Formation Process of Long-Period Stacking-Ordered Structures in Mg₉₇Zn₁Y₂ Alloy Comprising HCP and Cubic Phases Fabricated by High-Pressure High-Temperature Annealing

Abstract

As-cast Mg₉₇Zn₁Y₂ alloy consists of α-Mg matrix and 18R-type long-period stacking-ordered (LPSO) structures. We observed that the alloy undergoes a phase transformation to D0₃ superlattices and α-Mg matrix due to high-pressure high-temperature (HPHT) annealing at 3 GPa and above 773 K. Further, the alloy recovered after HPHT annealing, consisting of the α-Mg matrix and D0₃ superlattices, transformed into 18R-type LPSO structures during further annealing at ambient pressure. An fcc structure with a lattice parameter of 1.42 nm, which was twice that of D0₃, emerged in both the collapse process of the 18R-type LPSO structure under high-pressure, and the formation process of the 18R-type LPSO structure at ambient pressure. This fcc phase was an intermediate structure between 18R-type LPSO and D0₃. From the electron diffraction results, it is considered that 18R-type LPSO is continuously present with 2H including stacking faults, which almost corresponded with previous studies.

1. Introduction

Long-period stacking-ordered (LPSO) structures in magnesium (Mg) ternary alloys consisting of rare earth (RE) and another metal (M) have piqued interest as a potential strengthening approach [1,2,3]. The strength of Mg alloys consisting of α-Mg matrix and LPSO structures is greatly increased when the LPSO structure is kinked by plastic deformation. The LPSO structures in Mg–M–RE alloys have been studied using high-angle annular dark-field scanning transmission electron microscopy (HAADEF-STEM), and then the crystal structures of some kinds of LPSO were determined [4,5,6]. In these LPSO structures are characterized stacking faults (SFs) inserted periodically in ABAB stacking, and L1₂-type clusters of M₆RE₈ in the SFs [5,6]. Further, an interstitial Mg atom exists in a body center of the L1₂-type cluster, which makes the LPSO energetically stable [7].

The formation process of LPSO structures is interesting from the viewpoint of the physics of metals and for development of materials strengthened by kinked bands. Therefore, LPSO structure formation processes have been investigated on both experimental and computational methods [8,9,10,11,12,13,14,15,16,17].

Our group studied the pressure-induced phase transition of the 18R-type LPSO in Mg–Zn–Y ternary alloys. Mg₈₅Zn₆Y₉ possessed 18R-type LPSO monophase at ambient pressure and is in contact with a liquid phase at high temperatures [18]. However, by in situ X-ray diffraction (XRD) at 3 GPa, 18R-type LPSO structures in Mg₈₅Zn₆Y₉ transform into the α-Mg and D0₃ phases with an increase in temperature [19]. The alloy recovered after being subjected to 7 GPa at 973 K revealed a fine dual-phase structure composed of D0₃ and α-Mg (D0₃/α-Mg) [20]. The D0₃/hcp structure in Mg₈₅Zn₆Y₉ transformed into the 18R-type LPSO when heated at ambient pressure [13]. In situ XRD revealed that the lattice volume of α-Mg(2H) initially increased with an increase in the temperature due to the intrusion of Y emitted from D0₃ into the 2H lattice. After that, the 2H structure collapsed due to the formation of random SFs, causing the formation of 18R-type LPSO structures. Further, first-principles calculation explained that the formation process of 18R-type LPSO structure resulted from two factors: the thermodynamic instability of the 2H stacking and the segregation of Y and Zn into the SF layer [9,13].

In this paper, first, we reported the effect of pressure on the Mg₉₇Zn₁Y₂ alloy with the α-Mg matrix and an 18R-type LPSO structure. The 18R-type LPSO was decomposed to D0₃/hcp by high-pressure high-temperature (HPHT) treatments. Then, we reported the process of the 18R-type LPSO phase formation from D0₃/hcp fabricated by the HPHT-treatment.

2. Materials and Methods

Mg₉₇Zn₁Y₂ alloy ingot was made by casting, and an inductively coupled plasma optical emission spectrometer (Shimazu ICPS-8100, Kyoto, Japan) was used to ascertain its chemical composition. The alloy was cut into a columnar shape of 2 mm in diameter and 5 mm in length for high-pressure experiments. The high-pressure condition was generated by the multi-anvil high-pressure apparatus (CTF Co., Tokyo, Japan). MgO was used as the octahedron high-pressure cells and sample capsules, whereas graphite and tungsten carbide was used as a heater and secondary anvil, respectively. The annealing of the alloys after HPHT treatment was realized in an evacuated silica tube and subsequently quenched in water. XRD was conducted using Cu-Kα radiation (Rigaku., RAPIDII-V/DW, Japan). The X-ray incident angle was fixed at 2°, and the diffracted pattern was detected using an imaging plate. The microstructures were observed using a backscattered electron (BE) image of scanning electron microscopy (SEM: JEOL, JSM-7000F, Japan). Electron diffraction (ED) was observed via transmission electron microscopy (TEM: JOEL, JEM-2010, Japan). Sample preparation for TEM was realized by a focused ion beam (FIB; FEI, SCIOS, USA). XRD and ED results were analyzed using simulation software CrystalDiffract6 (CrystalMaker Software Limited., Begbroke, UK) and ReciPro ver.4 (open source software, Tokyo, Japan), respectively.

3. Results and Discussion

Figure 1 shows the XRD patterns of the Mg₉₇Zn₁Y₂ cast alloy and the alloy recovered after being annealed at 3 GPa for 2 h at various temperatures. The XRD peaks of the cast alloy may be attributed to the α-Mg, 18R-type LPSO, and some impurities (Y, and Y₂O₃). For the alloy annealed at 3 GPa and 773 K, however, certain peaks (designated as E₁, E₂, and E₃) were confirmed. The peaks almost corresponded to that of the E phase with a large cubic lattice, reported previously in the Mg₈₅Zn₆Y₉ alloy [13]. Only the peaks from the hcp and high-pressure phase (D0₃) were obtained by raising the annealing temperature beyond 823 K, whereas peaks from the 18R-type LPSO and E phase disappeared completely. The crystal information of D0₃ estimated from XRD intensity, which is similar to that the HPHT phase of Mg₇₅.₅Zn₉.₅Y₁₃.₈ [21]. These results indicated that the transformation from 18R-type LPSO/α-Mg to D0₃/α-Mg occurs at 3 GPa with an increase in temperature, and the phase suggested by E₁, E₂, and E₃ peaks existed as an intermediate phase during the phase transformation.

Figure 2a,b shows the compositional images in the BE mode of SEM for the Mg₉₇Zn₁Y₂ cast alloy and the alloy recovered after being annealed at 3 GPa and 823 K, respectively. Figure 2a shows the bright 18R-type LPSO phase and the boundary of the dark α-Mg matrix. The 18R-type LPSO surrounds α-Mg, which is characteristic of LPSO microstructures and agrees with previous reports [22]. Figure 2b shows that the thin D0₃ phase exists around the matrix. The volume fraction of the D0₃ phase is smaller than that of the 18R-type LPSO phase shown in Figure 2a. This result is plausible to insinuate that the atomic density of the D0₃ structure is higher than that of the 18R-type LPSO structure. In addition, the concentration of Y and Zn are higher than those of the 18R-type LPSO as shown in

Table 1. Crystal information of high-pressure phase (D0₃) estimated by XRD. Chemical composition: Mg₆₃Zn_16.5Y_20.5, Space group: $Fm\overline{3}m$, Lattice constant: 0.71 nm.

Element	X	Y	Z	Site Occupancy
Mg	0.50	0.50	0.50	1
Mg	0.25	0.25	0.25	0.67
Mg	0.00	0.00	0.00	0.18
Y	0.00	0.00	0.00	0.82
Zn	0.25	0.25	0.25	0.33

.

Subsequently, the process of the 18R-type LPSO structure formation from D0₃/α-Mg at ambient pressure was investigated. The Mg₉₇Zn₁Y₂ alloy recovered after being annealed at 3 GPa and 823 K for 2 h was made the starting material. XRD, SEM, and TEM were used to analyze the crystal structure and microstructural variations of the alloy caused by further annealing for 10 min at ambient pressure. The XRD patterns of the alloy annealed at various temperatures are shown in Figure 3. In the XRD pattern of the alloy annealed at 623 K, (0003)_18R, E₁, E₂, and E₃ peaks emerged. Further, at 643 K, a new peak indicated as ◇ emerged. Peaks labeled E₁, E₂, and E₃ disappeared in the XRD pattern of the alloy annealed at 673 K. Finally, the peaks from D0₃ disappeared, and the XRD patterns of the alloy annealed at 723 K completely corresponded to the as-cast Mg₉₇Zn₁Y₂ alloy.

Figure 4 shows the compositional images in the BE mode of SEM for alloys corresponding to the XRD patterns in Figure 3. Fine-threads bright phase precipitates were observed in the alloy annealed at 623 K in Figure 4a, which is identical to that in Figure 2b. XRD peaks corresponding to 18R-type LPSO structure were confirmed in the alloy, however, the characteristic microstructure of LPSO shown in Figure 2a, was not observed. The bright phase gets thicker in the alloy annealed at 643 K, as shown in Figure 4b. The microstructure in Figure 4c approaches that of the LPSO phase in Figure 2a when the temperature rises.

To investigate the crystal structure transformation, ED patterns were collected using TEM from the alloy annealed at 643 K. Figure 5a shows a bright-field image of the sample and the ED patterns shown in Figure 5b–d corresponded to the position B, C, and D in Figure 5a, respectively. The ED pattern in Figure 5b can be assigned to the 18R-type LPSO structure model in Reference [5] The (00018)_18R refractions, which were consistent with (0002)_2H, were divided into five refractions. In the ED patterns in Figure 5c, (0001)_2H refraction became clearer, however, the other four refractions became unclear. Further, in Figure 5d, only (0002)_2H and (0001)_2H refractions remained, whereas other refractions became streaks. The c-axis of three ED patterns in Figure 5b–d is the co-axis.

Here, the structure indicated by the ED pattern of Figure 5d will be discussed. Most refractions in Figure 5d can be assigned to the 2H-Mg, however, some refractions that have to disappear due to the extinction rule, such as 0001 of 2H, can be confirmed. Further, there are strong streaks along c-axis direction. Moreover, in the bright field image in Figure 5e, many thin lines perpendicular to the c-axis are confirmed. According to these results, ABAB stacking on the c-axis are disturbed by the plain defect perpendiculer to the c-axis. Therefore, it is considered that SFs may disturb the ABAB stacking of the c-axis.

In previous studies, similar ED patterns as in Figure 5d were reported [8,15,22]. In the case of Mg₉₇Zn₁Gd₂, the alloy forms 14H-type LPSO with an increase in the annealing temperature [8]. The similar ED pattern and bright field image to Figure 5d,e were observed in the alloy at a lower temperature than when the 14H-type LPSO emerged. They concluded the 2H structure including a lot of SFs emerge before 14H-type LPSO formation [8]. Furthermore, in the case of Mg₉₇Zn₁Y₂ annealed at 573 K for 120 h, a similar ED pattern to Figure 5d was observed in the α-Mg matrix [22]. This result also explained the ED pattern caused by 2H+SFs. A recent study for Mg₉₇Zn₁Y₂ also reported a similar ED pattern to Figure 5d [15]. Corresponding HAADEF-STEM images indicated a lamellae structure consisting of a dark α-Mg phase and thin bright line. In the paper, the thin bright lines were expressed as “thin LPSO”. Furthermore, recent atomic resolution HAADEF-STEM studied for the Mg–Zn–Gd and Mg–Zn–Y ternary alloy revealed the solute atoms enriched in SF before growing as an LPSO [14,16]. Considering thickness, the “thin LPSO” in Reference [15] would be corresponding to the solute enriched SF (SESF) reported in [16]. According to the Figure 5 and previous results [8,15,16,22], random SF emerged before LPSO formation. Taking these reports into account, we consider the solute atoms concentrate in SF at almost the same time, which is expected by theoretical calculations [10,13,16]. This will be a formation process where the solute concentration is lower than the ideal composition of 18R-type LPSO.

From here, the intermediate phase, the E phase, is reported. We discovered ED patterns corresponding to the E phase [13], which is double the D0₃ lattice constant (Figure 6). Figure 6a,b shows ED patterns for the same parts observed by different incident beam directions. In Figure 6, spots were assigned using the D0₃ lattice with a lattice constant of 0.71 nm. D0₃ refractions divided a spot (○), i.e., the ED patterns indicated the existence of cubic structure with a lattice constant twice that of D0₃, which corresponded to the lattice size of the E phase. Based on these facts, the XRD peaks of E₁, E₂, E₃, ◇ and those of D0₃ were indexed by a E phase with a lattice constant of 1.42 nm (

Table 2. XRD peaks list corresponding to E phase with a lattice constant of 1.426 nm. The XRD pattern is the alloy annealed at 643 K in Figure 3.

Peaks	2 Theta	d-Value (nm)	Hkl	Difference (%)
E1	10.7	0.8258	(111)	0.1493
Labeled ◇	17.4	0.5089	(220)	0.0924
E2	20.7	0.4286	(113)	0.0290
(111) of D03	21.6	0.4109	(222)	0.0147
(200) of D03	25	0.3558	(400)	0.0130
E3	27.3	0.3263	(133)	0.0139

). The E phase is considered as fcc lattice from the extinction rule of XRD. These results indicate the E phase exists in double the number of D0₃ lattices as an intermediate state between D0₃ and LPSO.

Next, we report on the process of the E phase collapse. A weak contrast change in the lamella pattern was observed in the alloy annealed at 643 K (Figure 7a). Lamellae parts were sectioned for TEM observation. The bright-field image and the ED pattern corresponding to the bright area indicated by an arrow are shown in Figure 7b. The ED pattern corresponded to α-Mg, where the incident beam was parallel to <0001>. The ED patterns obtained from the dark part indicated by C in Figure 7b are shown in Figure 7c. Many refractions exist, and the bright refractions can be analyzed by E phase with the incident beam direction parallel to [110]. The $\left[1\overline{1}1\right]$ and $\left[\overline{1}11\right]$ of fcc intersected at an angle of 109°. Enlarged views are given as an inset in (c). The obscure reflections indicated by yellow arrows, confirmed only among $\overline{1}11$ refractions, despite $\left[1\overline{1}1\right]$ and $\left[\overline{1}11\right]$ being equivalent directions in fcc. These refractions can be observed in a transformation process from fcc to long-period stacking, similar to a transformation from 3C to 6H in natural ZnS [23]. Figure 7d shows the ED pattern at the boundary between the fcc and α-Mg. The four spots on $\left[1\overline{1}1\right]$ fcc divided 1100 of hcp, and $\left[\overline{1}11\right]$ fcc is parallel to $\left[2\overline{3}10\right]$ direction of the α-Mg. In this case, both phases do not have shared stacking direction.

If the E phase transformed to LPSO, the microstructure and the orientation to the matrix would be different from those formed from α-Mg via SESFs. Previous research on Mg₉₇Zn₁Gd₂ highlighted the possibility of the existence of two kinds of LPSO formation processes [8]. One was mentioned concerning Figure 5; the LPSO precipitated into the matrix and increased the yield strength. Another is the decomposition of Mg₃Gd with the D0₃ structure; the LPSO has a block shape and does not increase yield strength. The latter feature is analogous to the formation of LPSO via the E phase.

4. Conclusions

• At a pressure of 3 GPa and above 773 K, as-cast Mg₉₇Zn₁Y₂ alloy consisting of α-Mg matrix and 18R-type LPSO structure undergoes structural phase transformation, forming D0₃ superlattices in the α-Mg matrix.
• The alloy consisting of the α-Mg matrix and D0₃ superlattices transformed into the α-Mg matrix and 18R-LPSO phase upon annealing at ambient pressure.
• Considering from TEM observation results, 18R-type LPSO was continuously present with the 2H phase containing a large number of SFs. When combined with previous report [14,16], the solute atoms concentrate in SF at the almost same time, which is expected by theoretical calculations [10,13,16]. This will be a formation process in the part where the solute concentration is lower than the ideal composition of 18R-type LPSO.
• In the collapse process of the 18R-type LPSO structure under high-pressure and the formation process of the 18R-type LPSO structure at ambient pressure, an intermediate fcc structure with a lattice parameter of 1.42 nm (E phase), which is twice that of D0₃, emerged. The E phase is considered an intermediate structure between 18R-type LPSO and D0₃.
• TEM observations for the E phase of the beam direction parallel to [110] confirmed obscure reflections on $\left[\overline{1}11\right]$ drections, but they were not confirmed in $\left[1\overline{1}1\right]$, though being equivalent directions in fcc. These results indicate the collapse of E phase. Further, these variations of refractions look like a transformation process from fcc to long period stacking, which is observed in natural ZnS from 3C to 6H [23].