Effect of Different Heat Treatments on the Evolution of Novel Al-Si-Cu-Ni-Fe-Re Alloy Fabricated by Selective Laser Melting

Abstract

In this study, Cu and Ni are successfully added to additively manufactured Al-Si alloy by the mixing process to improve the strength and ductility of the alloy. The effects of different heat treatments on the microstructural evolution and bending properties of selective laser-melted Al-Si-Cu-Ni-Fe-Re alloy are systematically investigated to optimize the mechanical properties. Nearly complete dense samples are initially additively manufactured with optimized parameters. The samples display a particular fiber network microstructure in which Cu-rich and Ni-rich phases distribute in an interwoven mesh around the eutectic silicon. After substrate plate heating (SPH) treatment, the network becomes denser, and the bending strength increases by 44.99 ± 1 MPa to 811.11 ± 29 MPa, despite the hardness decreases. Comparatively, solution aging (SQA) treatment results in the fiber network structures transforming into second-phase particles, which leads to a decline in bending strength and a significantly higher degree of ductility. Finally, the mechanisms of changes in microstructure and morphology, as well as mechanical properties after heat treatment, are discussed.

1. Introduction

Selective laser melting (SLM) is an additive manufacturing technology assisted by computer software that controls the laser beam to scan layer by layer in a specified path on a powder bed. The metal powder melts and cools rapidly under the energy input, thus allowing us to obtain parts with complex structures that are difficult to manufacture by conventional methods [1]. It is widely used in aerospace, automotive, and other industries and has become an integral part of high-end manufacturing [2,3]. Because of the strong demand in the aforementioned fields, lightweight and high-strength aluminum-based alloy materials have become the most widely used class of SLM technology. Liu et al. [4] investigated AlSi10Mg composite materials. After SLM treatment, the tensile and bending strength can reach 451 MPa and 674 MPa, significantly higher than aluminum alloys produced by traditional manufacturing methods, such as casting. With the in-depth research on SLM aluminum alloys, the disadvantages of aluminum alloys, such as high laser reflectivity during solidification [2], ease of oxidization, large solidification interval, and propensity for cracking, are gradually exposed to people. Moreover, the aluminum alloy powder and related supporting equipment suitable for SLM are too expensive, limiting its large-scale development in the commercial field [5,6,7]. Therefore, balancing the cost and performance of SLM aluminum alloys has become a hot topic.

AlSi10Mg in the SLM process rapidly decreases in strength after heat treatment. Hwang et al. [8] investigated the change in mechanical properties of AlSi10Mg after heat treatment at 280 °C. The ductility was improved, but the strength decreased by 25.51% (from 458.7 MPa to 341.7 MPa); Clement et al. [9] studied the decay of ultimate tensile strength of AlSi10Mg from 293 MPa to 217 MPa (a reduction of 25.94%) after solution treatment. This phenomenon also emerges in our preliminary work on TJ-AS19-01 (a high silicon aluminum alloy Al-22Si-0.2Fe-0.1Cu-Re aluminum alloy) [10].

Elemental Cu-Ni powders are often used to form diffusion alloy powders with intermediate alloys to improve overall performance [11,12]. In addition, Cu and Ni are also essential strengthening elements in the Al-Si alloy system [13,14,15,16,17,18]. Inspired by this, Cu and Ni powders are mixed as diffusion elements in TJ-AS19-01 to form Al-Si-Cu-Ni-Fe-Re alloy powder, which is then fabricated by selective laser melting. Therefore, in this study, the optimized heat treatment process is performed to enhance the ductility of TJ-AS19-01 mated with selected strengthening alloy elements. The microstructure morphology and composition of the alloy, as well as the heat treatment conditions, are investigated. Finally, the relationship between microstructure and mechanical properties (hardness and bending properties) is discussed, and the strengthening mechanism of the alloy is explored.

2. Materials and Methods

In previous work, TJ-AS19-01 has shown high performance [10]. However, its large solidification interval (due to a large amount of silicon phase) is unfavorable for forming stability, and it is not easy to perform further heat treatment to optimize its properties [19]. Therefore, in this work, TJ-AS19-01 is used as an intermediate alloy, and Cu, Ni and Al powders are used as additive elements to prepare a new Al-Si-Cu-Ni-Fe-Re alloy powder by a mixing process. This mixing process is widely used in SLM [20]. The powder used for mixing is the micron spherical Cu and Ni powder produced by Changsha Tianjo Metal Materials Co., Ltd., China, with a D₉₀ value of about 10–13 µm. Ningxiang Jiweixin Metal Powder Co., Ltd., China. produces spherical Al powder, whose D50 is about 20–25 µm. The purity is 99.9% for all of them. The corresponding masses of TJ-AS19-01, Al, Cu, and Ni powders are respectively weighed and thoroughly mixed in a V-shaped mixing drum. The nominal composition of the mixed powders is determined by calculation, as shown in

Table 1. The nominal composition of each element of the mixed alloy powder.

Element Weight (%)	Al	Si	Cu	Ni	Fe	Re
Al-Si-Cu-Ni-Fe-Re	bal.	12	3.8	2	0.2	0.05

.

Relative density is an effective and intuitive means of characterizing the quality of a print run. The relative density RD is calculated as follows:

$$\mathrm{RD} =\frac{{\mathsf{\rho }}_{\mathrm{s}}}{{\mathsf{\rho }}_{\mathrm{t}}} \times 100\%$$

(1)

ρ_s denotes the actual measured density of the sample, measured by Archimedes’ drainage method. ρ_t represents the theoretical density of the sample, calculated by Equation (2) [21]:

$$\frac{1}{{\mathsf{\rho }}_{\mathrm{t}}} =\frac{\mathrm{x}}{{\mathsf{\rho }}_{\mathrm{Al}}}+\frac{\mathrm{y}}{{\mathsf{\rho }}_{\mathrm{TJ}-\mathrm{AS}19-01}}+\frac{\mathrm{z}}{{\mathsf{\rho }}_{\mathrm{Ni}}}+\frac{\mathrm{u}}{{\mathsf{\rho }}_{\mathrm{Cu}}}$$

(2)

x, y, z, u represent the mass percentage of Al, TJ-AS19-01, Ni and Cu powder, respectively. ρ_Al, ρ_TJ-AS19-01, ρ_Ni, ρ_Cu are their true densities. Therefore, ρ_t can be calculated as 2.7816 g/cm³. The advantage of such a method of calculating the theoretical density of the sample is that we can block out the effects of inhomogeneous mixing (the mixing is actually homogeneous) and obtain a true density value with a minor error.

The equipment used for selective laser melting is HBD-150 (Shanghai Hanbang United 3D Tech Co., Ltd., Shanghai, China) with an IPG 500 w fiber laser beam. The laser beam diameter is about 60 μm. Argon is used as a protective gas in the printing chamber, with an oxygen content of less than 0.1%. The substrate of size φ150 mm is made of Al6061. Before printing, the powder is placed in a vacuum drying chamber (DZF-6090, Shanghai Shanzhi Instruments & Equipment Co., Ltd., Shanghai, China) at 70 °C for more than 12 h to remove surface moisture and allow sufficient drying [22]. Cubic specimens with dimensions of 10 mm × 10 mm × 10 mm are used for densification, microstructure, and morphology analysis. Rectangular specimens are mainly used for bending properties testing after the specimens are milled and polished by SiC sandpapers. The dimensions are shown in Figure 1. The optimal SLM forming parameters and scanning strategy of the sample are shown in

Table 2. Optimal parameters for as-built Al-Si-Cu-Ni mixed powder.

Laser Power	Scanning Speed	Layer Thickness	Hatch Distance	Scanning Strategy
290 W	1000 mm/s	30 μm	40 μm	Stripes (Initial angle 35°, Rotation angle 67°)

, and the relative density reaches 99.1%.

In the field of SLM, heat treatment can be in the form of substrate plate heating (in-situ) and external heat treatment (non-in situ, such as solution aging treatment). In substrate plate heating (SPH), the temperature of the substrate is kept at a constant value during the SLM process, which can have a similar effect to heat treatment.

The SLM-produced samples are divided into 3 groups on average, 2 of which are heat-treated under different conditions. The samples for solution aging (SQA) are fabricated in a tube furnace (GSL-1600X, Hefei Kejing Material Technology Co., Hefei, China). The samples for substrate plate heating (SPH) are obtained by setting the substrate heating parameter to 175 °C, and the other parameters are the same as the SLM process. The different heat treatment procedures are shown in

Table 3. Details of the heat treatment procedures.

Heat Treatment	Heat Treatment Procedure
SLM-produced (as-built)	N/A
SQA	560 °C/3 h/Water quench (WQ) + 175 °C/6 h
SPH	175 °C during SLM

.

The printed sample is separated from the substrate using a line cutter and then ultrasonically cleaned for 1 h to thoroughly remove surface oil and contaminants to ensure measurement accuracy. The density of the samples is measured based on Archimedes’ drainage principle [23]. The cubic specimens are milled and polished with SiC sandpaper under these 3 conditions. The defects of the polished samples are analyzed with an optical microscope (OM, 5XB-PC, First Factory of Shanghai Optical Instruments, Shanghai, China). The sample is etched with Keller (nitric acid 2.5 mL, hydrochloric acid 1.5 mL, hydrofluoric acid 1.5 mL, distilled water 95 mL) for about 1 min and then used for metallographic observation. The microstructure is observed by scanning electron microscopy (SEM, Zeiss Gemini 300, Oberkochen, Germany). Energy dispersive spectrometer (EDS, Bruker EDS QUANTAX, Karlsruhe, Germany) line analyses are performed to determine the chemical composition. The orientation of all microstructure images is perpendicular to the build direction. X-ray diffraction (XRD, Bruker D8 Advance, Karlsruhe, Germany) is performed at room temperature using a copper tube at 30 kV and 30 mA. Using an X-ray generator with Cu Kα1 (λ = 0.15406 nm) radiation, the scan rate is 2°/min, and the 2θ angle varies from 20° to 80°.

Due to the brittle characteristics of the Al-Si alloy produced by selective laser melting, bending tests are performed to determine the mechanical strength of the alloy under these conditions. The printed specimens’ bending strength and Vickers hardness are tested at room temperature using a universal hydraulic testing machine (CHT4305, MTS, Eden Prairie, MI, USA) and a Vickers hardness tester (HV-50, Shanghai Shangcai Testing & Manufacturing Co., Ltd., Shanghai, China). The 3-point bending method is used to measure the bending strength. The distance of the support cylinders on the fixtures is 25.4 mm, and the loading rate is fixed at 0.5 mm/min. For each condition, a total of 3 samples are tested for bending strength, and the average is calculated. The surface Vickers hardness of the specimens is tested under each of the 3 conditions. The loading duration of the hardness test is set to 15 s under at a load of 49 N. Three points are tested on the surface of each specimen, and the average value is calculated.

3. Results and Discussion

Figure 2 shows the XRD results of the three conditions of the alloy in as-built, SPH and SQA. The as-built alloy has no Cu and Ni single-element phases, indicating that the elements are fully alloyed after SLM. In addition to the peaks of the α-Al matrix and Si phase, a small amount of ternary reinforced Al-Cu-Ni phase and binary Al-Cu phase are found, similar to the results reported in the literature [16,24]. After SPH, this ternary phase is greatly diminished, and the peak intensity of the binary Al-Cu and Al-Ni phases gradually increases. After SQA, a large number of binary Al-Ni phases appear. Moreover, the peak intensity of each phase changes after heat treatment. It is evident that the peak intensity of the ternary Al-Cu-Ni phase is weakened, and the α-Al peak is shifted to a lower angle, indicating a distortion of the matrix, which may be related to the Cu atoms dissolved in the matrix. The silicon peaks’ intensity increases with the heat treatment’s deepening, attributed to the thermally activated growth of silicon along the most stable plane with the lowest free energy [25]. The silicon peaks’ width decreases with the heat treatment’s deepening, indicating that the silicon present in the supersaturated solid solution diffuses out of the aluminum matrix and increases in size [26], which is consistent with conventional findings.

The change in the microstructure phase often indicates a change in material properties. Therefore, the mechanical properties are measured under these three conditions, mainly including Vickers hardness and bending properties. Figure 3 shows the Vickers hardness results. The Vickers hardness at the top surface is 193.24 ± 9 HV, 175.01 ± 9 HV, 99.30 ± 2 HV, and the Vickers hardness at the side is 173.7 ± 11 HV, 157.66 ± 10 HV, 98.83 ± 2 HV for the samples in as-built, SPH, and SQA conditions, respectively. The hardness of the square specimens gradually decreases with increasing depth of heat treatment, which could be due to the diffusion of refined silicon and secondary phase precipitates from the aluminum matrix, resulting in a weakening of the solution strengthening effect [27]. It is worth noting that there is a significant difference between the top and side hardness of as-built and SPH samples, which is related to the anisotropy due to the unique SLM process [22]. The SQA specimens do not exhibit such differences, and the hardness is homogeneous. The Al-Si-Cu-Ni-Fe-Re as-built samples exhibit excellent hardness values, which are higher or almost the same as existing cast and other SLM aluminum alloys [10,28,29]. That is mainly due to the fine structure formed by the high cooling rate of the SLM process and the distorted stress field in the supersaturated matrix, which we will discuss in detail later.

After heat treatment, the bending properties at room temperature of this alloy prepared from the mixed powder are also significantly altered. Figure 4 shows the histogram of the bending strength-displacement of Al-Si-Cu-Ni-Fe-Re samples tested at room temperature under these three conditions. The bending strength of the as-built sample is 766.12 ± 30 MPa, which is much higher than that of the aluminum alloy produced by the conventional method [30,31] and also slightly higher than the SLM aluminum alloy reported in the existing literature [4,10]. It is obvious that the sample after SPH has the highest bending strength, measured to be 811.11 ± 29 MPa, and notably, the displacement after fracture (1.12 ± 0.11 mm) is also higher than that of the as-built alloy (0.84 ± 0.03 mm). The bending strength of the SQA sample is significantly reduced to 642.11 ± 6 MPa. Nevertheless, it has a relatively high displacement of 2.33 ± 0.02 mm.

In order to explore the reasons for the changes in the mechanical properties of this alloy before and after heat treatment, the microstructural characteristics of as-built, SPH, and SQA alloys are further investigated. OM images (Figure 5) depict the etched Al-Si-Cu-Ni-Fe-Re alloy in as-built and heat-treated conditions. It can be seen that the as-built alloy shows a unique molten pool structure. With the deepening of the heat treatment, this melt pool structure gradually becomes indistinct and eventually disappears.

SEM images depict the etched Al-Si-Cu-Ni-Fe-Re alloy in as-built and heat-treated conditions, respectively, as shown in Figure 6a–c. Moreover, Figure 6d–f offers enlarged views of the yellow dotted boxes in Figure 6a–c, respectively. At higher magnification, the microstructure of the as-built sample consists of a fine α-Al matrix and the forming fiber network structure, consistent with most descriptions in the literature [32,33,34]. The difference is that the literature mostly refers to eutectic silicon networks. At the same time, it is clear from the EDS results that these interconnected networks are Cu- and Ni-rich second phases (Figure 6g–i, an enlarged view of the yellow box). It can be considered to be uniformly distributed around the silicon network. This phenomenon is attributed to the high cooling rate and the unique laser scanning method during the SLM process [35]. Similar phenomena are also found in other systems, such as in Figure 6j,k [36] (as a comparison with Figure 6a). The fibrous cell borders are enriched with specific elements that are distinctly different from the composition of the cell core. During solidification, the low melting point phase (CoCr-rich phase in Figure 6j and Fe-rich phase in Figure 6k) solidifies before the high melting point phase (Mo-rich phase in Figure 6j and Mo-, Cr-rich phase in Figure 6k) due to the kinetic conditions. The high melting point phase diffuses with the help of surface tension effects (Bernard Marangoni Surface Instability, BMI mechanism). It is deposited along cell boundaries through the particle-accumulated structure formation mechanism (in short: the PAS mechanism). The combined effect of the two mechanisms results in forming of a specific sub-stable structure with different components in the cell core and cell boundary. That explains the formation of Cu- and Ni-rich phases (high melting point phase) in the fiber network. During the SLM process, the Cu-rich and Ni-rich networks are interwoven with the eutectic silicon network due to the combined effect of the BMI and PAS mechanisms, which significantly impacts the mechanical properties of the material.

It can be seen that the matrix is mainly divided into coarse and fine crystalline regions [10]. The coarse zone is due to grain growth as a result of repeated scanning with the laser. The SPH sample shows a similar microstructure to the as-built sample (Figure 6b). The tracks of the molten pool and its core are clearly visible. It is not easy to distinguish them from each other. Throughout the preparation process, the substrate plate is kept at the aging temperature of 175 °C, and the morphology and distribution of the phases are slightly changed. It can be seen that the size of the cells changes in comparison with the as-built alloy and becomes denser (shown in the dark blue dashed line in Figure 6b), reflecting the increase of the silicon phase as well as the second phase, which is consistent with the XRD results. Because of the reduced cooling rate and thermal gradient during the SLM process, these phases have sufficient time to precipitate out of the matrix. In addition, the presence of substrate heating makes the melt pool less reactive, which helps avoid undesirable phenomena such as splashing drops.

To better understand the microstructural evolution, the Al-Si-Cu-Ni-Fe-Re alloy is subjected to solution heat treatment followed by artificial aging (SQA). Figure 6c shows the microstructure obtained after SQA. Obviously, it does not exhibit the unique structural characteristics of SLM. The fiber network is wholly decomposed. The melt tracks disappear entirely and are replaced by a homogeneous microstructure. Al element mainly fills the matrix, and irregular silicon particles are dotted on it. Of particular note, the silicon grains and matrix are littered with irregular flakes and spherical particles of varying sizes, and these areas are enriched in nickel and copper. It is generally believed that the spherical second phase is due to precipitation from the supersaturated α-Al matrix during the subsequent aging process [39,40]. In conjunction with the XRD analysis results, it can be seen that the zone of elemental enrichment is mainly composed of Al-Cu, Al-Ni binary phase, and Al-Cu-Ni ternary phase. In addition, the second phase and silicon particles have grown significantly, with an average size of 1.29 μm, mostly between 1.0 and 1.5 μm, and a few reaches about 5 μm.

After solution treatment, the alloying elements are further dissolved into the Al matrix, which destroys the fiber network structure to a certain extent. In addition, with the influence of artificial aging, the supersaturated Al matrix gradually precipitates alloy atoms, making the original structure more and more indistinct. In the course of temperature increase, small particles have larger surface energy. In the trend of surface energy reduction, different particles merge and then grow into spheres, the number of which decreases. Fortunately, the presence of second-phase particles hinders the fusion between neighboring silicon particles and inhibits their growth. In comparison, the respective dimensions are suppressed to a relatively low degree, which positively affects the material’s mechanical properties.

Figure 7 shows the bending fracture surfaces of the as-built and heat-treated (SPH, SQA) alloys. The as-built alloy shows a relatively smooth fracture surface. Tiny dimples can be seen on the surface of the as-built alloy (Figure 7d) at high magnification [27,41], indicated by red arrows. Interestingly, the size of the dimples (0.5–1 μm) is similar to that of the fiber structures observed in the original microstructure. The boundary of the fiber structure is eutectic silicon with the second phase (Cu- and Ni-rich zones), which has the potential to act as a site of stress concentration. Therefore, microcracks tend to extend along the edges of the fiber structure. The image of the as-built alloy (Figure 7a) also shows a more typical cleavage step pattern, symbolizing brittle fracture [42,43]. Due to the mixing and SLM process characteristics, there are inevitably a small number of non-fused holes and powder particles. These defects often lead to crack propagation and expansion during the fracture of the material, which is not conducive to improving the mechanical properties.

Figure 7b,e shows the fracture characteristics of the SPH sample at different magnifications, respectively. Unlike the as-built alloy, the SPH sample does not exhibit obvious cleavage steps but rather a distinct ductile fracture feature–dimples. The size of these dimples (0.3–0.8 μm) also corresponds to the network structure after SPH. Due to the low heating temperature of the substrate (175 °C), the continuous distribution of the fiber network on the substrate is not destroyed. Numerous secondary phases and silicon phases are precipitated from the supersaturated aluminum matrix and successively distributed in the network, making the network denser [44]. In addition, there are still some cracks, which are shaping defects and are usually difficult to avoid.

The fracture characteristics of the SQA alloy are shown in Figure 7c,f. In contrast, it also shows a smooth fracture surface. Because of the higher heat treatment temperature, the network structure disappears, and a larger amount of silicon and second phase precipitate from the matrix. The possible reason is that voids nucleate at the large Si particles by de-cohesion, and then void coalescence occurs. The fracture occurs at the silicon particles, and cracks form along the fracture [45]. This feature indicates a typical ductile fracture. It is noteworthy that submicron and nanoprecipitates of the second phase are dotted in the dimples (shown by blue arrows in Figure 7f). They have a homogeneous shape and act as precipitation reinforcement, giving the material high ductility at the macroscopic level combined with high strength.

The schematic diagram in Figure 8 shows the microstructural evolution of the SLMed Al-Si-Cu-Ni-Fe-Re alloy under different heat treatment processes (SPH, SQA). As shown in Figure 8a, the solubility of Si, as well as Cu and Ni elements in the Al matrix, increase significantly due to the high cooling rate of the SLM process, resulting in lattice distortion. The stress field generated by the distortion hinders the movement of dislocations during deformation and provides a strong solid-solution strengthening effect. On the other hand, the second-phase particles distributed around the eutectic silicon network also play a role in dispersion reinforcement.

The SPH alloy still retains the structure of the fiber network. The reinforcement phase precipitated during the SLM process is distributed on the substrate, strengthening the connection with the network and refining the network structure (Figure 8b). It is expected that the fiber network structure is the most crucial factor determining the strength of the SLM alloy [44]. In addition, these precipitated second phases further fill the voids and interact with the eutectic silicon network. The second phase particles simultaneously act as a barrier to dislocation movement during the bending process. According to the Orowan strengthening mechanism, highly dispersed nano-reinforced phases in the matrix can also hinder the dislocation motion [44,46]. On the other hand, it is the intertwining of the two types of fiber networks that strengthens the ductility of the material. The reduction in hardness may be related to the solid solubility of the Al matrix. With the heating process of SPH, the stress field generated by the lattice distortion is released.

Due to the high heat-treated temperature, the microstructure of the SQA alloy is radically transformed (Figure 8c). The fracture of eutectic silicon networks, as well as Cu- and Ni-rich phase networks, is a factor in improving ductility [47]. This is because it changes the damage mode from crack extension along the brittle intercellular network to a more ductile microporous nucleation and agglomeration mechanism based on fine particles, which facilitates the reduction of local stress or strain [26,47]. On the other hand, the residual thermal stress during the SLM process can be reduced by the SQA process. These two factors contribute to the ductility of the alloy. At the same time, the disruption of the fiber network structure leads to a reduced contribution of boundary reinforcement [45]. During the bending process, Cracks extend along the large silicon particles (similar in size to the dimples in Figure 7f), and the strength is reduced. However, remarkable strength loss occurs in the SQA alloy, as in T6-treated AlSi10Mg [9,48,49,50]. Meanwhile, the fiber network structures have entirely transformed into isolated precipitation patterns.

The size of the dimples also indicates the structures brought by the different heat treatments [44]. In both the SPH and the as-built alloy, the dimples’ size resembles the fiber network’s structure. This means that the strength of the SPH alloy or the as-built alloy is dominated by the network structure since the failures of the two alloys originated from the network. In the SQA specimen, on the other hand, the alloy undergoes a simultaneous precipitation process, similar to the aging process. Under the action of a large thermal driving force, the size of the dimples increases sharply and reaches about 1–3 μm, which is comparable to the size of silicon particles after their growth. The failure is believed to be triggered by the constraint of the isolated precipitation phases, which reduce the strength after SQA treatment.

4. Conclusions

In this work, a novel, nearly complete dense (a relative density of 99.1%) Al-Si-Cu-Ni-Fe-Re alloy is fabricated by selective laser melting using optimized process parameters. The Al-Si-Cu-Ni-Fe-Re alloy is formed from a mixture of Al-22Si-0.2Fe-0.1Cu-Re alloy powders with additional Al powders and finer Cu, Ni powders and treated with different heat treatment processes (SPH and SQA). A comprehensive study is conducted on the evolution of this alloy’s microstructure and mechanical properties. The SPH heat treatment is able to promote the ductility of the alloy as well as the bending properties from 766.12 ± 30 MPa to 811.11 ± 29 MPa. This improvement is mainly due to the overlapping fiber network structure consisting of Cu- and Ni-rich phases and eutectic silicon precipitated on the matrix. Herein the hardness decreases from 193.24 ± 9 HV to 175.01 ± 9 HV. However, during alternative solution aging treatment, the fiber network structure is decomposed and converted into individual large particles. This transformation leads to a significant reduction in bending strength and hardness, as well as a marked increase in ductility. The evolution of the microstructure ranges from the initial fiber network structure (the as-built alloy) to the enhanced network (SPH alloy) to isolated precipitation patterns (SQA alloy). The evolution can also be explained by the sizes of the dimples in the corresponding bending failures.