IBM plots 'chip on a molecule'
Silicon on borrowed time?
A functioning processor on a single molcule has been created by IBM scientists, offering the possibility of super-fast processing by tiny devices.
The molecule in question is a carbon nanotube. The team at at IBM’s Thomas J. Watson Research Center, New York, created a type of logic device called a ring oscillator. It consisted of 12 bi-metal field effect transistors laid along the 18 micrometre length of the nanotube.
New Scientist reports developer Joerg Appenzeller explaining: "This isn’t about making the circuits smaller, it’s about making them faster. Nanotubes fit the characteristics we need to advance high-end processing."
Chip designers have long known that they are pushing the limits of conventional silicon technology. The effort to keep up with Moore's Law - which predicts that the number of transistors on a chip will double every 18 months - has meant more and more transistors have been packed onto silicon by improving manufacturing techniques. However this constricts the pathways electrons are able to take and increases resitance in the circuit, limiting its speed.
According to the research paper, published in the journal Science, the chief reason behind the problem is a phenomenon known as plasmonic resonance. An electron’s path is hindered when it becomes coupled with vibrations in the surrounding silicon lattice structure.
But because the carbon nanotube is a single molecule this problem is avoided and resistance much reduced. The upshot is the possibility of circuits running much faster. Technologists reckon it could one day pave the way for affordable near-terahertz processing.
This latest experiment was a proof of the principle that nanotubes can be used as a basis for electronics. The technology is a long way off doing useful computation - the nanochip achieved a 52MHz crawl.
Right now nanotube production is at best a hit and miss operation. The technology would have to be bettered controlled for the chip giants to invest in trying to put nanotech in their products. According to NS one of the team reckons it could be done in five years if a huge investment was made, but believes in reality it will take much longer.®
Sponsored: Hyper-scale data management