Larry Allard, Ted Nolan, and David Joy

olograms are familiar three-dimensional images seen in science museums, bookstores and specialty shops, and on magazine covers and credit cards. Each image is formed using specialized optical techniques and coherent visible light (typically from a laser) that illuminates the object of interest. The hologram, which is recorded on a photographic plate or on film, looks to the naked eye like a series of randomly sized, overlapping concentric circles. However, when the hologram is subsequently illuminated by an appropriate light source, a remarkable, three-dimensional (3-D) image appears within the plate. This image can be viewed from several angles, and parts of the object hidden when viewed from one direction come into view as the hologram is turned and viewed from another direction.  The optical hologram is formed by exposing the film simultaneously with coherent light that reflects from the surface of the object and by light that is split from the original illuminating beam and subsequently directed by mirrors so that it can overlap with the reflected beams as they are incident on the film. The interference pattern that is produced is directly related to the amplitudes (heights) of waves and phases (relative positions) of the waves reflected from the object. Because of the nature of this interference effect, 3-D information about the object is recorded in the hologram. By passing appropriately coherent illumination through the hologram, the imaging process is effectively reversed, and the object is "reconstructed." Changing the viewing direction relative to the plane of the hologram produces a continuum of differing reconstructions, and these appear as different views of the object.

Father of Holography

In 1947 Dennis Gabor, a Hungarian-British physicist, proposed the method of interference imaging and gave it the name holography, from a Latin word meaning "whole writing." The father of holography did not envision its use as an optical technique, however, because he was an electron microscopist who wanted to find a way to sharpen images produced in transmission electron microscopes (TEMs), which were in their infancy in the 1940s.

David Joy (left) and Ted Nolan discuss Bernhard Frost's poster on holography of electric fields.

An electron microscope produces a beam of high-energy electrons, using one of a variety of electron emitters in an "electron gun," and then directs the beam onto a thin specimen using electromagnetic lenses. The beam of electrons has the peculiar property that, because of the high speed of the electrons, it appears as if only one electron at a time is in the microscope and passes through the specimen. Thus, we consider not the particle properties of the electron beam, but the wave properties (moving particles have a wave nature according to quantum theory) in order to describe the formation of an image in the electron microscope. A "plane wave" is presumed to be incident upon the top surface of the thin specimen, and the wave is modified by passing through the specimen to form an image. The resulting image is magnified as it is projected through several additional lenses, creating a final image that can be directly viewed on a fluorescent screen in the microscope or that can be recorded on photographic film or on a charge-coupled device (CCD) detector.

For specimens of crystalline materials, the regular arrays of atoms aligned on crystal planes can scatter, or diffract, the beam in different directions. The diffracted beams are recombined in the modern TEM to form a high-resolution image that gives details of the atomic structure of a crystal and imperfections in the arrangement of atoms in the crystal. However, the technology of the 1940s gave images with resolution too poor to reveal the atomic structure.

Gabor sought a method to remedy this situation. He realized that the TEM image, as with an image formed with any type of radiation, is simply a superposition of wavelets from each point in the thin specimen. The image varies in amplitude and phase from point to point, depending on the nature of the interaction between the incident beam and the specimen as the beam passes through the specimen. A recording medium such as film does not record these two image components separately, however. The intensities recorded are a combination of the amplitude and phase, which cannot be separated in any standard way.

Gabor knew that one of the primary lens aberrations in the TEM, the spherical aberration, affects the phase components of the electron beam. He reasoned that, if the phase component could be separated from the amplitude component, then perhaps the spherical aberration of the microscope could be corrected and eliminated from the image forming process. This change would inevitably lead to a marked improvement in the microscope's resolution. The primary problem was how to separate the image components.

One day, while waiting to begin a tennis match, Gabor had a brilliant flash of intuition. The path leading to the formal development of his method of holography became instantly clear to him. The essence of the method was the combination of two waves within the microscope: the incident, undeviated electron wave and the image wave, which exits the bottom surface of the thin specimen. If the electron optical geometry is correctly set up, these two waves can be made to interfere. The interference pattern then would be processed using optical techniques to form optical holograms.

After describing his method of recording images from which the amplitude and phase components could be separately extracted, Gabor encountered an obstacle when he tried to use the technique. The electron microscopes of his era did not produce an electron wave with sufficient coherence to permit the proper degree of interference required to make a useful hologram. Similarly, holograms could not be produced from ordinary light because typical light sources of the time produced beams that spread over large angles or had a wide range of wavelengths. Thus, holography did not become practical until the invention of the laser, which produces light of a single wavelength moving in one direction. Recently, the development of TEMs using highly coherent field-emission electron sources have made Gabor's original dream come true.

Electron Holography Comes to ORNL

Progress in developing electron holography techniques was slow for a number of years. Electron holography was not pursued strongly until the introduction of the field-emission electron gun on electron microscopes in the late 1970s. Holography at Tübingen was then carried on primarily by Hannes Lichte, who trained under Wahl and Möllenstedt. At the same time, Akira Tonomura from the Hitachi Corporation in Japan studied at Tübingen and returned to Hitachi to lead a major developmental effort in electron holography. Tonomura had the advantage of working for a company that manufactured electron microscopes, and he prodded Hitachi into developing the cold field-emission electron gun, which culminated in the construction of a 350-kilovolt (kV) instrument designed to be optimized for electron holography. This instrument is housed in the new Hitachi Advanced Research Laboratory in Hatoyama, Japan, a northern suburb of Tokyo. It was used for research conducted in a recently completed 5-year, $20-million program supported by the Japan Research Development Corporation . The program achieved its goal: it developed electron holography as a major technique for characterizing materials. As a result of Hitachi's introduction of a commercial 200-kV version of this instrument in 1989, followed by the introduction of 200-kV and 300-kV instruments by other manufacturers, the number of laboratories that can undertake electron holography has increased from only a couple to several dozen.

In the United States, Oak Ridge National Laboratory has taken a leading role in this field beginning in late 1990, when Hitachi installed their 200-kV HF-2000 TEM in the laboratory of David Joy at the University of Tennessee at Knoxville (UTK). Joy, an ORNL-UTK Distinguished Scientist, collaborated with Larry Allard and Ted Nolan, both of ORNL's Metals and Ceramics Division, in initiating electron holography research. Preliminary work at UTK along with the acquisition and installation of an HF-2000 instrument in the High Temperature Materials Laboratory (HTML) at ORNL (see sidebar "The HF-2000 Cold Field-Emission Electron Microscope") led to an internally funded project. The goals of this 3-year project supported by the Laboratory Directed R&D (LDRD) Program were to further develop electron holography applications and to popularize its use both in the United States and internationally. The remainder of this article details some of the basic concepts of electron holography and describes various holography applications pioneered at ORNL.

Optics of Electron Holograms

Several electron optical geometries have been developed for making electron holograms, but the most popular one—the so-called "off-axis, image plane" geometry—is surprisingly simple, as illustrated in the figure at left. It employs the electron biprism, invented by Möllenstedt and Düker in 1955. This device is simply an ultrafine (0.3 micrometers in diameter) conductive fiber positioned in an imaging lens perpendicular to the electron beam so that it splits the field of view (left). A thin TEM specimen is placed over one side of the image field. When a positive voltage is applied to the fiber, the electron waves on either side of the fiber are bent toward the center, eventually causing them to overlap (right). The overlapping waves create an interference pattern of parallel fringes. These fringes are changed in position and contrast, depending upon how the specimen affects the electron beam. The pattern is recorded either on film, or in the case of the ORNL instrument, directly onto a digital CCD camera system. This interferogram, or hologram, is then processed to yield separate amplitude and phase images.

The digital processing software package currently in use in our laboratory was written by Edgar Völkl, who came to ORNL from the University of Tübingen as part of his postdoctoral assignment for the LDRD project. This powerful, sophisticated, easy-to-use Macintosh software package (called HoloWorks') has recently been licensed for commercial sale by Gatan, Inc. , the manufacturer of the digital camera system used on the HF-2000. It is in use in several laboratories in Europe and Japan, as well as in the United States.

In addition to providing a multifaceted hologram-processing software package, our electron holography project at ORNL also has pioneered applications of holography in a variety of materials science areas.

ORNL Applications of Electron Holography

Shapes of nanocrystals. The technique of electron holography offers a unique capability to characterize the three-dimensional structure of nano-sized crystals—crystals about a hundred-thousandth of a millimeter in diameter. We have used this technique (as described below) to study (1) nano-crystals of palladium supported on silica microspheres to simulate a catalyst material (see figure at left), and (2) shapes of nanocrystals of zirconium oxide (zirconia) to determine effects of particle shape on the behavior of the material during sintering.

Catalytic materials are typically composed of metallic nanocrystals that are supported on a highly porous material such as aluminum oxide. Such materials are used to control emissions from automobile exhaust and refine crude oil to produce gasoline. The morphology of these nanocrystals is of great interest to catalysis chemists because morphology (i.e., the nature of exposed crystal facets, or sides) affects the behavior and performance of the catalyst.

Although high-resolution TEM techniques have made it possible to image nanometer-sized particles at atomic resolution, extracting details of structure and morphology on an atomic scale is still a formidable challenge. The oxide support of a metal catalyst often obscures details of the morphology of such particles. One way to circumvent the problem of studying finely dispersed supported metal particles is to use a model support of simple geometry such as amorphous silica microspheres. This model structure permits metal particles deposited on the surface to be observed looking essentially parallel to the plane of contact between the particle and the sphere. The particle that protrudes from the edge of the sphere (see figure below) provides information on particle morphologies and interface structures. This geometry is also ideal for the formation of electron holograms from which pure phase images can be reconstructed.

Electron hologram of a single palladium nanocrystal at the edge of the silica sphere.

Professor Abhaya Datye, our colleague from the University of New Mexico, provides a model catalyst specimen consisting of fine palladium (Pd) nanocrystals that were deposited on amorphous silica microspheres 0.2 micrometers in diameter. The Pd particles were roughly spherical, often faceted, single crystals. Most Pd nanocrystals exhibited a central contrast feature, also often faceted, that typically extended over one-third the diameter of the particle. This feature was presumed to be an internal void, because the crystal lattice planes always extended through the feature, and no feature was ever observed intersecting with the particle surface. However, it cannot be determined from the direct TEM image whether the void is empty or filled with an amorphous material.

Comparison of the experimental (black line) and calculated (gray line) phase profiles. The close match confirms the presence of a void in the particle.

A phase image is reconstructed from the hologram, showing line for phase profile.

An electron hologram of a single Pd particle on the edge of the silica microsphere is shown at the right. The reconstructed phase image appears below. To analyze the phase image, we recognize that the intensity at any point in the phase image represents the change in phase of the electron wave front with respect to the reference wave. This phase change is directly proportional to the local thickness of the material through which the electron beam is passing, if the sample material is uniform in composition and very thin, as is the case for the Pd particles discussed here. Thus, any changes in the intensity of the phase image can be directly ascribed to variations in sample thickness.

The change in phase from point to point across the particle along the line shown (i.e., the phase profile) is shown in the figure above. Using the geometry of the actual particle (assumed to be spherical) and assuming a spherical void slightly offset from center as shown, the expected phase profile was computed (thick gray line). The excellent match of the experimental phase profile with the computed phase profile is convincing evidence that the Pd particles actually contain voids.

To the best of our knowledge, we have reported the first observations of nano-scale voids in metallic single-crystal particles and have made the first unambiguous determination of a void structure in any material.

Professor Altaf Carina of Pennsylvania State University is studying the mechanisms of sintering of fine ceramic powders such as zirconium oxide.

Fig. A
A high-resolution image of a zirconia nanocrystal.

Fig. B
A perspective view of the expected cuboctahedral shape of the crystal, and a view (right) looking down the crystal direction corresponding to the image of Fig. A.

Fig. C
Phase image showing lines X and Y for two phase profiles.

Figs. D and E
The phase profiles correspond to lines X and Y. The dotted lines show profiles expected from the cuboctahedral crystal.

Fig. F
A perspective view of the shape of the crystal (a right prism), as suggested by the phase image analysis.

Nanocrystalline zirconia (ZrO₂) particles often exhibit faceting—they have flat sides like larger crystals—and thus appear to be polyhedral (many-sided) from high-resolution images. However, the actual shapes of such particles have not previously been determined. The material has a cubic crystal structure, and the apparent surface-bounding planes generally lie along cube faces and cube diagonals (which make octahedra, or eight-sided shapes), so it is natural to assume they have a shape. Figure A shows a high-resolution TEM image of a nanocrystalline zirconia particle, and Fig. C shows the corresponding phase image. Analysis of the high-resolution image of the particle suggests that it is viewed looking perpendicular to one of the cube edges in the basic crystal structure. The particle appears to be cuboctahedral based on the bounding planes and overall projected shape. The relative lengths of the facets suggest a polyhedral morphology approximately halfway between an octahedron and a cube.

Figure B shows a sketch of the projection of a cuboctahedron looking in the same crystal direction as the particle is viewed, showing the correspondence between the particle and the model. The line profiles (Figs. D and E) through the phase image, however, are not consistent with the anticipated particle shape. If the particle were cuboctahedral, the profile along line X would show a continuous, linear change in phase, for example, as indicated by the dotted line in Fig. D. The abrupt phase change actually seen, coupled with a region of essentially flat phase seen in both the X and Y profiles, suggests that the particle is a "right prism" rather than a cuboctahedron. A sketch of the prism, which satisfies the experimentally observed phase changes, is shown in Fig. F.

Clearly, electron holography is a useful technique for characterizing shapes of nanoparticles. This type of information should prove invaluable in better understanding mechanisms of sintering in nanostructural ceramics and the behavior of catalysts that have similar-sized particles of heavy metals, which serve as the primary structures that initiate reactions.

Ferroelectrics. Ferroelectrics are materials that exhibit spontaneous electrical polarization, analogous to the way in which ferromagnetic materials show spontaneous magnetism. Spontaneous electrical polarization refers to pairs of separated positive and negative charges, each like a needle on a compass. When an electric field is applied, the polarization may change, just as when you turn from facing north to facing south while holding a compass and the needle shifts to point in the opposite direction. Because ferroelectric materials can both respond to and generate electric fields, they are important as potential information storage media, sensors and actuators, and light waveguides. They are a member of the class of "smart" materials identified as being of significant technological importance.

Although ferroelectrics have been studied quite intensively, the details of their electric domain structuresrandomly oriented regions of uniform electrical polarizationhave remained unclear because of the practical problems associated with imaging the domains and domain walls. Conventional diffraction techniques in the transmission electron microscope produce only weak contrast between regions of different polarization. This deficiency results from mechanical strain caused by the polarization rather than by the electric field itself. As a consequence, it was not known how wide the walls were between domains nor how the polarization rotated across the wallparameters that play an important role in determining the behavior of the ferroelectric material.

The schematic shows the arrangement of ferroelectric domains with polarizations at 90° to each other.

Electron holography provides a solution to these problems. We have demonstrated its capability in our studies of barium titanate (BaTiO₃), a ferroelectric material. As the electron wave transmits through the ferroelectric material, its phase is altered by the polarization of the material, because the phase is influenced in different ways by the varying directions of the electric potential in the ferroelectric domains. This phase change from area to area in the specimen can be measured directly from the hologram.The figure below shows schematically the arrangement used for observation. Two domains with polarizations at 90° to each other are separated by a domain wall. The specimen is tilted so that some component of the polarization is parallel, or antiparallel, to the beam direction, and the sample is thin enough so that electrons can be transmitted through it.

The figure below shows the hologram recorded from such an area. Note that, as the fringes in the hologram cross the position of the domain wall, they shift sideways. Each fringe represents a contour of constant phase so the lateral shift shows that the phase experienced by the transmitted electron waves changes as they cross the domain wall. The distance over which this change occurs, measured perpendicular to the domain wall, is the range over which the polarization is changing and is, therefore, the domain wall width. This parameter can thus be measured, easily and accurately, from the hologram. For barium titanate, the wall width was found to be between 2and 5 nanometers (nm), in good agreement with the most recent theoretical estimates but nearly a factor of 10 smaller than earlier experimental measurements made using standard diffraction contrast imaging methods.

The hologram recorded across a domain boundary clearly shows the fringe bending that results from the different polarizations. The inset shows the relationship between the degree of fringe bending and phase shift.

The exact way in which the polarization vector rotates from one side of the domain wall to the other can also be derived from the fringe bending shown in the above figure because the extent of the lateral shift is proportional to the component of the polarization parallel to the beam direction. The insert in the figure shows how the phase shift across the domain boundary can easily be measured.

We have used the same experimental method to study a variety of other ferroelectric materials and to investigate the behavior of ferroelectrics exposed to increasingly high temperatures. Subjecting the material to increasing temperature destroys the domain structure. The domain structure reforms as the temperature is lowered. We have found that the previous domain configuration usually is "remembered" when the domains reform, indicating the presence of defects and other characteristics that control domain size and location. An analysis of the electron holograms indicates that the defects responsible for this memory are probably positions in the crystal lattice where oxygen atoms are missing, resulting in a buildup of electrons, or "charged oxygen vacancies." This solution had been predicted theoretically, and our work provided the first supporting experimental observations.

Magnetic Field Measurements. With the advent of the new technique of magnetic force microscopy (MFM), surface magnetic field effects in different regions (domains) of a magnetic material can now be imaged at the micrometer level of resolution. In this special technique, which is related to atomic force microscopy (AFM), a sample is scanned in a raster pattern by a special AFM probe having a sharp tip, and the magnetic forces between probe and sample are monitored. However, to interpret quantitatively the magnetization effects in magnetic domains, both the field distribution around the tip apex and the separation between the tip and the sample must be known exactly. Bernhard Frost of the University of Tennessee has developed new holography techniques to image these magnetic fields.

The sketch shows the pyramid-shaped piece of silicon nitride that forms the tip of a magnetic force microscopy cantilever.

The magnetic probe shown on the left is a commercially available pyramid-shaped piece of silicon nitride positioned at the end of a cantilever. The cantilever attaches to an AFM, which provides the raster motion of the tip over the specimen. Prior to use the tip is coated with about a 100-nm-thick film of a magnetic material such as nickel-cobalt (NiCo) on the front face. After deposition of the magnetic layer, the probe is magnetized by applying an external field along the pyramid axis.

The phase image from the MFM tip. The view shows the tip in profile, and phase lines represent the magnetic lines of force emanating from the magnetic layer deposited on one face of the tip (thin gray line on right-hand edge).

The figure shown above is a phase image reconstructed from a hologram of the projected side view of the tip, which appears as the black triangle. This view looks edge-on to the magnetic layer, which is indicated schematically by the thin gray line that has been superimposed on the right-hand edge of the tip. The magnetic lines of force surrounding the tip have been made visible in this reconstruction, because the phase of an electron wave is sensitive to magnetic fields. The total "flux," or amount of magnetic field affecting the sample over a specific area imaged in an MFM, can be determined by the number of black lines revealed in the phase image, because the magnetic flux between two succesive black lines is known. For example, in one experiment at a working distance of 100 nanometers, we found the total magnetic flux over an area of 1 micron by 1 micron to be 6 millitesla. The magnetic field directly under the tip was 36 oersteds, while at a position 500 nm away laterally from this point to the right, the magnetic field was 26 oersteds. The shape and the strength of the field surrounding the tip are important quantities that govern the performance of the MFM. The use of electron holography for the first time to characterize these MFM tips now permits us to quantify the nature of the magnetic field effects around the tip. In this way, we can better quantify the magnetic domain images obtained when the MFM tip scans a magnetic specimen.

Conclusion

We have presented here only a few of the many uses of electron holography to provide new and unique information about materials structures and properties. As we gain further experience in the art of electron holography, we find that the technique can be applied to almost every specimen type that we observe in our laboratory. It is indeed gratifying that our work has gained significant recognition in the field, so that many well-known researchers in the field, including some from international laboratories, have come to work with us. As increasing numbers of the new generation of field-emission electron microscopes become available, we expect that many laboratories will offer this capability in the future and that electron holography will rapidly become a standard technique in materials characterization.

B I O G R A P H I C A L Sketches

Larry Allard is a senior research staff member in the High Temperature Materials Laboratory (HTML) of ORNL's Metals and Ceramics Division. He holds B.S.E., M.S.E., and Ph.D. degrees in materials science and engineering from the University of Michigan. While at the University of Michigan for 13 years, he co-founded the Electron Microbeam Analysis Laboratory and served for several years as group leader and associate director. In 1986 he joined ORNL as a member of the HTML's Materials Analysis User Center, where he has used various electron microscope techniques to study ceramic whiskers, structural ceramics, ceramic-matrix and metal-matrix composites, and catalysts. In recent years he has co-organized four symposia or workshops in holography and coherent beam imaging, and he is currently co-organizing the 2nd International Holography Workshop as well as a workshop on Direct Imaging of Catalytic Materials. He is a recent recipient of a Lockheed Martin Energy Systems Technical Achievement award, and he is co-editing with David Joy and Edgar Völkl a new textbook on electron holography.

Ted Nolan, a senior research staff member in ORNL's Metals and Ceramics Division, is manager of the division's Materials Analysis Group in the High Temperature Materials Laboratory. He holds degrees from Purdue University and the University of Louisville and has undertaken advanced studies at the University of Tennessee at Knoxville. He was one of the first staff members to join the team that created the HTML User Program. His research at the HTML has focused on studies of high-temperature structural ceramics, ceramic whisker growth, ceramic composite interfaces, and catalysts. More recently, he has been instrumental in implementing digital imaging and data analysis in the Materials Analysis Group, which is now a national leader in digital microscopy. Before coming to ORNL, Nolan spent many years characterizing materials systems for gaseous diffusion and advanced uranium isotope separation processes at the Oak Ridge K-25 Site. He is a recent recipient of two Lockheed Martin Energy Systems technical achivement awards and a best paper award.

David Joy is a University of Tennessee­ORNL Distinguished Scientist. He has an M.A. degree in physics from Trinity College, Cambridge University, and a D.Phil. degree in materials science from the University of Oxford . After 13 years at AT&T Bell Laboratories in New Jersey, he joined the staffs of ORNL and the University of Tennessee at Knoxville in 1987. Since that time, he has been involved in the development of electron holography and other advanced microscopy and microanalytical techniques. Joy is the author or co-author of more than 200 publications in electron microscopy and the author or editor of 7 books. He is past president of the Microbeam Analysis Society of America and winner of the Microscopy Society of America Burton Award for Early Achievement.

Where to?

[ Next article | Search | Mail | Contents Review Home Page | ORNL Home Page ]