27 September 2013

Scientists build the first carbon nanotube computer, change computing world forever

Stanford scientist Max Shulaker knows that when it comes to carbon nanotubes (CNTs), expectations are high. That’s why when he and his team made the first computer entirely from carbon nanotubes, he made sure to preempt the most obvious points of skepticism. He stressed that their process for fabricating his CNT transistors is leaps and bounds ahead of what’s come before, boldly claiming that the team’s approach “can actually start to compete with silicon.” Big words, but if he’s right, his primitive nanotube processor could be the first basic glimmer of a whole new era in computing.

The device harkens back to the early days of computer engineering, when a proof of concept could shatter paradigms even while remaining slower than a human with an abacus. The 178-transistor computer isn’t quite that slow, but its main virtue is still that it exists at all. A homemade operating system allows for three tasks: this computer can count, it can sort through numbers, and thanks to the multitasking OS, it can do both at the same time. That might not seem like much, but as little as a year ago it would have seemed totally beyond the pale.

In a basic sense, CNTs are just rubes of graphene.

That’s because, despite all the incredible properties of CNTs, they remain virtually impossible to work with. At such small scales and with the need to maintain a near-perfect physical structure, even the slightest glitch in production can leave the whole utterly useless. As the researchers point out, making 98% nanotubes correctly isn’t remotely good enough when you’re making chips with billions of transistors.

Foremost among the problems is the tendency of CNTs to arrange themselves rather chaotically as they self-assemble. Nanotubes are a crystalline form of carbon, one of the only few such forms in existence, and they grow in a criss-cross pattern that formed unpredictable connections. This would seem to beg for a breakthrough in manufacturing, but these Stanford researchers have opted to accept and work with this property of the material. An as-yet-unspecified algorithm can apparently get around that problem by designing chips to work with any possible alignment of CNTs.

The other major problem with nanotubes is that they randomly develop into metallic, ill-formed versions about a third of the time. Again, the researchers found a work-around as opposed to a solution, choosing to simply ablate any bad apples with a strong electric current. Combined with their resilient chip design, this results in a fabrication process they call “imperfection immune.”

Single CNT transistors have already been clocked at many times the speed of silicon transistors while consuming only a fraction of the power, but this marks the first time that the individual pieces of a hypothetical CNT computer have come together to create a fully functional whole. They also ran 20 different instructions from the commercial MIPS instruction set, proving the generality of their creation.

Their fabrication process does take a few concerted steps.

This isn’t exactly a quad-core gaming powerhouse, but it’s by far the most sophisticated electronic made with CNTs to date. Alternative methods are already cropping up for replacing silicon with carbon, however, such as this recent discovery that DNA can lay the foundation for graphene transistors, a close cousin of the carbon nanotube.

The carbon computers are coming, and not slowly either. Their raw ability to switch on and off make them attractive to speed demons, while their small size and efficiency with power are dreams come true for designers and engineers. It’s certainly possible that carbon nanotubes could give us an iPhone processor clocked at hundreds of gigahertz, but they could just as plausibly lead to a weaker chip small enough to fit in a contact lens.

CNTs have the potential to change not just how fast processors are, but what they can fundamentally do. Right now the research is focused on industrial processes, but soon enough it will be on world-changing implementation.

Now read: Graphene transistors based on negative resistance could spell the end of silicon and semiconductors

Research paper: doi:10.1038/nature12502 – “Carbon nanotube computer”


Post a Comment

Get every new post delivered to your Inbox.


Copyright © 2015 Tracktec. All rights reserved.

Back to Top