Percolation threshold of TCP clusters governs relaxation dynamics in Cu-Zr metallic glass-forming liquids

The origin of the abrupt dynamic slowdown observed during the concealed liquid-solid transition at temperature T_ls in rapidly cooled Cu-Zr metallic liquids remains poorly understood. Here, we resolve this longstanding puzzle by combining molecular dynamics simulations with the largest standard cluster analysis (LaSCA) to establish a direct causal link between topological cluster percolation and relaxation dynamics. Our findings reveal that topologically close-packed (TCP) clusters with an average size (S_TCP) exceeding a critical threshold (>3) trigger a dynamic arrest below T_ls. This arrest arises from cooperative atomic motions driven by the spatial percolation of TCP clusters, which dominate the structural evolution. The identified S_TCP threshold (3.0–4.0 across compositions) serves as a universal signature of rigidity onset in metallic glass-forming liquids. By bridging atomic-scale topological connectivity with macroscopic dynamics, this work provides a predictive framework for tailoring relaxation behavior in non-equilibrium alloys.