Improving density and strength-to-ductility ratio of a 3D-printed Al–Si alloy by high-pressure torsion

The studied material is a near eutectic Al-11Si alloy. Using the molten metal spray method, the metallic powder feedstock used in this study was produced at the Volgograd Aluminum Plant (UC RUSAL). The scanning electron microscopy image in Fig. 1a shows powders with spheric shape and an average particle size of 26.5 µm (D₁₀ = 17.3 µm, D₅₀ = 26.7 µm, and D₉₀ = 40.7 µm), as indicated in Fig. 1b. This particle size distribution, according to EPMA (European Powder Metallurgy Association (EPMA) and EuroAM (European Additive Manufacturing Group), is appropriate for good powder flowability during the L-PBF process (i.e., powders can spread evenly on the powders bed) [33]. The powder size distribution was measured using the ImageJ software thresholding black-and-white method [34]. 20 mm diameter and 100 mm length bars (building direction parallel to the bar length) were subsequently printed by L-PBF process using a selective laser melting SLM280HL printer from the SLM Solutions company in Lübeck-Germany. Table 1 indicates the alloy's chemical composition measured using a spectrometer. (The values indicated correspond to the average of five measurements.)

The material was built following the scanning strategy consisting of a 67° rotation after each layer in a controlled Argon atmosphere with a power (\(P\)) of 370 W, a scan speed (\(V\)) of 1650 mm/s, and hatch spacing (\(H\)) of 130 µm, equivalent to a laser energy density of 1.72 J/mm² according to Eq. (1) [35].

The microstructure was studied by optical microscopy (OM) using a Zeiss AX10 microscope. Keller's reagent (2.5% HNO₃, 1% HF, 1.5% HCl, and 95% distilled water) was applied to reveal the microstructure by immersing the samples for 5 s until the MPs were observed in detail. A SEM Jeol model JSM-7001F coupled with an electron backscattering diffraction (EBSD) detector was used to examine the microstructural characteristics. EBSD scans were performed at 15 kV and 15 mm working distance to quantify the material's microstructural characteristics, such as grain size, morphology, and crystallographic texture.

EBSD samples were prepared by mechanical grinding from 2500 SiC sandpaper, followed by 9 µm, 6 µm, 3 µm, and 1 µm diamond suspensions. To remove any scratch from the polishing process, one last polishing step using colloidal silica suspension (50% distilled water plus 50% colloidal silica in volume fraction) of 400 nm particle size in a vibratory polishing machine for three hours was carried out. 1 µm and 70 nm step sizes were used for the texture and microstructural characteristics analysis, respectively. The data were indexed using hkl Channel 5 and processed using OIM 7.3b and MTEX software [36]. Since the indexation rate in all the samples was not lower than 80%, the grain dilation method was used to correct the no-indexed points. Grains with less than 2 pixels in their interior were not considered for the post-processing analysis, and grain boundaries were defined as those with misorientations larger than 15°.

A JEOL 2100F transmission electron microscopy (TEM) operating at an acceleration voltage of 200 kV equipped with energy-dispersive X-ray spectroscopy (EDS) and the commercial precession system DigiSTAR tool and an external StingRay CCD by NanoMegas were set up for the microstructure analysis. TEM foils of 10 mm × 10 mm were subjected to polishing using sandpapers until a thickness of 50 µm was achieved. Subsequently, circular disks with a diameter of 3 mm were extracted from the foils. Then, the ion-beam process was used for thinning and perforation.

Porosity was accessed by X-ray microtomography using a microtomograph Skyscan 1272 by Bruker equipped with an X-ray generator of 10 W and a detector of 11 MP (4032 × 2688 pixels). For the microtomography analysis, samples with a volume of ~ 50.3 mm³ (8 mm diameter and 1 mm thickness) were scanned at a resolution of 2 µm and step size of 0.3°. 3D model reconstruction and rendering were carried out using Dragonfly™ 2022.2 software from Comet Group, Object Research Systems [37]. Porosity fraction was calculated using color thresholding between the pores and the solid material.

Phases and density of defects were calculated by X-ray diffraction using a Philips X pert Pro MPD diffractometer with Cu-Kα radiation at 40 kV and 15 mA by Panalytical Ltd, Malvern, UK. The intensities in the 2θ range between 38° and 118° were measured with a scan step size of 0.02° and analyzed with Maud software [38].

Differential scanning calorimetry (DSC) allows for evaluating the temperature effect on the precipitation kinetics of the as-built material. The Texas Instrument DSC 2920 Modulated DSC operating at a 20 °C/min heating rate was used. A thermal cycle between 20 °C and 550 °C in an Argon atmosphere and an AlSi11Cu sample mass between 10 mg and 80 mg were set up for the DSC heating ramp.

For the unconstrained HPT process operating at a rotation speed of 1 rpm, disks of 10 mm radius and 2 mm thickness were severely deformed at room temperature and 250 °C with an initial compression load of 2 GPa at three different degrees of rotation (\(\pi/2\), \(\pi\), and \(2\pi\), see Fig. 2a, b, which led to a final thickness of 1.7 mm. According to Eq. (2), these HPT turns are equivalent to shear strains (\(\upgamma\)) of 9.2, 18.5, and 36.9 [22]:where \(N\) is the number of revolutions, \(r\) is the disk radius, and \(h\) is the disk thickness after HPT processing.

In order to mitigate the formation of cracks and minimize the reduction in ductility, the HPT processing at a temperature of 23 °C was limited to a maximum of ¼ turn. Thus, HPT processing at a temperature of 250 °C allowed for processing the alloy at a number of turns above ¼. The temperature selection was based on a prior investigation using annealing heat treatments in the as-built condition, which did not cause coarsening of the Si-enriched particles while simultaneously enhancing ductility at the expense of reducing its strength [16].

Uniaxial tensile tests were performed on the as-built and HPT-processed materials to evaluate the alloy's mechanical performance. The tensile tests were carried out at room temperature at a constant strain rate of 1 × 10^–3 s⁻¹, using bone-shaped samples with 4 mm × 1.3 mm × 2 mm gauge dimensions, as indicated in Fig. 2c. HPT tensile samples were cut at a radius of 5.2 mm from the center of the disk. In order to establish a correlation between the mechanical properties of the HPT-processed samples and their respective microstructure, EBSD, SEM, and TEM samples were analyzed at the same position as the tensile samples (see Fig. 2c). The tensile strain was recorded using digital image correlation (DIC) and processed with the free, open-source software Ncorr™. In seek of statistics, image acquisition time was set up to 2 s employing a 12-megapixel digital camera.

Figure 3 shows the microstructure characteristics of the as-built sample. Figure 3a, b allows observing the characteristic overlapped melt pools (MPs) with a mean melt pool height of 150 µm separated by melting pool boundaries (MPBs) after the L-PBF manufacturing. Figure 3c reveals the formation of an interconnected network inside the MPs and MPBs that gives rise to a cellular structure coming from the eutectic reaction during the solidification process. Figure 3(b)–d corroborates the cellular structure coarsening (more like-columnar dendritic morphology with an average cellular size of 0.74 ± 0.21 µm) inside the MPBs due to the repetitive solidification and melting cycles between consecutive powder layers, while the heat-affected zone (HAZ) inside the MPB is more cellular (average cell size of 0.60 ± 0.30 µm).

By STEM-EDS observation, the EDS maps confirm that the cellular structure is made up of an Al cellular matrix, where the main alloying elements such as Si, Cu, Mn, and Fe are located in the eutectic phase, i.e., the cellular structure boundaries, as shown in Fig. 3e. The alloying elements distribution demonstrates that the interconnected eutectic network is mainly enriched with Si, although other elements like Fe, Mn, and Cu also appear in these zones.

The DSC curve in Fig. 3f for the as-built material indicates three exothermic peaks at temperatures of ~ 255 °C, ~ 337 °C, and ~ 527 °C due to the alloying elements. DSC reaction peak temperatures are highly influenced by the heating rate used. Therefore, careful interpretation is required. Hence, the mentioned peaks can be related to the alloy’s precipitation kinetics. In Al–Si alloys, the precipitation kinetic sequence during aging heat treatment is as follows [39]:where \(SSS\) represents the supersaturated solid solution, \(GP\) are the Guinier–Preston regions, \({\beta }^{{\prime}{\prime}}\) and \({\beta }{\prime}\) are metastable transient phases, and \(\beta\) is a stable phase.

In Al–Si alloys with low Cu content, the primary metastable precipitates are modifications of the Mg₂Si phase. However, adding Cu gives rise to additional precipitates of the metastable phases Al₂Cu (\({\theta }^{{\prime}{\prime}}\) and \({\theta }{\prime}\)), Al₂CuMg, and the stable phase \(\theta\) [40], which agrees with the alloy composition and the STEM-EDS maps.

EBSD analysis for the as-built material in Fig. 4 reveals several microstructural features inside the MPs. For instance, the as-built material microstructure has different grain sizes and morphologies depending on their crystallographic orientations inside the MPs, as shown in Fig. 4a, b for the columnar and equiaxed grains, respectively. According to Fig. 4a, most columnar grains are oriented with the \(\langle 001\rangle\) direction parallel to BD, while the equiaxed grains follow orientations between \(\langle 001\rangle\) and \(\langle 111\rangle\) directions. In this context, the grain size distributions in Fig. 4c clearly show a heterogeneous behavior, where columnar grains' average grain size is twice the average size of the equiaxed ones. This effect is also transferred into the texture evolution for both types of grains, where the \(\left\{111\right\}\) pole figures (PF) suggest that columnar and equiaxed grains present maximum texture intensities around the Goss and between Copper-S-Brass components, as shown in Fig. 4d.

These above observations match the corresponding hardening and softening phenomena associated with the different texture components in FCC materials. For example, Goss and Cube components have been linked to recrystallization and grain-grown phenomena after annealing treatments. In this case, they may appear due to the temperature gradients inside the MPs, so columnar grains appear and grow in the MPs' center where the temperature reaches the maximum values [41, 42]. On the other hand, the equiaxed grains with small grain sizes present texture components associated with deformation like Copper–S–Brass, which can be attributed to places with high residual stresses like the HAZ zones around the MPBs, as corroborated by the kernel average misorientation (KAM) analysis in Fig. 4e. The utilization of this map and the KAM profiles enables the confirmation of more significant misorientations along the MPBs compared to the interior regions of the MPs. Dhal et al. [43] used high-resolution nanoindentation mapping to demonstrate that MPB neighborhoods reached higher hardness values than grains inside the MP. They attributed this behavior to the heterogeneity in the grain size and precipitate distributions inside the MP. Thus, Fig. 4d indicates that the overall texture (\(\left\{111\right\}\) pole figure for all the grains) for the as-built material is controlled by Goss and Cube components, although some other components like Copper, S, and Brass with minor intensities than the later mentioned cannot be neglected. This results from the larger area occupied by the columnar grains, making them more representative when the overall texture is calculated.

Figure 5 shows SEM images at different zones along the radius of the HPT-processed disk. After ¼ HPT turn at 23 °C and 250 °C, Fig. 5a, b indicates that the cellular network remains interconnected in the central area (radius equal to zero), and it is disintegrated in the edge zones forming particles (maximum radius), respectively. At higher deformation magnitudes, the cellular structure breaks down in all zones, as shown in Fig. 5c, d. The main difference between ½ HPT and 1 HPT turns is that particles in the middle zone present an elongated morphology after ½ HPT turn, while after 1 HPT turn, particles are more like spheres in all the zones. This behavior is a consequence of the heterogeneous deformation state induced by the HPT process, as deformation is a function of the number of revolutions and the radial distance, as proved by Eq. (2). Therefore, at a low number of HPT turns is more likely to find a heterogeneous microstructure across the disk radius than at high plastic deformations, as several authors have indicated [31, 44, 45].

Through the cellular structure disintegration after HPT processing, the particle size refinement is observed on the insets of the figures. HPT gives rise to UFG particle sizes down to 0.4 µm till 0.29 µm after 1 HPT turn. According to Van Cauwenbergh et al. [46], these particles present a high amount of Si, which can increase their size with further temperature increments as solute migration can occur from the Al supersaturated solid solution. This confirms that the selected HPT processing temperature did not lead to Si-rich particles coarsening.

At the microscale, EBSD allows for deep examination of the microstructure evolution across the disk radius after different HPT turns. In the center zone, Fig. 6a, b shows no drastic grain refinement after ¼ HPT turn at room and warm temperatures, although some clockwise texture twist occurs in the \(\left\{111\right\}\) PFs. This observation confirms the heterogenous microstructure state at low HPT turns and the shear strain effect from the torsion load, which helps break the initial texture leading to fast grain refinement.

Through Fig. 6a1–f, the microstructure evolution between the central and edge zones at different HPT turns is observed. Figure 6c displays the grain size distributions for all the analyzed materials at the edge zones. This figure also shows the grain size reduction from the coarse-grained state in the as-built material to the UFG state after HPT processing. Comparing the EBSD microstructures in Fig. 6a1, b1 with Fig. 6d–f, the magnitude of the grain size reduction is evident together with the transformation from huge columnar to equiaxed UFG grains. Thus, HPT processing leads to grain size reductions of more than one order of magnitude as compared with the as-built material, passing from an average CG size of 15 µm to UFG averages values between 0.4 µm and 0.9 µm. Moreover, regarding texture evolution, the edge zones present more random orientations with intensities that decrease as the deformation degree increases. At the center zones, the texture presents the same components as that of the as-built material, with some small clockwise twists. However, at the edge zones, new texture components appear, which are dependent on the number of HPT turns. The samples with ¼ and ½ HPT turns at 250 °C show that components \(A/\overline{A }\) and \(B/\overline{B }\) are the most intense, as corroborated in Fig. 6d and f. These components have been associated with grain fragmentation and rotation, contributing to the texture strengthening in FCC materials processed by HPT [47]. After 1 HPT turn, the texture appears more symmetric, exhibiting a simple shear behavior, where components \(C\), \({A}_{1}^{*}\), and \({A}_{2}^{*}\) are the most representative, yet the overall texture experiences weakening. According to studies conducted by Orlov et al. [48], there was a texture evolution from \({A}_{1}^{*}\), \({A}_{2}^{*}\) to \(C\) components in pure aluminum processed by HPT up to a plastic strain of 4. These observations align well with the high density of subgrains observed until ½ HPT turn and the subsequent transformation into HAGB, leading to texture component rotations and texture weakening, respectively.

SPD processes like HPT produce a high density of defects and phase transformations that X-ray measurements can access. Figure 7a shows the X-ray diffractograms for the analyzed conditions and the main identified phases. The Maud fittings showed a good match with the experimental data, with Sigma and Rwp values less than 2 and 10%, respectively, as shown in Fig. 7b for the ¼ turn − 23 °C sample. At first glance, these diffractograms allow observing the presence of different intermetallic phases and the Al peak broadening after HPT processing, as indicated in the inset on the right-hand side. The main intermetallic components come from Si, Fe, and Cu, which agrees with the STEM-EDS maps shown in Fig. 3e. It is worth mentioning that the number of peaks related to intermetallic phases increases after the HPT processing at 250 °C (homologous temperature of \(0.38{T}_{M-Al}\)). This event can be related to the solute migration from the Al-matrix toward the particles due to the HPT processing temperature, as demonstrated by Muñoz et al. [16] for the same alloy after different annealing heat treatments, leading to the Al matrix purification, which has been found in Al alloys processed by HPT and after heat treatments [49, 50]. Conversely, the alloy processed by ¼ HPT turn at room temperature does not follow the same tendency. This diffractogram presents fewer peaks and a larger peak broadening than those processed at warm temperatures. According to Valiev et al. [51], SPD processes may hinder the formation of precipitates in Al alloys and instead promotes segregation and clustering of alloying elements, especially at the grain boundaries. Furthermore, the larger peak broadening suggests a higher density of defects due to a lower recovery rate than the warm HPT processing conditions.

Calculation of domain sizes and dislocation densities have deserved a great effort since the seminal works of Willianson and Hall (W–H) [52] and Warren (W–A) [53]. Further improvements on the last technique have been realized by what was called the modified W–H model and a few versions of the XRDCPLA largely developed by Ungar and Gubicza [54, 55]. Usually, the values calculated by these more elaborated approaches, besides their higher accuracy, are larger than the ones calculated by the simplest approach of using peak broadening calculated by Langford model and Williamson and Smallman (W–S) equations [56]. Due to its simplicity, they got a wide spread and are currently in use in the literature when the comparison between different alloys, materials, processes, etc., are not the main goals of the study, but rather to obtain relative values for similar processes in a single material.

In the current case, we determined dislocation densities for four different mechanical and thermal processes for a single material, taking advantage of the simplicity of the W–S equations to evaluate relative quantities. Despite the values obtained by this method could be even one order of magnitude less than better determined ones, they are a good guide to understand the whole set off simultaneous experiments.

Microstructural defects like dislocations, which are in part responsible for the peak broadening, can be calculated by Maud software using the Williamson–Smallman method, where dislocation density can be assessed by Eq. (3) as follows [56, 58]:where \(\varepsilon\) is the lattice microstrain, \(D\) is the size of the blocks building up the microstructure, and \(b\) is the Burgers vector of Al (2.86 × 10^–10 m).

Figure 7c plots the evolution of dislocation density and crystal size for all the analyzed conditions. The as-built material reaches a dislocation density of 1.2 × 10¹⁴ m⁻² and a crystal size of 124.1 nm. After ¼ HPT at room temperature, a considerable jump in the dislocation density and the crystal size occurs, changing to 4.0 × 10¹⁴ m⁻² and 61.2 nm, respectively. On the side of warm HPT processing conditions, the dislocation densities vary between 2.7 × 10¹⁴ m⁻², 3.0 × 10¹⁴ m⁻², and 2.7 × 10¹⁴ m⁻² after ¼, ½, and 1 HPT turns, respectively, and the corresponding crystal sizes values were 72. 7 nm, 75.4 nm, and 72. 3 nm. These values of dislocations and crystal sizes prove the SPD effect on the microstructural properties, refining the crystal size due to the multiplication of dislocations. The temperature effect can be observed between the room and warm processing conditions, so it is clear from Fig. 7c that processing the alloy at room temperature gives rise to higher dislocation density and smaller crystal size than at warm conditions. This behavior can be related to the dynamic recovery phenomenon that causes dislocation annihilation [59]. However, the dislocation annihilation and the crystal size differences concerning the room-temperature processed material are relatively small, suggesting a slow recovery rate for the alloy until this temperature.

As to the HPT nature, it is important to remember that this process involves high hydrostatic pressure, which highly reduces the dislocation annihilation rate, leading to more considerable grain size reductions than most SPD processes [60].

Materials density is one critical property for the successful structural application of metallic components obtained by AM. Figure 8a shows a panoramic image of one section of the as-built material, where red arrows identify several pores with different sizes. Porosity in AM materials can occur for different reasons, for example trapping of powder particles by oxide layers, gas trapping inside the powder particles, wrong manufacturing parameters, lack of fusion, and lack of penetration [61].

OM found that the as-built material has pore sizes ranging from 20 µm to 100 µm, as indicated in Fig. 8b. 3D-image reconstruction confirms a heterogeneous porosity size distribution (average porosity size = 0.042 ± 0.022 mm) that represents 0.13% of the material volume with an average sphericity of 0.9 ± 0.05, as illustrated in Fig. 8c, d. These observations prove the alloy's high initial density and the mainly spherical morphology of the pores after L-PBF manufacturing.

Figure 9 shows a similar analysis for the materials processed by HPT at room and warm temperatures. After ¼ HPT turn at room temperature, Fig. 9a indicates the panoramic optical micrography across the disk diameter, where several changes for the MPs are observed. MPs go through a high reduction and shear strain effect after HPT processing. As HPT deformation highly depends on the radial distance, zones close to the disk edges suffer a higher degree of deformation than the center zone, as indicated in Fig. 9b. MPs in the middle plane on the edge zones are highly deformed, while the zones far from the middle plane are less deformed due to the deformation in the HPT process also depends on the sample thickness, especially at low plastic deformations. On the other hand, MPs in the center zone are less shear strained and compressed. The HPT process involves two deformation stages, an initial compression load followed by the torsion stage. Therefore, many pores from the L-PBF process are closed or size-reduced after the first compression stage. Figure 9c displays the 3D-reconstructed volume section of the sample after 2 GPa load and before the torsion stage. This reveals a low fraction of porosity (0.01% of the analyzed volume) and the pores size reduction with an average value of 0.012 ± 0.0017 mm, as confirmed by the narrow peak on the pore size distribution. Moreover, Fig. 9d confirms a porosity morphology more sphere-like with an average sphericity value of 0.91 ± 0.048. In this context, the subsequent torsion stage and increasing HPT turns will lead to a neglectable porosity fraction.

Figure 9e –g confirms the more significant degree of deformation, higher densification, and more homogeneous strain distribution as the Al alloy is processed from ¼ turn to 1 turn at 250 °C. After ¼ HPT turn at 250 °C, Fig. 9e shows similar behavior to the one described in Fig. 9a. In contrast, at ½ HPT turn, it is evident that the size of the less deformed central zone is reduced (see Fig. 9f) until the point of 1 HPT turn, where the deformation seems to be more evenly distributed along all the disk directions and the MPs cannot be differentiated easily.

Figure 10a and Table 2 present the tensile properties after different number of HPT turns. HPT processing at room temperature increases the yield strength and elongation by 255 MPa and 1.5%, as compared with the as-built condition, respectively. On the other hand, HPT processing at 250 °C leads to yield strength and ductility improvements ranging between 91.4 and 205.7 MPa and 7—8% as a function of HPT turns, respectively. These values represent unusual properties for this type of alloy, as corroborated in Fig. 10b.

Al alloys with high Si contents are characterized by poor ductility and moderate tensile strength, as reported by Girelli et al. [62] for an as-cast AlSi10Mg alloy. With the advent of AM techniques, these alloys can reach better mechanical performance, reaching yield strengths between 200 and 350 MPa, as reported by several authors [63, 64]. However, the ductility remains the main issue for both manufacturing routes. Other authors have explored using heat treatments after AM to improve ductility at the expense of reducing strength. However, this has allowed to expand the properties window of AMed materials due to the high strength in the as-built condition [16, 65], but not to overcome the strength–ductility trade-off.

The exceptional strength–ductility combination exceeds the UFG low-carbon steels, which register similar tensile strength and lower ductility than the present alloy. Another proof of the standing properties obtained by the combination of AM and HPT over AM properties is the results shown in the review work of Mishra et al. [20], where yield strengths beyond 525 MPa were not reported for different families of Al alloys manufactured by L-PBF. It is noticed from Fig. 10b that the combination of AM and HPT produces better mechanical properties than the combination of AM and ECAP reported by different authors [26, 66]. This can be associated with the HPT capacity to introduce more significant plastic deformations and higher hydrostatic pressures than the ECAP process, leading to a higher density of defects. In this context, Zehetbauer et al. [60] observed that elevated hydrostatic pressures do not exert any influence on the initiation strains associated with various deformation stages. Still, it has been shown that the onset flow stresses exhibit a 10% increase relative to the values observed in low hydrostatic pressure processes.

Due to the size limitations of the HPT process, miniaturized tensile samples had to be used, leading to some ductility variations. According to Zhao et al. [75], when strain derives from the crosshead displacement, the uniform and the post-necking elongation increase with decreasing gauge length and increasing specimen thickness. However, in this study, DIC quantified strain, which leads to more reliable values. In addition, for samples with the size and geometry, the properties can be compared relatively.

The AlSi11Cu alloy can present different types of precipitates, as suggested by DSC and X-ray diffraction measurements in Figs. 3f and 7a, respectively. Figure 11a, b shows the image quality and phase maps obtained by setting the TEM in the scanning precession electron diffraction (SPED) mode, which confirms the formation/segregation of Al₂Cu around the Al grain boundaries. The high-resolution brief field TEM images in Fig. 11c–f also prove the presence of several particles and nanoparticles with different morphologies after ¼ HPT turn at room temperature. These particles are associated with the fragmented Si-enriched network and the precipitation effect during the L-PBF process.

Figure 11d indicates several regions with Si-enriched particles where twins appear inside, as highlighted by the yellow arrow in Fig. 11e. Also, the magenta and cyan color arrows highlight elongated and rounded nanoprecipitates in Fig. 11e, respectively, which may correspond with the Al_4.5FeSi and Al₂Cu indicated before with the X-ray diffraction patterns.

Precipitation hardening follows the Orowan mechanism as a consequence of interaction with dislocations [65], i.e., dislocation looping, and the following equation can evaluate it:where \(M\) is the Taylor factor (3.06), \(G\) is the Al shear modulus (27.9 GPa), \(b\) is the Burgers vector (0.286 nm), \(v\) is the Poisson’s ratio (0.33) [77]. \(\overline{r }\) is the mean radius of a precipitate in a random plane (\(\overline{r }= \sqrt{2/3}r\), where \(r\) is the mean radius of the precipitates), and \({\lambda }_{p}\) is the mean inter-precipitate spacing (\({\lambda }_{p}=2\overline{r }(\sqrt{\frac{\pi }{4{f}_{v}}}-1)\), with \({f}_{v}\) as the volume fraction of particles). Since most of the Al₂Cu phase was found at the grain boundaries (Fig. 11b), the precipitation calculation focused on the Si-enriched particles.

Here the mean precipitate sizes (radius) and the volumetric fraction of particles calculated from TEM for the as-built and ¼ HPT turn materials were 2.5 ± 1.04 nm, 0.5%, and 3 ± 2.1 nm, 1.1%, respectively. Thus, the calculated strength increments from the precipitates for both materials were 203 and 285 MPa.

The fast-cooling rates in L-PBF lead to a supersaturated Al-matrix that gives rise to a higher solid solution content of Si in the Al matrix than the theoretical value of 1.6 wt.%. Several authors have found a solid solution Si concentration of 3 wt.% using atom probe tomography in an AlSi10 alloy manufactured by L-PBF on a 35 °C building platform [78]. Therefore, the solid solution contribution can be evaluated through the following equation:where \({k}_{(Si)}\) is a constant for Si in solid solution (39.7 MPa/wt.%^2/3 [79]) and \({C}_{Si}\) denotes the wt.% concentration of Si in solid solution. In this investigation, solely the contribution of Si was considered due to the insufficient representativeness of the weight percentage of other elements, even under the assumption that all atoms of these elements were completely dissolved inside the matrix.

Assuming that all the Si content of the present alloy is dissolved in the Al matrix, Eq. (6) derives a value of 184.4 MPa, which can be regarded as an upper value for the solid solution contribution due to part of the Si atoms are not in solution but precipitated in the as-built and HPT processed conditions. This solid solution evolution has been reported by different authors in Al alloys obtained by AM [65, 76]. Thus, taking 3 wt.% of Si as the dissolved amount, Eq. (6) yields a strength contribution of 82.6 MPa, which is a more reliable and realistic magnitude.

Different grain sizes and distributions from the CG to the UFG regime were obtained for the as-built and HPT-processed materials, as indicated in Fig. 6. The grain size reduction obeys a mechanism of grain fragmentation due to the dislocation multiplication and aggrupation, creating new grain boundaries [80]. This mechanism was corroborated in Fig. 7b, indicating increased dislocation densities after HPT processing.

Precipitation-hardening Al alloys, unlike pure Al, can reach higher dislocation densities due to the precipitate–dislocation interactions, where precipitates can divide dislocations and act as new dislocation sources. Green arrows in Fig. 11c indicate some places where dislocation accumulates around the Si-rich precipitates and others where they form dislocations cells.

In this context, pure Al possesses a high stacking fault energy (SFE), which renders the dislocation generation more difficult than in low SFE materials; however, the alloying elements in Al alloys reduce the SFE and the recovery rate [81]. For that reason, the present alloy reaches UFG sizes down to 500 nm and high density of dislocations, as corroborated in Figs. 6c and 7b. It is worth mentioning that additively manufactured materials, especially those obtained by powder-based techniques, have been reported to possess high densities of dislocations in the as-built state due to the compression-tension cycles induced by the heating/cooling heterogeneity inside the MPs [42], which favors a faster grain refinement.

The inverse fast Fourier transform (I-FFT) representation from the red-dashed square in Fig. 11g confirms the above observations, indicating multiple edge dislocations and stacking faults (SFs) due to the intense plastic deformation (Fig. 11h). It allows observing the strong lattice distortion due to the precipitates, the dislocation accumulation, and the formation of stacking faults (SFs) after HPT processing at room temperature.

The most well-known equation to evaluate the grain size strengthening contribution in metallic materials is the one proposed by Hall and Petch, which establishes an inverse relationship between the yield strength and the average grain size as follows [82, 83]:where \({\sigma }_{0}\) represents the contribution from the friction and the Peierls–Nabarro stresses for Al alloys (10 MPa) [84], \({k}_{HP}\) is the Hall–Petch constant for Al alloys (0.14 MPa m^−1/2) [76], and \(d\) is the average grain size.

Taking the average grain sizes calculated by EBSD in Fig. 6c and using Eq. (7), the calculated grain boundary strengthening for the as-built and ¼ HPT materials are 46.2 MPa and 231.4 MPa, respectively. Taking into account all the contributions for both conditions as-built and ¼ HPT leads to calculated values for \({\sigma }_{y}\) of ~ 331.2 MPa and ~ 598.4 MPa, which agree well with the experimental values of 381 MPa and 640 MPa, respectively. The previous calculations indicate the dominant role of the precipitation component in both materials, followed by the grain boundary contribution, which is also a noteworthy component in the HPT-processed material, accounting for ~ 39.6% of the overall yield strength. It is important to recognize that the calculations above can only provide us with a rough estimate of the yield strength, as they fail to consider the interactions among different mechanisms, and the linear summation represents a preliminary approximation of the first order. Despite this, calculations offer insights into the potential contributions of specific microstructural characteristics.

The ductility improvement for all the HPT-processed materials can be related to different meso-, micro-, and nanoscale parameters. For example, HPT closes the porosity and decreases the Si-enriched network stiffness due to the cellular structure transformation into Si-rich particles of different sizes. In this regard, some investigations have corroborated that the partial disintegration of the cellular structure leads to ductility enhancements [18, 85].

It is worth mentioning that the warm HPT processing leads to larger elongations than the HPT processing at room temperature. This is attributed to temperature rise, modifying the cellular structure into UFG particles (see Fig. 5) and creating a more heterogeneous grain size distribution; more grains with sizes larger than 1 µm are presented in the samples processed at warm than at room temperature (see Fig. 6c). Thus, temperature creates more space for the dislocation motion, delaying the plastic instability and improving ductility to reach a better strength–ductility trade-off. Finally, the combination of AM and SPD processes can be a promising alternative to obtain unprecedented mechanical properties that cannot be obtained using only one of the mentioned techniques. These two processes complement each other improving densification and generating heterogeneous microstructural characteristics at the meso-, micro-, and nanoscales that conduct better strength and ductility values than the as-built state.