UCLA

The central scientific objective in this thesis is to develop experimentally-testable theoriesfor the dynamical behavior of magnetic systems. Our second objective is to use our understanding of these systems to investigate their potential for generating thermally-induced spin currents, ultimately having thermoelectric applications. This phenomena is known as the spin Seebeck effect. In Chapter 1, we derive the low-energy, long-wavelength spectra of Heisenberg ferromagnets (FMs) and antiferromagnets (AFs), in their strongly-ordered regimes, from an intuitive starting point in a classical theory. Once quantized, these spin excitations become magnons. The strongly ordered phases are understood as a dilute magnon gas, but interactions become increasingly complex as we approach the magnetic transition temperature in 3D. In order to pursue a theory of magnetism which treats all phases–both ordered and disordered–on equal footing, we turn to Schwinger boson mean field theory (SBMFT). The Schwinger boson transformation fractionalizes the spin operators into two bosonic field operators, and the language initially appears to be less intuitive. Nonetheless, we will show how two-spinon excitations reproduce magnons, and then press on to regimes where both magnon-like and paramagnetic-like excitations proliferate in thermal equilibrium. In Chapter 2, we investigate the dynamical linear response of insulating magnetic systems in equilibrium. In the ordered phases, where time-reversal symmetry is broken, we find that spin currents are present in equilibrium. The thermally-averaged spin-spin correlators, which make up spin currents, may be evaluated from the dissipative part of the dynamic susceptibility tensor using the semiclassical fluctuation-dissipation theorem (FDT). Since spin currents are not directly measurable, we interface the insulating magnet with a metal. The interface acts as weak link, and when the two are out-of-equilibrium with each other, a spin current flows across the interface. By the inverse spin Hall effect, this is picked up as a voltage drop across the metal. In the FDT method, we compute the interfacial spin current up to an overall phenomenological parameter called the (dissipative part of the) spin-mixing conductance. An alternative method, used for Schwinger bosons, is to compute the interfacial spin current directly using Fermi’s golden rule. In Chapter 3, we summarize and analyze our final results for the spin Seebeck coefficients in FMs, AFs, PMs, and discuss a novel SSE which is relevant at temperatures below where electronic spin dynamics freeze out – the nuclear SSE. In Chapter 4, we compare our results to experiments on the AF SSE in chromium oxide (Cr2O3), the paramagnetic SSE in gadolinium gallium garnet (GGG), and the nuclear SSE in manganese carbonate (MnCO3). We discovered remarkable quantitative agreement between our theory and the Cr2O3 data after comparing the SSE across a metamagnetic phase transition. Here, the overall thermal and electronic transport properties such as thermal conductivities and metallic resistivities which affect the measured SSE were eliminated in this comparison, because they are unaffected by the magnetic configuration of the AF. We then applied this technique to the paramagnetic and nuclear SSE analysis to extract additional, specific information from the overall magnetic field profile of the SSE. Finally, we conclude with predictions from SBMFT for future experimental proposals. These investigate the strength of paramagnetic fluctuations in ordered magnetic phases and signatures of magnetic correlations in disordered phases.