The Discrete World of discontinuous Potential Modeling

The Discrete World of discontinuous Potential Modeling

Jingyu Cui, Chandra Patni, and J. Richard Elliott, Jr.
Chemical Engineering Dept.
The University of Akron
Akron, OH 44325-3906
Phone*: 330 972 7253
Fax*: 330 972 5856


Discontinuous potential models characterize molecular interactions as discrete increments in potential energy. The simplest example is the square-well potential, but multi-step potentials can be easily conceived to resemble continuous potentials like the Lennard-Jones potential. Adaptations based on the work of Wertheim (1986) permit characterization of hydrogen bonding interactions. There are two advantages of discontinuous potential models: (1) The computational procedure for Discontinuous Molecular Dynamics (DMD) is very efficient, leading to CPU times that scale as Nln(N) where N represents the number of atoms explicitly represented. (2) The piecewise nature of the potential model facilitates the application of theories like perturbation theory. One theory can be adapted to a piece of the potential for which it works best while another theory is adapted to another piece of the potential.

A recent demonstration of isochoric integration (Elliott and Hu, 1999) as a reliable method of phase equilibrium calculation opens the door to several intriguing prospects.

(1) MD simulation can provide transport properties simultaneously with phase equilibrium properties.

(2) Coincident with the demonstration of isochoric integration was the finding that Thermodynamic Perturbation Theory (TPT) is much more accurate than previously appreciated. This means that the results of molecular simulations can be leveraged to a very high extent, so detailed simulations at specific conditions may be superfluous in many cases.

(3) Detailed features of the intermolecular potential model (aka. Force field) can be efficiently and precisely "tuned" to correlate macroscopic properties of a large database and those potential models can be used to predict unmeasured properties for new compounds. The role of specific simulations then becomes one of fine-tuning the characterization of perturbation contributions, rather than exhaustive computation of all properties at all conditions of interest.

This presentation shows recent results for force field characterization of n-alkanes, aromatics, alcohols, and water based on accurately matching experimental vapor pressures with those computed by isochoric integration. Vapor pressure results for several different step characterizations are illustrated along with results for other thermodynamic properties. The perturbation contributions are illustrated to show the variation in the contributions with varying molecular complexity.