For this thesis a scalable parallel finite element micromagnetics package has been developed. It is entirely based on free open source software packages, which have been selected for performance, scalability, portability, and ease of use. The combination of static energy minimization, time integration, and the nudged elastic band method makes it a very efficient and versatile tool, which has been used for the investigation of magnetic nanostructures.
In SmCo precipitation hardened magnets the coercivity is dominated by domain wall pinning on the precipitation structure. The detailed study of the pinning process has shown the influence of the material properties, the intercellular phase, and the cell geometry. Exchange decoupling can considerably improve the coercivity, but the thickness of the intercellular phase has to be in the optimum range between one and four times the domain wall width.
Magnetic nanoparticles of FePt are possible candidates for future magnetic storage media. The dependence of the nucleation and coercive fields on the distribution of easy axes has been investigated. It has been found, that a single misaligned axis reduces the coercivity by a factor of three. The coercivity is further decreased if more misaligned axes are assumed and if the particle size gets in the order of the domain wall width.
Another interesting system are magnetic nanodots, which exhibit curling magnetic structures (vortices). The calculation of their static properties shows the competition between exchange and magnetostatic energy. A phase diagram of the magnetic ground states has been obtained and compared with analytical and experimental investigations. The dynamic properties, which are important for high-speed, high-frequency applications, of vortex precession and radial modes have also been studied.
The behavior of chains of soft magnetic elliptical nanoparticles is also strongly influenced by magnetostatic interactions. These lead to antiparallel magnetization distributions, which remain stable even in the presence of exchange coupling between the particles.
In summary, the parallel micromagnetics code has proved its performance and scalability and provided insight into the magnetization reversal processes of magnetic nanostructures. The different solvers have been applied to different problems of magnetic domain wall pinning, nucleation, and magnetization dynamics. The high spatial resolution, which is required in hard magnetic materials like SmCo and FePt with very thin domain walls, could be achieved owing to the parallelization, data distribution, and efficient energy minimization method. The magnetization dynamics have been studied with a preconditioned time integration method, which allows large time steps and long integration times. Improved experimental investigations with higher spatial resolution will be necessary to obtain more accurate material parameters for the simulation of permanent magnets, and a higher time resolution is also required for the detailed investigation of the magnetization dynamics in magnetic nanoelements.
Still, many interesting effects have been neglected, such as thermal activation and the influence of eddy currents, for example. Efficient time integration methods for Langevin micromagnetic simulations including thermal activation are an active research area [138]. However, new solvers can be easily added to the program to consider also thermal effects. Eddy currents are often calculated using a magnetic vector potential [30]. A module, which calculates this vector potential and the resulting eddy currents and demagnetizing field, could replace the current implementation of the hybrid finite element/boundary element method.
The ever increasing computing power and availability of large clusters of workstations and parallel machines will support these developments and allow even larger systems to be simulated. A challenging problem, for example, is the simulation of magnetic hard disk media including the read/write head and their interaction [139]. High-frequency magnetization reversal during the writing process should also include thermal and eddy current effects. All these aspects are very important for the design and development of perpendicular recording media, and the micromagnetic package developed for this thesis can be a starting point for such a comprehensive micromagnetic model.