The silicon semiconductor industry has chugged along for more than 50 years. Like a steamroller, it has trundled over bumps and holes, while defying repeated warnings that it was running out of fuel or was about to be overtaken by flashier competitors.
So we technologists are understandably reluctant to speculate about the end of silicon. And yet, speculate we must. After decades of steady improvements to the efficiency and speed of our computer chips, brought about by physically shrinking the dimensions of silicon transistors, we’ve reached a point where the massive effort to miniaturize those switches ekes out only very modest gains in performance. The steamroller still rolls, but it’s slowing down, and the maintenance and upkeep on it are fast becoming unsustainable.
Researchers are pursuing many options to keep integrated electronics on its exponential arc. One possible approach is to simply give up on improving the silicon transistors themselves, and instead focus on changes to the architecture or design of computers. This strategy has had some success in the past, when, for example, we moved from single cores to multicore processors, or began incorporating specialized accelerators into systems. Yet there are only a limited number of design tricks that can be played, and we have used many of them already.
So future progress must still involve improving the underlying switches. Modifying materials and device geometries can create transistors with better electronic properties than silicon transistors have today. But such evolutionary approaches will yield only small benefits. To get bigger gains, we’re left with one option: use novel nanotechnologies to supplant silicon altogether.
There are many candidates for this role. But the most exciting and mature contender by far is the single-walled carbon nanotube, a rolled-up sheet of linked carbon atoms. If you’re experiencing some déjà vu now, it’s no wonder. There was great enthusiasm for carbon nanotubes in the 1990s, and a number of big-name semiconductor companies began investigating them. But the carbon nanotube fell out of fashion when researchers ran into obstacles in circuit fabrication that seemed too hard to overcome. When the first measurements of the electronic properties of graphene—flat lattices of carbon atoms—were reported in 2004, many were all too eager to abandon carbon nanotubes and take up that shiny new technology.
Now the carbon nanotube is back and better than ever. Unlike its much-hyped two-dimensional cousin, the carbon nanotube comes in a natural semiconducting form, which means it can be turned on and off to make a binary switch. And recent research into the material has moved well beyond the demonstration of simple logic gates. In 2013, for example, our group at Stanford University built the first complete digital system—a basic computer—entirely out of carbon-nanotube circuits. What’s exciting is that the fabrication and design techniques we employed to build that computer can be seamlessly integrated into a semiconductor fabrication facility today and used to manufacture chips with the billions of transistors needed to compete with state-of-the-art silicon.
(Source: IEEE Spectrum)