Effect of Interdendritic Precipitations on the Mechanical Properties of GBF or EMS Processed Al-Zn-Mg-Cu Alloys

The effects of nanoprecipitations on the mechanical properties of Al-Zn-Mg-Cu alloys after GBF (gas bubbling filtration) and EMS (electromagnetic stirring) casting were investigated. Dendritic cell structures were formed after GBF processing, while globular dendritic structures were nucleated after EMS processing. Equiaxed cell sizes were smaller in the EMS-processed specimens compared to the GBF-processed specimens, confirmed by EBSD (electron backscatter diffraction) analysis. Nanoprecipitations of η′ phases inside of dendrites were observed by TEM (transmission electron microscope), and other Fe-bearing compounds were located in the dendritic boundaries. The yield strength of the T4 and T6 heat-treated specimens was close to 400 MPa and 500 MPa, respectively. Fractographic analysis was performed to investigate the effect of precipitations on tensile fracture.


Introduction
Al-Zn-Mg-Cu cast alloys have been widely used for aeronautical and automotive applications due to their excellent combination castability, mechanical properties, and formability [1][2][3][4][5]. These alloys are mostly fabricated using ingot metallurgy, during which casting defects, such as shrinkage and pores, may develop and deteriorate the mechanical properties of the final product [6][7][8]. Several approaches are implemented in the aluminum alloy industry to eliminate these casting defects. Among these, gas bubbling filtration (GBF) and electromagnetic stirring (EMS) casting treatments are effective ways to remove casting defects [9][10][11]. However, these casting treatments can also cause additional effects on the microstructure of the alloy, such as grain refinement and intermetallic phase forming [5,12,13]. Therefore, demands for improved mechanical properties necessitate the understanding of these casting treatments on the microstructure of the alloy [14].
Many researchers have attempted to improve mechanical properties by optimizing the size and fraction of intragranular and grain boundary precipitates (GBPs) using casting treatment [28,29]. For instance, Wannasin et al. [30] reported non-dendritic microstructures refinement by controlling the cooling and gas flow rates. Oh et al. [14] suggested that longer stirring time, from 20 s to 80 s, during solidification refined the average grain size. Lee et al. [31] also reported effective grain refinement in the microstructure when the degassing time was increased during the GBF process. However, most of the past studies have only focused on the grain refinement effect caused by melt treatments during casting, while the information regarding casting treatment influence on the grain boundary and intragranular precipitates has been neglected. Therefore, a comprehensive analysis is required to observe and optimize grain refinement, as well as to precipitate the size fraction, in cast Al-Zn-Mg-Cu alloys.
In the present study, we developed Al-Zn-Mg-Cu alloys using two different melt treatments during casting, i.e., GBF and EMS. The variations in grain refinement and interdendritic precipitate size and distribution were evaluated using microstructural characterization. The tensile properties and strain-hardening behavior of the cast specimens were evaluated under different aging conditions, such as T4 and T6. The results were used to discuss the possibility of improving the mechanical properties of Al-Zn-Mg-Cu alloys by controlling the microstructure using casting treatments.

Experimental Details
Ten kilograms of alloy melt was produced in an electric resistance furnace. Metals of 99.9% purity (Al, Zn, and Mg) with two master alloys (Al-5Zr and Al-30Cu) were used for melting. The melt was either degassed for 10 min using Ar gas by the GBF process or electromagnetically stirred for 30 s or 50 s. A cylindrical mold with a 50 mm diameter and 200 mm length was pre-heated to 180 • C, and the 1.5 kg processed melts were poured into it at 700 • C. The chemical composition of as-cast ingots was measured by inductively coupled plasma optical emission spectrometry (ICP-OES, Oxford instruments, Abingdonon-Thames, UK) as Al-7.0Zn-2.5Mg-1.5Cu-0.2Zr (wt.%). The GBF-processed specimen is referred to as G-A, while the EMS-processed specimens are referred to as E30-A and E50-A, with respect to stirring time. Samples were solution-treated at 450 • C-8 h + 460 • C-8 h + 470 • C-8 h followed by water quenching. Some of the solution-treated specimens were naturally aged (T4) at 25 ± 2 • C for 1000 h, whereas some were artificially peak-aged (T6). T6 was performed immediately after solution treatment at 120 • C for 24 h.
The specimens for microstructure analysis and mechanical testing were cut from the center of each ingot. The specimens were polished and etched using Keller's etchant (H 2 O 95 ml + HCl 2 ml + HNO 3 2.5 ml + HF 1.5 ml) for microstructural observation in a scanning electron microscope (SEM) (JSM-7610F, JEOL, Tokyo, Japan). The volume fractions of the secondary phases were measured using an image analyzer. Five SEM images of each sample were used, and the average values are mentioned in Section 3. The chemical composition of secondary phases was determined by an energy dispersive X-ray spectrometer (EDS) (Oxford instruments, Abingdon-on-Thames, UK) attached with the SEM. Electron backscattered diffraction (EBSD) analysis was performed using a low accelerating SEM (JSM7900F, JEOL, Tokyo, Japan). High-resolution transmission electron microscopy (HRTEM) (TF30ST, FEI, OR, USA) was operated at 300 kV. Thin foils (∅ = 3 mm, thickness = 100 µm) for TEM were prepared by mechanical grinding. The foils were then electro-polished in a jet polisher (TenoPol III, Struers, Ballerup, Denmark) in a 3:1 methanol and HNO 3 solution at −20 • C. The Vickers hardness was measured using a hardness tester (HV-114, Mitutoyo, Tokyo, Japan) with a load of 1 kg and 15 s dwell time. The tensile test specimens were prepared with a gauge length of 30 mm and a diameter of 6 mm. Room temperature tensile tests were conducted using a universal testing machine (8516, Instron, MA, USA) at a strain rate of 10 −3 s −1 . The fractography of tensile-tested specimens was observed and charted via SEM. For hardness and tensile tests, five specimens for each condition were tested. The results, exhibiting properties and approach with average points, are reported in the present study. Figure 1a,b shows the SEM and EBSD micrographs of the representative specimens (G-A, E30-A, E50-A, G-T4, G-T6, E30-T4, and E30-T6). Dendritic cell structures existed in the G-series specimens while globular dendritic structures were observed in the Eseries specimens. Globular dendritic structures were formed by partial nucleation and crushing of slurries during solidification [32]. In the as-cast (G-A, E30-A, and E50-A) specimens, secondary phases were located on the interdendritic boundaries consisting of a network morphology. The volume fraction of secondary phases in the G-A specimen was 4.6%, whereas it was 4.8% and 5.4% in the E30-A and E50-A specimens, respectively. The equiaxed cell size of the specimens was measured using IPF maps shown in Figure 1b. The equiaxed cell size of the G-A specimen was 93.8 µm, and those of the E30-A and E50-A specimens were 82.7 µm and 69.3 µm, respectively, which decreased with increasing stirring time. According to the EBSD inverse pole figure (IPF) and grain boundary analysis of the G-A, E30-A, and E50-A specimens, equiaxed cell boundaries were composed of high-angle grain boundaries (HAGBs). The cell size was the largest (245 µm) in the G-A specimen, which also contained many sub-structures. In contrast, the E30-A and E50-A specimens showed a smaller cell size with different grain orientations. Figure 1c shows that the dissolution of the coarse secondary phase located on cell boundaries took place during solution treatment. The volume fraction of secondary phases in the G, E30, and E50 specimens was 0.9%, 1.0%, and 2.0% after T4, and 1.2%, 1.3%, and 2.9% after T6, respectively. The decrease in the volume fraction of secondary phases is attributed to the dissolution of Zn-and Mg-containing particles. These observations are consistent with past studies [33,34]. Furthermore, the equiaxed cell size of the as-cast specimens also decreased after the heat treatment, as shown in Figure 1d. Figure 2 and Table 1 show the EDS elemental mapping analysis of the G-and E30-series specimens in as-cast and heat-treated conditions. Figure 2a,b shows that secondary phases present on the dendritic cell boundaries of the G-A and E30-A specimens are Al 7 Cu 2 Fe and Al 2 (ZnMgCu) 3 . In Al-Zn-Mg-Cu alloys, solute atoms tend to segregate on dendritic cell boundaries during solidification and form Zn-, Mg-and Cu-rich phases. However, Cu extends its solubility in the Zn-Mg-rich intermetallic phase, i.e., the Al 2 (ZnMg) 3 -T phase, changing its composition to Al 2 (ZnMgCu) 3 [35].

Results and Discussion
After heat treatment, the Zn and Mg elements were readily dissolved in the matrix due to their high diffusion co-efficient, while the Cu and Fe velocities are low over wide temperature ranges [35]. Subsequently, the network type morphologies of secondary phases in as-cast specimens were changed to the island type. As mentioned previously, the total volume fraction of secondary phases decreased to 1~2%. Secondary phases were analyzed in G-T4 and E30-T4 specimens and characterized as Al 7 Cu 2 Fe phase and Curich remnants, as shown in Figure 2b,d. It is obvious in Figure 2a,b that the continuous interdendritic Al 2 (Zn, Mg, Cu) 3 or Al 7 Cu 2 Fe precipitates became discontinuous after heat treatment (Figure 2b,d). These discontinuous precipitates can act as crack initiation sites during the deformation of the alloy, degrading its strength and elongation. The precipitation hardening during T4 tempering occurs at a steady rate, with a lower density of strengthening precipitates and does not lead to higher strength. On the other hand, accelerating precipitation (η /η) hardening takes place during T6 treatment, which effectively increases the strength of the alloy.   Table 1).  Figure 3 shows the TEM analysis performed on the E50-T6 specimen confirming fine precipitates presence along [110] Al . Under T4 tempering, only GP zones were formed at room temperature after 1000 h of natural aging, having a coherent relationship matrix, as shown in Figure 3a. Under T6 tempering, a high density of nano-sized η phase was uniformly dispersed in the matrix [22][23][24][25]36]. As shown in Figure 3b  GP zones usually form during natural aging or at the early stage of artificial aging. GP zones serve as nucleation sites for the formation of metastable η phases [34]. The metastable η phase then transforms into the stable η phase (MgZn 2 ) which is solely responsible for the peak hardening of Al-Zn-Mg-Cu alloys [16,22,[25][26][27]. GP zones are categorized into two groups. GPI zones are coherent with the Al matrix, yielding the internal ordering of Mg/Al and Zn on the (001) matrix, which is dominant during natural aging. GPII zones are Zn-rich layers on a (111) matrix usually nucleated by artificial aging [37][38][39]. The GPII starts to form from the vacancy-related clusters and is transformed to the η phase during after-solution treatment [40,41]. The metastable η phase is mostly known as the major hardening phase in Al-Zn-Mg-Cu alloys. Localized strain fields were formed around the finely dispersed η phase, which act as barriers for dislocation sliding [42]. The η phase has a chemical composition of Mg 2 Zn 5-x Al 2+x [24,43]. Figure 3b confirms that η has a hexagonal crystal structure and the lattice parameters are a = 0.486 nm and c = 1.329 nm. η transforms into the η (MgZn 2 ) phase as the aging time increases. Figure 3c shows that the η phase has a hexagonal crystal structure with a = 0.523 nm and c = 0.856 nm. Table 2 summarizes the Vickers hardness of the G, E30, and E50 specimens under as-cast, T4, and T6 conditions. The Vickers hardness value is the highest under the T6 condition while it is the lowest in the as-cast form. The high hardness value of the T6 treated specimen compared to the T4 treated specimen is attributed to effective precipitation hardening during the η and η phases. When comparing the GBF and EMS processes, it was revealed that the GBF specimen had higher hardness than the EMS samples. According to microstructure analysis, the volume fraction of the interdendritic precipitates in the GBF specimen was smaller than in the EMS specimens. This lower fraction translated into better precipitation hardening during T6 tempering resulting in a higher fraction of strengthening precipitates, i.e., higher hardness. The higher hardness value of the E50 specimen is ascribed to its finer grain size, even though it had a higher fraction of interdendritic precipitates.  Figure 4 and Table 3 show the tensile test results of the G and E specimens under as-cast T6 and T4 conditions. The tensile properties of the GBF specimens were better than the EMS specimens under as-cast and heat-treated conditions. The better mechanical behavior of the GBF-treated specimens is attributed to their lower fraction of interdendritic precipitates, which act as crack nucleation sites during tensile deformation. Similarly, a lower fraction of interdendritic precipitates reflected better tensile properties in the E30 specimen compared to the E50 specimen. The average tensile strength values of the G, E30, and E50 specimens were approximately 492 MPa, 489 MPa, and 440 MPa, respectively. Compared to the T4 tempered specimens, the T6 tempered specimens had a much higher tensile strength. The average tensile strength values of the G, E30, and E50 specimens were approximately 545 MPa, 530 MPa, and 521 MPa, respectively. The higher strength of the T6 specimen is attributed to a large fraction of strengthening precipitates formed during artificial aging, as shown in the TEM images in Figure 3b,c. In contrast, all the specimens showed better elongation under T4 conditions than T6. This can be explained by the TEM image analysis of the T4 specimen in Figure 3a showing only nanosized GPII zones. It is well known that dislocations shear these nanosized clusters during plastic deformation. This shearing can effectively impede the dislocation motion, resulting in higher elongation. The effectiveness of nanosized GP zone shearing is further highlighted by strain-hardening curves in the Figure 4a,b insets. The strain-hardening rate in all specimens under the T4 condition is better than under the T6 condition, which is attributed to GP zone shearing by dislocations. Moreover, the strain-hardening behavior of the T6 specimen is lower than for the T4 specimen in all conditions displayed by the strain-hardening curve in Figure 4b. The precipitation sequence in an Al-Zn-Mg-Cu alloy usually begins with the formation of Mg-and Zn-rich clusters during natural aging at room temperature, or at an early stage of artificial aging. These clusters transform into coherent GP zones, followed by a semi-coherent η phase, and finally, an incoherent η phase, with time and temperature. Therefore, a peak-aged alloy is mostly comprised of η and η precipitates, while only GP zones form during natural aging (T4). Each precipitate imparts varying effects to the tensile behavior of the alloy based on their interplay with dislocation during plastic deformation.  According to precipitation strengthening theory, dislocations move forward by shearing precipitates [27,44,45]. In the shearing mechanism, the coherency strengthening (∆σ cs ) contributes to the yield strength; ∆σ cs is expressed as the following: [46,47] ∆σ cs = Mα ε (Gε c ) where M is the mean orientation factor (3.06 for fcc metals), α ε = 2.6 for face-centered cubic (fcc) metals, G is the shear modulus (26.9 GPa for the Al 7075 alloy), ε c is the constrained lattice parameter misfit, r is the radius of the precipitates, and f is the volume fraction of the precipitates. According to TEM analysis, the average diameter of the GPII zones was 2.3 nm and that of the η' phase was~6.6 nm; thus, they acted to increase the strengthening of the T4 and T6 specimens by shearing mechanisms, while the η phase, with a size of 20 nm, had difficulty enhancing strength. Therefore, the T4 specimen showed lower tensile strength compared to the T6 specimen due to the high volume of fracturing in the η phase. This observation confirms that η phases are not effective for the dislocation shearing mechanism.
Furthermore, the strain-hardening behavior of the GBF-treated specimen was better than for both the EMS-treated specimens. As observed in the microstructure analysis, the volume fraction of coarse interdendritic precipitates in the EMS samples was higher than in the GBF samples. The coarse interdendritic precipitates induced strain localization in their surroundings, resulting in crack initiation during plastic deformation, and alloy fractures with low elongation. Despite the presence of fine intergranular precipitates, a lower volume fraction of coarse precipitates is also significant for the strain-hardening behavior of the Al-Zn-Mg-Cu alloys.
According to the fractographic observation in Figure 5, all the specimens showed intergranular/interdendritic fractures. Porosities were not observed in the G-series specimen, but small pores existed in the E-series specimens due to severe EMS treatment. Stress was localized in the vicinity of pores leading to a decrease in tensile elongation. The grain size of the E-series specimens was smaller than that of the G-series specimens, as confirmed in Figure 1. The brittle fractures in the T4 and T6 treated specimens were associated with the continuous η phases combined with intergranular/interdendritic cracking. Primary precipitates along grain boundaries grow and aggregate to become coarse intergranular phases during the heat treatment process. Thus, the present study demonstrated that the lower volume fraction of coarse secondary phases in G-series specimens showed better elongation. A lower volume fraction of the coarse particles not only decreases the stress localization during plastic deformation but also the intergranular/interdendritic cracking.

Conclusions
In this study, the effects of casting methods (GBF and EMS) and heat treatments (T4 and T6) on the tensile properties of Al-Zn-Mg-Cu alloys were investigated. The grain size was reduced by a newly proposed EMS process compared to the conventional GBF method. The T6 tempered specimen had relatively higher strength and lower elongation than the T4 tempered specimen due to the presence of coherent and semi-coherent phases, such as the GPII and η phase. Grain refinement is essential to enhance the tensile strength of all the metals including the Al-Zn-Mg-Cu alloy. A network structure of grain boundary precipitates, in S(Al 2 CuMg) or Al 7 Cu 2 Fe phases, link together to create crack propagation during tensile deformation, leading to the premature failure of an Al-Zn-Mg-Cu alloy. The formation of grain boundary precipitates should be controlled to reduce the possible crack initiation sites.