Editor's Page

2020 vision

Your desktop computer will be as powerful as all of Silicon Valley's current computers put together. The descendants of today's microprocessors will actually contain the entire computer, with both memory and processing elements found on one chip. Wafers the size of extra-large pizzas will hold hundreds, even thousands of chips with linewidths in the atomic range. Welcome to the future, at least one envisioned by Dave Patterson, professor at UC Berkeley and author of "Microprocessors in 2020," one of the fascinating articles in Scientific American's recent special issue, "The Solid State Century."

At the center of Patterson's vision is the intelligent random-access memory, or IRAM, device. These superchips will merge processors and memories and achieve internal memory bandwidths nearly a hundred times faster than today's computers. A single IRAM will, with the fastest programs, contain most memory accesses. Patterson is one of the Berkeley IRAM project's prime movers, teaching graduate levels courses on this new paradigm.

As with all such optimistic prognostications, it comes back to this: can these devices be made in a cost-efficient, high-volume manner with the kind of product yields that will cover the immense capital expenditures required to build and equip succeeding generations of fabs? The volatile mixture of showstopping ingredients could boil over within the next decade, well before the arrival of Patterson's nanoelectronic nirvana. Smaller linewidths means even smaller killer defects (some of which are undetectable by current analytical tools), and larger wafer sizes translate into more area (on the front and back sides of the substrate) jeopardized by contaminants and other yield inhibitors. And then there's Industry Question No. 1 — will X-ray, soft X-ray, E-beam, enhanced optical, or another lithographic technique replace the current technology once the 0.13-µm wall cited by Gordon Moore and others is hit? These are some of the issues discussed by Jerry and Dan Hutcheson, VLSI Research's father-and-son team, in "Technology and Economics in the Semiconductor Industry" in the special issue.

Another, less-mainstream publication addresses the future of semiconductor manufacturing in the nearer term. The English translation of The Shape of Semiconductor Plants in the Years 2000/2005 was issued in September. A report of the survey activities of SEMI's Production Cost Savings Forum, the document was originally released in Japan in May. It is actually three reports in one, encompassing the surveys from the material savings, plant concept, and maximum output machine (MON) subcommittees as well as a lengthy appendix and copies of overhead foils from committee member presentations. The book packs reams of information in its 163 pages, with the actual survey response data found in the appendix offering some of the greatest insights.

A fresh approach based on improved capital effectiveness, touted by Stuart McIntosh, is discussed in this issue's lead Industry News story beginning on page 16. The Philips Semiconductor COO cites the need for upgradeable, modular tools and an integrated fab design that would move the industry away from the current scenario where equipment suppliers attempt to maximize the efficiency of individual tools. Says McIntosh: "We should be looking to maximize the whole fab's productivity."

Look to MICRO in the coming year to provide other thought-provoking ideas on the present and future of advanced semiconductor manufacturing.

Tom Cheyney
Editor

tom.cheyney@cancom.com

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit