Electric Motors, Semiconductors....Nanotubes?

Our economy has been driven by various components in the past. I remember Dave Carsten giving a speech at Semicon West (somewhere around 1985) that gave an analogy between the electric motor and semiconductors as drivers of our economy. (See if you can count the number of electric motors in your house.) Each driver represents a paradigm shift. The electric motor brought in literally thousands of applications that could employ the paradigm of rotary speed.. Semiconductors in the same way brought in millions of applications that can use the principle of programmability.

The end of each paradigm's reign is replaced by the next paradigm. With motors, we had the usual product life cycle curve where the product settled into a long, stable and sustainable commodity cycle. Hardly enough to support an economy. With semiconductors, we have Moore's Law [complexity doubles every 18 months]. This law has kept the semiconductor paradigm alive for over 40 years. However, Moore's Law is approaching a physical limit - the size of a transistor. Soon, our traditional, silicon-based technology will run out of steam and engineers and scientists will have to look for other ways to increase complexity.

That is not to say Moore's Law will fail; it only means that some other way must be found to achieve it. In our next paradigm shift, we will probably see silicon-based devices give way to carbon - carbon nanotubes. In particular, recent announcements by IBM (Nanotube "not" gate - Electronic News) and by Osaka University (Carbon nanotube crystal chains - Solid State Technology) has shown that integrated circuits using nanotube is within our sight.

Moore's Law with Nanotubes

What if Moore's Law continues? In 1985, the semiconductor industry shipped one brain. [This fact was observed by Howard Bogert of Dataquest at that time.] That's about a trillion bits. Assuming that Moore's Law kept working, we have had 10.6 doublings. That equates to 210.6 or 1625 brains this year. By the time that nanotubes can be developed, it will be 2005 or later. In 2005, we will ship 26,000 brains. By 2015, we will ship 27 million brains. With that kind of complexity, can anyone doubt that cybernetics will be one of the applications that creates growth in the next 20 years!

MEMS

One of the buzz words, that is seething below the surface of the semiconductor juggernaut is MEMS (micro electromechanical systems). This technology has the attractiveness of combining small mechanical devices with integrated circuits. The small mechanical devices are very small - sometimes less than 100 microns across. A couple of the first applications for this technology are inertial sensors for airbags and magneto-resistive devices for thin-film heads.

A few years ago, I helped found a start-up that targeted MEMS manufacturing. I attended the MEMS course at MCNC (They were spun-off from MCNC as Cronos Integrated Microelectronics and then acquired by JDS Uniphase in April 2000.) They used, and still use, the MUMPS (Multi-User MEMS Processes) for designing and fabricating MEMS devices. This is a three level planar process that uses silicon dioxide and polysilicon to provide a cost-effective access to surface micromachining for prototyping activities and a seamless transition into manufacturing. They also provide the LIGA process, a German acronym for lithography, electroplating, and molding. This can best be described as microfabrication processes which produce tall microstructures with vertical sidewalls. More information can be found at http://www.memsrus.com/cronos/svcsliga.html.

The MUMPs process has evolved elsewhere to a 5-level process named Summit-V. With 5 levels, the type and complexity of devices is greatly increased, albeit the process is more complicated and harder to prototype. A key organization that is developing applications for Summit V, along with LIGA, is Sandia National Laboratory. This national lab has spun off a company (MEMX, Inc.) and licensed its design tools (Microcosm, Inc.) in order to accelerate the commercialization of MEMS technology. It is hoped that the early commercialization will provide a reliable technology that can be used in our weapon systems. More information can be found at http://mems.sandia.gov/scripts/index.asp.

MEMS Manufacturing

The basic scenario for both MUMPs and Summit is to build the machines with silicon dioxide as a sacrificial material. For instance, a gear and bearing may be fashioned out of polysilicon with the space between made of SiO2. After the last layer is completed, the device can be placed in a solvent so that the SiO2 is dissolved and the gear is released. The range of devices that have already been designed and built is phenomenal.

Applications

The list for potential applications is legion. Current applications include, but are not limited to, accelerometers, pressure, chemical and flow sensors, micro-optics, optical scanners, and fluid pumps. MEMS devices are being studied for medical purposes. We will build little robots that will circulate through the blood system cleaning the plaque from blood vessels, thereby preventing arterial sclerosis. I saw a product in my wife's health magazine whereby a pill containing an electronic camera is swallowed in order to take pictures of cancers. This technique may uncover tumors that other methods would miss.

The applications for space technology are absolutely prodigious. I could go on and on, because this is such a facinating topic for me, but I will let you ponder this now if you have not already. You can find some excellent links to MEMS pages on my website (Some of them are obsolete, but most are still good.). There is a lot of government money chasing these technologies and there are a lot of scientists and engineers that would like to commercialize it.


Robert McGeary
Copyright August 2001

REM Enterprises, Inc.