Lee, Oscar (Sang Jin); (2023) Nonlinear magnonics and their applications for unconventional computing. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text OscarLee_PhD_Thesis_LibrarySubmit.pdf - Other Download (97MB) | Preview

Abstract

Nonlinear magnonics is a study using quantised magnetic excitations known as magnons for information processing. In the first study, strongly coupled photon-magnon hybrid quasiparticles, magnon-polaritons, are experimentally explored in nonlinear excitation regimes at the limit of first-order Suhl instability. A strong coupling between the cavity photons and the Kittel magnons is quantified by the gap size of the avoided crossing, which fully closes when bringing the system into nonlinearity. Observations are accompanied by a theoretical model attributing the suppression of the Kittel mode amplitude by the Suhl instability that effectively decouples the magnon and photon systems. GHz dynamics of nontrivial spin textures, i.e., magnetic skyrmions, are investigated in the second study. Variants of magnetic field cycling protocol are utilised to nucleate metastable unconventional low-temperature skyrmions (LTS). Experimental results highlight that the LTS phase exhibits a nonlinear population growth, with strong dependencies on the number of applied field cycles. Such tunability is not available in the conventional skyrmions phase and provides insights for their potential use in unconventional computing. Furthermore, the temperature-dependent analysis reveals that the predominant stabilisation mechanisms responsible for the LTS phase stem from the cubic anisotropy of the magnetic crystals. In the third study, the above discoveries are further utilised in different magnetic phase spaces to demonstrate task-adaptive physical reservoir computing (PRC). PRC is an unconventional computing technique facilitating the output response phenomena of a physical system as recurrent neural networks. The choice of magnetic phase spaces accommodates distinctive computing properties of the physical reservoir, thus, computational performance. Magnetic skyrmions excel in forecasting (prediction) tasks with strong history-dependent memory properties. In contrast, conical/ferromagnetic magnetic phase reservoirs provide high nonlinearity to the system, enhancing transformative tasks by multiple orders of magnitude. Such techniques are transferable to various chiral magnets and are versatile for near and above-room-temperature operations.

Download activity - last month

Export as JSONCSVXML

Download activity - last 12 months

Export as JSONXMLCSV

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item