Bottom-up coarse-grained models that accurately describe the structure, pressure, and compressibility of molecular liquids

• Published in: The Journal of chemical physics Vol. 143; no. 24; p. 243148
• Main Authors: Dunn, Nicholas J H, Noid, W G
• Format: Journal Article
• Language: English
• Published: United States 28-12-2015
• Subjects: Heptanes - chemistry; Molecular Dynamics Simulation; Molecular Structure; Pressure; Thermodynamics; Toluene - chemistry
• ISSN: 1089-7690
• DOI: 10.1063/1.4937383

The present work investigates the capability of bottom-up coarse-graining (CG) methods for accurately modeling both structural and thermodynamic properties of all-atom (AA) models for molecular liquids. In particular, we consider 1, 2, and 3-site CG models for heptane, as well as 1 and 3-site CG models for toluene. For each model, we employ the multiscale coarse-graining method to determine interaction potentials that optimally approximate the configuration dependence of the many-body potential of mean force (PMF). We employ a previously developed "pressure-matching" variational principle to determine a volume-dependent contribution to the potential, UV(V), that approximates the volume-dependence of the PMF. We demonstrate that the resulting CG models describe AA density fluctuations with qualitative, but not quantitative, accuracy. Accordingly, we develop a self-consistent approach for further optimizing UV, such that the CG models accurately reproduce the equilibrium density, compressibility, and average pressure of the AA models, although the CG models still significantly underestimate the atomic pressure fluctuations. Additionally, by comparing this array of models that accurately describe the structure and thermodynamic pressure of heptane and toluene at a range of different resolutions, we investigate the impact of bottom-up coarse-graining upon thermodynamic properties. In particular, we demonstrate that UV accounts for the reduced cohesion in the CG models. Finally, we observe that bottom-up coarse-graining introduces subtle correlations between the resolution, the cohesive energy density, and the "simplicity" of the model.