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.
38 _ MOST EVERY ASPECT of our lives at _. work, at home, and in recreation has been affected by the information rev- olution. Today, information is collected, proc- essed, displayed, stored, retrieved, and trans- mitted by an array of powerful technologies that rely on electronic microcircuits, light wave communication systems, magnetic and optical data storage and recording, and electrical inter- connections. Materials and devices for these technologies, along with photovoltaic materials and devices, are manufactured by sophisticated chemical processes. The United States is now engaged in fierce international competition to achieve and maintain leadership in the design and manufacture of these materials and devices. The economic stakes are large (see Table 4.11; national productivity and security interests dic- tate that we make the strongest possible effort to stay ahead in processing science and tech- nology for this area. In the manufacturing of components for in- formation and photovoltaic systems, there has been a long-term trend away from mechanical production and toward production by chemical processes. Chemists and chemical engineers have become increasingly involved in several areas of research and process development. Worldwide, however, many high-technology in- dustries, such as microelectronics, still have surprisingly little strength in chemical process- ing and engineering. The United States has a particular advantage over its international com- petitors in that its chemical engineering research TABLE 4.1 Worldwide Market for Materials and Devices for Information Storage and Handling (billions of 1986 dollars) Year Technology 1985 1990a 1995a Electronic semiconductors Light wave fiber and devices Recording materials Interconnections Photovoltaics Total electronics 25 60 160 5.5 55 58 20 10 21 0.8 550 a Market projection. SOURCE: AT&T Bell Laboratones. Compiled from var- ious published sources. FRONTIERS IN CHE^~AL ENGINEERING community leads the world in size and sophis- tication. The United States is in a position to exploit its strong competence in chemical pro- cessing to regain leadership in areas where the initiative in manufacturing technology has passed to Japan and to maintain or increase leadership in areas of U.S. technological strength. Table 4.2 illustrates some of the ways in which chemical engineering can contribute to research on information and photovoltaic ma- terials and devices. To fully achieve its potential contribution, though, the field of chemical en- gineering must strongly interact with other dis- ciplines in these industries. Chemical engineers must be able to communicate across disciplinary lines, as the technologies discussed in this chapter involve solid-state physics and chem- istry, electrical engineering, and materials sci ence. Electronic, photonic, and recording materials and devices may seem to be an exceedingly diverse class of materials, but they have many characteristics in common: their products are high in value; they require relatively small amounts of energy or materials to manufacture; they have short commercial life cycles; and their markets are fiercely competitive-conse- quently, these products experience rapid price erosion. The manufacturing methods used to produce integrated circuits, interconnections, optical fibers, recording media, and photovol- taics also share characteristics. All involve a sequence of individual, complex steps, most of which entail the chemical modification or syn- thesis of materials. The individual steps are designed as discrete unit or batch operations and, to date, there has been little effort to integrate the overall manufacturing process. Chemical engineers can play a significant role in improving manufacturing processes and tech- niques, and investments in chemical processing science and engineering research represent a potentially high-leverage approach to enhancing our competitive position. n.a. CURRENT CHEMICAL MANUFACTURING PROCESSES Before the invention of the transistor in 1948, the electronics industry was based on vacuum
39 Cal v ._ ._ so Ct ._ s~ o v v ._ o v o v - o Cq v C~ .= ._ - Ct v ._ V .O o .= Ct · ._ ~ ~ - .o oc Ct Ct ~ ~ (,, - Ct o~ s~ .o ~ o ~ o O V) ._ ;> ._ - o o o .~ C~ ._ C) ._ o :~ ._ ._ ._ oc C~ ~: o ._ ~ D ._ ._ s o ~ O V 0 1 ._ 0£ C~ O ~ ·~- E ~o 4, o ~ E ° E ~E ~, E o o ~ ~{,, ~ ~ =~V ~V ~ C~ ~ o ~ ~ ~ ~o E o 0, ~ o ~o s~ 0 ~ O _ 5 ·- ;_ ~ _ D ~ ~3 3 C~ ~8 Ct cn ~ ~ ~ o . ~ o o c, ~o~ c ~o .C o ~ .o ~ ~ ~ ~ ~ ~ ~ o o~= ~C =m C/7 ~ o -o 3 ~ o c E ' , C ~ C ,= C E ° Y - _ ~ ~ ~v ~ c ~ ~ E ~ E o ' E c, ~ ° . ° ~ ~ ~ ~ ~ E E ° 3 E - ~, _ _ ~oV ~ ~Z c<5 9 ~_ O O C) .o ~ o '' ' o V) .~_ _ ,C 2~ ~·C . ~ ~ ~ sO ~O ~
tube technology, and most electronic gear was assembled on a metal chassis with mechanical attachment, soldering, and hand wiring. All the components of pretransistor electronic prod- ucts-vacuum tubes, capacitors, inductors, and resistors were manufactured by mechanical processes. A rapid evolution occurred in the electronics industry after the invention of the transistor and the monolithic integrated circuit: · Today's electronic equipment is filled with integrated circuits, interconnection boards, and other devices that are all manufactured by chemical processes. · The medium used for the transmission of information and data over distances has evolved from copper wire to optical fiber. It is quite likely that no wire-based information transmis- sion systems will be installed in the future. The manufacture of optical fibers, like that of mi- crocircuits~ is almost entirely a chemical cro cess. · Early data storage memory was based on ferrite core coils containing a reed switch that mechanically held bits of information in either an on or off state. Today, most data is perma- nently stored through the use of magnetic ma- terials and devices, and the next generation of data storage devices, based on optoelectronic materials and devices, is beginning to enter the marketplace. Ferrite cores were manufactured by winding coils and mechanically mounting the individual memory cells in large arrays. Mag- netic and optical storage media are manufac- tured almost entirely by chemical processes. The importance and sophistication of current chemical manufacturing processes for elec- tronic, photonic, and recording materials and devices are not widely appreciated. A more detailed description serves to highlight their central role in these technologies. Microcircuits A semiconductor microcircuit is a series of electrically interconnected films that are laid down by chemical reactions. The successful growth and manipulation of these films depend heavily on proper design of the chemical reac · ~ ~ _ . _ ~ ~ ~ ~ _ ^\ :~ A ;1 ~ 4 = ~ ~ ~ ~ ^: ~ ~ ~ Al tip tors in which they are laid down, the choice of chemical reagents, separation and purification steps, and the design and operation of sophis- ticated control systems. Microelectronics based on microcircuits are commonly used in such consumer items as calculators, digital watches, personal computers, and microwave ovens and in information processing units that are used in communication, defense, space exploration, medicine, and education. Microcircuitry has been made possible by our ability to use chemical reactions and processes to fabricate millions of electronic components or elements simultaneously on a single sub- strate, usually silicon. For example, a 1-million- bit dynamic random access memory device (Figure 4.1) contains 1.4 million transistors and 1 million capacitors, with some chemically etched features on the chip being as small as 1.1 ~m. This stunning achievement is just one step in a long-term trend toward the design and produc- tion of integrated circuits of increasing com- plexity and capability. There is still considerable room for further increases in component density in silicon-based microelectronics (Figure 4.2), not to mention possible advances in component density that would result from alternative meth F1GURE 4.! Chemical reactions are used to achieve the fine structures seen in modern integrated circuits. This electron micrograph shows a transistor in a '~cell" of a I- megabit dynamic random access memory chip. The distance between features is about ~ Am. Courtesy, AT&T Bell Laboratories.
7, PHOTON`C, AV~ RE£~15iG MATER47~S ii.~D DE-~S 109 1o8 I ~ 106 CL Oh z z o o 107 105 104 103 1 o2 - ^ 1 960 SILICON CHIPS IN PRODUCTION \ PHYSICAL LIMIT Galas CHIPS UNDER ~ DEVELOPMENT 1970 YEAR 1 980 1 990 FIGURE 4.2 The chemical processes used for the manufacture of microcircuits have become progressively more sophisticated. This development is respon- sible for the large increases in the number of components that can be placed on a single chip. Trends in increasing component density are shown from 1960 for silicon chips (top line) and from 1975 for developmental chips based on gallium arsenide (bottom line). The rates of growth shown have been remarkable. From 1962 to 1972, silicon component density increased a thousandfold and from 1972 to 1982, a hundredfold. From 1975 to 1985, component density in developmental gallium arsenide devices grew by a factor of 40,000. Courtesy, AT&T Bell Labortories. oafs of storing and transferring information (e.g., three-dimensional circuits and Josephson (quan- tum) or optical devices). Chemical reactions and processes in the man- ufacture of microcircuits (Figure 4.3) begin with the basic material for integrated circuits, high- purity (less than lSO parts per trillion of impur- ities) polycrystalline silicon. This ultrapure sil- icon is produced from metallurgical grade (98 percent pure) silicon by the following steps (Figure 4.41: · reaction at high temperature with hydrogen chloride to form a complex mixture containing trichlorosilane; · separation and purification of of trichloro- silane by absorption and distillation; and · reduction of ultrapure trichlorosilane to polycrystalline silicon by reaction with hydro- gen at 1,100-1,200°C. To prepare single-crystal silicon ingots suit- able for use as materials in semiconductors, polycrystalline silicon is melted in a crucible at 1,400-1,500°C under an argon at mosphere. Tiny quantities of dop ants compounds of phospho rus, arsenic, or boron are then added to the melt to achieve the desired electrical properties of the finished single-crystal waf ers. A tiny seed crystal of silicon with the proper crystalline ori entation is inserted into the melt and slowly rotated and with drawn at a precisely controlled rate, forming a large cylindrical single crystal 6 inches (14 cm) in diameter and about as tall as an adult human being (1.8 m) with the desired crystalline orienta tion and composition. Crystal growth kinetics, heat and mass transfer relationships, and chem ical reactions all play important roles in this process of controlled growth. The resulting single crystal ingots are sawed into waf ers that are polished to a flatness in the range of from 1 to 10 ~m. The next steps in device fab rication are the sequential deposition and pat terning of thin dielectric and conducting films (Figure 4.51. The polished silicon wafer is first oxidized in a furnace at 1,000-1,200°C. The resultant silicon dioxide film is a few hundred nanometers thick and extremely uniform. The wafer is then coated with a photosensitive polymeric material, termed a resist, and is exposed to light through the appropriate pho tomask. 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 removes unwanted portions, and the resulting pattern serves as a mask for chemically etching the pattern into the silicon dioxide film. The resist is then removed with an oxidizing agent such as a sulfuric acid-hydrogen peroxide mix ture, and the wafer is chemically cleaned and ready for other steps in the fabrication process. The patterned wafer might next be placed in a diffusion furnace, where a first doping step is performed to deposit phosphorus or boron into
~2 the holes in the oxide. A new oxide film can then be grown and the photoresist process repeated. As many as 12 layers of conductor, semiconductor, and dielectric materials are de- posited, etched, and/or doped to build the three- dimensional structure of the microcircuit. Light Wave Media and Devices Photonics involves the transmission of optical signals through a guiding medium generally a FRONTIERS IN CHEMICAL ENGINEERI~G FIGURE 4.3 The manufacture of integrated circuits requires both expertise in electronic design and chemical processing. Chemical process steps are important to the preparation of silicon materials, to the steps from oxidation of silicon wafers through establishment of bonding pads, and to the final assembly of chips in individual packages. Excerpted by special permission from Chemical Engineering, June 10, 1985. Copyright 1985 by McGraw-Hill, Inc., New York, NY 10020. glass fiber for purposes that include telecom- munications, data and image transmission, en- ergy transmission, sensing, display, and signal processing. Optical fiber technology is less than 14 years old and only became a commercial reality in the early 1980s. It is now a $1 billion per year industry. The data-transmitting capac- ity of optical fiber systems has doubled every year since 1976 (Figure 4.61. In fact, optical fiber systems planned on the basis of the pre- vailing technology at that time are often obsolete
ELECTRONIC, PHOTONIC, AND RECORDING MATERIALS AND DEVICES Si + HCI · SiHCI3 + SkHyClz [2] SiHC13 + H2 Distillation 1,100°C HCI + SiXHyClz Metallurgical silicon ~,~ ~, ~_ SixHyClz FIGURE 4.4 The production of polycrystalline silicon for the electronics industry involves several chemical steps aimed at the reduction of impurities. These include (1) reaction of metallurigcal grade silicon to produce a mixture of chlorosilanes, (2) distillation of trichorosilane, and (3) reduction of trichloro- silane to polycrystalline silicon. Excerpted by special permission from Chem- ical Engineering, June 10, 1985. Copyright 1985 by McGraw-Hill, Inc., New York, NY 10020. by the time they are implemented. Typically, a given light wave technology is supplanted by an improved technology after one year. Other applications for light guides, such as optical fiber sensors and transducers, are re- ceiving a great deal of attention. Image trans- mission (e.g., endoscopes), energy transmission (e.g., light pipes), and display (e.g., decorative signs) are growing commercial areas. Light wave media and devices include the guiding medium (optical fibers), sending and receiving devices, and associated electronics and circuitry. The transmission of light signals through optical fibers must occur at wavelengths where the absorption of light by the fiber is at a minimum. Typically, for SiO2/GeO2 glass, the best transmission windows are at 1.3 or 1.5 Am (Figure 4.71. Optical signal processing for integrated optics and optical computing is in a rudimentary state but is certain to be an important area for future technological development. The processing in- volved in making optoelectronic devices is very similar to that used in microcircuit manufacture, but with considerable utilization of Group III- V compound semiconductors, lithium niobate, and a variety of polymeric materials. Devel- opmental manufacturing processes for optoelec ~3 SiHCI3 [3] Polycrystalline silicon tropics emphasize reactive ion etching, epitaxy (e.g., metalorganic chemical vapor deposition (MOCKED), vapor-phase epitaxy, and molecu- lar-beam epitaxy (MBE)), and photochemical and beam processing techniques for writing circuit configurations. All these processes are based on chemical reactions that require precise process control to produce useful devices. Optical fibers are made by chemical process- es. The critical feature of an optical fiber that allows it to propagate light down its length is a core of high refractive index surrounded by a cladding of lower index. The higher index core is produced by doping silica with oxides of phosphorus, germanium, and/or aluminum. The cladding is either pure silica or silica doped with fluorides or boron oxide. There are four principal processes that may be used to manufacture the glass body that is drawn into today's optical fiber. "Outside" processes-outside vapor-phase oxidation and vertical axial deposition-produce layered de- posits of doped silica by varying the concentra- tion of SiCl4 and dopants passing through a torch. The resulting "soot" of doped silica is deposited and partially sintered to form a porous silica boule. Next, the boule is sintered to a pore-free glass rod of exquisite purity and trans
~( SILICON INGOT ¢' SILICON WAFER -i\// Inky\\: -A Mask Pnsitiv~ / \ Negative \,~ Develop tCh ~ Strip FIGURE 4.5 Chemical steps in photolithography. A sim- plified series of steps in photolithography is shown. A silicon wafer, taken from a single-crystal silicon ingot, is coated with a polymer resist that is sensitive to light. A mask is placed over the wafer and the resist is thus selectively exposed to light. Depending on the type of polymer coating used, two things can happen. If the polymer is a positive resist, exposure to light makes the polymer easier to dissolve in a solution during the development step. After the development step, a protective film is left on the wafer that is the image of the mask used. If the polymer is a negative resist, exposure to light makes the polymer more difficult to dissolve during the development step. After- wards, a protective film is left on the wafer that is the opposite of the image on the mask used. A corrosive gas or liquid is then used to etch away those parts of the wafer unprotected by the resist film. The resist film is removed after etching in preparation for other process steps. Ex- cerpted by special permission from Chemical Engineering, June 10, 1985. Copyright 1985 by McGraw-Hill, Inc., New York, NY 10020. FRONTIERS iN CHEMICAL ~,YGI^~EER`~N'6 F 107 By e. 1 o6 - c~ 105 _ id '09 104 _ 103 lot 1 1 1 1 1 975 [PHYSICAL LIIIIIT 109tl I'm' 9: tOUBlES O ~ O ElERY YEAR . ~ ._ 1 980 1 985 1 g90 YEAR FIGURE 4.6 Since 1975, both the capacity of optical fiber and the distance a signal can be carried on optical fiber have steadily increased. Courtesy, AT&T Bell Laborato- r~es. parency. "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 (Figure 4.8~. In PCVD, the reaction is initiated by a microwave plasma. More than a hundred different layers with dif- ferent refractive indexes (a function of glass composition) may be deposited by either pro- cess before the tube is collapsed to form a glass rod. In current manufacturing plants for glass fiber, the glass rods formed by all the above processes are carried to another facility where they are E _ In 1 a) o 0.~ 0.1 _ \ . 1250nm 1390nm 1600nrr Sl-OH COMB Sl-OH P-OH SHARP SHARP BROAD POOH ` ~ADDED RAYLEIGH SCATTERI NG \-4 1 , 1 1 1 1 1 ~ 1 0.8 1.0 1.2 1.4 1.6 WAVELENGTH (~1 m) FIGURE 4.7 Transmission losses in glass fibers carrying optical signals are due to the interaction of light with chemical bonds. From 1.2 am to 1.6 am, losses due to Rayleigh scattering in the fiber are minimized, but trans- mission losses from Si-OH and P-OH bonds become large. The lowest transmission losses occur at wavelengths of 1.3 and 1.5 ~m. Courtesy, AT&T Bell Laboratories.
ELECTR0~G, PHOT0~C, AND RECORDING MATERIALS AND DEVICES FLOW METERS, MASS-FLOW CONTROLLERS O2 AND MANIFOLD ~5 r FUSED SILICATUBE 026], 02, 1 : ~1__O2 POCQ3 | GeCQ4 ~7 V ;;;;;;'; Cal MULTIBURNER ~ DEPOSITED LAYER TORCH ~OFCOREG~SS 1 O2 H2 I L BC]3 SiF4 SiC]A SF6 Ct FREON ~ / BUBBLERS drawn into a thin fiber and immediately coated with a polymer. The polymer coating is impor- tant; it protects the fiber surface from micro- scopic scratches, which can seriously degrade the glass fiber's strength. Current manufacturing technologies for op- tical fiber are expensive compared with the low cost of commodity glass. U.S. economic com- petitiveness in optical technologies would be greatly enhanced if low-cost means were found for producing wave guide-quality silica glass. The manufacture of glass lends itself to a fully integrated and automated (i.e., continuous) process. One can envision a fiber manufacturing plant that moves from purification of chemical reagents to a series of chemical reactions, glass- forming operations, and, finally, fiber-drawing steps. Intermediate products would never be removed from the production line. Sol-gel and related processes (see Chapter 5) are attractive candidates for such a manufacturing technology, which would start with inexpensive ingredients and proceed from a sol to a gel, to a porous silica body, to a dried and sintered glass rod, to drawn and coated fibers. Such a process TRANSLATION FIGURE 4.8 Modified chemical. vapor deposition (MCVD) is one of the principal processes used to manufacture optical fiber. In MCVD, a mixture of gases (O', POCK, SiCl4, GeCl4, BC13, SiF4, SF6, Cal, and freon) pass down the interior of a hollow silica tube that is being externally heated by a moving flame. The gases react to form a fine layer of silica glass doped with constituents of the gaseous mixture. Many layers can be deposited before the silica tube is collapsed and drawn into optical fiber. Courtesy, AT&T Bell Laboratories. 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 competi- tiveness of light wave technologies. At present the chemical steps involved in sol- gel 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-n~-mb-er of formats (e.g., magnetic tape, magnetic disks, or optical disks) and are made by a variety of materials and processes (e.g., evaporated thin films or deposited magnetic particles in polymer ma- trixes). The next generation of recording media will be based on optical storage of data (Figure 4.91. Already, read-only optical disks (or CD- ROMsJ are on the market for applications such as search and retrieval of information from large data bases. And optically based compact disks (CDs) are available in every record store. The possibility of creating optically based recording
~6 media for read-write storage of information has generated a tremendous amount of industrial research but, so far, no commercially viable products. Since a practical read-write optical disk has not yet been invented, it is hard to describe the processing challenges involved in making it. Thus, the remainder of this section examines the most challenging (from a processing stand- point) of the remaining forms of recording me- dia: magnetic disks and tape. Magnetic media are still an economically important part of the recording market and have a rich array of processing challenges with which chemical en- gineers have been involved. These challenges are relevant to the emerging technologies and materials in recording. In the manufacture of magnetic recording media, the chemical and physical properties of the magnetic particles or thin films coated on a 1o1o 109 - .~ y 1 o8 in ._ ~5 o .~ ~ 10 co a) 1o6 Optical disks _ ~ - - over halide| ~Limit Demonstrated Magnetic disk / Product density / ~ Low estimate of __ / / current optical / / density / 3370 ~ 3350 / 3033-11 ~/3033 C IBM 2314 05 1965 1 970 ^.~0 1980 1990 2000 FIGURE 4.9 The density at which information can be written on optical disks (measured in bits/in) was demon- strated in the early 1980s to be 10 times greater than the current highest performance magnetic disk. The gap be- tween optical and magnetic storage capabilities is projected to increase over the next 20 years. Reprinted with permis- s~on from Electronic Design, August 18, 1983, 141. .~NTIERS IN CHEMICAL ENGINEERING disk or tape are very important, for they deter- mine the density at which information can be recorded. Paramount among these properties are the shape, size, and distribution of the magnetic particles. An extremely narrow size range of magnetic particles themselves only a few tenths of a micrometer in diameter must be achieved in a reliable and economic manner. Furthermore, the particles must be deposited in a highly oriented fashion and lie as closely together as possible, so that high recording densities can be achieved. To accomplish this, a variety of challenging problems must be solved in the chemistry and chemical engineering of barium ferrite and the oxides of chromium, cobalt, and iron (e.g., the synthesis and pro- cessing of micrometer-sized materials with spe- cific geometric shapes). The manufacture of magnetic tape illustrates an interesting sequence of chemical processing challenges (Figure 4.101. A carefully prepared dispersion of needle-like magnetic particles is coated onto a fast-moving (150-300 m/min) poly- ester film base 0.0066-0.08 mm thick. The ability to coat thin, smooth layers of uniform thickness is crucial. The particles, after being coated onto the film, are oriented in a desired direction either magnetically or mechanically during the coating process. After drying, the tape is cal- endered (squeezed between 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 and handling of uniform dispersions, coating, drying, and calendering) are chemical processes and/or unit operations that are familiar territory to chemical engineering analysis and design. Materials and Devices for Interconnection and Packaging Interconnection and packaging allow elec- tronic devices to be usefully incorporated into products. The manufacture of complex elec- tronic systems requires that hundreds of thou- sands of electronic components be efficiently connected with one another in an extremely small space. In the past, this was accomplished by hand wiring discrete components on a chassis
ELECTRONIC, PHOTONIC, AND RECORDING MATERIALS A.~D DEVICES (4) (2) Drying oven air flotation (1) Coating head air flotation (5) Coating ~reverse roll (shown) (shown) Calender mix ~gravure knife ~l /1 ~ | (3) Ferrite magnets \\ Sandmill Dispersing If ~ ~ J ~ Coating head reverse roll (shown) (shown) gravure knife l it' Mix In: _ Enlargement of sandmill dispersing chamber Rotating Blades \; Media FIGURE 4.10 The manufacture of magnetic tape involves a series of steps including (1) forming a uniform dispersion of coating mix, (2) applying this coating to the film base, (3) orienting the magnetic particles, (4) drying the magnetic coat in an air-flotation oven, and (5) calendering and final wind-up on spools. Chemical processes are central to several of these steps. Excerpted by special permission from Encyclopedia of Chemical Technology, 3rd ea., Vol. 14, p. 745. Copyright 1978 by McGraw-Hill, Inc., New York, NY 10020. assembly. Today, interconnection technology is based on high-density printed wiring boards, often with as many as 30 parallel layers of interconnection. The board insulation substrate may be either polymer or ceramic, with appro- priate metal conductors. ~7 an\ T electro magnets Unwind Base film Breakaway of base film to ~_ / show air slots ;~ co: her Base \- Enlargement of alr-flotatlon oven creating sinusoidal wave form of base film - The dielectric and the conductors are selected to maximize data transmission speed while min- imizing signal loss. In addition, dissipating heat generated by the microcircuits is rapidly becom- ing an important consideration. If too much heat builds up in the microelectronic device,
the device will start to fail. The next generation of interconnec- tion substrates will likely require new materials and assembly techniques to cope with this chal- lenge. Plastic packaging of micro- electronic devices is the most cost-effective means of provid- ing electrical, mechanical, and environmental protection for an integrated circuit device. Ap- proximately 80 percent of the billions of these devices used in the United States each year are packaged in plastic, and most are packaged in thermoses molding compounds by a conventional transfer molding process. This process must be carried out with exceptionally high yield, produc- tivity, and reliability if the United States is to achieve a competitive advantage in the packaging of these devices. All modern interconnection devices are manufactured by chemical processes such as etch- ing, film deposition, and ceramic forming. Substrate formation is a vital part of the manufacturing process for dberglass-impregnated printed wir- ing boards. It utilizes chemical processes such as metal deposition, lithographic patterning, etching, and chemical cleaning. Process design, to improve quality and decrease cost, continues to be a challenge, particularly in terms of greater uniformity in plating and etching and the envi- ronmental problems posed by the disposal of spent etching and plating baths. Ceramic boards are currently widely used in high-performance electronic modules as inter- connection substrates. They are processed from conventional ceramic precursors and refractory metal precursors and are subsequently fired to the final shape. This is largely an art; a much bet- ter fundamental understanding of the materials and chemical processes will be required if low- cost, high-yield production is to be realized (see '~§T'~j,-~5 [any i~7~:,F~7p4~: ~ ,~, !./ '~t'7/I45 E - A reference plane layer ~ :~/ 1 tJ~ US Act ~ metallurgy and _ redistribution layers Signal distribution and signal reference t -~ ~ layers AT l: Power distribution and module I/O layers F - A power plane layer FIGURE 4.11 Cross-section of the IBM Multilayer Ceramic interconnection package. Various layers in this interconnection device are shown. Copyright 1982 by International Business Machines Corporation. Reprinted with per . . mission. Chapter 51. A good example of ceramic intercon- nection boards are the multilayer ceramic (MLC) structures used in large IBM computers (Figure 4.111. These boards measure up to 100 cm in area and contain up to 33 layers. They can in- terconnect as many as 133 chips. Their fabrica- tion involves hundreds of complex chemical processes that must be precisely controlled. Better insulating substrates will be required for the ultrahigh-speed interconnection modules of the future. Although organic polymers appear to offer cost advantages for substrate applica- tion, the large-scale production of organic sub- strates with sufficiently low dielectric constants has not yet been realized. Achieving this will require the scale-up of new chemical reactions and the development of process methods for new polymer substrates.
~ ~ -A ,< ,=\1 ~ ~ ~ ~v~x:~/ is- Ail Am. :~3 (4A: :~ i W5 Photovoltaics A photovoltaic cell is a solid-state device that transforms solar energy into electricity. Signif- icant research on photovoltaic cells began in the early 1970s and until now has generally focused on the invention and improvement of specific devices, including single-crystal silicon cells, amorphous silicon cells, heterojunction thin-film cells, and gallium arsenide cells. Gallium arsenide cells have achieved high- energy efficiencies (in excess of 20 percent) and have been successfully used in systems where there is magnification of the sun's rays, that is, concentrator systems that increase the light flux per unit area. Substantial research is being devoted to improving the conversion efficiency and lowering the cost per kilowatt of such devices. The materials and processes used in the manufacture of photoelectric energy conversion devices are almost identical to those used in manufacturing microelectronic devices and in- tegrated circuits. The photovoltaics industry could expand rap- idly if the efficiency of polycrystalline modules could be increased to 15 percent, if these mod- ules could be built with assurance of reliability over a 10- to 20-year period, and if they could be manufactured for $100 or less per square meter. Solar energy research has been largely directed toward only one of these issues: effi- ciency. All research aimed at reducing manu- facturing costs has been done in industry and has been largely empirical. Almost no funda- mental engineering research has been done on either the laboratory scale or the pilot plant scale for cost-effective processes for the pro- duction of photoconverters. Superconductors Superconductivity has been known since 1911, and superconducting systems based on various metal alloys (e.g., NbTi and Nb3Sn) are cur- rently used as magnets and in electronics. These materials exhibit superconductivity only at tem- peratures below 23 K and require cooling by liquid helium. The discovery of ceramics that exhibit superconductivity at temperatures up to 120 K, the so-called high-temperature super- conductors, has sparked a tremendous amount of scientific activity and commercial interest around the world. The key to the superconducting properties of these ceramics seems to be the presence of planes of copper and oxygen atoms bonded to one another. The significance of the other atoms in the lattice seems to be to provide a structural framework for the copper and oxygen atoms. Thus, in the superconducting compound YBa'Cu3O, the substitution of other rare earths for yttrium results in little change in the prop- erties of the material. Currently, superconducting materials are pro- duced by standard techniques from the ceramics industry: mixing, grinding, and sintering. The basic structure of the 95 K superconductors is formed at temperatures above 800-900°C, and then annealed with oxygen at a temperature below 500°C. More needs to be known about the effect of various synthesis conditions on the microstructure found in these materials. Alter- native routes to ceramics, such as sol-gel pro- cesses, may lead to significant improvements in the production of these materials. For micro- electronic applications, various chemical vapor deposition routes to these materials need to be investigated. MOCVD might be a particularly promising route if volatile precursor compounds could be discovered and synthesized. Chemical engineers have the needed background for de- veloping this technology (see Chapter 5) as well as finding the optimal procedure for drying, sintering, and calcining the final product. INTERNATIONAL COMPETITION In each of the technologies described in the preceding section, U.S. leadership in both fun- damental research and manufacturing is se- verely challenged, and in some cases the United States is lagging behind its foreign competitors. Microcircuits A recent report of the National Research Councili has assessed the comparative position
FRONTIERS IN AWAY ENGI1VEERING of the United States and Japan in advanced processing of electronic materials. The report, which focuses heavily on evaluating Japanese research on specific process steps in the man- ufacture of electronic materials, provides sig- nificant background for the following observa- tions: · The U.S. electronics industry appears to be ahead of, or on a par with, Japanese industry in most areas of current techniques for the deposition and processing of thin films chem- ical vapor deposition (CVD), MOCVD, and MBE. There are differences in some areas, though, that may be crucial to future technol- ogies. For example, the Japanese effort in low- pressure microwave plasma research is impres- sive 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 li- thography 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-beam lithography, it appears that U.S. equipment manufacturers 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. · The United States leads in basic research related to implantation processes and in the development of equipment for conventional ap- plications of ion implantation. Japan appears to have the initiative in the development of equip- ment for ion microbeam technologies. There is a penalty to be paid for falling behind foreign competitors in process equipment design and engineering. Early access to new prototypes of equipment allows a manufacturing firm to concurrently troubleshoot the equipment and integrate it into its existing process line. When the state-of-the-art processing equipment comes from overseas, companies in the country of origin gain a competitive advantage stemming from this early access. A look at the installation record for JEOL focused ion beam instruments (which are the best in the world) illustrates this phenomenon (Table 4.31. Much remains to be done, in both the United States and Japan, to solve the problems of process integration in microcircuit manufacture. Effort is being expended on equipment design for specific processing steps, but a parallel effort to integrate the processing of semiconductor materials and devices across the many individ- ual steps has received less attention in both countries. Yet the latter effort may have signif- icant payoffs in improved process reliability and efficiency- that is, in "manufacturability." The United States, with the strongest chemical en- gineering research community in the world, has the capability to take a significant lead in this area. Light Wave Media and Devices The Japanese are our prime competitors in the development of light wave 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. com- panies. 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 prod- uct from a U.S. subsidiary. Nonetheless, the Japanese are making strong efforts to surpass the United States and are reaching a par with us 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 hermetic coatings for fibers. Recording Media Japan is the America's principal technological competitor in the manufacture of magnetic me- dia, and Korean firms are beginning to make significant inroads at the low end of the magnetic tape market. U.S. companies producing mag- netic tape use manufacturing processes that
ELECTRONIC, PHOTONIC, AND RECORDIATO i'~ATERIALS AND DEVICES TABLE 4.3 Installation Record for Focused Ion Beam Instruments Made by JEOL Semiconductor Equipment Division Instrument Instrument Year Number Customer Country Type Installed 8 10 12 13 14 16 18 19 20 21 22 23 24 25 1 The Institute of Physical and Japan JIBL-34 1982 Chemical Research 2 The Institute of Physical and Japan JIBL-100 1983 Chemical Research 3 Optoelectronics Joint Japan JIBL-100 1983 Laboratory 4 LSI R&D Lab, Mitsubishi Japan JIBL-100 1983 Electric Corporation 5 Fujitsu Laboratories Ltd.- Japan JIBL-1OOA 1984 AtSugl Optoelectronics Joint Laboratory Institute of Laser Engineering, Osaka University The Institute of Physical and Chemical Research NTT Musashino Electrical Communication Laboratories Institute of Industrial Science , Tokyo University Fujitsu Laboratories Ltd. Atsugi Dainippon Screen Fujitsu Laboratories Ltd. Atsugi NTT Atsugi Electrical Communication Laboratories Tsukuba Research Center, Sanyo Electric Co., Ltd. NEC Corporation LSI R&D Lab, Mitsubishi Electric Corporation Optoelectronics Joint Laboratory Institute of Industrial Science, Tokyo University Matsushita Laboratory Nihon Denso Sony Denka Max Planck Institute IBM Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Japan Germany U.S.A. JIBL-1OOA JIBL-30 JIBL-200 JIBL-100 JIBL-100 JIBL-GP1 JIBL-100 JIBL-100 JIBL-100 1984 1984 1984 1984 1984 1984 1984 1984 1984 JIBL-1OOA 1984 JIBL- 140 JIBL- 140 IPMA-10 IPMA-10 JIBL-106 JIBL-106 JIBL-GPI JIBL-100 JIBL-1OOA JIBL- 106 1984 1984 1984 1986 1986 ? ? ? ? 1988 SOURCE: AT&T Bell Laboratories and JEOL. achieve higher integration through combined unit operations, but Japanese companies have a higher degree of automation in these separate operations. U.S. companies lead the Japanese in the use of newer thermoplastics in calender- compliant roll materials. Japan used to surpass the United States in the product uniformity of magnetic tape for professional applications; U.S. 51 firms have closed this gap in recent years and are now capturing worldwide market shares from the Japanese, even in Japan. The most significant development in Japan 1S the entry of photographic film companies (Fuji and Konishuroku) into the manufacture of mag- netic media. They are having a large impact because the heart of the manufacturing process
52 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 and disks. The United States still lags behind Japan in the treatment and manufacture of magnetic particles (except possibly for 3M, which man- ufactures its particles internally). There are disturbing signs that the Japanese may be ahead of the United States in the next generation of film base, especially for vapor-deposition mag- netic media. The situation is not entirely clear, because 3M and Kodak make their own pro- prietary film. Other U.S. magnetic media com- panies, though, may be buying their film tech- nology from Japan in the future. Optical recording media for read-write appli- cations are still in the research stage. U.S. companies are roughly on par with European and Japanese companies in such research. Read- only applications (e.g., CD-ROM disks and compact audio disks) are largely dominated by manufacturing technology from overseas. Interconnection and Packaging The United States leads its competitors in the design of central processing unit packaging for large computers. Companies such as IBM, Cray, and Amdahl are on the cutting edge of inter- connection design and manufacturing. Japanese companies are ahead in some interconnection technologies found in mid-sized and smaller computers (e.g., phenolic paper boards and epoxy-resin boards). Photovoltaics The U.S. photovoltaics industry serves more than 100 different countries. Major competition comes from Japan and, to a limited extent, from Europe. U.S. firms have a dominant position in the power module market (devices with photovoltaic areas greater than 0.5 m2) while Japanese firms have dominated the consumer market for small-photovoltaic goods (e.g., cal- culators, watches, and radios). Superconductors A recent report on high-temperature super- conductivity' characterizes international com FEgATIERS i^~\ CHEMICAL ENGINEERI.\G petition as intense, but the U.S. competitive position in science as good. Japan, China, a number of European countries, and the USSR are putting in place significant scientific and technological efforts. In Japan, industrial con- sortia are being organized by the government to begin initial development activities. The re- port concludes, "Japan offers perhaps the strongest long-range competitive threat to the U.S. position." General Observations The industries that manufacture materials and components for information applications are characterized by products that are rapidly superseded in the market by improved ones. This rapid turnover stems from the intense competition among these industries and results in rapid price erosion for products, once intro- duced. These industries also require rapid tech- nology transfer from the research laboratory onto the production line. Many of theirproducts cannot be protected by patents, except for minor features. Therefore, the key to their competitive success is thoroughly characterized and inte- grated 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 po- sition in these industries, it is essential that it develop the fundamental knowledge necessary to stimulate further improvement of, and in- novation in, processes involving chemical re- actions that must be precisely controlled in a manufacturing environment. In the next section, the principal technical challenges are set forth. INTELLECTUAL FRONTIERS A variety of important research issues require much more work if U. S. companies are to establish and maintain dominance in information storage and handling technologies. These issues are quite broad and cut across the spectrum of materials and devices. Process Integration Process integration is the key challenge in the design of efficient and cost-effective manufac
ELECTRONIC, PHOTONIC, AND RECORDING MATERIALS AND DEVICES luring processes for electronic, photonic, and recording materials and devices. Except for magnetic tape, these products are currently manufactured through a series of individual, isolated steps. If the United States is to retain a position of leadership, it is crucial that its overall manufacturing methodology be exam- ined and that integrated manufacturing ap- proaches be implemented. Historically, all in- dustries have benefited both economically and in the quality and yield of products by the use of integrated manufacturing methods. As indi- vidual process steps become more complex and precise, the final results of manufacturing (e.g., yield, throughput, and reliability) often depend critically on the interactions among the various steps. Thus, it becomes increasingly important to automate and integrate individual process steps into an overall manufacturing process. The concepts of chemical engineering are easily applied in meeting the challenge of pro- cess integration, particularly because many of the key process steps involve chemical reac , WAFER _ LASER ENTRY ~ READER Let i. 1 1 _ ~_ __ _ ~__ J RESIST PATTERN DEVELOP COAT & ~ TRANSFER ~ ETCH ~ TEST BAKE (EB, X-RAY STRIP IPOH )TO, CLEAN 1 1 53 lions. For example, in the manufacture of mi- crocircuits, chemical engineers can provide mathematical models and control algorithms for the transient and steady-state operation of in- dividual chemical process steps (e.g., lithogra- phy, etching, film deposition, diffusion, and oxidation), as well as interactions between pro- cess steps and ultimately between processing and the characteristics of the final device. As another example, in microcircuit manufacture, chemical engineers can provide needed simu- lations of the dynamics of material movement through the plant and thus optimize the flow of devices (or wafers) through a fabrication line (Figure 4.12~. The continuous production of photovoltaic devices will require similar studies with even more emphasis on automation. Reactor Engineering and Design Closely related to challenges in process in- tegration are those in reactor engineering and design. Research in this area is important if we MICROPROCESSOR CONTROL UNIT ~ _ ION IMPLANT _ , & THERMAL DRIVE .... : . , , ~, |PRODUCT| I REJECT| I 1 FIGURE 4.12 The integrated semiconductor processing line of the future will be a fully automated series of chemical processing steps. Chemical engineers will be needed to integrate individual process steps into a manufacturing line that can be operated free from human handling, and possible contamination, of the devices. Courtesy, AT&T Bell Laboratories. FILM DEPOSITION (SPUTTER, MBE, CVD EVAPORATE) l t~ T WAFER TRANSPORT ELECTRONIC CONTROL RF-ENTRY PATH EXIT PATH
54. are to automate manufacturing processes for higher yields and improved product quality. Processes such as CVD, epitaxy, plasma-en- hanced CVD, plasma-enhanced etching, reac- tive sputtering, and oxidation all take place in chemical reactors. At present, processes and reactors are generally developed and refined by trial and error. A basic understanding of fun- damental phenomena and reactor design would facilitate process design, control, and reliability. Because all these processes involve reaction kinetics, mass transfer, and fluid flow, chemical engineers bring a rich background to their study and improvement. For example, high-yield, continuous processes for film deposition and packaging are required if photovoltaic devices are to be manufactured at costs that are com- petitive with other energy technologies. New reactors and a better understanding of chemical dynamics in reactors are central to achieving this. An important consideration in reactor design and engineering is the ultraclean storage and transfer of chemicals. This is not a trivial prob- lem; generally, the containers and transfer me- dia are the primary sources of contamination in manufacturing. Methods are needed 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 1 part per billion. This requirement puts severe con- straints on the types of materials that can be used in handling chemicals. For example, ma- terials 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 of molecular dimensions. The ability to "tailor" films at the molecular level- to deposit a film and control its properties by altering or forming the structure, atomic layer by atomic layer opens exciting possibilities for new types of devices and structures. The fab- rication of these multilayer, multimaterial struc- tures will require deposition methods such as MBE and MOCVD. Depositing uniform films by these methods over large dimensions will FRONT`tERS 4~N CHEMICAL ElYrGINEERI^~!G require reactors with a different design from those currently used, especially 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., superlat- tices). To achieve economic automated pro- cesses, the reactor design must allow for the acquisition of detailed real-time information on the surface processes taking place, fed back into an exquisite control system and reagent delivery system. This problem gives rise to an exciting series of basic research topics. Ultrapurification A third research challenge that is generic to electronic, photonic, and recording materials and devices stems from the need for starting 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 find new, lower cost routes to ultrapure silicon and gallium arsenide and to purify other reagents used in the manufacturing process so that they do not introduce particulate contamination or other defects into the device being manufac- tured. For optical fibers, precursor materials of high purity are also needed. For example, the SiCl4 currently used in optical fiber manufacture must have a total of less than 4 parts per million of hydrogen-containing compounds and less than 2 parts per billion of metal compounds (Figure 4.131. Either impurity will result in strong light absorption in the glass fiber. For magnetic media, the challenge is to separate and purify submicrometer-sized magnetic par- ticles to very exacting size and shape toler ances. A variety of separation research topics bear on these needs, such as generation of improved selectivity in separations by tailoring the chem- ical and steric interactions of separating agents, understanding and exploiting interracial phe- nomena in separations, improving the rate and capacity of separations, and finding improved process configurations for separations. These are all research issues central to chemical en . . glneerlng.
ELECTRONIC, PHOTONIC, AND RECORDING MATERIALS AND DEVICES Irene N2/Ct 2 C A' ~ FEED ~ - h~ N2 ~Wi ~ ' ' 1~ ES PRoDUCT FIGURE 4.13 Schematic diagram for a purification plant for producing "optical fiber grade" SiCl4. The feed material is passed through a reactor (1) where chlorination takes place. Excess HCl generated in the reactor is removed (2) and the product stream is passed through two distillation columns (3,4) where contaminants are removed. On-line IR spectroscopy (5) is used to monitor final product purity. Plants built using this design currently produce about 27 kg/in of ultrapure SiCl4. Contaminants (e.g., compounds containing R-H, compounds containing C-H, and Fe) are reduced to below the limits of detection. Courtesy, AT&T Bell Laboratories. Chemical Synthesis and Processing of Polymeric Materials Although chemical engineering challenges re- lated to polymeric materials are discussed in Chapter 5, the special challenges for polymers in materials and devices for information storage and handling deserve some mention here. For the processing of microcircuits and in- terconnecting devices, improved radiation-sen- sitive polymers are needed for the formulation of better photoresists. Resists must be highly sensitive to the radiation used for exposure, but not to the microwave radiation used after de- velopment for other process steps such as plasma etching. Chemical engineering studies of poly- mer behavior during development steps are also needed. Details of the dissolution of the exposed (or unexposed) regions of the resist are at present poorly understood. There is a need for fundamental studies and modeling of the for- mation of a swollen gel layer at the solvent/ polymer interface and the subsequent diffusion of polymer chains into solution. Light wave technologies provide a number of special challenges for polymeric materials. Poly- mer fibers offer the best potential for optical Scommunications in local area network (LAN) applications, be- cause their large core size makes it relatively cheap to attach con- nectors to them. There is a need for polymer fibers that have low losses and that can transmit the bandwidths needed for LAN ap- plications; the acrylate and meth- acrylate polymers now under study have poor loss and band- width performance. Research on monomer purification, polymer- ization to precise molecular-size distributions, and well-con- trolled drawing processes is rel- evant here. There is also a need for precision plastic molding processes for mass production of optical fiber connectors and splice hardware. A tenfold reduction in the cost of fiber and related de- vices is necessary to make the utilization of optical fiber and related devices economical for local area networks and the telecommunications loop. Another challenge for polymer research in light wave applications is in the use of active coatings on optical fibers as transducers for sensors. Such coatings may have magnetostric- tive or piezoelectric properties. These coatings, or the fiber itself, may also incorporate dyes that would respond to chemicals, light, radia- tion, or other stimuli to produce transmission loss changes in the fiber. Such systems have enormous potential as sensors that would be ultrasensitive, capable of distributed sensing, able to operate in harsh environments, and unaffected by electromagnetic interference. Specialty fibers such as polarization-maintain- ing fibers, which have an asymmetric core and can double the bandwidth by transmitting two modes at once, may also play an important role in sensor technology. Techniques for fabricating low-cost optical components such as graded index lenses, mi- crolenses, couplers, splitters, and polarizers are needed to support optical fiber technology. Traditionally, amorphous inorganic materials have been used, but there are tremendous
56 opportunities for innovation with polymers, which offer manufacturing versatility that is not available with glass. For example, photoselec- tive polymerization techniques can be used to make branching wave guide circuits such as splitters and couplers. Photopolymerization and copolymerization of multiple monomer systems have been used to make radial, axial, and spherical graded-index lenses with a high degree of perfection (e.g., freedom from aberration). Large-scale, well-controlled chemical proc- esses will be needed to fabricate these struc- tures. For recording applications, new approaches to high-quality polymeric film substrates are needed. Improved automation and control of thin-film coating are also important. For interconnection and packaging technol- ogies, an important goal is to achieve high- purity molding and dielectric materials. Epoxy- Novolac prepolymers with ionic impurity levels below 20 ppm offer one approach. There is a further need for low-viscosity molding com- pounds to minimize the development of flow stresses during processing. Continued devel- opment of thermally stable polymers with low dielectric constants (such as the polyimides) is also necessary. Advances in our fundamental understanding of polymer chemistry and rheol- ogy are crucial for all these areas (see Chapter 51. Chemical Synthesis and Processing of Ceramic Materials Challenges for chemical engineering related to ceramic materials are also discussed in Chap- ter 5, but the potential contribution of chemical engineers to this area cannot be emphasized too strongly. A tremendous opportunity exists for chemical engineers to apply their detailed knowledge of fundamental chemical processes in the development of new chemical routes to high-performance ceramics for electronic and photonic applications. The traditional approach to creating and processing ceramics has been through the grinding, mixing, and sintering of powders. Although still useful in many appli- cations, this technology is being replaced by approaches that rely on chemical reactions to HERS IN CHEMICAL ENGINEERING create a uniform microstructure. Chemical routes to better ceramics have the advantage of being more amenable to continuous and automated processing. Among the typical examples of such approaches are sol-gel and related processes. (See Chapter 5 for a more detailed treatment of sol-gel processing.) Deeper involvement of chemical engineers in manufacturing processes for ceramics may be particularly important to the eventual commer- cialization of metal oxide superconductors. The current generation of such superconductors consists of planar structures formed during a conventional ceramic synthesis. The ability to precisely control complex phase structure and phase boundaries seems critical. It is by no means clear that the formulations and structures that may produce optimal performance in su- perconducting ceramics (e.g., room-tempera- ture superconductivity, capacity for high-cur- rent density) are accessible by these techniques. Rational synthesis of structured ceramics by chemical processing may be crucial to further improvements in superconducting properties and to efficient large-scale 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 much less than 1 Am (Figure 4.141. Although the lithographic tools to print such patterns exist, the exposure step is only one of a number of processes that must be performed sequentially in a mass pro- duction environment without creating defects. Precise and uniform deposition of materials as very thin films onto substrates 14 cm or more in diameter must be performed in a reactor, usually at reduced pressure. Particulate defects larger than 0.1 Am must be virtually nonexistent. Low-temperature methods of film deposition will be needed so that defects are not generated in previous or neighboring films by unwanted diffusion of dopants.
ELECTRONIC, PHOTONIC, AND RECORDING MATERIALS A.~D DEVICES 10 8 6 , -O 7 _ o E 4 _ r-~ 3 _ J I _ 2 _ 1 1 1 o o o o o 1 1 1 1 1 1 1 1 1 1 1 '74 '76 '78 '80 YEAR '82 '84 '86 FIGURE 4.14 Feature size on microelectronic devices has steadily declined over the years as improved chemical etching processes have been developed. This graph shows feature size as a function of the year in which the device with the smallest feature size was first produced. Courtesy, AT&T Bell Laboratories. For optical fibers, improved control over the structure of the thin films in the preform will lead to fibers with improved radial gradients of refractive index. A particular challenge is to achieve this sort of control in preforms created by 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 ultraviolet radia- tion, are needed for high-speed (>10 m/s) fiber- drawing processes. Both glassy and elastomeric polymers are needed for use over temperatures ranging from - 60 to 84°C or higher. Hermetic coatings are required to avoid water-induced stress corrosion of silica glasses, which pro- ceeds by slow crack growth. Materials under study include silicon carbide and titanium car- bide applied by chemical vapor deposition, as well as metals such as aluminum. A tenfold increase in the rate at which such coatings can be applied to silica fiber during drawing is needed for commercial success. Coatings must be free of pinholes, have low residual stress, and adhere well. Hermetic coatings will also be needed to protect the moisture-sensitive halide 57 and chalcogenide glasses that may find use in optical fibers of the future because of their compati- bility with transmission at longer wavelengths. Considerable progress in the science and technology of de- positing thin films is necessary if the U.S. recording media indus- try is to remain competitive with foreign manufacturers. New, fully automated coating processes that will generate high-quality, low- defect media are needed. Not only must considerable effort be mounted in designing hardware and production equipment, but complex mathematical models must be developed to study the kinetic and thermodynamic properties of film coating and the effect of non-Newtonian flow and polymer and fluid rheology. A better understanding of dispersion stability during drying, as well as of diffusion mechanisms that result in intermixing of se- quential layers of macromolecules, is important. Thin films are also critical to the performance of electrical interconnection devices (Figure 4.15) Better methods for depositing thin films conformably (for good sidewall coverage) and for achieving high-aspect-ratio trenches are needed for the interconnection of electronic devices for the high-frequency transmission of data. New processing strategies and device structures are required that use compatible layers of materials to minimize undesirable phenomena such as contact resistance; elec- tromigration; leakage currents; delamination; and stress-related defects such as cracks, voids, and pinholes. ~ , it, Modeling and the Study of Chemical Dynamics A challenge related to the problems of reactor design and engineering is the modeling and study of the fundamental chemistry occurring in man- ufacturing processes for semiconductors, opti- cal fibers, magnetic media, and interconnection.
~8 For example, mathematical models originally developed for continuously stirred tank reac- tors and plug-flow reactors are applicable to the reactors used for thin-film processing and can be modified to elucidate ways to improve these reactors. For these models to reach their full descrip- tive potential, detailed studies of the fundamental chemical reac- tions occurring on surfaces and in the gas phase are required. For example, etching rates, etch- ing selectivity, line profiles, de- posited film structure, film bond- ing, and film properties are determined by a host of varia- bles, including the promotion of surface reactions by ion, elec- tron, or photon bombardment. The fundamental chemistry of these surface reactions is poorly understood, and accurate rate expressions are particularly needed for electron-impact reactions (i.e., dis- sociation, ionization, and excitation), ion-ion reactions, neutral-neutral reactions, and ion- neutral reactions. The scale and scope of effort devoted in recent years to understanding cata- lytic processes need to be given to research on film deposition and plasma etching. Until we have a basic understanding of chemical reac- tions occurring at the surface and in the gas phase, it will be difficult to develop new etching systems. Research in this area has had a demonstrable impact on recent innovations in plasma process- ing. 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 fabri- cating integrated circuits. However, well-con- trolled processes could not be developed that would perform in a production environment. The work of chemists and chemical engineers in elucidating the relevant chemical reactions and their kinetics was crucial to the identifica- tion of the important chemical species in the FRONTIERS IN CHE.~CAL E~GINEERING Pb-Sn ~ Cu-Sn intermetallic solder pad ~ / Phased Cr-Cu 75~< Cal 3.8 rum SiO2 :///////////~ 2.4 ,um SiO2 i ·4 Am ~-4% Cu ~7~: //////////////////// ~ 7 ~ 1 Si3N4 /' Thermal SiO2' PtSi ~ 0.15 ,Um Cr-CrxOy FIGURE 4.15 Cross-section of multilevel interconnections for advanced bi- polar devices. Fourteen separate layers are laid down in the fabrication of interconnections such as the one shown. The precise orientation and com- position of these layers are controlled by chemical process steps. Copyright 1982 by the International Business Machines Corporation. Reprinted with . . permission. etching process and their reaction pathways. In addition, this work led to the discovery that the organic polymer photoresist contributed to plasma chemistry and selectivity in important ways. This in turn led to new, improved plasma processes that are currently being used in pro- duction. For magnetic media, mathematical models could enhance our fundamental understanding of the manufacturing processes used to make uniform high-purity magnetic particles. Models for the kinetics and mechanisms of reactions and an improved understanding of the thermo- dynamics of producing inorganic salts are re- quired. Modeling to describe the flows of viscous fluids could lead to better packaging of inte- grated circuits by assisting in the development of molding compounds and processes that will provide for lower thermal shrinkage stresses, lower permeability, and lower thermal conduc- tivity. Such modeling could also contribute to the development of packaging materials and processes amenable to automation.
ELECTRONIC, PHOTONIC, A^~D RECORDING IS Aims DEVICES Engineering for Environmental Protection and Process Safety Safety and environmental protection are ex- tremely important concerns that present de- manding intellectual challenges. The manufac- ture of materials and devices for information handling and storage involves substantial quan- tities of toxic, corrosive, or pyrophoric chemi- cals (e.g., hydrides and halides of arsenic, boron, phosphorus, and silicon; hydrocarbons and organic chlorides, some of which are sus- pected carcinogens; and inorganic acids). The expertise of chemical engineers in the safe handling and disposal of highly reactive mate- rials is much needed in the electronics industry. Recent studies in California indicate that the semiconductor industry has an occupational illness rate three times that of general manufac- turing industries. Nearly half of these illnesses involve systemic poisoning from exposure to toxic materials. Problems with groundwater contamination in Santa Clara County, Califor- nia, have also raised concerns about how well the semiconductor industry is equipped to han- dle waste management and disposal. If the semiconductor and other advanced materials 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. IMPLICATIONS OF RESEARCH FRONTIERS Industry has been the prime mover in ad- vancing technology in electronic, photonic, and recording materials and devices. It will remain so for the foreseeable future. University re- search groups need to develop and maintain good communication with counterpart research groups in industry. Collaborative mechanisms are needed to promote academic-industrial cou- pling. This coupling will become even more im- portant as the electronics industry hires ever greater numbers of chemical engineers. Since 59 TABLE 4.4 Employment of Chemical Engineers in the Electronics Industry, 1977-1 986 Year Number of Chemical Engineers 1977 1980 1983 1986 700 960 1,648 2,100 a Employment figures for Standard Industrial Classification code 367, "Electronic components and accessories." SOURCE: National Science Foundation.3 1977, the number of chemical engineers em- ployed by the industry has tripled (Table 4.4J. Up to 25 percent of the recent graduating classes of several leading chemical engineering depart- ments have been employed by the electronics industries. The increasing demand of these in- dustries for chemical engineers is one factor to consider in planning for the support of the field. Any new mechanisms proposed must address this need. In the electronics industry, a large number of relatively small firms play a key role in gener- ating new process concepts and equipment. These firms face important research problems in fundamental science and engineering that would benefit markedly from the insights of academic chemical engineering researchers. Ac- ademic researchers should seek out and forge links to these small firms that stand at the crucial step between laboratory research and production processes. Potential mechanisms for accomplishing this are described in Chapter 10. 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 instructional material and example problems relevant to the challenges outlined in this chapter. This would not require the creation of new courses, but rather the provision of material to enrich existing ones. This theme is echoed, more broadly, in Chapter 10.
6a NOTES 1. National Research Council, National Materials Advisory Board. State of the Art Reviews: Ad- vanced Processing of Electronic Materials in the United States and Japan. Washington, D.C.: National Academy Press, 1986. 2. National Academy of Sciences-National Academy of Engineering- Institute of Medicine, Committee on Science, Engineering, and Public Policy. "Re- search Briefing on High-Temperature Supercon- ductivity," in Research Briefings 1987. Washing- ton, D.C.: National Academy Press, 1987. 3. (a) National Science Foundation, Division of Sci- ence Resources Studies. Employment of Sci- entists, Engineers, and Technicians in Man- ufacturing Industries: 1977 (NSF 80-3061. Washington, D.C.: U.S. Government Printing Office, 1980. (b) National Science Foundation, Division of Sci- ence Resources Studies. Scientists, Engi- neers, and Technicians in Manufacturing and Non-Manufacturing Industries: 1980-81 (NSF 83-324~. Washington, D.C.: U.S. Government Printing Office, 1983. (c) National Science Foundation, Division of Sci- ence Resources Studies. Scientists, Engineers, and Technicians in Manufacturing Industries: 1983 (NSF 85-328~. Washington, D.C.: U.S. Government Printing Office, 1985. FRONTIERS IN CHEMICAL ENGINEERING (d) Preliminary data from the 1986 survey of manufacturing industries provided by the NSF Division of Science Resources Studies. SUGGESTED READING M. Bohrer, J. Amelse, P. Narasimham, B. Tariyal, J. Turnipseed, R. Gill, W. Moebuis, and J. Bo- deker. "A Process for Recovering Germanium from Effluents of Optical Fiber Manufacturing." J. Lightwave Tech., LT-3 (3), 1984, 699. T. Li, ed. Optical Fiber Communications, Vol. Orlando, Fla.: Academic Press, 1984. P. D. Maycock and E. N. Striewalt. A Guide to the Photovoltaic Revolution: Sunlight to Electricity in One Step. Emmaus, Pa.: Rodale Press, 1984. R. H. Perry and A. A. Nishimura. "Magnetic Tape Production," in Kirk-Othmer Encyclopedia of Chemical Technology, 3rd ea., Vol. 14, p. 744. New York: Wiley-Interscience, 1979. Solar Engineering Research Institute. Basic Photo- voltaic Principles and Methods (FT-290-14481. Washington, D.C.: U.S. Government Printing Of- fice, 1982. S. M. Sze. Semiconductor Devices: Physics and Technology. New York: John Wiley & Sons, 1984. W. Thomas, ed. SPSE Handbook of Photographic Science and Engineering. New York: Wiley-Inter- science, 1973. L. F. Thompson, C. G. Willson, and M. J. Bowden. Introduction to Microlithography (ACS Sympo- sium Series No. 2191. Washington, D.C.: American Chemical Society, 1983.
