The Role of Materials in Information Technology
The forum included topical sessions on the current and future role of materials in four selected areas. The first of these sessions, on information technology, began with an overview by Andrew E.Lietz, president and chief executive officer of HADCO Corporation, a manufacturer of printed circuits and electronic interconnection devices. He is also a member of the Board of Directors of the National Electronics Manufacturing Initiative (NEMI).
Paul S.Peercy of the University of Wisconsin gave a technical perspective. Dr. Peercy is dean of engineering at the University in Madison. Until September 1999 he was president of SEMI/SEMATECH, a nonprofit technical consortium of more than 150 U.S.-owned and -operated companies that constitute the equipment and supplier infrastructure for the U.S. semiconductor industry. He has also been director of microelectronics and photonics at Sandia National Laboratories.
Lawrence Dubois of the Defense Advanced Research Projects Agency (DARPA) gave a perspective from the federal government. Dr. Dubois recently became vice president of the Physical Sciences Division of SRI International. At the time of the forum he was director of the Defense Sciences Office at DARPA. The Defense Sciences Office is responsible for an annual investment of nearly $300 million in technology development, including the development of defense applications for advanced materials.
The session concluded with a panel discussion. The following are summaries prepared by the editors who adapted them from the remarks made by the individual presenters.
Materials and Electronic Interconnects
Interconnect technology is as fundamental to the electronics industry as semiconductors. Demand for electronics-based products is driving rapid growth in the interconnect industry, and the emergence of new applications is driving rapid change. Staying competitive requires meeting demands for circuit miniaturization, speed and bandwidth, environmentally friendly production, and lower production costs, all of which require new materials and processing technologies. Moreover, in today’s global marketplace, U.S. manufacturers need advanced technology to compete against products from countries with lower manufacturing costs.
Improved dielectric materials will increase circuit densities by reducing feature sizes for component interconnection and attachment and enabling microvias and higher layer counts. The electrical characteristics for these
materials include lower dielectric constants, improved bulk properties related to signal loss, and controlled impedance. Improved thermal characteristics include thermal stability, fire retardance, and higher operating temperatures. Mechanical requirements include improved dimensional stability, thickness control, and conductor adherence. Chemical properties include moisture stability and better resistance to process chemicals. Last but not least, materials must be available at low cost and in high volume.
A key cost-reduction goal is to reduce component counts for board assembly by embedding passive components into boards. This will also require new materials with specific capacitance, resistance, and inductance properties, as well as improved dimensional stability and adhesion. Mixed-polymer dielectric composites will be important, particularly as low-cost alternatives to the fluoropolymers now used for high-frequency circuitry.
Basic materials research for printed wiring boards must include improvements in the robustness of metal-polymer and matrix-polymer interfaces, improved photoresists and etchants and other process consumables, and active “smart” composites and interfaces to enhance interconnect performance and processability.
Manufacturers also need new materials and processing techniques to respond to environmental concerns. Materials substitution is one important approach, particularly lead-free alloys and nonhalogenated resins. Going beyond substitution will require the development of new processing chemistries.
U.S. interconnect companies have been less aggressive than their global competitors in implementing new technologies at high volume. Resources to support R&D are extremely limited, especially for smaller firms, so rapid commercialization of technological advances is critical to maintaining competitiveness. Substantial investment by the semiconductor industry has resulted in a deep materials and processing knowledge base that has greatly accelerated commercialization of integrated circuit designs. The interconnect industry must make analogous efforts to reduce the cycle time for implementing new materials from years to months.
Materials Research for Computing and Communication
The contribution of the electronics industry to the gross world product reached a new high of 2.8 percent in 1999 and seems sure to continue rising. Despite its worldwide sales of $998 billion, the industry depends on an inverted pyramid of successively smaller industries. Materials constitute the tip of this pyramid, at $22 billion, including $3 billion in the United States. For example, half the cost of electronic packaging is for materials. The United States leads the world in introducing new electronic materials, and according to Texas Instruments, high-technology products account for 45 percent of the growth of the U.S. gross domestic product.
The rate of progress in the speed of silicon chips, the communications capacity of optical fiber, and the density of information storage continues to accelerate—approximately doubling each year and following Moore’s law with remarkable fidelity. Several times more transistors already exist on an 8-inch wafer of state-of-the-art memory than there are people on Earth. Optical fiber is being installed worldwide at a rate of 2,000 kilometers per hour.
In about a decade, all of these technologies will be approaching what today appear to be fundamental limits. Transistor dimensions will approach a few atomic layers, and according to the road map of the Semiconductor Industry Association, Moore’s law will slow down. In place of continuing improvement in chip performance, progress will depend on continued improvements in chip architectures and more efficient utilization of semiconductors. Driven by improvements in materials technologies, prices will continue to drop, with the cost per instruction per second for processors, and per bit per second for communications, falling by four orders of magnitude in 20 years. Continued improvement will be achieved by the development of new dielectrics, better reliability, advances in interconnection technology, and the design of full systems on a chip.
Materials, Materials Processing, and the Future of Information Technology
New materials create new technologies, and new technologies create new materials. The bottom-up approach starts with new materials and phenomena, such as nanomaterials and spin effects, and develops new technologies that they enable. The top-down approach starts with new architectures and new algorithms, such as fault tolerance and quantum computing, and develops new materials to meet their requirements. The two approaches complement and reinforce each other.
DARPA-supported research follows both these paths. In an example of the former, giant magnetoresistance materials have enabled the development of low-cost magnetic random access memory (RAM) that is nonvolatile, radiation hard, as fast as static memory (SRAM), as dense as dynamic memory (DRAM), and infinitely cyclable. Spin effects in semiconductors will lead to revolutionary advances in photonics and electronics, including high-performance lasers, very dense quantum-dot memories, and other applications yet to be considered.
DARPA’s program in molecular electronics is an example of the second approach. Molecular electronics is made possible by new fabrication technologies and driven by technological needs, such as the skyrocketing cost of circuit fabrication, the demand for enhanced computing power, and fundamental roadblocks facing silicon technology. The goal of the program is to demonstrate computational functionality in defect-tolerant architectures fabricated by direct assembly of molecular devices. Molecular logic gates, single-bit and two-bit memories, and interconnects have already been demonstrated.
Panel Discussion
The above talks were followed by a panel discussion. A summary of the discussion follows.
Materials technologies become much more difficult at scales smaller than 100 nm. Moore’s law may continue to apply for several more decades, but it may be driven by improved architectures. Andy Lietz indicated that the short time horizons on Wall Street present industry with a difficult challenge when seeking money for development.
An emerging trend in the U.S. electronics industry is outsourcing of manufacturing, testing, and so on. Although this results in the effective use of capital and high productivity, it does not favor the support of long-term materials research, which typically takes 15 to 20 years to achieve maturity. In printed-circuit technology, more than 50 percent of manufacturing takes place in Asia. These trends suggest that the United States might not be competitive in this industry in the future.
Interconnect technologies that require materials development include backplanes, integrated circuit (IC) chip assembly, high-speed laminates, surface mount technology, and embedded passive and high-speed connectors—faster, lighter, cheaper, more reliable, and environmentally sound! In all cases, materials development is focused on current needs, and little longer-term research is under way. To overcome the dilemma, consortia of industry, government, and academia should be established to provide a long-term focus, establish road maps, reduce risk, and capture emerging product opportunities.