Effect of doping and pressure on magnetism and lattice structure of Fe-based superconductors

TL;DR

This study uses first principles calculations to analyze how doping and pressure affect the magnetic and structural properties of BaFe$_2$As$_2$, revealing limitations of density functional theory in capturing some experimental trends.

Contribution

It provides a detailed comparison between DFT predictions and experimental observations, highlighting discrepancies in structural and magnetic responses under pressure and doping.

Findings

DFT accurately reproduces ambient pressure magnetic and structural orderings.

Discrepancies in Fe-As bond length behavior under pressure between theory and experiment.

Magnetism's dependence on doping site is strongly site-specific.

Abstract

Using first principles calculations, we analyze structural and magnetic trends as a function of charge doping and pressure in BaFe$_2$As$_2$, and compare to experimentally established facts. We find that density functional theory, while accurately reproducing the structural and magnetic ordering at ambient pressure, fails to reproduce some structural trends as pressure is increased. Most notably, the Fe-As bondlength which is a gauge of the magnitude of the magnetic moment, $\mu$, is rigid in experiment, but soft in calculation, indicating residual local Coulomb interactions. By calculating the magnitude of the magnetic ordering energy, we show that the disruption of magnetic order as a function of pressure or doping can be qualitatively reproduced, but that in calculation, it is achieved through diminishment of $|\mu|$, and therefore likely does not reflect the same physics as detected in experiment. We also find that the strength of the stripe order as a function of doping is strongly site-dependent: magnetism decreases monotonically with the number of electrons doped at the Fe site, but increases monotonically with the number of electrons doped at the Ba site. Intra-planar magnetic ordering energy (the difference between checkerboard and stripe orderings) and interplanar coupling both follow a similar trend. We also investigate the evolution of the orthorhombic distortion, $e=(a-b)/(a+b),$ as a function of $\mu$, and find that in the regime where experiment finds a linear relationship, our calculations are impossible to converge, indicating that in density functional theory, the transition is first order, signalling anomalously large higher order terms in the Landau functional.