ULSI technology toward the next century: driven by DRAMs or MPUs?

This paper reviews the progress made in silicon-based ULSI technology, which has been mostly driven by dynamic random access memory, and, in part, very high speed RISC processors. It is interesting to compare those two technology drivers where DRAMs have always been driving microlithography and front-half technology, such as isolation, FETs and polysilicon capacitor/trench necessary for storage capacitor fabrication, while RISC processors have driven back-half technology, such as multi-level interconnection/metallization and planarization, which supports essentially high-speed logic circuit capability. DRAMs have started showing certain divergence from the rest of ULSI technology, and they will most likely become even more specialized to meet their requirement. A natural question that arises would be whether we should really have two different sets of technology drivers for ULSI technology. At the same time we should consider the possibility whereby both MPU and high density DRAMs can be integrated on a single chip in order to provide high density on-chip cache memory capability in the future. This may result in a more synergistic approach to identify technology drivers in the future. Even more complications would come with the need to integrate analog capability in broader meanings, which would open up great opportunities as well as serious concerns from process parameter control point of view in manufacturing.<
>