Effect of the Distance from Build Platform and Post-Heat Treatment of AlSi10Mg Alloy Manufactured by Single- and Multi-Laser Selective Laser Melting

Figure 2 shows the fine and inhomogeneous microstructure of SLM samples analyzed in the XY plane and in the XZ plane. As widely reported in the literature, the first plane mentioned is characterized by the characteristic ellipsoidal shapes of molten pool, elongated in the scanning direction; the second shows a “fish scale” structure along build direction. In the next layer, the laser scanning direction is rotated by 90°, as confirmed by the transversal line perpendicular to melt pools. Figure 2(a) and (b) shows the differences between the single laser case (Fig. 2a, D bar) and the multi-laser case (Fig. 2b, AB bar) in XY plane: it is possible to observe the greater number of different scan track directions in Fig. 2(b). In addition, all micrographs (Fig. 2) are characterized by the random presence of spherical pores that will be discussed later, but in Fig. 2(c) a keyhole pore is showed and it is located between two scan tracks at the edge of molten pool. The microstructure of molten pool (Fig. 3) is characterized by an increase in cell size from the center to the boundary, as widely reported in the literature. Considering the three SEM pictures showing the Fig. 3(b, c, d) (areas 1, 2 and 3 in Fig. 3), the eutectic Si network is gradually destroyed from the center to the coarse and heat-affected zones (HAZ) (Ref 21, 22).

Figure 4 shows the molten pool boundary and the overlap region among two adjacent track scans, which increases if the hatch distance decreases (Ref 22, 23). In this case of study, the sample was scanned with long bidirectional vectors in one layer with a hatch distance of 170 μm. On the other hand, considering the high magnification micrograph shown in Fig. 4(b), it can be emphasized the effect of temperature gradient (G) and the crystal growth rate (R) within the molten pool: the grains growth takes place along the thermal gradient and their orientations change within the molten pool because the G direction varies (Ref 5, 15, 24‐26). As matter of fact, from the bottom of molten pool, the G direction remains perpendicular to its boundary; on the contrary, the G direction will be deflected toward the center of the melt pool from the side of the same boundary (Ref 25). In fact, the cross section of molten pool is mainly composed of columnar grains and by small equiaxed grains near melt pool boundary. In the same context, considering a melt pool in the last scanned layer n, the grains direction can change from 0° to 90° compared to that which characterized the melt pool in the layer n-1 (Fig. 4b), as reported by D. Kong et al. (Ref 26). Finally, the characterization of this cellular structure can be done considering the G/R ratio and the product of G and R: the morphology and the size of solidification structure were described by the first and the second values, respectively, as reported by (Ref 25‐28).

Image analysis results in Fig. 5(a) shows that the bottom part of the multi-laser case (AB-bottom) has 60% of its porosity characterized by an equivalent diameter smaller than 1 μm, about 17% has a size ranging between 1 and 2 μm and about 7% is between 2 and 3 μm, decreasing to zero for larger areas. The overall density is 99.9% on overage. At the top of the same bar (i.e., approximately at the maximum distance from the pre-heated platform surface), porosity presents a similar size distribution: 40% of the porosity has an area smaller than 1 μm, about 20% between 1 and 2 μm and 15% ranges between 2 and 3 μm. The density is 99.2%.

Considering the single laser case (D-bottom) in Fig. 5(b), the bottom samples are characterized by 25% of porosity with equivalent diameter smaller than 1 μm, whereas 20% of the pores have diameter in the 1-2 μm range and 25% have a diameter between 2 and 3 μm. The density is 99.0%. As regards the top samples, pores with an equivalent diameter lower than 1 μm account for 45% of the overall porosity; this decreases to 17% and increases again to 22% for the diameter ranges 1-2 μm and 2-3 μm, respectively. The density is 98.5%.

The frequency of pores with equivalent diameter greater than 3 μm tends to 0 both for single and overlapping samples. However, it is possible to observe that the multi-laser case has the highest frequency of smaller pores, as also reported by Li et al. (Ref 6). Moreover, bottom samples show slightly lower porosity than top samples as also reported by (Ref 29); at the same time, in the bottom section there is a higher frequency of small pores (equivalent diameter < 1 μm) than at the top one. As regards the reported values, Aboulkhair et al. (Ref 22) and Buchbinder et al. (Ref 5) suggested a modification of scan strategy and an increase in laser power up to 1 kW, respectively, to increase density. Moreover, the density of as-built AlSi10Mg SLMed samples was between 95.3% and 99.9%, as reported by Lam et al. (Ref 9).

In the as-built conditions, very fine eutectic silicon is found at the border of α-grains, as shown in high magnification SEM micrographs of Fig. 3. In detail, the average eutectic thickness ranges from 0.045 to 0.065 μm in both kinds of samples (SL and ML) with a standard deviation of 0.020 μm as shown in Fig. 5(c) and (d). Considering such small thickness and the error associated with the measurements, the statistical distributions are similar at the top and bottom in the single and overlapping laser samples. So, in terms of eutectic Si distributions, there are no great microstructural variations in relation to machine setup (single or multi-laser) comparing AB and D bars, as also reported by Zhang et al. (Ref 24). It is inferred that pre-heating the platform at 150 °C does not induce a significant coarsening of eutectic Si. A higher platform temperature, on the other hand, could have caused coarsening of eutectic Si, as reported by Fiegl et al. (Ref 30) and Buchbinder et al. (Ref 31) through experiments with a platform heated at 200 and 250 °C.

Analysis of XRD patterns performed on top and bottom SL samples (Fig. 6) confirms that there are no significant differences in terms of structure. The Mg₂Si phase was detected through a weak peak at an angle (2θ) of 40.4° in both patterns as reported by (Ref 7, 9).

Figure 7 illustrates the microstructural differences inside the α-grains of top (Fig. 7a and b) and bottom regions (Fig. 7c and d) of the billets, in single (left column) and multi-laser samples (right column). There are small precipitates phase inside α-Al grains at the bottom (Fig. 7c and d), while precipitates are not visible at the top section (Fig. 7a and b) at the magnification of 80 kX. Higher-resolution views (Fig. 8) at 100 kX clarify that fine, sub-micrometer precipitates (arrows) do exist in both bottom (a) and top (b) samples, but they are more numerous in the former (Fig. 8a). However, it should be remarked that the amount of precipitates is not fully homogeneous: in both areas, the amount of precipitates varies among different grains, probably because of slight differences in thermal history, e.g., among center, coarse zone and HAZ as discussed previously. Considering two different studies of Kim et al. (Ref 32) and Hadadzadeh et al. (Ref 18), the fine precipitates visible within the α-Al grains can be both Si precipitates and Mg₂Si phases; so, this condition represents an additional strengthening of the matrix. All of these are also in agreement with the researchers reported by R. Casati et al. (Ref 15) and Li et al. (Ref 16). In this context, Y. Ji et al. (Ref 33) have demonstrated that the SLM process creates a supersatured solid solution (SSS) of Si in α-Al matrix due to the solidification process of molten pool. As matter of fact, the high cooling rate (10⁶-10⁸ K/s) reduces quantity of Si rejected into the liquid. With the successive scanned layers, the Si particles precipitate and grow in the previous solidified layer due to the high temperature that is generated Ref (33‐35). T. Kimura et al. Ref (34) reported the same scenario with a SSS formed by Si and Mg solvents.

Figure 9 shows the Vickers microhardness measured along the build axis (or Z-axis) of the two bars. Vickers microhardness is higher at the bottom of the bar (~ 125 HV at 0-30 mm distance from the bottom for both bars), but at the maximum height (300 mm), it decreases to 110 HV and 103 HV, respectively, for AB and D bars. In terms of time and according to HV values reported in Fig. 9, HV decreases by 12% and 16% for the single and multi-laser samples, respectively, in 40 h (i.e., the job duration time) during exposure at 150 °C (platform temperature).

The tensile properties measured in samples machined at different heights of the bar (according to Fig. 1a) are illustrated in Fig. 10(a) and (b), as a function of the distance from the platform (distance along Z-axis). The yield strength and the UTS decrease in samples extracted at higher Z, ranging from 290 to 220 MPa and from 440 to 360 MPa, respectively, for the Single laser case. For the multi-laser case, both σ_0.02 and UTS vary from 300 to 190 MPa and from 445 to 345 MPa, respectively. The greatest variation was, therefore, obtained in multi-laser samples (+ 110 MPa for yield strength and + 100 MPa for UTS), compared to the variations obtained for single laser samples, as reported in Table 2. The % engineering elongation, ε, shows average values of (7 ± 1) % in both kinds of samples and over the entire 300 mm-long investigation region along the Z-axis (Fig. 10). Unlike tensile strength, ductility is not affected by increasing distance along the build axis.

Hardness and tensile strength values are closely related to precipitation strengthening phenomena depending on distance from the heated platform as reported in Fig. 7 and 8 and by (Ref 7, 8, 9). In fact, there is a difference between the top and bottom samples in terms of intra-granular precipitates (Si or Mg₂Si phase), as explained previously. This can be regarded as an aging effect due to prolonged exposure to 150 °C. Yang et al. (Ref 36) confirmed that Mg₂Si phase induces a little strengthening at 160 °C for 8h. At the top of Z-axis, i.e., farther away from the heated platform, tensile strength and HV are lower because the matrix contains fewer precipitates. Therefore, it is mainly hardened by solid solution, which is a less effective mechanism than precipitation strengthening. The slight increase in strength (+ 17 MPa for SL case and + 27 MPa for ML case) of the last top samples (Fig. 10) can be attributable to lattice distortion due to Si dispersed in α-Al matrix during SLM process as previously reported (Ref 15, 24, 37). In fact, these samples are not affected by build platform which induces a stress relaxation and/or an overaging, as reported by (Ref 15, 16, 38).

The smaller decline of strengths along the Z-axis in the SL samples compared to ML samples could perhaps be also attributable to lower coarsening of the microstructure due to lower heat input. Anyway, the effect is barely visible by SEM (Fig. 7). In fact, mechanical properties and microstructure are largely comparable between SL and ML samples. In this context, the ML samples are affected by a greater amount of thermal energy than SL samples due to re-melting or partial re-melting of solidified tracks (Ref 15, 39). This situation induces a decrease in anisotropy that characterized the as-built samples; in fact, the thermal gradient induces a <100> texture orientation of microstructure along build direction, as reported by D. Dai et al. (Ref 35) and Guan et al. (Ref 39). Similarly, C. Zhang et al. (Ref 24) show a greater isotropy for the overlap samples (18.3% of the area) than single laser samples (12.6%). So, also considering the effect of build platform as previously reported, it can be explained how ML samples have a greater loss in strength than SL samples in the middle part of two bars printed. Therefore, in the following paragraphs, only the results of single laser samples will be considered.

Figure 11 presents the trend of tensile strengths and elongations for the AlSi10Mg samples subjected to unconventional treatments at 175, 200 and 225 °C for 6 h, designated as U-HT1, U-HT2 and U-HT3, respectively, for samples at the top and at the bottom. Values of mechanical properties are also listed in Table 3.

From Fig. 11, the gap reduction of tensile properties between the top and bottom samples is achieved by the U-HT3. In the bottom samples, which had already been aged at 150°C during the build process itself (as explained previously), additional aging due to the treatment at 175 °C has a barely perceivable effect. In the top samples, which were less affected by the pre-heated platform, the aging effect is comparatively more significant. In fact, after the U-HT1 treatment, the top and bottom samples develop comparable microstructures. Both are rich of fine precipitates within the α-Al matrix (Fig. 12a, b). A comparison between Figs. 8(b) and 12(b) particularly highlights the increased amount of fine, intra-granular precipitates in the top sample. The aging condition of both the top and bottom samples has thus become comparable after this heat treatment. On the other hand, the treatment seems not to affect the network of eutectic Si, which remains comparable to the one seen in as-built condition (compare Figs. 12 to 8). Fig. 13 shows that U-HT3 induces an increase in the size of eutectic Si compared to as-built condition. SEM micrographs shown in Fig. 14(a) and (b) highlight the incipient destruction of the Si eutectic network and confirm the coarsening of Si particles compared to Figs. 8 and 12.

After the aging treatment at 225 °C (U-HT3), the UTS and yield strengths have a maximum decrease compared to as-built AlSi10Mg samples for top and bottom regions, while ductility has a maximum increase. In detail, the largest variations occur for the bottom samples and are 24% and 26%, respectively, for UTS and σ_0.02. Fiocchi et al. (Ref 14) and Mertens et al. (Ref 40) reported comparable results.

Consistently, XRD patterns in Fig. 15(a), (b) and (c) show that the Si peaks become somewhat more intense and narrower after U-HT3 treatment, compared to the as-built condition, which confirms the coarsening of Si. Further, the relative intensity of the different diffraction peaks of α-Al changes after the U-HT3 treatment: the (200) peak at 44.7° becomes more intense than the (111) peak at 38.5°. This indicates that the α-Al phase is undergoing a change in crystal orientation.

The greatest UTS loss of 74 MPa is, therefore, explained by the combined effects of coarsening of the Si precipitates and interruption of Si eutectic network. The latter phenomenon allows easier movement of dislocations through the neighboring α-Al grains: dislocations are less likely to interact with coarser eutectic Si. Consequently, the strength needed to break the tensile specimen is also reduced; in fact, the required energy for plastic deformation must not be dissipated in the deformation of eutectic Si as reported by (Ref 32).

Due to such major microstructural re-arrangement, the values of tensile strengths become completely comparable between top and bottom samples. Any difference between the aging conditions of top and bottom after the build process is indeed lost through the alterations caused by the U-HT3 treatment. As regards the ductility, the results obtained on the U-HT3 samples can be regarded as optimal values, when compared both to the other samples tested within the present work (as shown in Table 3) and to literature data pertaining to analogous thermal treatment conditions (Ref 6, 14, 30, 41).

The U-HT2 condition, on the other hand, seems to provide an optimum compromise between retaining the mechanical properties and providing acceptable values of elongation, although the effect of the distance from the build platform is retained.

This compromise is achieved as the unconventional heat treatment causes a slight increase in the size of eutectic Si (Fig. 13), as well as a slight coarsening effect on the precipitates within α-Al grains (Fig. 14a, c), but it does not cause a destruction of the Si eutectic network (Fig. 14a, c), compared to the as-built condition (compare to Fig. 8).