A CALPHAD–stacking fault energy framework for predictive Design of Martensitic Steels with controlled Ms

This study presents a physics-based computational framework for the predictive design of martensitic steels, addressing the limited extrapolation capability of conventional empirical Ms. equations and driving-force-based models. The approach integrates stacking fault energy (SFE) predictions derived from Olson–Cohen nucleation theory with CALPHAD thermodynamic calculations to estimate the martensite start temperature (Ms). A comprehensive database of 249 experimental steel compositions was evaluated, contrasting the energetic pathways of both FCC → HCP and FCC → BCC transformation mechanisms. Our results demonstrate that the FCC → BCC-based transformation yields superior predictive fidelity across a diverse compositional spectrum, effectively bridging the gap between atomistic faulting processes and macroscopic phase stability. The framework's predictive capacity was validated using novel alumina-forming martensitic (AFM) steels, whose high aluminium content lies significantly beyond the calibration limits of traditional martensitic databases. This work establishes a versatile, mechanism-driven design route for precise Ms. control, enabling the exploration of non-conventional alloy spaces for high-performance energy applications.