Microstructure and Its Evolution of Solute-Enriched Stacking Faults in Kink-Deformed Mg₉₇Zn₁Y₂

Abstract

Microstructure and its evolution of solute-enriched stacking faults (SESFs) in kink-deformed Mg–Zn–Y alloys have been investigated by scanning transmission electron microscopy (STEM). Mass-thickness contrast in annular dark-field (ADF) STEM images and diffraction contrast in weak-beam dark-field (WBDF) STEM images visualized locations of solute-enriched regions and dislocations, respectively. Most of the SESFs are located at or near kink boundaries, and a variety of arrangements and morphology of the SESFs are followed by dislocations with the c component of α-Mg matrix and small-angle lattice rotations along the c-axis as well as the a-b axes. As notable microstructural features of the SESFs, rectangular-shaped and step-shaped arrangements of the SESFs, observed respectively before and after a heat treatment at 573 K for 168 h, were observed in detail, and their formation processes were discussed.

1. Introduction

Mg–TM–RE (TM: Al and transition metals, RE: Y and rare earth elements) alloys have garnered significant attention in the research field of lightweight metals due to their excellent mechanical properties. This exceptional mechanical performance is believed to be associated with the unique kink deformation promoted by the presence of the long-period stacking order (LPSO) structure in Mg–TM–RE alloys.^1–6) In addition to the LPSO structure, solute-enriched stacking faults (SESFs), which have a face-centered cubic stacking, also exist in the hexagonal close-packed α-Mg matrix. The distribution of such LPSO and SESFs forms microstructural variations in the α-Mg matrix.^7–9) Due to the existence of the LPSO structure, traditional dislocation motion is restricted, and instead, the unique kink deformation is significantly promoted.^10–14) Kink deformation is considered a primary factor in enhancing the alloy strength.^7,15) The characteristic feature of the microstructure resulting from the kink deformation is the crystal rotation at kink boundaries (KBs). In recent reports, for certain Mg–TM–RE alloys, KBs can exist independently of the LPSO structure and be randomly distributed in the α-Mg matrix. From a microstructural viewpoint, KBs manifest as periodic arrangements of nanosized SESFs that accompany geometrically necessary (GN) dislocations.^9,12)

As for the SESFs, Egusa et al.¹⁶⁾ investigated asymmetric X-ray diffraction peaks originating from the α-Mg matrix and the SESFs and evaluated their volume fractions. It was indicated that the majority of SESFs have less than a few unit cells of α-Mg in thickness and accompany dislocations with c-component (designated c′-dislocations herein-after). There is also a report on the distinctive characteristics of segmented KBs in heat-treated Mg–TM–RE alloys.⁹⁾ Such reports underscore the critical importance of observing the microstructure of segmented KBs. Further, the distribution of relevant dislocations to KBs, as well as the movement and alteration of dislocations and KBs after heat treatments, have also become highly significant issues in understanding the properties of Mg–TM–RE alloys. However, there is currently a lack of detailed reports on the spatial arrangement of SESFs and non-basal dislocations within such KBs. Dislocation evolutions at KBs before and after heat treatments for hot-extruded Mg–TM–RE alloys have also been rarely studied in detail.

This study focuses on the morphological changes of KBs and the distribution of non-basal dislocations in hot-extruded Mg–Zn–Y alloys before and after heat treatment, using three-dimensional (3D) electron microscopy. The 3D microstructures of SESFs and dislocation distributions before and after heat treatment will be discussed.

2. Experimental Procedure

2.1 Specimen preparation

The composition of Mg–TM–RE alloys used in this study was Mg₉₇Zn₁Y₂ (at%). Two types of specimens, before and after heat treatment, were prepared as the following procedure: A master alloy was prepared by induction melting of pure Mg (99.99 mass%), Zn (99.9 mass%) and Y (99.9 mass%) ingots in a carbon crucible. Subsequently, hot-extrusion was carried out at 623 K (specimens before heat treatment). The extrusion ratio was 10, and the extrusion speed was 2.5 mm/s.¹²⁾ For specimens after heat treatment, the hot-extruded specimens were further subjected to annealing at 573 K for 168 h in a glass tube under an argon atmosphere with a pressure below 3 × 10⁻³ Pa.⁹⁾ Both types of specimens were cut into square pieces with dimensions of 3 mm × 3 mm, and thin foil specimens for transmission electron microscopy (TEM) and scanning transmission electron microscopy (STEM) observations were prepared through mechanical polishing and standard argon ion milling (Gatan PIPSTM model 695).

2.2 Microstructural observation

In this study, a TEM/STEM apparatus, Titan Cubed G2 (Thermo Fisher Scientific Inc.), was operated at an accelerating voltage of 300 kV in the parallel beam STEM mode. The convergence semi-angle of the incident electron beam in this mode was 1.2 mrad. The small convergence angle allows a large depth of focus enough to ignore inevitable defocus in a field of field caused by specimen tilt.^17,18) Bright-field (BF) and weak-beam dark-field (WBDF) images were observed under two-beam and WBDF conditions, respectively. The transmitted electron detection range for the BF disc is 0–5 mrad. The atomic number-dependent Z-contrast images were obtained by the annular dark-field (ADF) detector ranging from 52–200 mrad.

For electron tomography (ET) observations, a high-angle triple-axis (HATA) specimen holder (Mel-build Co.) was used to adjust the crystal orientation and diffraction conditions on the specimen tilt axis.^19,20) The diffraction vector, g = 0002 or 0006, was aligned parallel to the specimen tilt axis. The tilt-series images were collected at intervals of 1–2° within a specimen tilt angle range of approximately ±30°. Subsequently, the tilt-series images were processed with the block-matching-and-three-dimensional filtering (BM3D) algorithm for denoising.²¹⁾ Because there is significant missing information about tilt-series images for the specimen tilt angle range of ±30–90°, the assessment of the 3D reconstruction results must be done with great care of the missing information artifact. Measurements and discussions on the ET results will focus solely on spatial distribution information, such as rotation angles and spacing of SESFs.

All electron microscopy images were acquired using Velox™ software (Thermo Fisher Scientific Inc.). Position alignment of the tilt-series images and 3D reconstruction using the simultaneous iterations reconstruction technique (SIRT) algorithm with 20 iterative operations were performed using Composer™ software (SYSTEM IN FRONTIER INC.). The Visualizer-evo™ software (SYSTEM IN FRONTIER INC.) and Avizo™ software (Thermo Fisher Scientific Inc.) were used to visualize 3D reconstruction results.

3. Results and Discussions

3.1 Hot-extruded microstructure before heat treatment

Figure 1(a) shows BF-STEM images and a corresponding electron diffraction pattern of the hot-extruded Mg₉₇Zn₁Y₂ alloy specimen before heat treatment. The two-beam condition, k = g(0002), was set to allow c′-dislocations to be visible.¹²⁾ The thin LPSO layers with different thicknesses are recognized in the α-Mg matrix. c′-dislocations are recognized in some regions of the thin LPSO, as indicated by the yellow arrows, while the other regions of the thin LPSO exhibit no c′-dislocation. Figures 1(b) and 1(c) show the magnified BF-STEM and ADF-STEM images, respectively, at the red-boxed region in Fig. 1(a). There are two LPSO regions, “LPSO-1” and “LPSO-2”. Only LPSO-1 is rich in c′-dislocations, and the c′-dislocations accompany distortions and thickness variations in the LPSO region, as indicated by the orange arrows and white dashed circles. The angular difference between the leftmost and rightmost layers in LPSO-1 is approximately 2°. This microstructural feature implies that the c′-dislocations induce local kinks in the thin LPSO region, coupled with small-angle crystal rotations. Additionally, the ADF-STEM images reveal the local segregation of solute atoms at the kinks in the LPSO regions. Figure 1(d) illustrates schematically the distribution of c′-dislocations, thin LPSO phase and solute atom segregation in the field of view.

Fig. 1

Microstructure of an α-Mg/LPSO region in hot-extruded Mg₉₇Zn₁Y₂ alloy before the heat treatment. (a) ADF-STEM image viewed from $[1\ \bar{2}\ 10]$ showing the location of c′-dislocations (arrowheads) and a corresponding diffraction pattern showing the two-beam condition, k = g(0002), (b) BF-STEM image of the white square area in (a) showing the change in LPSO layer thickness, (c) ADF-STEM image of the same square area indicating solute atom segregation, (d) schematic diagram of the distribution of c′-dislocations, thin LPSO layers, and solute atom segregation in “LPSO-1” denoted in (b), and (e) schematic 3D diagram of c′-dislocation arrays along the α-Mg/LPSO interfaces.

There have been some reports on the α-Mg/LPSO phase interface structure, indicating that Shockley partial dislocations with the same or opposite Burgers vectors periodically appear at the α-Mg/LPSO phase interface. Several models to describe the transitional region of the phase interface have been proposed.^22,23) There is also a report that the α-Mg/LPSO phase interface acts as an obstacle to the slip of both basal dislocations (a′-dislocations) and non-basal dislocations (c′- or a′ + c′-dislocations). As a result, c′-dislocations tend to accumulate on both sides of the LPSO layers during the dislocation slip process.^24–26) Similarly, the SESFs can also act as an obstacle to dislocation slip. Based on these previous reports, the observed distribution of c′-dislocations at the α-Mg/LPSO phase interface in Fig. 1(b) can be explained as follows: during the hot-extrusion process, the presence of thin LPSO regions hinders the slip of non-basal dislocations, causing a large number of c′-dislocations remaining at the phase interface. Simultaneously, the local thickness of the LPSO region changes with introducing c′-dislocations at the phase interface, as schematically shown in Fig. 1(e).

Figure 2(a) shows an ADF-STEM image at the KBs in the hot-extruded Mg₉₇Zn₁Y₂ alloy specimen before heat treatment. The LPSO layers are densely arranged and buckled in some regions. This microstructural feature implies that the α-Mg/LPSO crystal in this field of view is locally deformed. At the buckled LPSO region, segments of the KBs that do not arrange in line are observed in the α-Mg matrix, as indicated by the dashed lines. A close observation of Fig. 2(a) reveals that the KBs show arrangements of short bright-line segments aligning along [0001] and in contact with the neighboring LPSO regions. The short bright-line segments in the ADF-STEM image are SESFs containing partial a′-dislocations, as reported by Egusa et al.¹⁶⁾ It should also be noted that some of the SESFs extend from the KBs into the α-Mg matrix. These KBs can be categorized into two types: 1) k-SESFs extending from the KBs and 2) m-SESFs in the α-Mg matrix isolating from the KBs, as indicated by the orange and blue arrows in Fig. 2(a). In this field of approximately 2.5 µm², the width histograms of k-SESFs and m-SESFs are shown in Fig. 2(b). The average widths of k-SESFs and m-SESFs are about 80 nm and 45 nm, respectively. The number fractions of k-SESFs and m-SESFs are about 0.33 and 0.67, respectively.

Fig. 2

Microstructure of kink boundaries (KBs) and associating solute-enriched stacking faults (SESFs) in hot-extruded Mg₉₇Zn₁Y₂ alloy before the heat treatment. (a) ADF-STEM image viewed from $[1\bar{2}10]$, (b) width histograms of k-SESFs (extending from the KBs) and m-SESFs (in the α-Mg matrix isolating from the KBs), (c) ADF-STEM image after specimen tilt by 1.2° around $\text{y}[10\ \bar{1}\ 0]$ axis showing rectangular-shaped arrangements composed of two parallel k-SESFs, a sub-KB and a part of the KB, (d) enlarged view of the white square region in (c) showing small segments of SESFs as indicated by arrowheads, (e) BF-STEM image of the same square region showing c′-dislocations (arrowheads) and a corresponding electron diffraction pattern indicating the two-beam condition, (f) 3D reconstruction of the rectangle-shaped arrangement of SESFs formed from the KB, where the locations of c′-dislocations and SESF-broken zones are denoted, and (g) schematic diagram of the $\{ 10\bar{1}1\} \langle \bar{1}2\bar{1}0\rangle $ slip system in α-Mg.

Figure 2(c) shows an ADF-STEM image obtained by tilting the specimen around the y-axis by 1.2°, which gives information about the KB structure along the depth direction of the foil specimen. There are two couples of k-SESFs extending from a KB, and additional short KBs exist at the ends of these k-SESFs, referred to as sub-KBs. In other words, a KB, a couple of k-SESFs and a sub-KB form a rectangular region, as indicated in Fig. 2(c). Enlarged ADF- and BF-STEM views of the white boxed region in Fig. 2(c) are shown respectively in Figs. 2(d) and 2(e), where the two-beam condition, k = g(0002), is used to visualize c′-dislocations. The KBs and sub-KBs are composed of SESFs arranged in a periodic discrete array along [0001]. The arrangement of SESFs in the sub-KBs is noticeably denser than in the KBs. Similarly, although the k-SESFs extending from the KBs appear as a continuous line for the no specimen tilt in Fig. 2(a), specimen tilt in Fig. 2(d) reveals that the k-SESFs are a series of discontinuous, elongated fragment arrays.

Furthermore, the BF-STEM image in Fig. 2(e) exhibits c′-dislocations at the edge of each k-SESF segment, and the series of c′-dislocations form a dislocation wall. This k-SESF structure closely resembles nano-cluster arrays at a tilt boundary of the KBs reported by Shao et al.²⁷⁾ They pointed out that these nano-cluster arrays are formed by the segregation of solute atoms, Zn/RE, and are divided into multiple nanosized fragments by c′-dislocations to form small angle tilt boundaries, resulting in small-angle crystal rotations near the KBs. Therefore, if the observed k-SESF structure is considered a tilted boundary, the SESF fragments array enriched in Zn/RE solute atoms may effectively alleviate stress concentration around the kink-deformed area.²⁸⁾

To further analyze the spatial structure and distribution of the SESFs forming the KBs, sub-KBs and k-SESFs, ET observations were conducted under the ADF-STEM mode. The 3D reconstructed volume is shown in Fig. 2(f). The SESFs in the KBs and sub-KBs are flat and elongated thin sheets in shapes and neatly arranged along [0001]. The density of the SESFs in the sub-KBs is notably higher than in the KBs. Since the SESFs formed in the α-Mg matrix contain a′-dislocations, and the a′-dislocations are interpreted as GN dislocations, giving small-angle lattice rotations at the KBs.¹²⁾ Based on this fact, the SESF density difference between the KBs and the sub-KBs can be related to the lattice rotation difference, as follows. From the 3D reconstructed volume in Fig. 2(f), the average distances between the neighboring SESFs were measured to be 7.5 nm in the KBs and 4.4 nm in the sub-KBs, respectively. Suppose one assumes the KBs with a rotation angle θ as a symmetric grain boundary composed of a periodic arrangement of GN dislocations. In that case, the relationship between the GN dislocation spacing $\bar{D}$ and Burgers vector b is described as follows:¹²⁾

  

\begin{equation*} \tan \theta = |\boldsymbol{b}|/\bar{D}. \end{equation*}

From the lattice constant of Mg, a = 0.32 nm, here we use |b| = 0.32 nm, assuming $\boldsymbol{b} = 1/3\langle 1\bar{2}10\rangle $. Hence, the calculated θ values are 2.4° at the KBs and 4.2° at the sub-KBs. These calculated θ values show a qualitative agreement with the measured θ values 3D reconstructed volume: 2.0° at the KBs and 3.8° at the sub-KBs, respectively. This result indicates that as the crystal rotation angle increases at KBs, the GN dislocation density and SESF density also increase. Therefore, the higher density of the SESFs in the sub-KBs compared to the KBs can be explained by the fact that the crystal rotation angle at the sub-KBs is larger than at the KBs, leading to more GN dislocations at the sub-KBs, thereby generating more SESFs and further promoting the relaxation of local elastic energy accumulation caused by crystal rotation. Furthermore, compared to the KBs, the larger crystal rotation angle at the sub-KBs implies a higher local stress concentration.

The 3D reconstructed image in Fig. 2(f) also indicates that the SESFs in sub-KBs were broken along the slip plane $\{ 10\bar{1}1\} $ as indicated by the white broken lines. The angle between slip plane $\{ 10\bar{1}1\} $ and basal plane {0001} is approximately 62°, as shown in Fig. 2(g). If one assumes that the slip system, $\{ 10\bar{1}1\} \langle \bar{1}2\bar{1}0\rangle $, can activate under shear stress, these SESFs-broken zones would indicate a trace of shear deformation.²⁹⁾ It was also noted that the density of the SESF-broken zones at sub-KBs is higher than that in KBs.

As for the k-SESFs extending from KBs, the ET observation revealed that the k-SESFs are composed of nanosized fragments of SESFs extending along the specimen’s depth direction, and the average spacing of the nanosized SESF fragments is measured to be 3.4 nm. Because the c′-dislocation contrast was recognized along the nanosized SESF fragments in Fig. 2(e), it can be interpreted that the arrangement of the nanosized SESF fragments along $[10\bar{1}0]$ was formed by the introduction of c′-dislocations which segmented the solute Zn/RE atom segregation layers into their fragments.

The rectangle-like arrangement of SESFs and k-SESFs in Fig. 2 may have formed during the hot extrusion process. If the rectangle-like SESF arrangement is considered a hard region in the α-Mg matrix like the LPSO phase region,³⁰⁾ the c′-dislocations wall formed within the k-SESFs would be less likely to slip. Therefore, the Mg–TM–RE alloy would be more inclined to form kinks than twins. This interpretation could correspond to one of the reasons for the kink strengthening in Mg–TM–RE alloys.

3.2 Microstructure after heat treatment at 573 K for 168 h

Figure 3(a) shows an ADF-STEM image of the hot-extruded Mg₉₇Zn₁Y₂ alloy specimen after heat treatment. In this field of view of approximately 2.5 µm², the width histograms of k-SESFs and m-SESFs were measured, as shown in Fig. 3(b). The average widths of k-SESFs and m-SESFs are 48 nm and 25 nm, respectively. Compared to the measurement result before heat treatment in Fig. 2(b), the average widths of k-SESFs and m-SESFs decreased by 40% and 44%, respectively, after the heat treatment. The number fractions of k-SESFs and m-SESFs are 0.37 and 0.63, which have increased by 4% and decreased by 4% after the feat treatment. This opposite tendency could be interpreted as follows. The solute atoms Zn/RE tend to segregate at KBs during the hot-extrusion process, leading to lower nucleation energy for SESFs at the KBs than for those in the α-Mg matrix.¹⁶⁾ The subsequent heat treatment consumes the solute atoms at the KBs and promotes the growth of the k-SESFs into the α-Mg matrix. The m-SESFs in the α-Mg matrix would be consumed by the growth of the SESFs at the KBs and k-SESFs during the heat treatment like Ostwald ripening.

Fig. 3

Microstructure of KBs and associating SESFs in hot-extruded Mg₉₇Zn₁Y₂ alloy after the heat treatment at 573 K for 168 h. (a) ADF-STEM image viewed from $[1\bar{2}10]$ showing the step-shaped KBs, (b) width histograms of k-SESFs and m-SESFs, (c) BF-STEM image of the white square region in (a) showing the existence of c′-dislocations (arrowheads) and a corresponding electron diffraction pattern indicating the two-beam condition, (d) enlarged ADF-STEM image of the k-SESFs with and without c′-dislocations denoted with the yellow and red rectangles respectively in (c) without specimen tilt (0°), (e) ADF-STEM image after specimen tilt by 20° around x[0001] axis showing the continuous (red) and discontinuous (yellow) morphologies of the k-SESFs. (f) WBDF-STEM image in the same field of view for (e) showing c′-dislocations (arrowheads) and the corresponding electron diffraction pattern, and (g) superposition of the binarized images of (e) and (f) showing the locations of the SESF segments (white) and the c′-dislocations (red).

As shown in Fig. 3(a), the parallel arrangements of SESFs along [0001] to form KBs were preserved after the heat treatment. However, rectangular-shaped arrangements of a KB segment, a sub-KB and two k-SESFs, as shown in Fig. 2(c), were not conspicuously found. Instead, a unique step-shaped morphology of KBs was often recognized, as indicated by the broken lines in Fig. 3(a). The white boxed region in Fig. 3(a) was closely observed in the BF-STEM mode in Fig. 3(c), where the two-beam condition, k = g(0002), was set to visualize c′-dislocations. The c′-dislocations aggregate at the step-shaped KBs and some of the k-SESFs, as indicated by the yellow arrowheads in Fig. 3(c). We will discuss the relationships between the c′-dislocations, the k-SESFs and the step-shaped KBs in the following sections.

For observing the differences between k-SESFs with c′-dislocations and those without c′-dislocations, the specimen was tilted around [0001]. Figures 3(d) and 3(e) show ADF-STEM images of the two k-SESFs with and without c′-dislocations denoted with yellow and red rectangles respectively in Fig. 3(c), and the images were acquired respectively at 0° (d) and 20° (e) in specimen tilt angles. The ADF-STEM image contrast of the k-SESF with c′-dislocations changed from a continuous line in Fig. 3(d) to discontinuous segments of short lines after the specimen tilt by 20° in Fig. 3(e). On the other hand, the ADF-STEM image contrast of the k-SESF without c′-dislocation remained unchanged after the specimen tilt. To visualize the c′-dislocations at the k-SESF, a WBDF-STEM image in the same field of view was obtained under the WB diffraction condition, g/3g, g = 0002, as shown in Fig. 3(f). The c′-dislocations are recognized as bright dots, as pointed by yellow arrowheads. The ADF- and WBDF-STEM images in the yellow rectangle region in Figs. 3(e) and 3(f) were binarized and superimposed with each other, as shown in Fig. 3(g). The c′-dislocations are located at the apex of each segment of the k-SESF. This microstructure of the k-SESF with c′-dislocations is similar to that observed before the heat treatment shown in Fig. 2.

Figure 4(a) shows an ADF-STEM image of a step-shaped KB, and the white square region is enlarged in Fig. 4(b). The array of SESFs along [0001] in the KB starts to sequentially shift to $[10\bar{1}0]$ at the segment points denoted in Fig. 4(b). The four SESFs shift to $[10\bar{1}0]$, and the next SESFs again align along [0001]. The total shift of the SESFs along $[10\bar{1}0]$ is 17 nm. A WBDF-STEM image of the same field of view in Fig. 4(c) was binarized and superimposed onto Fig. 4(b), as demonstrated in Fig. 4(d). c′-dislocations are located at the apex of the SESFs where the shift along $[10\bar{1}0]$ occurs, indicating that the c′-dislocations form the step-shaped KB. This step-shaped KB is a common characteristic of the specimen after the heat treatment. In the step-shaped KBs structure, the c′-dislocations exist only at the segment points to prevent the occurrence of microcracks.¹²⁾

Fig. 4

Microstructure of a step-shaped KB in hot-extruded Mg₉₇Zn₁Y₂ alloy after the heat treatment. (a) ADF-STEM image viewed from $[1\bar{2}10]$ showing the step-shaped KBs, (b) enlarged ADF-STEM image in the white square region in (a), (c) WBDF-STEM image in the same field of view for (b) showing c′-dislocations and the corresponding electron diffraction pattern showing the following condition: g/3g, g = 0002, and (d) superposition of the binarized images of (b) and (c) showing the locations of the SESF segments (white) and the c′-dislocations (red).

The step-shaped KB structure observed after the heat treatment is likely related to the rectangular-shaped structure composed of a KB, a sub-KB and two k-SESFs observed before the heat treatment. Thus, the former structure is likely to have transformed from the latter structure during the heat treatment. Since the resistance to dislocation slip in the SESFs decreases with increasing temperature,⁹⁾ dislocation motion occurs during the annealing process, leading to the dissociation of k-SESFs and SESFs in sub-KBs. Simultaneously, with the redistribution of SESFs at high temperatures, original KBs transform into step-shaped KBs by introducing c′-dislocations, which may be a stress relaxation process.

Another example of the step-shaped KBs is shown in Fig. 5(a). A KB in the upper field of view is composed of three segments with two segment points, as indicated by the arrows in Fig. 5(a). The LPSO region lies across the KB and shows crystal rotation about 4.2° in the $xy(1\bar{2}10)$ plane. The white square region in Fig. 5(a) was observed under the BF-STEM and ADF-STEM modes, as shown respectively in Figs. 5(b) and 5(c), where, in the BF-STEM image, the two-beam condition, k = g(0006), was set to visualize c′-dislocations. The BF-STEM image, Fig. 5(b), clearly shows that c′-dislocations aggregate at the segment points of the step-shaped KB. In the ADF-STEM image, Fig. 5(c), the three parts of the KB are denoted as Segment 1, Segment 2 and Segment 3. A close observation reveals that Segment 2 can be divided into upper and lower parts, and they are distorted in different directions. In Segment 2, there are 18 layers of the SESFs in the upper part and 16 layers in the lower part. Such disordered arrays and morphology of the SESFs at KBs were frequently observed after the heat treatment.

Fig. 5

3D observation of a step-shaped KB in hot-extruded Mg₉₇Zn₁Y₂ alloy after the heat treatment. (a) ADF-STEM image viewed from $[1\bar{2}10]$ showing the step-shaped KBs, (b) BF-STEM image of the white square region in (a) showing c′-dislocations at the step regions and a corresponding electron diffraction pattern, (c) ADF-STEM image of the same field of view for (b) showing the three segments 1, 2 and 3 in the step-shaped KB, (d), (e) and (f) parts of ADF-STEM tilt-series images at specimen tilt angles of 4°, 6° and 7° respectively, (g) 3D reconstruction of the step-shaped KB denoting the locations of the c′-dislocations, and (h) schematic diagram of lattice rotation around the [0001] axis in each segment with different angles.

Figures 5(d)–(f) show parts of ADF-STEM tilt-series of the step-shaped KB, and Fig. 5(g) shows a 3D reconstruction result of the step-shaped KBs using the ET algorithm, where the locations of the c′-dislocations at the segment points are denoted. In Fig. 5(d), with specimen tilt by 4° around the c[0001] axis, the SESFs in Segment 3 are recognized as a bright-dot array, suggesting that the incident electron beam direction is parallel to the length direction of the SESFs. Similarly, in Figs. 5(e) and 5(f) with specimen tilt by 6° and 7°, respectively, the SESFs in Segments 1 and 2 appear as bright-dot arrays. The observed lattice rotations around the c-axis at the step-shaped KB are illustrated in Fig. 5(h). The tilt-series observation of the SESFs indicates that the α-Mg matrix containing the step-shaped KB is locally rotated around the c-axis with different angles, as well as the lattice rotation around the a- and b-axes of α-Mg.

Such lattice rotations in the multiple crystal axes are interpreted as follows: The crystal rotation around the a- and b-axes is a fundamental characteristic of the KBs in Mg–Zn–RE alloys. The original flat-shaped KBs transformed to step-shaped KBs during the heat treatment, followed by the movement of a′-dislocations and the formation of c′-dislocations. This microstructural change by the heat treatment mitigated the stress concentration due to crystal rotation at the KBs. It maintained the crystal integrity by introducing c′-dislocations at the segment points.⁹⁾ The crystal rotation angle around the c-axis varied for each segment of the step-shaped KBs. This fact may be attributed to the insertion of different numbers of SESFs into the original SESF arrays during the heat treatment, as seen in Fig. 5(c). As these segmented KBs remained after the heat treatment, one can assume that the lattice rotation around the c-axis induced by the insertion of additional SESFs assists in alleviating the stress concentration at the KBs.

The specimens before the heat treatment exhibited discontinuous arrangements of the SESFs within the KBs along the specimen-depth direction, denoted as SESF-broken zones in Fig. 2(c). This microstructural feature implies that a′-dislocations in the SESFs terminate within the specimen. After the heat treatment, the SESF-broken zones in the SESF layer arrays were not observed. This fact suggests that during the heat treatment, there was a redistribution of SESF layers, resulting in the disappearance of the SESF-broken zones. Recent studies have indicated that this configuration of SESFs may be related to the dislocation-disclination strengthening mechanism.³¹⁾

4. Conclusion

The microstructure and its evolution of SESFs associated with KBs were observed in the hot-extruded Mg–Zn–RE alloy before and after the heat treatment at 573 K for 168 h. 3D views and their models representing the spatial distribution of SESFs and c′-dislocations were constructed based on ET observations. Before the heat treatment, the original planar KBs were primarily composed of parallel arrays of SESFs. Some parts of the KBs formed rectangle-shaped arrangements of the SESFs in the α-Mg matrix, where c′-dislocation walls were recognized. Following the heat treatment, the original KBs transformed into step-shaped KBs. c′-dislocations existed at the step parts of the KBs. The step-shaped KBs produced lattice rotations along not only the a- and b-axes but also the c-axis in the α-Mg matrix by introducing additional segments of SESFs in the KBs. This lattice rotation behavior along the multiple axes is believed to mitigate local stress concentrations at the KB by redistributing SESFs during the heat treatment.

Acknowledgments

We thank Associate Professor Masatoshi Mitsuhara at Kyushu University for stimulating discussions. This work was supported by JST SPRING (Grant Numbers JPMJSP2136) and JSPS KAKENHI Grant Number (JP18H05479), Japan.

REFERENCES

• 1)   Y.  Kawamura and  M.  Yamasaki: Mater. Trans. 48 (2007) 2986–2992. doi:10.2320/matertrans.MER2007142
• 2)   E.  Abe,  Y.  Kawamura,  K.  Hayashi and  A.  Inoue: Acta Mater. 50 (2002) 3845–3857. doi:10.1016/S1359-6454(02)00191-X
• 3)   A.  Ono,  E.  Abe,  T.  Itoi,  M.  Hirohashi,  M.  Yamasaki and  Y.  Kawamura: Mater. Trans. 49 (2008) 990–994. doi:10.2320/matertrans.MC200763
• 4)   E.  Abe,  A.  Ono,  T.  Itoi,  M.  Yamasaki and  Y.  Kawamura: Philos. Mag. Lett. 91 (2011) 690–696. doi:10.1080/09500839.2011.609149
• 5)   D.  Egusa and  E.  Abe: Acta Mater. 60 (2012) 166–178. doi:10.1016/j.actamat.2011.09.030
• 6)   D.  Egusa,  H.  Somekawa and  E.  Abe: Mater. Trans. 61 (2020) 833–838. doi:10.2320/matertrans.MT-MM2019010
• 7)   M.  Egami,  I.  Ohnuma,  M.  Enoki,  H.  Ohtani and  E.  Abe: Mater. Des. 188 (2020) 108452. doi:10.1016/j.matdes.2019.108452
• 8)   M.  Egami,  I.  Ohnuma,  M.  Enoki,  H.  Ohtani and  E.  Abe: Mater. Trans. 61 (2020) 839–848. doi:10.2320/matertrans.MT-MM2019009
• 9)   D.  Egusa,  K.  Inoue,  Y.  Nagai and  E.  Abe: Mater. Charact. 177 (2021) 111153. doi:10.1016/j.matchar.2021.111153
• 10)   K.  Hagihara,  A.  Kinoshita,  Y.  Sugino,  M.  Yamasaki,  Y.  Kawamura,  H.Y.  Yasuda and  Y.  Umakoshi: Acta Mater. 58 (2010) 6282–6293. doi:10.1016/j.actamat.2010.07.050
• 11)   X.H.  Shao,  Z.Q.  Yang and  X.L.  Ma: Acta Mater. 58 (2010) 4760–4771. doi:10.1016/j.actamat.2010.05.012
• 12)   D.  Egusa,  M.  Yamasaki,  Y.  Kawamura and  E.  Abe: Mater. Trans. 54 (2013) 698–702. doi:10.2320/matertrans.MI201216
• 13)   T.  Matsumoto,  M.  Yamasaki,  K.  Hagihara and  Y.  Kawamura: Acta Mater. 151 (2018) 112–124. doi:10.1016/j.actamat.2018.03.034
• 14)   K.  Hagihara,  Z.  Li,  M.  Yamasaki,  Y.  Kawamura and  T.  Nakano: Acta Mater. 163 (2019) 226–239. doi:10.1016/j.actamat.2018.10.016
• 15)   K.  Hagihara,  R.  Ueyama,  M.  Yamasaki,  Y.  Kawamura and  T.  Nakano: Acta Mater. 209 (2021) 116797. doi:10.1016/j.actamat.2021.116797
• 16)   D.  Egusa,  R.  Manabe,  T.  Kawasaki,  S.  Harjo,  S.  Sato and  E.  Abe: Mater. Today Commun. 31 (2022) 103344. doi:10.1016/j.mtcomm.2022.103344
• 17)   Y.F.  Zhao,  S.  Koike,  R.  Nakama,  S.  Ihara,  M.  Mitsuhara,  M.  Murayama,  S.  Hata and  H.  Saito: Sci. Rep. 11 (2021) 20720. doi:10.1038/s41598-021-99914-5
• 18)   E.G.T.  Bosch and  I.  Lazić: Ultramicroscopy 207 (2019) 112831. doi:10.1016/j.ultramic.2019.112831
• 19)   S.  Hata,  H.  Miyazaki,  S.  Miyazaki,  M.  Mitsuhara,  M.  Tanaka,  K.  Kaneko,  K.  Higashida,  K.  Ikeda,  H.  Nakashima,  S.  Matsumura,  J.S.  Barnard,  J.H.  Sharp and  P.A.  Midgley: Ultramicroscopy 111 (2011) 1168–1175. doi:10.1016/j.ultramic.2011.03.021
• 20)   J.S.  Barnard,  J.  Sharp,  J.R.  Tong and  P.A.  Midgley: J. Phys. Conf. Ser. 26 (2006) 247. doi:10.1088/1742-6596/26/1/059
• 21)   Y.  Mäkinen,  L.  Azzari and  A.  Foi: IEEE Trans. Image Process. 29 (2020) 8339–8354. doi:10.1109/TIP.2020.3014721
• 22)   Q.Q.  Jin,  X.H.  Shao,  S.J.  Zheng,  Y.T.  Zhou,  B.  Zhang and  X.L.  Ma: J. Mater. Sci. Technol. 61 (2021) 114–118. doi:10.1016/j.jmst.2020.05.045
• 23)   T.  Mayama,  S.R.  Agnew,  K.  Hagihara,  K.  Kamura,  K.  Shiraishi,  M.  Yamasaki and  Y.  Kawamura: Int. J. Plast. 154 (2022) 103294. doi:10.1016/j.ijplas.2022.103294
• 24)   K.  Hiraga,  A.  Yasuhara and  K.  Saito: Mater. Trans. 53 (2012) 1385–1390. doi:10.2320/matertrans.M2012115
• 25)   J.  Wang,  G.  Zhu,  L.  Wang,  X.  Zhang,  M.  Knezevic and  X.  Zeng: Mater. Charact. 203 (2023) 113066. doi:10.1016/j.matchar.2023.113066
• 26)   Y.M.  Zhu,  A.J.  Morton and  J.F.  Nie: Acta Mater. 60 (2012) 6562–6572. doi:10.1016/j.actamat.2012.08.022
• 27)   X.H.  Shao,  Q.Q.  Jin,  Y.T.  Zhou,  H.J.  Yang,  S.J.  Zheng,  B.  Zhang,  Q.  Chen and  X.L.  Ma: Materialia 6 (2019) 100287. doi:10.1016/j.mtla.2019.100287
• 28)   A.  Datta,  U.V.  Waghmare and  U.  Ramamurty: Acta Mater. 56 (2008) 2531–2539. doi:10.1016/j.actamat.2008.01.046
• 29)   X.H.  Shao,  Q.Q.  Jin,  Y.T.  Zhou,  H.J.  Yang,  S.J.  Zheng,  B.  Zhang,  Q.  Chen and  X.L.  Ma: Mater. Sci. Eng. A 779 (2020) 139109. doi:10.1016/j.msea.2020.139109
• 30)   Y.  Kawamura,  K.  Hayashi and  A.  Inoue: Mater. Trans. 42 (2001) 1172–1176. doi:10.2320/matertrans.42.1172
• 31)   T.  Inamura: Acta Mater. 173 (2019) 270–280. doi:10.1016/j.actamat.2019.05.023