Van der Waals density-functional description of polymers and other sparse materials

Doktorsavhandling, 2006

Polymers are abundant in nature --- rubber, tar and latex have been known to mankind for thousands of years. Early in the 20th century, a systematic synthesis of polymers was developed, but the great potential of synthesized polymers was recognized first during World War II. Since then, polymeric materials of enormous variety have been developed and constructed, and research has been devoted to improve the production and to develop new polymeric materials. A prediction of the materials behavior of such complex systems requires insight at all length scales. Proper microscopic quantum-mechanical calculations are prerequisites but out of reach of covering all length scales. As a first step in the direction of such a general description, we treat one of the simplest polymer systems by help of first-principles density-functional theory (DFT) calculations. Specifically, interactions of chains of simple linear nonbranched polyethylene (PE) are investigated. PE represents an important class of systems that form complexes stabilized by weak but long-ranged dispersive interactions. Traditional DFT does not include the latter and predicts the PE crystal to be unstable, contradicting both experiments and intuition. A recently proposed density functional (vdW-DF), with a consistent account of the dispersive interactions for general geometries [Phys. Rev. Lett. {\bf 92}, 246401, 2004], is implemented to infinite crystalline systems and applied to crystalline PE. The vdW-DF does not only lead to a stable PE crystal structure but also predicts crystal-parameter values in promising agreement with experimental data. This motivates our application of vdW-DF to other technologically important sparse-matter systems, including dimers of parallel PE, PP, and PVC polymers, hydrogen and potassium intercalation in graphite and bundles of nanotubes. The adopted first-principles methods are based on electron-structure calculations and differ significantly from the simplified force-field approaches. These have potential parameters fitted to experimental data at the equilibrium separation and are widely used for complex polymer systems. Here they are explicitly shown to give general intermolecular parameters that lack any solid physical foundation, and which thus has no guaranteed success for systems outside the training set or at separation beyond the equilibrium separation. The first-principles insight gained makes possible a a well-founded interatomic description and in turn better predictive power of these fast force-field schemes.