Answers are grounded in this paper and cite their sources.
Identifying solid-solution compositions in multicomponent alloys requires exploring a highdimensional design space while accounting for competing phases. We present a differentiable search framework that uses composition gradients of a thermodynamic objective to identify promising alloy regions without exhaustive grid enumeration. In the present implementation, an alchemical machine-learning interatomic potential and ideal configurational entropy provide a differentiable free-energy approximation for random solid solutions. Gradient-based optimization is combined with metadynamics biasing to explore multiple basins, and a tangent-plane criterion screens candidates against precipitation of competing compounds. Applied to the Mg–Al–Sc–Ti–V–Cr–Mn–Fe–Co–Ni system on fcc, bcc, and hcp lattices at 1000 and 1600 K, the search recovers compositionally complex alloy (CCA) families supported by literature. The dominant Cr–Mn–Fe–Co–Ni candidate region expands at 1600 K to include V-containing families and six-principal-element compositions. Screened compositions show substantial departures from equimolarity, highlighting the value of continuous composition optimization. The search also identifies the less-explored Mg–Sc–Ti system, for which preliminary CALPHAD calculations suggest extended high-temperature bcc solid-solution fields under suitable pressure conditions. These results demonstrate a framework for efficient search of CCAs and illustrate how differentiable materials models can serve directly as tools for materials design.
No takes yet. Share an insight, caveat, or question.
Pin-Wen Guan (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: