IBM creates first 9nm carbon nanotube transistor

IBM creates first 9nm carbon nanotube transistor

IBM's carbon nanotube transistor - CNTFET - research has led to the first sub-10nm prototype in existence.

Researchers at IBM have produced what they claim is the first experimental evidence of a transistor material that will be viable at process sizes of below 10 nanometres: carbon nanotubes.

As process sizes shrink and components become packed ever more densely, several challenges present themselves. Not least of these is current leakage, which confuses nearby transistors and corrupts the signal, as Intel researchers found in a white paper (PDF) detailing the struggles with smashing what they called the '10nm physical gate length barrier.'

Shrinking process size is vital in continuing the trend for increasing semiconductor complexity: only by reducing the size of transistors and the spaces between them can increasing numbers be fitted efficiently into equivalent die sizes. The process is so important, in fact, that it forms a central part of Intel's 'tick-tock' development cycle: one year a 'tick' introduces a refinement into the processor's architecture, while the following year a 'tock' boosts performance with a move to a smaller process size.

According to a paper published in the Nano Letters journal, IBM believes it has found a way to effectively bypass the 10nm barrier: by replacing silicon with carbon nanotubes.

It's not the first suggestion for the move to sub-10nm process sizes. Intel's whitepaper, written when 130nm was the norm, looked to the tri-gate transistor technology the company is only now implementing in Ivy Bridge as a possible solution to the problem. Where Ivy Bridge is a 22nm technology and tri-gate as yet unproven at significantly smaller process sizes, however, IBM's carbon nanotube technology is already working at 9nm.

'Although carbon nanotube (CNT) transistors have been promoted for years as a replacement for silicon technology, there is limited theoretical work and no experimental reports on how nanotubes will perform at sub-10nm channel lengths,' the team explains in the paper's abstract. 'In this manuscript we demonstrate the first sub-10 nm CNT transistor, which is shown to outperform the best competing silicon devices with more than four times the diameter-normalized current density (2.41 mA/μm) at a low operating voltage of 0.5 V. The superior low-voltage performance of the sub-10 nm CNT transistor proves the viability of nanotubes for consideration in future aggressively scaled transistor technologies.'

While IBM's work on carbon nanotube field effect transistors (CNFET) is impressive, it's far from ready for commercialisation. With Intel already going to market with its tri-gate take on the FinFET concept, there's plenty of time for solutions to silicon's leakage issues to be found in order to avoid a change of material when it comes time to release sub-10nm parts into the market.

Do you think future chips will be based on carbon, silicon, or something else entirely? Share your thoughts over in the forums.


Discuss in the forums Reply
Bede 30th January 2012, 12:16 Quote
Meh, they have a vested interest in keeping silicon - I doubt that their existing fabs could easily transition to carbon nanotubes.
SexyHyde 30th January 2012, 12:57 Quote
While you have a point bede, they will also be aware that they need to move to carbon or be stuck with a technology that will shortly be end of life. They have known for a while silicon is problematic at the current size. Moving to carbon will give them a significantly better and longer future.
maverik-sg1 30th January 2012, 13:04 Quote
Developing <10nm in size transistors is a great acomplishment and should be applauded - kudos to you all for that achievement.

I'd like to see some active competition in large scale manufacturing, this one has the potential to be just that - in reality how far away are we from seeing this in mass production? 10years?

If we are looking at 14nm fabs in just two years based on silicone and the new Tri Gate designs, 10nm should effectively be just 4 or at maximum 6 years away - better get a shift on Carbon Nanotubes.
Tattysnuc 30th January 2012, 13:26 Quote
Great stuff IBM - Now let's see some actual working prototypes of complex chips!

Be interesting to see if the changes in manufacturing process actually yield the purported benefits calculated from "simple" models used in the lab, or even if this can be developed into a mass production process...
Dublin_Gunner 30th January 2012, 13:42 Quote
@ Bede: They must re-tool their fab facilities each time they move to a smaller manufacturing process anyway. The main difference will be in utilising the old equipment when they move on (which typically goes from creating state-of-the-art CPU's to chipsets, and other control IC's and older CPU designs). But this process in fact will help the adoption further down the line.

Brief on what I mean.

Equipment used for fabrication:
Current: CPU 22nm > Chipset 32nm > Chipset / other 45nm
Future: CPU 9nmCNT > Chipset 22nm > Chipset / other 32nm
Future 1: CPU 7nmCNT > Chipset 9nmCNT > Chipset / other 22nm

I used 7nm as an example, I do not know what node they could use beyond 9nm CNT.
dark_avenger 31st January 2012, 05:49 Quote
Had to come up with something new, can only shrink silicon so far.

Bring on faster, cooler chips :D
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.

Discuss in the forums