Abstract
The magnetic insulator Yttrium Iron Garnet (YIG) can be grown with exceptional quality, has a ferrimagnetic transition temperature of nearly 600 K, and is used in microwave and spintronic devices that can operate at room temperature. The most accurate prior measurements of the magnon spectrum date back nearly 40 years, but cover only 3 of the lowest energy modes out of 20 distinct magnon branches. Here we have used time-of-flight inelastic neutron scattering to measure the full magnon spectrum throughout the Brillouin zone. We find that the existing model of the excitation spectrum, well known from an earlier work titled "The Saga of YIG", fails to describe the optical magnon modes. Using a very general spin Hamiltonian, we show that the magnetic interactions are both longer-ranged and more complex than was previously understood. The results provide the basis for accurate microscopic models of the finite temperature magnetic properties of Yttrium Iron Garnet, necessary for next-generation electronic devices.
Introduction
Yttrium Iron Garnet (YIG) is a cornerstone of microwave technology and the emerging field of spintronics, prized for having the lowest known magnon damping. Despite its ubiquity, our theoretical understanding of its magnetic excitations has remained incomplete for decades. While YIG possesses 20 distinct branches of magnon excitations, historical measurements captured only the three lowest-energy modes. The standard theoretical framework, famously dubbed "The Saga of YIG," has long struggled to accurately describe the higher-energy "optical" modes. Without a complete map of these excitations, engineers cannot fully model device behavior at room temperature, where high-energy modes significantly influence phenomena like the spin-Seebeck effect.
Experimental Methodology
To address this gap, researchers employed time-of-flight inelastic neutron scattering (INS) at the ISIS Facility’s MAPS spectrometer. Unlike traditional triple-axis methods that sample specific points, this technique captured a massive 4D hypervolume of data (~100 GB) covering frequency and momentum across the entire Brillouin zone. The study utilized a high-quality, 12-gram YIG single crystal grown via the flux method.
Data interpretation was performed using a Heisenberg effective spin Hamiltonian and the SpinW software package. Crucially, the model was fitted to the data in absolute units (millibarns per steradian per meV per formula unit), providing a rigorous test of physical accuracy without the use of arbitrary scaling factors.
Revising the Magnetic Interaction Model
The investigation revealed that the magnetic interactions in YIG are significantly more complex and longer-ranged than previously believed. A key discovery was that the third-nearest-neighbor exchange ($J_3$) consists of two distinct interactions, $J_{3a}$ and $J_{3b}$. Although the physical distance between the atoms is identical, differing oxygen-mediated pathways lead to distinct magnetic strengths.
Furthermore, the second-nearest-neighbor interaction ($J_2$), previously estimated at 2.3 meV, was found to be much weaker (0.52 meV). To achieve an accurate fit, the model had to include isotropic exchange interactions up to the sixth nearest neighbor ($J_6$), proving that magnetic coupling in YIG has a much longer reach than the simple nearest-neighbor models typically found in literature.
Results: Mapping the Full Magnon Spectrum
For the first time, all 20 magnon branches were measured up to energies of nearly 100 meV. The new model reproduces the entire measured spectrum and absolute intensities with exceptional fidelity, establishing a new "gold standard" for YIG physics.
The data also clarified the limits of parabolicity for the lowest-energy "ferromagnetic" mode. While often assumed to be a simple parabola, it deviates after only 14.8% of the way to the zone boundary. Additionally, the absence of a predicted "flat" mode at ~35 meV necessitates a total rethink of the optical mode distribution, which directly impacts calculations for the spin-Seebeck effect and thermal energy harvesting.
Conclusion and Future Outlook
While this work provides a comprehensive map of YIG's magnons, some limitations remain. The instrumental resolution of ~3 meV may hide subtle magnon-phonon or magnon-magnon couplings. Future work could refine the model by considering trigonal symmetry instead of cubic, potentially matching bulk magnetometry more perfectly. Most importantly, this microscopic model provides the foundation for a "multi-magnon picture" to fully capture YIG's dynamical magnetic properties at the elevated temperatures required for practical electronic devices.
Summary updated