Numerical methods for frustrated magnetism: from quantum to classical spin systems

Abstract

In this thesis, we will concentrate on the numerical study of classical and quantum frustrated magnets. If we focus on the full set of problems which are contained in the class of quantum spin systems, we see that the techniques which can deal with them are plenty, but that the presence of frustration introduces several methodological hindrances. Whether it is the presence of a sign problem, or the extensive resources needed to perform the computation, many of these methods are efficient only on a reduced subset of frustrated systems. We will employ the pseudofermion-Functional Renormalization Group (pfFRG) method as our method of choice to tackle two dimensional frustrated quantum magnets, all of them modeled via a Heisenberg Hamiltonian. We will extend the pfFRG formalism to include spin systems where the spin length is unrestricted, $S \geq 1/2$. With this extension we are able to study in detail what the effect of quantum fluctuations are on the system. From the extreme quantum limit ($S = 1/2$) to the classical limit ($S \rightarrow \infty$), the large-S extension of pfFRG allows us to manipulate the strength of the quantum fluctuations and study in detail how the quantum to classical transition happens. We apply this method to the study of incommensurate phases in the Heisenberg honeycomb model. We map the phase diagram for different values of the spin length and analyze how this length, and subsequently quantum fluctuations, affect the phases found in the classical limit. Furthermore, we prove that in the classical limit, pfFRG reduces to the Luttinger-Tisza formalism.

When quantum fluctuations can be neglected, we model our frustrated magnet as a classical spin system. While this limit can easily be obtained from the large-S extension of pfFRG, this methodology is constrained to the study of two point correlators at zero temperature. To collect information regarding the behavior of our classical magnet beyond the two point correlators we study classical spin systems via Monte Carlo simulations. In this case we show how the method can be applied to continuous spin systems with strong anisotropic interaction. Furthermore, we employ this technique to numerically study $\alpha$-Li$_2$IrO$_3$, a material which exhibits an incommensurate ground state but for which a minimal model has not been determined. We study and compare the possible minimal models that have been proposed. We reduce the number of minimal models by showing that many of those that have been proposed do not reproduce the full set of experimental results. Furthermore, we predict the magnetic behavior (for those models which reproduce the experimental results) in the presence of an external magnetic field. With this study we obtain magnetization processes and propose experiments which can confirm if one of the studied models is the correct minimal model for $\alpha$-Li$_2$IrO$_3$.