Atomistic Insights into Metallic Fracture

DEREK WARNER
Department of Civil and Environmental Engineering
Cornell University

Abstract:
Since the late 1940’s, atomistic models for mechanical behavior have provided great insights into the fundamental mechanisms
controlling deformation and fracture. While the evolution of these models has been dramatic, from soap bubble rafts to billion
atom 3D simulations, the true predictive capability of atomistic models remains limited. Four longstanding challenges have been
the limited spatial and temporal domains associated with atomistic models, the accurate description of interatomic forces in a
computationally tractable manner, and the characterization of properties across the vast, complex, and influential interatomic
configuration space.

In this talk I will review these challenges and comment on the perspectives for overcoming them. Special attention will be
devoted to the immense discrepancy in the temporal domain. After introducing some common approaches for addressing this
challenge, a specific example involving crack-tip plasticity will be given that reveals an unexpected complexity and motivates
the use of more advanced approaches aimed at computing free energies of activation. Free energy based approaches will be
shown to be a requirement for predicting crack-tip plasticity as more approximate approaches are found to be unable of predicting
the correct mechanisms.