*3.6. Fractography Analysis*

Figure 6 illustrates the SEM morphology of the fractured surfaces from the tensile specimens before and after stress relieving. Figure 6a showed dense and smooth surface for the 0◦ orientation as-built samples, with fracture defects observed. Fracture observed to begin from the one end of the structure propagating to the other end.

**Figure 6.** SEM morphology of SLM produced AlSi10Mg samples, orientation 0◦ (**a**) and (**d**), orientation 45◦ (**b**) and (**e**) and orientation 90◦ (**c**) and (**f**) as-built and post stress relief respectively.

The stress relieved samples, (Figure 6c–f), showed dimples, which illustrated the ductility that was desirable for the manufacturing of industrial parts, but in this case the tensile testing strength was proven to have been compromised. There was also an increase in the number of surface pores observed on the stress relieved samples compared to the as-built samples, which was in support of the Archimedes density and porosity results presented in Table 1.

Various defects were observed for the as-built samples, for instance in Figure 6b, ballus defects were observed and Dheyaa et al. [30] explains these ballus defects, marked with yellow arrows, to be unmolten or partly molten powder. According to Read et al. [13,31], these unmolten powder particles are a result of thick oxides layer existing on the particles, which did not allow full consolidation during processing. Defects such as unmolten powder and pores (marked in red), impacts the fatigue life of a component severely by reducing the effective load bearing area thus causing stress concentrations that consequently form static and dynamic strength reduction [15]. In Figure 6e, post stress relieving for the 45◦ orientation samples, the unmelted powder disappeared but there seemed to not be an apparent improvement in the mechanical properties. This was observed in the tensile results for orientation B, which had the highest elongation post stress relieve.

Figure 6c (90◦ orientation) specimen showed a ductile fracture with dimples, which suggests forced fracture. It was determined that there were micro-cavities formed where there were defects visible in Figure 6d–f post stress relieve [15,25,31,32]. These micro-cavities lead to more micro-cavities that join together to cause a fast growing tear in the structure. The tear spreads laterally to the interface between the melt pool core and the boundary, constantly along the fracture sides. This is due to the fact that the melt pool boundary is weaker than the melt pool core, containing a coarser microstructure and minimal content of silicon for the reduced grain boundary, which were also observed on the OM and EBSD microstructures. Overall post processing by stress relieve has shown the microstructure of the fracture surfaces of all three different orientations were similar. Ductility dimples are virtually the same with voids that develop and merge together. Zaretsky et al. [33] believe that the homogeneous spreading of sites is suitable for void nucleation all over the sample and these are the characteristics of SLM processing.

#### *3.7. Fracture Toughness and Fatigue Crack Growth Rate Analysis*

The stress relieved samples for the 90◦ orientation were further analyzed for fracture toughness and crack growth in order to determine the effect that the heat treatment process has on the properties. One orientation was chosen because all samples after stress relieve exhibited more or less the same mechanical properties. The samples were machined to dimensions specified in Scheme 1 and ASTM399, before undergoing the tests. According to Rosenthal [34], as-built AlSi10Mg relative to traditionally produced AlSi10Mg exhibited inferior fracture toughness properties so the samples were stress relieved before testing.

Figure 7 was plotted based on the Paris equation: da/dN = C (ΔK)m and the three states were marked in the graph. The Kq obtained were 29.51, 30.47 and 29.99 MPa.m1/<sup>2</sup> for samples 1 to 3 respectively according ASTM399 standard for fracture toughness. Results show correspondence with those of Rosenthal [34], which were 30.4 MPa.m1/<sup>2</sup> after stress relieving. These results show that even though the stress relieving profile use in this case resulted in a decline in strength, it increased the fatigue life of the samples as presented in Figure 8. The samples have dimple fractures, which Kobayashi [35] equated to nucleation–growth–coalescence of voids. According to Brandl [15], additively manufactured AlSi10Mg demonstrated an increase in fatigue life after thermal treatment compared to cast AlSi10Mg.

**Figure 7.** Fracture toughness results of the SLM produced AlSi10Mg samples.

A representative sample was chosen from the three samples and is presented in Figure 8; it shows the propagation until the edge of the sample. The cleavage fracture was observed in these samples as "particle like" brittle phases where there were holes. These are called Griffith-like-microcrack as studied by Ruggieri and Dodds Jr. [36]. This is the microcrack that is nucleated then promptly spreads into the inner grain boundary to cause a fracture when it is not blocked by any obstructions in the grain boundary. It was observed in the gentleness of the steep of the graph in Figure 7 that the fracture was gradual. Therefore these samples failed due to nucleation overpowering various consecutive obstructions in the grains structures [37].

**Figure 8.** Illustration of the three fracture toughness samples (**a**), (**b**) and (**c**), SEM images of the SLM produced AlSi10Mg built in 90◦.

The fatigue crack growths results show that the samples were able to sustain the load for up to 75,000 cycles. The samples exhibited brittle propagation then dimples where it seemed to have tried to endure further before yielding to cracking. Aluminum alloys exhibited ductile fracture that formed dimples even though formation of dimples is dependent on shape, properties, volume fraction and condition of the particle matrix [35].

Figure 9 presents fatigue sample 1 (a,b,c), sample 2 (d,e,f) and sample 3 (g,h,i) of the fatigue crack growth sample. It was observed in sample 1 numerous pores that were anticipated to have assisted the crack to grow faster relative to sample 2 and 3.

**Figure 9.** Crack growth rate results of the SLM produced AlSi10Mg samples.

There seem to have been an increase of propagation possibilities even after a certain number of obstructions in the grain structure that arose from the pores [35,36] resulting to rapid failure. Samples 3 displayed a smoother propagation before fracture relative to the other two samples as also observed in Figure 10. Defects and pores at most were mainly the cause of rapid failure on components since

generally cracks were initiated where there were defects of some sort in the structure such as oxides in the case of SLM produced AlSI10Mg [37].

**Figure 10.** Crack growth rate SEM results of the SLM produced AlSi10Mg samples where ((**<sup>a</sup>**–**<sup>c</sup>**) is sample 1, (**d**–**f**) is sample 2 and (**g**–**i**) is sample 3).
