A comparison of computed and experimental neutron diffraction intensity at large momentum for MnO and NiO

TL;DR

This study compares experimental neutron diffraction intensities with theoretical calculations for MnO and NiO, highlighting the superior accuracy of fixed-node diffusion Monte Carlo in capturing magnetic properties.

Contribution

It provides a detailed benchmark comparing three theoretical methods against experimental data for magnetic neutron scattering in MnO and NiO.

Findings

Diffusion Monte Carlo outperforms DFT in NiO.

DMC shows the lowest RMS error for form factor and intensity.

Spin density in DMC is more diffuse, spreading away from bonds.

Abstract

Magnetic neutron scattering measures spin-spin correlations giving information about the long-range spin order as well as the shape of the spin density in magnetic materials. Similarly, detailed first principles calculations directly compute the spin density in materials. In this work, the authors carefully compare experimentally measured magnetic neutron intensities to three levels of theory: density functional theory in two approximations, and fixed-node diffusion Monte Carlo. While each theory performs similarly for the simple antiferromagnet MnO, there are significant differences between density functional theory and diffusion Monte Carlo in NiO. For both materials, fixed-node diffusion Monte Carlo shows the lowest RMS error with respect to experiment for the form factor and the intensity. Through connection of the intensities and form factors to the real-space spin densities, it is shown that diffusion Monte Carlo spin density becomes more diffuse by spreading spin away from the bond directions. This benchmark is of importance when considering the efficacy of an ab initio calculation in capturing both the long-range magnetic correlations and the microscopic spin details of magnetic systems.