A data‐physics integrated strategy for accelerated design of high‐temperature Ni–Mn–Ga–Cu–B superelastic alloys

The increasing demand for energy-efficient, high-temperature actuation materials has driven intensive research into advanced shape memory alloys. Ni-Mn-Ga-based systems, in particular, hold exceptional promise for superelastic cooling and actuation applications, but they face a critical hurdle: simultaneously delivering large recoverable strains and minimal thermal hysteresis at elevated temperatures under practical driving stresses. Conventional trial-and-error synthesis and testing workflows are prohibitively slow and resource-intensive to overcome this bottleneck. Here, we report a data-physics integrated strategy for the rapid, multi-objective design of high-temperature superelastic Ni-Mn-Ga-Cu-B shape memory alloys. Firstly, high-temperature Ni-Mn-Ga based shape memory alloys with a phase transformation temperature ranging from 125 to 225 °C were screened out using machine learning methods. SHapley Additive exPlanations has demonstrated that the valence electron number and electronegativity are the primary factors influencing the phase transformation temperature. Further, phase transformation temperature and superelasticity tests were conducted on eight high-temperature Ni-Mn-Ga-Cu-B shape memory alloys. The results indicated that Ni₅₀Mn₂₉Ga₁₆Cu₂B₃ composition exhibits a recoverable strain of 5.1 % at 135 °C under 575 MPa, while Ni₅₀Mn₂₉Ga₁₂Cu₄B₅ achieves 4.7 % strain at 231 °C. Mechanistic insights from first-principles reveal that strategic Cu and B doping stabilizes the martensitic variants and reduces lattice friction. This unified strategy not only accelerates the discovery of high-performance shape memory alloys but also provides a transferable theoretical framework for future materials optimization, paving the way toward energy-efficient, high-temperature actuation technologies.