The first part of the study has the aim to evaluate the influence of the main parameters of the SLM process on the quality of the produced material. Part of this investigation has been previously published in [186]. The process parameters taken into consideration for this study are the laser power (P), the scan speed (v) and the hatching distance (hd). The response of the material at each combination of these parameters is evaluated using two quality indicators: porosity and Brinell hardness.
An estimation of the porosity level in the material is evaluated through apparent density measurements with the Archimede’s method and image analysis of optical micrographs. The measured apparent densities of each sample are reported in figure 3.1 (see table 2.2 for the SLM process parameters associated at each sample).
Figure 3.1. Apparent density measurements of the SLM samples produced for each set of process parameters combination in comparison with the theoretical density values. Data published in
[186].
It is possible to observe that all the tested samples are near full dense and that the process parameters have not a clear effect on the apparent density of the material in the considered ranges. In general, the presence of small and uniformly distributed pores from 5 to 10 m in size is observable in figure 3.2 (some examples of them are marked with blue ellipses). Some larger and usually round shaped pores of 20-30 m (like the one marked with the red ellipse in figure 3.2) are detectable in the collected optical images.
Figure 3.2. Optical micrograph of sample n.18 (P = 185 W, v = 900 mm/s, hd = 0.09 mm) as example of the general porosity condition of the SLM samples with presence of small pores (blue
ellipses) and some bigger rounded pores (red ellipse).
The porosities ranges obtained for each tested sample through image analysis are reported in figure 3.3. The results of this analysis confirm that the porosity level of all the sample is very low on average (< 0.25% apart samples n.4 and n.11). Not evident influence of the SLM process parameters on the overall porosity level can be deduced from the plot in figure 3.3. However, a careful inspection of the collected micrographs reveals that the sample obtained with the lower scan speed (v = 600 mm/s) tend to have some bigger spherical porosities (red ellipses in figure 3.4-A), probably due to trapped gas. A low scan speed determines a high absorbed energy, which in turn can actually cause partial vaporization of the molten pool leading to the formation of porosities due to the entrapment of the formed vapor.
Nevertheless, sample n.22, that is the only one out of those obtained with a scan speed of 600 mm/s that was fabricated applying a low VED, don’t exhibit a large number of gas porosities. Therefore, the low value of laser power and the high hatching distance allow to prevent the formation of bubbles even when v is low.
Figure 3.3. Porosity area fraction ranges obtained on the SLM samples produced for each set of process parameters combination. Data published in [186].
On the sample n.11, the revealed porosity is comparable to the general one shown in figure 3.2, however there are some zones, especially near the edges, characterized by an anomalous density of pores (figure 3.4-B). This sample is associated to the highest VED value between the ones obtained with a scan speed higher than 600 mm/s. According to the present analysis such conditions lead to an unstable consolidation process and, as a consequence, they should be avoided.
Figure 3.4. Sample n.15 (P = 175 W, v = 600 mm/s, hd = 0.07 mm) as example of the porosity condition (red ellipses indicate gas porosities) in most of the samples obtained with v = 600 mm/s
(A) and sample n.11 (P = 195 W, v = 900 mm/s, hd = 0.07 mm) obtained with the highest VED between the samples produced with v > 600 mm/s (B).
On sample n.4, associated to the lower VED value applied, the porosity level is the highest found. A large number of big voids, mainly irregularly shaped and probably caused by lack of fusion, are present on this sample (some examples of them are marked with red ellipses in figure 3.5-A). This kind of porosity can also be observed on sample n.16, but with a much lower frequency. With respect to sample n.4, the number 16 is fabricated at the same scan speed and hatching distance, but with a higher laser power (195 W). This reduces drastically the problem of lack of fusion (figure 3.5-B).
Figure 3.5. Sample n.4 (P = 175 W, v = 1200 mm/s, hd = 0.11 mm) obtained with the smallest VED value (A) and sample n.16 (P = 195 W, v = 1200 mm/s, hd = 0.11 mm) obtained with a
higher value of laser power (B). The red ellipses indicate some lack of fusion voids.
In general, abnormal porosity is not found on samples obtained with intermediate VED values.
As far as the Brinell hardness is concerned, all the collected mean values fall between 250 ± 3 HB10 (sample n.12, obtained with P = 185 W, v = 600 mm/s and hd = 0.07 mm) and 266 ± 3 HB10 (sample n.5, obtained with P = 175 W, v = 1200 mm/s and hd = 0.07 mm). The sample produced by adopting the EOS recommended parameters has a mean Brinell hardness of 264 ± 9 HB10, very close to the highest measurement.
The main effects of the SLM process parameters and of the second order interactions between them on the Brinell hardness are shown in the plots reported in figure 3.6.
Figure 3.6. Main and interaction effects of the SLM process parameters on the Brinell hardness of Inconel 718 alloy. Data published in [186].
Porosity can affect the hardness, however the observed correlation between the collected data is weak because also concur to this property, for example the microstructure and the development of second phases. Table 3.1 contains the ANOVA treatment relative to the hardness of the tested samples. Linear model based ANOVA is probably not very suitable for describing complex nonlinear
phenomena involved in the SLM process that can influence the final hardness.
However, some rules of thumb can be derived from this analysis:
• v is the parameter that mostly affects the hardness, which decreases when v is lower in the analysed range; this is probably again related to the development of gas porosities, as noted on samples produced with v equal to 600 mm/s;
• higher hardness is obtained at higher v and lower values of hd, the latter parameter being effective in reducing the risk to develop voids due to lack of fusion;
• it’s better to adopt a higher value of hd when a high P is used and, conversely, a lower value of hd leads to higher hardness when P is lower, the balance between these two parameters allows to obtain an intermediate value of VED and prevents the formation of abnormal porosity;
• intermediate value of P leads to lower hardness, especially if a low scan speed is adopted.
Table 3.1. Analysis of Variance (ANOVA) on the dataset made by 5 Brinell hardness measurements for each sample of the 33 factorial plane with the main effects of P, v and hd and their interactions until the second order. DOF: degrees of freedom; SS: sum of squares; MS: mean
squares; F: F-ratio.
The optical micrographs of the samples at the vertices of the factorial plane showing the microstructure revealed after etching with Kalling’s solution are reported in figures 3.7 and 3.8. The segregation between laser tracks and at grain boundaries appears more intense when the higher value of v is adopted, probably because it determines a faster cooling rate during solidification. Furthermore, some voids between laser tracks are visible on the sample produced with the lower VED value (P = 175 W, v = 1200 mm/s, hd = 0.11 mm).
Figure 3.7. Micrographs on the horizontal plane of the samples produced with v = 600 mm/s. The red ellipses indicate gas porosities and other defects. Kalling's n.2 etchant.
No further differences in the general microstructural features which could be clearly ascribed to the modulation of the SLM process parameters are detectable at this inspection length scale.
Figure 3.8. Micrographs on the horizontal plane of the samples produced with v = 1200 mm/s. The red ellipses indicate some lack of fusion voids between laser tracks. Kalling's n.2 etchant.
The above reported results demonstrate that the production of Inconel 718 through SLM process has a high robustness around the EOS recommended parameters. Near full dense material can be obtained without abrupt changes in
residual porosity level with a limited modification of the main process parameters.
This is especially true if extreme value of VED are avoided. Even the Brinell hardness is only slightly affected by parameters values, although in this case some marked effects were detected, especially due to modulation of the scan speed. For these reasons, samples for all the following characterizations reported in this thesis were fabricated through SLM process parameters recommended by EOS, that also according to our investigation were confirmed to be the optimal ones for densification purpose.