*4.1. Influence of Chemical Composition, Investment Casting Conditions, and Heat Treatment on Microstructure and Weldability*

Moulds E and O with higher casting solidification rates (1.65 ◦C/s) and standard and low Si contents, respectively, did not present any Laves phases in solution annealing state after complete HIP + solubilisation annealing heat treatment. Area percentage of Laves phase observed in moulds N, NP, and P in the solution-annealed condition ranged from 0.14 to 0.35%, without significant differences between moulds N and NP and being the residual content of Laves phase higher in mould P, that is, the heat with higher Si content and faster solidification rate. Note that this mould also had the highest amount of Laves phase in the as-cast condition (Table 2).

In moulds P, N, and NP containing residual Laves phase in the solution annealing state, the onset of ductility drop in on-heating hot ductility tests was triggered at lower temperatures, and corresponding fitting curves were shifted to the left side of the chart. This was observed in curves displayed in Figures 7a and 14, with the latter superposing the on-heating curves of the five studied heats. Note that dashed fitting lines corresponded to those three moulds. Therefore, this early ductility drop was associated with incipient melting of Laves phase. It is worth mentioning that melting of Laves phase is a fast event which does not require any significant reaction time to form the liquid as in constitutional liquation of NbC [2,6], and, consequently, it readily melts upon heating at very fast rates, as in current hot ductility tests (111 ◦C/s heating rate).

**Figure 14.** Comparison of on-heating curves in terms of area reduction percentage and testing temperature for moulds E, O, P, N, and NP.

Effective dissolution of Laves phase in alloy 718 castings prior to welding and high temperature use is pursued in industrial manufacturing processes because this phase impairs both weldability and mechanical properties [4,5]. This has been particularly shown in castings with volume fractions of secondary particles (including Laves and NbC) higher than the residual values reported in this work.

The chemical composition of remaining Laves phase was remarkably modified in moulds P, N, and NP after HIP and solubilisation treatment. The considerable enrichment in Mo could be explained by differences in diffusivities of solute elements in the austenite matrix. The dissolution kinetics of Laves phase in IN718 has been studied by [28]. In this work, the authors concluded that the back-diffusion of molybdenum in austenite is the controlling micro-mechanism for dissolution of the undesirable Laves phase. On the basis of the Johnson–Mehl–Avrami–Kolmogorov (JMAK) analysis at different temperatures, the authors determined the activation energy of 274.5 kJ/mol for the dissolution of Laves phase, which was close to the activation energy for diffusion of Mo in Ni (288 kJ/mol). This was also supported by the diffusion calculations shown in Table 9, which shows lower diffusivity values for Mo in Ni than for Nb and Ti.

**Table 9.** Parameters for diffusivity calculation of Nb, Ti, and Mo in Ni and calculated diffusivities.


It is also interesting to note that the chemical composition of remaining Laves phase in these three heats was comparable with only minor differences in Si content.

HIP and solubilisation treatment did not modify original grain size and morphology resulting from investment casting. Consequently, different cooling conditions during casting yielded different grain sizes and aspect ratios which remained in the base material employed for the weldability tests. Thus, moulds O, E, and P cooled without ceramic blanket had highly columnar grains elongated along plate thickness. On the contrary, HIP and subsequent solution annealing heat treatment was effective in reducing segregation in interdendritic regions, leading to comparable segregation ratios in the five heats (Table 5).
