WHY THIS MATTERS IN BRIEF
As transistors reach their physical limits we need to develop fundamentally new computing architectures, and this is just one example of the future of computing.
Interested in the Exponential Future? Connect, download a free E-Book, watch a keynote, or browse my blog.
The world of computing is, frankly, getting weirder – after all how else would you describe the emergence of biological and DNA computers where videos are stored and replayed from living bacteria, and where researchers have turned human cells into dual core computers. And that’s before we discuss the emergence of chemical, liquid, molecular, neuromorphic, and quantum computers – all of which are insanely more powerful than today’s computer platforms in their own right and can pack the power of today’s supercomputers and hyperscale data centers into devices the size of a fingernail, and something the size of a traditional office desk.
Now though a team of academics at MIT have unveiled the world’s most advanced chip yet that’s made from carbon nanotubes – cylinders of carbon whose walls are the width of a single carbon atom – and, again, yet another quirky technology that’s already helping bioengineer plants that can grow on Mars, cure paralysis, create cables for future space elevators, and helping car manufacturers slobber at the prospect of an electric car that can drive 16,000km on a single charge.
The new microprocessor, which is capable of running a conventional software program, could be an important milestone on the road to finding silicon alternatives as the electronics industry struggles with a slowdown in Moore’s Law, which holds that the number of transistors that can be packed on a silicon processor doubles roughly every couple of years. This trend is facing its physical limits: as the sizes of the devices shrink to a few atoms, electrical current is starting to leak from the metallic channels that shuttle it through transistors. The heat that’s released saps semiconductors’ energy efficiency – and may even cause them to fail.
Carbon nanotubes could be the perfect solution. Not only are nanotube transistors faster than silicon ones, studies have found that chips made from nanotubes could be up to ten times more energy efficient. This efficiency boost could significantly extend electronic gadgets’ battery life.
Researchers have been working on alternative chips involving the molecules for decades, but manufacturing headaches have kept the processors stuck in research labs. In a paper published in Nature, the MIT team says it has found ways to overcome some of the biggest hurdles to producing them at scale.
One problem is that when carbon nanotubes are made, they come in two types mixed together: the first are semiconductors that are perfect for creating integrated circuits, but the second conducts electrical current like a wire, which sucks more power and can even undermine a circuit’s performance. To make the chips economically viable, a cost-effective way to minimize the impact of the latter group is needed.
Another problem is that to make the chips, a uniform monolayer of carbon nanotubes needs to be deposited over a wafer. But this has proven hard to do because nanotubes have an annoying tendency to bunch together. A bundle of them that lands on a transistor can knock it out of action.
These and other challenges intrigued Max Shulaker, an MIT professor who has worked on other notable projects in the field, and has received funding from the DARPA, the bleeding edge research arm of the US military, to develop nanotube technology.
The group of researchers he leads has developed a working 16-bit microprocessor built from over 14,000 carbon nanotube transistors that Shulaker claims is the most complex ever demonstrated. The techniques they have come up with can be implemented with equipment used for making conventional silicon chips, which means chipmakers won’t have to invest in expensive new gear if they want to make nanotube processors.
When they looked into the intermixing problem, the researchers discovered that some kinds of logic gates, which are fundamental building blocks of digital circuits, were more resistant to problems triggered by metallic-like nanotubes than others. That led them to develop a new circuit design that prioritizes these gates, while minimizing the use of more sensitive metallic ones.
To deal with the bundling problem, they coated a wafer in a polymer and then carefully washed it off in stages. This stripped off the nanotube clumps, leaving behind the monolayer needed to make the chip work most efficiently.
The chip that the MIT researchers produced using these techniques is capable of running a simple program that produces the message “Hello, World.” But if they are to displace silicon processors, carbon nanotube ones will ultimately need billions of transistors so they can run advanced software.
IBM, which a few years ago said it hoped to have carbon nanotube chips take over from silicon ones by 2020, is also working on projects involving the technology. But efforts have so far failed to come up with a way to translate lab breakthroughs into practical manufacturing. The new advances make the route towards doing this clearer.
“There’s no leap of faith required anymore,” says Shulaker.