PSI - Issue 7

Theo Persenot et al. / Procedia Structural Integrity 7 (2017) 158–165 Persenot et al. / Structural Integrity Procedia 00 (2017) 000–000

163

6

a)

b)

Sudden failure zone

b)

Stable crack growth zone

c)

c)

Fig. 6. SEM micrograph of the fracture surface of an as-built EBM fatigue sample ( σ max = 160MPa, N R = 10 6 cycles, R = 0.1). a) Global view. The region corresponding to the stable growth of the final crack is visible on the left side of the nearly vertical dashed line. The zone corresponding to the sudden failure zone is on the right side of this line; b) and c) Zoom on the two crack initiation sites.

4.2. E ff ect of chemical etching

Figure 4 shows the evolution of the specimen volume and roughness as a function of etching duration. After 45 minutes of etching, the Ra and Rt values decreased from 44 µ m (SD = 2.2 µ m ) and 390 µ m (SD = 39 µ m ) to 26.6 µ m (SD = 1.7 µ m ) and 228 µ m (SD = 24.5 µ m ) (reduction of 40% and 42% respectively). In the mean time, the volume changed from 28.3mm 3 (SD = 0.33mm 3 ) in the as-built cylindrical zone to 18.7mm 3 (SD = 0.2mm 3 ) after 45 min utes of chemical etching which represent a 32% loss. It is interesting to note that the etchant is still e ffi cient after 45 minutes even though its impact on the roughness seems to be disminishing between 30 and 45 minutes. Lhuissier et al. (2016) and Sun et al. (2016) have shown on the same material that after 60 to 90 minutes, the etchant e ffi ciency tends to level o ff . This “plateau” is probably not yet reached in our case. Tomographic scans allow to assess the impact of the chemical etching on the surface geometry. The chemical etching process is not homogeneous on the sample surface: partially melted powder stuck to the strut are first removed and in a second step the “plate-pile” stacking defects are smoothed. Lhuissier et al. (2016) link this discrepancy in dissolution rates to the local density of etchant near the surface. Convex surfaces (such as the stuck powder) are more impacted by the dissolution because of a higher local density of etchant in those regions. In our case, it is worth emphasizing that the notch-like defects responsible for the failure of as-built samples are not made more critical (deeper) by this post-treatment. On the other hand, the volume reduction brings internal defects such as pores closer to the surface and, hence, increase their propensity to initiate cracks. As expected, the surface modification induced by chemical etching on the surface state increases the fatigue perfor mance. The fatigue limit i.e. in this case the fatigue strength at 10 7 cycles is increased to a value of 220MPa after 30 minutes of chemical etching (60% increase with respect to as-built samples (fresh powder)) but an important scatter of the fatigue lives is observed: some samples are observed to fail around 200 000-300 000 cycles while others achieve more than 2.10 6 cycles. After 45 minutes of chemical etching, the same fatigue limit of 220MPa is obtained but the scatter of the fatigue lives is no longer observed. In term of crack initiation mechanisms, after 30 or 45 min of chemical etching, crack initiation still occurs at the same type of surface defects as in as-built samples but their size are smaller than the ones observed on as-built fracture surface. As pointed out on figure 4, even though the etchant e ffi ciency seems to be somewhat decreased after 45 min, it still has a significant impact on the roughness. This indicates that there is room for further improvement on the fatigue limit.