Thinking small has made it possible for light particles to replace electrons as the primary information carriers of the future.
Last year, inside the vast aerospace-industrial corporation known as United Technologies, a clean-room crew began manufacturing a new kind of integrated circuit. These circuits superficially resemble miniature electronic devices. But instead of maneuvering electrons through copper wires or silicon chips, they guide photons, or particles of light, through reflective channels.
Using light to carry information isn’t news. Photonics became a multibillion-dollar industry in the 1980s, when telephone companies converted long-distance lines to fiber optics--glass fibers that guide light beams. It’s the incredible shrinking act from a fistful of fibers to chips the size of a fingernail paring that has people talking about a photonics revolution. The analogy we like to use, says Fred Leonberger, general manager of United Technologies Photonics, is that we’re at a place comparable to the early days of silicon integrated circuits, which was the early sixties.
Though fiber optics was embraced by the long-distance carriers, it was too expensive for local phone networks and too bulky to replace the electronic devices inside telephones and computers. But throughout the past decade the deep pockets of the military assured that the photonics revolution would continue as part of sophisticated weapons research. Some of the more dazzling results were demonstrated in such Persian Gulf War wizardry as smart bombs and armored vehicles capable of pinpoint navigation in the trackless desert.
Now, thanks in part to Leonberger’s chips, photons are poised to replace electrons in more and more aspects of everyday life. Over the next decade or two Leonberger and his colleagues expect to see many local communications networks make the conversion. They also expect to see photons helping control a new generation of machines we don’t even think of today as smart, like highly efficient automobile engines or fly by light aircraft. And like any technical revolution worthy of the name, photonics promises miracles in computing.
The idea behind every smart machine, including computers, is to gather information, manipulate it, and move it. For example, in a smart airplane, lightweight fiber-optic nerves threading through the plane to a central processor or computer would carry information from sensors inside each engine keeping track of temperature, pressure, vibration, and pollutants. (A similar technique is already used in medical procedures to monitor the human body.)
Photons are the ideal particle for moving information--not because they’re faster than electrons (electronic signals also move at nearly the speed of light) but because they keep to themselves. Electrons carry information by nudging their neighbors. Loose electrons in a copper wire bump into neighboring electrons, which pick up the signal and pass it along. Electrons also jostle the copper atoms, which soak up a signal’s energy by converting it to useless heat. A signal entrusted to such meddlesome carriers gets weak in a hurry.
Photons in glass fibers, on the other hand, ignore one another, and they barely acknowledge the glass except to bounce off its internal surface, ricocheting from side to side as they rocket down the fiber. Thus photons can offer a much more efficient medium for transmitting information. If you launch a milliwatt of optical power on optical glass fiber with today’s technology, says Leonberger, the loss is very low. If you look at coaxial cable--the copper wire used for cable TV--the loss numbers are hundreds of times higher. As a result, you need to send pulses containing hundreds of times more electrons over a given distance to make sure each bit of your signal gets through. If you switch to smaller photonic pulses, you can shove far more information through a fiber using the same amount of power.
Leonberger has been exploiting the photon’s aloof style of travel since the late seventies, first at MIT’s Lincoln Laboratory and then, starting in the mid-eighties, at United Technologies. There his first challenge was to use photonics to improve the gyroscope. In fact, it was the need to confine photons to the very small space of a practical gyroscope that drove him to invent his photonic chip.
Gyroscopes typically contain a wheel or disk spinning on an axis that resists being turned in space. They’ve been used since the late 1800s to sense rotation and thus guide ships, aircraft, and missiles. Most of the gyros in the world today are still mechanical, Leonberger says. But the one his photonic chip made possible is nothing like that. It has no moving parts. Instead, two beams of laser light travel in opposite directions around a fiber-optic coil. As the beams pass through each other, light waves going one way interfere with light waves going the opposite way, making dark spots where the waves cancel and bright spots where they reinforce. What happens when this optical gyroscope moves, Leonberger says, is that if the coil rotates, it’s as if the light going in the direction of rotation is moving faster than the light going in the opposite direction. The interference pattern shifts, and that change can be detected.
No electronic device could perform such a trick. At the most fundamental level, Leonberger explains, electrons like to interact with each other, and photons don’t. So you can send two light beams through each other, and they both keep going. You try to send two electrons, two wires, through each other, and you get a short circuit. You have to go to great pains to keep electrical signals from interacting.
The optical gyro may be as close as anything will come to the perfect motion-sensing device: lightweight, simple, and robust, yet exquisitely sensitive. An earlier version is now standard equipment on commercial jetliners like the Boeing 767. Still, getting from concept to hardware wasn’t easy. Leonberger had to start with the existing design for an optical gyro and shrink the signal-processing part to the size of a chip.
In the 1970s the equipment needed for an optical gyro would have filled a tabletop three feet square: lasers, lenses, beam splitters, filters, and so on. In the 1980s most of those components existed as different types of optical fiber. With fiber-size components fused together, the optical gyro shrank to a cube three inches on a side. That made it practical to consider for aircraft and for the navigation systems some Japanese automakers are adding to luxury cars. Now Leonberger’s gyro- optic chip is the size of a shaving pared from the three-inch cube: it’s less than 40 thousandths of an inch thick, one-tenth of an inch wide, and an inch long. Such a sliver-size chip can go just about anywhere you’d want to control the position of something in space, from the pitching platform of an oil rig at sea to the suspension of an ordinary car, where it could activate stabilizers for an ultrasmooth ride.
The technique Leonberger’s team invented to miniaturize the gyro’s circuits is called the annealed proton exchange method. It involves masking a three-inch-diameter wafer of lithium niobate--a piece large enough for nearly a hundred circuits--and bathing it in acid. Lithium ions diffuse out of the unmasked areas while hydrogen ions, or protons, diffuse in. The proton-rich channels bend light more sharply than the surrounding material. This sharp bending traps the light: it reflects off the internal surface of the channels and ricochets down their length just as it does inside glass fibers. Electrodes straddling the channels apply electronic pulses that control how the light beams travel through the photonic circuit. This setup substitutes for bulky optical components like lenses and filters.
In short, Leonberger and his colleagues invented an integrated optical circuit that could be mass-produced. That invention, he says, was really the beginning of our business. We grew our activity from gyros to all applications for which integrated optics would work. In theory, that’s just about everywhere electrons are now employed as data mules.
One of the hottest markets for Leonberger’s chips right now is in cable TV, where they imprint television signals onto beams of light. When cable networks finish converting to fiber optics, a single fiber will conceivably deliver hundreds of television stations as well as give each home instant access to a video library containing, potentially, every movie you’d ever want to see. But the plug-in time is still years away. So far, cable companies are converting only their high-power trunk lines to fiber optics: that’s where huge amounts of information are transmitted to huge numbers of users, so power savings can quickly repay the high initial cost. The lines leading to individual homes remain copper.
The same logic holds true for telephone lines. It’s mainly the heavily trafficked lines--the interstate highways of communications--where installing fiber optics makes economic sense. Today a single hair-thin optical fiber in a long-distance line can transmit more than a billion bits of information a second. At that rate you could transmit the contents of the Encyclopaedia Britannica--all 29 volumes--from Boston to Baltimore in less than a second. But in the local byways, electrons take over. Traveling on a copper wire, data funnel through a fast computer modem at 9,600 bits a second, so the encyclopedia would take more than a day to ooze into a computer’s memory. The information age will really begin when photons carry data right up to each television and on-line computer.
In fact, there’s no reason why photons need to stop at the end of a transmission line. They could work just as well inside a computer. Today’s silicon chips are fast when they crunch data that are just sitting there. But shoving fat, sticky gobs of electrons through itty-bitty wires-- say, from a storage chip to a processing chip--eats up relative eons of time, as anyone knows who’s stared at a dead screen while a program boots up or a document tootles along on some obscure internal journey. People are looking at moving data within the computer or between workstations, says Leonberger, but it’s very price sensitive.
To bring prices down low enough to make photonics a real threat to electronics, Leonberger hopes to repeat the electronics revolution’s recent history. Today a thumbnail-size silicon chip has more computing power than a room-size computer of the sixties, and there’s been a comparable reduction in cost. Hoping that photonic chips will touch off a similar revolution in the nineties, United Technologies opened a new plant last June in Bloomfield, Connecticut, dedicated to photonics. You can begin to imagine, Leonberger says, that if you’ve got an honest-to-God manufacturing plant, as the volume demands go up, the prices can come down pretty rapidly.