Crystal structure search and electronic properties of alkali doped phenanthrene and picene

TL;DR

This study predicts the crystal structures and electronic properties of alkali-doped phenanthrene and picene using ab-initio methods, revealing metallic, insulating, and magnetic phases relevant to superconductivity.

Contribution

It introduces a comprehensive ab-initio approach to determine the crystal structures and electronic phases of alkali-doped aromatic hydrocarbons, highlighting new structural motifs and magnetic transitions.

Findings

Metallic phases are ionic with three electrons per molecule.

Insulating phases result from charge reduction to two electrons per molecule.

Both phenanthrene and picene exhibit structures that may support superconductivity.

Abstract

Alkali doped aromatic compounds have shown evidence of metallic and superconducting phases whose precise nature is still mysterious. In potassium and rubidium doped phenanthrene, superconducting temperatures around 5 K have been detected, but such basic elements as the stoichiometry, crystal structure, and electronic bands are still speculative. We seek to predict the crystal structure of M3-phenanthrene (M = K, Rb) using ab-initio evolutionary simulation in conjunction with density functional theory (DFT), and find metal but also insulator phases with distinct structures. The original P21 herringbone structure of the pristine molecular crystal is generally abandoned in favor of different packing and chemical motifs. The metallic phases are frankly ionic with three electrons acquired by each molecule. In the nonmagnetic insulating phases the alkalis coalesce reducing the donated charge from three to two per phenanthrene molecule. A similar search for K3-picene yields an old and a new structure, with unlike potassium positions and different electronic bands, but both metallic retaining the face-to-edge herringbone structure and the P21 symmetry of pristine picene. Both the new K3-picene and the best metallic M3-phenanthrene are further found to undergo a spontaneous transition from metal to antiferromagnetic insulator when spin polarization is allowed, a transition which is not necessarily real, but which underlines the necessity to include correlations beyond DFT. Features of the metallic phases that may be relevant to phonon-driven superconductivity are underlined.