Two potentially disruptive technologies were presented and another was loud in its absence.

Strained Silicon

E. A. Gene Fitzgerald of MIT (and Amberwave Inc) discussed the 13 year history of strained silicon. By placing traces of Ge and other elements in thin films of silicon on wafers, or other growth substrates, very high mobility transistor channels can be demonstrated. This faster mobility can result in faster transistors that may keep industry on the Moore's Law curve without having to go to yet finer channel lengths by lithography. This could be very economical in that the existing foundries could go for a few more generations with the same equipment, just different starting wafers.

He explained that there are only four big IC products that are surfing the advances of Moore's Law: FPGA, DSP, microprocessors, and memory. It was also pointed out that it is not clear that increasing the number of transistors to 400 million is really helping the already beleaguered system designers. Thus by making existing sized designs by faster transistors will have more impact than simply stuffing more transistors on the same chip size. Another manifestation of this problem is that most of the problem in layout of large ICs is the interconnections between transistors not the performance of the transistors.

In addition, Gene Fitzgerald looks at future microsystems as wanting to combine three vastly different functions: the digital, the fast analog, and the interface with E&M waves. Currently the last two functions are forming the bottle neck for today's microsystems since they are made of compound semiconductors which have not been obeying the same manufacturing history curves as silicon. He finds that by adding enough layers and enough Ge to some of these layers that all functions can be made on the same chip using existing manufacturing equipment. One such layering system has strains silicon on top, next Si(80%)Ge(20%), graded SiGe layer to reduce lattice mismatch with Si substrate on bottom.

Using variations of these strained layers gives electron mobility of 80% with another Ge rich layer which eventually increases hole mobility by 800%. The low temperature formation of some of these layers are currently being addressed. If one of the layers has Ge >70% one could add opto sources to the chip. One could also lower voltages evern further with increased Ge.

In summary, Prof Gene Fitzgerald considers that there is still room for 1000% performance improvements by making "designer" strained wafers for existing foundries. That message seemed to satisfy everyone in the audience that they had enough affordable technology advances to make it to retirement.

Polymer Transistors

Dr. Henning Sirringhaus of Plastic Logic Ltd followed with his presentation "Entering the Era of Polymer Transistors." Most of the audience was brought up familiarly fighting with inorganic crystalline semiconductors so this presentation made the first one seem almost possible.

Dr. Sirringhaus basic premise was that printing is a cheap, efficient well known technology. Thus if polymer materials can continue to be refined so that transistors of reasonable characteristics can be "printed" at the same dots-per-inch of the "National Geographic Magazine" there will be many applications that will be performed by polymers rather than silicon.

Most of the work to date has been on "pMOS" with special surface adhesivity pattern treatments to allow short channels to be easily printed. The best mobility have been achieved by vacuum deposition of pentacene (sp?). The mobility is 5 cm2/v.s. If this seems low one must understand that in the last 10 years the mobility in polymers has improved 4 orders of magnitude with no obvious wall in the future.

Dr. Sirringhaus convinced the originally skeptical audience of the viability of polymer electronics.

The Lack of the Barking Dog

Everyone expected to have another presentation on the leading edge of compound semiconductors. The Chair of the session, VP Phil Garrou, explained that several speakers had been lined up only to have their companies decline support. Since compound semiconductors have been the future of the industry for most of the audience engineers' careers, this may mean the beginning of the end of the compound industry (other than for niches like traffic lights). This would be another example of the silicon juggernaut grinding down any competition however promising it appears.