Introduction

Solid-phase processing (SPP) methods, such as friction stir processing (FSP)1 and shear-assisted processing and extrusion2, employ extreme shear deformation to achieve highly refined microstructures with improved mechanical properties in structural materials3,4,5,6,7. Extreme shear deformation is also prevalent in the field of tribology and occurs due to friction between two surfaces that are in contact and in relative motion8,9,10. Recently, the atomic-scale mechanisms of microstructural evolution under shear deformation have been a topic of growing interest11,12,13,14,15,16. Specifically, it has been hypothesized that there is competition in immiscible binary alloys between opposing forces of shear-induced chemical intermixing and thermally activated diffusion17. This competition has been postulated to result in increased solubility of elements, thereby causing the formation of metastable supersaturated solutions, even in alloys with moderately positive enthalpies of mixing (such as Cu–Ag)13,18. An et al. detailed the interfacial changes and dislocation transfer across the severely shear deformed Ag–Cu interfaces19. Nanolayer formation through morphological self-organization has also been reported after shear deformation of highly immiscible alloys such as Cu–Nb11,20. However, when considering alloy systems with a ductile matrix and a brittle second phase precipitate that has limited solubility in each other, it is currently unknown how shear deformation will modify the microstructure. Open research questions include: (1) will the shear deformation lead to hierarchical fragmentation of the brittle phase and inward diffusion of the matrix element; (2) to what extent will grain refinement occur for the ductile matrix; and (3) can true solute supersaturation of the matrix phase be attained at an atomic scale? Additionally, there is very limited understanding of the impact of shear deformation on microstructural evolution and mechanical properties.

To address these knowledge gaps, we selected a binary Al-4 at.% Si alloy, which is a castable alloy widely used in the defense, aerospace, and automotive industries. This lightweight alloy offers high strength, corrosion resistance, good thermal conductivity, as well as a low coefficient of thermal expansion21,22,23. The coarse eutectic structures, massive Si particles, porosity, and uneven shrinkage typically exhibited in this alloy system can be major drawbacks when processed using conventional casting methods24,25,26,27. Mechanical properties can be improved in Al–Si alloys by producing a fine-scale distribution of Si particles and eliminating porosity. Heat treatments and chemical modifications have been used to refine the eutectic microstructure of Al–Si alloys27,28,29. Heat treatments at high temperatures for extended time durations can be used to spheroidize Si, but such treatments also coarsen the Si particles, degrading mechanical properties and increasing production costs30,31. Additions of sodium (Na), antimony (Sb), and strontium (Sr) are commonly made as eutectic modifiers that can spheroidize the otherwise coarse acicular Si particles, effectively improving the mechanical performance of the Al–Si alloys1,32,33,34,35. However, these effects can diminish in the case of Na at elevated temperatures, or result in an increased amount of gas pickup and porosity in the case of Sr and Sb36,37,38. Sb also poses environmental safety concerns30 because of its health hazard and ability to cause severe poisoning. As solid-phase processing (SPP) techniques can result in refined microstructures and do not require the melting of the alloy, they can also be much less energy intensive than more conventional approaches. For example, the FSP of an Al–Si–Mg alloy was effective in uniformly distributing the acicular Si phase and completely healing casting porosities5,36. Nelaturu et al. demonstrated an improvement in fatigue life of an Al-7Si-0.3Mg (A356 alloy) by one order of magnitude after FSP. While the advantages of SPP approaches to Al–Si-based alloys have been demonstrated in past5,36,39, a clear understanding of transformation pathways and mass redistribution during SPP is lacking. Therefore, in this work, a detailed analysis of the shear-deformation-induced microstructural refinement and mechanical property changes in a model as-cast Al-4 at.% Si alloy is conducted using an experimental and computational study.

Results and discussion

The as-cast and homogenized Al–4Si alloy consists of a dendritic microstructure with a hypoeutectic Al matrix and a eutectic Al–Si inter-dendritic region (Fig. 1a–c and Supplementary Figs. 14). The grain size of the as-cast Al–4Si alloy ranges from 300 to 600 µm. The APT proximity histogram between the eutectic Al and Si phases (Fig. 1d) revealed the average composition of Si dissolved in Al to be 1.26 ± 0.01 at.%, while there was no Al dissolved in the eutectic Si phase. The crystal structure was verified using TEM-selected area electron diffraction, as shown in Supplementary Fig. 2a–c; the Al rich phase had a face-centered cubic and the Si rich second phase had diamond cubic crystal structure. The APT analysis identified 0.10 ± 0.02 at.% Si in the hypoeutectic Al matrix measured from the center of the dendrite (see Supplementary Fig. 2d, e). Thus, based on these observations, Si concentration ranged from 0.10 at.% in the hypoeutectic Al to 1.26 at.% in the Al adjacent to the eutectic Si phase. The TEM results also highlighted coherent Al-rich nanoscale precipitates (5–10 nm in diameter) within the eutectic Si phase (Fig. 1e–g). These Al nanoprecipitates were accompanied by a void at their periphery (Fig. 1f, g). Refer to the Supplementary videos 1 and 2 and Supplementary Figs. 3 and 4 showing the distribution of nanoprecipitates and the voids. Barrirero et al. observed similar nanoprecipitates in Al–7 wt% Si (i.e., ~6.8 at.% Si) using TEM and APT and suggested that these precipitates can have variable compositions ranging from 10 to 50 at.% Al 33,40.

Fig. 1: Microstructure of as-cast Al–4Si alloy.
figure 1

a Low-magnification SEM-BSE image of the as-cast Al–4Si alloy, with dendritic grains (dark) separated by inter-dendritic regions (bright) consisting of eutectic Al–Si phases. b Higher-magnification SEM-BSE image of the eutectic region. c SEM image of a FIB-prepared APT needle sample from the eutectic region showing stripes corresponding to Si precipitates in the needle. d An APT reconstruction of the eutectic region with the Si (red) and Al (blue) distributions (inset) and compositional profiles across the Al and Si phases. e Bright-field TEM (BFTEM) image showing the acicular Si phase in the eutectic region. STEM-EDS Al map at the same magnification (inset) reveals the Al-rich nanoprecipitates inside the eutectic Si phase (white arrow). f Nanoscale voids adjacent to the Al-rich nanoprecipitates inside the eutectic Si phase as imaged using Fresnel contrast imaging. g An Al-rich particle inside the eutectic Si phase in a STEM (high angle annular dark field) HAADF image. Arrows point to an Al-rich nanoprecipitate (red) and nanovoid (yellow). The inset (top left) is the FFT of the image with Si phase aligned on [112] direction while on right is the FFT from the Al-rich nanoprecipitate.

After characterizing the as-cast and homogenized alloy microstructure, we subjected the metallographically polished surface of the Al–4Si alloy to reciprocating shear deformation using a pin-on-disk tribometer (see schematic in Fig. 2a). The details of the tribometer test conditions are provided in “Methods” and Supplementary Fig. 5. After 5000 cycles of reciprocating shear deformation, the sample was cross-sectioned (see schematic in Fig. 2b), and the microstructure in the deformed region was analyzed using SEM-BSE imaging (Fig. 2c). The microstructure in the 30 µm thick shear-deformed region was substantially refined (Fig. 2c), and SEM Energy Dispersive Spectroscopy (EDS) mapping revealed fragmentation of Si precipitates in the shear-deformed region (Fig. 3d, low-magnification maps in Supplementary Fig. 6). The average Vickers hardness in the preprocessed region was measured to be 34 ± 05 Hv (28 Hv in center of dendrite and 37 Hv in the inter-dendritic eutectic region), while a hardness of 89 ± 11 Hv was measured in the post-deformation region (Supplementary Fig. 7). A nano indention hardness mapping was also performed which does not show conclusive trends as the alloy consists of a dual phase microstructure with very different hardness values of each phase and therefore the scatter in the hardness in the base material is very high, which makes it difficult to observe the trend in the hardness change due to deformation. A very high scatter was also observed by Stoyanov et al.41. Stoyanov et al. observed the change in nanomechanical properties of the shear deformed layer by tribometer as a function of contact pressure and observed that the lower pressure (15 MPa) showed a much higher degree of mechanical mixing and grain refinement compared to the higher pressure of 35 MPa in Al–Si alloy41.

Fig. 2: Cyclic shear deformation of Al–4Si.
figure 2

a The procedure used to generate reciprocal shear deformation using a pin-on-disk tribometer. b Schematic of a shear-deformed microstructure cross-section, the box highlights the location of microstructural and mechanical property assessment. c SEM-BSE image with the dashed curve highlighting the boundary between the 30 μm thick shear-deformed region and the undeformed microstructure below. The dashed rectangle highlights the region shown in d. d EDS map of the Si distribution in the shear-deformed and adjacent undeformed regions. A dotted white line is used to delineate the sheared and unsheared region in both c and d. e Micropillar compression stress–strain curves for the shear-deformed region and undeformed eutectic Al–Si and Al matrix. Inset: secondary electron SEM images of deformed pillars of all three regions.

Fig. 3: Characterization of shear deformed Al–4Si alloy.
figure 3

Al–4–Si microstructure in the sheared region. a SEM-BSE image of the fragmented eutectic Si phase (bright contrast phase). b, c APT reconstruction of the interface, as well as the compositional change across it, between the fragmented eutectic Si phase and the Al matrix. d STEM-HAADF image of the Al matrix. e Si STEM-EDS map of the same region as in d showing the distribution of nanoscale, Si-rich phases at Al grain boundaries. f APT reconstruction of a nanoscale, Si-rich phase within the ultrafine-grained Al matrix. g The compositional change across the Al matrix and nanoscale Si-rich phase. h APT reconstruction of Si distribution in intragranular regions of the nanograined Al matrix.

Micropillar testing under compression, shown in Fig. 2e, indicates that the 4% offset flow strength of the shear-deformed regions (0.33 ± 0.12 GPa) is two times higher than that of the as-cast Al matrix (0.14 ± 0.02 GPa) or the eutectic Al–Si regions (0.18 ± 0.01 GPa). An increase of 0.19 GPa in 4% offset flow strength is observed in our experiments could be due to a combination of Hall–Petch strengthening, matrix supersaturation, and second phase strengthening by nanoprecipitation, and fragmentation of Si phase. The grain boundary strength of the Al matrix (~100 nm), calculated using the Hall–Petch relation with Ko for pure Al to be ~0.002 GPa. mm1/2 and σo of ~0.01 GPa42 is ~0.213 GPa. Hence, the extra 0.17 GPa observed experimentally is likely due to matrix supersaturation and/or the refinement of the Si phase.

High-resolution characterization of the shear-deformed microstructure revealed three levels of microstructural hierarchy in terms of Si distribution. The first level of microstructural hierarchy is formed by the 500 nm to 1 μm fragmented eutectic Si phase, highlighted using white arrows in the high-magnification BSE image in Fig. 3a. The average composition of this phase was found to be 95.5 ± 1.5 at.% Si and 4.5% ± 0.5 at.% Al, as quantified using proximity histogram analysis across a 15 at.% Si iso-composition surface in the APT reconstruction (see Fig. 3b, c). Between these fragmented eutectic Si phases, ultrafine Al grains with an average grain size of 90 ± 35 nm were revealed using STEM imaging (Fig. 3d). As illustrated by the STEM-EDS map (Fig. 3e), these Al grains have a nanoscale, Si-rich phase at grain boundaries, that constitutes the second level of microstructural hierarchy. A proximity histogram across a 15 at.% Si iso-composition surface of APT-reconstructed regions (Fig. 3f, g) highlights that the composition of these Si-rich regions is 74.82 ± 1.8 at.% Si and 25.18 ± 0.7 at.% Al. In addition, the average composition of the grain interiors of the ultrafine-grained Al matrix in the shear-deformed region was measured to be 0.31 ± 0.01 at.% Si using APT. Si was distributed heterogeneously, forming nanoscale Si-rich clusters (Fig. 3h). These clusters constitute the third level of microstructural hierarchy. Shear-deformation-induced Si clustering in an Al matrix can modify the local stacking-fault energy and therefore potentially lead to the formation of stacking faults39,43,44,45. To verify this hypothesis, additional TEM analysis was conducted to analyze defect structures in the shear-deformed region.

The BFTEM image of the shear-deformed region shows the ultrafine-grained Al matrix and a fragmented eutectic Si phase (Fig. 4a). A high density of vacancy Frank loops was observed in the Al matrix (Fig. 4b). Similar Frank loops, 2–20 nm in size, were also observed by Wu et al. in pure Al after equal channel annular processing46. These Frank loops had stacking faults on the (111) planes, as highlighted in Fig. 4c. The phase contrast image in Fig. 4d shows the clear discontinuity in the (111) planes of Al, and the fast Fourier transformation inset of the phase contrast image depicts the [110] zone axis of the Al grain. In addition, we noticed a high density of microtwins in the fragmented eutectic Si phase along the {110} direction (Fig. 4e). The phase contrast image of the fragmented eutectic Si phase given in Fig. 4f also revealed a stacking fault that sheared an Al-rich nanoprecipitate, further confirming that these faults were formed during shear deformation. These stacking faults originated from the dissociation of 1/2〈110〉 type edge dislocations into two 1/6〈112〉 Shockley partial dislocations and are commonly observed in low-stacking-fault energy materials47. We note that the nanovoids associated with Al-rich nanoprecipitates in the eutectic Si phase were completely healed after shear deformation. Healing of nanovoids can be related to friction nano-welding, stress relaxation after processing, or the local heating due to friction. However, the degree to which this effect is observed can vary with the location of examination; hence, further investigation is required to fully elucidate the mechanisms behind this phenomenon.

Fig. 4: Defect structures in sheared region.
figure 4

a BFTEM image of Al nanograins and a fragmented eutectic Si phase (marked). b BFTEM image showing the dislocation loops in the Al grains. c Stacking faults in Al grains within the dislocation loops. d phase contrast image showing the discontinuity, or the fault, in Al stacking. The inset is the FFT of the image depicting the [011] zone axis of Al. e High-magnification STEM-HAADF image of the deformation twins in the fragmented eutectic Si. f STEM-HAADF image of the atomic columns of the fragmented eutectic Si depicting nanotwin shearing across the Al nanoprecipitate.

Microstructure-based finite element modeling was conducted to further simulate the stress and plastic strain accumulation at an early stage of cyclic deformation that can drive the type of microstructural evolution observed in the Al–4Si alloys examined in this study. An SEM-BSE image of a 200 × 200 μm representative area of the Al–4Si alloy (Fig. 5a) was discretized into finite elements using the open-source code OOF248, as shown in Fig. 5b. The two-phase representative volume element (RVE) mesh was then imported into the ABAQUS standard commercial finite element code. Material properties were assigned to each phase: for Al, Young’s modulus E = 70 GPa, Poisson’s ratio v = 0.33, yield stress σy = 117 MPa, ultimate stress σu = 200 MPa, and ultimate strain εu = 0.28; for Si, E = 130 GPa, v = 0.27, σy = 200 MPa, σu = 300 MPa, and εu = 0.05. Plane strain elements (CPE3) were used in the model.

Fig. 5: Microstructure-based finite element modeling.
figure 5

a 200 × 200 μm2 SEM-BSE image of an Al–4Si alloy and its representation as a computational representative volume element; b Si (red) and Al (green) occupying 18% and 82% of the mesh area, respectively; c the simulated loading cycle illustrates reciprocating shear deformation; d von Mises stress contours; and e accumulated equivalent plastic strain distribution from the 2-D, microstructure-based finite element shear model.

To simulate reciprocal shear deformation on the RVE domain, three loading cycles were considered using the loading pattern given in Fig. 5c. All the nodes on the RVE bottom boundary (y = 0) were fixed in both the x and y directions, while cyclic shear displacements were applied in the x direction at all the nodes along the top of the RVE domain. In addition, periodic boundary conditions were applied to the RVE domain, in which the nodes on the left and right boundaries were required to have the same displacements in the x direction. Figure 5d shows the von Mises stress distribution on the RVE after three loading cycles. Higher stress is predicted in the Si precipitates and along the grain boundaries in comparison to the Al matrix. The continued accumulation of von Mises stress can lead to eventual fragmentation of Si precipitates. The model predicted equivalent plastic strain contours after three loading cycles, which are presented in Fig. 5e. This plastic strain accumulation in the microstructure presumably drives dynamic recrystallization and the substantial microstructural refinement that improves the mechanical properties of the Al–4Si alloy.

In summary, we have demonstrated that shear deformation can be effective at modifying the microstructural hierarchy and defect structures in a binary Al–4Si alloy. A high degree of grain refinement in the Al matrix, from an initial grain size of ~500 μm to a ~100 nm final grain size, resulted from the reciprocating shear deformation. The eutectic Si phase was fragmented and redistributed in the shear-deformed microstructure in the form of (1) a 0.5–1 µm fragmented, deformed eutectic Si phase with ~95 at.% Si; (2) a nanoscale Si-rich phase along the Al grain boundaries with ~75 at.% Si; and (3) an heterogenous nanoscale clustering of Si within the Al matrix. The average dissolved Si concentration in the ultrafine Al matrix grains after deformation was 0.30 at.%. Both the Al matrix and Si phases displayed increased intermixing as a result of shear deformation. Dislocation Frank loops and stacking faults were observed in the Al matrix after deformation, and an abundance of nanotwins and stacking faults were found in the fragmented Si phase. Additionally, we demonstrated that the multiscale, microstructural hierarchy induced by shear deformation can further strengthen the alloy, in combination with the Hall–Petch strengthening by grain refinement. Micropillar compression tests indicated a twofold increase in the flow stresses due to shear-induced microstructural changes in of the alloy. A more detailed analysis of the deformation mechanisms activated in the highly refined shear-deformed alloy is currently ongoing and will be the subject of a future publication. The microstructural refinement and enhancement in mechanical properties observed in our study are analogous to observations from Al–Si alloys processed using FSP1,3,4,5,6,36,49. These results highlight the promise of using controlled shear deformation during SPP as a versatile approach to engineer hierarchical microstructures in alloys to achieve superior mechanical properties.

Methods

The binary hypoeutectic Al-4 at.% Si alloy (to be noted as Al–4Si through the remainder of the paper) was cast and remelted five times for homogenization. Samples with dimensions of 20 × 10 × 3 mm were cut using a slow speed saw, mechanically ground to 2000 grit, and metallographically polished using a 0.02 µm colloidal silica solution. The microstructure of the as-cast alloy was characterized using scanning electron microscopy (SEM) backscattered electron imaging. The as-cast alloys were subjected to shear deformation in dry air at room temperature using an Anton Paar pin-on-disk tribometer. A 6 mm diameter stainless-steel sphere with a load of 1 N and linear speed of 200 mm/s was used in reciprocating mode for 5000 cycles, with a stroke length of 14 mm. The microstructures of the undeformed and shear-deformed regions were characterized using SEM, transmission and scanning transmission electron microscopy (TEM and STEM), and atom probe tomography (APT). The SEM imaging and sample preparation for S/TEM and APT (using the focused-ion beam (FIB) lift-out technique) was performed using a Thermo Fisher Scientific Quanta 200 FIB-SEM outfitted with an Oxford Instruments X-ray EDS system for compositional analysis. An FEI Titan 80–300 operated at 300 kV and an aberration, probe corrected, JEOL ARM200F operated at 200 kV were used for S/TEM. A CAMECA LEAP 4000X HR APT was used in pulsed voltage mode at a 200 KHz pulse frequency with 20% pulse fraction, a specimen temperature of 50–60 K, and a detection rate that was maintained at 0.005 atoms/pulse.

To compare the local deformation behavior of the as-cast to the sheared microstructures, micropillars were fabricated from both regions using FIB micromachining (FEI Scios) and tested under compression (using an Alemnis SEM indenter system. A FIB workstation (Tescan Vela) was used for milling the micropillars using a three-step process. First, stepped circles with an outer radius of 20 μm and inner radius of 5 μm were milled using an accelerating voltage of 30 kV and a current of 6 nA, stopping when the crater floor first reached the interface between the film and the substrate. Second, an annulus was milled at 30 kV and 1.6 nA to reduce the pillar diameter and sidewall taper. Finally, a polishing circle was used under the same milling conditions as in the previous step to achieve the final pillar diameter. The diameter was selected to be 40% of the film thickness, resulting in pillars with a nominal aspect ratio of 2.5.

Detailed microstructural analysis and the mechanical property assessment were conducted on the region directly below the center of the wear track (highlighted in Fig. 2b). A 2-D, microstructure-based finite element reciprocating shear deformation model50,51 was also employed to analyze stress and plastic strain accumulation in the microstructure and to simulate the von Mises stress distribution and equivalent plastic strain after three cycles of reciprocal shear deformation.