Let's start down the road to nanotech
11/01/2004
While recent industry roadmaps have shown many red brick walls ahead, mapmakers say they are confident that the industry will still be able to follow the track of Moore's Law for another 15 years. They even feel that with new material sets and transistor configurations, this can be done using variations of today's CMOS, via an evolutionary rather than disruptive path.
But what happens then? If we continue to try to make devices based on manipulating charges, we will be working with so few atoms of width and thickness that the devices won't be repeatable and energy densities will become unmanageable. So the pat answer has been that we will go to "nanotechnology." But what is electronic nanotechnology? If it were simply a matter of size, the industry is already making nano-scale devices (100nm features are in the molecular size range). But most see it as far more than just shrinking today's technology. To do real nanotechnology, we need to learn to build devices and manipulate information at the molecular and atomic levels the way nature builds organized structures like crystals, polymers, and DNA. Then we will have to get them to interact in predictable ways, and extract readable outputs.
True nanotechnology means learning to use nature's own mechanisms and rules, rather than force-fitting our device concepts onto molecular-level structures.
That's a whole new approach to building intelligent systems. While 15 years seems far off now, at a recent Albany Nanotechnology Symposium, John E. Kelly III, group executive director of IBM's Systems and Technology Group, pointed out that it normally takes 15–20 years for an emerging technology to reach the commercial stage, and by 2020, he stated, "we will be in full nanotechnology."
That means we need to develop the emerging technology right now! So far, however, a Roadmap group examining all the prospective nanotech approaches currently being studied (such as spintronics and nanowires) concluded that none of them appears capable of meeting all requirements for a useful, manufacturable alternative.
While there is a National Nanotechnology Initiative in the US, it is very broad with only a fraction of the work applicable to future electronics. To focus efforts, the Semiconductor Industry Association (SIA) is advocating a new Nanoelectronics Research Initiative, eventually leading to new centers of excellence with programs complementary to existing nanotech efforts.
Frank Robinson of Intel, formerly head of Sematech, described a wide array of device concepts for nanotechnology; some, like nanocrystals, might first be integrated into traditional CMOS structures. The problem with all of them is that they are embryonic, and far more science is needed, let alone the technological work to turn promising science into commercial devices. Robinson echoed Dr. Kelly's call for a new research initiative.
If such a program is launched, it appears it could prove evolutionary as well as, eventually, revolutionary. Some new technologies might prove helpful soon — by integrating them into CMOS processing — as the industry struggles to break through some looming red brick walls. At the same time, in-depth studies could help sort out which nanotech approaches have the greatest potential. Then intensive efforts could be focused on forging the best ideas into a viable alternative when CMOS hits the final red brick walls.
Scientifically, the road to nanotechnology could be an exciting journey. Instead of force-fitting nature, we will have to probe much deeper into its often-mysterious bag of tricks, not only finding mechanisms for self-assembly, for example, but learning to harness them to create new ways of processing materials and structuring devices.
 |
The benefits of the electronics/ information/ communications revolution have proven to be so great that we can't afford not to do it.
Robert Haavind
Editorial Director