The first International Workshop on the Science and Technology of Thin Films for the 21st Century, held this past summer at Northwestern University, assembled the world's leading experts to assess the trends and opportunities in one of the most fruitful areas of materials research. A complex field, thin film science touches virtually all areas of technology and has had a major economic impact as well. Silicon integrated circuits, for example, form the basis of the computing and telecommunications industries, both noted economic cornerstones.

The diverse, highly qualified group of 150 invited speakers and registered participants came to several conclusions:

∎ The scientific, technological, and commercial opportunities in this field have never been more exciting. The panel chairs have articulated their findings, and they are presented here in separate reports.

∎ Although the thin film field is huge, complex, and interdisciplinary, there are common scientific linkages or driving forces in how the field has developed and is practiced. The discovery of new deposition and fabrication techniques has historically led to new thin film materials. Analysis of the basis molecules of these materials has led to the discovery of new physical and chemical phenomena. And the practitioners of thin film science have traditionally had a lively, realistic interest in applying their work.

∎ The funding situation and lack of a coherent national policy in the United States were causes for concern among U.S. workshop participants.

The advent of new materials processing and fabrication techniques has made it possible to control atomic-scale structure to an extent only dreamed of as recently as 20 years ago. This improved structure control takes us one step closer to synthesizing materials with a combination of properties tailored to the intended application. It is this ability to control structure that will accelerate the pace of technological advancement which is limited by the rate of materials development.

Mechanical performance of materials restricts their use, even for nonstructural applications. Magnetic composites used in recording heads are an example. The mechanical and tribological properties of these materials are a major consideration when designing reliable components.

Another example is electronic materials, where the strength of the interface between the polymer packaging and the semiconductor controls the lifetime of the semiconductor device. The semiconductor device undergoes thermal cycling as it is used, straining the polymer/semiconductor interface. Environmental impurities such as moisture can enter through cracks in the interface and degrade the semiconductor, causing device failure.

Three major superconducting parameters can characterize a type II superconductor: critical transition temperature, Tc, upper critical magnetic field, Hc2, and critical current density, Jc. Because of the variety of crystal structures, chemical stoichiometrics, and microstructures of the materials, these superconducting parameters differ greatly from system to system.

It has been found that the critical transition temperature is closely related to the crystal structure and stoichiometry. Previous studies have shown that compounds with a high degree of symmetry tend to be more favorable to superconductivity. Compounds with a cubic structure, such as a bcc structure, usually have higher Tc values than compounds with a hexagonal lattice. Among conventional superconductors, the so-called A-15 structure is most favorable for high transition temperatures. It is well known that many high Tc superconductors such as YBa2Cu3Ox and Bi2Sr2CaCu2Ox have an orthorhombic structure with a high degree of symmetry.

The upper critical field, Hc2, is more complicated in terms of its relationship to structural characteristics. Experimental results have indicated that Hc2 is more closely related to the chemical stoichiometry and crystal structure than to the microstructure.

However, the critical current density, Jc, can vary tremendously in a compound with a given crystal structure. The change in critical current density is associated with the so-called flux pinning that arises from the interaction between the flux lines and crystal defects. Thus, Jc is determined mostly by the microstructure of the materials.

Environmentally sensitive polymers and hydrogels exhibit sharp changes in behavior in response to relatively small changes in conditions such as temperature or pH. Typical changes include precipitation of the polymer in water or collapse of a hydrogel with expulsion of a large fraction of the gel pore water. These changes are depicted in Figure 1. If such a polymer is grafted or adsorbed on another polymer surface or within the pores of a porous membrane, then one observes sharp changes in wettability or permeability, respectively, when the polymer is environmentally stimulated. These changes are usually reversible, although the recdissolution or reswelling processes are often slower than the precipitation or deswelling steps. Other examples of environmental stimuli and polymer system responses are listed in Tables I and II.

A large number of polymers display sharp responses to such stimuli. Many are based on vinyl monomers, such as anionic and cationic monomers for pH, specificion, solvent, or electrically responsive systems, and N-alkyl substituted acrylamides (e.g., N-isopropyl acrylamide, or NIPAAm) for temperature-sensitive systems. Other copolymer compositions may contain the responsive moiety as a pendant group on one of the monomers. Photoresponsive groups have been incorporated in such a way.

The ratio in the responsive polymer of the responsive monomeric component to the “nonresponsive” comonomer (e.g., the pH-sensitive monomer, acrylic acid, versus acrylamide or the temperature-sensitive monomer, NIPAAm versus acrylamide) will control both the sharpness and intensity of the response as well as the specific environmental conditions where it begins.

A little more than 20 years ago we learned how to direct photons, electrons, or ions a few keV or lower energy onto a surface and measure the energy or mass distribution of backscattered or emitted species. By arranging conditions so that the bombarding or detected species is strongly attenuated by matter, the information obtained is restricted to not more than the outermost several atomic layers. We then learned how to infer the composition, chemical state, and (sometimes) the molecular structure of the surface from these spectra. So began the era of “alphabet soup.” The ensuing decades brought several things:

■ The commercial availability of major surface spectroscopies. Even if not totally user friendly, they aren't outright hostile and work for most people on most days. Instrument makers seeking increased sales volume will continue to improve their ease of operation and reliability.

■ The scrutiny of most materials and structures. There is reliable literature about the very large majority of applications. We know where to begin interpretation and what are the major pitfalls. We have little excuse for the egregious errors we sometimes still make.

■ Many good research groups. When something really new comes along (e.g., the high temperature superconductors), the correct understanding of the data is worked out quickly enough to use it as a tool, guiding research and development.

■ The seeming reluctance, outside certain specialized areas, of the materials community to integrate surface spectroscopies into other than basic research. Expense is often cited as a barrier, but this is hardly credible when the daily cost of top-notch spectroscopy isn't more than twice typical industrial researcher internal billine rates.

In 1888, the year commonly taken as the birthdate of liquid crystal research, F. Reinitzer wrote to O. Lehmann to describe the curious properties of cholesteryl benzoate, a “substance [which] has two melting points, if it can be expressed in such a manner.” Throughout most of the 33°C interval between these two “melting points” this material is in the birefringent fluid state now known as the cholesteric liquid crystal. Today it is common to find compounds showing a whole cascade of liquid crystalline mesophases as the temperature is increased, but it is not customary to refer to any of the phase changes between them as “melting points” except for the lowest temperature transition where the crystalline lattice dissolves. In recent years, however, it has been discovered that many cholesteric liquid crystals, includin g cholesteryl benzoate, do something very strange in a temperature interval of only a degree or so just before they yield up their last bit of liquid crystalline order: they form complex structures having the symmetries of cubic lattices — they “freeze”! – and then “melt” at a higher temperature into either the ordinary amorphous liquid or else into a new kind of amorphous liquid which in turn undergoes a sharp transition into the ordinary amorphous liquid at still higher temperature.

In unlubricated sliding contact, essentially all the mechanical work done to overcome friction is converted into heat produced in the vicinity of real contacts. The amount of frictional heat flux q is proportional to the friction coefficient ü, the normal force F, and the sliding velocity ν, but is inversely proportional to the nominal contact area An (e.g., q = (ü × F × ν)/An). The real areas of contact, being much smaller than the nominal contact area, give rise to much higher local heat fluxes in the vicinity of asperity contacts. Because the frictional heat flux enters the contacting bodies through these regions (or locations known as “hot spots”), their local temperatures (referred to as “flash temperature”) can be much higher than the overall or “bulk” surface temperature, as discussed in References 1-3.

Previous studies have demonstrated that frictional heat can profoundly affect the friction and wear behavior of both metallic and ceramic materials. In most steels and nonoxide ceramics, frictional heat was found to foster oxidation. The occurrence of phase transformations on or near the sliding surfaces was also cited in the literature for certain steels and ZrO2-based ceramics.

Except for SiC, BeO, and AlN, most ceramics have significantly lower thermal conductivity than do metals. When in sliding contact, ceramics cannot dissipate frictional heat generated at sliding interfaces as effectively as most metallic alloys. Large temperature gradients can often develop between areas of real contact and surrounding regions, thus creating high thermal stresses.

It is generally agreed that the mechanical performance of continuous fiber-reinforced composites will depend to a large extent on the nature of the interface between the fiber and the matrix. In the case of many inorganic composites, compositional and/or morphological gradients can arise over an extended region between the matrix and reinforcing fiber as a result of processing conditions, mechanical forces, or chemical interactions. The term “interphase” has been applied to such a region.

There have been numerous theoretical and experimental treatments of this subject in an effort to quantify the stress and strain in the interphase region, and to correlate the presence of such forces with the mechanical performance of the attendant composites. Much of the experimental work has emphasized controlling the nature of the interphase through careful control of the processing conditions or by the introduction of special interface layers, often by applying a coating to the reinforcing fiber prior to incorporating it into the matrix.

Lubrication may be defined as any method used to achieve control of friction and wear of interacting surfaces in relative motion under load. Gases, liquids, and solids have been used successfully as lubricants. To prevent surface contact, liquids and gases provide a film under hydrodynamic pressure to support the load.

When the load is high and/or the speed is low, the hydrodynamic or hydrostatic pressure may not be sufficient and the surfaces come into close contact. The amount and the extent of the surface contact depends on many factors: surface roughness, fluid film pressure, normal load, hardness of the materials, etc. When the surfaces come into close contact, many of the asperities undergo elastic deformation. The condition is generally referred to as elasto-hydrodynamic lubrication (EHL). EHL theories are well-developed. They describe and predict the surface temperatures, fluid film thickness, and hydrodynamic pressures. Contact pressure increases beyond the EHL conditions causes asperities to deform plastically and thinning of the fluid film. When the average fluid film thickness falls below the average surface roughness, the interaction between the contacting surfaces becomes the dominant factor in supporting the load.

This condition is referred to as the boundary lubrication (BL) regime. Theories for BL are not well-developed and the detailed processes are not understood. The classical view of boundary lubrication postulates the formation of a surface chemical film which is easily sheared and protects the surface.

The pervasive role of defects in determining the thermal, mechanical, electrical, optical, and magnetic properties of materials is biblical. Thermodynamic control of imperfection under equilibrium conditions dictates, for instance, the high temperatures needed to raise defect content for diffusion processes. Nonequilibrium treatments, such as work hardening, are used to control dislocation and grain boundary density and morphology to enhance mechanical properties. Both approaches represent the practice of defect engineering. Both are examples of a synergistic interaction between science and engineering in which an existing knowledge base is applied to its limits, stirring the development of new knowledge and new applications.

The purpose of this article is to convey the flavor of the defect engineering culture. The invention of the transistor can be traced to a triumph of defect engineering. Original explorations of semiconductor materials had the goal of controlling surface rectification properties to devise rectifiers, oscillators, and amplifier substitutes for vacuum tube counterparts. Schottky barriers, p-n junctions and metal-oxide-semiconductor capacitors—the products of the endeavor—are now the building blocks of today's microcircuits. The commercial success of these applications has fueled a boom in materials physics research during the last two decades. The work-hardening knowledge base can be traced from the Japanese swordmaking ritual to the discovery of dislocations (in theory first, and then by direct observation). Expansion of the dislocation knowledge base was a dominating concern in materials science prior to the transistor. As shown in this article, these two disparate areas are essential components of the defect engineer's tool kit.

For many years, drug delivery systems were composed of simple pills, eyedrops, ointments, or intravenous solutions. Recently, materials have begun to play a major role in improving drug delivery. Drugs are now chemically attached to polymers, entrapped in small vesicles that are injected into the bloodstream, or put in pumps or polymeric materials that are placed in the body. These new materials-based systems are beginning to change the way drugs can be administered and, in so doing, have improved human health. This article provides a brief review of the major classes of drug delivery systems; a recent paper discusses these issues in detail.

Chemically attaching a drug to a polymer may alter such properties as its distribution in the body, rate of appearance in certain tissues, solubility, or antigenicity. For example, drugs have been linked to soluble macromolecules such as proteins, polysaccharides, or synthetic polymers via degradable linkages. This alters the drug's size and other properties, resulting in a different bodily drug distribution pattern. One example involves coupling the antitumor agent neocarzinostatin to styrene-maleic acid copolymers. When this complex was injected intra-arterially in patients with liver cancer, tumor size decreased significantly. In animals, the antitumor agent, doxorubicin, bound to N(2-hydroxypropyl) methacrylamide copolymers reduced toxicity. The plasma half-life and the drug levels in the tumor increased while the concentrations in the rest of the body decreased.

Liquid crystals, discovered just a century ago, have wide application to electrooptic displays and thermography. Their physical properties have also made them fascinating materials for more fundamental research.

The name “liquid crystals” is actually a misnomer for what are more properly termed “mesophases,” that is, phases having symmetries intermediate between ordinary solids and liquids. There are three major classes of liquid crystals: nematics, smectics, and columnar mesophases. In nematics, although there is no correlation between positions of the rodlike molecules, the molecules tend to lie parallel along a common axis, labeled by a unit vector (or director) n. Smectics are more ordered. The molecules are also rodlike and are in layers. Different subtypes of smectics (labeled, for historical reasons, smectic A, smectic B,…) have layers that are more or less organized. In the smectic A phase, the layers are fluid and can glide easily over each other. In the smectic B phase, the layers have hexagonal ordering and strong interlayer corrélations. Indeed, the smectic B phase is more a highly anisotropic plastic crystal than it is a liquid crystal. Finally, columnar mesophases are obtained with disklike molecules. These molecules can stack up in columns which are themselves organized in a two-dimensional array. There is no positional correlation between molecules in one column and molecules in the other columns.