This issue of the MRS Bulletin aims to highlight the innovative and exciting materials science research now being done using in situ electron microscopy. Techniques which combine real-time image acquisition with high spatial resolution have contributed to our understanding of a remarkably diverse range of physical phenomena. The articles in this issue present recent advances in materials science which have been made using the techniques of transmission electron microscopy (TEM), including holography, scanning electron microscopy (SEM), low-energy electron microscopy (LEEM), and high-voltage electron microscopy (HVEM).

The idea of carrying out dynamic experiments involving real-time observation of microscopic phenomena has always had an attraction for materials scientists. Ever since the first static images were obtained in the electron microscope, materials scientists have been interested in observing processes in real time: we feel that we obtain a true understanding of a microscopic phenomenon if we can actually watch it taking place. The idea behind “materials science in the electron microscope” is therefore to use the electron microscope—with its unique ability to image subtle changes in a material at or near the atomic level—as a laboratory in which a remarkable variety of experiments can be carried out. In this issue you will read about dynamic experiments in areas such as phase transformations, thin-film growth, and electromigration, which make use of innovative designs for the specimen, the specimen holder, or the microscope itself. These articles speak for themselves in demonstrating the power of real-time analysis in the quantitative exploration of reaction mechanisms.

The first transmission electron microscopes operated at low accelerating voltages, up to about 100 kV. This placed a severe limitation on the thickness of foils that could be examined: Heavy elements, for example, had to be made into foils thinner than 0.1 μm. It was felt that any phenomenon whose “mean free path” was comparable to the foil thickness would be significantly affected by the foil surfaces, and therefore would be unsuitable for study in situ. However, technology quickly generated ever higher accelerating voltages, culminating in the giant 3 MeV electron microscopes. At these voltages, electrons can penetrate materials as thick as 6–9 μm for light elements such as Si and Al, and 1 μm for very heavy ones such as Au and U.

Static electron microscopy provides views of microstructures frozen in time. For many problems, such snapshots give a clear picture of the material's characteristics after a particular processing step, but for other problems it is the evolution of the microstructure itself, its mechanisms and kinetics, that are important to understand. For this class of problem, in situ electron microscopy is an indispensable tool because it can provide real-time dynamic observations of processes, showing, for example, where dislocations nucleate, how a particle grows, by what mechanism and how fast an interface migrates. The major limitation to such experiments is that foils must be extremely thin to be electron transparent. The proximity of the free surfaces can have a strong effect on the dynamic equilibrium. For that reason, high-voltage electron microscopes are particularly useful for in situ TEM experiments. The greater penetration depth of high energy electrons makes it possible to observe processes in foils that are thick enough to avoid the dominant influence of the surfaces.

This article will describe some experiments in which the dynamic behavior of precipitates in a simple alloy system was examined during in situ temperature cycling in order to understand the effect of bicrystal anisotropy on the characteristics of interface motion.

Processing has always been a key component in the development of new materials. Basic scientific understanding of the reactions and transformations that occur has obvious importance in guiding progress. Invaluable insight can be provided by observing the changes during processing, especially at high magnification by in situ microscopy. Now that this can be achieved at the atomic level by using high-resolution electron microscopy (HREM), atomic behavior can be seen directly. Accordingly, many deductions concerning reactions in materials at the atomic scale are possible.

The purpose of this article is to illustrate the level reached by in situ HREM. The essential procedure is to form a high-resolution image of a standard transmission electron microscope (TEM) sample and then to alter the structure by some means in a controlled manner, such as by heating. Continual recording on videotape allows subsequent detailed analysis of the behavior, even on a frame-by-frame (1/30 second) basis. The most obvious advantage is to follow the atomic rearrangements directly in real time. However, in addition, by continuous recording no stages in a reaction are missed, which can often occur in a series of conventional ex situ annealed samples because of the limited number of samples that can realistically be examined by HREM. One can be sure that the same reaction, in the same area, is being studied. Furthermore, by changing the temperature systematically, extremely precise kinetic measurements can be made (e.g., for activation energies and kinetic laws) and the whole extent of a material transformation can be investigated in one sample, something that would take months of work if studied conventionally. The information provided by in situ HREM is often unique and so it can become an important technique for fundamental materials investigations.

This article describes the application of transmission electron microscopy (TEM) to real-time, in situ dynamic observations of dislocations in strained epitaxial semiconductor heterostructures. Such experiments allow us to directly observe the formation, motion, and interaction of mis-fit dislocations. Preliminary extension of this work to the in situ measurement of the electrical properties of misfit dislocations will also be described.

The Fundamental Scientific Issue

It is well established that it is possible to grow a thin, coherent epitaxial layer on a substrate with a slightly different lattice parameter, as illustrated in Figure la. This concept is known as strained layer epitaxy. In the fields of semiconductor physics and device design, strained layer epitaxy offers many exciting new opportunities (see Reference 1 for a review). A coherently strained structure, however, will store an enormous elastic strain energy density in the epitaxial layer, due to the distortion of interatomic bonds. Therefore, as the epitaxial layer increases in thickness during growth, it will become increasingly energetically favorable to relax this strain energy. A number of relaxation routes exist: (1) roughening of the epitaxial layer surface (see, for example, Reference 2); (2) interdiffusion of the layers (this will generally only be significant at temperatures which are a large fraction of the layer melting temperatures (e.g., Reference 3)); and (3) introduction of a dislocation network into the substrate/epilayer interface, which as shown schematically in Figure lb, will allow the epitaxial layer to relax toward its bulk lattice parameter. This dislocation mechanism is the most prevalent strain relaxation mechanism at typical crystal growth and processing temperatures, and we concentrate on this mechanism in our experimental studies.

Electron microscopy has played an undeniably important role in the development and understanding of new materials. One particular method, plan-view transmission electron microscopy (TEM), has been used to reveal many aspects of materials and to help answer questions that are not accessible by other methods. In planview TEM, materials are studied either by the image contrast caused by diffraction or by direct diffraction analysis. The studies are often at a somewhat lower resolution than is possible with some other electron-based microscopies or, in the case of diffraction, must be interpreted in terms of an average effect. Nevertheless, plan-view TEM is very useful because properties at these intermediate length scales typically determine important materials characteristics. It is plan-view microscopy that gives us the big picture of what is happening in a material.

Until recently, TEM studies of materials have entailed comparing samples taken at various stages of processing and trying to observe and understand the mechanisms that determine the material behavior. With the advent of controllable specimen environments, materials processing has been taken into the TEM, allowing direct observation of the microstructural development. Here we will focus particularly on surface and near-surface processes, such as oxidation.

For surface science, the 1980s were the decade in which the microscopes arrived. The scanning tunneling microscope (STM) was invented in 1982. Ultrahigh vacuum transmission electron microscopy (UHVTEM) played a key role in resolving the structure of the elusive Si(111)-7 × 7 surface. Scanning electron microscopy (SEM) as well as reflection electron microscopy (REM) were applied to the study of growth and islanding. And low-energy electron microscopy (LEEM), invented some 20 years earlier, made its appearance with the work of Telieps and Bauer.

LEEM and TEM have many things in common. Unlike STM and SEM, they are direct imaging techniques, using magnifying lenses. Both use an aperture to select a particular diffracted beam, which determines the nature of the contrast. If the direct beam is selected (no parallel momentum transfer), a bright field image is formed, and contrast arises primarily from differences in the scattering factor. A dark field image is formed with any other beam in the diffraction pattern, allowing contrast due to differences in symmetry. In LEEM, phase contrast is the third important mechanism by which surface and interface features such as atomic steps and dislocations may be imaged. One major difference between TEM and LEEM is the electron energy: 100 keV and above in TEM, 100 eV and below in LEEM. In LEEM, the imaging electrons are reflected from the sample surface, unlike TEM where the electrons zip right through the sample, encountering top surface, bulk, and bottom surface. STM and TEM are capable of ~2 Å resolution, while LEEM and SEM can observe surface features (including atomic steps) with -100 Å resolution.

Electron holography was originally proposed as a means of improving the resolution of the transmission electron microscope (TEM). And within the last decade dramatic improvements in coherent field-emission electron sources, coupled with TEM instrumentation, have allowed the promise of holography to come to fruition. These improvements in TEM resolution notwithstanding, it has been the application of holography techniques and coherent-beam imaging to the investigation of magnetic and electric fields that has sparked the interest of materials scientists. Electron holography is a unique tool for probing phenomena in the microscopic world because all the information about the object is contained in the hologram (both the electron wave's amplitude and phase), whereas conventional microscopy records only the intensity. It is in this sensitivity to the phase that materials scientists can observe the true potential of electron holography.

Electromagnetic phenomena in a wide range of materials systems have been observed via electron holography; domain structures in magnetic particles and electric fields in p-n junctions are just two examples. Furthermore, fundamental physical properties, such as the existence of the vector potential predicted by Aharonov-Bohm (the AB effect), have been conclusively demonstrated by electron holography. For more information, interested readers are referred to several reviews of electron holography.

Since the early work of Ilan Blech and Gene Meieran more than 20 years ago, electromigration (mass transport produced in a metallic conductor by current flow) has been recognized as an important reliability hazard for solid-state electronic devices. The continuing shrinkage of dimensions in VLSI devices has resulted in a continuing increase in the current density in the interconnection lines and a corresponding increase in the potential electromigration hazard. Although there have been improvements in the materials and structures used for interconnections, notably the addition of copper to aluminum and the use of layered structures, the provision of an adequate margin of safety has become increasingly difficult and is an important constraint on both device design and process development.

A variety of performance indicators have been employed by the photovoltaic (PV) community to rate the performance of PV cells, modules, and arrays. The PV power conversion efficiency under a set of standard reporting conditions serves as a basis for meaningful comparisons of PV performance, within and among PV technologies. PV efficiency is defined as 100 times the maximum PV electrical power density produced divided by the incident light power at a reference PV temperature, total irradiance, and spectral irradiance. Standard reporting conditions commonly used by the PV community are summarized in Table I.

The direct and global reference spectra were generated by a comprehensive Monte-Carlo computer model and actually integrate to 768 W m−2 and 964 W m−2. The terrestrial PV community has arbitrarily assigned a 1-sun total irradiance of 1,000 W m−2 for the global and direct reference spectra. The term “direct” in Table I refers to the direct-normal (5° field of view about the sun) component of the global spectral irradiance distribution. The term “global” in Table I refers to the spectral irradiance distribution on a 37-tilted south-facing surface with a solar zenith angle of 48.2° (AM1.5). The terms AM1 or AM1.5 are often used to refer to standard spectra, but the relative optical air mass (AM) is a geometrical quantity and can be obtained by taking the secant of the angle between the sun and the zenith.