Major Technical Highlights

• Manipulating Magnetic Impurities. Researchers in the Electron Physics Group (EPG) have recently demonstrated the capability of manipulating individual cobalt atoms on a copper (111) surface (see fig. 1) and measuring their electronic properties on an individual-atom basis. This work is part of a study of the physics of magnetic impurities in non-magnetic host materials. Building magnetic nanostructures atom-by-atom allows EPG researchers to study the fundamental beginnings of magnetism at the nanometer level.

• A single magnetic impurity atom embedded in a non-magnetic host material represents the smallest magnetic structure and gives rise to the rich physics of the Kondo effect. At temperatures below the so-called Kondo temperature, the host conduction electrons condense into a many-body ground state that collectively screens the local spin of the magnetic impurity. This screening cloud displays a set of low-energy excitations at the Fermi-level known as the Kondo resonance.

  Atomically resolved measurements of the Kondo resonance are being performed on single magnetic impurities using a cryogenic scanning tunneling microscope system that is part of the Nanoscale Physics Facility in the Electron Physics Group. Building higher order magnetic clusters is achieved using atomic manipulation techniques being developed as part of the Nanoscale Physics Facility. Figure 1 shows a rendered scanning tunneling microscope (STM) image of the two cobalt monomers and two cobalt dimers (blue) that were positioned to form a square nanostructure pattern. The atomic manipulation was performed by forming a tunable chemical bond between the STM tip and the cobalt atoms, and then dragging them one-by-one to the desired location on the copper surface. The separate cobalt atoms in the dimers are not resolved, but appear higher than the cobalt monomers. The modulation in the copper-colored surface shows the quantum mechanical interference pattern resulting from the scattering of the quasi 2--dimensional electrons at the copper (111) surface off the cobalt atoms. The cobalt atoms are a magnetic impurity in the copper system and give rise to the Kondo effect. Future measurements concentrate on the dependence of the Kondo resonance on magnetic field and the interaction between magnetic atoms. (J.A. Stroscio, E.W. Hudson, R.J. Celotta, A.P. Fein, and S.R. Blankenship)

Figure 1. Rendered STM image of two cobalt monomers and two cobalt dimers on a Cu(111) surface. The image is 8 nm × 8 nm and was recorded at a temperature of 2.3 K and a tunneling current of 50 pA with a sample bias of –20 mV.

• Insight into Magnetic Coupling through Antiferromagnets. The study of magnetic coupling between two ferromagnetic films separated by a non-ferromagnetic metallic spacer has been an extremely active and fruitful area of research over the last decade. The interlayer exchange coupling, which causes the magnetization of the two ferromagnetic layers to be parallel or antiparallel to each other depending on the thickness of the spacer, is now well understood when the spacer is a diamagnetic or paramagnetic metal. However, when the spacer layer is an antiferromagnet, such as Mn, the observed coupling is non-collinear. J. Slonczewski (IBM, Yorktown Heights) proposed the torsion model to describe the more complicated interlayer coupling when exchange coupling within the antiferromagnetic spacer is important. The torsion model predicts that the coupling angle between the ferromagnetic layers is 90 ° for rough interfaces. For sufficiently smooth interfaces, the torsion model predicts that the coupling angle between the ferromagnetic layers will vary around a mean value of 90 ° by an amount that is very sensitive to the width of the thickness distribution of the spacer layer. We have performed scanning tunneling microscopy measurements of the growth of Mn on nearly perfect Fe single-crystal whisker surfaces to determine the thickness distribution of the Mn layer for particular growth conditions. The coupling angles actually measured for Fe/Mn/Fe(001) tri-layers, using scanning electron microscopy with polarization analysis (SEMPA), were in approximate agreement with the predictions of the torsion model. Scanning tunneling microscopy measurements of the lateral scale of the Mn thickness distribution provided insight into how to go beyond the torsion model to obtain a better explanation of the SEMPA results. (D. Pierce)

Figure 2. STM image of Mn film on an Fe whisker substrate. Mn regions that are 9, 10, and 11 atomic layers thick are indicated. Such STM images were used to determine Mn thickness distributions and quantitatively test the torsion model of magnetic coupling through antiferromagnets.

• SEMPA Images of Mesoscopic Ring Magnets. In collaboration with the University of Cambridge Thin Film Magnetism Group, we have used the NIST Scanning Electron Microscopy with Polarization Analysis (SEMPA) facility to directly image the magnetic domain structure of mesoscopic ring magnets. The micrometer sized rings and discs, patterned out of Co thin films, are the basis for new types of nonvolatile, magnetic random access memories. The SEMPA measurements provided the first images of various magnetic structures in these patterned films. Some of the magnetic structures agreed with predictions based on earlier non-spatially-resolved magnetization measurements of these films, but additional, unexpected domain wall structures were also observed. Knowledge about the nanoscale magnetic structure of the various magnetic states and how the states switch from one to another is a critical part of determining whether these patterned magnetic structures will make useful magnetic memories. (J. Unguris and T. Monchesky)

Figure 3. SEMPA image of a circular Co pad showing the two orthogonal in-plane magnetization components (top), the topographic image (bottom), and a schematic of the magnetic domain structure.

• Symmetry Properties of the Intrinsic Birefringence of Calcium Fluoride. Advanced microcircuit fabrication is now performed using DUV lithography for the pattern-forming step. The industry roadmap calls for a progression of DUV laser sources, from Kr:F 248 nm and Ar:F 193 nm lasers to the F₂ 157 nm laser as the critical dimension shrinks according to Moore's law. Presently, calcium fluoride is the material of choice for the 157 nm lithography stepper.

  To achieve diffraction-limited imaging, the combined stepper optics must have a very low value of integrated birefringence. Calcium fluoride is a crystal of cubic symmetry that was known to be practically optically isotropic for long wavelength light. Much to the surprise of the developers of DUV lithography, a NIST experiment performed in the Atomic Physics Division showed that an unacceptable level of birefringence develops in the DUV. In a collaborative theoretical effort with the Optical Technology Division, the experimental results were corroborated and the symmetry properties of the birefringence elucidated.

  The magnitude of the difference of the index of refraction for the two orthogonal polarizations for each propagation direction is shown in figure 4. At the largest, it is over one part per million – over ten times larger than the specification required for diffraction-limited imaging. Such a value has forced the developers of DUV lithography to reconsider their designs of 157 nm lithography systems. Solutions include trying to cancel the anisotropy by orienting different lenses in different directions, additional reliance on reflective optics, and the use of a mixed barium calcium fluoride. (Z.H. Levine, J.H. Burnett, and E.L. Shirley)

Figure 4. The magnitude of the difference of the index of refraction for the two polarizations for various directions of propagation. There are maxima along the 12 equivalent [110] directions and zeroes along the 14 equivalent [001] and [111] directions. Hence, there are seven optic axes in this system, in contrast to the permitted values of 1, 2, and infinity in standard crystal optics. A VRML version of this figure appears at physics.nist.gov/duvbirefring.

• Superfluid-to-solid crossover in a rotating Bose-Einstein condensate. At ultralow temperatures, a dilute gas of atoms can form a new state of matter: a Bose-Einstein condensate (BEC). When a trapped BEC is subjected to rotation, it cannot rotate as a whole due to its superfluid nature; rather, a few isolated quantized vortex lines are formed within the system. As the angular frequency increases, a larger number of vortices penetrate the sample and the cloud begins to expand in the plane perpendicular to the axis of rotation. At very high frequencies, the vortices arrange themselves into highly regular triangular arrays reminiscent of vortex lattices in superconductors. Figure 5 depicts the results of numerical simulations of this behavior, obtained by solving the Gross-Pitaevski equation. BEC densities are shown perpendicular (upper row) and parallel (lower row) to the axis of rotation for two sets of data with increasing rotational frequencies. This work was motivated by experiments at JILA, which, as of the time of this report, have not yielded resolved images of the vortex structures. However, the aspect ratio of the condensate density profile obtained in these calculations is in good agreement with that found in the JILA experiments. (D. Feder and C. Clark)

Figure 5. Vortices in a trapped rotating Bose-Einstein condensate.