Modeling rough energy landscapes in defected condensed matter

[摘要] This dissertation is a computational and theoretical investigation of the behavior of defected condensed matter and its evolution over long time scales. The thesis provides original contributions to the methodology used for simulating the slow evolution of complex condensed matter, as well as applications to three problems: hydrogen embrittlement of metals, radiation swelling, and cement setting. The problems are characterized by an increasing degree of complexity as the microstructure ranges from a crystalline alloy with few point defects to a highly damaged metal with partial amorphization to a semi-crystalline colloidal system. First, we investigate the interactions of hydrogen with point defect clusters (PDC) in Fe-C alloys via a combination of density functional theory and a statistical mechanics model. We cast our PDC concentration results in a novel PDC dominance diagram representation that can be generalized to any type of alloy and impurities. We also calculate the migration mechanisms and energy barriers for the most relevant PDC species in Fe-H. Our results demonstrate the essential role of hydrogen-vacancy interactions in mediating the formation and migration of PDCs, and the relevance of these crystalline defects to the problem of hydrogen embrittlement. Second, we study the effect of self-interstitial atoms (SIA) on radiation swelling at high dose rates. Using a combination of non-equilibrium molecular dynamics and two generalizations of the autonomous basin climbing (ABC) method, we characterize the structure and evolution of the defective species that result from SIA insertion during irradiation. Consistent with ion beam implantation experiments on surfaces, we show that, at high dose rates, swelling is a consequence of the nucleation and growth of disordered phases. This process is governed by a competition between defect generation and recrystallization, even at long times. Third, a binary colloidal model incorporating sticky interactions is developed to simulate chemomechanical hardening, with an application to cement setting. The model, inspired by a coarse-grained analogy with stress-corrosion cracking, captures gelation, diffusion, and percolation kinetics. The model is characterized and the effects of various parameters on setting kinetics are discussed. We find that, as observed experimentally, the induction time is dependent on the relative concentration and masses of the two colloidal phases. The application of ABC (static and dynamic) to the study of cement setting under more realistic conditions is also discussed, including objective means of calculating the effective elastic moduli of viscoelastic materials via atomistic simulations.

[发布机构] Massachusetts Institute of Technology