Improved implementation of Kirkwood-Buff solution theory in periodic molecular simulations

Abstract

Kirkwood-Buff (KB) solution theory is a means to obtain certain thermodynamic derivatives from knowledge of molecular distributions. In actual practice the required integrals over radial distribution functions suffer inaccuracies due to finite-distance truncation effects and their use in closed systems. In this work we discuss how best to minimize these inaccuracies under traditional KB theory. In addition we implement a method for calculating KB quantities in molecular simulations with periodic boundary conditions and particularly within the canonical ensemble. The method is based on a finite-Fourier-series expansion of molecular concentration fluctuations and leads to more reliable results for a given computational effort. The procedure is validated and compared to the original method for a nonideal liquid mixture of Lennard-Jones particles intended to imitate a real system, carbon tetrafluoride, and methane.