By Dario Gil
IBM Research scientists launched the nanotechnology revolution when they designed the scanning-tunneling microscope in 1981, and our researchers have achieved numerous nanotech breakthroughs since then–including being the first people in the world to move single atoms.
Now comes an advance that delivers on the promise of nanotechnology–potentially extending the life of Moore’s Law by enabling major performance improvements over today’s conventional chip technology.
A team at our Yorktown, New York, lab overcame one of the most daunting challenges facing the chip industry by demonstrating the first carbon nanotube transistors that don’t suffer from reduced performance as they’re shrunk to smaller dimensions. Read about their invention in the Oct. 2 edition of Science.
This is the kind of breakthrough that we’re committed to making at IBM Research via our $3 billion investment over 5 years in research and development programs aimed a pushing the limits of chip technology. Our aim is to help IBM produce high-performance systems capable of handling the extreme demands of new data analytics and cognitive computing applications.
Because of the limitations of silicon semiconductors and the laws of physics, the tech world faces huge challenges in extending the life of Moore’s Law, the formula that has driven much of the progress in electronics over the past 50 years. So scientists have been experimenting with new materials and new processes to pack more transistors on the chips that power everything from smartphones and connected cars to number-crunching supercomputers.
Carbon nanotubes are one of the most promising replacements for silicon in semiconductors. That’s because they can be used to build incredibly tiny devices and they use less power to switch the state of transistors–which is how we create the ones and zeros essential for digital computing.
The tech industry faces several major challenges, but we’ve overcome one of them with our latest breakthrough–where our scientists demonstrated a way to shrink transistor contacts without reducing the performance of carbon nanotube devices.
In traditional silicon-based chips, each transistor consists of a source wire for an electrical current, the “in” wire; a drain or “out” wire; and a gate between them, which is a sliver of metal or metal-like material that switches the transistor on or off in response to electrical charges. When the gate is switched on, electrons run on a metal strip through the transistor on top of a channel made of silicon.
The problem is that silicon doesn’t have the right properties to scale down to the smaller dimensions required to improve chip performance at a rapid rate and a low cost. Carbon nanotubes make for a great channel material at smaller scale, but there were problems with connecting them to the “in” and “out” wires of transistors. Our scientists overcame those impediments by using a new kind of metal contact between the nanotubes and wires and by running the channels straight into the sides of the wires. The conventional method is to run the channels under the wires, which requires curved connectors and causes traffic jams of electrons. Our solution makes it possible to shrink the channels to dimensions of less than 10 nanometers, and thereby shrink the overall size of the transistors, without sacrificing performance.
There’s still engineering required to make the new designs manufacturable at a massive scale. IBM and the rest of the scientific community still face two major hurdles before carbon nanotubes will fulfill their promise. 1) We’ve got to figure out how to purify carbon to the point where only semiconducting molecules remain–and not metallic, electricity-conducting ones. 2) We have to be able to manipulate nanotubes, which are 10,000 times thinner than a human hair, so we can build billions of tiny transistors on a single chip.
At IBM Research, we have teams working on these challenges. We’ve made great progress in purifying carbon, and we’re now within striking distance of the answers we need. We’re still seeking a breakthrough in the placement of nanotubes. That’s an incredibly difficult puzzle to solve.
But difficult puzzles are what make science so interesting–and so valuable . So we keep at it.
We believe that by remaining true to our commitment to advancing the science of computing at the most basic level, we’ll contribute not only to IBM’s success but to progress for business and society.