Computers get faster and communication signals get faster, but the interface between them–where the electrons in the computer circuits are converted into photons for the fiber-optic cable–remains clunky and slow. New transistors that rely on virtual particles called excitons could change that. An exciton is a state of electrical excitement that can pass from one atom to another, much as an electric current does. When an exciton loses energy, it emits a photon, so excitons are good at translating between electrical and optical signals.
“The problem in existing systems is the barrier at the interconnect between the optical signal and the electrical signal,” says Alex High, a graduate student at the University of California, San Diego (UCSD), who conducted the research along with colleagues there and at the University of California, Santa Barbara. “This cuts out that extra step. Because excitons are carriers of light, you can manipulate them, do logic processes on the light in exciton form, and then release that light in another place.”
The researchers have created tiny, supercooled integrated circuits made of gallium arsenide that can send exciton signals in different directions or merge two signals into one–jobs necessary to handle the rudiments of computer logic just as electronic circuits do. “The computation speed by itself may not be much faster” than a conventional chip’s, says Leonid Butov, who led the research. “Where we can gain speed is in the transformation of the photons.” Butov has so far demonstrated a switching speed of 200 picoseconds, which includes both computation time and the transformation of the photons into excitons. The speed of conventional conversion and switching varies with the material, but it’s about an order of magnitude slower than Butov’s switch. (Also on the market is an all-optical switch that doesn’t have to convert optical signals into electrical ones. It has a switching speed of 50 picoseconds, but due to its large size, it can perform only rudimentary operations.) And 200 picoseconds is “not even the final answer yet,” says Butov. “We may be able to make it considerably faster.”
A smoother optical-electronic interface has wide implications. Fiber optics is the most efficient way to carry large amounts of data at the speed of light, and it’s used in a myriad of applications, from telecommunications to temperature sensing to simply carting information from one computer chip to another. But at some point, optical signals almost always need to be converted into electrical signals–whether it’s so your desktop PC can understand them or so they can be amplified during a long trip. Not only is that conversion slow, but the traditional converters are expensive, relatively large, and power hungry.
The new integrated chip, however, takes in light as is, operates on it as necessary, and spits light out the other side. Whenever a photon hits the chip, it forces a negatively charged electron out of a semiconductor atom, leaving behind a positively charged “hole.” Without intervention, the electron and the hole simply recombine. But the UCSD team uses so-called quantum wells to keep the electron and hole separate yet close enough to remain bound into a single unit. This carefully bound “particle” is called an indirect exciton, and it has the odd property that it will move when placed in an electric field even though it is neutrally charged. Nudged by electric fields, the excitons scurry through the chip along a prescribed path until allowed to recombine. Then they release their energy in a flash of light that sends the communication signal off to its next destination.
The prototype chips have to be cooled to temperatures of less than -234 ºC. But the researchers are confident that they’ll be able to re-create their delicate quantum wells in semiconductor materials that allow excitons to form at room temperature.
Building the chip such that the electric fields didn’t rip apart the excitons was one of the main hurdles for the scientists. “If this field becomes too strong, it can tear apart the electron and hole. Designing the gates that define exciton circuits required new design ideas and extreme care in their implementation,” says Leonid Levitov, a scientist at MIT. “I believe this is a fundamental achievement in exciton physics which may also have very practical consequences and lead to applications.”
Butov and his colleagues agree. Their colorful pictures of their chips show streams of excitons that can be forced to veer either to the left or right, that can be split to go down both left and right paths, or–in reverse–that can travel in along two paths and combine into one stream. While the team would like to demonstrate additional tricks, such as amplification of a signal, the photos show that their circuits can be designed to perform any logical operation that a traditional electronic chip can do. And, the researchers say, their chip will do it better.