On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Modern lifestyles may not need to curb their appetites for smaller, faster smartphones and tablets when the digital age finally runs up against the physical limits of silicon-based computer chips. New research by Stanford University engineers might have just crowned a silicon successor by showing how to build a computer out of carbon nanotubes.
The 178-transistor computer only operates on one bit of information and a single instruction, which may seem unimpressive compared with modern computers that are based on 32-bit or 64-bit processors relying on millions to billions of transistors. But the Stanford University computer's use of carbon nanotubes—hollow cylindrical structures made from a sheet of carbon atoms—could pave the way for many future computing devices that run faster and use less energy. "It's a simple computer, but it's not a one-off computer," says Subhasish Mitra, an electrical engineer at Stanford and coauthor of a new paper detailed in the September 25 issue of Nature. (Scientific American is part of Nature Publishing Group.)
Smaller and faster electronic devices have long relied on engineers’ ability to shrink the size of silicon transistors—the tiny on/off switches that form the basis for modern electronics—so that manufacturers can pack more transistors onto each chip. Gordon Moore, co-founder of Intel Corp., predicted in 1965 that the density of transistors would double about every two years to allow for rapid progress in electronics.
But the smaller silicon transistors become, the more they end up wasting power as heat, leading to recent predictions about the end of "Moore's Law." Carbon nanotube transistors could maintain such progress by enabling electronics to become even faster and more energy-efficient at smaller sizes, because the nanotubes use very little energy to switch on or off. "We expect carbon nanotube transistors to offer up to a three times reduction in power and three times increase in performance—performance and power can be traded off against one another," explains Supratik Guha, director of physical sciences for IBM’s Thomas J. Watson Research.
IBM and Delft University in the Netherlands each built the first carbon nanotube transistors in 1997. IBM in particular wants to prepare carbon nanotubes for the day when the minimum feature size oftransistors shrinks from about 28 nanometers to something closer to five nanometers. (A nanometer is one billionth of a meter.) Such tiny transistors could allow for new mobile devices with better performance and improved battery life. "The Stanford result is a superb piece of research that shows that a universal computer could be made with carbon nanotube circuitry," Guha says. "We believed that this could be done—and this was a rudimentary, but clear demonstration."
Stanford's carbon nanotube computer only has transistors as small as eight micrometers (8,000 nanometers) because of limits in the academic lab's growing process for the carbon nanotubes. But the Stanford team headed by Mitra and H.-S. Philip Wong has excited Guha and other researchers because it developed a design that bypasses problems that have dogged researchers trying to grow carbon nanotubes: growing and aligning them perfectly in parallel lines. Even the smallest percentage of misaligned nanotubes could cause errors in the electronic circuitry. Although “we have carbon nanotube growth that guarantees 99.5 percent alignment," Mitra explains, that “0.5 percent is still a big number that can mess up digital logic function" in computer chips requiring millions or billions of nanotubes.
A second problem arises from a small number of carbon nanotubes acting like metallic wires that always conduct electricity, rather than semiconductors that can be switched on and off.
The Stanford engineers solved the alignment problem with a special algorithm that etches circuit patterns in the carbon nanotubes in a way that guarantees they can work despite misaligned nanotubes. The team also tackled the metallic conductivity problem by switching off the properly semiconducting nanotubes and running electricity through the metallic ones to vaporize the latter via the intense heat buildup.
Carbon nanotube computing probably won't appear outside the lab for another 10 or 15 years, says Max Shulaker, a Stanford Ph.D. candidate and lead author of the Nature paper. But he pointed out that the process of building carbon nanotube computers is compatible with the manufacturing processes of today's silicon-based semiconductor industry—a factor that could make it easier to reach industrial-scale manufacturing. "This research definitely gives an answer to the important question of whether carbon nanotube computing can scale for manufacturing, " Shulaker says. “It shows it is indeed scalable.”