Parameter-Free Prediction of Spinodal Decomposition via Nucleation Modeling and the Disorder Viscosity Correction

Spinodal decomposition can significantly enhance key material properties, including hardness and yield strength, but reliably predicting its onset and characteristic microstructure has remained difficult because traditional models depend on costly experimental inputs or extensive ab initio parameterization of interfacial properties. Here, we present a scalable, parameter-free framework to predict spinodal behavior and resulting length scales by combining two complementary ideas. First, we determine the spinodal temperature by modeling nucleation in a solid solution as it approaches the spinode boundary. Second, we stabilize the locally concave regions of bulk free energies—computed from finite representative cells—by introducing a disorder viscosity correction that approximates the energetic penalty for entering a disordered state. This correction enables interface formation while suppressing unphysical long-range concentration fluctuations, allowing the spinodal wavelength to be computed self-consistently using a small set of reasonable approximations. The resulting predictions show strong agreement with experiments and indicate that, in NiRh, spinodal microstructures can produce yield strengths surpassing those of Ni-based superalloys. The approach is well-suited for high-throughput screening and machine-learning workflows, accelerating exploration and optimization of compositionally complex and high-entropy materials governed by spinodal kinetics.