Computational materials scientists at ORNL are using the high-performance computing infrastructure at CCS to explore superconductivity and magnetic nanostructures.
![Click image for a larger view.](graphics/article06_mats_corral.jpg)
Using a CCS supercomputer, researchers calculated the magnetic structure
of a quantum corral nanostructure, which consists of magnetic
iron atoms deposited on a copper surface that "corral" copper
electrons.
|
|
The 1986
discovery of high-temperature superconductivity sparked the quest for
room-temperature superconductors that could transmit electrical current
without heat losses and without the need for an expensive coolant such
as liquid helium. Room-temperature superconductors could make possible
ultra-efficient power transmission lines, practical electric cars,
and superconducting magnets that could bring high-speed levitated trains
and smaller, more efficient, and less costly rotating machinery, appliances,
particle accelerators, electric generators, and medical imaging devices.
High-temperature superconductors are being used commercially. A few
urban utility companies have tripled their capacity to carry power
simply by replacing existing underground cables with liquid-nitrogen-cooled
superconducting cables. Cellular telephone towers have extended their
reception range and call-handling ability with superconducting signal
filters.
Understanding Superconductivity
No one understands why certain copper-oxide materials exhibit high-tem-perature
superconductivity, says ORNL Corporate Fellow Malcolm Stocks, co-leader
of the Materials Research Institute (MRI) at DOE's Center for Computational
Sciences at ORNL. Recently, MRI co-leader Thomas Schulthess has been
collaborating with Thomas Maier, Eugene P. Wigner Fellow at CCS, and
solid-state physicist Mark Jarrell of the University of Cincinnati.
|
![Click image for larger view.](graphics/article06_mats_maier.jpg)
Thomas Maier, recently named a Eugene P. Wigner Fellow at ORNL, used the Cray X1
supercomputer at CCS to confirm his suspicion that a widely used computer model was
insufficient to describe high-temperature superconductivity.
|
When Maier was a postdoctoral fellow at Cincinnati he, Jarrell, and other
colleagues were working on a theoretical understanding of high-temperature
superconductivity. Jarrell holds that a microscopic understanding of
why the current layered oxide (high-Tc) materials are superconducting
will lead to the design and synthesis of new room-tempera-ture superconductors.
Jarrell
is a co-developer of a new theoretical approach, called the Dynamical
Cluster Approximation (DCA). Here, a cluster of atoms, within which
the complex quantum mechanical interaction between electrons are treated
essentially exactly, is embedded into an effective medium that accounts
for the effects of the rest of the material in a computationally feasible
way without compromising the mathematical rigor of the theory.
Using a massively parallel implementation on the supercomputers
at CCS and at Pittsburgh Supercomputing Center, Jarrell and Maier studied
the two-dimen-sional (2-D) Hubbard model. This model has been widely
accepted as the theoretical framework for capturing the physics
underlying high-T c superconductivity. Because structurally the
high-T c materials are a series of copper-oxide planes, with no apparent
interactions between them, researchers believed they could be modeled
as 2-D systems.
However, based on their work, Jarrell and Maier began to suspect that
the 2-D Hubbard model may not provide a complete description of
high-T C superconductivity. When Maier came to ORNL as a Wigner
Fellow, he and his CCS colleagues were able to pursue this suspicion
using the power of the newly installed Cray X1, a machine ideally suited
to performing the complex DCA calculations. The most recent calculations
are confirming Maier's suspicions that the strictly 2-D Hubbard
model is insufficient, a conclusion that runs counter to most conventional
wisdom. If true it will be necessary to improve the underlying model
by, for example, introducing some 3-D features such as coupling
between the 2-D copper oxide planes.
The CCS team was able to carry out this work because the Cray X1 makes
the DCA calculations up to 35 times faster than the computers used previously.
This improved performance allows the modeling of larger DCA clusters
containing up to 64 (8 x 8) sites, rather than only 4 (2 x 2)
sites previously possible. According to Schulthess, if the Cray X1
had 4 to 10 times its current capability, the ORNL/Cin-cinnati collaboration
would likely have a model that describes the physics of high-temperature
superconductors, which eventually will enable computational design
and prediction of new materials that are superconducting at room
temperature.
|
![Click image for larger view.](graphics/article06_mat_poly.jpg)
The computed molecular structure of a nanoscale assembly of six polymer
nanoparticles.
|
Nanomagnetic
Modeling
The revolution
in magnetic storage that has occurred over the past 40 years has
always depended on the timely introduction of new materials that allow
researchers to cram increasing amounts of digital data into smaller
and smaller regions. In the very near future, whole new classes of
materials will be needed if the magnetic storage industry is to maintain
its current rate of increase in storage density for computers and digital
media. Increasingly, basic research is turning to nanoscience for solutions.
Nanoscience covers materials whose size is a few thousand to a few
million atoms. At the nanoscale, materials frequently exhibit new
and unexpected phenomena. Understanding and harnessing these "emergent" phenomena
could have untold future benefits to science and technology.
For some time Stocks, Schulthess, and others at CCS have been modeling
magnetism and electrical transport in magnetic materials as part
of the effort to develop new and improved magnetic materials. Now, the
group has also turned to studying the new
frontier of the properties of magnetic
nanostructures. Recently, using the IBM
supercomputers at CCS, the group performed
the first fully self-consistent ab initio
calculations of the magnetic structure
of scientifically fascinating nanostructures
called quantum corrals.
A quantum
corral is a magnetic nanostructure in which magnetic atoms,
are positioned in the shape of a stadium
on a copper surface. In the mid-1990s Don
Eigler of IBM used a scanning tunneling
microscope (STM) not only to make rings
of magnetic atoms but also to measure the
electron density, the concentration of the
copper surface's electrons that are corralled
by the magnetic atoms—hence, the
name "quantum corral." On a (111)-facet
of copper, some of the electrons form an
effective two-dimensional electron gas just
above the surface.
When the magnetic
atoms are deposited,
the 2-D electron gas
is perturbed, causing
standing waves that
are captured by STM
images.
The CCS researchers
have not
only calculated the
oscillations in the
surface charge obtained experimentally but have also predicted
oscillations in the magnetic density inside the corral as well as the
state of the magnetic atoms themselves. These predictions present
a challenge to next-genera-tion STMs that will measure the magnetic structure
of nanostructures. This achievement was made possible by use of
codes developed by the University of Tennessee's Balazs Ujfalussy
and his colleagues in Budapest and Vienna.
Materials
Institute
The rapid pace at which new complicated materials must be introduced
into technology, as well as the drive to miniaturize devices, increasingly
dictates a more systematic approach to design and fabrication that is
built on a firm theoretical underpinning. Here computation can be
expected to play a central role. Because simulation can study features
close to the atomic limit that are difficult to access experimentally,
computational materials science can accelerate the development of
new materials.
To facilitate
these goals, the Materials Research Institute has been created within
CCS. MRI's missions are to build and maintain a community of leading
computational materials scientists and to involve them in the development
of computational methods, algorithms, and simulation software, as well
as in the early evaluation of new computer hardware at CCS. MRI's long-term
goal is to enable the design of advanced materials by using leadership-class
computing envisioned by DOE.
|