Abstract
Neutron spectroscopy measurements have confirmed an earlier intimation of an 'anomalous' anisotropy in the dispersion of 'acoustic magnon like' excitations in yttrium iron garnet Y3Fe5O12. The present experiments have shown this anisotropy to be large and intricate, with energies at a fixed excitation wavelength 25 Å, varying from 5 to 20 meV. A contention that such anisotropy might arise from strong magnon-phonon coupling effects is not supported by experimental investigation of (i) a spin-reorientation transition in Tb0.3Y2.7Fe5O12 or (ii) applying variously directed magnetic fields to YIG itself. The experimental dispersion curves are replicated instead by linear spin-wave theory, with fitted values for all exchange interactions J1ad, J2ad, J3ad, J1aa, Δ1aa, J2aa, J3aa, J1dd, J2dd, J3dd, J4dd, J5dd closer than a/2∼6.2 Å. In this new model, the experimentally observed 'acoustic magnon' anisotropy is attributed to a complex coexistence of acoustic and low-energy optic magnon branches: ensuing computations suggest that possible 'soft modes' of such 'pseudo-acoustic' branches may relate to previously reported 'cantings' of the ferrimagnetic structure in various substituted garnets. The new model also helps in understanding some reported anomalies in existing magnetisation, specific heat and microwave measurements.