Multiscale Fracture Modeling of Amorphous Materials: Bridging Atomistic and Continuum Mechanical Perspectives

Fracture is a multiscale phenomenon in which material properties at the atomic level influence the mechanical behavior at the macroscale. Molecular dynamics (MD) simulations provide a tool for investigating the underlying mechanisms, taking into account individual atoms or representative superatoms. However, even with today's computational capabilities, it is impossible to model entire components with MD. This challenge is addressed by multiscale simulations. The Capriccio method used in this work is a partitioned-domain approach focusing on amorphous materials that couples three-dimensional MD domains with a surrounding continuum solved using the finite element method (FEM). In this way, FEM simulations are enriched by insights into the molecular level of detail. In recent years, first fracture simulations have been performed with the Capriccio method using a coarse-grained model of atactic polystyrene. The present thesis takes up on this and explores several aspects that have not yet been addressed. One of the main objectives is to derive fracture mechanical quantities using the Capriccio method. For this purpose, silica glass is chosen as an example material whose fracture process zone is significantly smaller than that of polymers. Using an MD model with atomic resolution, virtual experiments are performed on standardized test setups, in particular on an edge-notched rectangular panel loaded by all three fracture modes, three- and four-point bending specimens, and K-test setups. The studies conducted show that the Capriccio simulation environment is capable of determining values for the critical stress intensity factor and the crack opening displacement that are close to experimental findings and theoretical predictions. A further focus of this work is the consideration of crucial aspects regarding the modeling of fracture in polymers, especially the influence of bond scission on the failure of thermoplastics. For this purpose, the analytical potential functions of a generic, computationally efficient MD model are modified to include a length-based bond breaking criterion. This model is used to investigate the effect of various parameters that are essential for fracture simulations of polymers, such as sample geometry, temperature, and molar mass, but also the influence of nanoparticles in terms of their size and volume fraction. Despite the generic nature of the MD model, trends consistent with experimental observations are revealed. Ultimately, the computational efficiency of the model enables fracture simulations of polymers taking millions of superatoms into account. Overall, this work provides insights into the microscopic fracture processes of amorphous materials under realistic boundary conditions. At the same time, the Capriccio coupling is investigated and the influence of key methodological and numerical parameters is identified. This paves the way for future research activities on modeling and improving material properties based on molecular considerations.