarse non-DRX grains of the α-Mg phase and the fiber-shaped LPSO-phase, has been found to be beneficial for the resulting mechanical properties [4]. The stiffer LPSO phase reinforces the magnesium matrix, what results in excellent strength values. At the same time, fine DRX grains contribute to the ductility of the alloy.

Naturally, the amount of the alloying elements (Zn and Y) influences the microstructure evolution during both casting and thermo-mechanical processing and determines the volume fraction of the formed LPSO phase, therefore, strongly affecting the mechanical properties [6,7]. Among Mg-LPSO alloys, the best tensile properties at room temperature were achieved for the Mg₉₇Y₂Zn₁ (at.%) alloy prepared by powder metallurgy (tensile yield strength of 610 MPa, elongation of 5%) [2]. Using another processing route, such as rapid solidification (RS) of the ribbons and its consolidation via extrusion, the materials having ultrafine grain microstructures with dispersed stacking faults can be produced. The RS-extruded Mg_97.94Zn_0.56Y_1.5 (at.%) alloy can exhibit a yield strength of 362 MPa and elongation of 18.2% [8]. It is obvious that with a proper choice of alloy content and processing parameters (e.g., temperature, extrusion ratio, and extrusion rate) the microstructure and mechanical properties of the Mg-LPSO alloys can be varied in a wide range.

Several deformation mechanisms can operate in particular microstructural elements during the straining of wrought Mg-LPSO alloys. Activity of basal slip was reported in DRX α-Mg grains [5]. Concurrently, the non-basal slip as well as the presence of extension $\left\{10\overline{1}2\right\}\langle 10\overline{1}\overline{1}\rangle$ twinning has been observed in coarse non-DRX α-Mg grains [5]. Those mechanisms can also be active in DRX α-Mg grains, but at significantly higher applied stresses (the grain size effect) [9]. The LPSO phase is known to be deformed by kinking when concurrently shear and compression deformation act parallel to LPSO fibers. This mechanism is governed by the collective motion of basal dislocations. If the load is applied perpendicular to the LPSO fibers, rather non-basal slip dominates [5]. Thus, besides volume fraction, the orientation and distribution of the LPSO phase significantly affect the mechanical behavior of the material [5,7,10,11,12,13,14,15]. The orientation effect is given by the texture formed during the processing of wrought Mg-LPSO alloys. In general, both α-Mg and LPSO phases are characterized by their basal planes oriented parallel to extrusion direction (ED) [5]. The preferential activity of deformation mechanisms, with respect to the mutual loading axis and c-axis of the lattice, determines the resulting mechanical properties. For example, the polar nature of extension twinning [16,17,18,19] leads to extensive twin nucleation during compressive loading along ED (i.e., during compression perpendicular to the c-axis of the lattice of the α-Mg phase), and therefore, significantly lower yield stress comparing to one during tensile loading is observed [20]. During subsequent reverse loading of pre-compressed Mg alloy, detwinning could be activated in the twinned fractions [9,21,22]. This process is usually characterized by thickness reduction or complete disappearance of existing twin lamellae as their lattices are rotating back to the original orientation [23]. Thus, if the extension twins occur during in-plane compression along ED, detwinning takes place during subsequent reverse tensile loading. Thus, understanding the cyclic loading behavior is essential for engineering applications. However, there are only a few works dealing with this issue in Mg-LPSO alloys. Hagihara et al. [24] investigated low cycle fatigue properties for several Mg alloys with various content of the LPSO phase. They found that cyclic hardening is not significant. Furthermore, the apparent yield stress (YS) is gradually decreasing with increasing number of cycles for Mg₉₇Zn₁Y₂ alloys (in at%) in both (extrusion and transversal) loading directions. In contrast, a gradual increase in YS was found for the tensile side of the cycle in Mg_99.2Zn_0.2Y_0.6 alloy. In cast Mg-LPSO alloys, besides the influence of the composition, the effect of compressive pre-loading was studied [25]. Both the compressive pre-loading and the higher LPSO content increased the share of the kinematic hardening to cyclic hardening.

Similar to conventional Mg alloys, in Mg-LPSO-based materials, the twinning-detwinning phenomenon operated in the α-Mg phase plays a significant role during cyclic loading. To investigate this process, besides microstructure observations provided by scanning electron microscopy (SEM) (in situ and post mortem) [9,22,26,27], the in situ diffraction method has been found to be a powerful experimental tool [5,27,28]. The twinned volume can be estimated from the intensity variations of particular diffraction peaks [29]. At the same time, the activation stress of various dislocation slip systems in α-Mg and LPSO phases can be deduced from the evaluation of the lattice strain with applied stress [30].

The main aim of the present work is to reveal the single-cycle compression-tension properties of Mg-LPSO alloys with respect to the variation in the volume fraction of the LPSO phase. The experimental approach includes in situ neutron diffraction and mapping of the microstructure by electron backscattered diffraction (EBSD) technique. The novelty of the work consists in the identification of the active deformation mechanisms during the strain path change and revealing the evolution of the dislocation structure in the Mg matrix as a function of the applied load and fraction of the LSPO phase.

2. Experimental

For the present study, three alloy compositions WZ42-Mg_98.5Y_1.0Zn_0.5, WZ72-Mg_97.0Y_2.0Zn_1.0, and WZ104-Mg_95.5Y_3.0Zn_1.5 (nominal compositions in at%) were selected. The master alloys were cast at the Korea Institute of Industrial Technology and further extruded at 350 °C with 0.5 mm s⁻¹ extrusion speed and an extrusion ratio of 1:10 in National Center for Metallurgical Research (CENIM-CSIC), Madrid.

The in situ neutron diffraction (ND) measurements were performed on the VULCAN Engineering Materials Diffractometer beamline at Oak Ridge National Laboratory. The cylindrical specimens (thread-ended, gauge length and diameter were 20 and 9 mm, respectively) were fixed horizontally in a deformation rig manufactured by Measure Test Simulate (MTS) company. The specimens were loaded along ED at room temperature. The angle between the incident beam and the specimen was 45°. The diffraction patterns in the axial and radial directions were detected by two stationary detector banks located at ± 90° to the incoming beam. The neutron gauge volume was 245 mm³. The diffraction patterns were recorded in continuous and discontinuous modes, respectively. In the first case, the deformation cycle was executed without stopping at a strain rate of 10⁻³ s⁻¹, and the diffraction data (patterns) were recorded continuously. For the evaluation, obtained data were binned to 1 min long intervals. Since the proper evaluation of the LPSO peaks and the diffraction line profile analysis of the magnesium peaks requires data with good enough statistics, the tests in the discontinuous mode were stopped at pre-defined strain levels (0.1, 0.5, 1, 2, 3, 4%) for approximately 20 min for the collection of diffraction patterns. The ND patterns recorded in the discontinuous mode were evaluated by the Convolutional Multiple Whole Profile (CMWP) fitting method [31,32]. The diffraction patterns were fitted by functions constructed from the background spline, instrumental pattern, and a theoretical profile function. The theoretical profile function assumes dislocation-caused microstrains. As a result of the fitting procedure, the dislocation density (ρ) has been directly obtained.

The microstructure of the as-extruded material, as well as specimens deformed up to specified strain levels, has been investigated using SEM Quanta FX200 (Field Electron and Ion Company—FEI Company, Hillsboro, OR, USA) and Auriga Compact (Zeiss, Oberkochen, Germany) both equipped with an EBSD camera (EDAX Inc., Mahwah, NJ, USA). The specimen for microscopy observations were ground on SiC papers and subsequently polished by diamond paste with a particle size decreasing down to 0.25 μm. The final step of specimen preparation consisted of ion polishing using a Leica EM RES102 device.

3. Results and Discussion

The initial microstructure and texture of the specimens are presented in Figure 1. Investigated alloys are characterized by microstructure consisting of the LPSO phase (light contrast in backscatter electron (BSE) images) and α-Mg grains (dark contrast in BSE images), including small DRX and coarse non-DRX grains elongated along ED. The volume fractions of the LPSO phase has been estimated from the BSE images as 10, 21, and 35% for the WZ42, WZ72, and WZ104 alloys, respectively. The fraction of coarse non-DRX grains and their grain size decreases with the increasing amount of the alloying elements from the WZ42 alloy towards the WZ104 alloy. Those elongated grains in all of the studied alloys have an intensive texture with their c-axis perpendicular to ED. (Those few grains represented in the EBSD maps contributes to the higher texture intensities along the periphery of (0001) pole figures). Although the DRX grains show a more random orientation distribution, the overall texture has an ordered character with the basal planes oriented parallel to ED, see pole figures composed from EBSD maps and rather more informative intensity distribution evaluated from ND data in Figure 1d. The decrease in texture intensity with increasing alloying elements can be associated with the decrease in the volume fraction of coarse non-DRX α-Mg grains.

The obtained EBSD maps are indexed for α-Mg phase and black areas in the maps (related to pixels with a confidence index for α-Mg < 0.1) are therefore associated with the LPSO phase. The volume fractions of the LPSO phase estimated by BSE images are in good agreement with those obtained from the EBSD maps.

The deformation curves are presented in Figure 2. (The small waves on the curves around 0 MPa during unloading are given by the backlash of the deformation setup.) It is obvious that the flow stress increases with increasing volume fraction of the LPSO phase. First, during the compression part, a plateau follows the macroscopic yield, especially pronounced for WZ42 (see the inserted part in Figure 2), which can be a sign of enhanced activity of $\left\{10\overline{1}2\right\}\langle 10\overline{1}\overline{1}\rangle$ extension twinning [20]. During reverse tension, the stress–strain curve of the WZ42 specimen shows a pronounced S-shape, indicating a significant role of detwinning [9,33]. In the case of the WZ72 and WZ104 alloys, the S-shape is not observed, and their stress-strain curves have convex shape, typically observed during tensile loading of Mg alloys.

The microstructures of the alloys after the full-cycle deformation are presented in Figure 3. In the WZ42 alloy, cracks in the LPSO lamellae formed during straining are observed, Figure 3a. Kinking is characteristic for the WZ72 and WZ104 alloys, Figure 3b,c. The result of high twinning activity in WZ42 alloy can be spotted on the image quality (IQ) maps in Figure 3d. There is a relatively high number of leftover twin boundaries (highlighted by the red color). The twin boundaries are identified as extension $\left\{10\overline{1}2\right\}\langle 10\overline{1}\overline{1}\rangle$ twins with a misorientation angle of 86.3° with respect to the original lattice. In the case of the WZ72 alloy, this microstructural feature is negligible (cf. Figure 3e), and no twin boundaries have been observed in the WZ104 alloy (not presented here), which can be explained by the decreased amount and grain size of α-Mg grains in this alloy.

The appropriate volume fraction and grain size of non-DRX grains in the WZ42 alloy give a rise to investigate the evolution of twinning at particular stages of cyclic loading using the EBSD/SEM technique with high confidence. Therefore, an additional specimen was deformed by the same deformation loading cycle like those for ND tests. The microstructure has been examined just above the macroscopic yield, at 330 MPa of compression stress, and after a reverse tension up to 225 MPa, Figure 4. It can be seen that, after reaching the yield point, the twins appear mainly in elongated non-DRX grains (Figure 4a) and several twin variants are nucleated within a particular grain, forming families of twins (groups of twins with same orientation with respect to original grain). A schematic view in Figure 4a represents activated variants of twins with respect to the orientation of the parent grain. It can be seen that the tilt of nucleated twin lamellae with respect to ED (lenticular twin lamella is perpendicular to ED in “red-colored” grains and tilted by 45° from ED in “green-yellow” grains) is given by orientation of the twin plane with respect to the observation point of view. With increasing load (Figure 4b), the already existing twins become thicker, and new thin twins occur as well. It should be noted that, even at the compression peak stress, the non-DRX grains are still not fully twinned. Thus, twin boundaries are supposed to be mobile for further forward or backward movement. At the same time, twins in the small DRX grains can be also observed at higher compressive stresses, cf. BSE image in Figure 4b. During the reverse-tensile loading (Figure 4c), the twins nucleated during compression become thinner, and only a few narrow twins can be found in some non-DRX grains. The leftovers of these twin boundaries can be seen in Figure 4c. Besides, the trace of twin boundaries after complete detwinning are also seen, see gray lines in the white rectangular in Figure 4c. This could be explained by the localization of strain inside the lattice due to a high concentration of dislocations, which have been acting for twin growth and shrinkage. The operation of twins is also well seen in the texture development (see pole figures in Figure 4). In the initial state, basal planes are oriented parallel to ED, therefore texture intensity is distributed at the periphery of the (0001) pole figure. With increasing applied stress, a strong texture component in the middle of (0001) pole figure is formed due to the rotation of basal planes by almost 90° from the original orientation toward ED as a result of $\left\{10\overline{1}2\right\}\langle 10\overline{1}\overline{1}\rangle$ twinning. The intensity of this texture component increases with increasing compressive loading, what can be associated with massive twin growth. The texture of the specimen after reverse-tensile loading is comparable to the one in the initial state. The disappearance of the texture component is given by detwinning, i.e., rotation of the lattice in the twin fractions back to the original orientation.

Owing to the high fraction of the LPSO phase and a significantly low fraction of non-DRX grains, similar EBSD/SEM analysis is rather difficult for the WZ72 and WZ104 specimens. Thus, from a statistical point of view, the ND data, which characterize a large specimen volume, gives a better insight into the twinning development.

The change of the intensity of the (0002)–$\left\{10\overline{1}0\right\}$ diffraction peak pair is directly related to the extension twinning [29]. With respect to the initial texture of specimens and the diffraction geometry used in our experiments, the intensity of the (0002) peak is expected to increase in the axial detector, whereas the intensity of the $\left\{10\overline{1}0\right\}$ peak should decrease once the extension twinning is active. In Figure 5, the intensity changes in the axial direction of (0002)–$\left\{10\overline{1}0\right\}$ peaks are plotted as a function of the applied stress and strain. It can be seen that, in compression, the twinned volume starts to increase at significantly lower stress in WZ42 than that in the WZ72 and WZ104 specimens. Moreover, there is an effect of volume fraction of the LPSO phase (i.e., alloy content) on the resulting twinned volume; the twin volume fraction for WZ42 is significantly larger compared to that for the other two alloys. During the unloading, detwinning took place, however, it is not finished at zero stress and continue during tension. The intensity changes plotted against strain unambiguously indicate that the detwinning is terminated after reaching the zero-strain value (cf. Figure 5c).

The lattice strains plotted for axial detector as a function of applied stress, Figure 6, are determined using the equation

$$\epsilon =\frac{d-d_0}{d_0},$$

(1)

where d, d₀ are the lattice spacing for deformed and stress-free conditions, respectively. In this representation, a deviation from the linear Hooke’s elasticity indicates activation of specific deformation mechanisms. However, due to the plateau stress present for all alloys above the yield point, the applied stress-lattice strain plot does not provide an easy survey. Therefore, the applied strain-lattice strain plots are also shown here, which are more representative in the region of the plastic deformation.

The stress evolution of the (0002)–$\left\{10\overline{1}0\right\}$ lattice strain indicates the activity of extension twinning with respect to the composition of the alloys, and therefore the volume fraction of the LPSO phase. The lattice strain evolution is in a good agreement with the development of intensity of the (0002)–$\left\{10\overline{1}0\right\}$ diffraction peak pair presented in Figure 5.

The onset of the twinning takes place far below the macroscopic strain for the WZ42 and WZ72 specimens, and the activation stress increases with the increasing amount of alloying elements. The (0002) grains (that is, grains with their (0002) axis along the axial direction) are in “soft-orientation” during compression, and they relax after twin initiation, whereas the “hard-oriented” $\left\{10\overline{1}0\right\}$ grains accommodate a higher portion of elastic loading. The relaxation is well seen in the applied strain-lattice strain plot (Figure 6d,e).

During unloading, the compressive lattice strains in these grains relax. In the tension part, the soft–hard orientation roles interchange owing to the polar nature of the extension twinning [16,17,18,19]. Consequently, a hardening of (0002) oriented grains is observed, which are now unfavorably oriented for extension twinning with respect to the loading direction. This feature is the most significant for the WZ42 alloy. For the $\left\{11\overline{2}0\right\}$ orientation, the contribution of prismatic and pyramidal <a>-slip also cannot be excluded (Schmid-factor values for the slip in these systems are 0.43 and 0.38, respectively). For the WZ42 specimen in tension, the $\left\{11\overline{2}0\right\}$ lattice strain relaxes around 150 MPa, indicating significant non-basal <a>-slip.

In the case of the WZ104 specimen, the $\left\{4\overline{2}\overline{2}8\right\}$ $\left\{4\overline{2}\overline{2}10\right\}$, and $\left\{41\overline{3}1\right\}$ LSPO diffraction peaks were intensive enough for confident lattice strain evaluation. As it can be seen from Figure 6c,f,g, the $\left\{4\overline{2}\overline{2}8\right\}$ and $\left\{41\overline{3}1\right\}$ planes of the LPSO phase shares the highest load compared to other ($\left\{4\overline{2}\overline{2}10\right\}$) LPSO- and α-Mg-related planes below the yield point. In the plastic region, around −0.015 of applied strain further softening of the $\left\{4\overline{2}\overline{2}10\right\}$ planes takes place, whereas hardening of $\left\{41\overline{3}1\right\}$ is observed. In [34], it was shown that the change in the lattice strain distribution on the $\left\{4\overline{2}\overline{2}10\right\}$ planes can be associated with deformation by kinking. During unloading and reverse-tensile loading, relaxation takes place.

The evolution of the dislocation density in the α-Mg phase as a function of the applied stress for the investigated alloys is plotted in Figure 7. In the compression part, all of them behave similarly; above the yield point, the dislocation density increases due to continuously active dislocation slip systems. The highest ρ value is reached for the WZ104 alloy. After change of strain path (direction), for all investigated alloys, ρ decreases, what can be associated with annihilation of dislocation due the relaxation of internal stresses and closing sources of dislocations activated during compressive load. However, after reaching a certain stress level, the decrease in ρ terminates thanks to the activation of new sources of dislocations activated as a result of tensile loading. In the WZ42 alloy, detwinning significantly affects dislocation density development, Figure 5a. In the grain fraction, which reorients to the initial orientation, further slip is possible. Therefore, detwinning during tension is accompanied by a slight increment of the dislocation density. However, complete detwinning causes annihilation of a high number of dislocations on twin boundaries, leading to decrease in dislocation density in the secondary hardening part.

The results listed above indicate that the volume fraction of the LPSO phase significantly influences the deformation behavior of the Mg-LPSO alloys. For the WZ42 specimen, where the fraction of the LPSO phase is the lowest, the bimodal grain structure enhances the extension twinning. According to the observations for wrought Mg alloys [35,36], the coarse non-DRX grains twin first. The reason for this behavior is given by the low internal stress in those grains and by the long grain boundary, which serves as a nucleation site for twinning [37]. In contrast, the WZ104 alloy is characterized by a 35% fraction of the LPSO phase and a significantly smaller fraction of non-DRX grains than that in the WZ42 alloy (Figure 1). Accordingly, the highest twinned volume can be observed in the WZ42 alloy. During the unloading, detwinning takes place in all investigated alloys, especially pronounced for the WZ42 alloy. The detwinning is enhanced due to the limited twin thickening during compressive load as a result of the stored internal stress introduced into the alloy by the presence of the LPSO phase. Therefore, the majority of twins are in unrelaxed conditions, leading to their easier disappearance during reverse loading [9,38]. A detailed inspection of the lattice strain evolution for WZ42 and WZ72 specimen suggests that, in compression, the governing mechanism at the onset of plasticity is extension twinning (with significant dominance in WZ42). Basal slip also takes place, but its share is smaller than that of twinning. This finding is in good agreement with the work of Vinogradov et al. [39] in wrought ZK60 magnesium alloy, where dominance of twinning at the beginning of the plastic flow has been revealed by analysis of the acoustic emission signal. However, the macroscopic yield needs the activation of the non-basal <a>-slip as well, leading to the rapid increment of the dislocation density (cf. Figure 7) and relaxation of the lattice strains on the $\left\{10\overline{1}1\right\}$ and $\left\{10\overline{1}2\right\}$ grain families. At the beginning of reverse tension, the detwinning is dominating. Nevertheless, the <a>-slip on both the basal and non-basal planes is significant. Owing to the basal texture, significantly fewer grains are favorably oriented for extension twinning in tension than in compression. Consequently, the role of dislocation slip in plasticity increases.

In the WZ104 alloy, the development of the deformation mechanisms is different. Owing to the high fraction of the LSPO phase, a composite-like behavior is observed when the LSPO phase shares a larger portion of the load than the magnesium matrix. This behavior can be followed in Figure 6c. Due to this stress shielding effect, the twin nucleation, which is also suppressed by the small grain size of α-Mg grains and large internal stresses, can start only around the yield point. In compression, above the yield point, the lattice strain in the LPSO phase relaxes, indicating the deformation of this phase by the kinking mechanism, Figure 3c. During reverse tension, it is obvious (see Figure 6c) that the role of extension twinning is negligible. The strain is accommodated by <a>-slip and also by deformation of the LPSO phase. However, in tension, the kink formation is unlikely, and the LPSO phase is deformed rather by non-basal <a>-slip [40,41]. The enhanced dislocation activity both in the LPSO-phase and the magnesium matrix leads to high internal stresses and early failure of the material.

4. Conclusions

The in situ neutron diffraction (ND) technique has been employed to determine strain path dependences of the deformation mechanisms in Mg-Y-Zn alloys with various volume fractions of the LPSO phase.

The obtained ND data gives insight into the twinning-detwinning mechanisms operating in the investigated alloys. This mechanism is mainly realized in the non-DRX α-Mg grains and affects the overall development of deformation behavior of the Mg-LPSO alloys. It is the most significant in the WZ42 alloy with the high-volume fraction of α-Mg grains at the expense of the lowest volume fraction of the LPSO phase. The WZ104 alloy, having a high-volume fraction of the LPSO phase, behaves as a composite material, and the LPSO phase bears the main part of the applied compressive load. Further, the non-basal slip and kinking dominate the plastic deformation. Consequently, the twinning–detwinning mechanism plays a minor role, and the strain in the α-Mg-matrix is rather accommodated by non-basal slip.

Additionally, the analysis of ND profiles provides information about the development of dislocation density during cyclic loading. After reaching the yield point in compression, the dislocation density increases in all alloys, followed by a decrease during unloading. The highest dislocation density develops in the WZ104 alloy owing to the low twinning activity.