Spiking neurons and device synchronization for spintronic neuromorphic computing

The main goal of this thesis is the achievement of further understanding on various hot topics related to Spintronics and neuromorphic computing. Spintronics, also known as spin electronics, differs from traditional electronics in that, in addition to charge state, electron spins are used as a further degree of freedom. The connection between electron charge and spin admits changing the electronic transport by spins, and, on the contrary, to alter the magnetic properties by electron charges. Thus, nowadays, spin is fundamental for some of our technologies due to their interesting properties: nanometer dimension, low energy consumption, non-volatility, high scalability, and large speed. Various spin textures and devices have emerged and were deeply studied for the evolution of spintronic applications. Here we can mention the magnetic skyrmions and vortices where both exhibit intriguing and novel phenomena due to their topologically non-trivial spin textures, making them attractive for applications in spintronic devices. Moreover, the transformation of these 2D solitons into the 3D space, results in 3D magnetic solitons specifically known as “hopfions”. Recent studies concerning the dynamics of hopfions showed its promise in the field of magnetic data storage, topological photonics, and novel magnetic materials for spintronics applications. Alternatively, the non-volatility, high speed, and power efficiency of magnetic tunnel junctions, makes them primarily used as key components in magnetic memory applications. While both skyrmions and MTJs are important in spintronics, they serve different purposes and operate based on different principles. Skyrmions are topologically stable magnetic textures with potential applications in various spintronic devices, while MTJs are structures used primarily in magnetic memory and sensing applications. On the other hand, Neuromorphic computing has emerged as an alternative for conventional computing based on Von-Neumann architecture that suffers from various bottlenecks. Neuromorphic computing is meant to efficiently deal with the massive amount of data and computations in ubiquitous automobiles and portable edge devices. Such systems are based on a set of artificial neural networks (ANNs) with the most bio-realistic third generation neural network known as, spiking neural network (SNN). Spiking Neural Networks are highly power-efficient and have competitive capabilities to deal with numerous cognitive tasks. Spintronic based neuromorphic computing is an emerging field, holding promises for future technology. The main contributions of this thesis to the first topic have been about the different chiral spin textures characterized by a non-uniform distribution of the magnetization, including skyrmions and vortices. They have found a widespread range of applications because they can be easily nucleated, moved, and shifted by spin polarized current. The center of our attraction is the dynamics of these textures driven by Dzyaloshinskii-Moriya Interaction (DMI). We carried out a theoretical study based on micromagnetic simulations, in absence of thermal fluctuations. Our results show that under the influence of linear DMI gradients, Néel and Bloch-type skyrmions and radial vortex exhibit motion with finite skyrmion Hall angle, while the circular vortex undergoes expulsion dynamics. We provided a deeper and crucial understanding of the stability and gradient-driven dynamics of magnetic solitons and paved the way for the design of alternative low-power sources of magnetization manipulation in the emerging field of 2d materials. The second topic was about magnetic tunnel junction (MTJ) neuron, where MTJ is considered as a major spintronic device, composed of two ferromagnets separated by an insulating material. The designed MTJ neuron performs firing for spiking neural networks without the need of any resetting procedure. We leverage two physics, magnetism, and thermal effects, to obtain a bio-realistic spiking behavior equivalent to the Huxley-Hodgkin model of the neuron. Numerical simulations using experimental-based parameters demonstrate firing frequency in the MHz to GHz range under constant input at room temperature.