to develop techniques for fabricating
nanostructures and measuring their electronic and magnetic
properties.
INTENDED OUTCOME AND
BACKGROUND
The intended outcome of this work is the continuous improvement
of methods for fabricating and characterizing nanometer-scale
electronic and magnetic structures, as required to meet current
and future needs of the semiconductor and data storage industries.
Our main tools for pursuing this program are scanning electron
microscopy with polarization analysis (SEMPA) and the scanning
tunneling microscope (STM). The Electron Physics Group in the
Division has been a leading innovator in both of these methods,
which are outgrowths of work begun at NIST in the 1970s.
SEMPA enables us to use conventional scanning electron microscopy
(SEM) to image nanometerscale magnetic structure, through
spin-polarization analysis of secondary electrons ejected
from the sample. It has several unique capabilities that
distinguish it from other magnetic imaging techniques: it
is a highly sensitive, nonperturbative method, and thus
is especially well suited for in situ studies of
surface and nanostructure magnetization; it provides a
direct measurement of the magnetization of a material
region, rather than of a magnetic field; it has the high
spatial resolution (about 10 nm), long working
distance, and large depth-of-field characteristic of SEM;
and it facilitates simultaneous measurements of the
magnetization and the topography. SEMPA studies have led
to a number of breakthroughs in understanding the basic
mechanisms of magnetism on the micro- and nanoscale, and
have also addressed near-term measurement issues faced by
the magnetic data storage industry.
Our STM program is focused on understanding the electronic
and magnetic properties of nanostructures on surfaces.
In recent years the STM program has been particularly
concerned with the magnetic multilayer materials that
have been investigated by SEMPA. The complementarities
of SEMPA and STM measurements have elucidated many
connections between conditions of layer growth and magnetic
device performance.
The main, current direction of the STM program is
the course of research made possible by the recently
completed Nanoscale Physics Laboratory. This laboratory
permits us to measure quantum electronic structures with
atomic-scale imaging resolution and high electronenergy
resolution. Samples grown in situ can be measured
in an ultrahigh vacuum environment with magnetic fields
of up to 10 T at temperatures down to 2.3 K.
Additionally, a program in Autonomous Atom Assembly is
underway that will allow us to fabricate highly complex
and perfect nanostructures on demand.
Accomplishments
Understanding Single Atom Manipulation
| |
Figure 1. Binding site image for a cobalt atom
on a copper (111) surface, showing three-fold
symmetry. |
A low-temperature STM can move atoms about on clean, flat surfaces.
However, our limited understanding of the interactions involved
restricts our ability to make complex nanostructures.
We recently studied the atom dynamics that occur during atom
manipulation— important if we are to make real nanodevices on
technological substrates. We found that the atoms at the end
of an iridium STM tip create a potential trap sufficient to
capture a cobalt atom adsorbed on a clean copper crystal and
to limit its lateral motion to less than 0.2 nm. We also
discovered that the vibrational level of the cobalt-copper
bond could be varied dramatically by changing the electron
current tunneling into the cobalt atom.
By adjusting the proximity of the STM tip to the cobalt atom,
as well as the current flowing through it, it is possible to
translate the cobalt atom across the copper crystal surface
in a controlled and well-understood manner. It is even possible
to force the cobalt atom into what is normally not a binding
site on the surface. As the conditions necessary for forcing
the atom into a nonbinding site are approached, the atom begins
to jump back and forth between the nonbinding site and a
neighboring binding site, giving rise to audio frequency noise
in the tunneling current. This noise can be monitored in the
laboratory and correlates with the locations of the nonbinding
sites on the surface.
Using our newly acquired understanding of the atom manipulation
process, we were able to: obtain a direct observation of a
characterizable random two-state fluctuator, demonstrate the
ability to modify single atom surface dynamics, show how chemical
binding sites can be modified to change unstable ones to stable
and vice versa, provide the first explanation for the origin of
excess tunneling noise, and present a new type of STM image that
reveals the surface binding sites of the crystal. (See Fig. 1.)
This work is an example of a new class of measurements referred
to as "Atom-Based Metrology," where a single atom probes its local
environment while nanoscale techniques position it within that
environment. Here, the probe is the cobalt atom and the STM tip
moves it across the copper surface. The atom responds to the
surface potentials through its lateral motion, which is
reflected in the measured tunneling current. This allows
measurement of the binding site map of the copper surface
through observation of the dynamical response of a single
scanned atom to its environment. We are planning additional
experiments using atombased metrology.
Geometrically Constrained
Magnetic Nanostructures
| |
Figure 2. SEM image of a silicon nanopillar,
along with a SEMPA image of the epitaxial,
cobalt thin-film disk at the top of the pillar.
The SEMPA measurement shows a geometrically
constrained domain wall whose width varies
with radial distance. |
Small, patterned, magnetic thin-film structures play a critical
role in emerging magnetoelectronic technologies, which range from
nonvolatile random access memories to magnetic field sensors. One
potential advantage of miniaturizing these devices is that the
size and geometry of the patterned structure may be used to shape
the internal magnetic nanostructure and thus control the device operation.
Imaging this internal nanostructure is a measurement challenge,
since the structures consist of only a small amount of magnetic
material and can easily be perturbed by conventional magnetic
imaging methods such as Magnetic Force Microscopy (MFM).
In collaboration with the University of Cambridge Thin-Film Magnetism
Group we have used our SEMPA facility to noninvasively image the
magnetic nanostructure of epitaxially grown thinfilm Co disks.
Isolated, micrometer-sized Co disks were grown by depositing a
several-nanometer-thick Co film onto a Si substrate with etched
circular pillars. Unlike bulk Co, these epitaxially grown Co thin
films have a cubic crystal structure that leads to a four-quadrant,
closed-flux domain structure, characteristic of a film with cubic
anisotropy.
The SEMPA images (e.g., Fig. 2) showed how the internal magnetic
nanostructure of these disks depends critically on the size and
shape of the element. Magnetic domain wall widths, for example,
do not depend on sample geometry in bulk ferromagnets. But in
these structures the domain walls are geometrically constrained
by the significant shape-related magnetostatic energies, and the
wall widths vary dramatically with radial distance from the
magnetic vortex core.
Imaging Magnetic Sensors| |
Figure 3. SEMPA image of a zigzag magnetic
sensor element, compared with the OOMMF
micromagnetics simulation. Line scans through
the centers of the elements show the magnitude
of the magnetic oscillations. |
Recent advances in thin-film and multilayer magnetism offer the
exciting possibility of creating a new generation of magnetic
sensors that are small, inexpensive, and as sensitive as currently
used, more cumbersome sensors such as SQUIDs. Development of this
new generation of sensors requires understanding and control of
the magnetic nano-structure, in order to enhance sensitivity and
eliminate potential sources of magnetic noise.
We are working with NIST magnetics groups in EEEL, MSEL, and ITL
to test and measure possible new materials and devices for use
as sensors. One potentially useful device is a zigzag-shaped
magnetic sensor developed by EEEL. The device’s magnetic sensitivity
is based on anisotropic magnetoresistance (AMR), and therefore
requires the magnetization to be tilted with respect to the current
passing through the device. In the zigzag sensor, the shape
anisotropy forces the magnetization to tilt relative to the current.
We have used SEMPA to measure how well various zigzag geometries
accomplish this angular biasing and to look for potential sources
of magnetic noise. (See Fig. 3.) SEMPA imaging is especially
useful since it not only images the magnetization directly, it
provides quantitative measurement of magnetization directions
that can be compared to the results of micromagnetic calculations.
The SEMPA images also reveal potential magnetic trouble spots such
as trapped magnetic singularities, which can interfere with
reproducible magnetic response, as well as defects that can
produce magnetization-related noise. Work is currently underway
to allow imaging of electrically active magnetic sensors, which
will allow a better understanding of noise sources and failure
modes in real devices.
CONTACT:
Dr. John Unguris
(301) 975-3712
john.unguris@nist.gov
Current Induced Magnetic Switching
| |
Figure 4. Electron diffusing near a ferromagnetic
interface. The spin of the electron diffusing in
the copper has its direction changed when it
scatters from a cobalt interface with an in-plane
spin wave. |
In a magnetic multilayer, changing the relative orientation of
the magnetizations of two layers changes the current flowing
through the multilayer. This effect, called giant magnetoresistance,
is used in magnetic sensors, magnetic random access memory (MRAM),
and read heads in magnetic disk drives. An inverse effect, where
the current changes the magnetic configuration, is being studied
for possible applications. In this case, a current passing through
a multilayer exerts a torque on the magnetizations of different
layers causing the magnetizations to rotate.
Large enough currents passing through such multilayers can switch
the magnetizations of the layers between parallel and antiparallel,
or cause one to rapidly precess around the other layer. If the size
of the current necessary to reverse magnetizations can be reduced
sufficiently, spin-transfer torques could provide a way to switch
bits in MRAM. In other devices, it should be possible to make
current-controlled oscillators.
We have developed quantitative models for these spin-transfer
torques to help the development of devices based on this effect.
(See Fig. 4.) In particular, we recently developed an analytic
formula for the torque as a function of the device geometry and the
magnetic configuration. Such a formula will allow rapid simulation
of prototypes for device optimization.
We have also shown how spin-transfer torques can cause a magnetic
instability even with a single ferromagnetic layer. This effect was
first seen in point-contact experiments and later in more device-like,
lithographically fabricated nanopillars. We explained how in-plane
inhomogeneities in the magnetization lead to the observed
instabilities. We are presently studying the effect of these
inhomogeneities on more typical device geometries.
End States in One-Dimensional Chains
| |
Figure 5. STM images of one-dimensional atom
chains on Silicon, showing empty states (upper
panel) and filled states (lower panel). End
states cause the chains to appear shorter in the
upper panel. |
The abilities to fabricate structures on the nanoscale and to measure
their electronic properties present the opportunity to discover new
phenomena. We have observed a new kind of electronic state at the
ends of a one-dimensional nanostructure. This "end state" is a
direct consequence of the lower dimensionality of the structure. Such end
states can be thought of as zero-dimensional analogs to two-dimensional
states that occur at the surface of a crystal.
In fabricating these structures, we exploit the self-assembly of
atom chains that occurs when we deposit gold on stepped silicon
surfaces at elevated temperatures. Rows of atom chains of varying
length can be seen in the STM images of Fig. 5, which are of the
same area but measured at opposite polarity. The chains in the
upper panel, measured at a sample potential of +0.5 V, appear
shorter than the chains in the lower panel, measured at a sample
potential of -1.0 V, as emphasized by the white box around a
seven-atom chain. Such a polarity contrast in STM suggests an
underlying difference in the density of states for the empty and
filled states near the ends of the chains, indicating the presence
of end states.
To characterize this new kind of electronic state, we made
spatially resolved scanning tunneling spectroscopy measurements
along finite chains to map the density of states. These
measurements reveal quantized states that form in isolated chain
segments. Furthermore, a transfer of spectral weight from the
filled to the empty states over the atoms at the ends of the
chains is directly attributable to the formation of end states.
These end states cannot be described by a simple particle-in-a-box
model for states along the chains. A comparison to a tight-binding
model demonstrates how the formation of end electronic states
transforms the density of states and the quantized levels within
the chains. The end states effectively lower the energy levels of
the filled states within the chains, suggesting a possible driving
force for their formation. As a further confirmation of the
tight-binding model and the end electronic effects, we calculated
STM topography profiles at positive and negative biases and
reproduced the experimentally observed contrast at the end atoms
in Fig. 5.
First strategic focus |
Second strategic focus |
Third strategic focus
"Technical Activities 2004" - Table of Contents |
