Machine learning-guided revelation of Detwinning-dominated cryogenic plasticity in Nano-twinned Cu-Ag alloys

Abstract

High-strength high-conductivity (HSHC) copper alloys are urgently needed for cryogenic applications such as superconducting cables and pulsed magnets operating at liquid-nitrogen temperature and below. By integrating large-scale molecular dynamics simulations containing 472,689 datasets with interpretable machine learning, we reveal that detwinning-induced grain boundary migration rather than conventional dislocation - twin boundary interactions dominate plastic deformation below 150 K in nano-twinned Cu-18 wt% Ag alloys. Cryorolling followed by low-temperature annealing produces dense nanotwins with mean spacing of 5.2 nm, as confirmed by TEM. Atomistic simulations show that Ag segregation dramatically reduces the critical detwinning stress from 800 MPa (pure Cu) to 220 MPa, resulting in a pronounced Hall-Petch plateau below λ = 7.0 nm. SHAP analysis quantitatively demonstrates that detwinning accounts for >72% of plastic strain at T < 150 K. A random forest-derived (R² = 99.4%) temperature–twin spacing phase diagram identifies a cryogenic service window (T < 150 K, λ = 3.0–12.0 nm) where flow stress stabilizes at 5.1 ± 0.2 GPa. These findings provide explicit design guidelines for next-generation HSHC copper alloys in extreme cryogenic environments.