According to Hunts steady state model , the CET mainly depends on three factors: the thermal gradient G, the solid/liquid interface velocity V and the number of potent nucleants ahead of the columnar front. Such analysis has been performed for AM Nickel alloy 718 by Raghavan et al. for electron-beam melting process . Based on a heat-transfer model and considering a constant nucleation density of 2.1015 nucleus.m−3, the authors showed that CET is more likely to occur at the end of the solidification of the layer, i.e. close to the free surface. Contrariwise,the high G and low V at the interlayer boundary is not favorable to a CET. Yet this is the position where the CET has been observed in this work, about 100 µmfrom the interlayer boundary. Assuming the G-V path determined by Raghavan et al. is valid, this means that the number of potent nucleus varies in the layer thickness, and is much more significant close to the interlayer boundary where the liquid phase comes from remelting of the n−1layer. From the twin frequency measured in this work, and following the analysis of Kurtuldu and Rappaz , ≈13% of twin GBs relates to fraction of icosahedral nucleants over the total number of nucleants of 10%, each icosahedral cluster giving birth to potentially 20 fcc grains. The ISRO-mediated nucleation mechanism induced by the fast melting can explain why a CET occurs in high G and low V regions, by increasing strongly the local number of nucleation sites close to the interlayer boundary. This shows that the melting step can be employed beyond its usual role in the building process.

Our explanation for the occurence of small grains in Nickel alloy 718 suggests a new route to refine the grain structure in alloys displaying ISRO mediated nucleation, in between the current strategies based on either the control of the thermal conditions , or the enhancement of nucleation. Indeed, the route based on ISRO mediated nucleation requires some melting that remains to be optimized. It is similar in spirit to what has been observed in Al alloys containing Sc, but with a different mechanism. Finally, it is worth noting that the reported grain sizes associated with the ISRO mediated nucleation seem in average lower than the sizes achieved by controlling the thermal conditions in Nickel alloy 718. Even if this observation must be consolidated by quantitative comparisons, it suggests that ISRO mediated nucleation is worth to be further investigated for AM of Ni base superalloys.

To conclude, Nickel alloy 718 alloy has been additively manufactured with a laser-wire co-axial process. Multiple equiaxe zones were observed in the sample. Based on detailed EBSD analysis, it has been shown that the equiaxed grains ensue from an ISRO-mediated nucleation mechanism, previously observed in Al-based and Au-based fcc alloys only. The icosahedral symmetry deduced from EBSD analysis of the fcc phase is expected to have originated from the fast remelting of the previous layer, providing metastable configurations in the liquid favorable to the nucleation of the fcc phase. These results open up new prospects for additive manufacturing of nickel based alloys. Indeed, controlling the ISRO nucleation mechanism in these alloys so as to obtain homogeneously fine equiaxed grains would (i) avoid solidification cracking, (ii) weaken texture to get isotropic mechanical properties, and (iii) benefit from the grain refinement to increase room temperature mechanical properties.