Crystal–Glass High-Entropy Nanocomposites with Near  Theoretical Compressive Strength and Large Deformability

 

The long-standing trade-off between strength and ductility in structural metals remains one of the central challenges in alloy design. High-entropy alloys (HEAs) based on multiple principal elements offer broad compositional freedom and generally good ductility through dislocation slip, twinning, or deformation-induced phase transformation, but their shear strength remains well below G/10 — the theoretical limit. Metallic glasses (MGs), by contrast, achieve shear strengths approaching G/37, yet fail catastrophically through shear band localization at ambient temperature. We sought to transcend both limitations simultaneously by merging the two material classes at the nanoscale into a single crystal-glass high-entropy nanocomposite architecture.

The key physical insight underpinning our design is twofold. First, the stacking fault energy (SFE) of the crystalline HEA phase governs twin nucleation and density; keeping the SFE low — as in CrFeCoNi-based systems (~27 mJ m⁻²) — promotes the formation of high-density nanotwins that act as strong barriers to dislocation motion while simultaneously providing additional shear carriers and kinematic degrees of freedom for plastic flow. Second, MGs become intrinsically capable of homogeneous plastic flow when their characteristic size is reduced below ~100 nm, a regime in which shear bands remain confined and cannot percolate. By combining both effects in one material, we aimed at a nanocomposite in which the crystalline and amorphous phases deform compatibly rather than sequentially or catastrophically.

We produced the nanocomposite by magnetron co-sputtering of a CrCoNi target and a Fe₇₈Si₉B₁₃ metallic glass target, both of 99.9 at% purity. By tuning the substrate-to-target distance ratio, we accessed the compositional window 23.1 Cr – 20.7 Fe – 24.8 Co – 26.7 Ni – 1.7 B – 0.3 C – 0.6 O – 2.1 Si (at%), as confirmed by atom probe tomography (APT). The rationale for adding Fe-Si-B into the CrCoNi base is grounded in glass-forming-ability (GFA) theory: Cr content exceeding ~20 at% in Fe-Co-Ni mixtures forces near-eutectic conditions in the binary sub-systems (Fe-Cr, Co-Cr, Ni-Cr), which strongly suppresses crystallization; B, C, O, and Si further enhance the GFA through their directional bonding character.

The resulting microstructure consists of nanocolumnar fcc grains approximately 8 nm in width, with weak crystallographic texture as confirmed by selected-area electron diffraction. Between the grains, an amorphous phase of average thickness ~1 nm forms preferentially at triple junctions and along a subset of grain boundaries — a topology consistent with faster elemental diffusion along triple lines. APT and annular bright-field STEM (ABF-STEM) analysis confirms that this amorphous phase is enriched in Cr (Cr iso-concentration surface at 12 at nm⁻³), while the crystalline grains are enriched in Fe. B, C, O, and Si distribute essentially uniformly across both phases. Critically, the average amorphous region size is well below the ~100 nm threshold below which MGs exhibit homogeneous plastic flow.

Within the crystalline grains, low-angle annular dark-field (LAADF)-STEM reveals an extraordinarily high density of nanotwins and stacking faults (SFs) on {111} habit planes, with a Kurdjumov–Sachs orientation relationship ([1-10]twin ∥ [-110]γ) relative to the adjacent matrix. Because the columnar grains carry only weak texture, twinning operates on all available {111}<112> systems. The average twin boundary density is 1.0 × 10⁹ m⁻¹, and twin/SF/matrix lamellae are sub-2 nm thick, subdividing the already ~8 nm grains into an exceptionally fine lamellar sub-structure. Higher-order fivefold nanotwins are also observed. This density of twin boundaries, far exceeding what is typically achievable in bulk processing, is made possible by the extreme quench rate (~10¹⁰ K s⁻¹) inherent to sputter deposition and by the low SFE of the CrFeCoNi system.

 

Mechanical behavior was assessed by micro-pillar compression at pillar diameters of 380, 400, 500, and 1000 nm, all giving the same yield strength — confirming the absence of a mechanical size effect in this regime, as the smallest pillar diameter is still more than 40× the grain width. For comparison, we tested single-crystalline CrCoNi, nanocrystalline CrCoNi (average grain width ~20 nm), and nanocrystalline CrCoNi-Fe-Si-B without an amorphous phase (~15 nm grain width) under identical conditions.

The crystal-glass high-entropy nanocomposite achieves a compressive yield strength of 4.1 GPa (0.2% offset criterion), compared to 3.8 GPa for nanocrystalline CrCoNi-Fe-Si-B, 3.3 GPa for nanocrystalline CrCoNi, and 0.7 GPa for single-crystalline CrCoNi. The Young's modulus, obtained from nanoindentation, is E = 124 GPa, giving a shear modulus G = E/[2(1+ν)]. The shear strength τ = σy/2 yields a G/τ ratio of 24, approaching the theoretical shear strength limit of G/10. None of the reference alloys reach this ratio.

Equally important, the nanocomposite sustains homogeneous plastic strain above 45% in compression without shear band percolation or macroscopic fracture. The reference alloys, despite also surviving to 50% engineering strain in the pillar geometry, do so through inhomogeneous flow: post-mortem SEM and STEM confirm the formation of discrete shear and slip bands — a compression instability artifact — rather than genuine homogeneous plasticity. The nanocomposite pillar, by contrast, deforms uniformly throughout its cross-section with no localization features.

The origin of the near-theoretical strength lies in a synergistic hierarchy of mechanisms: (i) the thin amorphous interfacial phase raises the stress barrier for dislocation nucleation at grain boundaries; (ii) the extremely fine twin spacing (~2 nm) within ~8 nm grains maximizes twin boundary strengthening, which scales inversely with twin thickness; (iii) the amorphous phase surrounding the grains suppresses the softening normally associated with grain boundary–twin intersection events; and (iv) the overall duplex crystal-glass nanostructure stiffens the composite response from the outset of loading.

 

To elucidate the mechanisms responsible for the exceptional combination of strength and ductility, we performed post-deformation APT and LAADF-STEM on pillars compressed to ~50% engineering strain. Atom probe reconstructions using Cr iso-concentration surfaces (10 at nm⁻³ threshold) reveal three structural zones: undeformed base material with straight columnar grains; a ~80 nm-thick transition region with bent columnar grains; and the heavily deformed pillar interior where columnar grains have been transformed into globular units ~8 nm in diameter. No grain growth is detected — a significant departure from the load-driven grain growth typically observed in nanocrystalline metals.

Within the deformed pillar, the Cr-enriched amorphous phase uniformly wraps the now-globular grains, maintaining topological connectivity and showing no shear-band-like localization. The Cr concentration within the amorphous phase progressively increases from the base material through the transition zone into the fully deformed region, as confirmed by 2D Cr contour maps of 2 nm-thick APT slices. This Cr redistribution reflects dislocation drag: partial dislocations generated in the crystalline grains carry solute Cr toward the crystal-glass interface, where it is incorporated into the amorphous phase as dislocations are absorbed.

The deformation sequence we identify is as follows. Partial dislocations are emitted from crystal-glass interfaces, traverse the ~8 nm nanograins, and interact with pre-existing nanotwins and SFs, driving twin boundary migration via Shockley partial motion — a mechanically induced de-twinning process clearly resolved in LAADF-STEM of the transition region. After 50% deformation, the average twin boundary density decreases from (1.0 ± 0.2) × 10⁹ m⁻¹ to (5 ± 1) × 10⁸ m⁻¹. De-twinning by itself would normally produce strain softening; however, in our nanocomposite, the progressive subdivision of grains and the concomitant increase in crystal-glass interfacial area provide additional barriers to dislocation motion, generating a compensating strain hardening increment that keeps the flow stress approximately constant.

The amorphous phase plays a dual active role. Its nanoscale size — well below the critical dimension for shear band propagation — enables intrinsically homogeneous plastic flow with near-ideal glassy strength, making it simultaneously a strong interface and a compliant buffer that accommodates dislocation flux without forming shear bands. The absence of dislocation pile-ups (dislocations are continuously nucleated, traverse the grain, and annihilate at the opposite interface) prevents stress concentrations that would otherwise trigger localized failure. The mutual kinematic compatibility between viscoplastic flow in the amorphous phase and partial dislocation activity in the crystalline phase is thus the central mechanistic feature enabling co-deformation.

 

We demonstrate that by deliberately engineering compositional and structural conditions — low SFE for nanotwin promotion in the crystalline phase, near-eutectic composition and metalloid doping for GFA in the amorphous phase, and nanoscale confinement of both phases — it is possible to simultaneously approach the theoretical shear strength limit and maintain large homogeneous plastic strainability exceeding 45% in compression. This performance is not accessible in either pure nanocrystalline HEAs or monolithic metallic glasses.

Beyond mechanical properties, the same compositional space produces soft magnetic behavior (exploiting the Fe-Si-B glassy constituent) and good thermal stability, suggesting applicability in mechanically loaded micro-electromechanical systems and flexible devices where multifunctionality is essential. The crystal-glass high-entropy nanocomposite concept thus defines a new design space at the intersection of HEA alloy metallurgy, metallic glass science, and nanoscale materials engineering — one in which the limitations of each constituent class are overcome by the structural and chemical synergies arising from their nanoscale integration.