Thermostatting Open Systems in Non-Equilibrium Computer Simulations

点击翻译按钮获取中文摘要

Molecular Dynamics (MD) simulations, in which atoms move according to Newtonian's dynamics, have been extensively used to study various processes in matter in which materials are under considerable mechanical stress, subjected to a temperature gradient, irradiated upon by high-energy particles, etc. For instance, in tribology applications, two surfaces are sheared with respect to each other, and bonds are constantly formed and broken in the contact area leading to friction; this results in the contact area being much hotter than the bulk of the two materials leading to constant energy flow from the contact region outwards into the bulk. Correct treatment of the effects responsible for describing energy dissipation from the hot region into the bulk of the materials is crucial for a realistic description of friction. As another example, in irradiation processes high-energy particles impinged on a crystal surface (e.g. in coatings of nuclear reactors) go through the material forming canals. Locally along the canal large amounts of energy are released leading to considerable damage (deformation, defects formation, etc.); at the same time, a large portion of that energy is dissipated inside the material propagating the damage radially out of the canal. Correct treatment of such dissipation effects is vital for simulating the realistic damage to the material and hence finding new materials, which can sustain higher radiation doses. In these and many other cases, due to considerable computational cost, only a small fragment of the material can actually be studied at the atomic level by practical MD simulations. At the same time, violent processes releasing considerable amounts of energy, which dissipates into the bulk of the material(s), require considering essentially infinite systems. Thermostatting is meant to solve this problem by providing mechanisms whereby a finite fragment of the real system is simulated, however, the environment is mimicked as a heat bath (kept at constant temperature), which can take or give energy in accordance with the laws of thermodynamics. Unfortunately, in very many cases, equilibrium MD simulations are still frequently used although this can hardly be justified! This still happens because there is no viable alternative. The well-known Generalised Langevin Equation (GLE) method is specifically designed to take care of the energy exchange with the environment (the heat bath), in spite of its long history, GLE has not yet been exploited sufficiently and implemented into a code to offer a practical solution. The know-how generated by this research project offers this solution. The methodology to be generated and the computer codes (both the full implementation of GLE and the approximate schemes) will be applicable to simulations of a wide class of non-equilibrium phenomena such as (but not limited to): tribology; thermal transport through bulk materials and layered systems, nanowires and molecular junctions; relaxation of point defects in the bulk and surfaces of crystals; irradiation problems; film and crystals growth on substrates at elevated temperatures (e.g. epitaxial), etc.