We describe joint atomistic and continuum studies of deformation and failure in brittle solids and thin film systems. The work is organized in four parts. In the first part, we present a review on atomistic modeling and analysis tools, including a summary of recent research activities in the field.

The second part is dedicated to joint continuum-atomistic modeling of dynamic fracture of brittle materials, where we employ one-, two- and three-dimensional models. The main focus is a systematic comparison of continuum mechanics theory with atomistic viewpoints. An important point of interest is the role that material nonlinearities play in the dynamics of fracture. The elasticity of a solid clearly depends on its state of deformation. Metals will weaken or soften, and polymers may stiffen as the strain approaches the state of materials failure. It is only for infinitesimal deformation that the elastic moduli can be considered constant and the elasticity of the solid linear. However, many existing theories model fracture using linear elasticity. Certainly, this can be considered questionable since material fails at the tip of a dynamic crack because of extreme deformation. We show by large-scale atomistic simulations that hyperelasticity, the elasticity of large strains, can play a governing role in the dynamics of fracture and that linear theory is incapable of capturing all phenomena. We introduce the concept of a characteristic length scale for the energy flux near the crack tip and demonstrate that the local hyperelastic wave speed governs the crack speed when the hyperelastic zone approaches this energy length scale. This length scale implies that in order to sustain crack motion, there is no need for long-range energy transport. Instead, only energy stored within a region defined by the characteristic energy length scale needs to be transported toward the crack tip in order to sustain its motion. This new concept helps to form a more complete picture of the dynamics of fracture. For instance, the characteristic energy length scale explains the observation of crack motion faster than all wave speeds in the solid, including recent experimental reports of mode I cracks faster than the shear wave speed. The existence of this novel length scale is verified for mode I and mode III cracks. Further, we show that hyperelasticity also governs dynamic crack tip instabilities. Stiffening material behavior allows for straight crack motion up to super-Rayleigh speeds, and softening material behavior causes the crack tip instability to occur at speeds as low as one third of the theoretical limiting speed, in accordance with experimental results. Additional studies focus on the dynamics of suddenly stopping cracks as well as the dynamics of fracture along interfaces of dissimilar materials. An important result in this area is the discovery of a novel mother-daughter mechanism of mode I cracks moving along interfaces of stiff and soft materials, leading to supersonic mode I fracture.

The third part is devoted to the mechanical properties of ultra thin submicron copper films. We discuss a novel material defect referred to as a diffusion wedge, recently proposed theoretically and observed indirectly in experiment. The theory predicts that tractions along the grain boundary are relaxed by diffusional creep and a diffusion wedge is built up. Due to traction relaxation, the diffusion wedge behaves as a crack along the grain boundary in the long-time limit. As a consequence, large resolved shear stresses on glide planes parallel to the film surface develop that cause nucleation of dislocations on glide planes parallel to the film surface and close to the film-substrate interface, referred to as parallel glide dislocations. This new dislocation mechanism in thin films, though standing in contrast to the well known Mathews-Freund-Nix mechanism of threading dislocation propagation, has been observed recently in experiments of ultra thin submicron copper films subject to thermal stress. We discuss joint atomistic-continuum modeling of such diffusion wedges, with a focus on the relation of diffusion and nucleation of dislocations. We propose a Rice-Thomson model for nucleation of parallel glide dislocations, and report a critical condition for initiation of grain boundary diffusion in thin films leading to a threshold stress for diffusion initiation independent of the film thickness. We extend the existing continuum model to account for the new concept of a threshold stress and model experimental thermal cycling curves. The new model improves the stress-temperature curves particularly at high temperatures. By large-scale atomistic modeling, we study the atomic details of buildup of the diffusion wedge and subsequent parallel glide dislocation nucleation. Based on our atomistic simulation results, we calculate a critical stress intensity factor as a condition for nucleation of parallel glide dislocations. We show that this criterion can serve as input parameter for mesoscopic discrete dislocation modeling of constrained diffusional creep. By atomistic studies of polycrystalline thin films, we study the transition from classical threading dislocations to parallel glide dislocations. In agreement with experimental findings and the classical understanding, threading dislocations are found to dominate when tractions are not relaxed by diffusion. If grain boundary tractions are relaxed by diffusional creep, parallel glide dislocations dominate due to the crack-like deformation field near the diffusion wedge. Another result is that the structure of grain boundaries has impact on dislocation nucleation and on the motion of dislocations along grain boundaries. Low-energy grain boundaries provide more fertile sources for dislocations than high-energy grain boundaries. We also discuss the role of the grain boundary structure on the diffusivities and show by large-scale atomistic studies of diffusional creep in polycrystalline thin films that high-energy grain boundaries provide faster diffusion paths than low-energy grain boundaries. Finally, a deformation map summarizes the range of dominance of different strain relaxation mechanisms in ultra-thin films. We show that besides the classical “threading dislocation” regime, there are numerous novel mechanisms once the film thickness approaches nanoscale.

In the fourth and last part of this thesis, we emphasize the potentials and limitations of molecular-dynamics simulations in studying small-scale materials phenomena, and include a critical assessment of the simulation methods employed in this work and the validity of the results. Finally, the most important results of this thesis are briefly summarized and an outlook to possible future research is provided.