Chemical similarity description of liquid kinetics across composition space in magnesium alloys

Liquid kinetics is vital for designing and processing advanced metallic materials, yet linking atomic diffusivity with solute chemistry across an extensive composition space remains challenging. In this work, an automated high-throughput ab initio workflow is developed to efficiently calculate the diffusion coefficients of 31 binary liquid Mg alloys. Mapping liquid kinetics onto the periodic table reveals distinct chemical similarity trends, which are quantitatively captured by a robust quadratic scaling with the Pettifor chemical scale. This correlation enables the construction of a unified model for quantifying liquid diffusivities across temperatures and compositions. Our proposed phenomenological model achieves accuracy comparable to that of the data-driven symbolic regression model while retaining physical interpretability, and outperforms the physics-based excess-entropy scaling approach. This work bridges the gap between static chemical descriptors and dynamic liquid properties, offering an accurate and structure-free paradigm for screening liquid kinetics of alloys.