Synergistic strengthening of crack-free Al–Zn–Mg–Cu alloys with hierarchical microstructures achieved via laser powder bed fusion

Crack-free Al–Zn–Mg–Cu alloy, processed by laser powder bed fusion, displays a characteristic microstructure comprising fine equiaxed grains (EGs) and coarse columnar grains (CGs). Notably, the CGs show a high hardness compared to EGs due to the high density of in-situ precipitates. Analysis of mechanical properties takes into account the precipitates in the grain and hierarchical differences in grain morphologies near the melt pool boundary. Strengthening mechanisms are elucidated through synergistic hetero-deformation-induced strengthening, where multi-modal-sized EGs induce significant strain gradients within EG regions. Our findings demonstrate the potential to enhance strength of Al–Zn–Mg–Cu-based alloy in as-built state through architecting microstructures. GRAPHICAL ABSTRACT IMPACT STATEMENT The Al–Zn–Mg–Cu alloy, fabricated by laser powder bed fusion, exhibiting hierarchical microstructure consisting of hard coarse and soft fine grains shows synergistic in-situ precipitation and hetero-deformation-induced strengthening.


Introduction
The demand for lightweight and high-strength components has led to increased interest in laser powder bed fusion (LPBF) for fabricating parts of the high-strength Al-Zn-Mg-Cu alloy, particularly in automobile and aerospace industries [1,2].However, the vulnerability of conventional high strength Al alloy to hot cracking introduces a significant hurdle in the quest for flawless part production.To address this issue, the implementation of inoculants to facilitate grain refinement has emerged as a potent approach for effectively countering the challenges posed by hot cracking [2][3][4].Elements such as Zr [2], Ti [5], Nb [6], and Sc [7] serve as grain refiners by forming the primary Al 3 X (X = Zr, Ti, Nb, and Sc) phases, which accelerates the nucleation of α-Al phase by providing abundant nucleation sites.An inoculant-added Al-Zn-Mg-Cu alloy can form crack-free microstructures if it forms fine equiaxed grains (EGs) instead of coarse columnar grains (CGs).Recently, samples with controlled fractions of EGs and CGs by adjusting the amount of inoculant and processing parameters demonstrated good performance in LPBF [7][8][9][10][11].Zhu et al. [7] have demonstrated that Al-Zn-Mg-Cu-Sc-Zr alloys with hierarchical microstructures through LPBF possess excellent strength-ductility combinations.This hierarchical microstructure induces hetero-deformationinduced (HDI) strengthening and contributes to the superior mechanical properties [12].
HDI strengthening occurs due to differences in strengths between soft and hard domains.During deformation, soft domains yield earlier than hard domains, initiating plastic deformation [13][14][15].Geometrically necessary dislocations (GNDs) accumulate in the boundary regions of soft domains to maintain strain continuity, generating back stress in the soft domains and forward stress in the hard domains [15][16][17].Consequently, soft domains can withstand higher stress because the HDI stress counters the applied shear stress.Wang et al. [18] revealed that grain boundary strengthening creates differences in tensile properties between CGs and EGs in LPBF-processed Al-Mg-Sc-Zr alloys.Fine EGs exhibit higher strength than coarse CGs due to grain boundary strengthening, resulting in localized deformation primarily in coarse CGs.However, Zhu et al. [7] reported that the localized deformation was higher in find EG than in coarse CG regions in Al-Zn-Mg-Cu-Sc-Zr alloys, suggesting that the origin of strength differences between the two regions may vary depending on alloy composition.
Despite these findings, the HDI strengthening behavior observed in the inoculant-added Al-Zn-Mg-Cu alloy has not been elucidated yet.Thus, we have designed and fabricated Al-Zn-Mg-Cu alloy with the addition of a ZrH 2 inoculant to confirm the possibility of synergistic HDI strengthening when the difference in strain between EG and CG regions is maximized.

Materials and methods
The 1 wt% ZrH 2 -containing Al-Zn-Mg-Cu powder was additively manufactured using a laser powder bed fusion machine (Model: M2, Concept Laser GmbH).This alloy was designed to prevent solidification cracking during LPBF [2,8].Details of properties and mixing methods can be found in Ref. [8].The following process parameters were chosen: 325 W laser power, 600 mm/s scan speed, 100 µm hatch space, 30 µm layer thickness, and 100 µm focused beam diameter.These parameters were determined based on our previous work to ensure a high relative density with crack-free fabrication during the LPBF process [8].Thin plate-shaped samples were built with dimensions of 100 × 1.5 × 11 mm 3 in length, width, and height, respectively.The samples are denoted as the Al-Zn-Mg-Cu-Zr alloy.
The microstructure of the samples was analysed on a plane parallel to the build direction using XL30 FEG (Philps N.V.) equipped with electron backscatter diffraction (EBSD) and Aquilos 2 (Thermo Fisher Inc.) equipped with a backscatter electron (BSE) detector.Energy dispersive spectroscopy (EDS) was used to characterize the micro-scale precipitates and segregation of alloying elements.Transmission electron microscopy (TEM) analysis was performed using JEM-2100F (JEOL Ltd.), and specimens for TEM analysis were prepared using Helios 5 UX (Thermo Fisher Inc.).
The micro-Vickers hardness was measured using a load of 5 g and a holding time of 15 s.Thirty indentation traces were made in each region and the average hardness values were obtained.The micro-scale digital image correlation (DIC), monotonic tensile test, and loading-unloading-unloading (LUR) test were conducted using samples with a 1.5 mm gauge length.Scanning electron microscopy images for micro-scale DIC, obtained using a JSM-7800F (Jeol Ltd.) before deformation and at global strains of 2% and 4%, were analysed using GOM Correlate 2018 software (Carl Zeiss GOM metrology GmbH).The monotonic tensile test and LUR test were performed at a strain rate of 10 −3 s −1 using an Instron 5582 machine.

Results and discussion
Figure 1(a and b1) show the mixed grain structure.Along the melt pool boundaries, regions of CGs and EGs are consistently observed.Figure 1(b2) exhibits the CG region.Figure 1(b3) exhibits the EG region characterized by a blend of ultrafine equiaxed grains (UEG) with sizes on the order of 1 µm and medium-sized equiaxed grains (MEG) several µm in size.The element segregation and precipitates between CG and EG are shown in Figure 1(c  and d), respectively.The number of countable precipitates within the grain interior of CG and EG regions were measured as 5.92 and 1.06 precipitates/µm 2 , respectively, indicating a higher density of precipitates in the CG regions compared to the EG regions.Precipitates rich in Zn, Mg, and Cu, marked with yellow boxes in Figure 1(c), are observed in the CG regions.At the grain boundaries of the CG region, marked with blue arrows in Figure 1(c), a high concentration of Cu is detected.The selected area electron diffraction (SAED) pattern in the inset of Figure 1(c) confirms that this phase is θ-Al 2 Cu [19,20].At the same time, two types of precipitates were identified in the EG regions, although with a lower number density of precipitates than in the CG regions.The precipitates included Zn, Mg, and Cu-rich precipitates, also found in the CG regions.The Zr-rich precipitates, marked with a magenta box in Figure 1(d), were identified as the L1 2 -Al 3 Zr phase based on the SAED pattern, shown in the inset in Figure 1(d) [21].Instead of less precipitation in the EG region, Zn, Mg, and Cu were found to segregate at the grain boundaries.Based on the aforementioned results, Figure 1(e) exhibits a schematic of the global and unit regions, representing the melt pool (MP) with melt pool boundary (MPB) and melt pool interior (MPI).An MP unit is composed of EG (mixture of UEG and MEG) and CG regions.The hierarchical microstructure of this alloy was attributed to the differences in thermal history between the MPB and the MPI regions.The finite element method (FEM) simulations of the thermal conditions within MP have suggested that the cooling rate of MPB is lower than that of MPI [22,23].The range of the cooling rate distribution in MPI (1.23 × 10 6 K/s ∼ 3.58 × 10 6 K/s) is assumed to fall within the critical cooling rate (1 × 10 6 K/s ∼ 8.7 × 10 6 K/s) required for the nucleation of the Al 3 Zr phase (at 1.0 wt% Zr) [8,24].Accordingly, the absence of Al 3 Zr in the CG region was attributed to the supersaturation of Zr resulting from the high cooling rate [6,25,26].Furthermore, the thermal histories of the EG and CG regions play a crucial role not only in the formation of the primary L1 2 -Al 3 Zr phase but also in the precipitation and segregation behaviors of other alloying elements [27].The higher cooling rate in the CG regions likely caused the entrapment of Mg, Zn, and Cu within the grain interior.The maximum solid solubilities of Mg, Zn, and Cu in the Al matrix have been reported as 9.6, 83.1, and 5.7 wt%, respectively, based on the binary phase diagrams of the alloy systems [28].Thus, as the tendency for grain boundary segregation is inversely proportional to the solid solubility, the grain boundary segregation of Cu in the CG regions was attributed to its lower solid solubility compared to the other elements [29].Figure 1(f) exhibits the hardness measurement results.The average hardness of the CG region is higher than that of the EG region, contrary to the expectations based on the Hall-Petch relationship, which represents the strengthening effect of grain boundaries [30,31].This difference was attributed to the reinforcing effect of a higher number density of precipitates containing Zn, Mg, and Cu in the CG region, along with the effect of their solid solution.At the same time, the hardness distribution in the EG regions well matches the Hall-Petch relationship.The average hardness of the UEG region is higher than that of the MEG region but lower than that of the CG region.
Figure 2 shows the engineering stress-strain curves of the Al-Zn-Mg-Cu-Zr alloy and other Al alloys with inoculants as reported in [18,24,32].All these alloys contain Mg as an alloying element and were inoculated with the primary Al 3 X (X = Zr, Sc) phase.The Al-Zn-Mg-Cu and Al-Cu-Mg alloys exhibit higher strength and lower ductility than Al-Mg alloys, indicating precipitation hardening or solid solution hardening effects of Zn and Cu, even in the as-built condition.The Al-Zn-Mg-Cu-Zr alloy exhibits a yield strength (YS) of 426.8 ± 1.0 MPa, an ultimate tensile strength (UTS) of 487.2 ± 10.3 MPa, and a total elongation of 11.5 ± 1.8%.
Figure 3(a) shows the local equivalent strain distributions, measured using a micro digital image correlation (micro-DIC) method.As global strain (ε g ) increased, the local strains were concentrated near both the CG/UEG and UEG/MEG interfaces, known as interface-affected zones (IAZs).Figure 3(b) more clearly shows strain concentrations in interfaces as a line profile of the local equivalent strain (ε e ) along the white line in Figure 3(a3) at the global strains (ε g ) of 2% and 4%.The greater strain gradients in the IAZs of UEG/MEG than those in the CG/UEG were attributed to two reasons: (1) the higher mechanical incompatibility between UEG and MEG compared to that between CG and UEG [33,34]; (2) the lower average hardness of EG regions compared to that of CG regions.The latter can be explained by the linear relationship between the strain gradient within the IAZ and the applied strain [35].Based on the hardness measurement results (Figure 1(f)), it was anticipated that the YS of the EG regions would be lower than that of the CG regions.Therefore, it is logical to expect that the EG regions, with lower YS, would undergo larger deformation than CG regions.The average equivalent strains in each domain are presented in Table 1.The ε e in UEG and MEG are higher than that in CG. Figure 3(c and d) present the GNDs density maps of the undeformed and deformed samples, respectively.It is evident that the GND density in the EG region is higher than that in the CG region in both undeformed and deformed samples.The phenomenon is presumed to arise from the lower hardness of the EG region in Al-Zn-Mg-Cu-Zr.The accumulation of GNDs at the interfaces, as shown by the micro-DIC results and GNDs density maps, is schematically presented in Figure 3(e).The n1, n2, and τ a are the number of GNDs in the pile-up in CG/UEG and UEG/MEG interfaces, and the applied shear stress, respectively.In the CG regions, which represent the hardest domain in the Al-Zn-Mg-Cu-Zr alloy, dislocation movement was hindered by precipitates and solute atoms, resulting in the accumulation of GNDs in the UEG regions at the CG/UEG interface [36].At the same time, in the EG regions, the high grain boundary strengthening effect caused by the high volumetric density of grain boundaries in the UEG region led to mechanical incompatibility at the UEG/MEG interface, causing the accumulation of GNDs in the MEG region [37].Nevertheless, the back stresses are expected to be generated at both CG/UEG and UEG/MEG interfaces.[24,38,39], Al-Mg-Sc-Zr [18,40], and Al-Mg-Zr alloys [32].
effect resulting from the changes in microstructural heterogeneity rather than the strengthening effect of precipitation and solid solution, as illustrated in Figure 4(c), where σ p represents additional precipitation strengthening, σ s represents additional solid solution strengthening, and σ HDI represents additional HDI strengthening. Figure 4(d) shows an Ashby plot of YS vs uniform elongation of the as-built Al alloys [7,18,24,32,[38][39][40].The Al-Zn-Mg-Cu alloys demonstrate higher yield strength while minimizing the reduction in uniform elongation.This can be attributed to the additional HDI strengthening effect, which arises from the accumulation of GNDs within the soft EGs due to the presence of hard CGs.We expect that the additional HDI strengthening effect can be maintained even if changes are made to the heat treatment process or materials, provided that this characteristic microstructure can be preserved.

Conclusion
In summary, the present study introduces a novel approach for achieving a synergistic effect through the HDI strengthening driven by strain differences in hard CG and soft EG in the control of the mechanical properties of the Al-Zn-Mg-Cu-Zr alloy.The formation of a CG structure, attributed to a higher cooling rate within the melt pool, effectively suppressed the formation of the primary Al 3 Zr phase, leading to the entrapment of other alloying elements (Zn, Mg, and Cu) within the grain interior.This resulted in a significantly higher number density of precipitates in the CG regions.At the same time, the lower cooling rate near the MPB favored the formation of distinct EG regions but resulted in the segregation of Zn, Mg, and Cu at the grain boundaries.The EG regions were further classified into UEG and MEG regions based on grain size, exhibiting notable mechanical incompatibility due to grain boundary strengthening effects.Micro-DIC and microstructure analysis revealed that significant strain gradients were generated near the CG/UEG and UEG/MEG interfaces during tensile deformation due to the hierarchical microstructure consisting of hard CGs and soft EGs.Notably, the strain gradient near UEG/MEG introduced higher HDI stress, thereby contributing to the excellent strength-ductility combination of the Al-Zn-Mg-Cu-Zr alloy.Overall, these findings provide valuable insights for the design of novel high-strength Al alloys with tailored bimodal grain structures intended exclusively for LPBF.

Figure 1 .
Figure 1.(a) EBSD inverse pole figure map of the as-built alloy.(b1) BSE image of the boundary between CG and EG regions.Highmagnification BSE image of (b2) CG and (b3) EG regions.TEM-EDS results for (c) CG and (d) EG regions.SAED patterns in (c) and (d) correspond to the diffraction patterns of θ-Al 2 Cu and L1 2 -Al 3 Zr, respectively.(e) Schematic of grain morphology and distribution of precipitates in the Al-Zn-Mg-Cu-Zr alloy.(f) Hardness measurement results at each region in as-built Al-Zn-Mg-Cu-Zr alloy.

Figure 3 .
Figure 3. (a) Micro digital image correlation results.(a1), (a2), and (a3) Local equivalent strain (ε e ) maps at the initial state and the global strains of 2% and 4%, respectively.(b) Line profile of ε e from A 0 to A 20 along the white line in Figure 3(a3).(c) and (d) GNDs density maps of the initial state and at the global strains of 4%, respectively.(e) Schematic of the accumulation of GNDs at the interfaces.

Figure 4 (
Figure 4(a and b) present the results of the LUR test with the contributions of HDI stress and effective stress on flow stress, respectively.The data for

Table 1 .
Average equivalent strains in each domain.