Euro PM2018: The processing and properties of additively manufactured aluminium alloys

Effects of Si content on densification and properties of Al-Si alloys processed by Selective Laser Melting
The first paper, from Takahiro Kimura and Takayuki Nakamoto (Osaka Research Institute of Industrial Science and Technology, Japan) and Kazuki Sugita, Masataka Mizuno and Hideki Araki (Osaka University, Japan), investigated the effects of silicon content on the densification and properties of Al-Si alloys processed by Selective Laser Melting (SLM), otherwise known as Laser Powder Bed Fusion (L-PBF) [1].

L-PBF with aluminium alloy powders is attracting attention as a manufacturing method for lightweight components and thermal control parts (heat exchangers, heat sinks) for aerospace, machinery and automotive applications, taking advantage of their low density and/or high thermal conductivity. However, the effects of alloy elements on the processability and properties of aluminium L-PBF materials have not been thoroughly understood. It is well known that Si, one of the main alloying elements in aluminium casting alloys, can strongly affect both the flowability of the melt and mechanical properties of cast materials. Based on this knowledge, the authors have proposed the hypothesis that the Si content in Al-Si L-PBF materials may have significant effects on densification behaviour during the L-PBF process and on the properties of the materials. However, systematic studies on the effects of Si content in Al-Si L-PBF materials have rarely been reported.

In the reported study, the effects of Si content on the relative density, microstructure and mechanical and thermal properties of Al-Si binary alloys processed by L-PBF were systematically investigated.

The aluminium powders used in the study were Al-xSi (x = 0, 1, 4, 7, 10, 12, and 20 mass%) alloy powders. As shown in Fig. 2, these powders were of solid solution (x = 0, 1), hypo-eutectic (x = 4-10), near-eutectic (x = 12) and hyper-eutectic (x = 20) compositions in the Al-Si binary alloy system. All the powders had almost the same particle diameter distribution and similar spherical morphologies. An L-PBF machine, the EOSINT M280, having a 400 W class ytterbium fibre laser with a beam diameter of approximately 0.1 mm, was used to fabricate the L-PBF specimens. The preheating temperature of the base plates was 35°C.
Optimum laser scan parameters for each Al-xSi powder were obtained by changing the values of the laser power, scan velocity and scan distance (interval between laser scan lines) at a fixed layer thickness of 0.03 mm. The L-PBF specimens were cylindrically shaped (diameter: 8 mm, length: 15 mm). The optimum laser scan parameters were defined as the conditions under which the highest density could be obtained. Table 1 shows the optimum laser scan parameters for each Al-xSi alloy.

From optical microscopy assessments, it was determined that the Al-0Si (industrial pure aluminium), Al-4Si, Al-7Si, Al-10Si, Al-12Si and Al-20Si L-PBF specimens contained no defects. However, micro-cracks were generated along the stacking direction in the Al-1Si L-PBF specimen. The relative density values of the Al-0, 4~20Si L-PBF specimens were almost 100%. On the other hand, the relative density of the Al-1Si L-PBF specimen was lower (at approximately 97%) due to the micro-cracks.

To investigate the mechanism behind the generation of the micro-cracks, verification experiments were conducted. A single bead, formed by a single-line laser scan of an Al-1Si powder layer with a thickness of 0.05 mm, was deposited. A few cracks were observed, regardless of laser scan parameters. This suggested that the cracks were generated in the single bead state, meaning extremely brittle characteristics of the Al-1Si melt during the laser melting process. Fig. 3 shows an IPF (inverse pole figure) map of a vertical plane of the Al-1Si L-PBF specimen, analysed by EBSD. It is important to note that the cracks occurred along the grain boundaries and the crystallographic directions on either side of the cracks were different. The fracture surface of a vertical tensile L-PBF specimen of the Al-1Si alloy showed an equiaxed solidified microstructure with grain sizes smaller than 200 μm. These results show that the cracks were generated before the solidification was complete (i.e., during crystal growth in the solid-liquid co-existing state).

From these verification results, the authors concluded that the micro-cracks found in the Al-1Si L-PBF specimen were solidification cracks, which were formed due to thermally induced tensile stresses caused by local heating (thermal gradient) from the laser irradiation, i.e., the solidification cracks in the Al-1Si L-PBF specimen were generated when the thermally induced tensile stresses and/or strain exceeded the tensile strength and/or elongation of the Al-1Si alloy melt, which had the brittle characteristics in the solid-liquid co-existing state. Additionally, due to the low flowability of the Al-1Si melt, there was no possibility of cracks healing by liquid infiltration.

Optical microscopy also showed that all the Al-xSi L-PBF specimens had a semi-circular macrostructure on the vertical planes. The size and shape suggested that the macrostructure was of laser traces surrounded by solidification boundaries. From SEM images, it was noted that the microstructure changed significantly with increasing Si content. The Al-0Si (industrial pure aluminium) L-PBF specimen had a finely dispersed granular microstructure (~0.3 μm), which was considered to be Al, Si and/or Fe oxides. The Al-1Si L-PBF specimen exhibited a granular microstructure of the oxides with partially-linked crystallised Si phase. The images of the Al-4~12Si L-PBF specimens showed fine elongated cellular dendrites (approximately 0.5 μm in cell size) parallel to the stacking direction (i.e., the thermal flow direction). These cells were primary crystallised α-Al phase and the cell boundaries were crystallised Si phase. The volume fraction of the crystallised Si phase increased with increasing Si content. This fine cellular microstructural morphology is a peculiarity of L-PBF materials, due to the ultra-rapid solidification caused by laser irradiation. The Al-20Si L-PBF specimen, on the other hand, showed a characteristic petaloid microstructure of primary crystallised β-Si phase within an Al/Si eutectic phase matrix.

Fig. 4 shows the ultimate tensile strength (TS), 0.2% proof stress (PS) and elongation (EL) of the Al-xSi L-PBF specimens. The Al-0Si L-PBF specimen had a TS of 110 MPa, which is higher than that of pure aluminium wrought material, due to the observed granular microstructure. Meanwhile, the Al-1Si L-PBF specimen had an extremely low TS of 25 MPa due to the micro-cracks generated along the stacking direction. As the Si content increased above 4 mass%, the TS increased monotonically and finally reached 580 MPa for the Al-20Si L-PBF specimen. The PS showed a similar trend to the TS. The PS increased steadily with increasing Si content from 125 MPa for the Al-4Si L-PBF specimen to 425 MPa for the Al-20Si L-PBF specimen. The EL of the Al-0Si L-PBF specimen was almost as high as that of pure aluminium wrought material (approximately 30%) and the EL then steadily decreased with increasing Si content above 4 mass%, in contrast to the TS and PS behaviour. On the other hand, the EL of the Al-1Si L-PBF specimen (approximately 1%) was significantly lower than the other specimens due to the micro-cracks.

In the Al-4~12Si L-PBF specimens, the volume fraction of the crystallised Si phase (cell boundaries) increased with increasing Si content. The increase in the TS and PS of the Al-xSi L-PBF specimens with increasing Si content was therefore attributed to a composite reinforcement effect due to the increased amount of secondary crystallised Si phase. Additionally, the increase of solid-solute Si in the aluminium matrix may also have strengthened the Al-xSi L-PBF specimens. On the other hand, the crystallised Si phase caused discontinuities in the aluminium matrix and this led to a decrease in the EL since such discontinuities can lead to fracture. In the Al-20Si L-PBF specimen (hyper-eutectic composition), the primary crystallised phase was β-Si instead of α-Al. Since the primary crystallised β-Si phase is generally hard, the TS and PS values became higher and the EL value lower.

Fig. 5 shows the thermal conductivity (TC) of the Al-xSi L-PBF specimens. The TC shows a similar trend to that of the EL in Fig. 4, i.e., the TC decreased from 200 W/m·K for the Al-0Si L-PBF specimen to 105 W/m·K for the Al-20Si L-PBF specimen. The decrease in the TC was mainly attributed to the increase in the solid-solute Si in the aluminium matrix. This is because the lattice strain, induced by the solid-solute Si, acted as scattering sites for conduction electrons, leading to an increase in the thermal resistivity.

How porosity is affected by different residual oxygen concentrations in the building chamber during Laser Powder Bed Fusion (L-PBF)

Next, Kai Dietrich and Gert Witt (University of Duisburg-Essen, Germany), Dominik Bauer and Pierre Foret (Linde AG, Germany) and Veronika Krumonova (Technical University Munich, Germany) reported on a study of how porosity is affected by different residual oxygen concentrations in the build chamber during L-PBF [2].

In L-PBF, the laser process takes place in an argon or nitrogen inert gas atmosphere depending on the material used. Therefore, the build chamber is purged with the appropriate gas until an oxygen concentration of around 1000 ppm is reached. This leaves an additional 3728 ppm of nitrogen and traces of hydrogen and humidity, from the powder or the filter, in the atmosphere if the chamber is purged with Argon 5.0. Aluminium-based alloys form oxides, leading to different properties than were expected. Although gases can have a great influence on the chemical and mechanical properties of metals, little research has been reported with regard to the L-PBF process. Due to the importance of the atmosphere during L-PBF, the reported study has therefore investigated the variation in hardness, porosity and microstructure obtained with different oxygen concentrations (100 ppm and 1000 ppm) during the laser process.

The experiments were carried out with Electrode Induction Melting Gas Atomised (EIGA) AlSi10Mg powder. SEM assessment of this powder revealed many satellites attached to the surface, possibly influencing the particle flowability. Table 2 shows the particle size distribution (PSD) as well as the chemical composition. The D10 of 11 μm points towards a high number of small particles in the powder.

Using a Trumpf TruPrint 1000 machine, a parameter study was performed with two different oxygen values in the build chamber (100 ppm and 1000 ppm), while flushing with Argon 5.0. 300 Cubes (8 x 8 x 10 mm³) were produced with different parameters (Table 3), while keeping the build layer thickness stable at 30 μm.

Fig. 6 shows an overview of the investigated densities in relation to the laser speed. For most of the curves, porosity of the 100 ppm oxygen samples is lower overall than for 1000 ppm. Fig. 7 reveals that the density at lower laser speed (around 900 mm/s) is lower at 100 ppm oxygen than with the higher oxygen concentration. At around 1100 mm/s, there seems to be a cross-over point and the density of the 100 ppm samples becomes higher than those with 1000 ppm. At 100 ppm oxygen, the porosity at lower speed mainly consists of spherical pores, which are most likely gas pores. Increasing speed seems to reduce gas porosity at first, resulting in a lack of fusion pores at higher speed. The highest density was reached at around 1300 mm/s. Enlarging hatch distance does not appear to influence gas porosity significantly. For 1000 ppm oxygen samples, the gas porosity is similar for all laser speeds.

The observed results can be explained by means of the Marangoni flow. Surface tension gradients lead to a driving force for a liquid to flow from a region of low surface tension to one of higher surface tension, as there is a larger force pulling it in this direction. In general, the surface tension of liquids decreases with temperature due to increased thermal vibrations, leading to a lower level of cohesion. Within L-PBF melt pools, the highest temperature will be at the centre of the pool where the laser is incident on the metal powder and will decrease outward, leading to a flow from the melt pool to the outside. This can lead to widening and flattening of the liquid zone. At the edge of the melt pool, a flow along the Solidus-Liquidus line is caused by shearing forces. These flows clash at the middle of the melt pool and rise to the surface.

Oxygen is known to influence the surface tension of the melt pool. The more oxygen in the atmosphere, the lower the surface tension will be. This will inhibit the Marangoni flow, because of more similar surface tensions of the melt pool and the surrounding region. Forming an oxide layer on the melt pool will also lead to a reduced flow effect. At lower oxygen concentrations such as 100 ppm, the surface tension, and therefore the Marangoni effect around the melt pool, will be higher. This leads to increased gas pore formation as reported in the literature. On the other hand, the melt pool will become deeper. At lower laser speed, the laser spot will remain on one point for longer. The Marangoni flow might therefore exert an influence for a longer time, which can lead to more gas pores. By increasing the laser speed, the effect diminishes, therefore increasing part density. By obtaining the same or better density while using around a 50% higher scanning speed, it is possible to decrease build time.

Table 4 shows the average hardness of the 100 ppm and 1000 ppm samples at hatch distances between 0.08 mm and 0.09 mm. This range was chosen because the densities of these samples are similar and the hardness is not greatly affected by the porosity.

The hardnesses of the 100 ppm and 1000 ppm oxygen samples are similar. Differences in oxide concentrations in the samples seem to be too low to greatly affect hardness.

The authors drew the following conclusions from the reported results:

• Reducing the oxygen concentration in the build chamber will result in an overall higher density
• Gas porosity at low oxygen concentrations seems to depend on the laser speed; the higher the speed, the lower the gas porosity
• At lower oxygen concentrations, the laser speed can be increased while maintaining or even increasing density
• Oxygen might have a small influence on the hardness

In terms of future work, simulation might be carried out to confirm the influence of oxygen on surface tension during the L-PBF process.

Mechanical properties of aluminium alloys produced by metal AM

Laura Cordova and Tiedo Tinga (University of Twente, Netherlands) and Eric Macia and Monica Campos (University Charles III of Madrid, Spain) turned their attention to the mechanical properties of Al alloys processed by metal Additive Manufacturing [3].

L-PBF is a promising alternative for the fabrication of aerospace components where weight reduction, high accuracy and complexity are required to achieve optimal performance. Since the development of L-PBF technology, aluminium alloys have proved to be challenging materials to process due to their high conductivity and high reflectivity. Additionally, when aluminium powder is processed by L-PBF, density is usually compromised by a high level of porosity, products can suffer deformation and hot cracking as a consequence of stress concentration, and high surface roughness is obtained due to poor flowability.
The reported work offered an innovative method for evaluating the effect of porosity and inclusions on fracture behaviour. In-situ testing was presented as a valuable approach to relate local plastic deformation and fracture mechanisms to various porosity distributions.

The two alloys included in the reported study were AlSi10Mg and Scalmalloy^®, an aluminium-scandium alloy with enhanced mechanical performance developed specifically for the L-PBF process and aerospace applications.

Al-Si alloys are characterised by excellent weldability and high corrosion resistance. Their attractive mechanical properties and low weight make them suitable for a large number of applications, especially in the aerospace industry. The addition of magnesium to the Al-Si alloy strengthens the matrix by Mg2Si precipitation. The Al-Sc solid solution in Scalmalloy forms Al₃Sc precipitates acting as an age hardener, leading to high strength. Sc is reported to increase the weldability of Al alloys, mainly for alloys susceptible to hot cracking and extended heat-affected zone formation.

AlSi10Mg and Scalmalloy tensile test specimens were manufactured by L-PBF in the XY and Z directions using an SLM Solutions SLM 280 AM machine. Prior to the build, an overnight (12 h) vacuum drying process was carried out to reduce the moisture level in the metal powders. The printing parameters used to build the specimens were optimised to obtain maximum density. The AlSi10Mg alloy had a stress relief at 300°C for 120 min and Scalmalloy a stress relief for 90 min at 180°C and a precipitation hardening treatment of 240 min at 325°C, since the strength of this material is derived mainly from precipitation hardening.

In order to compare the performance of AlSi10Mg and Scalmalloy in two build directions (XY and Z), the mechanical properties were evaluated. Micro tensile and Vickers hardness tests were conducted. Fig. 8 (left) shows the geometry of the micro-specimens tested with the micro machine. The thickness of these specimens was 1 mm. Although this geometry is not standard, due to the nature of the L-PBF process these specimens are more cost-effective to build than the larger dog-bone shaped specimens which are traditional. Note that, although micro tensile tests are useful in determining the stress-strain curve, maximum strain and ultimate tensile strength (UTS), the results should not be compared with results from other methods.

Sample preparation was a key step before beginning tensile testing, due to the high roughness of L-PBF specimens. Silicon carbide (SiC) paper, diamond solution and OPS colloidal silica were used to grind and polish the flat surfaces (Fig. 8, left). The cross-sections were shaped and polished with a three-square file. The rougher part of the printed specimens corresponded to the area where the supports were placed. For the XY specimens, the supports were located on the 10 mm side and, for the Z specimens, they were on the 4 mm side (Fig. 8, left). The supports help to attach the specimens to the build plate while scanning and melting the metal powder.

Fig. 8 (right) shows the micro machine used to perform the tensile tests at IMDEA Materials, Spain. This equipment has a maximum load of 10 kN. Since aluminium alloys have a relatively low tensile strength, a load of 1 kN was sufficient for the experiments. The clamping system speed during the test was set to a strain rate of 2 μm^s-1.

Ex-situ and in-situ tests were conducted with the micro machine. UTS and maximum strain were obtained from the ex-situ tests using four specimens for each material and building direction. The in-situ tests, one for each condition, were conducted inside a Scanning Electron Microscope (SEM) to capture images of the specimen at different points on the stress-strain curve.

An automatic micro indentation testing system was used to compare the hardnesses of AlSi10Mg and Scalmalloy.. Since aluminium alloys are relatively ductile, the indenter was loaded to 1N (HV0.1). A patented visual method automatically traced the sample edge of a live image, enabling the positioning of indents and patterns directly onto the overview image of a part.

The tensile test and Vickers hardness measurements provided quantitative results to compare the studied materials built in the XY and Z directions. From the tensile tests, σ – ε (stress-strain) curves were obtained (see Fig. 9). On average, Scalmalloy showed tensile strength values almost double those obtained for AlSi10Mg. Due to the high porosity and anisotropic properties, the materials showed some scatter in the results. The building direction appeared to have little influence on the mechanical properties for AlSi10Mg and Scalmalloy, although the maximum strain (ε_break) was generally higher for the XY direction.

Fig. 10 shows the in-situ test of an AlSi10Mg specimen, manufactured in the XY direction, deforming in tension. In Fig. 10(III), neck formation can be noted, with the material elongating until final fracture (IV). The images taken with the SEM also showed the superficial defects present in this specimen. Although there were large-sized inclusions from the manufacturing process, the fracture occurred in the middle of the neck. The tensile forces formed a neck close to the top side instead of crossing areas of high porosity.

Fig. 11 shows the average microhardness (HV0.1) results for Scalmalloy and AlSi10Mg in the XY and Z build directions. Five indentations for each condition were conducted in the clamping area, over flat surfaces without porosity. Similarly to the tensile strength and strain, there were not large differences between the XY and Z build directions for the hardness values. Nevertheless, Scalmalloy exhibited higher hardness (HV) values than those obtained for AlSi10Mg; almost double, as for the tensile strength values. The error, obtained for Scalmalloy Z build direction, was relatively high due to the porosity and anisotropy arising from the L-PBF process.

Table 5 shows the mechanical properties of conventionally manufactured (cast) AlSi10Mg and the state-of-the-art AM properties for AlSi10Mg and Scalmalloy. Conventionally manufactured AlSi10Mg shows similar properties to those reported in the literature for AlSi10Mg AM; however, the AlSi10Mg tested in this study showed considerably higher strain values and lower UTS values.

The values stated in Table 5 for AlSi10Mg and Scalmalloy refer to the minimum properties, reached using Additive Layer Manufacturing in the weakest direction of the material. The values of the mechanical properties were generated with tests on specimens that had been heat treated and machined.

The differences between the values from the literature and the results from this study were mainly due to differences in heat treatment/time and specimen geometry. The results showed slightly lower UTS values but higher strain than the values from the literature. In the case of the XY direction, this value was twice that from the literature for Scalmalloy and five times that for AlSi10Mg. Also, the hardness values found in the literature were higher than those obtained from the tests.

The fracture surfaces of the AlSi10Mg and Scalmalloy specimens showed a ductile fracture micro-mechanism. The fracture surfaces of both materials were covered by dimples. However, in the case of AlSi10Mg, there were two different sizes of dimples and, due to the high porosity of this material, the crack could propagate faster through the pores. On the other hand, Scalmalloy showed a less homogeneous surface, with the presence of limited cleavage areas. In both alloys, the shape and size of dimples in the XY direction differed from that in the Z direction.

Both alloys showed constriction before fracture. In general, AlSi10Mg contained more porosity and larger voids. Porosity on the surface helps to initiate and propagate the cracks. In L-PBF, the specimens are built layer-by-layer, but each layer is also processed line-by-line. So, if either a low laser power or a high scan speed is used, or the spacing between the lines is large, the melting of the additively manufactured layer can be insufficient, resulting in the formation of pores. Small spherical pores can also be attributed to gas entrapped during melting. This gas may originate from the powder bed or from powder reactions. In the case of Al10SiMg, Mg can be evaporated in the L-PBF process. Density optimisation is therefore necessary to avoid a lack of fusion, resulting in keyhole pores, or gas entrapment, resulting in metallurgical pores.

A digital microscope was used for identification of the porosity at 100 x magnification. Generally, the porosity measured on the surface near the crack exhibited higher values than in the cross-section. This can be explained by the fact that the surface belongs to the printing surface and therefore contains more imperfections. In the cross-section for direction XY, the porosity was due to gas entrapped within the melting pool, whereas, for direction Z, this could happen because of scanning defects.

Heat treatment of additively manufactured aluminium alloys

Finally, Jukka Simola, Maija Nystrom, Eero Virtanen and Juha Kotila (EOS Finland Oy, Finland) reported on a study of the heat treatment response of additively manufactured aluminium alloys [4].
Direct Metal Laser Sintering (DMLS), otherwise known as Laser Powder Bed Fusion (L-PBF), of aluminium is being adopted extensively by the machine building, automotive and aerospace industries. The unique characteristics of the layer-by-layer AM microstructure provide equal or even enhanced properties, when compared to conventional manufacturing methods, such as casting. In the reported research, EOS AlSi10Mg and EOS F357 aluminium-silicon (Al-Si) alloys, manufactured by L-PBF, were studied. Their AM microstructures can be characterised by finely-dispersed alternating phases of aluminium cell-like structures, surrounded by a silicon-rich eutectic network. Melt pools, i.e., laser scan tracks, make slight variations across the microstructure. The heat affected zones along melt pool lines show slight coarsening of the otherwise homogeneous microstructure. Build orientation can be observed from the tendency of the alloy to form more elongated cells along the direction of the heat flow (from top to bottom in Fig. 12 for the EOS F357 alloy).

Heat treatment of aluminium alloys has traditionally been used for strengthening as well as for improving the thermal stability of aluminium alloys. The response to heat treatment is dependent on the microstructure. Shorter T6-type heat treatments have been reported in the literature and some of these concepts were applied in this research.
The traditional T6 heat treatment for aluminium consists of solution annealing and ageing steps. A supersaturated solid solution of solute atoms in the aluminium matrix is achieved by fast quenching. Some Al alloys may harden naturally, but, in order to ensure stable properties at room temperature, an artificial ageing step is typically employed at elevated temperatures. Strength and thermal stability of the alloy can be optimised by varying ageing times and temperatures. Peak ageing (T6) will usually produce the highest yield strength, thanks to the formation of coherent fine-scale precipitates.
In many cases, the solution annealing step is not necessary. For example, many Al alloys can be quenched directly from hot working operations and the alloy is then already in a near-supersaturated state. For AM materials, the extremely high solidification rates can also increase solute supersaturation in the as-built condition. Peak-aged or over-aged tempers, such as T5 or T7, could also be produced without a need for the solution annealing step in order to provide higher strength or better thermal stability for the AM parts.

In AM processes, the current trend is towards higher productivity and high build rates, sometimes at the expense of increasing the overall porosity and defect rates. The amount of porosity in AM parts is, in most cases, higher than for wrought grades but lower than for castings. Microstructural changes during conventional heat treatments can also have negative effects on the extraordinary properties of AM parts, and new approaches to heat treatment are needed. A shortened solution annealing step was therefore studied in the reported research, in order to balance the mechanical properties achievable by heat treatment.

Within the study, EOS M290 and EOS M400 L-PBF machines were used to manufacture horizontal and vertical tensile bars. Nitrogen was employed as the shielding gas for the EOS F357 sintering process and argon was used as the shielding gas for the EOS AlSi10Mg process. The chemical compositions of the two alloys are given in Table 6.

A build job layout, consisting of 42 vertical and 42 horizontal samples, was prepared and manufactured using the EOS M400 L-PBF and EOS M290 machines. Samples were heat treated using a laboratory furnace. The target temperatures were of 540°C for F357 and 525°C for AlSi10Mg. The furnace was pre-loaded with a thermal ballast (28 mm thick 2.7 kg aluminium plate) to promote faster heating of the specimens loaded into the hot furnace. Solution annealing treatments of 6 h (conventional SA) and 30 min (short solution annealing, SSA) were used for the tensile samples. Microscopy samples were subjected to SA, holding times ranging from 15 to 360 min. After SA/SSA, all samples were quenched in 20°C water with a volume of around 30 l. No significant heating of the quenching medium was observed. Warping during quenching was not observed, thanks to the good racking procedures used. Artificial ageing was performed after an inoculation time of 20-24 h at room temperature, used to promote the formation of finer G.P. zones.

Ageing was carried out at 160–165°C for 6 h. Solution annealing and ageing heat treatment onset times were calculated from the point in time, when samples reached a temperature within 6°C of the target. Quench delay was kept to a minimum and samples were immersed in water within 10 sec after opening the furnace door. After ageing, samples were left to cool in open air.

Tensile test results for as-built and heat treated samples are presented in Table 7. The samples heat treated for 6 h, according to traditional convention, had good strength values, but elongation values, especially in the xy-direction, were seen as having room for improvement.

Metallographic examination showed that the onset of microstructural changes was almost immediate after reaching the target temperature in solution annealing. The break-up of the as-built lamellar cell structure led to the formation of dispersed silicon particles, located on grain boundaries. The build direction also became more apparent due to the elongated appearance of individual aluminium grains.
The break-up of the continuous silicon network was found to be linked with a ductility increase.

ractographs were prepared from vertical (Z-direction) tensile bars of AlSi10Mg in the as-built and SSA T6 conditions. The behaviour in the heat treated condition could be seen from SEM fractographs Larger areas of continuous Al-matrix could be seen to help in the formation of fracture dimples or voids. Some cracked silicon particles could also be observed in the fracture surfaces. The as-built fracture surface comprised faceted surfaces, aligned along the melt pool tracks. The fracture was therefore deemed to have propagated along the melt pools.

Solution annealing soak time tests were performed for the EOS F357 and EOS AlSi10Mg alloys. From micrographs of EOS AlSi10Mg, it could be noted how the particle sizes and the size of pores both grew with the coarsening microstructure. The development of spheroidised dark grey silicon particles from the Si network was also visible. Similar observations were made for the EOS F357 alloy. Formation of new pore sites was already observed after roughly 10–15 min of solution annealing. Longer soak times at solution annealing temperatures were found to increase the maximum pore size as well as the overall grain size. According to published data, the grain growth rate should reach a plateau between 3–4 h of solution annealing and similar observations were made in this study.

The formation of pores and their subsequent growth can be attributed to many possible sources and was not within the scope of this reported study. However, by reducing the overall porosity by 30% with a simple process improvement, the elongation was found to improve further and, at the same time, the scatter of the mechanical property results was reduced. In order to determine how the shorter solution annealing time affected the distribution and size of the strengthening Mg2Si phase, EDS maps were created for different heat treatment states of the EOS F357 samples: these states comprised as-built, SSA for 10 min and quenched, SSA for 30 min and quenched, solution annealed for 5 h and quenched and, finally, SSA for 30 min, quenched and aged for 6 h. In Fig. 13, the distribution of main alloying elements Si, Mg and Al in the as-built sample are shown. A rather uniform coverage of magnesium was found, with a slightly higher intensity of Mg observed in conjunction with the silicon network.

According to the literature, 15 min of solution annealing is sufficient to dissolve all available Mg in the Al matrix in AlSiMg alloys. However, a large portion of magnesium was seen to agglomerate with silicon in close relation to the newly formed, spheroidised Si particles. Some Mg₂Si particles apparently formed at this stage and their size would become coarser as the solution annealing was continued. At 10 min and 30 min, a significant amount of Mg was found in small clusters. The longer 5 h solution annealing time was found to help the remainder of Mg to dissolve and be more evenly distributed into the Al matrix. The coarsening of the microstructure was very apparent at this point, as evidenced in the distribution of Si phases and also in the size of the Si particles. The formation of larger Mg2Si particles during the solution annealing from the earlier Mg clusters could be observed.

The silicon needed to form Mg₂Si particles would be more sparsely distributed after longer solution annealing times, although some of it may have remained in solid solution. It is questionable whether the silicon from these particles in a coarser microstructure would be available to be taken back into solution in the matrix in order to be able to form fine Mg₂Si precipitates during artificial ageing. It has been reported in the literature that, by the rupture of the Si network, the spheroidised Si particles originate mainly from high angle grain boundaries. Si particles dissolved in the matrix would therefore have originated from low angle grain boundaries only. The resulting mechanical properties may not be optimal for alloys with higher Mg contents, such as F357, due to the lack of free silicon in the matrix.

Overall, the authors drew the conclusion that, for AM parts manufactured from aluminium-silicon casting alloys, a soak time of 20–40 min (depending on the thickness of the parts) at solution annealing temperature can be proposed. After a rapid quench by water immersion, the ageing treatment can be designed to provide the desired properties of under-aged, peak-aged or over-aged and stabilised conditions.

References

[1] Effects of Si content on densification and properties of Al-Si alloys processed by Selective Laser Melting, Takahiro Kimura et al. As presented at the Euro PM2018 Congress, Bilbao, Spain, October 14-18, 2018, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

[2] How porosity is affected by different residual oxygen concentrations in the building chamber during Laser Powder Bed Fusion (L-PBF), Kai Dietrich et al. As presented at the Euro PM2018 Congress, Bilbao, Spain, October 14-18, 2018, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

[3] Mechanical properties of aluminum alloys produced by metal Additive Manufacturing, Laura Cordova et al. As presented at the Euro PM2018 Congress, Bilbao, Spain, October 14-18, 2018, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

[4] Heat Treatment of additively manufactured aluminium alloys, Jukka Simola et al. As presented at the Euro PM2018 Congress, Bilbao, Spain, October 14-18, 2018, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

Author and contacts

Dr David Whittaker
Tel: +44 1902 338498
Email: [email protected]

[1] Takahiro Kimura, Osaka Research Institute of Industrial Science and Technology, Japan, [email protected]

[2] Kai Dietrich, University Duisburg-Essen/ Linde AG, Germany, [email protected]

[3] Laura Córdova, University of Twente, The Netherlands, [email protected]

[4] Jukka Simola, EOS Finland Oy, [email protected]

Euro PM2018 Proceedings

The full proceedings of the Euro PM2018 Congress is now available to purchase from the European Powder Metallurgy Association. Topics covered include:

• Additive Manufacturing
• PM Structural Parts
• Hard Materials & Diamond Tools
• Hot Isostatic Pressing
• New Materials & Applications
• Powder Injection Moulding

www.epma.com

Euro PM2019

The Euro PM2019 Congress and Exhibition will be held in Maastricht, the Netherlands, from October 13-16, 2019.

www.europm2019.com