Phase switch Monte Carlo

• Main Author: McNeil-Watson, Graham
• Format: Dissertation
• Language: English
• Published: University of Bath 2007

Computational studies of phase behaviour have always proved difficult, since phase . transitions are inherently slow processes compared to accessible simulation timescales. Despite valiant efforts by researchers there remains a dearth of efficient, robust and scalable methods for determining phase equilibria,· especially in the case of fluid-crystalline solid transitions. This thesis is about such phase coexistence problems, the existing solutions, and more advanced methods that have only recently come into their own. ... Extended sampling methods are examined in detail, and applied to a testbed system, the critical point Lennard-Jones fluid, leading to an estimate of the system free energy in the thermodynamic limit. Then a comparatively new technique, phase switch Monte Carlo (Phys. Rev. Lett. 85, 5138) is applied initially to the venerable hard sphere system. The method overcomes many of the shortcomings present in other works by directly connecting the coexisting phases in a single simulation, and doing so without creating an artificial inter-phase route but rather affecting a direct 'phase leap' from one phase to the other. Finally, phase switch is generalised to soft potentials and applied to the Lennard-Jones freezing transition, resulting in an extensive mapping of the phase boundary for a variety of system sizes (J. Chern. Phys 124, 064504).