Understanding Complex Magnetic Interactions Using Diffuse Neutron Scattering

Many important magnetic materials have complex bond-dependent magnetic interactions, such as Kitaev and frustrated spin-liquid candidates [1,2]. Traditionally, spin Hamiltonians of such systems are determined by fitting to spin-wave spectra measured below the magnetic ordering temperature T_N. Unfortunately, this approach is often impractical in spin liquids, either because T_N is unmeasurably low, or the ordering is poorly understood.

I discuss an alternative approach to parameterize the spin Hamiltonian that does not assume long-range magnetic order: fitting interaction models to diffuse magnetic neutron-scattering data measured above T_N. I demonstrate the sensitivity of this method for canonical models of bond-dependent interactions on triangular and honeycomb lattices. I discuss my results in the context of developments in reverse Monte Carlo refinement [3], pair distribution function analysis [4], and machine learning [5].

[1] Science 356, 1055 (2017); [2] Nat. Phys. 13, 117 (2017); [3] Phys. Rev. Lett. 108, 017204 (2012); [4] IUCrJ 5, 410 (2018); [5] arXiv:1906.11275 (2019).