This section describes and gives examples of processes developed at the DOE laboratories to make LDMMs. In general, these processes fall under one of two schemes: (1) phase separation of polymer or polymer-like solutions, or (2) replication of sacrificial pore formers. Both schemes produce a liquid-filled precursor. The LDMM can be obtained only if this precursor can be “dried” without collapsing the underlying structure. This key synthetic step is rarely a trivial task, and because of its importance, this section starts with a description of LDMM drying technologies.

Most liquid-filled LDMM precursors will collapse into a high-density mass if subjected to evaporative drying. This occurs because large capillary forces are generated when the liquid meniscus moves through the small cells and pores of the precursor. Only very strong and relatively large-celled precursor structures (e.g., inverse emulsions) can withstand these forces.

Some LDMM precursors of moderate strength can be dried by evaporation if the precursor is exchanged into a liquid of low surface tension. Liquified carbon dioxide is an example — at room temperature, liquid CO2 has a surface tension of 1–2 dynes/cm, compared to ˜70 dynes/cm for water, and 20–40 dynes/cm for most organic solvents. CO2 evaporative drying should not be confused with C02 supercritical drying discussed later.

This article discusses the fundamentals required to produce narrow size distribution fine ceramic powders, make suspensions of fine ceramic powders, and make green bodies with a uniform packing of these particles. In all cases, the interface that fine ceramic powders present to their environment is a very important parameter in controlling the properties of the powders during processing.

There are two major classifications for ceramics: structural and functional. The former includes high and low temperature applications. High temperature ceramics are needed for kiln furniture, ladles, catalyst substrates, and insulations. Low temperature uses are represented by the traditional white ware, as well as hardness applications, such as coatings, armor, and cutting tools. Electrical functions include superconductivity, dielectrics, piezoelectrics, and varistors; magnetic functions are represented by ferrite magnets and SQUIDs (Superconducting Quantum Interference Devices); and optical functions include optical and infrared windows, as well as radar windows. Each class of ceramics has different processing problems and, therefore, different research and development directions. The major areas of research advances for structural and functional ceramics are described below.

Electron tunneling measurements have proven enormously valuable in studies of conventional superconductors. Very early measurements confirmed, in an especially convincing way, the existence of the superconducting energy gap, and more detailed studies demonstrated the spectral form of the gap and its temperature dependence. These measurements were instrumental in confirming in some detail the predictions of the Bardeen, Cooper, Schrieffer (BCS) theory of superconductivity in simple metals. For example, it was shown very clearly that the ratio of the energy gap (2Δ) and critical temperature Tc was close to the BCS value (2Δ/kTc = 3.5). As the sophistication of the technique improved, deviations from this BCS weak coupling limit became apparent (2Δ/kTc was measured to be >4 in materials like Pb, for example), and subtle structure in the current-voltage characteristics of tunnel junctions unearthed a signature of the electron-phonon interaction—the microscopic mechanism responsible for superconductivity in these traditional materials. Through a quantitative analysis of this structure, people were able to extract a function α2(ω)F(ω), which is the phonon density of states F(ω) modulated by the electron-phonon coupling function α2(ω). This function gave a quantitative description of the electron-phonon interaction and confirmed beyond a doubt that the electron-phonon interaction was responsible for superconductivity.

Metallic alloy thin film media and ever decreasing head-to-media spacing make severe demands on storage devices. Decreasing head-to-media separation is critical for high storage densities but it also leads to increased slider-disk interactions, which can cause slider and disk wear or even head crashes. Wear can also occur when drives start and stop when the slider contacts the disk at relatively high speeds. The reliability and durability of thin film disks, which provide much higher areal density than conventional oxide disks with particulate media, are achieved by the use of very thin overcoat materials and surface lubricants. This article summarizes the approaches taken in the industry to enhance the tribological performance of magnetic media, with special emphasis on the basic understanding of the processes occurring at the slider-disk interface.

The continuous rise in the demand for storage capacity at a competitive price is the prime motivator of the changes we have seen in the data storage industry. It is clearly stimulating the present move away from particulate media, which has long dominated all fields of data storage, i.e., tape, rigid, and flexible disks, to the thin film storage media. Particulate storage devices use magnetic media formulated by dispersing magnetic particles, usually iron oxides, in an organic binder. In thin film storage devices, the storage medium is a continuous magnetic film, usually a cobalt alloy, made either by sputtering or by electroless plating.

Wall materials for fusion research must meet stringent requirements. The purity of a fusion plasma is critical to obtaining meaningful results during fusion plasma experiments. The concentration of impurity ions in the plasma must be kept low because impurities result in radiative losses that at best decrease the efficiency of the plasma “burn” and at worst prevent ignition completely. Radiative losses scale with the effective charge, Z, of the plasma. Radiative losses due to bremsstrahlung are proportional to Z2; those due to recombination are proportional to Z4. Figure 1 shows the impurity concentration above which ignition cannot be achieved as a function of impurity atomic number for an experimental power reactor. The advantage of using a low-Z material as the first wall of a fusion machine (i.e., as the plasma-facing material) is evident.

The major plasma impurities observed in most experiments are oxygen, carbon, and first-wall material. These impurities are introduced into the plasma by many mechanisms, including outgassing, desorption, chemical interaction, sputtering, and evaporation. Chemical and physical interactions between the first wall and the energetic hydrogenic ions produced in the plasma are important factors in impurity production. (Here “hydrogen” refers to hydrogen and its isotopes, deuterium and tritium.)

The energy contained in the plasma can be quite large (many megajoules) and can interact with the first wall on a short rime-scale (e.g., a millisecond). It is possible that no material will prove suitable for all operational scenarios. Therefore, the burden of reducing the plasma/first-wall interaction to an acceptable level rests on both plasma and materials scientists.

This issue of the MRS BULLETIN is devoted to the technique of neutron scattering and its role in materials characterization. Compared with electrons and x-rays, the other major scattering probes, neutrons seem exotic and expensive. Why then are they so important? This article will address this question generically, with a few specific examples. Subsequent articles will address in greater detail a few of the more prominent techniques that are important for materials scientists.

We will begin by reviewing why scattering methods are used to study structure, when for many materials it is possible to employ electron microscopy (where, of course, the images are reconstructed from the interference among the scattered waves). One reason is that direct imaging often tells very little about the relationship between an observed structure and the material properties that derive from it. Figure 1, for example, shows a stereo image of a colloidal latex which is clearly crystalline. The stereo view of the same material in the melt seems random to the eye. If many thousands of such images are taken, however, and the number of pairs of particles separated by a given distance are plotted against the separation (this is the so-called pair distribution function), it is found that certain separations are highly probable, while others are equally improbable. In the systems of Figures 1 and 2, the pair distribution functions are not very different (particularly for separations up to several interatomic distances), and it is this function which governs the thermodynamic properties of the material.

The ability to study, and to some extent control, the magnetic properties of surfaces and interfaces is an extremely interesting development in the field of magnetism. Along with experimental advances in techniques to accurately prepare magnetic structures by such methods as sputtering and molecular beam epitaxy (MBE) have been theoretical and computational advances in the ability to calculate properties of realistic model systems. At the same time, increasingly sophisticated characterization techniques have become available (e.g., intense synchrotron sources, pulsed sources for magnetic neutron scattering, polarized electron microscopy, scanning tunneling microscopy, etc.), which enable sample characterization at a level not before possible.

In recent years, the study of magnetism near surfaces and interfaces has been driven by the observation of significant differences from bulk behavior.1 An understanding of these phenomena is fundamentally important to the field of magnetism and provides the potential for technological applications.2 Although the real situation is more complicated, for the purposes of the discussion here, it is convenient to think about the causes of the observed physical effects at interfaces and in superlattices as falling into two categories: “dimensionality effects,” due to the reduced coordination of the magnetic atoms; and “substrate effects,” due to interactions between the differing materials at an interface. An ultrathin film on an insulator, or the free surface of a bulk crystal, are examples of the first category. However, the magnetic atoms at an interface between two different materials are affected both by the reduced number of similar neighboring atoms (i.e., the dimensionality) and by the crystal and electronic structure of the neighboring layer (i.e., the substrate).

The adhesion of particles at solid surfaces in liquid media has attracted the attention of scientists because of its various applications as well as the theoretical significance of the processes involved. Early studies were characterized either by poorly defined systems or limited by the properties of a few morphologically well-defined model colloids, such as latex dispersions. Consequently, results were either of semiquantitative nature or were related to some specific cases, which eluded general conclusions. New methods for preparing uniform particles of different compositions, shapes, and sizes make it possible to approach the problem in a more comprehensive manner. For example, to demonstrate difficulties caused by polydispersity, it is sufficient to mention that the electrostatic interaction energy between a plane surface and a particle is approximately proportional to the particle radius, yet the rate of deposition depends exponentially on the height of the energy barrier.

In principle, static and dynamic approaches may be employed in the study of particle adhesion. The static method yields the force required to detach an adhered particle, while kinetic investigations of attachment and detachment give the rates of the respective processes. Both methods offer information on the stability of the system in terms of the bond strength of adhered solids. For small colloid particles, which are the subject of thermal random Brownian motion, the dynamic approach is more appropriate. This article emphasizes the kinetics of deposition and detachment of small colloid particles in liquid media.

Aerogels are a special class of open-cell foams derived from the supercritical drying of highly cross-linked inorganic or organic gels. These materials have ultrafine cell/pore sizes (less than 1,000 Å), continuous porosity, high surface area (400–1000 m2/g), and a microstructure composed of interconnected colloidal-like particles or polymeric chains with characteristic diameters of 100 Å. This microstructure is responsible for the unusual optical, acoustic, thermal, and mechanical properties of aerogels. For example, aerogels can be prepared as transparent, porous solids because their ultrafine cell/pore size minimizes light scattering in the visible spectrum. Figure 4.1 shows the different aerogels that will be discussed in this article.

The hydrolysis and condensation of metal alkoxides is the most common synthetic route for the formation of inorganic aerogels. Inorganic aerogels have been prepared from monomers such as tetraisopropoxy titanate, aluminum secbutylate, and zirconium isopropoxide. Nevertheless, the majority of scientific research has concentrated on the sol-gel polymerization of tetramethoxysilane (TMOS), or the less toxic tetraethoxysilane (TEOS). The resultant silica aerogels are being investigated for applications ranging from window insulation to the collection of hypervelocity partis cles in space.

The sol-gel polymerization of a multifunctional monomer in solution, leading to the formation of an aerogel, is not unique to metal alkoxides. Organic reactions that proceed through a sol-gel transition have been discovered recently.

Metal-ceramic interfaces play an important, sometimes controlling, role in composites, multilayer substrates, capacitors, electron tubes, and automotive power sources. Often bonding and adhesion between the ceramic and metal are critical to the components' performance. Interface geometry and chemistry play a dominant role in determining the mechanical and electrical integrity of composites. Furthermore, unique properties may be developed from multilayer ceramic-metal structures.

Systematic studies of metal-ceramic interfaces started in the early 1960s. Such studies were directed toward identifying general rules that govern bonding and interface behavior both theoretically and experimentally, including the thermodynamics of interfacial reactions and crys-tallographic relationships, and toward evaluating atomistic structure at the interface. This article summarizes results concerning the interrelation between atomistic structure and the macroscopic fracture resistance of metal-ceramic interfaces. More details are published in a recent conference proceedings.

Determining atomistic structures of metal-ceramics interfaces is, in general, complicated since the two materials that have to be matched exhibit different atoms (ions) and possess different crystal symmetries, crystal structures, and lattice parameters. The adjacent lattices are not commensurate, the two different structures can be described as being just quasiperiodic. However, examples exist where the lattice mismatch is small, and both components possess the same lattice symmetry. Ag/MgO and Nb/Al2O3 interfaces are examples that serve as model systems for experimental studies as well as theoretical calculations. The interfaces can be formed either by diffusion-bonding, internal oxidation, or epitaxial film growth.

For many classes of materials, neutron diffraction is the best way to obtain detailed atomic-level structural information. Diffraction experiments on single crystals provide the most precise data, but sufficiently large specimens (>0.1–0.5 mm3) are often not available. Steady development of instrumentation and data analysis techniques, however, has now made it possible to obtain comparably precise structural information from neutron diffraction experiments on powder samples. Such studies have played a prominent role in solid state physics, chemistry, and materials science in recent years. The special capabilities that have contributed to the success of this technique include atomic cross sections that are often favorable for a particular structural problem, high neutron penetrating power, the excellent resolution achieved with state-of-the-art diffractometers, and steadily advancing analysis techniques that facilitate obtaining structural information from a diverse range of polycrystalline materials.

As Axe, Pynn, and Hayter note in their introductory article in this issue of the MRS BULLETIN, atomic scattering cross sections for neutrons are not simply a function of atomic number, as is the case for x-rays. The scattering is predominantly from the nuclei (thus avoiding the form factor diminution observed for x-ray scattering), and coherent neutron scattering cross sections can, generally, be as large for light atoms as for heavy atoms. Light atoms, such as hydrogen (deuterium), oxygen, nitrogen, carbon, or lithium, can therefore be located in the presence of heavier atoms. This advantage has led to the widespread use of neutron powder diffraction for studing metal hydrides and, more recently, oxide superconductors.

Phase-change erasable optical recording uses a focused laser beam as a heat source to reversibly switch a micron-sized area in a thin film between the amorphous and crystalline states. A bit of information is stored as an amorphous spot in a crystalline background, and the state of the bit is determined by the differing optical properties of the amorphous and crystalline phases. This concept was first demonstrated in 1971 and then, after about a decade of exploratory work, the field accelerated throughout the 1980s at several research laboratories. Currently the subject of number of reviews, the field of phase-change materials promises to broaden and intensify in the 1990s.

The active layer, where the storage occurs, is typically a tellurium-based alloy with a variety of solute species. Early work studied the recording properties of single-layered films, but it has been clearly shown that multilayered films, where the active layer is sandwiched between two or more dielectric layers, have superior recording properties and resistance to irreversible damage caused by laser heating. The dielectric layers (typically SiO2, Si3N4, or ZnS) provide barriers to active-layer oxidation and contamination, help prevent the hole formation associated with ablative write-once storage methods, and act as crucibles and heat sinks which contain the molten spot and influence its cooling properties, respectively. A typical multilayer structure is shown in the cross-sectional transmission electron micrograph of Figure 1.

Much attention has again been brought to the question of undergraduate education for the “materials” field. The present reawakening relates to the release last fall of the report by the National Research Council's Committee on Materials Science and Engineering. This far-ranging study, popularly known as the MS&E Study, revealed the vitality, opportunities, challenges, and needs in materials science and engineering for the future. Education was one area that received much attention in this study because it is the enabling aspect for the field: no educated, trained personnel — no progress.

When one considers the breadth of materials science and engineering, the problem of designing a unique undergraduate program becomes a major hurdle. The diversity of the field is, on one hand, a major asset in addressing problems but, on the other, a major obstacle to unifying an educational approach. This problem is not new since we have faced it over time as information and knowledge expands. However, the problem becomes magnified as the time frame for the doubling of information decreases. The explosion of information over the past half-century has drastically altered our outlook on materials, and the educational programs have been evolving with the changes. We are now at one of those crossroads where revolution versus evolution becomes a factor, i.e., the need to change basic philosophy and/or approach becomes the issue. If one doesn't consider a breakdown of the current materials classes into subgroups, then a reunification approach becomes not only desirable but absolutely necessary.

Among the many surprises in the held of high temperature superconductivity, new features discovered in the magnetic phase diagram are among the most exciting and controversial, generating many new physical concepts and impacting practical applications. This brief review complements several other recent reviews and refers mostly to the bulk crystal (not ceramic) Y1Ba2Cu3O7 and Bi2Sr2Ca1Cu2Ox materials, from now on denoted YBaCuO and BiSrCaCuO.

The magnetic phase diagram of a conventional type II superconductor as a function of magnetic held H and temperature T is well known and understood in the mean-field Ginzburg-Landau and London theories. As shown in Figure 1a, a Meissner phase characterized by complete flux exclusion appears at low fields, delineated by a mean-field phase transition line called the lower critical field Hc1(T), which increases linearly with decreasing temperature below Tc and then saturates at low temperature. A second mean-field phase transition line, called the upper critical field Hc2(T), delineates the transition between the normal and superconducting states and shows a T-dependence similar to Hc1(T). In a strongly type II superconductor (in which the penetration depth λ is much larger than the coherence length ξ), the large region intervening between Hc1 and Hc2 is called the Abrikosov mixed phase. Here magnetic field penetrates the superconductor in the form of tubes of magnetic field called flux lines or vortices. In an ideal isotropic superconductor, these vortices self-organize into a hexagonal array. Defects disturb the hexagonal long-range order, causing the array to break up into a kind of glassy state with more or less short-range order.

An interface is an internal surface, the boundary between two media which may be crystalline, amorphous solid, or liquid. Its close similarity with a surface, a solid-vacuum boundary, suggests that many of the powerful techniques available for studying surfaces might be applied to the interface structure problem. The extent to which this is possible is the subject of this article.

The techniques to be discussed in this article include low energy electron diffraction (LEED), medium energy ion scattering (MEIS), x-ray diffraction, and x-ray reflectivity. (The most widely used method, transmission electron microscopy (TEM), is the subject of a separate article in this issue of the MRS BULLETIN.) To summarize what we will find, surface methods were developed to be nonpenetrating in order to have surface sensitivity. This works against us in the interface situation by requiring the use of extremely thin samples, at least on one side of the interface. This means special handling of samples in some cases and raises the possibility of artifac-tual results. Of the three methods, x-ray diffraction is the most penetrating and least surface sensitive; it probably has the greatest potential for widespread use in interface science.

This article defines structure as “atomic structure” for this purpose: we are interested in the coordinates of atoms at the interface and their relation to bulk structures on one or both sides. For this reason, we will consider only interfaces that are crystalline on at least one side. Since crystals are by far our strongest structural reference point, much less can be said about other interfaces. We will also consider the morphology of an interface, defined as the boundary of the crystal(s) that demarcates the interface, also at the atomic level. This is most apparent in the form of interface roughness. The roles of strain and misfit dislocations in interface formation, also studied by these techniques, are outside the scope of this article.

In order for the new generation of long pulse D-T (deuterium-tritium) burning tokamaks to be successful, it is necessary to minimize the helium “ash” that accumulates during the thermonuclear burn. Since tritium is radioactive as well as relatively expensive, the tritium inventory stored at any given facility must also be minimized. It is therefore necessary either to preferentially remove the helium from the plasma, or to recycle all of the gas, extracting the deuterium and tritium for reuse as fuel. This latter approach requires incorporating a high pumping speed and correspondingly large pumping ducts into the fusion device design, as well as an external facility for separating helium from the D-T fuel. Since the He/H ratio in the plasma should not exceed ~5%, the pumping speed must be 20 times higher than required for a system which removes the helium only. The resulting complexity of the pumping system and the need to handle large quantities of tritium imposes a cost penalty of ~100M$ for a typical commercial fusion reactor design. It is therefore desirable to design a pumping system which removes helium only, leaving the D-T fuel in the fusion device. Recall, however, that existing pumping technologies generally provide better pumping of hydrogen, and many pump designs are totally ineffective at pumping helium.

Despite the low solubility of helium in many metals, permanent trapping of helium atoms occurs at defect sites, culminating ultimately in the formation of helium bubbles. Retention of trapped inert gas is determined by diffusional processes and is consequently temperature-dependent. Hydrogen has not only a much higher solubility, but also a much higher diffusion coefficient than helium in most metals. Consequently, it is to be expected that for some metals, there is a temperature range in which implanted hydrogen is released much more quickly than implanted helium. Brooks and Mattas proposed using this “self-pumping” effect as the basis for a pump which preferentially removes helium in the presence of hydrogen.

Magnetic recording is a very useful and versatile technology, and one that is continuously evolving to serve the increasing demand for information storage and to meet the challenge of competitors such as optical recording. The basic principles of magnetic recording are described in detail elsewhere and briefly below.

The elements of a magnetic recording system are a magnetizable surface layer carried on a flexible tape or on a rotating disk, and a transducer that can both write information to and read information from this surface. The tape or disk, often called a recording medium, and the transducer, called a head, move with respect to each other. The information to be stored is originally contained in an electrical signal, either by direct analog representation or via frequency, phase, amplitude, or pulse-code modulation. In response to this signal, the head in the writing mode generates an intense, localized magnetic field that is capable of changing the direction and degree of the magnetization in the surface material. Each time the input signal changes sign, the writing field changes direction and a transition between regions of opposite magnetization is created. As the head moves along the surface, a series of these transitions is created along a track. The resulting magnetization pattern of the tape or disk becomes itself the source of a spatially varying magnetic field.