Grain boundary (GB) engineering is an effective tool for improving the mechanical properties of metallic materials. In this study, we added 30 appm boron (B) to an equiatomic TiNbZrHfTa refractory high-entropy alloy (HEA), and found that the GB chemistry has been altered. B preferentially segregates to GBs with the co-segregation of Zr and depletion of Nb and Ta, as revealed by atom probe tomography probing. This change in GB chemistry affects the macroscopic load-bearing performance of the HEA, where a reduction in the yield strength by ∼6.2 ± 0.7% is observed, i.e. , 869.8 ± 10.4 MPa of the B-doped HEA vs. 920.7 ± 3.7 MPa of the B-free HEA. Additionally, a correlation between GB chemistry and the plastic deformation at GBs is found, indicated by changes in the crystallographic slip traces and the Luster-Morris compatibility factor among adjacent crystals. More specifically, we find that (i) B-decorated GBs accommodate higher adjacent plastic strain levels, which is manifested as an out-of-plane offset (via GB shear localization) between two neighboring grains; and (ii) in the B-doped case, slip transfer across GBs requires a lower misalignment between slip planes or slip directions as compared with the B-free sample. This study thus enhances our understanding of the role of GB chemistry in the mechanical properties of polycrystalline body-centered cubic HEAs, leveraging opportunities for GB segregation engineering in this field.

This work aims to predict the microstructure of recrystallized medium and high-entropy alloys (MEAs and HEAs) with a face-centered cubic structure, in particular the density of annealing twins and their thickness. Eight MEAs and five HEAs from the Cr-Mn-Fe-Co-Ni system are considered, which have been cast, homogenized, cold-worked and recrystallized to obtain different grain sizes. This work thus provides a database that could be used for data mining to take twin boundary engineering for alloy development to the next level. Since the stacking fault energy is known to strongly affect recrystallized microstructures, the latter was determined at 293 K using the weak beam dark-field technique and compared with ab initio simulations, which additionally allowed to calculate its temperature dependence. Finally, we show that all these data can be rationalized based on theories and empirical relationships that were proposed for pure metals and binary Cu-based alloys.

Physical properties of ten single-phase FCC Cr_xMn₂₀Fe₂₀Co₂₀Ni_40-x high-entropy alloys (HEAs) were investigated for 0 ≤ x ≤ 26 at%. The lattice parameters of these alloys were nearly independent of composition while solidus temperatures increased linearly by ∼30 K as x increased from 0 to 26 at.%. For x ≥ 10 at.%, the alloys are not ferromagnetic between 100 and 673 K and the temperature dependencies of their coefficients of thermal expansion and elastic moduli are independent of composition. Magnetic transitions and associated magnetostriction were detected below ∼200 K and ∼440 K in Cr₅Mn₂₀Fe₂₀Co₂₀Ni₃₅ and Mn₂₀Fe₂₀Co₂₀Ni₄₀, respectively. These composition and temperature dependencies could be qualitatively reproduced by ab initio simulations that took into account a ferrimagnetic ↔ paramagnetic transition. Transmission electron microscopy revealed that plastic deformation occurs initially by the glide of perfect dislocations dissociated into Shockley partials on {111} planes. From their separations, the stacking fault energy (SFE) was determined, which decreases linearly from 69 to 23 mJ·m⁻² as x increases from 14 to 26 at.%. Ab initio simulations were performed to calculate stable and unstable SFEs and estimate the partial separation distances using the Peierls-Nabarro model. While the compositional trends were reasonably well reproduced, the calculated intrinsic SFEs were systematically lower than the experimental ones. Our ab initio simulations show that, individually, atomic relaxations, finite temperatures, and magnetism strongly increase the intrinsic SFE. If these factors can be simultaneously included in future computations, calculated SFEs will likely better match experimentally determined SFEs.