IBM Is Shrinking Transistors With Rows of Tiny Carbon Nanotubes

Illustration for article titled IBM Is Shrinking Transistors With Rows of Tiny Carbon Nanotubes

In the semiconductor industry, size matters — and people are worried that it won’t be able make transistors any smaller. But a team of IBM scientists has now published research showing how carbon nanotubes could help.

The size problem is a simple one: continue shrinking down the circuity required by transistors, and the electrical resistance — particularly through the connections to the tiny electronic devices — increases. As resistance increases, temperatures increase, efficiencies decrease and the circuits begin to fall over. In the past, IBM researchers have claimed that carbon nanotubes could provide a useful alternative. Now, in a new paper published in Science, they’ve described a new way to make tiny transistors using parallel rows of carbon nanotubes.

Carbon nanotubes are theoretically useful cylinder made from a one-atom-thick sheet of carbon atoms. Theoretically useful because — like USB cables, headphones and other wires — they form a tangle when left to their own devices. But the researchers have managed to align them, lay them them out regularly spaced rows and deposit them on silicon wafers. Once that was all done, they managed to weld tiny metal contacts made from molybdenum to their ends.


As the carbon nanotube works as a semiconductor, the result is a transistor that works much like its silicon counterparts — but the end connection in combination with the carbon tube seems to cut resistance at the connections, as the researchers show in Science. In turn, the IBM researchers reckon that the technology could be used to build transistors at least as small as 5 nanometers. For some context, until recently manufacturers have been able to produce them at a scale of 14 nanometers, and IBM recently announced that switching to a silicon and germanium mix could cut that to 7 nanometers at best.

In other words, the use of such end-connected carbon nanotubes could help engineers pus the limits of chip design a little further. And that might just allow Moore’s Law to cling on for a little longer yet.

[Science via New York Times]

Image by IBM


Share This Story

Get our `newsletter`


I’m really curious how things will go once we hit the physical limits of silicon. Will everyone have time to catch up to Intels manufacturing process lead, or will Intels R&D budget allow them to leapfrog the limits with another technology sooner?

By the way other foundries advertising 14nm at the same time as Intel is shipping it doesn’t mean they’ve caught up for various reasons. Plus Intel uses 14nm for both front end of line and back end of line transistors while everyone else uses an n-1 process for back end