*3.3. Superplastic Fracture Surface Morphologies*

The superplastic fracture specimens were naturally aged, and the SEM results of the fracture surface morphologies at different temperatures and strain rates are shown in Figure 7. Fracture surfaces that formed under the GBS exhibited the characteristics of ductile failure at grain boundaries, and a large density of submicrometer filaments was evident on the fracture surfaces. Nevertheless, these filaments protruded from the fracture surface along the tensile direction. Similar sub-micrometer filaments were previously observed on the fracture surfaces of deformed superplastic alloys AA5083 [38] and AA7475 [44], as well as for aluminum-matrix composites [45].

**Figure 7.** SEM micrograph of the superplastic fracture surface morphologies at *<sup>T</sup>* = 400 ◦C, . *<sup>ε</sup>* = 1 <sup>×</sup> <sup>10</sup>−<sup>3</sup> <sup>s</sup>−<sup>1</sup> (**a**), *<sup>T</sup>* = 450 ◦C, . *<sup>ε</sup>*= 1 <sup>×</sup> <sup>10</sup>−<sup>3</sup> <sup>s</sup>−<sup>1</sup> (**b**), *<sup>T</sup>* = 500 ◦C, . *<sup>ε</sup>*= 5 <sup>×</sup> <sup>10</sup>−<sup>4</sup> <sup>s</sup>−<sup>1</sup> (**c**), and *<sup>T</sup>* = 550 ◦C, . *<sup>ε</sup>*= 5 <sup>×</sup> <sup>10</sup>−<sup>4</sup> <sup>s</sup>−<sup>1</sup> (**d**).

A variety of explanations have been reported for the origins of these sub-micrometer filaments. Superplastic tensile tests were performed below the temperature of the initial phase inversion (*T* < 556 ◦C). Therefore, the formation of filaments at 400 ◦C was actually due to sliding transitions from solute drag creeping to high angle grain boundaries. However, this cannot be from the incipient melting or the formation of a glassy phase at grain boundaries [46]. Figure 8, however, provides apparent evidence from which higher concentrations of magnesium and oxygen were observed on the fracture surfaces, indicating the formation of an Mg-rich oxide. Oxides were ridged on early exposed surface regions, as shown in Figure 8, with a composition similar to the filaments on the fracture surface. Ritchie et al. [47] reported that the Mg-rich oxides MgO and MgAl2O4 were preferentially formed on Al-Mg alloys at 500 ◦C. Chang et al. [48] confirmed that filaments were observed at the fracture surfaces when the specimen was tested in air rather than under vacuum conditions.

Previous works observed that filament formation occurred in AA5083 within the cracks in the tribolayers that correspond to locations above the grain boundaries of the near-surface subjected to the GBS at *T* > 300 ◦C [49,50]. Consequently, the Mg5Si6 phase precipitated at the grain boundaries. This was the reason for the increased Si content from 2.95 (in the substrate) to 11.69 wt.% (in the filaments). In addition, the Mg5Si6 phase particles that precipitated along the grain boundaries and dominated the filament formation were the result of grain boundary sliding during superplastic deformation. Therefore, this is a reasonable explanation for the increased Si content in the filaments. These results suggest that the formation of sub-micrometer filaments of the fine-grained 5A70 alloy during superplastic deformation occurred from the growth of Mg-rich oxides and precipitated second phases at the grain boundaries.

**Figure 8.** EDS analysis for the superplastic fracture surface of the substrate (**a**) and the filaments (**b**) of the fine-grained 5A70 alloy.
