Precisely known atomic scattering factors are essential for accurate atom location by channelling enhanced microanalysis (ALCHEMI) based on inner-shell ionization.1 For ALCHEMI using energy dispersive x-ray analysis (EDX), first principles calculations of ionization cross sections, realistically modeling the “delocalization” of the ionization interaction, give excellent agreement with experiment.2 Such calculations are complex and computationally intensive. Hence, simple analytic forms are often assumed to describe the ionization potential. However such analytic forms require prior knowledge of the “delocalization” of the effective ionization interaction. Such an approach assumes that the precise shape of the ionization potential is not important but that at least the half width at half maximum (HWHM) should be accurately estimated, for example using estimates of the HWHM from root-mean-square impact parameters for ionization. However this is generally not a good approximation3 and we have provided more realistic estimates (Fig. 1).

Two-dimensional Atom Location by Chanelling Enhanced Microanalysis ( Alchemi ) maps of characteristic Xray emission as function of the diffraction conditions of a 200kV electron beam hav been obtained from thin crystals by several workers . Reciprocity shows that these are equivalent to the diffraction patterns produced at infinity by 200kV point sources of electrons on atom sites, sue! as D in figure 1, for CsCl. Here ray SCD at left has been time-reversed at right. Using multislici superlattice simulations which launch a spherical wave from an atom site, we find that the resultinj Kikuchi pattern breaks up into separated blobs along the K-lines at small thickness, with one blol generated at infinity along each atomic string direction . Each blob (e.g. ID) is a Gabor in-lim hologram (e.g of atom C, formed by source S at right). Each source atom on equivalent sites (E, D produces identical holograms, whose intensities add.

Contrast in zone axis channelling patterns, formed by variations in characteristic X-ray emission rates as an electron beam is scanned in angle, is related to projected sublattice site symmetries of ionized atoms within the unit cell. Whereas overall Brillouin zone geometry is identical to that observed in large angle convergent beam patterns, this incoherent X-ray channelling contrast is related to ADF or BSE contrast by integration over thickness of the signal generated within the specimen. Contrast is thus relatively stable and easily interpretable.1 This is useful in separating the response of dilute atoms from abundant atomic species, since the specific channelling pattern of each atom has its own overall fingerprint. Correlation of patterns formed from Ta-doped TiAl with bright field LACBED is shown in Fig. 1 (clearly Ta occupies Ti sublattice sites). No simulation of channelling contrast is necessary if minority atomic species are distributed over sublattice sites in a way that can be reconstructed from a linear superposition of majority (or host) atom sites that form part of the regular lattice structure. The distribution over sublattice sites may be determined by any one of a variety of analysis methods.2,3

It has been long known that orientation effects in crystalline materials can influence characteristic x-ray emission and microanalysis1-7. High Angular Resolution Electron Channeling X-ray Spectroscopy (HARECXS)6-7. a variation of ALCHEMI4-5, has been used at ANL for the last few years to investigate the effects of channeling on quantitative XEDS analysis of materials. More recently we have also been using HARECXS to carefully measure elemental disordering in a number of systems and have found that it can be used very successfully to elucidate the various stages of disorder.

Perovskite (nominally CaTiO3) is a host phase for actinides in various wasteforms for the immobilization of high level radioactive nuclear waste. Over geologic time, alpha decay damage of the actinides in perovskite will cause displacive effects that influence the dimensional and chemical stability of the wasteform. in the past, the progression of damage has been studied by monitoring changes in selected area electron diffraction (SAED) patterns with increasing dose (e.g. 11).

The mineral spinel (MgAl2O4) is a candidate material for proposed fusion reactors because of its resistance to structural damage during irradiation. When irradiated to high damage levels, spinel is eventually driven amorphous. However, at lower fluences the nearly cubic close packed arrangement of the oxygen anions remains intact, as indicated by electron diffraction patterns consistent with a face centered cubic Bravais lattice. in this regime, the damage to the material is manifested primarily in the redistribution of the magnesium and aluminum cations among the interstices of the anion sublattice. For ion irradiation, the study of these early stages of damage is well suited to electron beam characterization methods, because the damage is spatially localized at the nearsurface region. Electron diffraction methods have been used for the characterization of cation disorder, but are notoriously unreliable because the similarity in the elastic scattering cross sections offers little discrimination between the magnesium and aluminum cations.

Garnets were one of the first materials in which an occupation of separate lattice sites by different atomic species was determined with an ALCHEMI technique proposed by Spence and Tafto in l982. The reason of so much interest in this material was twofold, i.e. first its known high sensitivity of X-ray generation depending on orientation especially in the axial orientation and second its complicated crystal structure allowing different atomic arrangements in the unit cell depending on its chemical composition. The dodecahedral (X), octahedral (Y) and tetrahedral (Z) sites between the relatively large oxygen atoms can be filled with a variety of small cations in accordance with the formula X3Y2Z3O12. Partial substitution of one cation with another is common in this structure. The results presented in the previous literature indicated that ALCHEMI can only separate the Y-sites from the sum of the other two (X+Z), while the latter has to remain unresolved.

Everyone presently answers “no” to this question, since this methodology for specimen preparation and preservation commonly is believed to be inherently limited to 20-25Å (2.0-2.5nm) resolution. Achievement of resolution levels smaller than this figure have been published only rarely. This report presents experimental evidence from electron diffraction showing that negative staining can preserve periodic protein structure to the level of at least 4Å.

A suspension of bovine liver catalase crystals (orthorhombic: a= 69Å, b= 174Å, c= 206 Å [1]) is deposited on a hydrophilic carbon support and negatively stained using an on-grid protocol [2,3]. For low-dose electron diffraction with a JEOL 100CX transmission electron microscope (l00kV), the set-up for routine selected area diffraction is modified to keep the second condenser lens maximally overfocused; by a defocusing of the diffraction pattern, the grid can be surveyed and suitable thin single crystals centered in the diffraction aperture with only negligible irradiation.

EBSD analyses are routine for understanding and quantifying the orientations of materials. Analyses of single-phase cubic materials are by far the most prevalent reported in the literature. However, the understanding of the orientations of multi-phase materials are just as important from a technological standpoint. The analysis of non-cubic and non-hep materials is also of interest to investigators. However, analyses of these materials provide some unique challenges for accurate and precise orientation and phase determination compared to cubic materials. The ORKID analysis system was designed from the beginning to address these materials and analysis concerns.

Single-phase materials, especially with cubic symmetry, do not require any special treatment for indexing of diffraction patterns. Because only one crystal system is under investigation, the algorithm for matching the measured bands and the expected bands can have some relaxation in error limits. Furthermore, d-spacing measurements of the bands are not strictly required because the bands angles are sufficiently unique.

The Hough Transform is widely used to detect linear features in image processing techniques. However, the features sought in the EBSD patterns are typically Kikuchi bands from low-index crystal planes, which are not lines but are bands of above average intensity bordered by dark bands. This gives rise to a characteristic shaped peak in the Hough transform that has been called the “butterfly” shape. in most cases, to reduce noise, the Hough Transform is convoluted with a mask having a matching “butterfly” shape. Unfortunately, this method sacrifices resolution through averaging the intensity of neighboring pixels. We have been concerned to use the Hough transform to locate the position of the Kikuchi bands with the highest possible precision. in order to achieve this goal we have looked into the way the Hough transform works on Kikuchi bands in detail, and found some anomalies that must be considered if accuracy is to be achieved.

The long-term aim of our research is to detect and image dislocations in the SEM with EBSD. We will apply it first to the characterization of threading dislocations in Si/Si-Ge structures. As an initial step, we set out to detect the shift of the Kikuchi bands in EBSD patterns to sub-pixel precision.

In the standard analysis of EBSD patterns, to find the positions of Kikuchi bands, it is usual to use the Hough transform. This transform displays the intensity along all lines in the EBSD pattern, as a function of the position and the angle of the line. Thus linear features in the pattern (Figure 1) appears as peaks in the transform (Figure 2a). in normal EBSD analysis, the aim is to determine the orientation of the grain to a precision of perhaps a degree or two. Therefore, many simplifications are made in the software with the aim of gaining speed.

An accuracy of better than 1% is needed to measure the changes in charge density due to bonding. Here we report an accuracy up to 0.025% (random error) obtained in rutile crystal structure factors measurement by QCBED. This error is the standard deviation in the mean value obtained from ten data sets. Systematic errors may be present. Figure 1 gives an example of the (200) refinement results. Table 1 lists several low order structure factor refinement results. The accuracy of the measured electron structure factors was 0.1-0.2% but after conversion to x-ray structure factors, the accuracy for low orders improved due to the Mott formula [1] For (110) and (101) reflections, the accuracy in x-ray structure factors became 0.025% and 0.048% respectively. This accuracy is equivalent to that of the X-ray single crystal Pendellosung method on silicon crystals [2].

The experiments were done on a Leo 912 Omega TEM.

A study of the initial stage of AIN growth on SiC substrate by the sublimation re-condensation technique was undertaken to understand the origin of crystal defects in bulk AIN crystals. When SiC was used as a seed crystal the AIN crystals did not completely coalesce, and individual grains were observed. Such crystal growth features were determined by the initial nucleation of the crystals. in this study, the nucleation of AIN on 6H-SiC(0001) was investigated by Scanning Electronic Microscopy (SEM), S-4700 Hitachi, and Electron Backscattered Diffraction (EBSD), Thermo Noran Phase ID.

At high temperatures (>1800°C) in the presence of AIN, the surface of the SiC substrate first decomposed along remnant polishing scratches[l], ultimately leaving hexagonal hillocks; simultaneously, the AIN nucleated on these SiC hillocks. Surprisingly, the AIN nuclei appeared out of alignment with the underlying SiC hillocks. By measuring several points from the SEM image (figure 1), this misorientation is in the range of 15° to 30°.

A free computer program was introduced recently to process electron diffraction ring patterns. The program (called ProcessDiffraction) produces XDR-like output from the digitized version of the measured diffraction patterns. The enhanced functionality of V1.2 is summarized in the present paper.

1. 1. The program starts with a hint-window giving suggestions what can be done next. Hints change permanently as processing progresses. The hint-window can be switched off optionally. It is a complement to the Windows-style Help-system.

2. 2. The center of the rings is found by manually shifting the measured pattern until it overlaps with a generated reference circle. The computer can further refine the position of the center.

3. 3. Elliptical distortion of the measured pattern (NOT of the display, which is irrelevant) is corrected by adjusting the eccentricity and axis direction of the reference ellipse (distorted from the reference circle) until visually coincide with the measured ring.

Electron backscatter diffraction (EBSD) is a powerful tool used for the characterization of deformed microstructures and has been successfully applied to determine the microtexture of specific regions of bulk samples. The work presented here is a continuation of previous analyses investigating the microstructural response of 6xxx aluminum extrudates to complex die designs. Specifically, the grain orientations across longitudinal welds produced from porthole dies is being studied.

Previous work examined the cross section of a sample that was extruded using a specially designed die that would position the weld in the center of the part. This specimen, shown in Figure 1, is a simple, solid, rectangular shape produced under controlled conditions. The other sample being studied, seen in Figure 2, was part of an actual extrusion run, conducted with typical industrial parameters. This sample is part of a complex shape that uses a porthole die to produce hollow sections.

Elastically-filtered transmission electron diffraction patterns of La0.5Ca0.5MnO3-δ have been obtained between 105K to 288K using the Leo 912 Omega. On heating La0.5Ca0.5MnO3 undergoes a transition from antiferromagnetic insulator to ferromagnetic metal (TN ∼ 150K) and then to paramagnetic insulator (Tc ∼ 230K) . The q modulation wave vector along the b-axis direction arises from charge ordering of Mn3+ and Mn4+, and orbital ordering of dz22 orbitals associated with Jahn-Teller distortions in the Mn3+O6 octahedra at low temperature. Both the intensity and the position of superlattice peaks change as a function of temperature during heating. The large incommensurability we see between 105K to 260K differs from that seen previously and proposed model. Figure 1 a) shows the lattice constants a, b, c related to the simple perovskite lattice parameters ap (ap ∼ 3.9Å) by a≈b≈ap and c≈2ap, and indicates the octahedral coordination.

In our initial studies of the phase identification analysis of individual particles by EBSD, we observed that the quality of the EBSD patterns obtained from particles less than 1 micrometer in size decreased with decreasing particle size, density, and atomic number. This effect is shown in FIG. 1 by comparison of an EBSD pattern from a 12 um Al2O3 particle (FIG. la) and a 200 nm Al2O3 particle FIG. lb. Both particles were mounted on 2 mm thick carbon substrates. The Kikuchi bands from the larger particle are clearly visible while the bands from the smaller particle are not. This decrease in image quality is believed to be the result of a combination of 2 factors related to the electron scattering associated with these very small particles. First, as particle size decreases below about 1 micrometer, an increasing number of electrons are transmitted through the bottom of the particles.

Electron backscatter diffraction (EBSD) in the SEM has become a widely used technique for both automated orientation mapping and the identification of unknown crystalline phases. Despite the rapidly growing use of this technique for the above applications, there has been relatively little fundamental research concerned with determining the thickness of the surface layer that generates the EBSD pattern. EBSD patterns are related through reciprocity to electron channeling patterns. Based on this assumption the thickness of the surface layer that an EBSD pattern is generated in is 2 or 3 extinction distances. Extinction distances at low SEM voltages (10-20 kV) are quite short. Thus, EBSD patterns are generated in very thin surface layers. This paper will discuss the results of a study of the information depth of EBSD in the SEM.

Previous research has attempted to determine the information depth of EBSD by obtaining patterns from samples that have had layers of other crystalline or microcrystalline material deposited on the surface.

In a landmark paper Henderson compared X-ray, neutrons and electrons for protein structure determination. He showed that electron microscopy should be superior to X-ray or neutron diffraction in terms of dose for a given resolution. in addition he presented a theoretical analysis to determine the smallest size molecule whose structure could be determined by phase contrast microscopy. Although he qualitatively considered amplitude contrast mechanisms and concluded they were inferior to phase contrast, no explicit numerical analysis was performed. It has been implicitly assumed that bright field phase contrast imaging is the optimal technique for imaging small biological molecules. Protein specimens are usually embedded in some medium such as ice or glucose. Since they must give a very low contrast it seems reasonable to expect that bright field techniques for these weakly scattering objects would be inferior, given that a weak signal is sitting on large background.

The structure and composition of low angle grain boundaries produced in sapphire by a liquid phase sintering process were investigated by conventional and high resolution transmission electron microscopy (CTEM and HRTEM, respectively). Considering the current emphasis on producing ceramics with textured microstructures for various applications, the question of grain boundary wetting vs. dewetting has become a relevant issue to determining the microstructure development and the properties of these liquid phase sintered materials. Accordingly, the present study was designed to cover a wide range of tilt misorientations, twist misorientations, and boundary orientations.

The boundaries were formed by the directed growth of two sapphire plates, both having nominal <0001>, , or surface orientations through an alumina tape-cast containing an anorthite composition glass phase. After an initial hot-pressing stage, followed by an anneal at 1600° C for 200 hours, the samples typically contained a single boundary delineated by isolated pockets of entrapped glass, Fig. 1.

Silica and titania based ceramics and their analogs are some of the most fundamental in crystal chemistry and ceramic science Our interests include applications of nanostructures and chemical composites of the ceramics in nanoelectronics, chemical processes and as scaffolds in biotechnologies. Finely divided titania is used in a vast array of products including paper, paint, food and clothing. Novel microscopy methods including dynamic environmental-high resolution transmission EM (EHREM) at the atomic level, FESEM and cathodoluminescence are leading to striking progress in the development of the ceramic nanotechnologies.

Phase transformations in the cristobalite form of silica, from the tetragonal a phase (low or room temperature form) to the cubic β phase (high temperature, (270°C) form) result in discontinuous thermal expansion and are not conducive to nanotechnology. Here we report fundamental in situatomic resolution studies of the phase transformations using EHREM and have used the results to design a number of stable, single-phase structures at room temperature (RT).