Room-Temperature Single-Step Production of Ultrafine-Grained Bulk Metallic Sheets From Al Powder

Components developed from metallic powders are extensively used in the automobile, aerospace, and cutting tool industries. The automotive industry alone accounts for 70% of the application of powder metallurgy [1]. However, there are two major drawbacks in these processes: the requirement for (1) elevated temperatures for long durations and (2) multiple processing steps to get the final product, which in most cases is also cumbersome and often requires different equipment. Additionally, the high temperatures used for consolidation also induce unfavorable grain growth and undesirable phase changes in the material during the processing which affects the properties of the material negatively [2,3]. The usage of high temperatures and multiple processing steps leads to increased production costs while making the whole process highly energy inefficient [4]. It is extremely important to satisfy these two conditions to make the future sustainable. Hence, it is most desirable to develop manufacturing processes that produce components in a single step without the use of high temperatures.

There is another category of processing techniques that have been utilized to consolidate metallic powders/chips. These are friction-based processes in which frictional heating is used to create bonding between the metallic constituents [4,9,10]. Tang and Reynolds [11] produced Al wires of diameter 2.5 mm from machining chips using friction extrusion. However, Whalen et al. [12] produced well-consolidated extrusions of 5 mm diameter from Al powder using the same method. Li et al. [9] consolidated Al chips into a disc of ∼25 mm diameter and thickness of roughly 10 mm using friction consolidation. Ma et al. [4] recently used friction extrusion to produce meter-long wires of oxide dispersion strengthened (ODS) copper of diameter ∼2.5 mm. Although both friction consolidation and friction extrusion are excellent one-step methods for powder consolidation, they have their limitations as well. For example, in the case of friction consolidation, the sample size is again a problem like HPT. On the other hand, in friction extrusion, the extrusion diameters cannot be large. Moreover, since both processes involve frictional heating, additional power is required to generate this heat through spindle rotation for the duration of the process.

Recently, the present authors have re-adopted a new metal-forming technique called the friction-assisted lateral extrusion process (FALEP) in which very large shear strains (∼20) can be applied in a single step to transform bulk metal billets into sheets [13]. This process evolved independently from another manufacturing technique developed by the present authors called the plastic flow machining (PFM) process [14–16]. In PFM, sheets were produced from a bulk specimen by a controlled lateral extrusion process in which the surface layer of about 1 mm thickness was removed by shear as the machined chip. However, unlike conventional machining chips, the chip formed using this method was constrained in the die to transform them into sheets. To produce the chip, the specimen had to be moved against the friction generated between the specimen and the die, which led to an increase in the pressures required to push the material. However, in FALEP, the friction force generated is used to aid the movement of the specimen and its subsequent deformation. This is achieved by inserting another moving piece between the die and the sample. This piece had a generated rough surface on the side facing the sample, which allowed it to pull the specimen with it rather than push it against the die. This drastically reduces the pressures required to deform the material as there is no work being done against friction. Another important aspect of FALEP is that it can produce extremely high plastic deformation to a level where the grain fragmentation process, which is a process of grain refinement during large deformation, reaches its saturation state. FALEP is also not limited by the sample size and meter-long sheets can be produced making it favorable for industrial scale-up. Due to the grain refinement occurring, this process also produces UFG microstructure. Moreover, FALEP does not induce strain heterogeneity, unlike SPD techniques. Such strain homogeneity is exhibited through the uniformity in grain shape, crystallographic texture, and disorientation distribution [13]. Furthermore, it also increases the Lankford parameter value (r-value), especially in the case of Al (r = 1.28). This parameter is a ratio of the strains along the width and the thickness, which characterizes the plastic anisotropy in metallic sheets and is important in determining undesirable effects like thinning and earing during deep drawing [17]. It is highly desirable to have high Lankford parameter values. In the case of FALEP of Al, this value measured is the highest compared to any metal-forming process [13].

In this work, we have attempted to consolidate Al powder in a single step at room temperature without any form of heating. We successfully obtained a bulk sheet of Al from powder with complete densification, UFG microstructure, and superior mechanical properties. Since there was no heating at any stage, the strength evaluated was the material’s green strength. We also compared the results with an Al specimen that was initially in bulk form (a solid billet) and was processed by FALEP. Interestingly, the consolidated powder specimen exhibited substantially superior strength compared to the Al billet after FALEP. We also performed crystal plasticity simulations using the Taylor-type lattice curvature-based approach to understand the deformation process better by comparing the texture evolution during FALEP for both materials. Another consequence of these simulations was arriving at a different origin narrative for the 45 deg-rotated cube (RC) component.

Al-1050 powder with a particle size of ≈50 µm and a grain size of ≈5 µm and Al-1050 billet with a grain size of ≈100 µm were the starting materials. The powders were poured into the FALEP die and consolidated at room temperature. The billet was inserted into the die and processed under similar conditions as powders.

The experimental platform and equipment used to perform FALEP is shown in Fig. 1(a). There were three punches that were driven by their respective hydraulic piston-cylinder system. The maximum force capacity of each hydraulic punch was 72 tons. The normal punch and the driving punch were driven by piston 1 and 2, respectively. These two pistons controlled the movement of the main punches used in FALEP. The horizontal punch driven by piston 3 was used to close the gap between the channel and the die and prevent the loss of Al powder from the gap. In between the driving punch and the horizontal punch, a plunger was placed. The dimensions of this plunger could be varied, which allowed for different extrusion ratios and shear strain induced in the material.

The experimental setup for Al powder is shown in Fig. 1(b). The vertical channel was in a square shape of 20 mm × 20 mm, while the channel between the plunger and the die was set with a gap of 1.5 mm in order to obtain an extrusion ratio of 13.3. As mentioned earlier, this channel width could be varied to obtain different extrusion ratios. The smallest gap used so far was 0.5 mm. During processing, the normal punch was controlled by a constant normal load (P_N). The process parameters for the experiments are presented in Table 1. The driving load (P_O) on the driving punch increased progressively to the maximum value (provided in Table 1). These loading values were measured in real time during processing by the various sensors installed in the machine. The displacements of all punches and the length development of the produced consolidated sheet were also obtained in real time via this sensor system.

The processing of FALEP is shown in Fig. 1(c). In the present configuration, FALEP was performed first by moving the normal punch downward to compress the Al powder. The movement of the normal punch was determined by the limiting load that was set, which in this case was 45 tons. Then while keeping the normal punch in position, the driving punch was moved (note that with improvements in the machine, the movement of the normal and driving punches occurs simultaneously in the latest configuration of the machine). The driving punch pushed the plunger in the horizontal direction. Since the upper surface of the plunger had higher surface roughness due to sand blasting, it aided in extruding out the compressed Al powder. Hence, there was a friction force F_fr between the upper surface of the plunger and the bottom of sample in the direction of its extrusion, thus assisting the material to flow into the gap and eventually flow out laterally throughout the outlet between the die and the plunger. During such an extrusion, the direction of the material flow changed by 90 deg, which induced shearing in the sample. The final machine configuration after FALEP is shown by a schematic diagram in Fig. 1(d).

The microstructure of the specimen produced from the bulk sample was examined by electron backscattered diffraction (EBSD) using a JEOL JSM-6500F field-emission gun-scanning electron microscope (SEM). For the initial powder sample, focused ion beam-scanning electron microscopy and focused ion beam-electron backscatter diffraction (FIB-EBSD) (Zeiss Auriga) were used. For EBSD analysis, atex software [18] was used. The minimum disorientation angle between neighboring pixels was set to 5 deg for the detection of grain boundaries (GBs). Also, transmission electron microscopy (TEM) (Philips CM 200) was employed for a better understanding of the microstructure produced by the Al powder sample. To examine mechanical properties, tensile tests were performed at room temperature using a universal testing machine (Zwick-Roell Z050). The tensile direction was set parallel to the longitudinal direction of the sample, i.e., along the extrusion direction, and the strain rate was 0.001 s⁻¹.

In Fig. 2, the in situ load–displacement curves of the driving punch for the initial powder specimen after FALEP (henceforth termed powder-FALEP) and of the sample that was initially an Al bar but then processed by FALEP (henceforth termed bulk-FALEP) are shown. The values of the load and displacement were obtained directly from the machine in real time via various force and displacement sensors built in the machine. In the case of the powder-FALEP, it can be observed that for the initial displacement of 3.8 mm, there was a rapid increase in the load from 0 to ∼129 kN. After that, there was a steady increase in the load from 129 kN to 157 kN for a displacement of 51.2 mm. The initial rapid increase can be attributed to the initiation of plastic flow within the shear zone of the powder. On the other hand, the steady increase after the first 3.8 mm, although quite small, could be due to the improvement in the consolidation as the process progressed. The load–displacement curve of the bulk-FALEP sample also exhibited a similar trend with an initial rise to ∼89 kN for the first 3.8 mm. However, there was no steady increase in the driving load after the initial rapid rise. This is expected as the sample was already in the bulk form. The total work done for the process could be calculated by taking the area under the curve as shown in Fig. 2, which were 7894 J and 5446 J for the powder-FALEP and bulk-FALEP samples, respectively. The duration of the movement of the driving punch for powder-FALEP was 2 min and 5 s, and hence, the power required for this part of the process was 46 W. On the other hand, for the bulk-FALEP, the power required was 72 W. It is to be noted here that such values have been provided not for comparison purposes between powder-FALEP and bulk-FALEP, but rather to get a sense of the power required for this process as a whole. The values will vary depending on the process parameters used.

In Figs. 3(a) and 3(b), the initial powder specimen after FALEP is shown. The former shows the top view and the latter shows the side view. It can be observed that the sample was well consolidated with a good surface finish. The density value of the specimen measured by Archimede’s principle was 2.69 ± 0.01 g/cm³. In the SEM micrographs in Figs. 3(c) and 3(d), it can be observed that there were no porosities present, which is also a proof of proper consolidation. These micrographs were obtained near the top edge of the specimen which has been indicated in Fig. 3(b). Additionally, particles with higher atomic contrast compared to the rest of the specimen could be observed. These particles appeared white in color in the SEM micrographs and were observed to be distributed throughout the specimen. Such white particles were possibly fragmented aluminum oxide layers found on the surface of the Al powder particles. Similar reports on the presence of fragmented aluminum oxide particles have been made by others as well and have been observed to affect the mechanical properties of the material [19,20]. The effect of such particles has been discussed further in this work in Sec. 3.3.

With the shear of γ = 13.41, the particle surface increased 6.87 times. Under these conditions, the oxide coating fragmented during deformation, and the virgin metallic surfaces became cold welded between each elongated metal particle. In this way, the fragmented oxide particles appeared as a dispersed phase throughout the specimen (Figs. 3(c) and 3(d)).

In Fig. 5(a), the inverse pole figure (IPF) map of the powder-FALEP sample in the ND-FD plane (containing the normal and forward directions) is shown. The projection axis was along the transverse direction (TD). The specimen possessed the UFG microstructure with an average grain size of 500 nm (number-weighted average). Figure 5(b) shows the IPF map of the initial bulk specimen after FALEP in the ND-FD plane. The microstructure of this specimen was similar to the powder-FALEP specimen but with a larger average grain size: 600 nm (also number-weighted average). One of the unique features of FALEP is a lack of deformation gradient in the specimen produced due to the absence of strain heterogeneity during the processing. This results in a more homogenous microstructure with uniformity in grain structure, texture, and disorientation distribution. Such findings have been discussed by Vu et al. in their work [13]. The lack of a deformation gradient can be observed in the present work also through the uniformity in the grain structure in both powder-FALEP and bulk-FALEP specimens in Figs. 5(a) and 5(b), respectively.

As explained earlier, FALEP results in the elongation of each Al powder particle; however, this also results in the fragmentation of the oxide layers present on the surface of those powder particles. Some of the oxide particles when present at the GBs (Figs. 3(c) and 3(d)) would pin the moving GBs, thus leading to an overall smaller average grain size in powder-FALEP. In Figs. 5(c) and 5(d), the grain boundary maps of the powder-FALEP and bulk-FALEP specimens are shown. The maps have been divided by grain boundary fractions for HAGB (high angle grain boundaries, misorientation angle 15 deg–61 deg), LAGB (low angle boundaries, misorientation angle 5 deg–15 deg), and subgrain boundaries (misorientation angle 3 deg−5 deg). In both specimens, the largest fraction was for the HAGB, which indicates that recrystallization occurred. However, this is confirmed by the grain orientation spread (GOS) maps that are used to distinguish between recrystallized and nonrecrystallized regions. Higher GOS values are registered by those grains that have higher geometrically necessary dislocation content [22]. Hence, recrystallized grains have low GOS values. From the maps in Figs. 5(e) and 5(f), it can be observed that the majority of the grains possessed low GOS values, typically less than 2 deg (only grains with values less than 2 deg are considered recrystallized grains [22]). This confirms that room temperature recrystallization occurred in the specimen due to the high shear strain generated during the process. Since, the recrystallization occurred at room temperature under very high shear conditions for Al, which is a high stacking fault energy material [23], new strain-free grains formed by the continuous dynamic recrystallization mechanism.

In Fig. 6, the TEM micrographs of the powder-FALEP specimen are shown on the ND-FD plane. In Fig. 6(a), particles could be observed both within the grains and at the grain boundaries (marked by yellow boxes). Such particles are shown at higher magnification in Fig. 6(b). The ones within the grains acted as obstacles to the dislocation motion (Figs. 6(c)–6(e)). Hence, it can be understood that such particles would contribute to the strengthening of the material. The particles did not show any diffraction contrast when the specimen was tilted at different angles in the TEM, which indicated that the particles were amorphous. The bright field image of the region for obtaining the selective area diffraction (SAD) pattern (figure inset) of the particle is shown in Fig. 6(f), which is marked by a red circle. It can be observed in the SAD pattern, and a diffuse ring was present along with the diffraction spots from the Al matrix. The diffuse ring from the particle confirmed its amorphous nature. In the examined material, the only possible origin for such amorphous particles is the initial oxide layer around each powder particle. Previous studies have shown that it is possible to have amorphous aluminum oxide layers around such aluminum particles at room temperature [19,20].

The presence of oxide particles within a grain can be understood by the shear-coupled grain boundary migration theory [24–26]. It has been experimentally found that GBs can migrate due to the applied shear stress [27]. This motion is perpendicular to the direction of shear stress when the grain boundary is parallel to it. Molecular dynamics simulations have shown that this phenomenon occurs for all GBs under all conditions, irrespective of the temperature [26]. In the present scenario, due to the large shear stresses involved during FALEP, such grain boundary migration was possible, which led to the oxide particles being engulfed within the grain interior. In some regions, the oxide particles appeared to be laid in a sequence (Figs. 6(g)–6(i)). Such features can be expected when the fragmented pieces of the original oxide layer remain attached to the Al particle boundary during shearing.

In Fig. 7(a), the positions of the ideal texture components of face-centered cubic (FCC) shear textures are shown. In Fig. 7(b), the orientation distribution function (ODF) of the powder-FALEP sample, measured in the ND-FD plane, is shown. The specimen exhibited a simple shear texture with a strong C-component {100} <110 > . Additionally, the B and $B¯$ components were also observed, with lower intensities, compared to the C-component. Moreover, the 45 deg-RC component was also observed, which is not a stable shear texture component.

Here, $Ω˙_int$ is the lattice spin vector in the center region of the grain, μ is a lattice curvature controlling parameter, and N_GB is the number of GBs that affects $Ω˙_ext$ due to the lattice curvature. μ can range from 0 to 1 and controls the extent of grain fragmentation. The strain mode of the FALEP deformation was idealized by the velocity gradient of simple shear along the intersection plane of the two channels.

Here, $τ0i$ is the reference strength of the slip system indexed by i, $γ˙j$ is the shear rate in the slip system, and q^ij expresses the hardening interaction between slip systems. Based on the geometrical configuration between two slip systems, four q^ij coefficients were distinguished; they correspond to coplanar (g₁), collinear (g₂), perpendicular (g₃), and other slip geometry cases (g₄). The hardening parameter values were tuned on a separate torsion test and are shown in Table 2 [30]. The initial texture was represented by 500 randomly oriented grains. This number progressively increased up to 1,079,005 grains due to the grain fragmentation process. Simulations were performed by varying the strain rate sensitivity index m in the viscoplastic slip law along with the lattice curvature parameter μ. By comparing with experimental measurements, m = 0.125 and μ = 0.07 were found as the most appropriate values. With these parameters, the polycrystal grain fragmentation code reproduced the experimental decrease of the average grain size (from ≈5 µm to 0.5 µm). Figure 7(c) shows the simulated texture, which is very close to the experimental one. Quantitative comparison made using the texture correlation index in the atex software [18] gave a value of 0.922, which indicates a high agreement in texture simulations because the crystallographic texture is a 3D quantity in orientation space (maximum correlation index value is 1.0).

Figure 7(d) shows the experimentally measured ODF of the initial bulk Al specimen before FALEP, which exhibited nearly random texture. Figure 7(e) shows the experimentally measured texture of bulk-FALEP, which also exhibited a simple shear texture with the C-component being dominant. However, in this specimen, the A1 and A2 components were also prevalent, together with the B and $B¯$ components. The RC component was also observed in this specimen; however, its intensity was much lower compared to powder-FALEP.

For bulk-FALEP texture simulation, the initial texture comprised 100-grain orientations selected from the measured texture of the initial bulk Al sample. Due to the very high degree of grain refinement in the experiment from 100 µm to 0.6 μm, the simulation was performed in two stages to reproduce the same final average grain size. This was achieved by setting the values of m and μ in the first stage to 0.125 and 0.3 up to a shear strain of 10, whereas in the second stage to 0.167 and 0.2, respectively, up to the final shear strain of 13.4. All other parameters were kept the same in the two-stage simulation. The second stage started with 500-grain orientations, which were discretized from the texture of the first stage. This was necessary because the grain number increased to an extremely large value (1,247,464) in the first stage and because the three-level fragmentation code could not reach the 0.6 μm final average grain size in one stage. Figure 7(f) shows the simulation results. The texture correlation index for bulk-FALEP was 0.801, which is also a high value for texture simulation.

In the simulations for both specimens, the RC texture component could be reproduced similarly to the experimental one. Although the development of the RC component was previously associated with dynamic recrystallization (DRX) [13], the fact that this component could be produced without DRX simulation suggests that it is in fact a deformation-induced texture component. The RC component is situated along the ϕ = 0 plane of Euler space along which all FCC crystal orientations have zero plastic spin, so their lattice orientation rotate continuously with the rigid body rotation, imposed by the shear deformation (the lattice spin is the difference of the rigid body spin and the plastic spin [33]). Therefore, one would expect that a texture component could not be formed along ϕ = 0, so the RC component should not be appearing as a stable component. However, the constant lattice rotation is only valid on the exact ϕ = 0 plane, while there is an orientation flow near the ϕ = 0 plane at the vicinity of the location of the RC component with low velocity [33]. This orientation flow, which is coming from positions far from the ϕ = 0 planes, is producing a temporary accumulation of orientations by its low motion, making in this way an apparent texture component. This component is captured in the grain fragmentation simulations because an extremely high grain number is developing during large strain, which makes the appearance of fine details of the texture possible.

In Fig. 8, the tensile curves of powder-FALEP and bulk-FALEP are compared with the initial bulk Al. Interestingly, the powder-FALEP specimen exhibited higher strength compared to the latter, which suggests that the powder was very well consolidated. The 0.2% yield strength (YS) and ultimate tensile strength (UTS) values for the powder-FALEP specimen were 128 MPa and 262 MPa, respectively. The uniform elongation of the powder-FALEP specimen was ∼9% (which corresponds to the peak value in the engineering stress–strain curve), after that the specimen underwent necking and the deformation continued leading to fracture at an elongation of ∼14%. The YS and UTS values for the bulk-FALEP specimen were 116 MPa and 193 MPa, respectively. The uniform elongation was ∼7.2%, and after necking, the specimen continued to deform, leading to fracture at ∼24%. The YS and UTS of the initial bulk material were 31 MPa and 69 MPa, respectively. The uniform elongation was ∼33%, and after necking, the specimen continued to deform, leading to fracture at ∼54%. The values show that both the powder-FALEP and the bulk-FALEP specimens exhibited substantial improvement in strength compared to the initial bulk Al specimen. However, there was a reduction in the uniform elongation values, a consequence of the strength–ductility paradox.

In most SPD-processed specimens, the strength improves due to the Hall–Petch mechanism [34–38]. Hence, in both powder-FALEP and bulk-FALEP, this mechanism would have been active, which improved their strengths significantly compared to the initial bulk Al sample. However, the difference in their grain size would not produce a significant variation in their YS. Using the Hall–Petch parameters of aluminum, the ratio of their yield strengths can be estimated to be only about 1.1 [39,40]. Hence, another strengthening mechanism must have been active. It is well known that the presence of oxide particles can strengthen a material, especially if they are present in the grain interior [41–44]. Such particles were visible in the TEM micrographs (Fig. 6), acting as barriers to dislocation motion. Hence, oxide particles could have strengthened the powder-FALEP sample. Furthermore, the oxide particles at the GBs could also contribute to the strengthening of the sample by increasing the density of geometrically necessary dislocations (GNDs) near the GBs to accommodate the strain incompatibility during deformation. Hence, both these mechanisms would enhance the strength of the powder-FALEP sample compared to the bulk-FALEP.

In this work, consolidation and complete densification of Al powder with UFG microstructure at room temperature was successfully performed in a single step using FALEP. Investigations on the microstructure, crystallographic texture evolution, and mechanical behavior were made and compared with an initial bulk specimen that underwent the same processing conditions. The following conclusions can be deduced from the results obtained:

1. The Al powder was well consolidated with a density value comparable to a bulk Al specimen. Moreover, the sample also possessed an excellent surface finish, which indicated no further processing was required to improve the surface quality.

2. The powder-FALEP specimen exhibited the distribution of oxide particles throughout the specimen thickness. In the TEM micrographs, these particles were observed to be pinning the dislocations within the grains. However, they were also present at the grain boundaries. Their presence within the grains could be explained by the shear-coupled grain boundary migration theory.

3. The powder-FALEP specimen exhibited equiaxed grain morphology with an average grain size of 500 nm. However, the bulk-FALEP specimen exhibited similar morphology with an average size of 600 nm. The decrease in the grain size in the former could be attributed to the presence of the oxide particles that could act as pinning points for the moving grain boundaries.

4. Crystallographic texture measurements of the powder-FALEP specimen exhibited the dominance of the C-component of the FCC shear texture followed by the B and $B¯$ components. However, the 45 deg-rotated cube component was also observed in this specimen, which is not an FCC shear texture component. The bulk-FALEP sample also exhibited similar textures; however, it also possessed the A1 and A2 components.

5. The texture evolution was studied by performing simulations using the Taylor-type viscoplastic grain fragmentation model. There was good agreement between the experimental and simulated textures for both powder-FALEP and bulk-FALEP with texture correlation indexes of 0.922 and 0.801, respectively. These results exhibited that the material deformed by shear with the activation of both the octahedral slip family {111} <110 > and the nonoctahedral slip family {100} <110 >.

6. The simulations could also produce the 45 deg-rotated cube texture component. This component was earlier thought to be produced due to DRX. However, in this work, it was revealed that this component was produced due to deformation, which could be attributed to the orientation flow during the grain fragmentation process.

7. The stress–strain curves revealed that the powder-FALEP specimen exhibited the highest strength. Interestingly, it was higher than the bulk-FALEP specimen, which could be attributed to the presence of the oxide particles that were observed to be barriers to dislocation motion. Moreover, there was also a possibility of them producing GNDs due to strain incompatibility between the particles and the Al matrix.

This work was supported by the French State through the program “Investment in the future” operated by the National Research Agency (ANR) and referenced by ANR-11-LABX-0008-01 (LabEx DAMAS). The authors would like to thank Thai Nguyen University of Technology (TNUT) for the support, Dr. Olivier Perroud (LEM3, LabEx-DAMAS, Metz, France) for the help provided in the XRD texture measurements, Dr. Yudong Zhang (LEM3, LabEx-DAMAS, Metz, France) for the help provided during TEM studies, and Dr. Julien Guyon (LEM3, LabEx-DAMAS, Metz, France) for the help in TEM sample preparation.

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.