3.1 Microstructure
Figure 4 shows the typical morphologies of the LPBFed AlSi10Mg specimens, including “horizontal section” and “vertical section”. For the “horizontal section”, the distribution of the molten pools was tightly arranged, which was consistent with the chessboard scanning. Besides, the width of the molten pools was about 138 µm, and the shape of molten pools was oval in Fig. 4(a). In addition, the light-color precipitates were distributed at the molten pool boundaries. For the “vertical section”, the molten pools were crescent-shaped in Fig. 4(b). Apart from that, the molten pools related to the powder or the solidified part, resulting in the high cooling rates at layer-to-layer transition regions. Therefore, there were obvious differences of microstructures for the LPBFed AlSi10Mg samples were caused by the LPBF principle and the chessboard scanning.
Figure 5 shows the inverse pole figure (IPF) maps on horizontal section, IPF maps on vertical section and pole figure (PF) maps on the horizontal section under different conditions including as-printed, T6 treatment, T2 treatment and T6-T2 treatment. Figure 6 shows the grain size distribution histograms on horizontal section and vertical section under different conditions including as-printed, T6 treatment, T2 treatment and T6-T2 treatment. It can be observed from Fig. 5 that the as-printed AlSi10Mg specimen on vertical section presented a typical columnar grain structure on vertical section and a strong < 001 > texture. However, the as-printed AlSi10Mg specimen on horizontal section mainly consisted of equiaxed grains. By comparing Fig. 5(a) and Fig. 5(c), it can be found that the T2 sample on vertical section remained < 001 > texture parallel to the building direction, but the average grain size became larger. After the heat treatments, the average grain size on horizontal section had little change as shown in Fig. 6. The formation of equiaxed grains would interrupt the growth of the columnar grains resulting in decreasing the length of some of columnar grains. Besides, the intensity of the < 001 > fibre texture was increased after heat treatments, as the maximum pole-density projection values of {001} plane were larger than those of the as-printed sample, as shown from Fig. 5, which was different from the result of the LPBFed AlSi10Mg specimens after T6 heat treatment [21]. Notably, the proportion of equiaxed grains on vertical section increased, which was favorable to reduce anisotropy degree of the mechanical properties.
Figure 7 illustrates the microstructures, including “horizontal section” and “vertical section” after different heat-treated conditions. The heat treatment could significantly affect the difference of microstructure between “horizontal section” and “vertical section”. Moreover, to reveal the effect of the heat treatment on the element distribution, Fig. 8 displays the spot EDS measurements of the “horizontal section” specimens and comparison of Si content in Si particle and Al matrix.
For the as-build specimens, according to the Al-Si binary phase diagram [21, 28], the solidification pathway of the LPBFed AlSi10Mg has experienced a crystallization reaction L→L + α(α-Al matrix) and a eutectic reaction L→α(α-Al matrix)+ β (eutectic Al-Si phase) at 577℃. The primary α-Al matrix was first generated (see Fig. 7a). Apart from that, the eutectic Al-Si phase was generated and distributed along the molten pool boundaries with the decrease of temperature. Besides, the microstructure at the center of the molten pools performed typical cellular dendrite structure. The existence of the dark area was resulted from the generation of the α-Al matrix, in which the mass of Si was about 10.26%, as shown in Fig. 8. However, the mass of Si in the white bands increased to 14.64% and generated as the eutectic Al-Si phase. Additionally, there was little difference between the content of Mg element in the grain and the grain boundary, indicating that Mg has no segregation tendency and is evenly distributed in the matrix.
As expressed by Wei et al. [20], the sequence of precipitation of the LPBFed AlSi10Mg specimens during heat treatment was: AlSSS (supersaturated solid solution) → aggregation of Si/ Mg atoms (GP zones) → dissolution of Mg → formation of Si/Mg atoms cluster → a small number of unknown phases → precipitation of β´´phase → precipitations of β´ and β´´ phases → stable precipitation of Si particles.
Apart from that, Rao J. et al. [29] found that the network Si phase of A357 alloy had disappeared, after holding at 535℃ for 15 min. with the extension of time, the Si begined to gather and grow. Besides, the Size of Si particles was the largest at 24h. However, As the holding time continued to increase, the Si particles in α-Al matrix no longer continued to grow.
The process of partial dissolution, fracture and passivation of eutectic Al-Si phase could be explained by the Gibbs-Thompson effect, expressed as:
$$\text{l}\text{n}\frac{{C}_{R}}{C\left({\infty }\right)}=\frac{2\sigma {V}_{m}}{RTr}$$
1
Where C represents the solubility of the solute, r represents the radius of the particle, σ represents the surface tension on the particle boundary, \({V}_{m}\) represents molar volume, R represents the molar gas constant and T represents the thermodynamic temperature.
The increment of free energy of the particle is expressed as:
$$\varDelta p=2\sigma /r$$
2
where σ represents the surface tension on the particle boundary, and r represents the radius of the eutectic Al-Si phase.
According to Formula 1 and 2, it is found that the smaller the radius of the eutectic Al-Si phase is, the greater the curvature is, the greater the free energy is, the easier the diffusion is. So, the grooves or corners of eutectic Al-Si phase are easy to diffuse and dissolve, and the corners of eutectic Al-Si phase are passivated to make the outer surface of dendrites smooth in the dissolution process.
The T6 treatment transformed the eutectic Al-Si phase into Si particles uniformly distributed in α-Al matrix as shown in Fig. 7(b). Beyond that, the atomic distortion of supersaturated solid solution of the eutectic Al-Si phase called β´ phase or β´´ phase generated by solution treatment might result in large residual stress and hinder the movement of dislocations, where would store high distortion energy. Moreover, the β´ phase and β´´phase in an unstable state would form the new phases easily, in order to reduce distortion energy. Thus, during the T6 treatment process, the eutectic Al-Si phase was finally transformed into Si particles after first fused and then spheroidized. Besides, the mass fraction of Si in the particles could reach 31.48wt% whereas the mass fraction of Si in the Al matrix was about 9.41wt% in Fig. 8.
During the T2 treatment process, the Si particles had enough time to precipitate under the condition of high temperature. As shown in Fig. 7(c), the morphology of Si particles was small and spherical. The size of Si particles spheroidized decreases to nanoscale. Moreover, the Si particles were fully refined and evenly distributed in the Al matrix. Meanwhile, the mass fraction of Si in the particles could reach 35.93wt%, while the mass fraction of Si in the Al matrix was about 10.08wt%, as displayed in Fig. 8.
For the samples treated by the T6-T2 treatment, the spherical Si and flake Si were distributed in the Al matrix in Fig. 7(d). According to the XRD results, the contents of Si in the particles and the Al matrix were completely different. To be specific, the mass fraction of Si in the particles reaches 51.68wt%, whereas the mass fraction of Si in the Al matrix was about 6.92wt% in Fig. 8. Compared with the LPBFed samples, the trend of Si distribution along the molten pool boundary was significantly weakened after the heat treatments. Besides, the color of Si particles became shallower, and the mass fraction of Si in the particles was the largest by T6 - T2 treatment than the other heat treatments as shown Fig. 8, indicating that the Si element experienced the whole process of “fully diffusion - nucleation-precipitation - growth” and reached a stable state finally.
3.2 Mechanical Properties
The obvious differences in the properties of the LPBFed AlSi10Mg under different conditions are determined by the evaluation of the microstructure, as shown in Fig. 9. Obviously, the mechanical properties of the horizontal direction were better than those of the vertical direction, due to the distribution of the microstructure in layers and the weaker bounding force between layers in Fig. 4. However, the LPBFed AlSi10Mg specimens were easy to form the Al supersaturated solid solution, due to the fast-cooling rate in the LPBF process. The mechanism of solid solution strengthen was good for improving the strength of the LPBFed AlSi10Mg specimens, but the disadvantage was poor plasticity.
Furthermore, the heat treatments play an important part in improving the plasticity at the expense of strength. Rao J et al.’s study [29] suggested that the strength of LPBFed A357 alloy decreased and the plasticity increased after T6 treatment. Besides, Han Q et al. [30] found that the yield strength of the LPBFed AlSi10Mg specimens was reduced from 200 to 100MPa while the elongation increased from 6 to 22%, after soluiton-treated at 550℃ for 2h. In comparison to the LPBFed AlSi10Mg samples, the strength of the T2 treatment samples reduced by over 50%, while the elongation was twice as large as the LPBFed samples. For the T6 -T2 treatment, the strength was further reduced and the plasticity was also increased, related to the formation of Si particles as shown in Fig. 7. After the T6-T2 treatment, the network-like Mg2Si particles were dissolved and broken to grow the finely disappeared Si particles resulting in weakening the effect of hindering the motion of the dislocation, which would increase the plasticity. However, the splitting effect of the finely disappeared Si particles on the α-Al matrix increased, which would decrease the strength.
In contrast, the properties of vertical specimens by T6 treatment were a tensile strength of 292 MPa, a yield strength of 245 MPa, and a percentage elongation after the break of 11.8%. Furthermore, the properties of horizontal specimens were a tensile strength of 298 MPa, a yield strength of 250 MPa, and a percentage elongation after the break of 13.1%. This suggests that the T6 treatment can not only eliminate anisotropy of the LPBFed AlSi10Mg specimens basically, but also obtain a better combination of strength and plasticity. In order to understand the fracture mechanisms of the T6 treatment samples including “horizontal” and “vertical” ones, fracture morphology and EDS analysis of tensile specimens were carried out using SEM, as presented in Fig. 10. There was no defect on the tensile fracture of the vertical samples in Fig. 10(a-c). Besides, the distribution of the dimples closely connected on the tensile fracture indicates that, the fracture mode is mainly arbitrary fracture. Apart from that, the dispersed precipitation particles were found on the bottom of the dimples, which was favorable to increase the tensile properties.
The fracture morphology of horizontal specimens is shown in Fig. 10(d-f), and obvious pores appeared on the tensile fracture. However, the dimples on the tensile fracture reveal that the fracture mode is dominated by the arbitrary fracture. Likewise, the bottom and side walls of the dimples were dispersed by the enhanced precipitates. At the same time, the morphology of pores that appeared on the fracture was enlarged to investigate its mechanism. Besides, the inner wall of the pores was smooth, where there were irregular particles. The EDS analysis was performed on the inner wall of the pores to clarify its chemical composition in Fig. 10(g-h). The oxygen content in this area reached 17.4% owing to the splash generated by the laser during the LPBF process. It is noteworthy that the splash produced is easy to fall on the formed part during the LPBF process of the horizontal specimens. The splash partly melted together with the next powder layer may remain in the specimens, but this is related to the cross-sectional area of specimens. The larger the cross-sectional area is, the greater possibility of generating the pore defects caused by the splash is.
The kernel average misorientation (KAM) was analyzed according to EBSD results and performed in Fig. 11. The average value of KAM on vertical section (0.6916) was smaller than that of the horizontal section (0.75975) for the LPBFed AlSi10Mg samples, as shown in Fig. 11(a). In general, a higher average value of KAM indicates a higher dislocation density; therefore, a higher dislocation strengthening in the sample. Therefore, the strength and plasticity of the vertical section was smaller than that of the horizontal section for the LPBFed AlSi10Mg samples.
For the T6 treatment samples, as shown in Fig. 11(b), the average values of KAM on vertical section (0.72425) and horizontal section (0.73675) were similar. For the T2 treatment samples, it can be observed from Fig. 11(c) that the average values of KAM on vertical section (0.66272) and horizontal section (0.66037) were similar. For the T6-T2 treatment samples, it can be observed from Fig. 11 (d) that the average value of KAM on vertical section was 0.6637, while that on horizontal section was 0.78397. It can be found that there was smallest difference in the average value of KAM between the vertical section and the horizontal section for the T6 samples, which was consistent with the difference of tensile strength and plasticity. That is to say, the T6 treatment weakens the direction difference to the greatest extent from dislocation strengthening and metal deformation mechanism. Besides, the average value of KAM for T2 heat treatment was the smallest, suggesting that the T2 treatment minimizes the stress concentration.