1998 Annual Report
Grand Challenge Projects
Research Objectives
To develop first-principles quantum mechanical methods for addressing
materials problems microscopically, especially the relationship
between technical magnetic properties and microstructure. Towards
this goal are major problems involving microstructure (independent
of magnetism), magnetism (independent of microstructure), giant
magneto-resistance, and thermal properties. Computational Approach
A number of different first-principles techniques, including tight-binding
molecular dynamics (TBMD), an iterative pseudopotential (IP) method,
and the locally self-consistent multiple scattering (LSMS) method,
are used to perform fundamental studies of the atomistic, electronic,
and magnetic structure of microstructural defects in metals and
semiconductors that involve the interactions between large numbers
of atoms (TBMD 20,000 atoms, IP > 200, LSMS 250 to 3000 atoms).
In addition, we are developing spin dynamics based on both model
Hamilitonians and (local) spin density calculations as a fundamental
theory of the magnetic properties of metals and alloys. Accomplishments
A new constrained local moment (CLM) theory of non-equilibrium
states in metallic magnets has been developed that places a recent
proposal of our co-workers at Ames Laboratory for first-principles
spin dynamics (SD) on firm theoretical foundations. In SD, non-equilibrium
"local moments" (for example, in magnets above the Curie
temperature, or in the presence of an external field) evolve from
one time step to the next according to a classical equation of
motion. As originally formulated, the instantaneous magnetization
states that are being evolved were not properly defined within
density functional theory. The CLM theory properly formulates
SD within constrained density functional theory. Local constraining
fields are introduced, the purpose of which is to force the local
moments to point in directions required at a particular time step
of SD. A general algorithm for finding the constraining fields
has been developed.
The existence of CLM states has been demonstrated by performing
calculations for large (up to 1458 atom) unit cell disordered
local moment (DLM) models of bcc Fe above its Curie temperature.
This DLM model can be considered prototypical of the state of
magnetic order at a particular step in a finite temperature SD
simulation of paramagnetic Fe. Figure 1 illustrates the calculated
magnetic moments (arrows in the left frame) and constraining fields
(spikes in the right frame) corresponding to a 512-atom DLM model
of paramagnetic bcc Fe. Magnitudes of magnetic moments and constraining
fields are color coded. These calculations represent significant
progress towards the goal of full implementation of SD and a first-principles
theory of the finite temperature and non-equilibrium properties
of magnetic materials.
The record-setting computational performance we obtained during
our work demonstrating the existence of CLM states for large unit
cell models (up to 1024 atoms) resulted in our receiving the 1998
Gordon Bell Award for parallel processing applications. The calculations
that resulted in our nomination were performed using the LSMS
method extended to treat CLM states. The basic LSMS method is
an O(N) local density approximation (LDA) method that was specifically
designed for implementation on MPPs and for treating the quantum
mechanical interactions between large numbers of atoms. The constrained
non-collinear LSMS code showed near linear scale-up for system
sizes from 2 to 1024 atoms per unit cell, utilizing 2 to 1024
processing elements (PEs). We obtained a performance of 657 gigaflops
on a Cray T3E-1200 LC1024/512 at a U.S. government site, and 276
and 329 Gflops on T3E-900 and T3E-1200 LC512 machines at NERSC
and Cray Research, respectively. After our award nomination, we
obtained a performance of 1.02 teraflops during a CLM calculation
of a 1458 atom/unit cell CLM state of paramagnetic Fe (a 9x9x9
repeat of the underlying bcc unit cell). The calculations were
performed on a 1,480-processor Cray T3E at the manufacturer's
facility in Minnesota. The 7x7 reconstruction of the Si(111) surface is one of the most fascinating surface structures. The presence of adatoms is a unique feature of the 7x7 reconstruction as compared to other reconstructed structures on the semiconductor surface. |
Figure 2. Calculated STM images of the adatom vacancy on the
Si(111)-(7x7) surface in the energy window of 0.6 to 0.1 eV below the Fermi
energy level. Significance
The availability of powerful and accurate first-principles techniques
permits the study of quantum interatomic interactions on a length
scale not previously accessible, opening up the possibility of
relating these fundamental interatomic interactions to the strength,
ductility, transport and magnetic properties of materials. Applied
to magnetic materials, these techniques should help establish
the foundations for understanding the relationship between the
technical magnetic properties (permeability, coercivity, remenance)
of magnets and microstructure. Publications
B. Ujfalussy, Xindong Wang, Xiaoguang Zhang, D. M. C. Nicholson,
W. A. Shelton, G. M. Stocks, A. Canning, and B. L. Gyorffy, "High
performance first principles method for complex magnetic properties,"
in Proceedings of the ACM/IEEE SC98 Conference, Orlando, Florida,
November 7-13 (IEEE Computer Society, Los Alamitos, CA 90720-1264),
CD-ROM (1998).
M. Alatalo, M. Weinert, and R. E. Watson, "Stability of Zr-Al
alloys," Phys. Rev. B 57, R2009-R2012 (1998).
K. M. Ho, A. Shvartsburg, B. C. Pan, Z. Y. Lu, C. Z. Wang, J.
Wacker, J. L. Fye, and M. F. Jarrold, "Structures of medium-sized
silicon clusters," Nature 392, 582 (1998).
|
INDEX | NEXT >> |