Introduction

Additive manufacturing (AM) is a technology that depends on manufacturing parts in the final shapes, from loose powders directly to the bulk form. This technology is regarded as the next predominant technology in the industry for the various advantages it is adding, including less design constrains compared to conventional manufacturing methods providing space for innovation and product customization, and material saving since AM allows reusing the non-consolidated raw material in the following building processes, which reduces the manufacturing cost, and more importantly the short lead-time by which AM processes can produce the final required product [1, 2]. AM includes different processes that are classified based on the heating source and the form of supplied material for fabrication [3]. One of which is selective laser melting (SLM) , which has a broad range of applications in different fields such as aerospace, military and implants sectors [1]. SLM depends on the interaction between the laser and material powder. Laser is used to melt the powder layer based on the CAD data provided. The input energy sufficiently heats the powder until it melts creating the melt pool. Moving through the powder layers, while overlapping the heat between the adjacent melt pools, creates the final structure. Layer by layer deposition is how a complete product of interest is build [4].

Various alloys are processable by SLM, such as Ti-based alloys, stainless steels, Cobalt-gold alloys and Aluminum alloys. Al-alloys are considered one of the most interesting engineering alloys, due to the ability of SLM structures to meet the required performance, with generally high mechanical properties [5]. However, Aluminum alloy powders have low absorptivity and high thermal conductivity, which requires supplying higher input energy levels to compensate the heat loss and reflection, in addition to the high oxygen affinity of Aluminum [6]. AlSi10 Mg alloy is a widely used alloy in casting, and has been one of the main candidates for the SLM. It has been extensively studied from mechanical and densification behavior point of view [7,8,9]. The presence of silicon increases the fluidity of the alloy in addition to decreasing the thermal expansion coefficient. However, it is hard to machine Al–Si alloy for the existence of hard silicon particles. This problem is mitigated by the presence of magnesium, which enhances the machinability of the alloy and improves the corrosion resistance [10]. Several factors affect the output from the SLM process and they are mainly the laser power, scan speed and hatch spacing. All the mentioned parameters are usually integrated in a single function (the input energy density function) as shown in Eq. (1)

$$ \text{E}_{\text{density}} = \frac{p}{{\text{h}\, * \,\upnu\, * \,\text{t}}} $$
(1)

where p represents the laser power, h is the hatch spacing between the adjacent laser tracks, \( \upnu \) is the scan speed and t the layer thickness [8].

Input energy has a great influence on the densification behavior of the AM materials as will be illustrated later. With the continuous development in the industry, alloys with their limited properties became insufficient for fulfilling the challenging material requirements. For that reason it has become mandatory to introduce new materials that can cope with the industry needs, and that was when metallic matrix composites were introduced to be fabricated using SLM. Aluminum matrix composites combine lightweight and high strength, which is needed in various fields due to the improvement in mechanical properties, which is induced by the ceramic particles. Studies focusing on the densification and mechanical behavior of different AMC are limited [5]. For example, a study was performed by Gu [6] investigating SLM of TiC/AlSi10 Mg nanocomposite using process parameters of 110 W, 50 µm layer thickness, and scan speed ranging from 100 to 350 mm/s. The study proved that the input energy directly affects the densification behavior, at low (volumetric) input energy of 126 J/mm3, the process produces bulk-form composite of density 91.7%, which was attributed to insufficient input energy. Near full dense samples were achieved upon increasing the input energy above 293 J/mm3, as the densification behavior is directly related to the amount of liquid phase formed. According to Dadbakhsh et al. [11], who performed a comparative study between different Al-alloys namely (Al, AlMg1SiCu and AlSi10 Mg), which were used as matrices for reinforcement using 15 wt% Fe2O3, AlSi10 Mg showed best results compared to other candidates. This was attributed to the high percentage of alloying elements, which induces changes in the thermal characteristics of the alloy, such as reducing the melting temperature which facilitates the melting and heat diffusion during SLM process, and accordingly improved density results were achieved. In a study for the densification behavior of AlSi10 Mg/SiC, Chang et al. [12] showed that the SiC particle size significantly affects the SLM-processed samples, whereby using SiC with a range of 5–50 µm, fine SiC particles (5 µm) created a high specific surface area which enhanced the powder absorptivity compared to coarse particles (50 µm) this leads to improving the wettability between the SiC particles and Al. Moreover, this resulted in increasing the in situ reaction between the matrix and reinforcement resulted in the formation of AlSiC4 phase that acts as an interface between SiC particles and the aluminum melt.

This paper focuses on studying the densification behavior of AlSi10 Mg/Al2O3 composite using SLM process through analyzing the porosity morphology of sectioned samples build within a range of energy density of 49–309 J/mm3, and investigating the influence of input energy on the balling phenomenon of the samples.

Processing and Experimental Procedure

Material

Aluminum powder was supplied by LPW Ltd, with an average particle size ranges between 12–56 µm (Fig. 1a). Chemical composition and further characterisation of the powder was reported elsewhere [8]. The powder showed a generally irregular morphology, as shown in Fig. 1a. This might result in some agglomeration of the powders during processing, as suggested by Manfredi et al. [13] who found that spherical powders enhance the follow ability of the powder and ensure homogenously distributed layers. Alumina particles ranges between 0.16 and 0.38 µm as shown in Fig. 1b. The alumina powders were supplied by Sigma Aldrich. Figure 1c shows SEM micrograph for the composite powders (2.5wt% Al2O3 mixed with AlSi10 Mg alloy powders) using a tubular mixer.

Fig. 1
figure 1

SEM image for powder morphology a AlSi10 Mg, b Aluminum oxide powder, c powder mixture (arrows point at Al2O3 particles)

Full size image

Processing

19 Cubic samples were fabricated by SLM with dimensions 10 mm × 10 mm × 10 mm using Concept Laser M2 CUSING®, equipped with a Yt-fiber laser beam, and a maximum laser power capability of 400 W, scan speed up to 4000 mm/s, and 150 µm is the laser track width. The process was performed in Argon atmosphere with oxygen content less than 0.1%. A simple scanning strategy was used in building the samples as shown in Fig. 2, with powder layer thickness of 20 µm. The main parameters of the building processes that were varied in the experiments are the laser power, scan speed and hatch spacing. Statistical experimental design was used to develop the building parameters matrix upon which building parameters were chosen. The used process parameters as shown in Table 1.

Fig. 2
figure 2

schematic presentation for SLM building strategy

Full size image
Table 1 Building parameters matrix
Full size table

Characterization

Samples were sectioned through their X-Z direction, mounted, ground, polished, and etched using Keller’s etchant for microstructural analysis. Polished samples were investigated using optical microscope (Leica DM IRM) for producing panoramic images to give a clear overview of the porosity morphology. Density measurement (porosity percent) was conducted using Image J software [14], through measuring the area fraction of the porosity. A field emission scanning electron microscope (FEG-SEM) Leo Supra 55—Zeiss Inc., was used for powder size investigation at accelerating voltage of 20 kV and for investigation of the SLM cubes surface morphology. Energy dispersive X-ray (EDX) detector attached to the SEM was employed for phase analysis of the sectioned samples. For hardness test, Vickers micro-hardness (Hv) testing machine (Mitutoyo HM-112) was used with a load of 0.5 kg for 30 s.

Results and Discussion

The densification behavior of the AlSi10 Mg/Al2O3 composite was extensively studied using electron microscopy and image analysis. Addition of ceramic particles (Al2O3) was expected to enhance the mechanical properties. Nonetheless, Alumina particles addition to the Al-alloy powders significantly influenced the consolidation behavior of the processed cubes producing high porosity. This was directly reflected upon the mechanical properties of the SLM cubes. The HV-values reported in the current study for the composite cubes was 109 Hv, which represent 14% reduction in the Hv-values measured for the plain Al-alloy, which was about 127 ± 3 Hv [15, 16]. The low Hv-values were attributed to the combined effect of the metallurgical defects induced by the SLM process along with the formed porosity associated with the Al2O3 addition. In the following sections, the metallurgical defects and the Al2O3 induced porosity are demonstrated.

Porosity Analysis

Figure 3 presents a bar chart for the variation of the processed samples relative densities percent as a function of input energy, according to the processed parameters shown in Table 1. In addition, demonstration of panoramic images for samples 5, 17, 16 and 9, representing the four consolidation mechanisms of the Al/Al2O3 obtained over different ranges of input energies (Fig. 3). The four main mechanism which were observed are classified as follows: (1) First mechanism, occurred over energy density range of 49–53 J/mm3 at which isolated porosity was formed; (2) Second mechanisms occurred over energy density range of 77–100 J/mm3, where interrupted porosity was formed; (3) Third mechanism, occurred over energy density range of 123–163 J/mm3, where longitudinal connected porosity (gaps) was formed along the building direction; (4) Fourth mechanism occurred over energy density range of 185–309 J/mm3 where disrupted non-uniform porosity was formed.

Fig. 3
figure 3

Density variation as a function of input energy ranges representing the 4-mechanisms a First, b Second, c Third and d Fourth

Full size image

Figure 3a shows the first mechanism in which the input energy was not insufficient to allow for consolidation between the adjacent melt pools. This was manifested in the form of isolated porosity across the whole section [8]. Second mechanism (Fig. 3b) showed improved consolidation due to the overlap between melt pools, which was observed in the increased relative density value. However, the input energy was not high enough to create fully dense bulk samples, which resulted in forming longitudinally aligned interrupted porosity along the building direction. Nevertheless, the second mechanism input energy resulted in the increase in density compared to the first mechanism. It is worth noting that increasing the input energy (100 J/mm3) within the second mechanism resulted in a gradual decrease in the relative density due to change in the porosity volume fraction. Increasing the input energy over the range of 123–163 J/mm3 (third mechanism, Fig. 3c) produced tracks almost of equal width separated by connected porosity along the building direction forming gaps between the adjacent tracks. At relatively higher energy input density, it seems that steep temperature gradients developed within the melt pool, which promoted surface tension gradient between their centers and edges. This initiates Marangoni flow, which is the flow of the fluid outwards from the region of high temperature (melt pool center) to that of low temperature (melt pool edges) [6, 17]. Olakanmi [18] suggested that increasing the oxygen content causes a change in the nature of the Marangoni flow form flowing outwards creating shallow and wide melt pools to inwards melt flow forming narrow and deep melt pools as shown (Fig. 4). The change in the melt pool dimensions has a direct impact on the scan track dimensions leading to reducing the scan track width [19]. The formation of Oxide films within the aligned gaps along the building direction were evident by EDX analysis (shown in section “Microstructure and EDX Analysis”). It is suggested that the formed oxide films between the scan tracks are responsible for hindering the consolidation creating the third mechanism [20].

Fig. 4
figure 4

Effect of marangoni flow on melt pool shape [18]

Full size image

Although, it is hard to visualize the Marangoni flow nature transition through the dimensional changes of the melt pool since all samples have the same Al2O3 content, a comparison was made for the evaluated melt pool shapes of representative samples for: (1) the first mechanism at which input energy was insufficient for initiating the Marangoni flow Fig. 5a and (3) the third mechanism Fig. 5b at which high energy density and high oxygen content the two main conditions for changing the Marangoni flow are present. As shown in Fig. 5, the difference in melt pool depth between both mechanisms is clear. Measuring the depth of the melt pools of the representative conditions, it was found that the first mechanism produced an average melt pool depth of 54 µm (Fig. 5a), while that for the third mechanism was 90 µm (Fig. 5b). These findings clearly manifested the influence of Marangoni flow nature on the melt pool size and accordingly on the scan track width.

Fig. 5
figure 5

OM images for samples representing a First mechanism (Sample 5), b third mechanism (sample 4)

Full size image

For the fourth mechanism based on the porosity morphology analysis, the disrupted non-uniform porosity as shown in Fig. 3d. Unlike the lower input energy mechanism, input energy range of 185–309 J/mm3 resulted in liquid phase with low viscosity and higher temperature, leading to the localized tearing of the oxide films [11], which allowed partial consolidation between the scan tracks. Accordingly, the porosity morphology changed from continuous connected longitudinal gaps between scan tracks to the distorted porosity.

Microstructure and EDX Analysis

The etched cross sections cut parallel to the building direction revealed the melt pool microstructure as shown in Fig. 6. It was observed that spherical metallurgical pores were formed within the melt pools, which size varied as a function of increasing the input energy. Shows the relation between energy density and the metallurgical pore size. It is well known that the spherical shaped pores are formed during the solidification process of the melt pool, which occur by forming gas bubble surrounded by liquid state [21]. Accordingly, metallurgical pores are formed when high input energy is employed for consolidation [22]. When comparing the metallurgical pores size generated in different energy ranges it was concluded that the pores size increased from 10–72 µm for the first (Low energy density) and fourth mechanisms (highest energy density), respectively. It is clear that for low input energy range (first mechanism) the metallurgical pored are too small to be observed at the employed magnification as shown in Fig. 6a. For higher input energies (second, third and fourth mechanisms), the metallurgical pores can be easily observed as pointed out by arrows in Fig. 6b–d, respectively.

Fig. 6
figure 6

OM images showing influence of input energy range on the metallurgical pores size for the a First, b Second, c Third, d Fourth mechanisms and e A bar chart for the variation of Pore size with Energy for the 4-mechanisms. Arrows point at Metallurgical pores

Full size image

SEM imaging and EDX analysis were employed on representative samples for the four-mechanism as a supportive evidence for the proposed scenario. Figure 7a is representing the first mechanism. Although, no oxide films were observed clearly in Fig. 7a oxides were evident in Point 2 at the boundaries of the pores in EDX analysis. It is expected that Alumina did not melt when low input energy is employed. Figure 7b represents the second mechanism, oxygen content is higher in the interrupted porosity this may be attributed to the high oxygen affinity of aluminum powders in addition to the presence of Al2O3 particles. The third mechanism image and EDX analysis (Fig. 7c) showed the formation of a continuous oxide layer on the peripheries of the longitudinal gaps along the building direction. This could be attributed to the melting of alumina particles, which was associated with the relatively high input energy. The direct mixing of both powders rather than using ball milling deprived the mixture from having the alumina particles impingement in the soft Al-powder. Instead the Al2O3 particles stuck on the surface of the Al-particles as shown in Fig. 1c, which facilitate segregation of the Alumina particles within the gaps creating oxide film along the formed gaps. Figure 7d represents the fourth mechanism at which distorted non-uniform porosity associated with high oxygen content can be observed within which overlapping between separated scan tracks was formed. Moreover, evidence for loose unconsolidated powders was observed within the distorted porosity.

Fig. 7
figure 7

SEM images for cross section of representative sample for each mechanism a First, b Second, c Third and d Fourth with their corresponding EDX analysis for the regions listed in tables below for each

Full size image

Balling Effect and Oxide Film Formation

The input energy showed significant influence on the surface morphology of the SLM cubes. The combined effect of Marangoni flow, which resulted in the melt pool instability and the low wettability induced by the addition of Alumina particles resulted in the liquid spheroidization, this phenomena is known as balling effect. The uneven surface created by the balling acts as an impediment to having homogenous layer of deposited powder [17, 23]. Two types of balling phenomena were formed under the selected SLM conditions in the current research, one of which is the formation of regular ball shapes of ~0.8–1 mm in average size as shown in Fig. 8a, this morphology appeared in the third consolidation mechanism range. And the second balling type appeared at the fourth consolidation mechanism range at which large irregular agglomerates with inter-connected porosity between them as in Fig. 8b. Deposition of successive powder layers the presence of balling phenomena, Al2O3 particles with their relatively small size compared to Al-particles. Al2O3 particles segregates and agglomerate inside the connected porosity creating the oxide film upon melting.

Fig. 8
figure 8

SEM for surface morphology showing types of balling phenomena a Third mechanism, b Fourth mechanism

Full size image

Conclusions

The input energy is a major factor influencing the densification behavior of AlSi10 Mg/Al2O3 SLM-processed composite. Within the energy densities range employed in the current work, 4-mechanisms were identified based on the porosity morphology/input energy. The porosity morphology changes depends on the combined effect of Al2O3 and the input energy. The size of metallurgical pores increases with increasing input energy density. The surface morphology reflects the change in energy through the changes in the type of balling phenomena. Ball milling of the reinforced particles is essential for eliminating the probability of having Al2O3 agglomerations rather than homogenous dispersion.