Moore no more?
By Bill Schweber, Executive Editor - September 2, 2004
Since scientist Gordon Moore's seminal and fairly accurate conjecture in 1965, our industry has lived and prospered by his eponymous "law" (Reference 1). From the earliest ICs to today's mega-ICs, it's been a long road of problems and advances in all aspects of technology and tools, as our industry developed an enveloping supply chain encompassing silicon, optics, physics, metrology, software, and more. We have faced and overcome innumerable, substantial unknowns and technical challenges.
But, as we approach the next milestone of the road map—90 nm—some industry experts are starting to worry, and these people know more about solid-state physics and processes than I do. They are concerned either that we won't be able to attain the next steps or that the realizable gains won't outweigh the technical and cost pains. For example, Bernard Meyerson, PhD, chief technologist at IBM, says that classical scaling is dead and that feature shrinkage no longer offers benefits that outweigh its problems. Meanwhile, John East, president and chief executive officer of Actel, argues that 130 nm will be the last great process reduction and that the 90-nm step will be impractical due to financial problems and technical limits.
I have mixed feelings about their comments. On the one hand, we've all heard over the past decades many dire warnings about the impossibility or difficulties of these next leaps with respect to basic materials, production equipment, design tools, or test capability, yet our industry has overcome each obstacle and moved ahead. On the other hand, we know that things can't go on forever; at some point, every technique or process reaches a fundamental limit at which radically new, disruptive approaches take over.
Plastic transistors, printed using an ink-jet-like method, may become feasible on a large scale for producing LCDs with fairly low density (Reference 2). Perhaps some biologically based technology that researchers are now studying in a lab will replace our solid-state world, as difficult as that is for us to believe.
Another force is making me wonder whether we are reaching the end of a long road. Our industry has always been cyclical, with swings between boom and bust, and these swings seem to be getting larger. When business is good, it's very good; fabs run at full capacity. When business is bad, fab usage drops significantly, and a lot of expensive capacity sits idle.
Yet, as engineers, you know that it is usually better to have a closer coupling between a system's inputs and capacity, or line and load, along with smoother changes to each, as a condition for overall stability and more efficient operation. As the swings get wider, the absolute difference between demand and capacity gets bigger, because there is always a lag time between those two factors. This situation occurs whether demand ramps more quickly than capacity or manufacturers build capacity in anticipation of future demand.
Perhaps we are reaching the point at which the matched forces of our solid-state-technological progress and the laws of physics are no longer in close enough balance to make things work out as they have in the past. If that's the case, then the next few years are going to be exciting, surprising, and a learning experience for all of us.
Contact me at firstname.lastname@example.org.