Click for next page ( 114


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 113
Semiconcluctors INTRODUCTION There are only a handful of scientific or technological discoveries that have revolutionized society. Within the past few decades, none has held as central a role as the computer and communication technologies. Spectacular progress in these is directly connected to materials research in semiconductors and other materials used in electronic devices (Figure 5. 1~. If the rate of progress that has characterized this technology is to continue into the next decade, our scientific understanding of such subjects as semiconductor surfaces, interfaces, and defects and of deliberately structured materials-either geometrically or spatially will be indispensable. instead of reaching a plateau after its initial explosive growth following the discovery of the transistor based on semiconductor physics and materials, materials research related to semiconductor technology is expected to receive another impetus to further growth from the advent of very-large-scale integration (VLS11. Alongside these exciting technological developments semiconductor physics has continued to be a surprisingly rich and fertile field of scientific inquiry. Current experimental and theoretical developments offer tantalizing suggestions that we may be able to understand some of the properties of these materials at a microscopic level. However, recently discovered new phenomena such as the quantum Hall eject or 113

OCR for page 113
114 A DECADE OF CONDENSED-MATTER PHYSICS FIGURE 5. ~ An experimental I-megabit dynamic random-access silicon memory chip. The width of the smallest features is just 1 micrometer. Access time is 150 nanoseconds. (Courtesy of IBM Thomas J. Watson Research Center.) deeper insights into the transport properties of disordered solids are constant reminders that not everything can be anticipated or claimed to be understood. The ability to prepare materials deliberately with atomic arrangements not found in nature is another reminder that science and technology are often symbiotic. The tools for preparing these materials were developed for computer technology and are now used to prepare and characterize materials that may, in the future, provide even more powerful and cost-effective computer components. In this chapter, it is not our intent to discuss technology in any detail or the essential role of semiconductor materials in it, but rather we hope to convey a brief perspective of the status in semiconductor science. However, as it is a field that is characterized by a relatively unique interplay between science and technology, it is useful in this introductory section to indicate the relationship of science, to be described in the following sections, to technological development. Semiconductor science includes the growth and characterization of materials as well as the study of physical phenomena. Spurred by technology, the study of growth and characterization of materials will remain a competitive area. New and improved materials for applica

OCR for page 113
SEMICONDUCTORS 1 15 tions ranging from infrared detectors, solar cells, and solid-state lasers to transistors and VLSI circuits will continue to be in great demand. The properties of surfaces of semiconductors is an active area of solid-state physics. New theoretical calculations and experimental techniques show promise of resolving many of the outstanding issues related to the structure of semiconductor surfaces. The interactions of ions, atoms, and molecules with semiconductor surfaces play a signif- icant role in the processing of semiconductor materials, ranging from epitaxial deposition to reactive ion-beam etching. In related studies, most semiconductor device materials require contacts of one kind or another. Semiconductor-solid interfaces are therefore an important area of investigation. The role of point, line, and planar defects in the electronic properties and yield of semiconductor devices has long been recognized, and much has been understood about such defects. However, there are still outstanding issues that need resolution, and it is expected that these will continue to interest scientists and technologists for the foreseeable future. One of the more exciting areas of solid-state physics in the past few years has been the role of disorder and dimensionality on transport in solids. Lithographically produced structures, two-dimensional inver- sion layers, and high-mobility semiconductors have been widely used in the investigation of phenomena related to quantum transport local- ization, Coulomb interaction effects, and the integer and fractional quantum Hall effects. Heterostructures continue to be investigated for their optical and electrical properties. Recent technological develop- ments in the production of high-speed transistors in GaAs-based epitaxial multilayers has brought renewed and increasing emphasis on this class of materials. Over the coming decade we expect these materials to be explored intensively for electronic applications requir- ing high speed and high-density integration. The search for new semiconductor materials or ingenious methods for fabricating known semiconductors with the potential for novel device geometries is expected to be an active area of interest. The desire to obtain an understanding of amorphous semiconductors re- mains strong. Technological applications such as copying, solar cells, and optical storage will continue to drive this field. SURFACES AND INTERFACES Scientific interest in surfaces and interfaces of semiconductors is worldwide. The use of ultrahigh-vacuum technology over the last decade has provided data on clean surfaces, surfaces with controlled

OCR for page 113
116 A DECADE OF CONDENSED-MATTER PHYSICS exposure to impurities, and interfaces. The availability of synchrotron radiation has made possible the generation of data on the electronic properties of semiconductors with unprecedented resolution. The recently developed scanning tunneling microscope has provided, for the first time, a direct image of the arrangement of atoms on one of the more complicated surfaces of silicon t(111) 7 x 7] (see Figure 7.1 in Chapter 7~. Theoretical approaches have provided reliable estimates of the energies of semiconductor surfaces as functions of atomic posi- tions. The latter enable one to rule out a variety of possible surface models by comparing their relative energies. The last decade can be characterized as one in which a great variety of experimental and theoretical techniques were developed. It can also be characterized as one in which it was realized that understanding surfaces and interfaces is difficult but important problems to solve. A combination of the results of a variety of experimental techniques using x rays, electrons, ions, and atoms has provided evidence that our understanding of the atomic arrangement at surfaces is only now beginning. In fact, it is generally agreed that only one semiconductor surface that of GaAs (1101- is currently known reliably and accu- rately. A particularly encouraging development in the theory of clean surfaces has been the ability to calculate the total energy of crystals as a function of atomic geometry. Such calculations have already pro- vided evidence (confirmed experimentally) that the notion of the buckling of surfaces a long-held view in this field is valid more for ionic semiconductors such as GaAs than for the covalently bonded Si. More surprisingly, it has been proposed, and current experiments support this view, that a surface of Si rearranges its atomic positions to form bonding more characteristic of carbon (pi bonding) than of Si. Parallel to the study of clean surfaces, the effects of impurities, both physisorbed and chemisorbed, have been investigated on a number of different semiconductors and their surfaces. These impurities include H. O. C1, and F. as well as metals such as Al or Pd. on Si. The spatial location and electronic effects of impurities have been investigated both by structural studies and, for example, by vibrational high- resolution electron energy loss and infrared absorption spectroscopies. Semiconductor-metal, semiconductor-oxide, and semiconductor- semiconductor interfaces have been intensely investigated over the last decade. With the availability of high-resolution probes and controlled environments under which such interfaces can be prepared, the number of theoretical models that can explain the properties of such interfaces has been sharply reduced. In the case of semiconductor- metal contacts, a particularly important result has been the demonstra

OCR for page 113
SEMICONDUCTORS 1 17 tion that chemical reaction between semiconductor and metal is a dominating factor in interracial properties. This reactivity studied at a monolayer level by atomic and electronic structural techniques has posed some fundamental questions about the formation of Schottky barriers. DEFECTS IN SEMICONDUCTORS Defects in semiconductors are essential for the operation of semi- conductor devices as well as detrimental. They have therefore been studied for a number of decades. The properties of shallow impurities were well understood during the 1960s primarily through optical absorption experiments and effective-mass theory. In contrast, progress in understanding deep impurities and other point defects (deep centers) has been slow, largely because of technical difficulties. Most experimental techniques using bulk samples ran into problems associated with the presence of shallow impurities at concentrations greater than the defects of interest. Theoretical techniques using primarily the cluster approximation, which simulates the infinite crys- tal by a small number of atoms, often yielded poor results. Major advances in both experimental and theoretical techniques for the study of deep centers occurred in the last decade. During the early and mid 1970s, a new family of experimental techniques was developed using junctions (pen junctions and metal-semiconductor junctions, for example) instead of bulk samples. The advantage here is that one can use electric fields to sweep mobile carriers out of the junction region, thus effectively simulating a material without shallow impurities. A variant of these techniques, known as deep-level transient spectros- copy, is particularly powerful because individual deep levels appear as peaks on a continuous spectrum. In the past 5 years or so, we have seen the evolution of a large number of hybrid techniques. For example, optical detection of magnetic resonance combines electron spin resonance (ESR) with luminescence and is thus capable of simultaneously probing the local symmetry and the chemical identity of atoms (which is an ESR feature) and electronic energy levels (which is a luminescence feature). Also during the past 5 years or so, a new theoretical technique has been developed, based on the mathematical tool called Green's func- tions, which enables one to avoid the cluster approximation and treat an isolated defect in an infinite crystal with accuracy comparable with that achieved in the study of the perfect host crystal. This has provided a detailed picture of the electronic structure of many classes of deep

OCR for page 113
118 A DECADE OF CONDENSED-MATTER PHYSICS centers. For each deep center, the number, relative energy positions, and wave-function character of localized states can now be explained in terms of simple physical models. Trends for classes of impurities (or for the same impurity in different hosts) are better understood. In the area of extended defects, the most noteworthy developments have been the achievement of very-high-resolution micrographs and of theoretical simulation techniques, which lead to more reliable identi- fication of the nature of the defects. Major advances have been made in understanding the role of some extended defects in electronic devices. Most notable is the appreciation of the role of dislocations in Bettering defects from the active region of devices and the similar role played by oxygen precipitates. REDUCED DIMENSIONALITY IN SEMICONDUCTORS Advances in semiconductor technology, especially in the silicon metal-oxide-semiconductor technology used to make the devices that are central to computer memories and other applications, have also made possible the observation and study of two-dimensional electron systems in which the electron density can easily be varied over two orders of magnitude (from about 10" to about 10'3 cm-21. These systems are two dimensional in the sense that the motion of the electrons in the direction perpendicular to the semiconductor-insulator interface is constrained to a region of about 10 rim by the interface barrier and externally applied electric fields. The first clear demonstra- tion of the two-dimensional character of these electrons was made in 1966. Many kinds of structure have now been shown to exhibit the reduced dimensionality first seen in metal-oxide-silicon devices. They include heterojunctions (structures in which two different materials adjoin, usually epitaxially), quantum wells formed by two heterojunctions, and superlattices formed by periodic arrays of quantum wells or by periodic variations of impurity concentrations. If the confining potentials quan- tize the energy levels to give level spacings comparable with or greater than the thermal energy kBT and the energy level broadening, then the motion of the electrons will have a two-dimensional character. The metal-insulator-semiconductor structure has the advantage that the carrier density is easily varied by changing the voltage across the device. Although work on superlattices has concentrated on the GaAs-(Ga,Al)As system, because of the favorable growth conditions and simple band structures that they present, there is also considerable work on so-called type II superlattices, in which occupied energy

OCR for page 113
SEMICONDUCTORS 1 19 levels in one material lie at the same energy as empty levels in the other. Intensive studies of electron transport, hot-electron effects, strong and weak localization in one and two dimensions, impurity-band erects, piezoresistance, many-body erects, charge-density-wave ef- fects, and magnetotransport effects including the quantum Hall effect have been carried out on these systems. The ability to vary the electron density has been of great help in allowing meaningful comparisons between theory and experiment to be made. Structures involving GaAs-(Ga,Al)As heterojunctions have the advantage of high mobility, especially when donor impurities are spatially separated from elec- trons. This can lead to electron mean-free paths of the order of I lam at low temperatures, more than an order of magnitude larger than the best values attained in silicon. The smaller electron effective mass in GaAs leads to larger energy splittings both for the quantum levels induced by confining fields and for the Landau levels induced by magnetic fields, which means that lower magnetic fields and higher temperatures can be used than for comparable phenomena in silicon. The reduced dimensionality of the systems being discussed here has made possible the experimental observation of a number of important physical erects. For example, weak localization effects and the re- markable quantized Hall conductance phenomena, both discussed in Chapter 1, have been observed in these systems. The two dimension- ality of these systems also leads to a situation where the electron- electron interactions make a major contribution to the electronic energy levels, as has been verified in far-infrared spectra of silicon inversion layers. OPTICAL PROPERTIES OF COMPOUND SEMICONDUCTORS As interesting (and important) as the optical properties of elemental semiconductors are, compound semiconductors, mainly the Ill-V materials (A~BV), add much more scope to this area of work. The llI-Vs cover a wide energy-gap range (0.172-2.24 eV), are direct gap (not just indirect gap) in much of the range, possess high electron mobilities, can be made into alloys, and, above all, can be made into heterojunctions and are powerful light emitters. Thus, they have the potential to be made into light-emitting diodes (LEDs) and lasers, not to mention photodetectors (and various high-speed transistors as well). In fact, optoelectronics is totally dependent on these materials, i.e., on their optical properties. The binary crystals GaAs (the prototype), GaP, and InP have become important bulk substrate materials, and

OCR for page 113
120 A DECADE OF CONDENSED-MATTER PHYSICS their optical properties, in their own right, are heavily studied. Also, their bulk optical properties serve as a reference for an entirely new and large area of work that is unique: lII-Vs permit the construction of heterojunctions, and this in turn makes possible the construction of quantum-well heterostructures (QWHs) and superlattices (SLs), and thus deliberately designed quasi-two-dimensional structures. This achievement (i.e., quasi-two-dimensional heterostructures) of mainly the past decade puts III-V materials, and their technology, in a special category that promises to be of a revolutionary nature. Also this development is of immediate and long-range use in devices. The two-dimensional nature of a QWH or SL breaks the crystal bulk symmetry and, for example, removes the heavy-hole, light-hole degen- eracy of, let us say, GaAs of thickness smaller than 500 A. In addition, the confined-particle states, electron and hole, partition the conduction and valence bands and permit excitor absorption (and recombination) to be observed in an abnormally large range, including (300 K, 0-10 kbar) up into the region (energy) of higher band minima (L and X). All the usual optical properties are modified by the quasi-two-dimen- sionality of QWHs or SLs. This, of course, is becoming an intensive area of study for a variety of III-V heterostructures, which prefera- bly are lattice-matched (e.g., AlxGa~-xAs-GaAs)' but even in some cases can be strained-layer heterostructures (e.g., GaAs-ln~Ga~_yAs or GaAs~_xPx-GaAs). The undoped QWH or SL is of interest at low and at high carrier levels and serves, moreover, as a reference and comparison for similar heterostructures with impurities intro- duced into the wells or barriers, as is necessary for device applica- tions. In the area of optoelectronics, III-V QWHs and SLs promise to have a profound effect. Already major improvements have been effected in semiconductor laser performance. In the form of QWHs, monolithic, single-diode structures have achieved laser power levels from 100 mW to over 2 W. In these heterostructures the large asymmetry in electron-hole behavior permits a major redesign of valence photo- detectors and makes possible other unique hot-electron devices. Of further interest, impurity-induced disordering can be used to convert, selectively, quantum-layer regions to bulk-layer regions, or lower gap to higher gap, thus making possible interesting device geometries (and microgeometries) and consequently integrated optical and electronic structures. There is little doubt that the optical properties of III-V materials will be a major area of study for 10 and more years and, in general, will be the basis for many further developments in optoelectronics (more sophisticated lasers, LEDs, detectors, real

OCR for page 113
SEMICONDUCTORS 12 1 space negative resistance devices, higher-speed transistors, and inte- grated versions of all of these processing simultaneously charge and photons). Clearly, progress in this area of work will depend on the skill and progress in III-V crystal growth and development. It is not unreasonable to assume also that QWHs and SLs will be constructed in Il-VI and other semiconductor crystal systems with interesting optical properties. AMORPHOUS SEMICONDUCTORS Amorphous-semiconductor physics is concerned with the structural, vibrational, and electronic properties of noncrystalline semiconduc- tors. By material, the field divides into two principal subfields: (1) tetrahedrally bonded group IV elements, mostly Si or Ge, and alloys with each other or with H and (2) the chalcogenides S. Se, or Te, alloyed with each other or with group IV or V elements. By phenom- ena, the field has numerous subtopics that parallel much of semicon- ductor physics as a whole. Within the past decade, by far the most important discovery has been n- and p-type doping of hydrogenated group IV amorphous semiconductors, abbreviated a-C:H, a-Si:H, and a-Ge:H for hydro- genated amorphous carbon, silicon, and germanium, respectively. Related technological applications have rapidly followed, led by world- wi~le efforts in nhotc~voltaics but also including demonstrated applica- tions in xerography, vidicons, and thin-film transistors. Most of the attention for both the physics and technology is focused on a-Si:H because in this system more than in others there is the promise of studying the intrinsic disorder of a prototypical amorphous semicon- ductor. Because of overconstrained bonding conditions, true glasses cannot be expected with fourfold coordination. Experimentally this is seen in the form of incomplete or dangling bonds and other local structural inhomogeneities, which lead to gap states that obscure the basic semiconducting properties of interest, for example, activated conductivity, doping, and distinct band gaps. For Si, hydrogenation heals the dangling and other weak bonds and thereby permits the study of most of the previously observed effects. In the last decade there has been a burst of activity worldwide to capitalize on the scientific and technological promise of doped, hydrogenated group IV amorphous semiconductors. The bonding between atoms in amorphous chalcogenides is different from that present in the group IV semiconductors. Hence, their atomic arrangement and the defects associated with this arrangement are also

OCR for page 113
122 A DECADE OF CONDENSED-MATTER PHYSICS quite different. It is generally believed that electrons are paired at dangling bonds (defects) in the chalcogenides in contrast to, say, amorphous Si where the dangling bond is associated with a single charge. The chalcogens show strong photostructural changes at photon energies comparable to band gaps rather than bonding energies. For example, volume changes of several tens of percent are observed. The microscopic origins of these changes are not known. Chalcogenide materials are being explored for photoresist and optical storage appli- cations. While real qualitative advances have been made in the past decade, quantitative and predictive understanding of the basic phenomena associated with disorder are still lacking in both classes of amorphous semiconductors. There is considerable scientific challenge in the problems of knowing the principal sources of disorder (bond angles, intrinsic defects, and role of impurities, to name a few) and in discovering which semiconductor phenomena are unique to the disor- dered, amorphous state rather than remnants of analogous effects in the ordered, crystalline state. Key to the physics is the sorting out of the innumerable chemical and materials-science preparation and char- acterization aspects. FUTURE PROSPECTS It is our belief that the rate of progress in understanding phenomena and materials and in manipulating materials to obtain deliberately arranged geometrical and spatial structures will accelerate in the next decade. We expect semiconductor research to be an active area of interest not just because of technological forces but also because of our increased experimental and theoretical capabilities. Semiconductor Surfaces and Interfaces A variety of experimental and theoretical techniques will be applied to investigate the nature of atomic rearrangement or reconstruction on semiconductor surfaces. There will be an increasing trend toward understanding the nature of gas-surface interactions for both scientific and technological reasons. Reactions such as etching or deposition of materials with directed external radiation such as that introduced by ions or lasers are likely to be of increasing importance in semiconduc- tor technology. Basic research on semiconductor interfaces will grow in the coming

OCR for page 113
SEMICONDUCTORS 123 years for the following two reasons: (1) The essential role played by semiconductor interfaces in microelectronic devices will become even more significant as the trend toward miniaturization continues; (2) experimental techniques and computational methods are available now for interracial studies that bridge theory and experiment. Key areas of interest and progress can be identified by the following interfaces. SEMICONDUCTOR-SEMICONDUCTOR INTERFACES Of special interest is the low-temperature growth of pen (or n+-n, p+-p) junctions. Since the diffusion distance (which increases with temperature) is an intrinsic limitation on device dimension, low- temperature processing is likely to become crucial for achieving submicrometer structures in VLSI devices. The low-temperature epitaxial growth of high-quality Si on Si with a controlled doping is a key issue. This will require understanding of the atomic process of growth of pure Si in ultrahigh vacuum, the addition of dopant, and interracial defect formation and control during growth. Ion implantation and laser annealing have been combined to obtain a dopant concentration in Si much higher than its equilibrium solubil- ity. Ultrafast interracial growth by energetic beam annealing will be a subject of continuing interest. The motion of a liquid-solid interface at high speed, its nonequilibrium nature, the heat dissipation, and the atomic mechanism involved will be subjects of study. Interfaces in man-made superlattice structures will be another sub- ject of study. Epitaxial growth of materials will be of particular interest, as will the growth of defect-free and stoichiometric GaAs layers for VLSI devices. SEMICON DOCTOR-INSULATOR INTERFACES Currently the most important semiconductor-insulator interface is the Si-SiO2 interface. This is almost entirely due to the use of this combination of materials in curved electronic devices. However, other insulators on Si or GaAs will be actively explored as new device concepts are explored. More studies are required in order to under- stand the atomic structure, composition, and property correlation of the interface, for example, its charge-trapping states. Defect formation on Si during high-temperature oxidation is a subject of current study; it may lead to a better understanding of the intrinsic defects in Si and also of the structure and kinetic processes that determine the behavior

OCR for page 113
124 A DECADE OF CONDENSED-MATTER PHYSICS of the interface. Laser annealing of Si on SiO2 and thermal growth of large grains of Si on SiO2 are subjects of technical importance. SEMICONDUCTOR-METAL INTERFACES The research on metal-St interfaces has experienced rapid growth. This growth has been fostered by technological demands. We expect it to continue for several years. Also, we expect an increasing emphasis on the study of metal-compound semiconductor interfaces. The link between ultrahigh-vacuum studies and those carried out in ambient environments may bridge basic research and technological applications of these interfaces. A desire to seek fundamental understanding of the origin of Schottky barrier formation will motivate continuing research on this topic. Defects in Semiconductors Ion implantation is currently used to introduce controlled amounts of shallow impurities in semiconductor devices. The samples are then annealed in order to redistribute the impurities to electrically active sites. This process depends on the type of annealing used. Shallow- impurity diffusion is also affected by oxidation, the growth of a silicide, and other processing steps. The understanding of migration mecha- nisms both with and without thermal equilibrium is a challenging and important problem for both science and technology. The problem of identifying deep centers is essential for a complete scientific understanding of defect processes and valuable for technol- ogy in enabling appropriate processing steps in the fabrication of devices to be chosen. The study of defect reactions under external stimulation (electron injection, temperature, and laser irradiation, for example) is only beginning, and many effects are not understood. Extended defects such as dislocations are detrimental in device per- formance. The electrical properties, for example, of the core structure of dislocations and the role of impurities in making them conducting are still not understood. Overall, the study of extended defects is closely related to materials processing for devices. The main problems are the understanding of the conditions under which these defects grow, their identification, migration kinetics, and role in reactions. Surface and interface defects are becoming more important as technology evolves toward the use of shallower junctions. Understanding of the pinning of the chemical potential at Schottky barriers is an outstanding problem

OCR for page 113
SEMICONDUCTORS 1 25 whose resolution may involve defects. The nature of defects at the Si-SiO' interface still poses a difficult problem. Process-induced sur- face defects are important for technology, but current understanding is limited. The theory of surface and interface defects is primitive. Systems of Reduced Dimensionality Where might one expect further activity in systems of reduced dimensionality? QUANTIZED HALL EFFECT Work on the fractional and integer quantum Hall effect will continue, and its observation in a wider range of materials can be expected. If the precision being attained now is confirmed by additional work, the quan- tized Hall resistance may become a resistance standard or a secondary standard. GROWTH TECHNIQUES AND LITHOGRAPHY Continuing improvement in semiconductor growth techniques and in lithography can be expected to lead to use of a wider range of materials and to new device structures. In particular, it is possible to reduce dimensions to the order of 10 rim by lithography and to the order of 5 rim by shadowing techniques. This means that it is possible to construct conventional devices small enough so that electrons have a low probability of being scattered during their motion from one contact to the other and should behave ballistically. SMALL STRUCTURES The increased ability to fabricate small structures now makes it possible to reduce the effective carrier dimensionality even further. Narrow lines on surfaces can be expected to lead to corresponding confinement of the carriers inside the semiconductor. For conductivity processes to appear one dimensional, it is only necessary that the relevant mean-free-path parameter be large compared with the channel width. Such one-dimensional behavior has already been observed in a variety of samples. One-dimensional behavior in a quantum sense requires that the carriers be confined in a distance comparable to the electron wavelength at the Fermi surface, which is just out of reach at

OCR for page 113
126 A DECADE OF CONDENSED-MATTER PHYSICS present but can be expected to be achieved in the coming decade. Studies of magnetic-Hux quantization in small structures of normal metals are also being pursued. HETEROSTRUCTU RES Superlattices of type II are materials in which there is energy overlap between filled states in one material and empty states in another; InAs-GaSb is the prototype. They are likely to be studied more extensively and in a wider range of materials. In these systems, electrons and holes lie in adjacent layers. This makes possible new experiments involving excitonic effects and collective effects. Strained- layer superlattices, in which the conditions on lattice-constant match- ing across an interface are relaxed, should also extend the range of materials and structures that can be studied. Given the ability to make small surface structures, it is possible to construct surface superlattices in which the carrier density and the strength of the potential can be varied. THE TWO-DIMENSIONAL WIGNER CRYSTAL The elusive two-dimensional Wigner crystal, the electron crystal expected in a degenerate low-density electron system at low tempera- tures, may finally be observed in inversion layers at semiconductor surfaces in the coming decade, as its classical analogue was observed in electrons on liquid helium in the past decade. Exciting new possi- bilities arise if electrons are placed on a thin film of liquid helium on a substrate in which a periodic- or random-potential is imposed.