Computational Design of Grain Boundary Topology for Fracture Mitigation in Additively Manufactured Tungsten
Please login to view abstract download link
The intrinsic brittleness and intergranular cracking of tungsten pose significant challenges for its application in extreme environments, such as nuclear fusion reactors. While conventional mitigation focuses on thermal stress, this work introduces a data-driven "Grain Boundary (GB) Topology Engineering" framework to enhance fracture resistance. Integrating large-scale Molecular Dynamics (MD) with high-fidelity Atomic Cluster Expansion (ACE) machine-learning potentials, we systematically explore the fracture energy landscape of diverse GB triple junction configurations. Our atomistic simulations reveal a definitive topology-performance correlation: triple junctions with large dihedral angles promote intensive crack-tip blunting and dislocation-mediated plasticity, effectively arresting crack propagation. In contrast, small-angle junctions provide negligible resistance. To bridge atomistic discovery with manufacturing, thermo-mechanical Finite Element (FE) simulations are employed to decode microstructural evolution during multi-cycle laser scanning. Results demonstrate that cyclic high-temperature plasticity drives GB network reconstruction toward these desired high-angle configurations. By establishing a multiscale "Process-Structure-Property" linkage, we provide predictive design rules to engineer "crack-free" brittle metals through local topological control. This research demonstrates how computational mechanics transitions from descriptive modelling to a predictive tool for the sustainable design of high-performance refractory materials.
