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.
APPE D CES
A Electronics Industry Workshop ON NOVEMBER 13-14, 1997, the Committee on Materials Science and Engineering: Forging Stronger Links to Users of the National Materials Advisory Board hosted a workshop on linkages and the exchange of information in the electronics industry. This was the first of three workshops intended to identify (1) user needs and business practices that promote or restrict the incorporation of materials and process innovations, (2) how priorities in materials selection are determined, (3) mechanisms to improve links between the materials community and the engineering disciplines, and (4) programs (e.g., education, procedures, information technology) to improve these linkages. As shown in the agenda in Box A-1, the workshop was divided into four sessions. The first three sessions were devoted to different aspects of the electronics industry: the magnetic hard-disk-drive (HDD) industry, the chip manufacturing industry, and the packaging industry. Each session included presentations by rep- resentatives of consortia, academia, industrial research and development (R&D) organizations, supply manufacturers, and primary manufacturers. The fourth ses- sion was devoted to a discussion of the characteristics of the electronics industry that distinguish it from other industries and the importance of road maps and linkages among primary industries, supplier industries, and universities in the development of advanced technologies. MAGNETIC HARD-DISK DRIVE INDUSTRY The HDD market in 1997 was estimated to be $35 billion. Sixty-one million HDDs were shipped worldwide in the first half of 1997. The companies involved were Seagate (24.5 percent), Quantum (20.4 percent), Western Digital (19.7 percent), IBM (11.7 percent), Fujitsu (7.7 percent), Maxtor (4.9 percent), 81
82 MATERIALS SCIENCE AND ENGINEERING BOX A-1 Agenda for the Electronics Industry Workshop November 13, 1997 8:30 a.m. Convene and Introductions, Dale F. Stein, Committee Chair MAGNETIC STORAGE SESSION (W. Doyle, Session Chair) 9:00 a.m. Road Map Development and Maintenance, B. Schechtman, NSIC 9:20 a.m. Source of Invention: University, Sheldon Schultz, UCSD 9:40 a.m. Source of Invention: Industry, David Thompson, IBM Almaden Research Center 10:00 a.m. Supply Industry Perspective, R. Rottmayer, Read-Rite 10:20 a.m. User Industry Perspective, Thomas Howell, Quantum 10:40 a.m. Discussion CHIP MANUFACTURING SESSION (J. Shaw, Session Chair) 1:00 p.m. Road Map Development and Maintenance, Paul Peercy, SEMI/ SEMA TECH 1:20 p.m. Source of Invention: University, Woodward Yang, Harvard University 1:40 p.m. Source of Invention: Industry, Don W. Shaw, Texas Instruments 2:00 p.m. Supply Industry Perspective, Alain Harrus, Novellus 2:20 p.m. User Industry Perspective, Pier Chu, Motorola 2:40 p.m. Discussion 5:00 p.m. Adjournment November 14, 1997 8:30 a.m. Convene, Dale F. Stein, Committee Chair PACKAGING SESSION (J. Decaire, Session Chair) 8:35 a.m. Road Map Development and Maintenance, James McElroy, NEMI 8:55 a.m. Source of Invention: University, Michael G. Pecht, University of Maryland 9:15 a.m. Source of Invention: Industry, William T. Chen, IBM 9:35 a.m. Supply Industry Perspective, Jack Fischer, Interconnection Technology Research Institute 9:55 a.m. User Industry Perspective, Robert MacDonald, Intel 10:1 5 a.m. Discussion DISCUSSION SESSION 1:00 p.m. Generic Linkages in the Electronics Industries 2:00 p.m. Strengths and Weaknesses of Linkages in the Electronics Industries 3:00 p.m. Strategies for Improving Linkages in the Electronics Industries
APPENDIX A 83 Toshiba (4.7 percent), and others (6.4 percent). Although drive design is still concentrated in the United States, more than 50 percent of head and media devel- opment and manufacturing occurs elsewhere. Most HDDs are assembled in the Far East (50 percent in Singapore). The basic components of an advanced HDD system are (1) moving magnetic media with two remnant states (representing "1" and "0" bits in digital systems); (2) a magnetic head with a miniature transformer for recording and magne- toresistive (MR) field sensors for reading; (3) an interface between the head and media to achieve high reliability at extremely small spacings (25 nary); (4) a servo system for tracking previously written data; and (5) electronics to detect data at an error rate of less than 10-~2 errors/bit. The first three components which include disk substrates and surface overcoats, magnetic film media, wear- resistant overcoats, topical lubricants, head carriers (sliders), magnetic films for record heads, complex multilayer MR structures for reading, conductors, insula- tors, and planarization materials are materials intensive and were the focus of the HDD session of the workshop. HDD industry characteristics were identified as follows: (1) high volume at low cost; (2) short product cycles (less than 2 years); (3) low profit margin; (4) successful incremental improvements; (5) complex supplier networks; and (6) proprietary manufacturing process "art." The current metrics for evaluating HDD systems are cost, capacity, and access time. Incredible progress has been made in capacity, driven by the areal storage density, which has increased six-fold in the past 50 years. This progress has been driven primarily by scaling dimensions, which required significant changes in materials and processes. In the last 15 years, six major changes have been implemented in media (i.e., thin-film media and glass substrates) and heads (i.e., metal-in-gap, thin-film inductive, thin-film MR, and thin-film giant MR). Until the early 1980s, almost all HDD technology was originated by IBM. Since then, as competition has increased and IBM has scaled back its R&D, the responsibility for the development of new materials and processes has fallen more to other manufacturers. Sources of information on new materials include industry alliances, mergers, technical conferences, publications, and contract re- search with university faculty members. In the 1980s, only a few university faculty members were interested in mag- netic devices, so most new graduates were trained by industry. Two substantial changes have improved the linkages between universities and the HDD industry since then: . the establishment of industry-supported, multidisciplinary university centers devoted to magnetic storage (first at Carnegie-Mellon University and the University of California at San Diego, and later at the University of Alabama, the University of Minnesota, the University of Washington, the University of
84 MATERIALS SCIENCE AND ENGINEERING California at Berkeley, Ohio State University, the University of Nebraska, and Rice University) · the formation in 1990 of the National Storage Industry Consortium (NSIC) to provide mechanisms for industry-university collaborations on focused problem areas NSIC's mission is to increase the worldwide competitiveness of the U.S. storage industry by: (1) conducting joint research on high-risk, precompetitive storage technologies; (2) procuring government funding; (3) developing technol- ogy road maps; (4) maximizing the value of university research; and (5) acting as an industry spokesgroup. The following characteristics of the HDD industry were identified by indi- vidual workshop participants as contributors to the introduction of advances in materials and processes: · competition in an industry that produces high-technology products · extraordinary improvements in performance fueled by materials inno- vations · extensive industry-university collaborations facilitated by NSIC · strong university centers focusing on storage technologies, often staffed by faculty with industry experience · NSIC-developed technology road maps that identify challenges for con- tinued progress · no regulatory issues to interfere with efforts to develop new technology high employee mobility, which provides rapid equilibration of technology . The following characteristics of the HDD industry were identified by work- shop participants as inhibiting the introduction of materials and process advances: . . . strong interdependence of heads, media, interface, servo, and channel electronics, which makes it difficult to make changes that could affect more than one component · low profit margins in a commodity market, which shifts the emphasis to evolutionary rather than revolutionary changes · complex intellectual property agreements that inhibit universities from obtaining patent protection inconsistencies between university "blue sky" research and focused in- dustry goals · insufficient federal support for the mainstream magnetic storage industry · requirements for extensive empirical investigations because materials modeling capabilities are insufficient ineffective accelerated tests for the evaluation of long-term reliability
APPENDIX A 85 · requirements for large capital investments ($250 million/year) 12 to 18 months in advance of orders · shrinking customer base for materials suppliers CHIP MANUFACTURING INDUSTRY The second session was devoted to the chip manufacturing industry. The increased density of silicon, with feature sizes being reduced at a rate of about 10 percent per year, will lead to a predicted market value of $200 billion by the year 2000. Complementary metal oxide on silicon (CMOS) technology domi- nates more than 90 percent of the market. The dimensions of current CMOS chips are approximately 2 cm x 2 cm, with 0.35 micron critical feature sizes and three to five layers of metal wiring to interconnect the devices. The chips are fabricated on an 8-inch wafer, which requires about one month of processing time (three to six steps/day, running 24 hours/day). Terabucks (on the order of $10~2) are cur- rently invested in infrastructure, including raw materials, equipment, and R&D. Since 1992, the Semiconductor Industry Association has coordinated a pro- cess to develop a road map of industry technology requirements with a 15-year horizon. The market has grown at a rate of 15 percent per year for the past 35 years, following Gordon Moore's prediction of a 20 to 25 percent per year improvement in cost performance through (1) shrinking feature sizes (which increases performance), (2) increasing wafer sizes, (3) improving yield, and (4) increasing manufacturing productivity (which lowers costs). As the industry moves into the production of feature sizes of 0.1 micron within the next 10 years, however, innovative technologies will have to be developed. The Na- tional Technology Roadmap for Semiconductors, which identifies the key tech- nology needs, was devised by a large cross-section of the semiconductor com- munity. As many as 600 engineers from industry, government, universities, and suppliers participated in the technical working groups (TWGs), which included lithography, interconnects, front-end processes, factory integration, assembly and packaging, design and testing, process integration, devices, and structures. Crosscutting TWGs focused on environment, safety, and health; metrology; defect reduction; and modeling and simulation. The 1997 road map identified six difficult challenges facing the semiconductor industry that will require major initiatives to overcome: (1) continued affordable scaling; (2) affordable lithogra- phy at and below 100 nm; (3) on-off chips that operate at GHz frequencies; (4) new materials and structures; (5) measurements, metrology, and testing; and (6) R&D challenges. New materials are being explored at all levels for future silicon chips, from new substrate materials to new gate and gate oxide systems to high dielectric constant electrode materials to low dielectric constant insulators for wiring interconnections. The incentives to develop these materials are performance
86 MATERIALS SCIENCE AND ENGINEERING enhancement, product needs, cost reduction, competitive advantage, and regula- tory issues. The obstacles to material implementation are manufacturability and cost, high risk, long development times, lack of control on tool development (long lead time), cross-contamination in fabrication, difficulty in predicting time- to-market, and inadequate information on critical materials. Suppliers of manufacturing equipment do not dictate materials choices, but they enable their use in manufacturing. The mean time from development of a new material to manufacturing implementation is six to seven years, and the time to develop processing equipment increases this to ten to fifteen years. These long development times reflect the extreme conservatism of the semiconductor industry, which prefers to make progress through careful evolutionary tweaking. Major changes that are considered revolutionary are implemented only when absolutely necessary (e.g., when mandated by performance, regulation, or cost requirements). From the industry perspective, some of the risk and cost can be reduced by leveraging university research and ideas to complement internal R&D and de- velop science and technology for future products. Although communication has been improved in recent years, there are still a number of weak links between the industry and universities: differences in culture and objectives (e.g., system solu- tions and manufacturability are not usually objectives for universities); differ- ences in policies and practices on intellectual property; and the lack of material and simulation/prediction techniques. These links could be improved in several ways: addressing integrated-product issues early; promoting personnel exchanges/ long-term visits; clearly defining objectives and milestones; providing adequate project reviews and industry mentors; and strengthening the industrial/university research community through participation in consortia and jointly sponsored con ferences. The development of material technology, which is essential to the micro- electronics industry, is a high-risk, high-reward undertaking that requires vision and a long time span. University research is an important part of industrial materials technology development and can reduce the risk and costs to industry. The imple- mentation of new materials in products could be accelerated by (1) leveraging strategic programs to provide early learning and materials/modeling predictions; (2) closely adhering to product-technology road maps, (3) focusing on manufac- turability and cost issues early; and (4) developing a systems approach to materials development. The following factors were identified by workshop participants as contribut- ing to the introduction of advances in materials and processes: . Road maps provide a technological "stake in the ground" so that chip manufacturers can focus on accelerating time-to-market. · Road maps provide a framework for industry, suppliers, academia, and
APPENDIX A 87 government to "buy into" goals and research directions and provide a tool to help funding agencies decide which projects to fund. · Road maps and short product cycles allow universities to focus on long- term goals to extend the fundamental limits of silicon devices. · Industrial laboratories can focus on solving critical near-term problems. · Equipment suppliers can develop hardware from concept stage to produc- tion tools in approximately four years. The following factors were identified by workshop participants as inhibiting the introduction of materials and process advances: Technology road maps, even if they take a long-term view, are essentially evolutionary. Production conditions are difficult to duplicate in a university research environment because of the costly tooling or processing of real microelec- tronics fabrication facilities. · Increasing standardization places bounds on materials innovation. · Availability of equipment and processes are setting the pace of materials . . ~nnovahon. · Equipment manufacturers depend on the rapid development and accep- tance of new materials for the timely development of new tools. · Equipment manufacturers take a substantial risk in the introduction of new materials because there is no infrastructure available to test integra- tion in a manufacturing line. PACKAGING INDUSTRY The third session was devoted to a discussion of the packaging industry. Microelectronic packaging can be considered at several levels: silicon wafer/ chips, chip carriers, printed wiring boards and other interconnection substrates, circuit card assembly/test, and final product assembly/test. At one end, packaging technology is being driven by advances in silicon technology and at the other end by customer acceptance in the marketplace (e.g., form, fit, function, cost). Trends in electronics packaging technologies were illustrated through se- lected examples. The most advanced packaging technologies are being imple mented in portable hand-held products, such as the Sony camcorder and the Motorola cellular phone. These products are exploiting small, lightweight chip carriers with high input/output (I/O) pin densities, such as chip-scale packages and ball grid arrays as well as flip-chip and chip-on-board assembly technologies. Trends in packaging for integrated circuits included the following: ball grid array and chip-scale packaging area-array chip interconnections
88 MATERIALS SCIENCE AND ENGINEERING · migration from ceramic to organic materials · high-density substrates for ball-grid array assemblies · multichip packaging · package design for area/volume and weight constraints New materials are generally introduced one at a time to minimize require- ments for new tooling and to reduce capital investment and materials risks. The key decision factors are low technical risks, readily available material sources, and low costs. The National Electronics Manufacturing Initiative (NEMI) is a private, industry-led consortium dedicated to the advancement of the electronics manufactur- ing infrastructure in North America. NEMI members include a broad spectrum of electronic equipment manufacturers, components suppliers, manufacturing equipment suppliers, contract manufacturers, material suppliers, software suppliers, consortia/ trade associations/consultants, universities, and government organizations. NEMI's road map structure is derived from a comprehensive manufacturing system plan (i.e., a manufacturing system model representative of a comprehen- sive supply chain and factory for a "virtual product" target anticipated to be representative of future high-volume products). NEMI has customized the generic target product according to different market sector drivers. Five product market sectors are characterized as follows: low cost, hand held, cost/ performance, high performance, and harsh environment. Manufacturing infra- structure needs for each element of the plan were assessed to develop road maps. The overall NEMI road map is a coordinated set of road maps for each of the following manufacturing system elements: packaging; board (circuit card) as- sembly; final product assembly; interconnection substrates; displays; energy stor- age systems; radio-frequency components; passive components; semiconductor devices; magnetic mass data storage; optical mass data storage; optoelectronics; factory information systems; modeling, simulation, and rapid prototyping; and test, inspection, and measurement. NEMI has developed interorganizational link- ages to coordinate these road maps (e.g., magnetic and optical storage; semicon- ductor devices; displays; optoelectronics; and interconnection substrates). The manufacturers of electronic end products have embraced technology road maps as a way to establish linkages with their suppliers, including the R&D community. The NEMI road map does not explicitly address materials as a sepa- rate element of the manufacturing system plan, but materials issues are embedded in each element. Some workshop participants were concerned that the road map- ping process tends to emphasize evolutionary development and thus may over- look revolutionary innovations. Several mechanisms for research collaboration between industry and universi- ties are being implemented. University centers with research agendas based on problems identified by industrial partners are attracting more industrial participants and funding. Presumably, their research results will be more rapidly adopted.
APPENDIX A 89 FINAL DISCUSSION The final afternoon of the workshop was devoted to a discussion of (1) the characteristics that distinguish the electronics industry from other industries and (2) the importance of road maps and linkages among primary industries, supplier industries, and universities in the development of advanced technologies. Work- shop participants identified two characteristics that distinguish the electronics industry from the automotive and turbine-engine industries: The motivation to improve product performance has been based more on internal industrial road mapping (i.e., Moore's Law) than on customer relations (i.e., demands of the buyers and users of computers). Because of the emphasis on performance qualification, developers of elec- tronic components have greater freedom to incorporate new materials and processes. Some participants questioned whether the electronics industry actually intro- duced new materials faster than other industries. The basic materials and pro- cesses used in the electronics industry have remained relatively constant for the past 20 to 30 years. Silicon technology has improved incrementally, which has enabled the reduction of component sizes and improvements in performance. Many workshop participants felt that road maps are especially important for complex products. They identified four ways in which road maps could further technological development: · by defining the issues facing industries and gaps in technology · by coordinating all segments of the industry in the development process from fundamental research and development to final assembly · by providing a framework for industries and researchers to plan potential materials changes and technological innovations · by providing a level playing field among researchers and industries and lowering overall risk Consortia play two important roles in the road-mapping process: (1) provid- ing neutral territory on which competing industries can meet to develop and maintain industrial road maps and (2) acting as links between industries and universities to ensure that the necessary short-term research is conducted. In the opinion of some participants, consortia and industry have been too effective in promoting short-term research at universities, to the detriment of long-range research and teaching. Individual workshop participants identified three weaknesses in the industry-university collaboration:
9o MATERIALS SCIENCE AND ENGINEERING · Many primary manufacturing companies support university research, but supply industries have not been as active. This linkage could be strength- ened by the development of more supplier-oriented road maps. Most universities conduct research on different equipment than the equip- ment used in industry, which reduces the compatibility and ultimate use- fulness of some research results. If research at universities leads to the development of better process models, and thus a better general under- standing of the fundamental principles of new systems, industrial param- eters could be determined and the transition of new materials and pro- cesses accelerated. Many universities are devoting considerable time and resources to the establishment of university-industry links. Universities and industries should develop a standard methodology for interactions to eliminate the need to reinvent contracts with each new project. . . Most participants felt that the linkages between the primary and supply companies were stronger in the magnetic-head industry and the chip- manufacturing industry than in the packaging industry. An enormous amount of information is exchanged between all electronics industries to ensure that suppli- ers' products meet the needs of the primary manufacturers. One of the strengths of this relationship has been the standardization of many features (e.g., inputs, outputs, performance indicators). The participants noted that SEMI/SEMATECH has been instrumental in the establishment of this close integration.
