A dangerous over-confidence now prevails in the assumption that detailed all-atom or united-atom models which are used to represent the properties of fluid molecules (e.g. the OPLS-type potentials) are sufficient to describe molecular systems with a precision that supplements experiments. More than 1% of all recent articles published in the open science and engineering community deal with molecular simulations at this level and in some cases the accuracy of the results is taken for granted. The fitting of parameter of the force fields is, however, still rather unsophisticated as compared to other aspects of computer modelling. Common practice is to hand fit a few parameters to a few experimental data points (e.g., a radial distribution function, solubility data and/or enthalpies at a given temperature or phase state). In this contribution we propose a new way of obtaining the required force field parameters. In our methodology one requires access to a physical-based equation of state that describes the complete Helmholtz free energy in closed algebraic form, i.e., an equation of state (EoS) that is based on a defined intermolecular potential. Such an equation can then be used to explore a very large parameter space to estimate the locally optimal parameter set that provides an optimal description of the available macroscopical experimental data. This parameter set represents not just a unique fit to a single temperature or density, but rather an over-arching average. If the equation of state is expressed in terms of the free energy of the system for a well defined intermolecular potential, it can be used to develop a “top-down averaged” intermolecular potential. Here we follow this line of thought and present a proof-of-concept of such methodology, employing a recently developed EoS of the Statistical Associating Fluid Theory (SAFT) family using the so-called Mie intermolecular potential.
view more