Businessweek Archives

The Light Fantastic

Cover Story


Embedded in the new dam spanning Vermont's Winooski River are four miles of glass fibers. Laser light racing through them will warn of strains and stresse s long before failures occur. In late April, in fact, the fiber optic sensors alerted dam operators to a turbine gear that was about to break.

In a university of Pennsylvania lab, scientist Britton Chance and Arjun Yodh have deciphered the chaotoc light patterns created when harmless low-power laser beams shine through the human body. Their goal: light-based medical imaging that will detect brain umors and other deseases far more cheaply than today's magnetic-resonance systems.

Laser beams flash through the AT&T Bell Laboratories room where Alan Huang is designing a computer that moves data via light instead of electrons. His latest optical processor can run a washing machine. "We've come from the crazy to the impossible to the impractical," says Huang. Someday, he believes, his technology will spawn astonishingly powerful computers

In only a few decade, the enormous power of silicon chips has transported much of the world from Industrial Revolution to the Information Age. The next great leap, many scientist argue, will come from a technology called optoelectronics--the marriage of light and electricity. Once it's fully harnessed, they say, light will supercharge the speed of communications and omputers and bring new waysof probing everything from the atmosphere to the body. The possibilities "are truly amazing" says John Day, president of market researcher Strategies Unlimited. 'We are moving from the world of the electron to the world of the proton"--the most basic unit of light.

This light fantastic is already shaping everyday life. Each hours, millions of voices and megareams of data zip through fiber-optic lines on steams of phontons. Every time you use a laser printer, buy from a store that has bar-code scanners, gaze at a laptop computer, or hear Modonna on compact disks, the words, purchases, data, and tunes are communicated by light beams. Sales of optoelectronic devices from AT&T, Hewlett-Packard, Sony, NEC, Fujitsu, and countless small innovators have been rising 15% a year and will total $39 billion in 1993, estimates the Optoelectronics Industry Development Assn. (OIDA). "Optoelectronics is becoming the backbone of the Information Age," says Arpad A. Bergh, a top official at Bellcore, the research and development arm of the Baby Bells. "Without it, we could not collect, store, or transmit information."

And yet, the Optoelectronic Age is just dawning. As there was a formative leap in the 1960s from separate transistors wired together to thousands of transistors etched on silicon chips, so optoelectronics is poised to move from individual lasers and devices "toward integrated-circuit technology," says University of Illinois photonics pioneer Nick Holonyak Jr. At Bell Labs, Bellcore, and other facilities, scientists are building prototype chips that contain thousands of microscopic lasers. Instead of relying on electrons coursing through tiny wires, these chips will send and receive messages by flashing the lasers on and off millions of times a second.

Foreseeing such advances, the Clinton Administration has made a nationwide fiber-optic "information superhighway" a centerpiece of its technology policy. The Pentagon, meanwhile, is pouring millions into optoelectronics for military gear and civilian spin-offs such as flat-panel displays (page 48).

Both efforts are being spurred by intense rivalry with Japan. The fundamental optoelectronic inventions were U.S.-made. But today, Japan dominates markets for CD players, CD-ROMS, and flat-panel displays. It is investing billions in cutting-edge R&D and plans to link every home and business with fiber optics by 2015. "Japan is betting the country on optoelectronics," says Peter F. Moulton, vice- president of the research unit at Schwartz Optics, a Boston-area laser maker.

INTERCEPT. Such a commitment is not surprising, given the potential gains. In telecommunications alone, replacing poky electronics and copper wire with photons can boost capacity of transmission lines 10,000 times. The same approach could make today's computers the equivalent of Model Ts. At IBM, AT&T, Martin Marietta, and Honeywell, engineers are building hybrid systems in which beams of light replace wires that connect chips and computers. Big advances also lie ahead for computer-memory technology, says Scotty R. Neal, president of AT&T CommVault Systems, which makes optical storage devices. "Revolution is an overused word, but it's happening. You'll be able to put more information than all the world's books contain in a box and have it accessible in seconds."

The technology's impact will go well beyond that. The coming decade will bring optosensors that monitor factory processes, spot stresses on buildings and other structures, or detect submarines (page 50). In cars, optical sensors may control engine performance and help avoid crashes. Add it up, and Tokyo's analysts foresee optoelectronics sales in Japan alone of more than $300 billion by 2010. Because of this technology, "the 21st century will be a very different place," muses Phil Anthony, head of optoelectronics research at Bell Labs.

TOO MUCH HYPE? How different is still a matter of conjecture. As with any nascent technology, "our predictions will almost certainly be wrong," says Bell Labs physicist David A.B. Miller. When the semiconductor laser--a building block of optoelectronics--was invented in 1962, few expected that its first mass market use would be for listening to music through CD players so complex "we shouldn't let ordinary people buy them," jokes Robert W. Lucky, Bellcore's vice-president for applied research. Still, Lucky and others wonder if visions of optoelectronic miracles are mostly hype. "Photonics is so beautiful, it's got to be good for something else," he says. "But I don't know what."

To the technology's boosters, Lucky's skepticism is reminiscent of the pessimists who called electronic integrated circuits impossible. But it's easy to see where he's coming from. To examine the details of the technology is to enter a world of exotic materials and esoteric devices largely governed by the topsy-turvy rules of quantum mechanics, where light simultaneously acts as particles, or photons, and as waves.

At its simplest, optoelectronics involves transforming electricity to light and back again, largely through the marvel of semiconductor materials. Through intricate engineering of their structure, semiconductors such as gallium arsenide can be made to give off light when a current is passed through them. The simplest of these devices are light-emitting diodes (LEDs), invented in 1961. Today, 20 billion are made a year, for everything from the blinking lights on cellular phones to the high-mounted center brake lights on newer cars.

The next advance came in 1962. The earliest lasers required a ruby or cumbersome tube of gas to generate light. To make smaller devices, researchers at IBM, GE, and other labs put mirrors on each end of tiny, light-producing regions in the semiconductor. When light bounces back and forth between the mirrors, it's amplified until it's bright enough to shine through one mirror--and make a pure beam of a single wavelength, or color, of light. This device, the size of a grain of salt, is a semiconductor laser. Today, these are at the core of CD players and many other products.

Semiconductor lasers are also the engines for fiber-optic transmission. Through what Bell Labs' Anthony calls an "accident of nature," some of them produce light at precisely the infrared wavelengths at which optical fibers are most transparent--the quality that lets them carry data great distances. Until recently, however, there was a bottleneck: Light pulses fade as they travel the fiber. So every 35 miles, they were converted back to electricity, amplified, reconverted to light, and re-sent.

This limitation has disappeared with the invention of the optical amplifier. Emmanuel Desurvire's bosses were skeptical when, as a new Bell Labs researcher in 1986, he attempted to make such a gadget. Desurvire persevered, however. Building on advances in England and Japan, he laced glass fibers with a rare earth element, erbium, and devised a way to pump energy continuously into the fiber from an external laser. This boosts the brightness of light pulses--and ends the need for electronic repeaters. Dramatically faster fiber-optic networks are thus possible--as is light- based computer processing. Desurvire's device "is about to revolutionize the industry," says Tingye Li, who directs light-waves system research at Bell Labs.

'FLY BY LIGHT.' Among the most avid followers of all this are defense contractors. In the past decade, Hughes, Boeing, Honeywell, and Lockheed have begun or boosted optoelectronic R&D. The Lockheed-Boeing F-22 fighter and Boeing-Sikorsky Comanche helicopter will use fiber optics to ferry huge amounts of data between their various systems. In addition, the Pentagon's Advanced Research Projects Agency (ARPA) and some companies want to replace electronic controls with a "fly-by-light" system. Both approaches would save weight and keep information free of electromagnetic interference from other equipment, lightning, or nuclear blasts.

Defense companies also are working on sensors and optoelectronic components for radars so sensitive that they'll read license plates from satellites. And they're creating wings that are "smart" enough to telegraph dangerous stresses--for instance, through sensors that analyze how laser light is absorbed or scattered. A Lockheed Corp. system bounces light off of dust particles in the air to spot wind shears, the violent downdrafts that can cause plane crashes.

Essentially the same approach is being applied to such medical jobs as monitoring blood sugar. In fact, when Hewlett-Packard Co.'s Mark Chandler began assessing the biomedical uses of the technology for the OIDA last year, he "was surprised by the size of the market opportunities," he says. The group predicts that the medical market will exceed $10 billion a year by 2013. Lasers are already used to blast plaque from clogged arteries and perform surgery. Now, they're showing promise for imaging. "We could take a big bite out of health-care costs," says Penn's Britton Chance, who thinks optoelectronic detection of brain tumors could cost 99% less than magnetic-resonance imaging.

Perhaps the greatest impact of optoelectronics, however, will be in telecommunications and computing. Even as fiber-optic networks are extended to homes--New Jersey Bell Telephone Co., for example, plans to reach all of its customers by 2010--their capacity will soar. At IBM's Thomas J. Watson Research Center in Yorktown Heights, N.Y., scientists are creating an astonishingly powerful fiber network for rollout by the late 1990s. Their key innovation is boosting the number of wave- lengths of light that travel the fiber. Now, optical fibers carry only pulses of light at one wavelength, or color, like a one-lane highway. Using an array of lasers, each producing light with a slightly different wavelength, creates thousands of lanes within the same fiber.

DIALING FOR DATA. Each channel is big enough to send the contents of the Encyclopaedia Britannica in a second. So the system's overall capacity will be prodigious--large enough to profoundly alter the business of communications. Today, customers typically pay a phone company by the bit to transform information into light pulses and route it through the fiber-optic system. But with what IBM's Paul Green, manager of advanced optical networks, calls a "dumb, dark" network, a company could simply lease unused--or dark--fiber for a flat fee. It would use its own lasers to send data, as though on a private highway.

Optoelectronics will also help solve a resulting problem: where to store such vast amounts of information. Optical CD-ROMs, in which lasers etch information as tiny pits on the disk's surface, offer 100 times the capacity of today's floppy disks. That explains why sales of computer CD-ROM drives have doubled since 1990. And the potential of optical storage has barely been tapped. Scientists at IBM and elsewhere are devising ways to make the pits smaller and closer together--including using lasers that emit short-wavelength blue light--to boost disk storage capacity three- or fourfold.

More exotic solutions lie ahead. Researchers at Microelectronics & Computer Technology Corp. in Austin, Tex., have formed a company called Tamarack Storage Devices Inc. to commercialize a holographic memory system. The gadget shines two laser beams onto a material that changes chemically when struck by light. Much like the wakes of criss-crossing boats, the light waves interfere with each other. The result is a distinctive wave pattern, or hologram, that holds information and can be retrieved with another laser. This technology boosts storage capacity and speeds access times to the data. "It could replace floppy disks, tape drives, CD-ROMs, and even some hard disks," says Tamarack President John Stockton, who hopes the first products will reach the market next year.

NO WIRES. Optoelectronics also promises the computer power needed to move this mountain of data. In the 1980s, some researchers began laying the groundwork for purely optical computers--entirely light-based machines with no wires inside. In these, data would be moved exclusively by light beams that cross through each other without interference. The dream of an optical computer still lives: Bell Labs' Huang has built an all-optical processor. And in January, University of Colorado researchers showed off a rudimentary optical computer that can do simple mathematics. It's hard to program the machine for more complex tasks, however. So a practical version is still decades away.

In the meantime, most researchers have shifted to an easier goal--a hybrid computer that combines electronics and optics. "What makes sense is optoelectronic computing," says Sadik C. Esener, an engineering professor at the University of California at San Diego. IBM, Martin Marietta, Honeywell, and AT&T have formed the Optoelectronic Technology Consortium, funded with $8 million from ARPA, to develop optical links among a processor, two memory units, and the machine's communications ports. This so-called optical backplane will remove wiring bottlenecks in current machines to let them run faster. "Many people think this will be the next thing in system design, with broad applications for supercomputing and high-speed telecommunications," says David Lewis, head of optoelectronics at Martin Marietta's electronics laboratory in Syracuse, N.Y.

Optoelectronics also offers a radical shift in the way a computer processes information. "The real advantage is that an image not a bit is the unit of information," says Purdue University physicist David Nolte. "One image can contain 1 billion bits." To match fingerprints, for example, computers now search for similarities between two pictures by comparing the thousands of individual pixels, or dots, that make up each image. An optical processor, by contrast, can overlay two images and immediately spot similarities. University of Colorado computer scientist Kristina M. Johnson is using this to design an optical chip that can identify cancer cells in Pap smears.

The main hurdles to such bold ideas are complexity and cost. Because current lasers emit light from the side of a chip, each must be cut from the parent gallium arsenide wafer, then aligned by hand with a glass fiber about the width of a human hair--a costly job. A better approach would be to move from single lasers to integrated systems.

That leap will require some wizardry. At startup Photonics Research Inc. in Boulder, Colo., co-founder Jack L. Jewell, in collaboration with researchers at Bellcore, is using techniques adapted from chipmaking to fashion a new breed of microscopic lasers. These emit light upward from the wafer, making it possible to create chips that aim hundreds or thousands of lasers in the same direction, creating a powerful beam. Moreover, the job of aligning this array of lasers with a corresponding array of optical fibers should be much simpler, dramatically reducing manufacturing costs. "There are huge numbers of markets" for the improved devices, says Jewell. These include laser printers, optical memory for computers, and advanced telecommunications networks.

EAGLE 'EYES.' Another promising computer technology involves so-called smart pixels--chips with arrays of lasers and transistors--that send and receive information using hundreds or thousands of light beams. These hybrid devices are already capable of such rudimentary processing as routing signals from one path to another in a network. At Bell Labs, Miller's team has built experimental smart pixels that process a hundred million bits per second. Very fast conventional chips can do as well. But the advantage of smart pixels is that they can contain hundreds or thousands of channels operating at that speed. They should thus make it possible to build faster hybrid optoelectronic computers and zippier high-speed switches in communications networks. And smart pixels could be used in computer-vision systems capable of matching fingerprints or serving as the "eyes" for self-guided vehicles.

Smart pixels are a big step toward practical integrated optoelectronic circuits, but they're still tricky to build. The Bell Labs devices, for example, are fashioned from complex wafers that are processed to make both transistors and lasers. Another approach uses less exotic wafers and a technique called selective-area epitaxy. This allows the lasers and transistors to be made separately on the same chip, says Colorado State University electrical engineer Henryk Temkin. He expects the technique to take several years to perfect but predicts it will greatly lower the cost of making many optoelectronic devices.

Some researchers are even trying to coax light from silicon itself. If that becomes possible, manufacturers could build optoelectronic devices without using materials such as gallium arsenide, which is expensive and hard to work with. So far, scientists have gotten a few rays of light by riddling silicon with holes or lacing it with sulfur or germanium, then applying a current. But they're still in the dark about what's going on inside the chips. "We're looking for a breakthrough," says Dennis G. Hall, professor of optics at the University of Rochester, who is working with sulfur in silicon. One lesson from the past, he adds, is that in optoelectronics, it's reasonable to expect new discoveries.

Taken alone, each of these advances represents only a small leap. It may not even be obvious, for example, when optical fibers replace wires in computers--except that the machines will work faster. But the remarkable attribute of optoelectronics is its enormous breadth. It will touch everything from medicine to the waging of war. As the technology creeps into use, the quality of images on videophones will improve dramatically and the amount of information that data bases can hold will rise exponentially. Airplanes made of smarter materials will be safer, medical devices will be better and cheaper, until one day, opto-electronics will have fomented a revolution as far-reaching as that wrought by the silicon chip. John Carey in Washington, with Neil Gross in Tokyo

Tim Cook's Reboot
blog comments powered by Disqus