…Many think silicon’s long run as the king of computing could be nearing an end. That’s because continually jamming more tiny transistors on a chip has become more difficult, expensive, and inefficient, not to mention the inevitable physical limitations — you can’t keep shrinking transistors forever.
Carbon nanotubes — long chains of carbon atoms thousands of times thinner than a human hair — have the potential to be more energy-efficient and “take us at least an order of magnitude in performance beyond where you can project silicon could take us,” according to H.S. Philip Wong, another member of the Stanford team.
Problem is, carbon nanotubes aren’t perfect either. They don’t always grow in perfectly straight lines, and a fraction of the tubes grown aren’t able to “switch off” like a regular transistor.
The Stanford team used a technique of “burning” off some of the imperfect carbon nanotubes while also working their way around other imperfections by using a complex algorithm. The final design consists of a very basic computer with 178 transistors that can do tasks like counting and number sorting and switch between functions. The computer’s limited power is due in part to the facilities available to the team, which did not have access to industrial fabrication tools.
So what we have now is basically a proof of concept for the first carbon nanotube computer, which is about as powerful as Intel’s 4004, the first single-chip silicon microprocessor released in 1971. But if this technology turns out to be a worthy successor, we’ll likely see devices that can not only compete with, but greatly exceed, the potential of silicon systems.