Design-Coupled Manufacturing
- PDF / 726,948 Bytes
- 8 Pages / 414.72 x 648 pts Page_size
- 60 Downloads / 174 Views
Introduction Over the past several decades, the continuous advances in lithographic and
other semiconductor processing equipment have led to the resolution of finer
features and ever increasing integration levels. According to trends, a new generation of DRAM has generally emerged about every three years with a fourfold increase in bit capacity (i.e., 64k then-256k then 1M then 4M, and now 16M in production). As manufacturing capabilities advance, the manufacture of dynamic random access memories (DRAMs) has led the production of other components such as microprocessors (MPUs) and application-specific integrated circuits (ASICs). DRAMs are considered commodity components and are consumed in relatively large quantities by both the computer industry and the consumer electronics industry. Since all DRAMs are functionally identical, DRAM market economics have demanded that these commodity components be manufactured as inexpensively as possible via the economy of scale. One reason that DRAMs have led the production of other components has been that manufacturing the leading-edge of semiconductor microelectronics requires substantial experimental adjustment of the process, often referred to as "volume learning." The economy of scale model of manufacture is well suited for production of certain components, especially a single product/single process component, such as the DRAM. Unfortunately, as semiconductor manufacturing technologies have advanced, the cost associated with constructing, equipping, and maintaining a semiconductor factory appear to be increasing nonlinearly. Extrapolating the present trends suggests that the cost of a state-of-the-art factory 339 Mat. Res. Soc. Symp. Proc. Vol. 389 ©1995 Materials Research Society
at the turn of the century may be in excess of $2 billion. Additionally, process development and integration costs are of the same magnitude. At this price, it is unclear how many companies could afford the risks associated with this industry and construct such facilities, and support the DoD's projected requirements for a large number of differentiated, leading-edge ICs. This is of particular concern to ARPA since much of the tactical advantages of US defense and military systems are due to the use of state-of-the-art components. At one time, it was largely believed that to be at the leading-edge of semiconductor technology, a company had to be a volume producer of DRAMs. While this conclusion is in some doubt due to product differentiation caused by design, the values of volume learning are real and demonstrated. As the world economy is understood today, it may reasonably be expected that success in the marketplace may be determined by the capability to simultaneously optimize a number of
product parameters, including performance, reliability, cost, and time-tomarket. The synergy of these with military and defense needs is clear. Without a new approach toward semiconductor manufacturing, it will be difficult for many companies to remain at or near the leading-edge. The dangers in falling a
Data Loading...
