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.
EXECUTIVE SUMMARY Electronic packaging has become an important element in electronic systems and may very well be a critical pacing feature in the future. Modern packaging technology embodies a host of materials science and technology issues and necessitates an integrated design approach that includes packaging considerations from the very outset. Integrated circuit performance can be limited or enhanced by packaging features, and the designer no longer has the latitude to ignore such considerations until the electronic portion of the design is complete. Packaging, with all its implied materials issues, is a basic design feature. The materials of packaging and interconnection that are intrinsic to the design effort are the subject of this report. The ability to incorporate these issues into microelectronics products will be a key factor in maintaining this nation's competitive position in the world market for advanced electronics Electronic systems needed in the next few years will require unprecedented packaging technology. The rapid advances in integrated circuit chip capabilities will continue to increase demand for enhanced interconnect capability as regards numbers of connections (pinouts), pinout configuration, heat removal, signal rise time, signal transit time, power lead inductance, power supply current, and environmental protection. The projected evolution of chip parameters is presented in the format suggested by scaling theory for three families of chips "bipolar, CMOS, and GaAs], and the concomitant implications for packaging and interconnection are discussed. Appendix A contains a list of many special terms used in this report. No attempt is made to explain all terms commonly used in the industry. For individual chips, there will be hundreds of pinouts, tens of watts, and subnanosecond rise times in the l990s. Clearly, interconnect structures will require considerable enhancements to translate these chip capabilities to system performance. Heat dissipation in thousands of watts and power supply requirements of thousands of amperes are projected at the board level. Special physical design problems arise with very-high-frequency and very high-speed circuits. There is concern that the United States is relying too heavily on foreign sources of packaging and interconnection materials for high-density electronic circuitry. The United States has lost significant market share in advanced chip technology, and the process continues. As domestic production is lost (e.g., in DRAlls I, advanced packaging and interconnection strategies will be handicapped by reluctance of foreign chip makers to supply ICs in unpackaged or other non- star~dard forms . Printed wiring board (PWB) technology continues to evolve with the introduction of efficient surface-mount technology, finer 1
2 patterns, greater numbers of signal layers, and improved board resins for greater dimensional control. Forty-layer boards have been achieved. PWBs will certainly be important for many years to come, but their limitations in very-high-density electronics will have to be addressed in terms of new design strategies in which materials innovations will play a key role. Packaging is now approaching a turning point at which single-chip packages cannot be assembled directly onto conventional circuit boards without impairment of performance. A new level of packaging, the multichip module (MCM), is coming into prominence. MCMs consist of inorganic base layers, to provide power, ground and decoupling capacitances, and signal interconnect patterns fabricated of high-conductivity metals and low-dielectric-constant organic polymer dielectrics. The individual chips are assembled on top, either in unencapsulated form or in low-impedance single-chip packages. The signal interconnection density achievable is very high, owing to the fineness of the patterns. Two layers of MCMs can replace dozens of layers in conventional Pubs. Materials support for MCM designs must be strongly encouraged in the United States. Materials issues emphasize both process and final property aspects of design, and material compatibility is a critical issue. Many different properties and compatibilities must be optimized simultaneously. Issues of importance include the following: · coefficient of thermal expansion · dielectric properties · thermal conduc t ivi ty · electrical conductivity · interracial chemis try · adhesion ~ mechanical strength and toughness · impact strength · long-term stability purity (including absence of radioactivity) · vapor permeability (especially water) · corros ion ~ metal migration · process control and reproducibility ~ process compatibility The engineering-design-manufacturing process sequence is somewhat difficult to describe briefly while emphasizing materials factors. Discussing some specific packaging and interconnection materials (see Chapter 5) can give a flavor for the complex compromises that must be made. These systems include epoxies for encapsulation and PWBs, ceramic materials for packages and co- fired circuits, polyimides for dielectric layers, and more exotic materials, such as superconductors, synthetic diamond layers, and composites. Beyond the domain of engineering, packaging and interconnection materials are strongly affected by business, organizational, and government policy issues. International competition presents implications in terms of economics and national security. These interacting business and government
3 issues have much in common with many other high-technology areas, but packaging materials is a field in which the organizational problems are important, competition is fierce, and the financial consequences of losing out are massive. The United States has already lost consumer electronics and the entire computer market could follow. Packaging is one of the essential ingredients in the future U.S. position. In certain respects, the United States is poorly positioned in regard to materials employed for packaging and interconnection. In the sequence of raw materials, intermediates, formulated materials, completed piece parts, and assembled modules and systems, materials and information move from one company to another in a highly incomplete and imperfect way. The sequence is fragmented, and the technology is migrating to other countries, leaving U.S. manufacturers dependent on foreign supply houses. Some form of domestic supply chain that accomplishes the advantages of a vertically integrated corporation is urgently needed. A further problem exists in connection with government-funded research and development. The most substantial source of federal money is the Department of Defense, and many important initiatives have received timely funding from that agency. Unfortunately, electronic packaging of military systems has evolved along lines that differ significantly from computer, consumer, automotive, and other electronic subareas. There are some hopeful signs of rationalization, but much remains to be done. In recent years, many consortia have been formed to address problems faced by the United States in regard to international competition in technological matters. Groups of companies and other organizations come together to address common issues in a coordinated way and to pool resources. A consortium specifically addressed to materials for electronic packaging and interconnection could be an effective approach to some of the problems faced. Coupling to system design and engineering and the involvement of first-class engineering talent are essential features. An alternative to an independent consortium on packaging might consist of an expanded emphasis on packaging and interconnection by existing consortia. The United States has enormous intellectual resources in the university system, and it will be important to bring this potential to bear on the critical area of electronic packaging. Coordination of efforts of the various relevant university engineering and science departments will not be easy for this interdisciplinary field. Coupling of university programs to industrial design, development, and manufacturing projects in productive arrangements will require creative management. Similarly, involvement of national laboratories and other organizations also must be implemented. Issues of organizational culture complicate communication among the potential contributors. The National Science Foundation's program on engineering research centers is a recent approach to finding a useful format, but it is directed to support large centers. Smaller grants that encourage materials innovations and industrial collaboration are needed. The United States has lost significant segments of the electronics market in recent years. The future position of the United States as a world
follow 4 power and the U. S. standard of living depend on a broadly-based response to foreign competition. Any response that does not include aggressive materials and packaging programs will fail to regain and sustain the U. S . position. Both domestic comfort and national security are at stake The committee's conclusions and recommendations regarding specific areas · National policies. Antitrust laws still wield considerable influence in U.S. business circles. The need for strict interpretation of the laws has been largely eliminated by the growth in overseas competition, which will prevent any U.S. combine from raising prices to the disadvantage of the public. Modernization of these laws should be undertaken with emphasis on the features that affect the competitiveness of U. S . corporations in critical world marke ts . The introduction of advanced packaging technology and materials in U. S . industry could be facilitated by selectively removing antitrust restrictions on "buying cooperatives . " Specificat ly, if U. S. companies were permitted to aggregate demand for new products (e. g., packaging materials), U. S . producers would be stimulated to risk their capital and develop the needed manufacturing capability. The committee concludes that a maj or impediment to U. S . competitiveness In computer and other electronic systems is the unwillingness of domestic material suppliers to invest capital for research, development, and production in long-term ventures. By aggregating market demand (as is done in Japan), U.S. electronics manufacturers would provide a more attractive market that would stimulate production of advanced materials and build a self- sustaining market. This would help compensate for systemic advantages of similar actions enjoyed by foreign materials companies. Foreign companies also enjoy low-interest loans that encourage greater patience in market development. Some form of selective capital encouragement should be considered as a part of a broadly-based U.S. strategic plan to revive, sustain, and create critically-important base technologies. This could be done at the state level as well as on the federal level. It is difficult to overstate the need for actions that will encourage the long- range research and development necessary to provide the technology for future industries. Low-interest money is needed. Tax credits for research and development are an alternative approach that should be considered. In an era of emphasis on short-term financial results, it is important to provide incentives for long-term investment in technologies that will build wealth for the nation. Although it has not been discussed widely, the much- admired U. S . corporate research and development laboratories are increasingly under pressure to shorten their sights and turn over quick profits. The current frenzy for mergers, acquisitions, and other forms of corporate churning is destabilizing and destroying the climate for long-range technical development. Some form of legislation that would discourage these practices seems desirable Legislative solutions by their nature are long term. Thus, it is all the more important that a national strategiic program be undertaken to find remedies for our present malaise . The problems do not show s igns of self-cure.
5 The committee discussed some changes in regard to patent procedures that could enhance U.S. competitiveness in the packaging materials area. Legislation would not be required for a system that gives priority on Patent Office and Patent Court dockets for submissions in the packaging area. (This already has been done for high-temperature superconductivity.) The ad hoc nature of this approach is a negative aspect. Legislation to allow the formation of patent pools and licensing arrangements by removing antitrust constraints for a period of 10 to 20 years has been proposed. As U.S. corporations become owned by and allied with foreign companies, legislative and procedural efforts to favor U.S. firms become awkward and difficult to administer. · User consortium. The committee exhibited considerable enthusiasm for an industry-led consortium of packaging-materials users, with the objective to develop a stronger U.S. base of packaging-materials suppliers. This consortium would develop materials requirements and materials application technology, and would cultivate domestic sources of supply. The analogy with Sematech, the U.S.-based consortium aimed at integrated circuit process equipment, is strong, and the term "Sepatech" was coined for purposes of discussion among committee members. Table 1 illustrates the analogy. Table 1 Analogy of U.S.-Based User Consortia Parameter Intended to motivate Sematech "Sepatech" _ Equipment suppliers to integrated circuit manufacturers Set up by: Chip makers Activity: Vehicles for activity: Full-scale manufactur- ing by: Prototype and build leading-edge parts, working with equipment suppliers to integrated circuit manufacturers Memories (advanced DRAM and SRAM) Chemical suppliers, compounders, and other materials suppliers System companies using packaging and inter- connection technology Perform materials and process evaluation, build some demonstration systems, and establish sources of supply Advanced work stations US Memories, Inc., System houses employing employing the apparatus, materials and processes materials, and processes developed
6 The consortium approach offers an increasingly acceptable way for manufacturers to pool their resources and aggregate their markets at a precompetitive stage. The approach places responsibility and control in the hands of the user industry and, therefore, ensures development along relevant paths. Financing must be substantially industrial ("earnest money"), but government investment can significantly encourage and facilitate establishment of the combine. Indeed, government funding at some level is probably essential and justified. Consortia are controversial, and there is a lot to be learned in their organization, strategic objectives, and operation. Clearly, success depends on the participating organizations, which must provide quality people at all levels and viable long-term career paths for individuals. The committee is aware that consortia are subject to diverse forms of criticism, but the consortium approach is a positive mechanism to relieve the problems of U.S. industrial fragmentation. There are not many readily implementable alternative approaches. · Military packaging. The packaging of integrated circuits for U.S. military systems must be hermetic, which has caused military electronics systems to lag considerably in terms of overall capabilities. The committee urges that alternative means be found to allow military hardware to move into the mainstream of electronic packaging, while still preserving reliability over long periods in difficult environments. Some study programs are under way in specific areas (e.g., silicone gel coatings), but a broadly-based action group should confront the issue and initiate studies that will lead to needed change. Although it is possible to obtain high levels of performance under hermetic constraints, the committee believes that pursuit of other options is a promising approach that should be encouraged. ~ Industry-national laboratory-university coupling. Although systems manufacturers have been very active in work on packaging and interconnection materials and structures, relatively little activity in this area is evident in universities and the national laboratories. The committee feels that interconnection and packaging are regarded as insufficiently exciting by many members of the academic community. Some means is needed to stimulate university work in the area. The National Science Foundation's Engineering Research Center program would be well suited for establishing a close coupling of industrial scientists and engineers as an explicit part of the basic format. A program of smaller grants, similarly structured, could be highly useful. Also, coupling of national laboratory talent in the area should be pursued. Clearly, new and innovative means must be found to focus more of the considerable U.S. technical power on the issues electronic packaging. · Specific materials for support. The following specific materials and process areas are recommended for support in connection with electronic packaging and interconnection. In any such list, there is the danger of important omissions, statement of the obvious, and possible emphasis on areas
7 that lie close to the interests of the committee members. The following list had general support: 1. Present mainline materials should be pursued to bring U.S. capability up to a position of world leadership. Briefly, this includes copper, gold, and aluminum metallization with progress on high reliability in high current density; alumina and aluminum nitride, and other ceramic processing for preparation of advanced interconnect details; further development of epoxies for encapsulation, board resins, and adhesives; and development of polyimides and other high-temperature polymers for substrates, interlayer dielectrics, and other uses. Benzocyclobutanes are a very promising class of low dielectric constant polymers that are beginning to appear in electronic products. 2. Glass-ceramic substrate compositions that can be co-fired with copper are viewed as an area of the highest priority. Much work has been done, but the problem has not been solved. 3. Low-dielectric-constant materials and interlayers are becoming increasingly important because signal transit time limits circuit performance. Polymeric materials are available with electric per~ittivity as low as 2, but substantial development is needed to bring the many other properties and process variables into useful ranges. Ceramics seem less likely for low permittivity. 4. High-thermal-conductivity materials for packaging are needed. Composites offer advantages in this area. Thermoelectric cooling offers an alternative approach. 5. Materials for high-reliability encapsulation are extremely important. Hermetic structures are expensive and not entirely reliable. Silicone gels are currently under study, with promising initial results. 6. New solder compositions that are creep resistant would be valuable. Solder substitutes (e . g. , anisotropically conducting polymers and ceramics) can be used for low- temperature assembly. 7. Composite materials can be employed to engineer combinations of properties not achievable with homogeneous substances. 8. Materials amenable to environmentally benign processing (e.g., dry processing, aqueous-based systems) will become increasingly important. Circuit-board cleaning is ~ major source of solvent loss to the atmosphere, which leads to ozone depletion and other ills. 9. High-temperature superconducting oxides are exciting candidates for interconnects. Chemical stability, electrical contacts, and other problems remain to be solved . (Bear in mind that copper, aluminum, and other metals also are considerably better conductors at liquid nitrogen temperature than at room temperature.)
8 10. Synthetic diamond films offer great promise for mechanical protection, chemical isolation, and electrical i nsulatior~. The thermal conductivity is uniquely high. 11. The ability to design materials with thermal expansion tailored to the appl ication is emerging and should be extremely important for interconnect s true tures
