A molecular density functional theory to study solvation in water

TL;DR

This paper develops a classical density functional theory approach to efficiently study solvation in water, accurately reproducing molecular dynamics results at a fraction of the computational cost.

Contribution

It introduces an approximate excess functional for water based on correlation functions, enabling fast and accurate solvation studies of various solutes.

Findings

Reproduces MD results for solvation free energy

Achieves at least three orders of magnitude reduction in computational cost

Successfully models complex solutes like proteins and clay

Abstract

A classical density functional theory is applied to study solvation of solutes in water. An approx- imate form of the excess functional is proposed for water. This functional requires the knowledge of pure solvent direct correlation functions. Those functions can be computed by using molecular simulations such as molecular dynamic or Monte Carlo. It is also possible to use functions that have been determined experimentally. The functional minimization gives access to the solvation free energy and to the equilibrium solvent density. Some correction to the functional are also proposed to get the proper tetrahedral order of solvent molecules around a charged solute and to reproduce the correct long range hydrophobic behavior of big apolar solutes. To proceed the numerical minimization of the functional, the theory has been discretized on two tridimensional grids, one for the space coordinates, the other for the angular coordinates, in a functional minimization code written in modern Fortran, mdft. This program is used to study the solvation in water of small solutes of several kind, atomic and molecular, charged or neutral. More complex solutes, a neutral clay and a small protein have also been studied by functional minimization. In each case the classical density functional theory is able to reproduce the exact results predicted by MD. The computational cost is at least three order of magnitude less than in explicit methods.