Atomistic spin dynamics

The project has the overall objective to study magnetization dynamics on atomic scale using a combination of electronic structure methods and atomistic simulations. Of particular interest are materials aimed for future applications in information technology, like memory devices using spin-transfer torque and nanomagnets for storage.

Magnetic materials are a very important class of materials in industry and modern technology. They are the basis for many electrical and electromechanical devices such as electromagnets, electric motors, generators, transformers, magnetic storage such as hard disks. If magnetism and electronics are simultaneously combined to create novel applications it is typically referred as spintronics. The basic principle in many of these applications is to change the orientation of magnetic moments with an external source that in turn affects the electrical properties of the system. The most successful application of spintronics so far is read heads in hard drives based on Giant Magneto Resistance (GMR), an effect that was awarded Nobel Prize in Physics 2007. It is expected that spintronics will play an even more important role in the future since the dimensions of the devices continues to shrink making it difficult to continue with the semiconductor route that has been employed for decades.  


In collaboration with Uppsala University, we have developed a reliable theoretical framework for studies of magnetization dynamics. It is based on a two step approach. In first step we perform electronic structure calculations based on density functional theory of the system of interest to extract relevant material parameters such as magnetic moment, anisotropies and magnetic exchange interactions. These material specific parameters are used to set up a simplified yet accurate model (extended Heisenberg model) describing the magnetism in the system. In the second step, we use this model to perform finite temperature atomistic spin dynamic studies by solving the equation of motion (Landau-Lifshitz-Gilbert) for each atomic moment as implemented in the UppASD program.


Recently, we adapted the program to massive parallel computers by implemeting sophisticated domain decomposition techniques where all communications are overlapped with computations. This allows for studies of very large systems consisting of billions of atoms. We have found excellent scaling up to at least 13 824 processors on the Lindgren computer on PDC (Cray XE6) and micrometer length scales are within reach.  


Some selected topics that we study using this methodology:


  • Finite temperature magnetism, in particular magnon dispersion 
  • Switching dynamics for next-generation memory devices (STT-MRAM)

  • Stability and switching of nanomagnets (wires, clusters, nanostructures etc)

  • Magnetic response due to thermal gradients

Project link


PI: Lars Bergqvist
Member: Fan Pan