Chemical potentials of hydrogen-bonded aqueous mixtures from adaptive resolution simulations and Kirkwood–Buff theory

Abstract

We combine Kirkwood–Buff (KB) theory with the Hamiltonian adaptive resolution simulation (H-AdResS) method to efficiently compute chemical potentials in aqueous mixtures of urea, ethylene glycol, methanol, ethanol, and propanol over a wide concentration range. In this approach, explicit chemical potential calculations using H-AdResS are required at only a few state points. In contrast, the concentration dependence of the chemical potential is obtained from an expression containing KB integrals (KBIs). In practice, evaluating KBIs amounts to computing fluctuations of the number of particles from conventional molecular dynamics trajectories, thereby significantly reducing the computational load compared to explicit free-energy calculations. The reliability of the KB analysis is assessed by computing isothermal compressibility, as well as density and concentration structure factors, quantities that are sensitive to long-range density fluctuations. For the monohydroxy alcohol–water mixtures considered, the resulting chemical potentials are then used to determine the alcohol activity coefficient as a function of concentration and to extrapolate the water activity coefficient via the Gibbs–Duhem relation, thereby enabling the computation of the excess Gibbs free energy and excess mixing entropy. Hence, the combined KB/H-AdResS framework provides a computationally efficient, thermodynamically consistent method for calculating chemical potentials in complex aqueous mixtures.