Fine-structured aluminium products with controllable texture by selective laser melting of pre-alloyed AlSi10Mg powder

TL;DR: In this paper, the high thermal gradients occurring during SLM lead to a very fine microstructure with submicron-sized cells, which can be modified to a weak cube texture along the building and scanning directions when a rotation of 90° of the scanning vectors within or between the layers is applied.

About: This article is published in Acta Materialia.The article was published on 2013-03-01 and is currently open access. It has received 1431 citations till now. The article focuses on the topics: Texture (crystalline) & Selective laser melting.

Summary

1. Introduction

• This article discusses the material structure at micro as well as macro scale.
• In addition, also the influence of the scanning strategy is taken into account.
• The scanning strategy and more particularly the influence of the way of scanning on the produced microstructure and texture is investigated.
• This however, will not be dealt with in this article.
• This article will focus on the solidification structure and the texture formed in AlSi10Mg SLM parts.

3.1 Microstructure

• Outside the melt pool, the intercellular network is broken by coarsening of the silicon phase into idiomorphic particles.
• This is caused by the increase in diffusion rate of the Si in the heat affected zone.

3.2 Texture

• Between the border and the centreline, the grains are larger (around 6µm) and have a green-blue colour.
• And finally, large (around 12µm) reddish grains are present around the centreline of the melt pool,.
• When speaking in terms of orientations, small grains with random orientations are present at the melt pool border.
• At the centreline, grains have their <001> orientation parallel to the building direction and in between, the grain directions between <110> and <111> are oriented along the building direction.

4. Discussion

• The direction of scanning, however, is having a significant influence in SLM.
• When the scanning direction is rotated over 90°, the texture is significantly reduced and a weak cubic texture along the building direction arises.
• Furthermore, no difference is observed between rotating the scanning directions between consecutive layers, or rotating the direction within one layer by using island scanning.
• Using other rotations than 90°, e.g. 60° or 45°, is expected to be able to lower the texture even further and make the SLM product more isotropic.
• Unfortunately, this could not be checked since the Concept Laser M1 does not allow other rotations than 90°.

5. Conclusion

• This study shows that AlSi10Mg parts with an extremely fine microstructure and hence a high hardness can be made by SLM.
• Furthermore, by also considering the macrostructure, it is shown that a morphological and crystallographic texture is present and that a more anisotropic or isotropic part can be obtained by choosing the applied scanning strategy.

Figures

Figure 3: Defects in AlSi10Mg SLM parts. Keyhole pores at the end of the scan track and oxide material in a top view micrograph of sample D etched by Kellers are indicated by respectively white and grey arrows.

Figure 1: Overview of the scanning strategy used for the different samples. The building (BD), scanning (SD) and transverse direction (TD) are indicated. Sample A is scanned with long unidirectional vectors; sample B with long bidirectional vectors; sample C is first scanned with long bidirectional vectors in TD direction and secondly scanned with long bidirectional vectors in SD; sample D is scanned with island strategy with 90° rotation but without shift and sample E is scanned with island strategy with 90° rotation and 1 mm shift between the layers.

Figure 2: Macrostructure of AlSi10Mg by SLM. In this figure, the macrostructure in the top (a), side (b) and front (c) view cross sections of sample A is shown. Views are schematically presented in (d).

Figure 5: Microstructure of a AlSi10Mg SLM part in front and side view by SEM. Front (top) and side view (bottom) SEM micrographs of a sample produced with long bidirectional scanning tracks (sample B). The graphs on the right are a magnification of the areas indicated by the white box. The sample was etched with Keller’s. The white lines are indicating the different zones and the arrow is indicating the general dendrite orientation.

Figure 4: Microstructure of an AlSi10Mg SLM part in top view by SEM. Top view SEM micrographs of a sample produced with long bidirectional scanning

Figure 8: EBSD orientation maps in two perpendicular views for an AlSi10Mg SLM part. EBSD maps of the front view (left) and top view (middle) of sample B. The scanning and building direction is indicated by the black arrows on the pictures and some of the melt pool borders are pointed out by the dashed lines. The specimen coordination system and the crystal orientation - colour relation map are shown at the right.

Figure 7: Grain structure of AlSi10Mg SLM part revealed by polarized light. Polarized light optical micrographs of the (a) front view and (b) side view cross section of the part produced with unidirectional scan vectors (sample A); (c) micrograph of the view perpendicular to the firstly scanned vectors and parallel to the secondly scanned vectors (sample C); micrographs from parts produced with islands (d) without (sample D) and (e) with 1 mm shift of the island borders between the successive layers (sample E) are shown.

Table 1: Overview of the scanning strategy and scanning parameters of the produced samples and their resulting relative Archimedes density.

Frequently asked questions

Q1. What was used to calculate the pole figures?

The MTM-FHM texture processing software and MTEX Software Toolbox [13] were used to calculate the pole figures, texture indices and normalized texture differences. 

Q2. What is the reason for the morphological and crystallographic texture of the AlSi?

Due to the partial remelting of previously scanned neighbouring tracks, a morphological as well as crystallographic texture arise. 

Q3. What was the method used to correct the measured pole figures?

Measurements on virgin powder were used to correct the measured pole figures and the calculated pole figures took into account the orthorhombic symmetry of the process. 

Q4. How can the SLM process be carried out in an inert argon atmosphere?

Although measures are taken to carry out the SLM process in an inert argon atmosphere, the atmosphere can still contain 0.1-0.3% residual oxygen and oxides may still be formed. 

Q5. How many pixel counts are used to determine the melt pool height?

Based on pixel count, the melt pool height is determined to be 110 ± 10 µm (calculated based on microstructure pictures of all samples) and the width (based on the measured half-width) to be 170 ± 15 µm (for samples A and B) and 180 ± 30 µm (for samples D and E). 

Q6. Why is the texture reduced in SLM?

due to the partial remelting the morphological texture is increased since the majority of the elongated grains will now be oriented along the building direction. 

Q7. How is the hardness of the as-built AlSi10Mg parts?

Due to the fine dispersion of Si in the Al phase, the Vickers hardness of the as-built AlSi10Mg SLM parts is very high, namely 127 ± 3 Hv0.5. 

Q8. What is the effect of the direction of scanning on the texture?

When the scanning direction is rotated over 90°, the texture is significantly reduced and a weak cubic texture along the building direction arises. 

Q9. What are the only phases that could be detected in the XRD patterns?

Although this alloy can precipitate out Mg2Si, the only phases that could be detected in the XRD patterns are the face centred cubic aluminium and the diamond like cubic silicon phase. 

Q10. What is the main reason for the difficulty in producing dense aluminium parts with SLM?

The presence of Al oxides in the SLM products was also noted by E. Louvis et al. [6], who attributes the difficulty in producing dense aluminium parts with SLM mainly to the presence of these oxides. 

Q11. How much is the texture difference between samples D and C?

Because of the more equal volume percentage of grains resulting from scanning in each direction, a slightly more isotropic structure is obtained in sample D than in sample C. However, the normalized texture difference between samples D and C with reference to sample C is only 3.5%.