TECHNICAL ACTIVITIES 1998 - NISTIR 6268

MISSION
ORGANIZATION
CURRENT DIRECTIONS
TECHNICAL HIGHLIGHTS
FUTURE DIRECTIONS

TECHNICAL HIGHLIGHTS

X-Ray Microtomography of Integrated Circuits. Three-dimensional imaging of buried structures with sub-micron resolution represents an important tool for analyzing modern engineered structures such as those found in microelectronics, quantum wells, and multilayer magnetic devices. We have begun a program to perform x-ray microtomography of buried structures in integrated circuits, in collaboration with the Intel Corporation, Rensselaer Polytechnic Institute, and Argonne National Laboratory. We have used a 200 nm x-ray microprobe at 1.7 keV photon energy at Argonne's Advanced Photon Source to produce a series of two-dimensional microradiographs with different views of a pair of microelectronic interconnects. Tomographic algorithms were then applied to these microradiographs to synthesize a three-dimensional image of the buried circuit. With a capability of resolving features as small as 400 nm, this work represents the highest resolution tomography ever achieved at a photon energy above 1 keV (A. Kalukin, Z.H. Levine, and T.B. Lucatorto)

Figure 1. X-ray images of an electromigration sample. This circuit was exposed to high current density, resulting in electromigration damage, which is seen as a break in the line pattern in the frame on the left. The left frame is one of the two-dimensional micrographs that have been tomographically analyzed to generate the three-dimensional image shown in the right frame. The spatial frame width is 10 microns, and the image resolution is 60 nm. The three-dimensional image incorporates data from twelve two-dimensional views, and has a voxel resolution of 180 nm.

• Improved Radiometric Scale Between 125 nm to 400 nm. In collaboration with Division 844, we have established a new radiometric facility at SURF for the calibration of photodetectors in the 50 nm to 400 nm wavelength range. The several methods of calibration formerly used in this wavelength range are to be replaced with procedures based on the Absolute Cryogenic Radiometer (ACR). Initial work in the region from 125 nm to 400 nm, performed before the SURF decommissioning last year, shows that we are able to reduce the maximum uncertainty in this range from 10% to 2%. Our plans are to extend the ACR method down to 116 nm as soon as SURF III is operational, and then to use the method to improve the accuracy in selected intervals in the region from 5 nm to 116 nm. For example, the EUVL-LLC consortium spearheaded by Intel to develop EUV lithography has expressed great interest in improved accuracy in the interval 11 nm to 13 nm. UV and EUV detectors are particularly susceptible to damage and contamination. A program to develop new detector materials and configurations that are more resistant to these deleterious effects is also underway so that the accuracy of the improved scale can be more reliably transferred to the point of application. (U. Arp, L.R. Canfield, T.B. Lucatorto, C.S. Tarrio, and R.E. Vest)

  

• Far UV Transfer Standard Detectors Activity. During the upgrade of SURF, several improvements were made to the BL-9 detector radiometry system, addressing all of the known marginal hardware sub-systems identified. This system is used to perform absolute calibrations of NIST working standards and relative calibrations of transfer standards in the 5 nm to 50 nm spectral region. Laboratory investigations of possible alternate detector materials in the 50 nm to 254 nm spectral region have continued, with a new photodiode based on platinum silicide/ silicon emerging as a leading candidate. Prototypes of these detectors have shown radiation hardness far beyond the best silicon photodiodes, and may be useful for direct monitoring of the vacuum ultraviolet excimer lasers now being used for medical applications and for lithography. An SBIR Phase I contract was awarded to a manufacturer specializing in the production of far ultraviolet photodiodes, and the early progress has been excellent. For the first time, silicon avalanche photodiodes have been investigated in the far ultraviolet. Design modifications indicated by these studies were conveyed to the manufacturer, and improved structures will be tested when they become available. An absolute trap detector for the region just longer than 160 nm was designed, constructed, and tested. The results confirm that absolute measurements of incident radiation can be made without the need for device calibrations. Twenty calibrations of transfer standard calibration detectors were performed during 1998 for applications in solar physics, astronomy, aeronomy, and plasma diagnostics. (L.R. Canfield and R.E. Vest)

• SURF III Upgrade. The SURF accelerator upgrade project, begun in September 1997, is nearing completion. The principle objectives of this project are to improve the accuracy of the SURF electron storage ring as a primary standard source of spectral irradiance, and to extend the spectral range. These goals were to be achieved through complete replacement of the electromagnet components and control systems.

  The main components of the SURF electromagnet are the upper and lower yokes, the two backlegs, the upper and lower poles, the upper and lower main coils, and the correction or trim coils. All the steel magnetic components for the SURF II electron storage ring were of laminated construction to prevent large eddy current production. Lamination was necessary since the steel was originally designed for the SURF I synchrotron, which operated at 60 Hz. The SURF III steel is constructed of solid blocks, permitting a larger on-orbit field in its slightly larger footprint.

  The new magnet material was delivered to NIST in December 1997, and construction began immediately. Each yoke consists of five blocks, each weighing approximately 15 tons, which were carefully aligned to each other within the exacting tolerance of ±25 µm (0.001 in). The assembled lower yoke section, weighing approximately 75 tons was centered to the same location as the SURF II lower yoke to within ±250 µm. Construction of the magnet was completed in May 1998, and a magnet mapping system was installed at that time.

  The careful construction of the electromagnet and a new design for the magnetic poles was intended to create a highly circular orbit for stored electrons and the capability to accurately measure the characteristics of the magnetic field produced. The critical measurement for orbital circularity is azimuthal uniformity of the magnetic field. Horizontal and vertical fiducial surfaces were accurately machined near the outer perimeter of the upper and lower pole pieces. An air bearing rotary table was centered on the lower yoke and made parallel to the lower yoke’s upper surface. A rod with a dial indicator attached was bolted to the rotary table and used to position the upper and lower poles to the same center as the lower yoke center. Final measurements showed that the pole centers and lower yoke center were all aligned to within ±10 µm (0.0004 in).

  Another critical parameter for operation of SURF III is the radial magnetic field gradient. The poles were designed with a particular shape to produce the gradient for optimum performance. In order to achieve the design gradient it is critical for the magnetic poles to be parallel. Measurements of the gap between the horizontal fiducial surfaces were made with an inside micrometer to determine parallelism. Analysis of these measurements reveals that the pole faces are parallel to within 3 µrad. The rotary table was removed and replaced by a magnet mapping system consisting of two probes that could be moved both radially and azimuthally and a stationary probe attached to the upper pole. This system was used to measure the field gradient and to determine the azimuthal uniformity of the magnetic field. The field gradient was well within the design parameters. The magnetic field in the range corresponding to electron energies from 105 MeV to 405 MeV is azimuthally uniform to better than 2 parts in 104. A sample azimuthal field measurement is shown in Fig. 3 for an electron energy of 388 MeV. For comparison, the dotted line shows the field for SURF II at its highest energy, 302 MeV)

  During the construction of the SURF III magnet, several modifications were made to the storage ring vacuum chamber: new instrumentation for beam diagnostics and ion neutralization were installed; two new beamports were added; and two existing beamports were modified to increase their angular acceptance, optimizing the beamlines for use in the infrared spectral region. On October 21, 1998 the vacuum chamber was installed and the magnet top was set in place for the final time. The control system has been updated and is ready to run SURF III in the standard operational mode. Testing of the rf system at the increased power levels necessary to store electrons at energies greater than 300 MeV is complete. Initially, the maximum electron energy will be limited to 350 MeV due to the characteristics of the accelerating cavity. Even at this energy, the photon flux in the "water window" from 2.3 nm to 4.4 nm is increased by an order of magnitude over the flux available from SURF II. As of November 1998, assembly of the injector vacuum system is nearly complete, and the commissioning process is expected to begin in late November 1998. All indications are that SURF III will be operational before the end of 1998. (M.L. Furst, R.M. Graves, A. Hamilton, L.R. Hughey, R.P. Madden, A. Raptakes, and R.E. Vest)

• Patterning of Silicon by Metastable Atom Impact Depassivation. In a collaboration with researchers at Rice University and Cornell University, we have demonstrated a new way to pattern silicon. Instead of coating a wafer with resist and exposing it to radiation (such as photons, electron, or ions), the new technique uses hydrogen-passivated silicon and exposes it to a beam of metastable rare gas atoms. Hydrogen-passivated silicon consists of an ordinary silicon surface on which the dangling bonds have been terminated with hydrogen atoms. With this termination the silicon surface is extremely passive and will not form an oxide. When a metastable atom strikes the surface, however, its internal energy is released and a hydrogen atom is ejected. This depassivation allows oxide to form (provided oxygen is available), which can then serve as an etch resist. Patterning of the surface is achieved by imposing a pattern on the incoming metastable atom flux—either by using a mask, as was done in the first demonstrations of the process, or by manipulating the metastable atoms with laser light. The latter approach holds promise for pushing the resolution of this patterning technique well into the sub-100 nm regime, as has been demonstrated in related work on laser-focused atomic deposition of Cr atoms. Because the "grain size" of the hydrogen passivation is a single hydrogen atom, the ultimate resolution limit of this new technique is likely to be on the atomic scale. (J.J. McClelland and R.J. Celotta)
  

  Figure 4. Atomic force microscope image of a silicon surface patterned by metastable atom impact depassivation. A beam of Ar(³P_0,2) atoms passes through a square mesh and strikes the hydrogen-passivated surface. The surface is exposed to oxygen and then etched in KOH. In the regions struck by metastable atoms, the passivation is removed, allowing oxide to form. The oxide resists etching, so the pattern is transferred into the surface.

Magnetic Exchange Coupling Strengths of Antiferromagnets. As part of a continuing program to understand the physical basis of magnetic coupling in magnetic multilayers, we measured the exchange coupling between Fe films separated by Ag, Au, Cr, Mn, V, Cu, or Al spacer layers. The films were grown epitaxially on nearly perfect Fe whisker substrates in order to achieve the atomic scale precision necessary to make meaningful comparisons with theory. The coupling through antiferromagnetic spacers, Cr and Mn, is especially of interest because of the current use of antiferromagnets to exchange bias spin valve structures. Figure 5 shows a measurement of the Fe/Cr/Fe exchange coupling strength. The sample for this measurement consists of a variable thickness Cr wedge deposited on the Fe whisker and topped with a thin Fe film. The figure consists of a series of magneto-optic Kerr images taken at different applied magnetic fields. The dark bands correspond to Cr thicknesses at which the Fe film is antiferromagnetically coupled to the whisker. When the applied magnetic field exceeds the exchange coupling strength, the Fe film magnetization becomes aligned with that of the whisker. The composite image therefore graphically shows the thickness dependence of the antiferromagnetic exchange coupling strength. Although Cr coupling strengths have been measured before, this is the first time that an Fe/Cr/Fe multilayer has been grown with sufficient precision to clearly see the spin density nature of the Cr in the coupling strength measurement. Cr is not an ideal ferromagnetic with a 2 layer periodicity, but an incommensurate spin density wave antiferromagnet with a 2.05 layer periodicity that leads to the modulated envelope of the coupling strength in Fig. 5 and the node at about 24 layers. This figure also shows how difficult measuring coupling strengths of well-ordered Fe/Cr/Fe structures one thickness at a time can be: a change in the Cr thickness of only a tenth of a monolayer can lead to over an order of magnitude change in the exchange coupling strength.

Figure 5. A series of MOKE images from an Au(10 ML)/Fe(15 ML)/Cr wedge/Fe whisker sample taken at various applied magnetic fields, showing the field and Cr thickness dependence of the reversal of the antiferromagnetic regions (dark bands). The Fe/Cr/Fe exchange coupling strength is determined from the switching field. A SEMPA image of the same wedge at zero applied field is shown at the bottom for reference.

We were also able to measure the temperature dependence of the Fe/Cr/Fe coupling in our nearly ideal Cr wedge structure. Figure 6 shows SEMPA measurements as a function of temperature. This data clearly shows oscillatory coupling well above the Neel temperature of bulk Cr, T_N=311 K. The coupling periods and hence the nodes in the exchange coupling are very sensitive to temperature. Figure 6 also includes curves showing the thickness at which the phase slips in bare Cr/Fe occur. Like the Cr/Fe case, the short period oscillations in Fe/Cr/Fe exist to nearly twice the bulk T_N and the phase slips have nearly the same temperature dependence. Locating the phase slip is somewhat difficult in Fe/Cr/Fe because the short period coupling strength appears to drop off more rapidly with temperature than that of the long period. The temperature dependence of the coupling also leads to reversals in the direction of the coupling, i.e., below 420 K the coupling at 30 layers is antiferromagnetic, while above 500 K the coupling switches to ferromagnetic. Our measurements are qualitatively similar to neutron scattering measurements which find a commensurate spin density wave below and incommensurate spin density wave above the first phase slip. However, the boundary dividing the two regions of spin density wave behavior saturates at about 300 K in the neutron measurements but extends to well over 500 K for our samples. (J. Unguris, R.J Celotta, D.T. Pierce, and D. Tulchinsky)

Figure 6. The temperature dependence of the bilinear exchange coupling in Fe/Cr/Fe. The phase slips measured on bare Cr are shown by the solid gray line; the dashed line is the estimated position of the next phase slip. Note that the short period oscillations, where visible, have opposite direction at temperatures below and above these lines.

• Model for Exchange Bias. We have developed a model for exchange bias, an effect that arises from the coupling between a ferromagnetic thin film and an antiferromagnetic film. Exchange bias is the shift in the hysteresis loop of the ferromagnetic thin film due to this coupling. While this effect has been known for a long time, there has been a great deal of recent interest because the shift in the hysteresis loop is useful for pinning chosen ferromagnetic thin films in thin film devices. In particular, devices called spin valves, which incorporate such films, are used in the most recent read heads for magnetic disk storage. In these devices, the magnetization of a free magnetic layer rotates in response to applied fields while the magnetization of another magnetic layer is pinned by exchange bias. The changing relative magnetization of the two layers gives rise to a changing resistance through the giant magnetoresistance effect. The interaction between the ferromagnetic and the antiferromagnetic films is poorly understood. Many experiments carried out to understand these systems indicate that there are irreversible effects occurring in the antiferromagnetic layers. We have developed a model for the coupled system that explains both the exchange bias and the irreversible processes. The model is based on winding partial domain walls in the antiferromagnet as the ferromagnetic magnetization is rotated. In some grains, these partial domain walls are stable; these grains give rise to the exchange bias. In other grains, the partial domain walls become unstable at some critical thickness and give rise to the irreversible effects. (M.D. Stiles).

  Figure 7. Partial domain walls, which are wound up by coupling to a ferromagnetic thin film, in a model antiferromagnetic grain.

Pages designed and maintained by the Office of ECSED.
Inquiries or comments: Feedback.
Online: April 1999