The influence of binary mixtures' structuring on the calculation of Kirkwood-Buff integrals: A molecular dynamics study

Abstract

The Kirkwood-Buff integrals (KBIs) provide a link between structuring on an atomic scale accessed by simulations and thermodynamics, which makes it a powerful tool for bridging the gap between microscopic and macroscopic quantities. Here we present the simulation study of the concentration dependence of KBIs for various types of binary ethanol mixtures which range from simple to complex. The thermodynamically ideal but extensively clustered methanol-ethanol mixture shows linearity in KBIs. The non-ideal mixtures, ethanol-water and ethanol-benzene, have the extrema in KBIs which are linked to the formation of the domains due to the association of molecules of the same species. We also explore some of the issues in KBI calculation, as KBI, a property of an open ensemble, is usually obtained by using the running integrals of a pair correlation function, calculated from the simulation of constant number of particles placed in a finite simulation box. The study includes a multifaceted analysis of the LP shift (J. Mol. Liq. 159, 52, 2011) and Ganguly and van der Vegt correction (J. Chem. Theory Comput. 9(3), 1347, 2013) of the radial distribution function, and compares different approaches in the computation of running KBI (J. Phys. Chem. Lett. 4(2), 235, 2013), both of which compensate for finite-size and ensemble effects. The results show how the best approach to the KBI calculation varies depending on the structural complexity of the simulated system.