Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Report of the Research Briefing Panel on Chemical Processing of Materials and Devices for Information Storage and! Handling
Research Beefing Panel on Chemical Processing of Materials and Devices for Information Storage and Handling Larry F. Thompson (Chairman), Head, Organic Materials and Chemical Engineering, AT&T Bell Laboratories Lee L. Blyler, Supervisor, Plastics Applied Research, Properties, and Processing, AT&T Bell Laboratories lames Economy, Manager, Polymer Science and Technology, IBM Almaden Research Center Dennis W. Hess, Professor of Chemical Engineering, University of California, Berkeley Richard Pollard, Professor of Chemical Engineering, University of Houston T. W. Fraser Russell, Professor of Chemical Engineering, University of Delaware Michael Sheptak, Senior Staff Engineer, Magnetic Tape Division, Ampex Corporation 26 Staff Robert M. Simon, Project Director, Board on Chemical Sciences and Technology AlIan R. Hoffman, Executive Director, Committee on Science, Engineering, and Public Policy
Report of the Research Bneftug Panel on Chemical Processing of Materials and Devices for Information Storage and! Handlling INTRODUCTION Almost every aspect of our lives-at work, at home, and in recreation has been affected by the information revolution. Today, infor- mation is collected, processed, displayed, stored, retrieved, and transmitted through the use of an array of powerful technologies that rely on electronic microcircuits, lightwave communication systems, mag- netic and optical data storage and recording, and electrical interconnections. Materials and devices for these technologies are manu- factured using sophisticated chemical pro- cesses. The United States is now engaged in a fierce international competition to achieve and maintain supremacy in the design and manufacture of materials and crevices for in- formation storage and processing. The eco- nomic stakes are large (see Table I); national productivity and security interests dictate that we make the strongest possible efforts to stay ahead in processing science and tech- nology for this area. This briefing explores the chemical pro- cessing required in three of these key tech- nologies: electronic microcircuits, lightwave communication systems, and magnetic re- cording media. This briefing also explores 27 briefly some potential needs for advanced chemical processing that may be required to realize more fully the promise of supercon- ducting metal oxides. In high-technology manufacturing of com- ponents for information systems, there has been a long-term trend away from mechani- cal production and toward production using chemical processes. In several of these in- dustries, chemists and chemical engineers have become increasingly involved in re- search and process development. WorId- wide, though, many high-technology indus- tries, such as the microelectronics industry, still have surprisingly little strength in chem- ical processing and engineering. The United States has a special advantage over its inter- national competitors its chemical engineer- ing research community leads the world in size and sophistication. The United States is in a position to exploit its strong competence in chemical processing to (1) regain leader- ship in areas in which the initiative in manu- facturing technology has passed to Japan, and (2) maintain or increase leadership in ar- eas of U. S. technological strength. To achieve their potential contribution fully, it is of paramount importance that chemical engineers strongly interact with
TABLE 1 Total Estimated Worldwide Market for Materials and Devices for Information Storage and Handling (billions of 1986 dollars) Year Technology Electronic semicon cluctors 1985 1990 1995 25 Lightwaves Recording materials 720 Interconnections 1021 Photovoltaics 0.30.S Total Electronics 397550 60160 3 55 58 3 (n. a) Source: AT&T Bell Laboratories. Compiled from various published sources. other disciplines in high-technology indus- tries and have the ability to communicate across clisciplinary lines. The technologies discussed in this report cross over disci- plines such as solid-state physics and chem- istry, surface ancT interracial science, electri- cal engineering, and materials science. Materials and devices for information stor- age and handling are exceedingly cliverse, yet they have many characteristics in com- mon: the products are high in value; they re- quire relatively small amounts of energy or materials to manufacture; they have short commercial life cycles; anc! their markets are fiercely competitive consequently, these products experience rapid price erosion. The manufacturing methods used to produce in- tegrated circuits, optical fiber, and recording media also have common characteristics. Each of these products is manufactured us- ing a sequence of individual, complex steps, most of which entail the chemical moclifica- tion or synthesis of materials. The inclividual processes are designed as discrete unit or batch operations and, to date, there has been little effort to integrate the overall manufac- turing process. Because chemical reactions and processes are used in the manufacture of this broad array of materials and devices, 28 chemical engineers could play a significant role in improving manufacturing processes and techniques, and investments in chemi- cal processing science and engineering re- search represent a potentially high-leverage approach to improving our competitive posi- tion. CURRENT CHEMICAL MANUFACTURING PROCESSES MICROCIRCUITS The use of chemical reactions and pro- cesses in the manufacture of microcircuits begins with the basic material for integrated circuits, high-purity (less than 150 parts per trillion of impurities) polycrystalline silicon. This ultrapure silicon is produced from met- allurgical grade (98 percent pure) silicon, by (~) reaction at high temperature with hydro- gen chIoricle to form a complex mixture con- taining trichIorosilane; (2) separation and purification of trichIorosilane by absorption and distillation; and (3) reduction of ultra- pure trichlorosilane to polycrystalline silicon by reaction with hydrogen at Il00-1200°C. To prepare single-crystal silicon ingots suit- able for use as materials in semiconductors, polycrystalline silicon is melted in a crucible at 1400-1500 °C under an argon atmosphere. Tiny quantities of dopants compounds of phosphorus, arsenic, or boron are then added to the melt to achieve the desired elec- trical properties of the finished single-crystal wafers. A tiny seed crystal of silicon with the proper crystalline orientation is inserted into the melt and slowly rotated and withdrawn at a precisely controlled rate, forming a large (15 cm x I.3 m) cylindrical single crystalwith the desired crystalline orientation and com- position. Crystal growth kinetics, heat and mass transfer relationships, and chemical re- actions all play important roles in this process of controlled growth. The resulting single-crystal ingots are sawed into wafers that are polished to a flatness in the range of from ~ to 10,um.
CHEMICAL PROCESSING OF MATERIALS AND DEVICES The next steps in device fabrication are the sequential deposition and patterning of thin dielectric and conducting films. The pol- ishecl silicon wafer is first oxidized in a fur- nace at 1000-1200°C. The resultant silicon di- oxide film is a few hundred nanometers thick and extremely uniform. The wafer is then coated with an organic photosensitive mate- rial, termed a resist, and is exposed to light through the appropriate photomask. The purpose of the photolithographic process is to transfer the mask pattern to the thin film on the wafer surface. The exposed organic film is developed with a solvent that re- moves unwanted portions, and the resulting pattern serves as a mask for chemically etch- ing the pattern into the silicon dioxide film. The resist is then removed with an oxidizing agent such as a sulfuric acid-hydrogen per- oxide mixture, then the wafer is chemically cleaner! and is ready for other steps in the fabrication process. The patterned wafer might next be placed in a diffusion furnace, where a first eloping step is performed to deposit phosphorus or boron into the holes in the oxide. A new ox- ide film can then be grown and the photore- sist process repeated. As many as 12 layers of conductor, semiconductor, anct dielectric materials are deposited, etched, and/or doped to build the three-ctimensional struc- ture of the microcircuit. Thus a semiconcluc- tor device is a series of electrically intercon- nected films, the successful growth and manipulation of which depends heavily on proper reactor design, the choice of chemical reagents, separation and purification steps, and the design and operation of sophisti- cated control systems. LIGHTWAVE MEDIA AND DEVICES Optical fibers are also made by chemical processes. The critical feature of an optical fi- ber that allows it to propagate light down its length is a core of high refractive index sur- rounded by a cladding of Tower index. The higher index core is procluced by doping sil 29 ice with oxides of phosphorus, germanium, and/or aluminum. The cladding is either pure silica or silica doped with fluorides or boron oxide. Four processes are principally used to manufacture the glass body that is drawn into today's optical fiber. "Outside" pro- cesses, such as outside vapor-phase oxida- tion and vertical axial deposition, produce layered deposits of doped silica by varying the concentration of SiCl4 and dopants pass- ing through a torch. The resulting "soot" of doped silica is deposited and partially sin- tered to form a porous silica boule. In a sec- ond step, the boule is sintered to a pore-free glass rod of exquisite purity and transpar- ency. "Inside" processes, such as modified chemical vapor deposition (MCVD) and plasma chemical vapor deposition (PCVD), deposit doped silica on the interior surface of a fused silica tube. in MCVD, the oxidation of the halide reactants is initiated by a flame that heats the outside of the tube. In PCVD, the reaction is initiated by a microwave plasma. Over a hundred different layers with different refractive indexes (a function of glass composition) may be deposited by either process before the tube is collapsed to form a glass rod. In current manufacturing plants for glass fiber, the glass rods formed by all of the abovementioned processes are then carried to another facility where they are drawn into a thin fiber and immediately coated with a polymer. The polymer coating is important; it protects the fiber surface from microscopic scratches, which can seriously degrade the glass fiber's strength. Current manufacturing technologies for optical fiber are relatively expensive, com- pared to the low cost of commodity glass. U.S. economic competitiveness in optical technologies would be greatly enhanced if low-cost means were found for producing waveguide-quality silica glass. The manu- facture of glass lends itself to a fully inte- grated and automated process (i.e., a contin- uous process). One can envision a fiber
manufacturing plant that starts with the purification of chemical reagents, which is then followed by a series of chemical reac- tions, glass-forming operations, and finally fiber-drawing steps. in such a plant, inter- mediate products would never be removed from the "production line." Sol-gel and re- lated processes are attractive canctidates for such a manufacturing process, which would start with inexpensive ingredients anct pro- ceed from a so! to a gel, to a porous silica body, to a dried and sintered glass rod, and finally to drawn and coated fibers. Such a process could reduce the cost of glass fiber by as much as a factor of 10, a step that would greatly increase the scope, availability, and competitiveness of lightwave technologies. At present the chemical steps involvec! in sol-ge} processes are poorly understood. Methods are being sought to manipulate these processes to produce precisely layered structures in a reliable and reproducible way. RECORDING MEDIA Recording media come in a variety of for- mats (e.g., magnetic tape, magnetic clisks, or optical disks) and are made using a variety of materials anc! processes (e.g., evaporated thin films or deposited magnetic particles in polymer matrixes). To illustrate the chemical reactions and processes in the manufacture of recording media, this section focuses on magnetic particle technology, an economi- cally important part of the market for which the processing challenges are easy to dis- cuss. Chemical reactions and processes are equally relevant to emerging technologies and materials in recording. The manufacture of magnetic recording media depends heavily on chemical process- ing. The density at which information can be recorded is determined by the chemical and physical properties of the magnetic particles or thin films coated on a disk or tape. Para- mount among these properties are the shape, size, and size distribution of the mag- netic particles. An extremely narrow range 30 in the size of magnetic particles themselves only a few tenths of a micron in size must be achieved in a reliable and economic man- ner. These particles must be deposited in a highly oriented fashion, so that high record- ing densities can be achieved by having the magnetic particles lie as closely together as possible. Accomplishing this requires the so- lution of a variety of challenging problems in the chemistry and chemical engineering of barium ferrite anc! the oxides of chromium, cobalt, and iron (e.g., the synthesis and pro- cessing of micron-sized materials with spe- cific geometric shapes). The manufacture of magnetic tape illus- trates an interesting sequence of chemical processing challenges. A carefully prepared dispersion of needIe-like magnetic particles is coated onto a fast-moving (150-300 miming polyester film base that is 0.0066- to 0.08-mm thick. The ability to coat thin, smooth layers of uniform thickness is cru- cial. The coated particles are oriented in a de- sired direction either magnetically or me- chanically during the coating process. After drying, the tape is calendared squeezed be- tween microsmooth steel and polymer rolls that rotate at different rates, providing a "microslip" action that polishes the tape surface. These manufacturing steps (i.e., materials synthesis, preparation anct han- dling of uniform dispersions, coating, dry- ing, and calendaring) are chemical processes and/or unit operations that are familiar terri- tory to chemical engineering analysis and clesign. INTERNATIONAL COMPETITIVE ASSESSMENT INTRODUCTION In each of the technologies described in the preceding section, U.S. leadership in both fundamental research and manufacturing is severely challenged, and in some cases the United States has been judged to lag behind foreign competitors such as Japan.
CHEMICAL PROCESSING OF MATERIALS AND DEVICES MICROCIRCUITS A recent report of the National Research Council* has assessed the comparative posi- tion of the United States and Japan in ad- vanced processing of electronic materials. The report, which focuses heavily on evaTu- ating Japanese research on specific process steps in the manufacture of electronic mate- rials, provides significant background for the following observations. · The U. S. electronics industry appears to be ahead of, or on a par with, Japanese in- dustry in most areas of current techniques for the deposition and processing of thin films chemical vapor deposition (CVD), metalorganic chemical vapor deposition (MOCVD), and molecular beam epitaxy (MBE). There are differences in some areas, though, that may be crucial to future tech- nologies. For example, the Japanese effort in low-pressure microwave plasma research is impressive and surpasses the U.S. effort in some respects. The Japanese are ahead of their U.S. counterparts in the design and manufacture of deposition equipment, as well. · Japanese industry has a very substantial commitment to advancing high-resolution lithography at the fastest possible pace. Two Japanese companies, Nikon and Canon, have made significant inroads at the cutting edge of optical lithography equipment. In the fields of x-ray and electron-bean~ lithog- raphy, it appears that U.S. equipment man- ufacturers have lost the initiative to Japan for the development of commercial equipment. · Japanese researchers are ahead of their U.S. counterparts in the application of laser and electron beams and solid-phase epitaxy for the fabrication of silicon-on-insulator structures. * Panel on Materials Science, National Materials Ad- visory Board. Advanced Processing of Electronic Ma- terials in the United States and Japan. Washington, D.C.: National Academy Press, 1986. · The Unitecl States leads in basic research related to implantation processes and in the development of equipment for conventional applications of ion implantation. Japan ap- pears to have the initiative in the develop- ment of equipment for ion microbeam tech- nologies. Neither the United States nor Japan has satisfactorily solved the problems of process integration in microcircuit manufacture. As the previous comparisons indicate, much ef- fort is being expended on equipment design for specific processing steps, but a parallel ef- fort to integrate the processing of these ma- terials across the many individual steps has received less attention in both countries. Yet the latter effort might have significant pay- offs in improved process reliability and effi- ciency that is, in "manufacturability." The United States has the capability to take a sig- nificant lead in this area. LIGHTWAVE TECHNOLOGY The Japanese are our prime competitors in the development of lightwave technology. They are not dominant in the manufacture of optical fiber thanks in part to a strong overlay of patents on basic manufacturing processes by U.S. companies. In fact, a major Japanese company manufactures optical fiber in North Carolina for shipment to Japan. This is the only example to date of Japan importing a high-technology product from a U. S. sub- sidiary. Nonetheless, the Japanese are mak- ing strong efforts to surpass the United States, and are reaching a par with the United States in many areas. The United States still significantly leads Japan in producing special purpose and high-strength fibers, in preparing cables from groups of fibers, and in research on her- metic coatings for fibers. RECORDING MEDIA Japan is the United States' principal tech- nological competitor in the manufacture of 31
magnetic media, anct Korean firms are be- ginning to make significant inroads at the Tow end of the market for magnetic tape. U.S. companies producing magnetic tape use manufacturing processes that achieve higher integration through combined unit operations, but Japanese companies have a higher degree of automation in these sepa- rate operations. U. S. companies are ahead of the Japanese in the use of newer thermoplas- tics in caTendar-compliant roll materials. la- pan used to surpass the United States in the product uniformity of magnetic tape for pro- fessional applications; U.S. firms have closed this gap in recent years, and are now capturing worldwide market share from the Japanese, even in Japan. The most significant development in Japan is the entry of photographic film companies (i.e., Fuji and Konishuroku) into the manu- facture of magnetic media. They are having a large impact because the heart of the manu- facturing process is the deposition of thin layers, and chemical processing technology from the photographic film business can be used to improve the quality and yield of magnetic tape. The United States still lags behind Japan in the treatment and manufacture of magnetic particles (except possibly for 3M, which manufactures its particles internally). There are disturbing signs that the Japanese maybe aheact of the United States in the next gener- ation of film base, especially the film base for vapor-deposition magnetic media. The situ- ation is not entirely clear, because 3M and Kodak make their own proprietary film. Other U.S. magnetic meclia companies, though, maybe buying their film technology from Japan in the future. GENERAL OBSERVATIONS AS noted previously, the industries that manufacture high-technology materials and components for information processing and storage are characterized by short product life cycles, enormous competition, and rapid erosion of product value. These industries also need rapid technology transfer from the research laboratory onto the production line. Many of their products cannot be protected by patents, except for minor features. The key to their competitive success is thoroughly characterized and integrated manufacturing processes, supported by process innovations. in the past, much of the process technology on which these industries depend has been developed empirically. If the United States is to maintain a competitive position in these industries, it is essential that we develop the fundamental knowledge necessary to stimu- late further improvement of, anc! innovation in, processes involving chemical reactions that must be precisely controlled in a manu- facturing environment. In the next section the principal technical challenges are set forth. GENERIC RESEARCH ISSUES INTRODUCTION A variety of important research issues are ripe for a substantially increased effort to en- able U.S. companies to establish and main- tain dominance in information storage and handling technologies. These research is- sues are quite broad and cut across the spec- trum of materials and crevices. PROCESS TNTEGRATION Process integration is the key challenge in the design of efficient and cost-effective manufacturing processes for electronic, photonic, and recording materials anc! de- vices. Currently, these products are manu- facturecl by a series of individual, isolated steps. If the United States is to retain a posi- tion of leadership, it is crucial that the overall manufacturing methodology be examined and integrated manufacturing approaches be implemented. Historically, all industries have benefited both economically and in the quality and yield of products by the use of in 32
CHEMICAL PROCESSING OF MATERIALS AND DEVICES tegrated manufacturing methods. As indi- vidual process steps become more complex anc! precise, the final results of manufactur- ing (e.g., yield, throughput, and reliability) often depend critically on the interactions among the various steps. Thus, it becomes increasingly important to automate and inte- grate individual process steps into an overall manufacturing process. The concepts of chemical engineering are easily applied in meeting the challenge of process integration, particularly because many of the key process steps involve chemi- cal reactions. For example, in the manufac- ture of microcircuits, chemical engineers can provide mathematical models and control al- gorithms for the transient and steady-state operation of individual chemical process steps (e.g., lithography, etching, film depo- sition, diffusion, and oxidation), as well as models and associated control algorithms for the interactions between one process step and another, and ultimately between pro- cessing and the characteristics of the final de- vice. As another example, in microcircuit manufacture, chemical engineers can pro- vide needed simulations of the dynamics of material movement through the plant, and thus optimize the flow of crevices (or wafers) through a fabrication line. REACTOR ENGINEERING AND DESIGN Closely related to challenges in process in- tegration are research challenges in reactor engineering and design. Research in this area is important if we are to automate man- ufacturing processes for higher yields and improved product quality. Contributions from chemical engineers are needed to meet this challenge processes such as CVD, epi- taxy, plasma-enhanced CVD, plasma-en- hanced etching, reactive sputtering, and oxi- dation all take place in chemical reactors. At present, these processes and reactors are generally developed and optimized by trial and error. .4~ basic understanding of funda- mental phenomena and reactor design in 33 these areas would facilitate process design, control, anc! reliability. Because each of these processes involves reaction kinetics, mass transfer, and fluid flow, chemical engineers can bring a rich background to the study and improvement of these processes. An important consideration in reactor de- sign and engineering is the ultraclean stor- age and transfer of chemicals. This is not a trivial problem; generally, the containers and transfer media are the primary sources of contamination in manufacturing. Meth- ods are neecled for storing gases and liquids, for purifying them (see the next section), and for delivering them to the equipment where they will be used all the while maintaining impurity levels below ~ part per billion. This purity requirement puts severe constraints on the types of materials that can be used in handling chemicals. For example, materials in reactor construction that might be chosen primarily on the basis of safety often cannot be used. Designs are needed that will meet the multiple objectives of high purity, safety, and low cost. The ultimate limit to the size of microelec- tronic devices is that of molecular climen- sions. The ability to "tailor" films at the mo- lecular level to deposit a film and control its properties by altering or forming the struc- ture, atomic layer by atomic layer opens ex- citing possibilities for new types of devices and structures. The fabrication of these mul- tilayer, multimaterial structures will require more sophisticated deposition methods, such as MBE and MOCVD. Depositing uni- form films by these methods over large dimensions will require reactors with a dif- ferent design than those currently used, es- pecially for epitaxial growth processes. The challenge is to be able to control the flow of reactants to build layered structures tens of atoms thick (e.g., superIattices). To achieve economic automated processes, the reactor design has to allow for the acquisition of de- tailed real-time information on the surface processes taking place, fed back into an ex- quisite control system and reagent delivery
system. This problem gives rise to an excit- ing series of basic research topics. ULTRAPURIFICATION A third research challenge that is generic to electronic, photonic, and recording materi- als and devices stems from the need for start- ing materials that meet purity levels once thought to be unattainable. This need is particularly acute for semicon- ductor materials and optical fibers. For semi- conductor materials, the challenge is to fin ct new, Tower cost routes to ultrapure silicon and gallium arsenide, and to purify other re- agents used in the manufacturing process so that they do not introduce particulate con- tamination or other defects into the device being manufactured. For optical fibers, pre- cursor materials of high purity are also needed. For example, the SiCi4 currently used in optical fiber manufacture must have a total of less than 5 parts per million of hy- drogen-containing compounds and less than 2 parts per billion of metal compounds. Either impurity will result in strong light ab- sorption in the glass fiber. For magnetic me- dia, the challenge is to separate and purify submicron-sized magnetic particles to very exacting size and shape tolerances. A variety of separation research topics have a bearing on these needs. These include generating improved selectivity in separa- tions by tailoring the chemical and steric in- teractions of separating agents, understand- ing and exploiting interracial phenomena in separations, improving the rate and capacity of separations, and finding improved pro- cess configurations for separations. These are all research issues central to chemical . . engineering. CHEMICAE SYNTHESIS AND PROCESSING OF CERAMIC MATERIAES The traditional approach to creating and processing ceramics has been through the grinding, mixing, and sintering of powders. 34 Although still useful in some applications, this technology is being replaced by ap- proaches that rely on chemical reactions to create a uniform microstructure. Among the typical examples of such an approach would be sol-ye! and related processes. A tremen- dous opportunity exists for chemists and chemical engineers to apply their detailed knowledge of funciamental chemical pro- cesses in developing new chemical routes to high-performance ceramics for electronic and photonic applications. Deeper involvement of chemical engi- neers in manufacturing processes for ce- ramics may be particularly important in the eventual commercialization of metal oxide superconductors. The current generation of such superconductors consist of structures that are formed during a conventional ce- ramic synthesis. It is by no means clear that the structures that may produce optimal per- formance in such superconducting ceramics (e.g., room-temperature superconductivity, capacity for high current density are accessi- ble by these techniques. Rational synthesis of structured ceramics by chemical process- ing may be crucial to further improvements in superconducting properties and in afford- ing efficient large-scare production. DEPOSITION OF THIN FILMS Precise and reproducible deposition of thin films is another area of great importance in the chemical processing of materials and devices for the information age. In microelectronic devices, there is a steady trend toward decreasing pattern sizes, and by the end of this decade the smallest pattern size on production circuits will be less than ~ ,um. Although the litho- graphic tools to print such patterns exist, the exposure step is only one of a number of pro- cesses that must be performed sequentially in a mass production environment without creating defects. Precise and uniform depo- sition of materials as very thin films onto
CHEMICAL PROCESSING OF MATERIALS AND DEVICES substrates 15 cm or more in diameter must be performed in a reactor, usually at reduced pressure. Particulate defects larger than 0.! ,um in these films must be virtually nonexis- tent. Low-temperature methods of film dep- osition will be needed so that defects are not generated in previous or neighboring films by unwanted diffusion of dopants. For optical fibers, improved control over the structure of the thin films in the preform will lead to fibers with improved radial gradi- ents of refractive index. A particular chal- lenge might be to achieve this sort of control in preforms created using sol-gel or related processes. Another challenge in depositing thin films on optical fibers occurs in the final coating step. improved coating materials that can be cured very rapidly, for example, by ultravio- let radiation, are neecled for high-speed ~ > 10 mist fiber-drawing processes. Both glassy and elastomeric polymers with Tow glass transition temperatures are needed for use over temperatures ranging from-60 to 85°C or higher. Hermetic coatings are re- quired to avoid water-induced stress corro- sion of silica glasses, which proceeds by slow crack growth. Materials under study inclucle silicon carbide anct titanium carbide applied by chemical vapor deposition, as well as metals such as aluminum. A lO-foIcl increase in the rate at which such coatings can be ap- plied to silica fiber during drawing is needed for commercial success. These coatings must be pinhole-free, have low residual stress, and adhere well. Hermetic coatings will also be needed to protect the moisture-sensitive halide and chaTcogenide glasses that may find use in optical fibers of the future because of their compatibility with transmission at longer wavelengths. Considerable progress in the science and technology of depositing thin films is neecled if the U.S. recording media industry is to remain competitive with foreign manu- facturers. New, fully automated coating pro- cesses that will generate high-quality, Tow- defect media are needed. Not only must con 35 siderable effort be mounted in designing hardware and production equipment, but it is also necessary to develop complex mathe- matical models to gain an understanding of the kinetic and thermodynamic properties of film coating, as well as the effect of non- Newtonian flow and polymer and fluid rhe- ology. A better understanding of dispersion stability during drying, as well as of diffu- sion mechanisms that result in intermixing of sequential layers of macromolecules, is important. MODELING AND THE STUDY OF CHEMICAE DYNAMIC S A challenge related to the problems of re- actor design and engineering is the model- ing and study of the fundamental chemistry occurring in manufacturing processes for semiconductors, optical fibers, and mag- netic media. For example, mathematical models originally developed for continu- ously stirred tank reactors and plug-flow re- actors are applicable to the reactors used for thin-fiIm processing, and can be modified to elucidate ways in which thin-film reactors can be improved. Enabling these models to reach their full descriptive potential will re- quire cletailed studies of the fundamental chemical reactions occurring on surfaces and in the gas phase. For example, etching rates, etching selectivity, line profiles, deposited film structure, film bonding, ancT film prop- erties are determined by a host of variables, including the promotion of surface reactions by ion, electron, or photon bombardment. The fundamental chemistry of these surface reactions is poorly understood, anct accurate rate expressions are particularly needed for electron-impact reactions (i.e., dissociation, ionization, or excitation), ion-ion reactions, neutral-neutral reactions, and ion-neutral reactions. The scale and scope of the effort devoted in recent years to understanding catalytic processes needs to be given to re- search related to film deposition and plasma etching. Until a basic understanding is
achieved of chemical reactions occurring at the surface and in the gas phase, it will be dif- ficult to clevelop new etching systems. Research related to this area has had a de- monstrable impact on recent innovations in plasma processing. Five years ago, it was well known that a fluorine-containing plasma etches silicon at a rate significantly greater than the rate for SiO2, thus offering significant advantages for fabricating inte- grated circuits. However, well-controlled processes could not be developed that would perform in a production environ- ment. The work of chemical engineers in eTu- cidating the relevant chemical reactions and their kinetics was crucial to the identification of the important chemical species in the etch- ing process, their reaction pathways, and, in addition, to the discovery that the organic polymer photoresist contributed to plasma chemistry and selectivity in important ways. These studies led to new, improved plasma processes that are currently being usec! in production. For magnetic media, mathematical models could enhance our fundamental under- standing of the manufacturing processes used to make uniform high-purity magnetic particles. Models for the kinetics and mecha- nisms of reactions and an improved uncler- standing of the thermodynamics of produc- ing inorganic salts are required. ENVIRONMENT AND SAFETY Safety and environmental protection are extremely important concerns in all of the high-technology areas already discussed. They present demanding intellectual chal- lenges. The manufacture of materials and devices for information handling and stor- age involves substantial quantities of toxic, corrosive, or pyrophoric chemicals (e.g., hy- drides and halides of arsenic, boron, phos- phorus, and silicon; hydrocarbons and or- ganic chlorides, some of which are cancer suspect agents; and inorganic acids). Unfor- tunately, the industries involved in manu 36 factoring these materials and devices have only recently begun to employ significant numbers of chemical professionals, and have suffered from a lack of expertise in the safe handling and disposal of dangerous chemicals. Recent studies in California indi- cate that the semiconductor industry has an occupational illness rate 3 times that of gener- al manufacturing industries. Nearly half of these illnesses involve systemic poisoning from exposure to toxic materials. Problems with groundwater contamination in Santa Clara County, California, have also raised concerns about how well the semiconductor industry is equipped to handle waste man- agement and disposal. If the semiconductor and other advanced material industries are to continue to prosper in the United States, it is important that the expertise of chemical engineers be applied to every aspect of chemical handling in manufacturing, from procurement through use to disposal. RECOMMENDATIONS Pursuing the research frontiers discussed in the preceding section will significantly benefit our national standard of living, de- fense, education, and trade balance. How can we best use resources to foster work in these areas, and to foster communication and collaboration among researchers in in- dustrial, academic, and federal laboratories? The following goals should be set for im- proving national research capabilities that will result in improved manufacturing pro- cesses for electronic, photonic, and record- ing materials and devices. · Federal agencies involved in the support of basic materials research (for example, the National Science Foundation tNSF], the U.S. Department of Energy, and the U.S. Department of Defense) should consider un- dertaking new initiatives in the support of fundamental research addressing the ge- neric intellectual issues in the chemical pro
CHEMICAL PROCESSING OF MATERIALS AND DEVICES cessing of electronic, photonic, and record- ing materials and crevices (see the preceding section). · It is particularly important to involve chemists and chemical engineers in research related to ceramic synthesis and processing. Researchers trained in traditional ap- proaches to ceramic materials may not have the optimal background to pursue the new challenges in the molecular clesign, synthe- sis, and engineering of ceramics. · Fundamental research and training to meet the needs of industry for chemical process engineers and scientists should be broadly based in many academic institu- tions, for two reasons. First, many of the re- search areas mentioned in the preceding sec- tion lend themselves to research groups led by single principal investigators or by small teams of two or three coprincipal investiga- tors. The magnitude of support given to such research groups should be enhanced to provide access to the sophisticated instru- mentation neecled to pursue effective re- search on fundamental phenomena impor- tant to research areas such as separations, processing, and reactor design. Second, the demand from the electronics industry alone for personnel with chemical backgrounds is sufficiently large that the founding of a few large centers is not likely to meet the need. Some chemical engineering clepartments, for example, are reporting that up to a quar- ter of their baccalaureate graduates are being hired by electronics firms. · University research, particularly in engi- neering, should be effectively coupled to in- dustry through collaborative mechanisms. industry has been the prime mover in ad- vancing technology in materials and compo- nents for information storage and handling, and will remain so for the foreseeable future. It is important, then, for university research groups to develop and maintain good com- munication with counterpart research groups in industry. The NSF Engineering Research Centers program and Tndustry-University Cooperative Research program are two effec 37 five means to stimulate such communication and collaboration. · A few of the existing NSF Engineering Research Centers are addressing research is- sues that touch on the topics covered in this briefing. Where appropriate, these centers should be encouraged to seek broader partic- ipation in their programs from chemical sci- entists and engineers. · The current undergraduate curriculum in chemical engineering, although it provides an excellent conceptual base for graduates who move into the electronics industries, could be improved by the introduction of in- structional material and example problems relevant to the challenges outlined in this briefing. This would not require the creation of new courses, but the provision of material to enrich existing ones. Seminal texts often serve to redefine the boundaries of a clisci- pline and to direct teaching and research to- ward new frontiers. The NSF should create incentives for select researchers at the cutting edge of chemical engineering to write the next generation of textbooks for their field. The existing network of programs in fund- ing agencies do not address some important problems in the generation and transfer of expertise and ideas from the research labora- tory to the production line. For the technolo- gies cliscussed in this report, a key role in generating new process concepts and equip- ment is played by a large number of rela- tively small firms. These firms are generally not in a position to make financial contribu- tions to Engineering Research Centers or to retain academic consultants, yet face impor- tant research problems in fundamental sci- ence and engineering that would benefit markedly from the insights of academic re- searchers. The United States could signifi- cantly boost its competitive position in the technologies discussed in this report by facil- itating information transfer between aca- clemia and this segment of industry. The problem for funding agencies with an inter- est in promoting U. S. capabilities in this area
is how to create incentives for academic re searchers to seek out and forge links to the small firms that stand at the crucial step be- tween laboratory research and production processes. Examples of two possible mecha- nisms that provide these incentives follow. · Agencies such as the NSF could create a new sabbatical award for academic research- ers to spend up to a year in the laboratory of a small process technology firm. The rationale for such a program would be both to provide a critical sector of U.S. advanced process technology firms with the latest insights from university research, and to provide university researchers with insights into the ways in which fundamental science and en- gineering can contribute to the practice problems of high-technology processing of materials and devices. · A limited number of "incubator re- search programs," providing state-of-the- art facilities cohabited by researchers from 1 38 advanced process technology firms and re- searchers in process engineering associated with universities, could be set up in close proximity to academic research campuses. Key to these programs would be the contri- bution by industry of high-quality research personnel, in lieu of providing financial sup- port for academic research conducted under these programs. The government might pro- vide a significant portion of the facility costs to those university applicants that could as- semble a critical mass of researchers from their own departments and from high-tech- nology firms. The concept of "incubators" is not novel, and past attempts to translate such a concept into reality have met with success on some occasions and failure on others. The panel believes that a solicitation of proposals emphasizing interactions be- tween academia and the small process tech- nology companies that are capital-poor but problem-rich would prove a worthwhile ex- periment with a good chance of success. /
