Search

The scanning electron microscope (SEM) is the most widely used, as well as the most versatile, of electron-optical instruments. In all of the imaging modes of the instrument it is now possible to achieve a spatial resolution on the nanometer scale from a bulk specimen, provided that the electron-optical performance of the instrument is of the necessary quality. In practice this means that a field emission gun (FEG) as well as a highly excited probe forming lens must be used. With state-of-the-art instrumentation, image resolutions as good as 4 - 5 nm at lkeV, and lnm at 20keV, are now achievable. This paper outlines the design criteria for a high resolution FEG SEM and discusses the type of image information available. The performance achieved is compared to that of other types of microscope that offer similar capabilities.

In this paper we will present recent experimental results, using electron holography and electron interferometry, of studies of ferroelectric domain walls in BaTiO3 thin films. Unlike conventional TEM diffraction contrast imaging of ferroelectrics, the new technique that we have developed not only allows direct visualization of ferroelectric domain walls and electrostatic field distributions in the vicinity of the domain wall, but also enables quantitative measurement of domain wall width and local polarization. We have measured 90° domain wall width to be between 20 to 50 Å for a BaTiO3 thin specimen. The variation of polarization across the domain wall will be shown to be close to that predicted by the Zhirnov model. The value of the measured spontaneous polarization is about 1.5×10−5 C/cm2, which is closed to the bulk macroscopically measured value.

The theory of imaging crystallographic defects in solid specimens through the use of electron channeling contrast is reviewed and the necessary conditions for observation are deduced. It is shown that current high performance field emission scanning electron microscopes can meet these requirements and produce dislocation images from suitable materials.

We discuss the use of electron beam induced current measurements in a scanning electron microscope to deduce local Schottky barrier height with high spatial resolution. For theNiSi2/Si system, using UHV-prepared thin “templates”, wedemonstrate the uniformity of barrier heights for A or B single crystal films. In comparison, there is evidence for local variation of barrier height in mixed A+B films. Quantitative models for EBIC dependence on barrier height are discussed. Local variations in barrier height can be overlooked by other techniques and may be much more common than previously suspected.

In many semiconductor materials problems, structural characterizations must be achieved in both the lateral and vertical dimensions. Although a combination of cross-sectional and planar transmission electron microscopy can provide this information, the sample preparation time is demanding and only relatively small volumes of material are examined. We describe here an alternative approach in which the charge collection (‘CCM’) imaging mode of the scanning electron microscope (SEM) is used. It is shown that, by varying the incident electron beam energy, electricallly active defects at different positions beneath the entrance surface of the material can be imaged and their depth estimated.

Silicon-on-Insulator (SOI) films on SiO2 scan-melted at low velocities (20 to 300 µm/sec) with reduced thermal gradients at the melt-freezing interface are found to have qualitatively different properties from similar films melt-scanned at higher gradients and scan velocities. The transition between the two regimes appears to be abrupt. Scanning at intermediate velocities often results an admixture of patches containing one or the other type of material. The slow scan regime is characterized by long straight non-branched subboundaries having a lateral spacing 50-60 µm, and very small tilt misalignments of 0.1° or less. These slow scan subboundaries are found to consist largely of threading dislocations in contrast conventional subboundaries which are tilt boundaries of up to 3° and typically consist of edge dislocations running in the plane of the film.

Interfaces between Si substrates and epitaxial Si buffer layers grown by Molecular Beam Epitaxy (MBE) are shown to contain a high density of SiOx pockets for certain sustrate preparation conditions. It is also shown that post-deposition thermal annealing of these structures grown upon Czochralski wafers can lead to a greatly increased defect density at the interface. The primary model proposed for this increase is trapping of background oxygen diffusing from the bulk of the Czochralski substrate wafers.

Rapid thermal annealing (RTA) has been shown to be a useful technique for improving the epitaxial quality of CaF2 films grown on Si(100) by molecular beam epitaxy. We have modified the annealing procedure to prevent crack formation, while still maintaining high film quality. Our work also shows RTA to be useful for improving the crystalline quality, chemical stability, and electrical characteristics of CaF2 films grown on Si(111). After RTA, the crystallinity of CaF2 films on Si(111), as measured by electron channeling, is indistinguishable from bulk single crystal CaF2, and the interface state density is reduced by nearly two orders of magnitude from that found in as-grown films.

We review recent results of our Graphite Strip Heater Si-on-Insulator (SOI) effort: (i) recrystallization of SOI films on 100 mm wafers, (ii) model of subboundary pattern formation in SOI films, (iii) low defect density SOI films by ultra slow scanning of the melt zone, (iv) low defect density SOI films by patterned openings in the cap oxide overlayer.

Selected area electron channeling patterns were used to identify the cleavage planes in a polycrystalline Al3Ti-base alloy having the L12 structure. In order to do this unambiguously in the scanning electron microscope (SEM), one needs to know that the cleavage facet from which any given channeling pattern is obtained is indeed normal to the electron beam. We accomplished this by utilizing a recently-developed technique in which an optical microscope with a short depth of focus is inserted in the SEM column and used to measure the elevations of several points on the cleavage facets. By appropriately tilting and rotating the sample, and using the optical microscope to measure elevations, it was possible to orient the facets normal to the beam. The cleavage planes in a cast and extruded alloy having an equiaxed grain structure were compared with those in a directionally-solidified (DS) alloy of the same composition. Of the eight cleavage facets examined in the DS material, six were of the {110} type and two were of the {111} type. Of the six facets examined in the cast and extruded material, two each were of the {110} and {111} types, and one each were of the {100} and {013} types. Although it cannot be said that all possible cleavage planes have been identified in this alloy, the availability of several low-strength cleavage planes apparently exacerbates its brittleness.

With X-ray diffraction techniques, it is possible to routinely measure lattice parameters to several parts in 104 for macroscopic specimens. However, measurements of lattice parameter changes for quaternary (InGaAsP) device structures several microns in width are not usually feasible with X-ray diffraction techniques. Convergent Beam Electron Diffraction (CBED), which is one of the techniques available on a modern transmission electron microscope (TEM), may be sensitive to these small, localized lattice parameter changes. Unfortunately, dynamical diffraction effects prevent direct extraction of changes in the lattice parameter from CBED patterns which are obtained from high atomic number materials. For this reason, we have chosen to calibrate the relative position of CBED features with X-ray lattice parameter measurements which were obtained from planar quaternary layers grown on InP substrates. For the active quaternary region of an electro-optical device structure, it is shown that this approach may be sensitive to a relative change in the lattice parameter as small as ±2 parts in 104, which is the uncertainty in the X-ray calibration measurements.