Effect of Microstructure on Multiscale Mechanical Properties of Scalmalloy Produced by Powder Bed Fusion-Laser Beam

Abstract

For additive manufactured parts, it is important to measure homogeneity and demonstrate representative parts can be printed faster while maintaining key mechanical properties. In this work, a multiscale characterization of microstructural and mechanical properties was carried out to gain a thorough understanding of a range of powder bed fusion-laser beam (PBF-LB)-manufactured Scalmalloy for future optimization of the processing parameters. The relationship between microstructure, including porosity, grain structure, and precipitates, and mechanical properties, is investigated. The stress-relieved samples were characterized mainly using scanning electron microscopy (SEM) suite, uniaxial tensile tests and nanoindentation. The results show the multiple strengthening mechanisms in Scalmalloy, including solid solution strengthening, grain size, precipitates and dislocations strengthening, demonstrated through a combination of the nanoindentation measurements with microstructural analysis at the local scale. The current work suggests potential mechanisms for further improvement of the strength and ductility in PBF-LB-Scalmalloy.

1. Introduction

Scalmalloy is specifically designed for the additive manufacturing process of selective laser melting and is typically deployed in aerospace and automotive applications. For the structural efficiency of high-performance aerospace applications, Scalmalloy has shown remarkable properties such as excellent corrosion resistance, high strength-to-weight ratio, hardness, high plasticity, and fatigue resistance [1]. The mechanism to achieve these advantageous properties of Scalmalloy is due to the high cooling rates that can be achieved in the PBF-LB manufacturing process and the effect this has on the resultant microstructure. The fast cooling rates achieved in the manufacturing process enable a hypereutectic Al-Scandium solid solution to form [2,3,4,5], which increases the hardness by between 200 and 500% over that of a pure aluminium film [6]. This alloy offers high strength and ductility in its processed condition and can be further improved by thermal aging due to the precipitation of (additional) Al₃Sc particles from solid solution [7]. In contrast to other AM alloys, such as Nickel-based alloys, published results for Scalmalloy have consistently demonstrated that anisotropy plays a negligible role in the mechanical performance of these alloys from both micro-hardness and tensile test [1,4,8,9]. These superior mechanical properties and light weight give potential for PBF-LB-Scalmalloy as automotive and aerospace products.

PBF-LB-Scalmalloy shows two distinct grain structures, namely, fine grain (FG) regions along melt pool boundaries, coarse grains (CG) regions in melt pool areas. This is due to the localised heating and consolidation effect in an AM process, which also produces bimodal associated distribution of precipitates, which has been investigated thoroughly in more detail in previous studies [4,5,7,8,10,11,12,13,14,15]. These characteristic microstructures give Scalmalloy excellent elongation as well as increased tensile strength when compared with the other Al alloys. The outstanding mechanical properties are attributed to enhanced grain refinement, precipitation (including coherent precipitates) hardening properties, and combination of inhibition of recrystallisation and grain growth during repetitive aging [1,5,16]. However, only recently have sufficient investigations on the contributions of multiple strengthening mechanisms been carried out [14,15], which enables calculation of mechanical properties with microstructure information obtained from state-of-art electron microscopy analysis.

Due to the layer-by-layer building inherent in PBF-LB, the AM-manufactured materials are, in general, inhomogeneous and non-continuous at different length scales. This indicates the need to characterise the microstructure and properties at different length scales. Despite the bimodal microstructure of the PBF-LB-Scalmalloy, the investigation of mechanical properties and strengthening mechanism has mainly been through static uniaxial tensile testing [5,7,8,11,12,13,14,15,17,18]. These provide a reasonable correlation with the microstructure and strengthening mechanism of the Scalmalloy, but the high elongation properties have not been fully explored. It was found that the elongation changes (reduces) with aging [7,13,15], but there has not been a sufficient study on the effects of localised microstructural properties, which could play an important part role. There is a limited number of studies on the mechanical homogeneity at the local scale, for example, with nanoindentation [8,10,19], despite its ability to show a correlation between hardness and elastic modulus with microstructural features such as precipitates and grain size. In addition, the Portevin–Le Chatelier (PLC) effect, which is typical for Al-Mg alloy, was occasionally observed in PBF-LB-Scalmalloy [7,14,18], and the effect is related to plastic instability. However, the addition of the Sc, Zr, and their concentration, which has the opposite PLC effect, would reduce or eliminate the effects. Nanoindentation is particularly well suited for investigating these effects because of its high-resolution force–displacement capabilities, which makes it an effective tool for probing nanoscale perturbations such as plastic instability [20,21].

Traditionally, the tailoring of the density of material for weight-saving requirements, the fabrication of structures with a controllable number of sizes, and the distribution of pores have been difficult and expensive. In addition, the rapid development of complex parts in industry is gaining great interest. Advanced manufacturing enables less waste and provides energy efficiency. Combining the two could offer cost savings and accelerate part manufacture, enabling lower-cost components to be made.

In this work, the focus is to understand the relationship between microstructural features (relates to processing parameters) such as porosity, grain structure, and precipitates, as well as the mechanical properties measured with uniaxial tensile tests and nanoindentation. This promotes a thorough multiscale measurement from sub-micron to bulk mechanical response and provides useful load–displacement curves and quantified data in terms of hardness and Young’s modulus, and yield strength. The results are analysed to demonstrate the grain size, precipitates and dislocation strengthening mechanisms at small scales, and the contributions of these properties to the ductility of the alloy are discussed.

2. Experimental

2.1. Materials

Selective Laser Melting (PBF-LB) Scalmalloy

Scalmalloy industrial grade powder is produced and marketed exclusively by APWorks [22], and its typical chemical composition is (Mg, Sc, Mn, Zr of 4.5, 0.6, 0.5 and Zr 0.4 wt.%, respectively, with other element <0.2 wt.%, and Aluminium for balancing). In this study, a range of manufacturing parameters was used to vary the microstructure and properties of the alloy. These various selective laser melting conditions are as listed in

Table 1. Summary of PBF-LB printing parameters and density results (Archimedes method, ASTM B311 standard was used for <2% porosity) with constant laser power of 487.5 and laser spot size of 80 μm.

Sample ID	Velocity (mm/s)	Hatch Distance		Relative Print Speed **	Density (g/cm3)
		X (mm)	Y (mm)
S1	875	0.175	0.020	109%	2.6578
S2 *	1250	0.1125	0.009	100%	2.6473
S3	1250	0.175	0.014	156%	2.6037
S4	875	0.3	0.034	187%	2.6036
S5	1250	0.3	0.024	267%	2.5062
S6	2000	0.175	0.009	249%	2.4582
S7	2000	0.3	0.015	427%	2.3375

* Normal processing conditions, ** relative to normal processing conditions.

, with constant laser power and spot size.

The schematic of the printed tensile test specimen and dimensions is given in Figure 1, which illustrates the specimen orientation of parallel and perpendicular to build up directions.

2.2. Microstructure

To investigate the microstructural and local micromechanical properties with nanoindentation, samples with surfaces parallel and perpendicular to the build direction were prepared from off-cuts of the tensile specimen (end parts of the specimen as shown in Figure 1), with different surface finishes. First, samples were progressively mechanically polished with SiC paper, followed by diamond suspensions, with a final finish using silica polishing suspension (OPS) for 5 min. The same samples were further polished in a Hitachi IM4000 argon ion beam polishing system to try to remove any fine scratches due to the mechanical polishing. In general, significant material removal by ion beam polishing required an Ar + ion beam of 6 KV for a minimum of 30 min, followed by lower beam energies of 3 or 1.5 KV for similar times. The argon ion beam polishing reveals the microstructural features, such as grain boundaries, which are milled at different speeds compared with grains. Optical (Nikon Measuring Microscope MM-60, Shinagawa, Japan) and field emission scanning electron microscopy (SEM) (Zeiss Auriga, Carl Zeiss, Oberkochen, Germany) were used to examine the microstructures. Crystallographic texture analysis was conducted using electron backscattered diffraction (EBSD) (Symmetry2 detector, Oxford Instruments, Oxford, UK) mounted on the SEM at an acceleration voltage of 10 kV. The EBSD data obtained were analysed using Aztec analysis software (Oxford Instruments, UK) to obtain pole figures and equivalent diameter grain sizes. The chemical composition was investigated via energy-dispersive X-ray spectroscopy (EDS, X-Max80 detector with Aztec software, Oxford Instruments, Oxford, UK) mounted on the SEM at an accelerating voltage of 10 kV. The size and distribution of precipitates were calculated based on SEM secondary electron (SE) images using ImageJ 1.54.

2.3. Mechanical Tests

2.3.1. Tensile Testing

Tensile tests were conducted using an Instron 5969 universal mechanical test machine fitted with a 50 kN load cell. The test pieces were gripped using a pair of wedge action grips with serrated grip faces. The alignment of the test machine and grips was set prior to testing the samples by gripping a stiff steel bar and loading it to 1 kN before firmly locking the position of the grips. The test pieces were loaded for initial tests at a crosshead speed of 1 mm per minute until failure (with a strain rate of about 0.001/s).

Prior to testing, the test pieces were speckled using white and black spray paints. The strain was subsequently measured on opposing faces of the test piece using a non-contact video extensometer (Imetrum^®, Bristol, UK). For each building set, five repeat measurements were conducted, and the results were averaged.

2.3.2. Nanoindentation

Indents were performed with a diamond Berkovich indenter using a NanoTest Xtreme system (Micro Materials Ltd., Wrexham, UK) in a relative humidity and temperature-controlled room (ambient conditions). For nanoindentation mapping, a maximum load of 4 mN corresponded to a maximum indentation depth of about 300 nm. For fast mapping, a Nano position controller was used on a 32 × 32 matrix (1024 indents on each map) with 3 µm between adjacent indents. Pre- and post-calibrations of the indenter area function via indirect verification were performed by indenting a fused silica reference material sample at varying load/depth. Each nanoindentation cycle consisted of applying the load in 30 s, holding the maximum load for 30 s and unloading within 30 s. In addition to indentation mapping, the Portevin-Le Chatelier (PLC) effect was investigated at different strain rates (at about 0.01/s and 1/s) at a maximum indentation depth of about 1.5 μm at 100 mN and held at a maximum load for 60 s and standard unloading time of 30 s. For the indentation on the aluminium alloy, the final contact depth to maximum depth ratio is >0.7, indicating pile-ups at the outer perimeter of the indents, which is not corrected in this study. The load–displacement curves are analysed using the Oliver–Pharr method [23]. This leads to slightly higher hardness and Young’s modulus values than pile-up corrected values, and its use in this study does not affect subsequent analysis since the comparative difference in hardness variation is investigated.

3. Results

3.1. Tensile Test and Fracture Surface

3.1.1. Tensile Test

Figure 2 shows the Young’s modulus, 0.2% yield stress, and elongation at failure as a function of density. The Young’s modulus and yield stress increase almost linearly with the density in all samples (S1–S7). The variation in the density has less of an effect on the elongation in the samples which have relatively low densities, less than 2.6 g/mm³; above this value, the effect on elongation becomes more nonlinear, leading to large variations (S1–S4, >95% density). The 0.2% offset yield stress of the relatively dense samples (S1–S4) is similar to those reported in the literature for materials with enhanced mechanical properties after specific heat treatments. The specimens have both high strength and an elongation above 10% [7]. For lower-density specimens (S5–S7), the porosity dominates the mechanical properties, and their distribution and size affect the elongation. Therefore, it is of interest to understand the strengthening mechanism which gives the high strength even in relatively porous structures (S3 and S4) and the superior and consistent mechanical response of S1 compared with S2.

Unlike some typical Al-Mg alloys [21], including Scalmalloy [7], which exhibit inhomogeneous plastic deformation due to the interaction of diffusing solutes and dislocation motion (the PLC effect), the current strain-stress curves (Figure 2b) show little serrations despite of slow strain rate of about 0.001/s. The serrations, as shown in the enlarged insertion, could be a combination of the small sample size and the resolution/noise of both the strain and load measurements. The PLC effect appears within a certain region of applied strain and strain rates and is also affected by precipitates and the grain size in the material, so it’s important to investigate the effect in relation to microstructure at the local scale.

3.1.2. Fracture Surface

Figure 3 shows SEM images of the fracture surface from the tensile test pieces. All the samples exhibited a similar fracture surface with common features such as fracture of the melt pools and defects. In sample S1, the fracture surface presents features that show the failure of the sample occurred in a ductile manner, as seen in the lower right feature in Figure 3a, which shows the necking before fracture. The rest of the samples (Figure 3c,e,g) show irregular fracture surface without noticeable necking. In samples S3 and S4, several large pores were seen to be ‘broken’, which appear to have been produced during the fracture. In addition, the fractured surfaces of S1 and S2 exhibited uniformly distributed deeper dimples of different sizes (Figure 3b,d), indicating an increased amount of ductility in the deformation. Despite samples S3 and S4 showing some fine dimpled structures (Figure 3f,g), there are large areas of defects, such as open pores and unmelted particles, which dominated the fracture surface.

3.2. Microstructure

To optimise the processing parameters, it’s important to investigate the relationship between the microstructure and the mechanical properties. In the PBF-LB-processed Scalmalloy, once the powder is melted, it solidifies very quickly, with the resultant microstructure formed being due to the cooling rates and heat transfer direction, and also due to reactions (i.e., among particle precipitation, particle, and grain growth due to cyclic heat treatment from next laser scan on the next powder layer) taking place in the melt-pool before and after solidification, leading to a complex microstructure. The following section provides information on the porosity, grain structure and precipitate distribution in the PBF-LB Scalmalloy.

3.2.1. Porosity

Figure 4 shows the porosity (size and distribution) varies as a function of different scan speeds and hatch distances. The average pore area fraction in S1, S2, S3 and S4 is about 0.61, 0.54, 2.30, and 1.29%, respectively, with the size and shape varying from micropores in S1 and S2 to a few hundred micrometres of irregular-shaped pores in S3 and S4. In general, increasing printing speed (as shown in Table 1) increases the total area and size of porosity. The micropores in S1 have a slightly higher variation in size and shape compared with S2 on the surface perpendicular to the print direction (Figure 4a,b). Larger, close-to-spherical pores are observed across sample S3 surface, while pores in S4 are connected, forming larger elongated pores. The distribution of pores is related to the processing parameters, of which more details can be found in references [24,25].

3.2.2. Grain Structure and Distribution

Figure 5 shows the typical microstructure found in PBF-LB-processed Scalmalloy, where two dense samples exhibit a very similar appearance. In sections parallel to the build direction (Figure 5a,c), weld line cross-sections can be seen with the weld pool dimensions of width and depth ranging between 100–200 µm. The alternative coarse-grained (CG) and fine-grained (FG) structures can be clearly seen, with the darker areas being regions of very fine-grained material. The darker regions are boundaries and interfaces where partial fan-shaped structures, which are coarser grains, merge. This indicates significant turbulence occurred in the melt pool, and it is a result of the repetitive re-melting. As the laser power and powder layer thickness were kept the same, the main difference between the two samples (Figure 5a,c) is different hatch distances (Table 1).

In sections perpendicular to the build direction (Figure 5b,d), the band contrast imaging from the EBSD scan was used. These reveal better microstructural features compared with IPF images, where alternative and overlapping laser lines can be seen, with a width of about 150 µm corresponding to the size of melt pools. Similar to the observations made in the parallel direction, the boundary of the melt lines (Figure 5b,d) are darker regions, corresponding to finer grains. According to previous research [25], the patterns shown here indicate that the basic principles of the laser-substrate interaction are the transition mode, between heat conduction (HC) mode and keyhole/deep-penetration (KH) mode. In HC mode, heat conduction is the dominant heat transfer mechanism with negligible vaporisation of metal and reduced melt pool dynamics, resulting in a weld pool geometry—weld pool depth/weld pool width ratio ≤ 0.5. In KH mode, the formation of cavities is based on increased metal vaporisation and highly dynamic melt pool behaviour, where the weld pool depth/weld pool width ratio is >0.8. According to Figure 5, the ratio is 0.5–0.8, which is defined as the transition mode. The transition mode was found to have a remarkable fatigue-to-yield strength ratio compared to other alloys, and Scalmalloy printed with another mode [25], which indicates the microstructural homogeneity of the materials. The larger welding penetration depth in transition mode compared with HC mode increased the probability of re-melting certain material inhomogeneities. The main crack initiation sites are mostly defects resulting from a lack of fusion and/or defects.

Figure 6 gives more detailed information on the grain structures in samples S1 and S2 from EBSD scans of surfaces parallel (Figure 6a,c) and perpendicular (Figure 6b,d) to the printing direction. The microstructure of the processed Scalmalloy specimen exhibits the typical bimodal microstructure, as already reported in many references (i.e., Refs. [4,5]). In Figure 6a,c, the grains in the FG region are very small equiaxed grains, with grain diameter of approximately 200–1400 nm, whereas the CG regions show columnar grains of about 2 μm to 20 μm in length. In the FG region, grains have uniform crystallographic texture, while in the CG region, the columnar grains grow perpendicular to the melt pool boundary, generating a localised texture. The FG regions are also observed in the middle of the weld pools, which were observed in previous studies, indicating different solidification behaviour for the two distinct grain regions. The average circle equivalent grain size in CG and FG regions in sample S1 are about 1.3 and 4.5 μm, respectively. In Figure 6b,d, the FG and CG regions on surfaces perpendicular to the build direction are distinguishable, confirming the equiaxed geometry and size of the fine grains and elongated geometry of the coarse grains. Statistically, the microstructure in S2 shows larger variation in the grain size (i.e., larger elongated grains) compared with sample S1.

3.2.3. Precipitates

The secondary electron SEM images, Figure 7a, show a region containing both fine and coarse grains, with a sharp transition from fine to coarse grains. The concentration of bright spots in Figure 7b is much higher in FG than in the CG regions. Corresponding energy-dispersive X-ray spectroscopy (EDX) scans in this region show a higher concentration of Mg, Si, Sc, and Fe (Mn) elements compared with the larger grain regions, which form Al (Si, Mg, Fe, and Mn) and Al₃Sc precipitates. The variation of the distribution of Zr is not detected in this region. A further investigation of the FG region (Figure 7c) shows many of the precipitates are concentrated along grain boundaries and intragranular particles (Figure 7c). Previous research [4] provides a more detailed investigation of the microstructural properties arising from this region. Despite the different deposition processes, the microstructure of this current study exhibits a very similar appearance, which is a result of the repetitive re-melting. These are defined as primary precipitates, which cannot be dissolved after re-melting in the short term and therefore act as heterogeneous nucleation sites.

In Figure 7d, the fine bright particles, nanometre in scale, are precipitates dominating CG regions, both within the grains and along the grain boundaries, which are also shown in some of the FG regions. Previous research with transmission electron microscopy (TEM) was able to identify these as mainly Al₃(Sc_xZr_1−x) or other particles that dissolved during coarse grain growth [4,15]. Due to their size (<100 nm) and lack of TEM facility, current SEM-EDX is unable to identify the specific elements and their bonding status.

In summary, the microstructure formation mechanism in CG and FG regions is clearly different. In FG- regions, solidification takes place almost simultaneously, while in the CG region, the columnar grain growth proceeds to the top of the melt-pool. Different types of primary intermetallic phases were found in a high number of densities in the FG regions, and they act as seeds for competitive growth of FCC aluminium grains and pin and stabilise grain boundaries. The size of precipitates is larger in the FG region compared with the CG regions, while the distribution of larger precipitates along grain boundary in the FG region occurs with a greater spacing between the precipitates when compared to the CG regions.

3.3. Local Mechanical Properties

Nanoindentation is particularly suitable for investigating the mechanical properties of materials at a fine scale and has been used to investigate the variation of mechanical properties and plastic instabilities of alloys [10,20,21].

Hardness maps have been generated with an area of 96 μm × 96 μm. Figure 8 shows two mapped areas, with the distinguished features of FG and CG regions in Map 1 and a scratch across Map 2 in Figure 8a. In Figure 8b, the CG grain at the inner side of the melt pool shows a reduced hardness value when compared with the fine grains in the melt pool boundary. It shows higher hardness values in FG regions at the boundaries of melt pools, with an average hardness of about 2.05 GPa, higher than CG regions with average hardness values of about 1.93 GPa by about 6%. The lower hardness regions (i.e., blue dots) could be due to a combination of macro-scale porosity. The hardness results in Map 3 are dominated by the effect of the scratch, where the hardness values are higher in the scratch than the rest of the surface.

In the current study, the load–displacement curves of all the indents show pop-ins at different indentation depths, which are bursts in displacements (change of displacement varies from a few to tens of nanometres); these are characteristic of energy-absorbing or energy-releasing events occurring beneath the indenter tip. The size and magnitude of the current pop-ins indicate they are dominated by the long- and short-range dislocation interactions in the dislocation network in Scalmalloy (not dislocation initiation), which are driven by mechanical forces.

Our observation is different from previous studies, where no pop-in effects were observed under nanoindentation [10]. This difference could be due to a difference in strain and/or strain rate. Nanoindentation at a larger indentation depth (>one μm) at a lower strain rate (0.01/s) was carried out at different locations on samples S1 and S2 under load–controlled mode; no serrations in the load-displacement curves associated with the PLC effect were observed. Nanoindentation has previously been used to correlate with the PLC effect observed under tensile testing [20,21]; the displacement burst event was observed at indentation depths larger than 1 μm, where steps are very well-defined and repetitive. These events correspond to discontinuities resulting from the interaction of lattice dislocations with solute atoms. The PLC effect in aluminium alloys is sensitive to the solute (Mg) concentration, which leads to instantaneous negative strain rate sensitivity, while the solute of (Sc) has the opposite effect [26,27]. The combined effect of Al-Mg and Al-Sc indicates reduced plastic instability in Scalmalloy; this observation is consistent with our current tensile test results and load–displacement curves from nanoindentation measurements, which did not show a clear PLC effect.

The hardness maps for samples S1 and S2 at different locations (96 μm × 96 μm) to analyse the effect of microstructure on mechanical properties are given in Figure 9. Except for slightly higher hardness in the FG regions compared with the CG regions, there are locations where hardness in the CG region is also higher, reflected in the load–displacement curves in Figure 10 (i.e., indent 369 and 370). Similar observations were found in previous nanoindentation maps of Scalmalloy, which require further explanations for the mechanisms. In addition, the histogram shows a statistical distribution of hardness values in S1 and S2, with a wider distribution in S2. The lower average hardness value in sample S2, more than that of S1, is consistent with the tensile strength data.

4. Discussion

A thorough study of the microstructural and mechanical properties of Scalmalloy alloys with different densities was carried out. The tensile test results show a good correlation between strength and density, while the elongation at failure is a more complex issue. The microstructure in samples with >95% density shows porosity, characteristic alteration of fine and coarse grain structures (bimodal), and associated distribution of precipitates. The high strength and microstructural properties indicate that sufficiently high cooling rates occurred during the laser additive manufacturing and that a supersaturated solid solution was realised. [1,4,7,11,25,28]. Plastic deformation in Scalmalloys with a bimodal microstructure is highly localised, attributed to the shear localization, cavitation, and necking under the tensile tests. In previous studies, the mechanical tests on Scalmalloy were limited to tensile tests; thus, in this work, nanoindentation was also carried out to investigate local mechanical properties in relation to the bimodal microstructure. To investigate the relationship between microstructure and mechanical properties, the effect of porosity, strengthening mechanism and factors contributing to higher elongation are discussed.

4.1. Effect of Porosity

Due to the nature of selective laser melting (melt track-by-track and layer-by-layer), a variety of different material inhomogeneities can be defined, such as spherical porosity and lack of fusion porosity (irregular shape). This work shows the correlations between density and mechanical properties such as Young’s and yield strength; reduced density leads to lower mechanical properties and porosity. Figure 2 shows that for specimens with a lower density (<95%), the strength and Young’s modulus are clearly dominated by porosity. When density is high (>95%) in S1–S4, strength is less affected by the porosity, and the values are higher than traditional groups of Aluminium alloys despite the presence of the porosity. Even when the scanning speed is increased by more than 50%, in S3 and S4, the tensile strength of the samples is comparable with nearly fully dense samples S1 and S2. However, the effect of porosity on elongation at failure during the tensile test is complex. At low density, the elongation is insensitive to density. It shows a significant reduction in elongation in samples with larger pores in samples S3 and S4 compared with S1 and S2, as shown in Figure 2. Even for a nearly fully dense sample S2, the elongation varies significantly compared with S1.

In conclusion, large porosity (i.e., millimetres in S3 and S4, as shown in Figure 4) appears to have some correlation to the ultimate tensile strength, yield strength and modulus, leading to very low elongation. The existence of micro voids and process pores did not influence the mechanical properties that much throughout the samples but did significantly affect the elongation [8,11,13,22,28]. The elongation difference between S1, S2, S3 and S4 could be due to the presence and distribution of pores up to hundreds of micrometres in size. The influence of the material inhomogeneities on the quasi-static tensile properties in AM aluminium alloys was investigated in references [29,30,31], and the relationship between porosity and ductility of AM metal fits with casting theory [30]. The elongation difference between S1 and S2 could not be explained by porosity, which will be further discussed in Section 4.3.

4.2. Strengthening Mechanism

There are signs of multiple strengthening mechanisms in PBF-LB-Scalmalloy, with contributions from solid solution, grain boundary/refinement (Hall-Petch), precipitation and dislocations. The formation of the FG is related to the presence of intermetallic particles, and the particles and their distribution are affected by local temperature variation in the melt pool [4], and the solid solution obtained during PBF-LB solidification provides excellent conditions for the formation of precipitates during repeated heating [7]. In this study, the refined microstructure and precipitates from SEM images are observed in both FG and CG regions. The dislocation is common in additively manufactured materials, and the density can be estimated using EBSD scanning (GNDs). The Al₃Sc coherent structure within the Al matrix could not be resolved using SEM-EDX; however, the EDX spectra within FG and CG regions show a strong Sc peak. This indicates the solid solution status of the Al₃Sc, which was observed with TEM in Scalmalloy [4,14,15]. Therefore, the solid solution strengthening in the FG and CG regions is considered the same.

The nanoindentation mapping shows the difference in hardness in FG and CG regions; for instance, in sample S1 in Map 1, the average hardness in the FG regions is about 128 MPa (6.6%) higher than in CG regions. According to the Hall–Petch relationship, ${\sigma }_G=k{d}^{-{\displaystyle \frac{1}{2}}}$, the strength ${\sigma }_G$ increases when the grain size $d$ decreases, $k$ is the strengthening coefficient (0.17 MN/m^3/2). If the strengthening mechanism is purely due to the grain refinement, the contribution should be above 8.6% (assuming averaged circular equivalent fine and coarse grain size of about 1.3 and 4.5 μm, respectively). This indicates that other strengthening mechanisms are in place, such as dislocation density and precipitate size and distribution.

Considering dislocation strengthening, ${\sigma }_D=M\alpha Gb{\rho }^{{\displaystyle \frac{1}{2}}}$, $\alpha$ is a constant of 0.2, $M$ is Taylor factor of 3.06, $b$ is Burger’s vector (0.286 nm), $G$ shear modulus of Al matrix (26 GPa) and $\rho$ dislocation density. According to previous studies on the dislocation density in PBF-LB-Scalmalloy [12,18], higher dislocation density was found in CG regions, while only a few dislocations were observed in FG regions, and the dislocation density is in the order of 10¹⁴/m⁻². This indicates higher dislocation strengthening in coarse grain; the difference is up to 10 MPa.

According to the Orowan–Ashby equation, which was previously used to measure the effect of precipitate strengthening [15], the mechanical properties of CG regions with precipitates strengthening would lead to a strength increase, compared with the FG region of $∆{\sigma }_{Orowan}={\displaystyle \frac{0.13Gb}{\lambda }}ln{\displaystyle \frac{r}{b}}$, where $r$ is the particle radius, and $\lambda$ is the interparticle/precipitate spacing. The precipitate size and spacing were measured according to the SEM imaging, as shown in Figure 7, and the estimated precipitate strengthening in CG and FG regions is about 27 and 15 MPa, respectively.

Table 2. List of strengthening mechanisms contributing to the variation of yield stress in FG and CG (Calculation is based on data from S1 Map1).

Strengthening Mechanisms	Parameters	$\mathbf{Calculated} {\mathsf{∆}\mathit{\sigma }}_{\mathit{F}\mathit{D}-\mathit{C}\mathit{G}}$ (MPa)
Hall-Petch ${\sigma }_G=k{d}^{-{\displaystyle \frac{1}{2}}}$	k is 0.17 mN/m3/2 (FG grain size 1.3 and CG grain size 4.5 µm)	68
Dislocation strengthening ${\sigma }_D=M\alpha Gb{\rho }^{{\displaystyle \frac{1}{2}}}$	$\alpha$$\mathrm{is} \mathrm{a} \mathrm{constant} \mathrm{of} 0.2, M$$\mathrm{is} \mathrm{Taylor} \mathrm{factor} \mathrm{of} 3.06, b$$\mathrm{is} \mathrm{Burger}’\mathrm{s} \mathrm{vector} (0.286 \mathrm{nm}), G$$\mathrm{shear} \mathrm{modulus} \mathrm{of} \mathrm{Al} \mathrm{matrix} (26 \mathrm{GPa}) \mathrm{and} \rho$ dislocation density	−10
Precipitate strengthening $\mathsf{∆}{\sigma }_{Orowan}={\displaystyle \frac{0.13Gb}{\lambda }}ln{\displaystyle \frac{r}{b}}$	$r$$\mathrm{is} \mathrm{the} \mathrm{particle} \mathrm{radius}, \mathrm{and} \lambda$ is the interparticle/precipitate spacing	−12
Combined	${\mathsf{∆}\sigma }_D+\mathsf{∆}{\sigma }_{Orowan}+\mathsf{∆}{\sigma }_G$	46

is a summary of strengthening mechanisms in Scalmalloy that could contribute to the different yield stress in FG and CG, as reflected by the hardness variation. As the above strengthening mechanisms are affected independently, the yield strength is a simple summation of the individual contributions [32,33]. According to Tabor relation $H=3Y$, converting the hardness $H$ to yield stress $Y$, the combined strengthening in the FG region is about 43 MPa higher than the CG region in S1 (Map 1). This value is close to the calculated combined strengthening contribution of grain refinement, dislocation, and precipitation variation in FG and CG region is $\mathsf{∆}\sigma ={\mathsf{∆}\sigma }_D+\mathsf{∆}{\sigma }_{Orowan}+\mathsf{∆}{\sigma }_G$, of about 46 MPa. Given that the solid solution strengthening in the FG and CG region is the same, the calculated strengthening mechanism from microstructure properties matches well with the nanoindentation data. The results indicate the local variation of mechanical properties is dominated by grain refinement and compensated by the precipitation and related dislocation variation.

4.3. Effect of Microstructure on Elongation

In Al-Mg alloys with a bimodal microstructure (uniform distribution of coarse grain within fine grain structure), a combination of high strength and good ductility [34] is exhibited. The toughening mechanism is through the fine grains that provide the high strength, while the relatively large grains of the order of micrometres contribute to the ductility. Despite the bimodal structure, the above theory does not seem to work when comparing samples S1 and S2 in the current study and thermal-treated samples in previous research [7,13,15]. There is no evidence of pronounced grain growth in PBF-LB-processed Scalmalloy, which contrasts with other conventional aluminium alloys processed by PBF-LB [4,9,13,35]. This is because grain boundaries are pinned by small Al₃(Sc,Zr) precipitates. Such grain boundary pinning helps to stabilize grain growth during heating, which occurs when subsequent layers are deposed. The thermal treatment affects precipitate size and distribution and slightly increases the grain size, which could lead to reduced elongation. Nanoindentation mapping results show more homogeneous mechanical properties in S1 compared with S2; this indicates grain refinement strengthening and precipitate strengthening are balanced. Under tensile testing, local strain concentration shifts between CG and FG regions to accommodate stress concentration with different mechanisms. Figure 10 shows that, in S2, the average hardness is slightly lower than in S1 due to increased precipitate strengthening and reduced grain refinement strengthening; their combination leads to a larger variation of hardness distribution. The above could explain the different elongation between sample S1 and S2, as S2 has smaller hatch distances, as shown in Figure 5, which leads to a stronger ageing effect when compared with S1 and slightly larger grain sizes, as shown in Figure 6.

5. Conclusions

A methodology was developed to thoroughly extract the mechanical data from a representative AM material, Scalmalloy, and to study the effect of porosity, precipitates, and grain structure on mechanical properties. Large nanoindentation arrays have been performed to investigate the mechanical variation within the sample, together with statistical analysis. By combining with microstructure characterisation, our study allowed for the mechanical investigation and fundamental understanding of the strengthening mechanism. In addition to the superior strength and high ductility of the Scamalloy, factors that affect the elongation are illustrated and discussed, and these are further related to the PBF-LB scanning parameters. The main conclusions are:

(1)

Yield strength of 450 MPa at 0.2% is insensitive to porosity when density is above 95%, while the size of pores is above hundreds of micrometres, and the elongation is severely reduced from average of 11.5% to 4.5%.

(2)

Characteristic two distinct regions of fine and coarse grained (bimodal) microstructure and associated precipitates are observed with SEM and EBSD, which are affected by hatch distance.

(3)

The high strength is attributed to the synergy of multiple strengthening mechanisms composed of solid-solute, grain refinement, precipitates, and dislocations.

(4)

The local mechanical variation (i.e., average hardness of about 6%) is measured with nanoindentation mapping, which is explained by different strengthening mechanisms.

(5)

Through discussions and analysis, the current work concludes that the a balance between grain refinement and precipitates reflected from small variations in hardness distribution would increase elongation.