Powered by Indico
v3.2.9
Exotic spin textures such as skyrmions, are proved to be promising candidates for the development of sustainable, next-generation spintronic devices. Despite extensive research in this domain, the quest for efficient computational methodologies for the automated discovery of novel functional magnetic materials, and identify the intricate topological spin textures they can host, remains a formidable challenge in solid-state physics. This thesis work introduces a promising end-to-end computational approach, employing the Heisenberg spin Hamiltonian model, aimed at overcoming this challenge and discovering novel exotic spin textures. Our approaches encompass an automated Density Functional Theory (DFT) calculation workflow designed to predict candidate functional magnetic materials for hosting spin textures, with those candidate material we calculate their magnetic exchange interactions for constructing the spin Hamiltonian. Subsequently, a computational workflow that integrates new developed metaheuristic optimization algorithms with Atomistic Spin Dynamics (ASD) simulations is employed to identify spin textures in targeted systems. Additionally, a post-processing tool for the visualization of these textures is presented.
In the computational approach part of this work, we have developed several tools, including the high-throughput workflow code and the scientific visualization software, dedicated to studying exotic spin textures. On the algorithmic front, we introduce the metaheuristic conditional neural network and a controlled assembly approach for investigating the long-lifetime metastable states of magnetic systems with long-range interactions. Through these novel approaches, we have identified and predicted the constructing pathways to several novel high-order antiskyrmions and three types of skyrmionic metamaterials (i.e., lattice-like, flake-like, and cell-like). Furthermore, we applied evolutionary algorithms for identifying the ground states of skyrmionic systems, developing both the genetic tunneling optimizer (GTO) and a neuroevolutionary algorithm.
In summary, the main results are:
1. Discovery and analysis of the forming mechanism of novel high-order antiskyrmions in the PdFeIr system.
2. Introduction of the evolutionary algorithm to the atomistic spin Hamiltonian model for the first time, combined with Markov chain Monte Carlo and atomistic spin dynamics simulations.
3. Prediction of skyrmions in 4d and 5d doped B20 systems.
4. Discovery and revealing the construction pathway of new skyrmionic textures, i.e., 2D skyrmionic metamaterials.
5. Development of a general visualization and post-processing code for computational magnetism, which offers new opportunities for analyzing complex spin textures.