van der Waals density functional calculations of binding in molecular crystals

TL;DR

This paper evaluates the performance of the van der Waals density functional (vdW-DF) methods, including vdW-DF1 and vdW-DF2, for accurately modeling binding in molecular crystals, providing insights into their effectiveness and implementation details.

Contribution

The study compares vdW-DF1 and vdW-DF2 for molecular crystal binding, analyzing their non-local correlation enhancements and providing computational implementation details.

Findings

vdW-DF2 maintains enhanced non-local correlations similar to vdW-DF1.

The methods effectively model binding in molecular crystals like C60 and graphite.

Implementation details improve computational efficiency for sparse matter simulations.

Abstract

A recent paper [J. Chem. Phys. 132, 134705 (2010)] illustrated the potential of the van der Waals density functional (vdW-DF) method [Phys. Rev. Lett. 92, 246401 (2004)] for efficient first-principle accounts of structure and cohesion in molecular crystals. Since then, modifications of the original vdW-DF version (identified as vdW-DF1) has been proposed, and there is also a new version called vdW-DF2 [ArXiv 1003.5255], within the vdW-DF framework. Here we investigate the performance and nature of the modifications and the new version for the binding of a set of simple molecular crystals: hexamine, dodecahedrane, C60, and graphite. These extended systems provide benchmarks for computational methods dealing with sparse matter. We show that a previously documented enhancement of non-local correlations of vdW-DF1 over an asymptotic atom-based account close to and a few A, beyond binding separation persists in vdW-DF2. The calculation and analysis of the binding in molecular crystals requires appropriate computational tools. In this paper, we also present details on our real-space parallel implementation of the vdW-DF correlation and on the method used to generate asymptotic atom-based pair potentials based on vdW-DF.