Computers are always improving, getting smaller and faster with bigger and brighter screens. To forward-thinking scientists, however, all the rah-rah about the latest tablet innovation is really just tinkering at the margins. If you really want to build a better computer, they say, replace electricity with light.
“Light can compute a lot faster and carry more information than electrons,” says Eric Mazur, a physicist at Harvard University who recently developed a material that moves light-based computing slightly closer to reality.
Contemporary electronics are, well, electrical. In a computer, information is encoded in electrical current, which itself is a stream of electrons: That electrical current can be either on or off, which corresponds to the 0s and 1s that form the foundation of computer logic.
Optical communications, by contrast, have many more “degrees of freedom” than electronic communication. Instead of being simply on or off, light can take different wavelengths, for example. You can have a red beam, a blue beam, an infrared beam; they each carry a separate signal and don’t interact with each other when sent simultaneously down a fiber optic cable, conveying a denser bundle of information than an electrical signal could.
Light is already used to convey information in much of our long-haul data infrastructure — it encodes phone calls and carries Internet information via fiber optic routes around the world. But when light-based information gets to your house, it has to be translated into an electrical signal in order to be used by electrical equipment like phones and computers.
“Once the Internet gets to the home, there’s a box that takes the light signal out of fiber optics and converts it to electric,” says Mazur. “This is the biggest bottleneck. From the end of the fiber optic cable into a computer, that’s where everything slows down.”
A better way would be to have the light signal go straight into your computer, but several significant hurdles stand in the way.
“Changing to a completely photonic system implies computing completely with photons, which obey different laws compared to electrons. We have to rethink the whole process,” says Alessandro Salandrino, an electrical engineer at the University of Kansas who works in this area.
One of the biggest challenges involves finding a way to, in essence, shrink or squeeze light so that it can be processed on the nanoscale of a silicon chip. To think about this, consider the way your computer works now. Electrical information comes into the computer through, say, a USB cable, which might be a centimeter wide. But that same electrical information has to be processed on a silicon chip, which contains transistors whose channels might be only 15 nanometers long.
Electrical signals can be shrunk very easily from the macroscale of a USB cord to the nanoscale of a chip. Light is less accommodating.
“It’s difficult to squeeze light from larger dimensions down to the nanoscale. Light doesn’t like to be squeezed quickly,” says Mazur.
The difficulty lies in the fact that light has a certain characteristic length, its wavelength. Wavelength is the distance from peak to peak in a frequency of light. A wavelength of light might be 1,550 nanometers. This is much longer than the size of some of the computer components that are meant to manipulate it, which makes the two somewhat incompatible. “The way electromagnetics work, it’s difficult to confine light to regions that are smaller than its wavelength,” says Salandrino
There are ways, though, to shorten wavelengths. Light changes when it moves through different kinds of materials, and one way these changes are measured is using what’s known as the “refractive index.” Air has a refractive index of 1 — it’s the baseline medium. Water has a refractive index of 1.3, meaning light slows down (its wavelength shortens) when it moves through it. A material with a refractive index below 1 accelerates light and lengthens its wavelength.
And that is the kind of material Mazur has created. In a study published online in October in Nature Photonics, he and his colleagues reported that they’d fabricated what’s called a “zero-index metamaterial” (a material with a refractive index of 0). The material is made from silicon wafer, polymers, and gold mirrors. When light enters it, something bizarre happens — the wave spreads out over the entire material and in a sense the speed at which the crests move becomes infinite. In this translated state, light becomes easy to manipulate. You can “shape the wavelength of your light at will,” says Salandrino.
As Mazur puts it, “You can squeeze it down instantly, twist it, bend it. It’s the perfect coupling between a nanoscale circuit and ordinary connections. It opens the door to a much better interface between nanophotonics and world we live in.”
The creation of this zero-index metamaterial does not mean that optical computers will be available tomorrow. Mazur and Salandrino both said it’s impossible to predict when they might become real products, though the scientific community is fully engaged in pursuit of that vision. And when and if optical computers arrive, you’ll know they work, only because physicists and engineers figured out how to direct light on the smallest imaginable scale.
Kevin Hartnett is a writer in South Carolina. He can be reached at firstname.lastname@example.org.