Multi-paradigm modeling of mode I&II dynamic fracture mechanisms in single crystal silicon

Abstract

In addition to its semi-conducting properties, silicon has the ability to be manipulated with high precision at very small length- scales. This property makes it very useful in the design of Nano/Micro-Electromechanical Systems (N/MEMS) and similar technologies. The understanding of fracture of silicon is crucial for the engineering process and the development of robust devices. However, the mechanisms of fracture in silicon are complex and are still not fully understood. Several experimental studies of fracture have been reported, however, these often lack insight into atomistic mechanisms of fracture. Ab initio computational methods (e.g. based on Density Functional Theory) to study silicon that are able to provide a fundamental description of the complex fracture mechanisms remain an open challenge. In particular, the mechanisms that lead to brittle cleavage or to the transition to ductile behavior of silicon at higher temperatures remains an open question. Empirical molecular dynamics (MD) studies have proven successful in simulating silicon fracture, but are unreliable and most models could not be validated against experimental results. Here we propose to use MD modeling based on a novel first principles reactive force fields ReaxFF, which has shown to be an accurate model to describe fracture processes of silicon. Two numerical methods are used here to study fracture mechanisms in silicon: a multi-paradigm model employing reactive and non-reactive force fields, and a fully reactive model. The CMDF and GRASP are used for the simulation of brittle fracture mechanisms in mode I and mode II loading conditions, as well as simulations of the brittle-to-ductile transition (BDT). Our results indicate that CMDF is suitable for modeling silicon brittle fracture, but has limitations during the study of the mechanisms involved in the BDT. GRASP provides a suitable framework for BDT study, and the results in this study provide for the first time an observation of the BDT without the use of an empirical model. In this thesis we report, for the first time, the direct atomistic simulation of the BTD in silicon, revealing the microscopic atomistic mechanisms that explains this drastic change in the behavior of silicon.