|
|
|
|
|
|
 |
The illustrations show results of a calculation of the ground state (T=0K) magnetic structure of a large cell (256-atom) model of Fe65Ni35. In order not to prejudice the outcome, the magnetic moments associated with the Fe (big moment/big arrows) and Ni (small moment/small arrows) sites are initialized to point in random directions. As the calculation proceeds (left to right) most of the moments align ferromagnetically. However, some Fe moments align antiferromagnetically (anti-parallel) and some remain non-collinear. This is particularly apparent in the "rose" plots (lower frames) where the moments associated with all the Fe(Ni)-sites are projected onto a common Fe(Ni)-origin. |
A project involving cooperative research among scientists at several National Laboratories uses high performance computing platforms such as the Intel Paragon XP/S 150. A massively parallel code for studying magnetism in metals and alloys has been developed based on the formulation of ab initio spin dynamics and a scalable method for performing large scale first-principles electronic structure calculations of solids (100-1000 atoms). The new method has the potential to allow investigation of a wide range of magnetic properties previously not accessible to ab initio theoretical study. Preliminary investigations of the ground state magnetic structure of large cell models (256 atoms) of disordered Fe65Ni35 alloys (the classic Invar system) are pointing to unusual magnetic behavior (see caption above) involving non-collinear arrangements of the local magnetic moments associated with Fe-rich clusters within the otherwise disordered alloy.
Magnetic materials represent a multi billion dollar industry. Despite this great driving force, a microscopic understanding of metallic magnetism has proved to be elusive. Although much progress has been made in understanding magnetism in the 3-D transition metals using first-principles quantum mechanical methods based on the local spin density approximation, many significant scientific and technologically important problems remain unsolved. These range from a first-principles theory of the magnetic phase transition, to magnetism in inhomogeneous materials, to a microscopic theory of technologically important extrinsic properties such as permeability, coercivity and remanence. The methods that are being developed, coupled with the availability of massively parallel computers, will provide the necessary tools to perform realistic simulations on many of these problems, and have the potential to enable significant advances in magnetic materials research and development.
 |
 |