The first DSC analysis was carried out to identify the melting and solidification ranges of the alloy. The presence of the non-equilibrium eutectics in the as built material can lead to the formation of liquid phase at temperatures lower than the theoretical expected one. Figure 4.1 reports the DSC analysis into the fusion range.
In comparing the heating and cooling curves, it is important to consider that the signal peaks associated to thermal phenomena are slightly shifted to higher temperatures during the heating run and to lower ones during the cooling stage. This is due to kinetics reasons and these shifts depend on the heating and cooling rates, respectively [210]. As a consequence, the melting and resolidification peaks (figure 4.1) are not perfectly overlapped because of these shifts.
Figure 4.1. DSC analysis across the melting/solidification range. The detected peaks are indicated by the arrows.
During the heating step, two distinct endothermic peaks are detected. They are probably due to the formation of the first liquid phase at about 1230-1250°C in the interdendritic zones, where the low melting eutectics products are preferably located, and the final melting of the phase at about 1320-1340°C, respectively.
For comparison, the melting ranges of Inconel 718 manufactured through conventional processes are 1205-1345°C for the cast and 1260-1335°C for the wrought products [66].
During the cooling stage, the exothermic peak related to solidification of the phase is the prevalent part of the collected signal. A very slight peak is also present at lower temperature and it is probably due to the occurring of the eutectic reactions that complete the solidification. The asymmetry between the heating and the cooling detected curves is due to the initial far from the equilibrium condition of the AM as built material, characterized by a large volume fraction of eutectic products. The solidification range during the cooling stage was assessed using a 28°C/min cooling rate, that is a very low cooling rate with respect to the one that occurs during the SLM process. Therefore, this cooling condition leads to a markedly different solidified microstructure and, in particular, to a lower level of eutectics formation than in real AM process.
Once determined the starting point of melting, thermal runs until 1200°C were carried out to detect the thermal phenomena that take place in the solid state using DSC and TMA combined techniques. DSC recorded curves for as built samples carried out at different heating rates are reported in figure 4.2. These samples were removed along the horizontal plane, i.e. on the plane perpendicular to the building direction. It is possible to note that the detected peaks become more intense and slightly shifted to higher temperature when the heating rate is increased. Therefore, a clearer signal of such peaks is obtained by adopting a higher heating rate.
Figure 4.2. DSC curves of heating and cooling relative to the horizontal samples. The arrows indicate the detected peaks.
The DSC analysis reported in figure 4.3 indicate that the growth direction of the sample doesn’t affect the DSC signal and related analysis, since no significant differences were detected between samples grown along the BD (vertical samples) or perpendicular to BD (horizontal samples).
Figure 4.3. Comparison of the DSC curves of vertical and horizontal samples at the heating rate of 5°C/min and 20°C/min.
The signals collected during the TMA analysis of the as built samples are reported in figure 4.4. In this experimental set, it is possible to observe that clearer signals can be generally obtained using lower heating rate, it is to say in the experimental conditions opposite to what applied for DSC analysis. Furthermore, the comparison between the TMA curves collected on vertical and horizontal samples shows a difference starting from about 870°C, in particular an extra contraction peak (marked as “Contraction 3”) is observable in the vertical sample and the last expansion peak (Expansion 2) is greater than the one detected in the horizontal peak. Therefore, an anisotropy effect can be revealed by TMA, whereas DSC analysis were not capable to detect any difference along the different directions analyzed. An explanation of this difference will be given later.
Figure 4.4. TMA heating curves of horizontal and vertical samples at the heating rate of 5°C/min and 20°C/min.
Apart from these differences in the sensitivity to anisotropy effects, the signals recorded with DSC and TMA analysis (Figure 4.5) are in good agreement and provide an efficient crosscheck. It is possible to observe that exothermic phenomena provide a reduction of the CTE, whereas endothermic ones lead to an increase of the CTE. Actually, the exothermic peaks are related to precipitation of second phases with consequent reduction of the solute dissolved in the phase.
Therefore, the lattice distortion of the phase is reduced during precipitation with consequent decrease of the volume detected by the TMA equipment. Conversely,
endothermic peaks indicate dissolution of second phases in the matrix that causes increase of the solute in the matrix and lattice expansion.
Figure 4.5. Comparison between the collected DSC and TMA curves at the heating rate of 5°C/min and 20°C/min. Dotted boxes indicate the overlapping signal peaks.
The above reported thermal analyses reveal the following thermal phenomena occurring in the as built material:
• precipitation of ’ (EXO 1) at 450-600°C;
• precipitation of ’’ (EXO 2) at 650-720°C;
• dissolution of ’ and ’’ (ENDO 1) at 780-930°C;
• precipitation of (EXO 3) at 860-920°C;
• dissolution of and preexisting second phases, in particular Laves phases, (ENDO 2) at 950-1040°C.
The above reported temperature ranges are obtained from the analyses performed with a heating rate of 5°C/min. They are quite compatible with the CCT curves of a conventional Inconel 718 alloy reported at figure 1.2. Furthermore, a similar sequence of exothermic peaks is also reported by Niang et al. [43].
During the cooling ramps, all conducted at the same cooling rate of 20°C/min, an exothermic peak is detected. Such peak is probably related to the co-precipitation of ’ and ’’. By increasing the heating rate of the heating stage (dotted arrow in figure 4.2), during cooling an extra exothermic peak become visible. This extra peak is probably related to the growth of precipitates. When the heating is faster, some nuclei remain after passing through the high temperatures and can grow during the following cooling stage. On the other hand, a slower heating dissolves completely the phase, which have no time to nucleate again during the cooling ramp.
The DSC analysis doesn’t reveal any difference between the response of the vertical and horizontal samples to the thermal ramps. Therefore, the difference that is observed in the TMA curves for vertical and horizontal samples (figure 4.4) cannot be due to an effective different evolution of the second phases. It is more likely to ascribe such effect to the starting heterogeneity and anisotropy of the as built material. The peak of contraction, that indicates solubilization at high temperature (ENDO 2), is lower in the horizontal sample. This occurs because, as already observed in the previous chapter, the second phases are not evenly distributed in the as built material, but they form predominantly at the interdendritic boundaries. Therefore, the solutioning of the pre-existing second phases leads to a lattice expansion that is not uniform in the material volume, but is greater at the interdendritic boundaries, where the release of solute in the matrix is more intense.
Since the dendrites are predominantly aligned along the building direction, the increase of the measured CTE would be greater along this direction, where the interdendritic boundaries are disposed in parallel on average. On the other hand, along the transversal directions, where the interdendritic boundaries are disposed in series, the measured CTE will be lower. The simplified scheme in figure 4.6 explains such concept.
Figure 4.6. Scheme illustrating the not uniform expansion during the dissolution of the interdendritic phases in the SLM material. The expansion is lower of a along the transversal
direction of the interdendritic boundaries.
A similar explanation can be given also to explain the suppression of the contraction peak indicating the precipitation of phase (EXO 3). As it would be shown later in this chapter, phase tends to form precipitates with plate-like morphology and with a certain relationship with the matrix (see paragraph 1.3.1.3) that leads to the formation of a parallelepiped grid oriented along the building direction (figure 4.35). Since the plate-like precipitates are not randomly oriented, their formation leads to non-uniform lattice contraction that is therefore revealed in the vertical samples but not in the horizontal samples.