Exploring the deformation behavior of nanotwinned Al–Zr alloy via in situ compression

Aluminum alloys are a mainstay for lightweight structural applications. However, like most metallic materials, increasing their strength is typically accompanied by sacrificing ductility.^1–3 Nanocrystalline (NC) metals have high strength due to a lack of dislocation pileups in nanograins, but they often have poor ductility manifested by shear banding, as conventional dislocation plasticity mechanisms are difficult to operate in nanograins.^4–7 Softening is often observed in nanocrystalline metals below critical grain sizes due to the activation of grain boundary (GB) sliding and GB diffusion based deformation mechanisms.^8,9 A significant amount of research aimed at mitigating strength-ductility trade-off has led to some profound discoveries and innovations. A classic example is twinning induced plasticity (TWIP) steels, where deformation twins offer simultaneous strength and plasticity.^1,10–13 High entropy alloys (HEAs) are a newer family of metallic materials that have recently incorporated TWIP effects to avoid sacrificing ductility.^14,15 Another avenue for simultaneous enhancement of strength and deformability is through gradient microstructures.^16–20 The combination of fine grains and coarse grains in these structurally graded metals helps to overcome the strength-ductility trade-off.

The influences of twin boundaries on strength and ductility have been extensively investigated in face-centered-cubic (FCC) metals with low stacking fault energy (SFE), predominately Cu (45 mJ m⁻²)^21,22 and Ag (16 mJ m⁻²).^23,24 When deposited using electrodeposition or magnetron sputtering, these metals form abundant Σ3 (111) coherent twin boundaries (CTBs) with nanoscale twin spacing, and CTBs strongly inhibit dislocation transmission,^25–33 as demonstrated using in situ nanoindentation experiments and molecular dynamics simulations.^34–37 Σ3 {112} incoherent twin boundaries (ITBs) and Shockley partial dislocations are also important for maintaining ductility and high strengths.^32,35,38,39 Nanotwinned (NT) metals also possess a high strength-to-resistivity ratio^21,33,40 and have remarkable thermal stability^41–44 compared with their nanocrystalline counterparts as CTBs store much lower energy than high-angle GBs, and electron scattering coefficient at CTBs is one order of magnitude lower than high-angle GBs.⁴⁰ TBs are difficult to form in FCC metals with a high SFE due to the high ratio of the stable SFE to unstable SFE $(γsfγusf)$⁠, which prevents the formation of stacking faults by limiting the distance between leading and trailing partials.^45,46 Thus, twins in high SFE metals can rarely form except under extreme conditions, such as severe plastic deformation or at the high stress concentration of crack tips.^46–48 Recent discoveries in high SFE metal films, such as Ni (120 mJ m⁻²) and Al (160 mJ m⁻²), show the formation of a high density of growth twins.^19,49–58 Xue et al. demonstrated a texture dependence on the formation mechanisms of growth twins in Al^55,59 and Bufford et al. identified a template method for replicating TB structures from a low SFE coherent seed layer into high SFE Al films.^53,60 More recently, alloying sputtered Al films with various transition metal solutes has demonstrated the capability of forming NT microstructures in Al composed of abundant ITBs.⁵⁰ These Al alloys possess extremely high strengths (up to ∼2 GPa compressive flow stress^61,62) while maintaining impressive plasticity due to the presence of a high density of partial dislocations making up the periodic 9R phase.^63,64

This work investigates the deformability of NT Al–Zr alloy films with the aim of demonstrating simultaneous improvements to strength and deformability. In situ microcompression tests explore this phenomenon, with a multifaceted look into the role of Zr solute on the microstructure of these alloys using density functional theory calculations and microscopy. This study underscores the role of TBs in the strength and work hardening ability of Al alloys.

Magnetron sputtering was used to deposit 2 μm thick Al–xZr (x = 0–10 at. %) films with a 40 nm Ag seed layer onto HF etched Si (111) substrates. Al (99.999%), Ag (99.999), and Zr (99.995%) targets were used and the chamber base pressure was below 4 × 10⁻⁸ torr before deposition. Out-of-plane θ-2θ x-ray diffraction (XRD) scans were completed using a Panalytical Empyrean X'pert PRO MRD diffractometer operated at 40 kV using Cu Kα₁ x rays to explore the texture and structure of the films. Transmission electron microscopy (TEM) samples were prepared by mechanical grinding, dimpling, and low-energy ion milling, with care taken to avoid sample heating. The microstructure and composition were examined using an FEI Talos 200X analytical microscope operated at 200 kV with a Fischione high-angle annular dark field (HAADF) detector and a super X energy-dispersive x-ray spectroscopy (EDS) detector. Crystallographic orientation mapping and GB misorientation measurements were made using Nanomegas ASTAR^TM with a camera length of 205 mm, a precession angle of 0.6°, and a step size of 5 nm in the same TEM. The hardness and moduli of these films were measured using a Hysitron TI Premiere nanoindenter under displacement control. 75 indents at different depths were performed on each sample to guarantee data reliability, and the indentation depth was less than 15% of the film thickness to exclude any substrate effects. Micropillar compression was performed to probe the deformability and compressive strength of these films. Pillars and TEM samples of deformed pillars were fabricated using an equipped focused ion beam (FIB) inside an FEI Quanta 3D scanning electron microscope. The pillar height and diameter were ∼800 and 1600 nm, respectively, with the taper angle minimized to below 2° during pillar fabrication. Pillars were compressed using a Hysitron PI88 nanoindenter to ∼20% strain using displacement control, with a strain rate at 5 × 10⁻³. Pillars were compressed with numerous partial unloading segments to assess the validity of the alignment during compression.

The XRD patterns in Fig. 1(a) reveal the strong (111) texture. The lack of additional peaks also confirms the formation of a super-saturated solid solution across all compositions. The bright field (BF)-TEM image in Fig. 1(b) shows that Al–4.3Zr is composed of low-angle grain boundaries (LAGBs) and the corresponding selected area diffraction (SAD) pattern demonstrating “single-crystal-like” texture. In comparison, the inserted SAD pattern of Al–10Zr in Fig. 1(c) reveals a diffraction ring indicative of high-angle grain boundaries (HAGBs). The addition of Zr reduces the average grain size in Al from 98 nm (with primarily low-angle GBs) to 50 nm with dominant high-angle GBs. Figure 1(e) shows the lattice parameter measured by x-ray diffraction follows Vegard's law with Zr composition, which further confirms the formation of a solid solution.⁶⁵ The inverse pole figure (IPF) maps collected from ASTAR in Fig. 2 show the texture progression from single-crystal-like {111} texture in for Al–4.3Zr to randomly oriented nanograins in Al–10Zr. The results from these IPF maps correspond well with the SAD evolution and clearly show the strong grain refinement induced by Zr solutes.

Cross-section TEM analysis was also conducted along the 〈110〉 zone axis to study the microstructure in further detail. Comparing BF-TEM images for the Al–4.3Zr [Fig. 3(a)] and Al–10Zr [Fig. 3(d)] samples, it is obvious that the microstructures are both composed of an abundance of vertically oriented columnar boundaries. The inserted SAD patterns in Figs. 3(a) and 3(d) confirm that these samples are highly twinned. Additionally, the dark field (DF) TEM images (Fig. S2 in the supplementary material) taken with $g¯=[2¯00]$ also show the twin relationship across a columnar ITB, with the matrix (M) and twin (T) variants labeled. The high-resolution TEM (HRTEM) image of Al–4.3Zr in Fig. 3(c) shows a sharp ITB taken from the location outlined by the box in Fig. 3(a). The HRTEM image in Fig. 3(f) taken from the box location in Fig. 3(d) of the Al–10Zr sample shows the formation of diffuse ITBs or 9R phase. This 9R phase stabilization leads to a larger volume fraction of 9R phase at higher Zr content, which directly impacts the mechanical properties, to be discussed later.

Figures 4(a)–4(c) contain in situ SEM snapshots from micropillar compression tests conducted on pure Al, Al–4.3Zr, and Al–10Zr (see Videos S1–S3 in the supplementary material for more details). Looking at the real-time morphological evolution of the pillars provides insight into the deformability of these alloys under compression. The pure Al [Fig. 4(a)] micropillar displays evident shear band formation. In comparison, the Al–4.3Zr micropillar exhibits pillar barreling indicative of significant plasticity. The Al–10Zr pillar is also highly deformable; however, the morphology transitions to limited dilation of the pillar top. The true stress–strain curves in Fig. 4(d) show Al reaches a maximum flow stress of ∼150 MPa with a highly serrated flow curve, commonly seen due to shear banding. The NT Al–4.3Zr and Al–10Zr samples reach significantly higher flow stress, ∼600 MPa and 1.1 GPa, respectively. Interestingly, they maintain significant compressive deformability at these high strengths. Additionally, the work hardening response is noticeably improved from Al–4.3Zr to Al–10Zr, highlighted in the work hardening rate plot in Fig. 4(e), which will be discussed in greater detail later.

Figure 5 contains the post-mortem TEM analysis of the deformed Al–4.3Zr micropillar. Figure 5(a) shows a BF-TEM image overview of the deformed pillar taken along the [011] zone axis. Figure 5(b) shows a DF TEM image taken along the $g→=13[1¯1¯1]$⁠, revealing the extent of detwinning occurring throughout the top half of the pillar. The microstructure remains relatively unchanged at the base of the pillar. Figures 5(c)–5(e) show the SAD pattern evolution from the pillar top [Fig. 5(c)] down to the base [Fig. 5(e)], corresponding to the regions highlighted by the yellow circles in Fig. 5(a). The progression of the diffraction patterns shows the same trend depicted in the dark field, with the pillar top experiencing almost complete detwinning and significant grain rotation. The SAD pattern of the pillar base reflects little grain rotation, and the twin and 9R superlattice spots remain intact. Figures 5(f) and 5(g) contain HRTEM images taken from the regions highlighted using the yellow squares in Fig. 5(a). Figure 5(f) taken from the top region of the deformed pillar confirms the extent of detwinning that takes place as the entire region is composed of single-crystal FCC structure without 9R phase. In comparison, HRTEM image taken from the pillar base in Fig. 5(g) shows pristine 9R phase that remains intact through deformation.

The BF-TEM image for the deformed Al–10Zr pillar in Fig. 6(a) reveals a difference in post-deformed microstructure with higher Zr content. The columnar ITB structure is continuous throughout the pillar, unlike the Al–4.3Zr sample which experienced extensive detwinning. Figure 6(b) shows an EDS map, which exhibits no solute segregation after deformation. Figures 6(c)–6(e) show the SAD pattern evolution moving from the pillar top down to the base. Figure 6(c) shows clear grain rotation near the pillar top. Both the twin and 9R spots are retained in each diffraction pattern throughout the pillar. HRTEM images were also collected at four representative locations highlighted using yellow squares in Fig. 6(a). Figure 6(f) taken at the top of the deformed pillar shows 9R phase that is highly deformed due to interactions with dislocations. Figure 6(g) reveals minor detwinning and discontinuous 9R phase in the middle of the pillar. The pillar base shown in more detail in Figs. 6(h) and 6(i) contains a significant fraction of 9R phase, similar to the as-deposited Al–10Zr microstructure.

To understand the difference in deformation response with evolving Zr content, it is first necessary to extract the difference in microstructure in these alloys. As evident in the grain-size evolution [Fig. 1(d)] and XTEM images (Fig. 3), Zr addition is a significant factor in both grain refinement and alteration of the ITB structure. Density function theory (DFT) calculations were performed to explain the influence of Zr solute on the formation of ITBs and 9R phase during deposition. A schematic in Fig. 7(a) illustrates the atomic configurations used in the DFT calculations for determining the migration energies of trimers and heptamers. Magnetron sputtering inherently possesses an extremely high quenching rate.⁶⁶ This far-from-equilibrium process promotes an extension of solid solubility limit, a significant factor of the microstructural evolution presented in this study.^66,67 As is evident in the XRD presented in Fig. 1, these Al–Zr films present no evidence of intermetallic phase formation and are complete solid solutions even at 10 at. % Zr. Another aspect of this high degree of solubility relates to the defect formation energies. Figure 7(b) contains a plot detailing the defect formation energies corresponding to Zr solute residing at either interstitial (octahedral vs tetrahedral) or substitutional sites. The DFT calculations reveal that Zr prefers to reside at substitutional sites in the Al lattice. The slow diffusing Zr solute⁶⁸ obstructs Al self-diffusion, providing an increased energy barrier for the correction of any stacking faults or twins formed during deposition. Additionally, the migration energies of surface trimer [Fig. 7(c)] and heptamer [Fig. 7(d)] clusters were calculated using DFT. Multiple pathways were calculated for the trimer case, with the red symbols indicating a cluster shifting from a hollow site to residing over an HCP atom and black showing the reverse movements. Each of the conditions calculated in Figs. 7(c) and 7(d) results in an increase in migration energy for these surface clusters, with the heptamer migration energy increasing by 0.19 eV. These calculations suggest that Zr increases the energy barrier for defaulting during film growth. In spite of the high SFE inherent to Al and its alloys, the DFT presented here provides a compelling explanation for how Zr additions aid in stabilizing the unique NT microstructure synthesized in this study.

The high flow strengths achieved in these Al–Zr alloys can be attributed to the microstructural changes induced by the addition of Zr solute. The microstructural changes of the as-deposited films mentioned here are discussed at length in a preceding study exploring the change in nanoindentation hardness with evolving Zr content.⁶⁹ The two mechanisms contributing to the high strengths of these alloys are solid-solution strengthening and Hall–Petch strengthening. The hardening increment induced by solid solution strengthening based on the Fleischer equation^70,71 is relatively minor (28 MPa), and so the majority of the strengthening is derived from the ITB structure.⁶⁹ The boundary strengthening provided by ITBs is evident in the Hall–Petch plot presented in Fig. 8(c), with a clear rise in Hall–Petch slope (k) for NT Al alloys compared to the NC Al samples from the literature.^72–76 Molecular dynamics simulations³¹ and in situ nanoindentation experiments in the TEM⁶⁴ have both been used to demonstrate that TBs are strong barriers to dislocation propagation. Our previous work⁶⁹ also underscores the importance of diffuse ITBs (9R phase) on the improved strength in an analogous manner to thick GBs documented in NC Ni-Mo samples.⁷⁷ The flow stress recorded during the microcompression experiments was also used to compare these NT Al–Zr alloys to other classes of metallic alloys in the specific strength–specific modulus plot in Fig. 8(d). The unique columnar ITB microstructure combined with the low density of Al leads to these NT Al–Zr alloys far outperforming other Al alloys and other classes of materials. The study conducted by Li et al. on the anisotropic mechanical behavior of NT Al–Fe alloys underscores some potential issues of limited plasticity in tension.⁶² However, the performance documented here highlights the great potential of NT Al alloys in applications such as coatings under compression loading conditions.

Zr solute plays a further role in altering the plasticity exhibited by these alloys. The extent of these changes is heavily tied into the relationship between Zr solute content and the fraction of 9R phase stabilized in the microstructure as a result. Figure 8(a) shows that the overall phase fraction of 9R phase (f_9R) increases to 60% with 10 at. % Zr. Figure 8(b) illustrates the microstructural characteristics of NT Al–Zr alloys with ITBs and 9R phase. The values shown in Fig. 8(a) were determined based on extensive HRTEM analyses that measured the average column size (d) and average 9R width (d_9R). The spacing between 9R phase (L), which was used to determine the contribution of 9R to the increasing Hall–Petch slope in the previous study,⁶⁹ was calculated based on

As shown in Fig. 8(a), the width of 9R increases with Zr content until ∼6 at. % Zr, beyond which it plateaus at ∼30 nm. L continues to drop since Zr further refines the column size, reducing the matrix width between each patch of 9R phase. Thus, f_9R increases with the further addition of Zr solutes in the alloys.

The increase in the work hardening rate determined from micropillar compression testing [Fig. 4(e)] can be attributed to several possible sources tied into this microstructure evolution. First, the increase in f_9R leads to a larger percentage of the matrix composed of the complex stacking fault array of diffuse ITBs. As a result, dislocations will have an increasing rate of interaction with defects, leading to a slight jump in work hardening response. These types of dislocation-9R phase interactions were captured using molecular dynamics simulations carried out in NT Al–Fe by Li et al.⁶³ Second, the reduction in grain size from 100 nm (Al–4.3Zr) down to 50 nm (Al–10Zr) has further implications on the work hardening. Decreasing grain size has been shown to play an important role in work hardening. The correlation between work hardening and grain size can be expressed by⁷⁸ 

where ρ is the dislocation density, M is the Taylor factor, b is the magnitude of the Burgers vector, f is a constant, sensitive to the strain rate and temperature, d is the grain size, and k₀ and k₁ are Hall–Petch constants related to the nature of the grain boundaries. Sun et al. utilized a modified Kocks–Mecking model to explain the increase in work hardening seen in NC FeCrAl alloys.^79,80 The increasing work hardening rate in their UFG FeCrAl alloys was attributed to a higher dislocation storage rate at finer grain sizes, which is reflected in Eq. (2). This phenomenon provides another supporting argument for the augmentation of work hardening in the NT Al–10Zr with finer grains.

Third, we examine the evolution of the ITB structures induced by mechanical loading and its correlation to the work hardening. The post-mortem TEM of the deformed Al–4.3Zr pillar [Figs. 5(a) and 5(b)] shows the top half of the pillar experienced extensive detwinning (softening). In contrast, the pillar base maintains the as-deposited microstructure, leading to hardening. The concomitant high strengths and large deformability demonstrated here can, thus, be linked to the simultaneous strain hardening (explained earlier) and strain softening (detwinning) during microcompression. However, Al–4.3Zr and Al–10Zr possess similar as-deposited microstructures yet a noticeable difference in the work hardening response [Fig. 4(e)]. Such a difference lies in the extent of detwinning that occurred in the two alloys. In contrast, as shown in Fig. 6, the twin and 9R phase retain even at the top of the deformed pillar in Al–10Zr, which experiences the largest stress state due to pillar taper inherent to the fabrication process.⁸¹ The resulting reduction in strain softening in the Al–10Zr alloy is because the higher Zr content stabilizes the mobile partial dislocations composing 9R phase and inhibits detwinning, leading to more extensive work hardening. The stabilization of 9R phase by Zr solutes is consistent with DFT calculations showing the increase of migration energy barriers for trimers and heptamers in Al. Despite geometrical constraints (pillar taper) limiting the ability to accurately determine the work hardening exponent, the pillars in this study exhibit similar geometries and a self-comparison on work hardening ability should still be valid.⁸¹ Overall, the extended degree of solid solubility achievable through magnetron sputtering leaves its fingerprint on both the microstructural evolution and the mechanical response of these NT Al alloys. This study highlights the significance of solutes on stabilization of 9R phase and consequent impact on accomplishment of high strength and work hardening ability in NT Al alloys.

Sputtered Al–Zr alloys with abundant ITBs and 9R phase have high strength (1.1 GPa flow stress) and high deformability. The post-mortem TEM analyses of the deformed pillars uncover the importance of the columnar ITBs in the work hardening response of these alloys. DFT calculations explain the role of Zr solutes in formation and stabilization of ITBs in Al. The interplay of ITBs and 9R phase with dislocations induces high strength and significant work hardening ability in Al alloys. This work highlights the mechanical properties of NT Al alloys and underscores an important method for fabricating high strength Al alloy coatings.

See the supplementary material for x-ray pole figure analysis (Fig. S1) to expand the XRD spectra in Fig. 1 and further identify the twinned relationship of the texture in these NT Al–Zr alloys. Also included are EDS maps collected using the Talos 200X showing complete solid solution (Fig. S2) and dark field (DF) TEM images highlighting the matrix/twin orientation relationship across columnar boundaries (Fig. S3). In situ videos collected using the SEM during the micropillar compression experiments (Fig. S4) are also provided (Videos S1–S3) showing live proof of the morphological changes and validity of the tests conducted.

This project is primarily funded by DoE-BES (Basic Energy Sciences) under Grant No. DE-SC0016337. The ASTAR crystal orientation system in TEM microscope is supported by ONR-DURIP under Award No. N00014-17-1-2921. Access to the Microscopy Facilities at Purdue University and Center for Integrated Nanotechnologies (managed by Los Alamos National Laboratory) is also acknowledged. Atomistic simulations were completed utilizing the Holland Computing Center of the University of Nebraska, which receives support from the Nebraska Research Initiative.

The authors have no conflicts to disclose.

N. A. Richter: Data curation (equal); Formal analysis (equal); Methodology (equal); Writing – original draft (equal); Writing – review and editing (equal). M. Gong: Formal analysis (equal); Validation (equal); Writing – review and editing (equal). Y. F. Zhang: Formal analysis (equal); Validation (equal); Writing – review and editing (equal). T. Niu: Data curation (equal); Formal analysis (equal); Writing – review and editing (equal). B. Yang: Data curation (equal); Writing – review and editing (equal). J. Wang: Conceptualization (equal); Formal analysis (equal); Funding acquisition (equal); Project administration (equal); Supervision (equal); Validation (equal); Writing – review and editing (equal). H. Wang: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – review and editing (equal). X. Zhang: Conceptualization (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Supervision (equal); Validation (equal); Writing – review and editing (equal).

The data that support the findings of this study are available within the article and its supplementary material.