Dinâmica de spins em nanoestruturas metálicas

• Autor(a) principal: Guimarães, Filipe Souza Mendes
• Data de Publicação: 2011
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Federal Fluminense (RIUFF)
• Texto Completo: https://app.uff.br/riuff/handle/1/19126
• topic: Magnetismo; Nanotubos de carbono; Nanoestrutura; Spin; CNPQ::CIENCIAS EXATAS E DA TERRA::FISICA

In the present work, we investigate the spin pumping mechanism. In 2002, Tserkovnyak et al. proposed that a precessing magnetization of a magnetic unit coupled to a non-magnetic metal can transfer angular momentum to the conduction electrons, creating a spin flow that propagates across the non-magnetic metal [1]. This flux of angular momentum, or spin current, contributes to the damping of the precessing magnetization and propagates without a net charge current. The spin current can be used to excite a second magnetic unit, far away from the source, transporting information through conductors. To study this phenomenon, we developed a fully quantum-mechanical approach, based on linear response theory, to calculate the expected value of the spin current emitted by the precession of a magnetization of a magnetic unit in contact with a non-magnetic metal. We showed that this quantity is related to generalized dynamical transverse magnetic susceptibilities. In this work, we also detail the semi-classical theory developed by Tserkovnyal et al. [1], and compare it with our formulation. To illustrate this comparison quantitatively, we investigate some relatively simple systems. An excellent agreement between the two theories is revealed, considering the differences between the approaches: while our theory is completely based on quantum mechanics to obtain the spin current that emanates from a precessing magnetization, the semi-classical theory is based on scattering matrices and makes use of the adiabatic approximation to calculate the spin current pumped inside the contacts, away from the magnetic unit. Moreover, we calculate the spatial distribution of the spin currents, as a function of frequency and time, for magnetic impurities embedded in unidimensional systems, and we show that quantum interferences play a central role in this phenomenon. We also use our theory to study the propagation of spin currents in carbon-based nanostructures. We show that carbon nanotubes are able to carry information stored in a precessing magnetic moment for long distances with very little dispersion and with tunable degrees of attenuation. These systems, known to function as conduits for electrons and for phonons, are also efficient spin-current waveguides. Pulsed magnetic excitations are predicted to travel with the nanotube Fermi velocity and are able to induce similar excitations in remote locations [2]. In addition to the perturbation in carbon nanotubes, we also demonstrate that graphene can function as gate-controllable transistors for pumped spin currents. Furthermore, we propose as a proof of concept how these spin currents can be modulated by an electrostatic gate [3]. Because our proposal involves nano-sized systems that function with very high speeds and in the absence of any applied bias, it is potentially useful for the development of transistors capable of combining large processing speeds, enhanced integration and extremely low power consumption. As the spin current tends to travel omni-directionally, a large fraction of this information never reaches the probe and is lost. We propose, in analogy to optics systems, that a curved boundary between a gated and a non-gated region within graphene acts as an ideal lens for spin currents despite being entirely of non-magnetic nature. We show as a proof of concept that such lenses can be utilized to redirect the spin current that travels away from a source onto a focus region where a magnetic probe is located, saving a considerable fraction of the magnetic information that would be otherwise lost.