Intel toys with carbon nanotubes

RECENTLY, CARBON nanotubes have been getting a lot of attention as a replacement for, or more likely an addition to, normal semiconductor materials. What are they, and what do they do?

Carbon nanotubes are basically carbon atoms that form into a straw-shaped tube molecule which gives them a bunch of interesting properties, two of which are relevant to chip makers. There are two forms of the tube one is a good ballistic conductor, the other a semiconductor, and they only differ in their chirality (look it up).

In nature, they form at about two-third semiconductor, one-third conductor, making them pretty useless in aggregate for either function. Intel research is hard at work trying to determine how to not only make them at 100 per cent one form or the other, but also how to place them on a chip.

The chirality can be strongly influenced by the seed material or catalysts for the formation of the tubes, so you can nudge it one way or the other. For a chip with a billion transistors, you need to improve the error rate to one wrong molecule ever few billion, one off could kill a chip. Going from 33 per cent to .00000000001 per cent or so is not easy, and missing by a tiny bit means dead parts.

Once that is solved, there comes the question of what to do with the damn things. Currently there are two options, neither is really the clear winner right now, and both or neither may pan out. You can use them in the semiconductor form for traditional semiconductor roles, or you can use the conductive form for interconnects.

No matter what you use them for, there is another question, or more to the point, problem to be solved. Grow or place? You can make the tubes in a vat and place them with sub-micron accuracy a billion times, or grow them where they need to be. Both are not trivial things to do, and will be another research topic once you figure out how to make them in the first place.

In any case, there are three problem and two uses for carbon nanotubes in chips. With any luck, they will be solved before traditional methods run out of headroom, and progress can continue. They may pan out, they may not, that is what research determines

Source Techbuzz