January 17, 2013

Speaker: William R. Smith, Faculty of Science, UOIT

Title: Some Specialized Molecular Simulation Algorithms and their Applications

Abstract: Molecular simulation algorithms for predicting the thermodynamic properties of classical fluids and their mixtures were first developed about 50 years ago. An ultimate goal is to predict system properties at both the macroscopic and molecular levels with a minimal need for experimental data. The two main approaches are Molecular Dynamics (MD) and Monte Carlo (MC); both are based on specifying an underlying mathematical model for the molecular interactions, and are typically implemented for several hundred particles (molecules) in a box with imposed periodic boundary conditions. MD algorithms solve Newton’s equations of motion for the particles, and macroscopic properties are calculated from time averages along the trajectories. MC algorithms assume an ergodic hypothesis and properties are calculated by averaging over a Markov chain using the probability density function for the states of the trajectories. MD has the advantage that it can also calculate transport properties, and since its implementation is relatively straightforward, many commercial and public domain computer packages exist.

On the other hand, MC is limited to the calculation of thermodynamic properties, but for which it is in principle much more powerful than MD. However, implementations are often complex and both system and property dependent, and as a result only a few general computer packages exist. In this general level talk, I will describe two specialized MC algorithms developed in my research group over the past 15 years, which are making their way into general simulation packages. These have expanded the application of such methods to new classes of problems of both scientific and practical significance. The methods permit the direct simulation of chemical reaction equilibria and of isoenergetic processes such as those involved in Joule-Thomson expansion and in the determination of adiabatic flame temperatures. Simulation of the former iso-energetic processes can be used to design refrigeration cycles, and hence to screen for new environmentally benign refrigerants; and the latter has applications to the design of industrial processes and to explosives. I will conclude by discussing recent ongoing work whose goal is to predict the thermochemical properties of aqueous electrolyte systems. These are of considerable importance, both for biological systems in general and for industrially important systems such as occur in thermochemical hydrogen production processes