How We Became Wired With Glass
Today fiber optics is the backbone of the Internet and global communications. Thirty-five years ago most people thought it was impossible.
DANIEL COLLADON DIDN’T SET OUT TO CREATE A telecommunications revolution in 1841. The Swiss physicist simply wanted to show an audience in Geneva how a horizontal jet of water broke up into droplets as gravity pulled the liquid in a downward arc. The water was hard to see with the meager illumination available at the time, but Colladon had a bright idea. He piped a beam of sunlight into the darkened lecture hall with a hollow tube poking through closed window shutters. He then aimed the beam along the water jet. The smooth surface of the jet guided the light until turbulence broke the falling arc into droplets, which sparkled with light. In a paper published the following year, Colladon called the effect “one of the most beautiful, and most curious experiments that one can perform in a course on optics.”
In 1853, using an electric arc light instead of sunshine, Colladon designed a water-jet display for the Paris Opéra, tinting it red to portray a stream of fire. The crowds loved it. The Paris Opéra soon hired one of Colladon’s assistants as a full-time lighting engineer. The engineer also designed illuminated fountains as playthings for the rich. Elaborately lit fountains became common public spectacles toward the end of the nineteenth century.
The phenomenon that guided light along Colladon’s water jets was total internal reflection, the same principle that makes cut gems glitter and bounces light around the inside of water droplets to separate its colors and create a rainbow. It occurs under certain circumstances when two transparent materials—water and air, for example—meet along an interface. Normally, when light crosses the interface, it is refracted, or bent, because of its differing speeds in the two materials. This tendency to slow down light waves is quantified in a substance’s refractive index.
But if the second material has a lower refractive index, and if the light strikes the interface at a low enough angle, it will not cross into the second medium but will instead be reflected back into the first. This behavior is called total internal reflection. A century and a half after Colladon, the same principle that lit up his jet of water now guides light around the world on the fiber-optic backbone of the global telecommunications network.
It was well known in Colladon’s time that bent glass rods could guide light. In the late 1800s inventors patented devices using the technique to illuminate microscope slides and the inside of the mouth. It was also known that glass could be stretched into thin fibers, which are essentially fine, flexible glass rods. But no one seems to have thought of combining the two phenomena to make light-guiding fibers until the 1920s, when two young men independently came up with the idea. One was Clarence W. Hansell, who managed RCA’s radio-transmission laboratory at Rocky Point, New York. The other was Heinrich Lamm, a medical student at the University of Munich.
Hansell needed to read an instrument dial that was in an inaccessible location. He had seen glass rods guide light and knew that individual glass fibers could do the same thing. On December 30, 1926, he sat down and put the pieces together in his laboratory notebook: “It is only necessary to make up a cable of parallel laid quartz hairs or strands of similar material, the ends of which can be cut off plain. One end of this cable can face the instrument and another can face toward the observer. The light from the instrument will fall on the ends of the quartz fibers and be transmitted through them coming out as an image in the other end visible to the observer.”
The idea is simple. Each fiber picks up light from a spot on the object being observed and transmits it to the other end of the bundle. If the fibers are arranged in the same way on both ends, the illuminated fibers will display an image of the object. Leaving the fibers loose in the middle makes the bundle flexible.
Hansell refined the idea before turning it over to RCA’s patent lawyers. He proposed a flexible periscope for military uses and a medical instrument to peer into the human stomach. He envisioned scanning an array of fibers across a sheet of paper as the input for a facsimile system and designed some vacuum-tube circuits for the job. Then he went back to building long-distance radio-transmission systems. No surviving records say he ever saw an image through a bundle of fibers.
Lamm was inspired by Rudolf Schindler, a pioneering gastroenterologist who had developed a “semiflexible” instrument to look down the throat. It wasn’t flexible enough for Lamm, who talked Schindler into giving him money to buy some commercial glass fibers for an experiment. Lamm meticulously combed the fibers into a fat bundle several inches long. He could not see through the finished bundle easily, but he could photograph the bright Vshaped filament of a clear light bulb through it. The image was ragged because the fibers were not perfectly aligned, but it was nonetheless a first. The excited young Lamm hurried to apply for a patent in 1930, only to find that Hansell already had one. Lamm had to settle for publishing a paper in a German journal, where he pleaded for an optical company with more resources to carry on from where he had left off.
His timing could not have been worse, for the world had just tumbled into the Great Depression. Then the rise of the Nazis forced Lamm, a Jew, to flee to America, where he settled in southern Texas as a surgeon. Meanwhile, Hansell’s electronic inventions multiplied—he eventually received more than 300 patents—but he never returned to fiber optics.
In the aftermath of World War II, three Europeans independently reinvented fiber-optic imaging. The first was Holger Møller Hansen, a Danish inventor trying to build a flexible gastroscope on a shoestring budget in his home laboratory. He tried using bare glass fibers and soon noticed a problem: Light could leak between them where they touched. He realized that covering the fibers with another material could overcome that difficulty. Total internal reflection occurs naturally in bare fibers because they are surrounded by air, which has a lower refractive index than glass. Cladding the fibers with some other material that met the same criterion but kept light from leaking would also cause total internal reflection.
With few resources to develop clad fibers, Møller Hansen did the best he could by testing oils and margarine as coating materials. In 1951 he announced his invention to reporters and applied for a Danish patent. His story went out on the Reuters wire, but the Danish patent office rejected his application after discovering HanselPs patent. Unable to find investors for further development, Møller Hansen moved on to other projects.
Abraham van Heel, an eminent professor of optics at the Technical University of Delft in Holland, was seeking a new type of periscope to help rebuild the Dutch submarine fleet, which had been shattered by World War II. After briefly considering arrays of reflective metal tubes, van Heel tried transparent fibers with metallic coatings. They didn’t transmit much light, because the coatings absorbed a small amount with each reflection. Van Heel asked for help from the Defense Ministry, which in 1951 referred him to Brian O’Brien, head of the Institute of Optics at the University of Rochester in upstate New York.
O’Brien suggested a transparent cladding. He realized that the refractive index needed only to be a little lower than that of glass, allowing the use of plastic or some other solid. Van Heel returned home and experimented with glass and plastic fibers coated with beeswax and plastic. However, progress was slow. Van Heel was busy with many other projects and had no idea that an English physicist had also thought of fiber-optic imaging.
At a 1951 dinner party in London, Harold H. Hopkins, an optics specialist at the Imperial College of Science and Technology, had encountered a physician who had just performed, in Hopkins’s words, “a particularly distressing endoscopy using the old rigid type of gastroscope.” The doctor asked Hopkins if he could make a flexible instrument to examine the stomach. Some pondering led Hopkins to realize that a bundle of flexible glass fibers could do the job. With a small research grant he bought some glass fibers and hired an Indian-born graduate assistant, Narinder Kapany.
They didn’t try cladding the fibers, but one day in early 1953 Hopkins showed one of his fiber bundles to a visiting Dutch physicist, Fritz Zernicke, who later that year would be awarded the Nobel Prize. Zernicke quickly told van Heel about the English researchers’ work. News of the competition lit a fire under van Heel, and he rushed a long report to a Dutch journal and a brief letter to the British journal Nature . He also wrote O’Brien but heard nothing in response. O’Brien was notorious for ignoring paperwork and was overwhelmed with his new job starting up a research laboratory for the American Optical Corporation. In addition, his wife of 30 years was dying. So the letter from van Heel, accompanied by one from Hopkins and Kapany, in the January 2, 1954, issue of Nature came as a great surprise. O’Brien was furious that van Heel had mentioned cladding without crediting him with inventing it, but he hoped he could still salvage a patent from the debacle.
The two Nature letters put the idea of fiber-optic image transmission in the public eye. They also caught the eye of Basil Hirschowitz, a young physician from South Africa who was at the University of Michigan on a fellowship. The poor light transmission of Hopkins and Kapany’s bundle disappointed Hirschowitz, but, he says, “it was flexible, and did transmit an image, and that was enough to set one dreaming.” Neither Hopkins nor van Heel found support for further development, but Hirschowitz persuaded his department to invest a little money and enlisted the help of Prof. Wilbur Peters of Michigan’s physics department.
Hirschowitz and Peters tried to lure Kapany to Michigan, but when he took a better job at the University of Rochester, Peters hired an undergraduate physics student named Larry Curtiss as a laboratory assistant. Curtiss, a sophomore, heated glass rods to their softening points, pulled them into fibers, and wound the fibers onto round cartons of Mother’s Oats. (Peters collected the oatmeal and took it home for family breakfasts.) However, they managed to see only weak images that faded quickly. Light leaked out at places where the uncoated fibers touched, at scratches they made by rubbing against each other, and at every fingerprint, because fingerprint oils had the same refractive index as their glass. Coating the glass fibers with instrument lacquer helped some, but not enough.
In the summer of 1956 Curtiss suggested a novel cladding method. He would slip a rod of glass inside a tube of a different glass with a lower index of refraction, melt the two together, and draw a fiber from the composite rod. An informal committee of physics faculty members told him it wouldn’t work, because the glass would crack or the coating would be nonuniform. They suggested using plastic or lacquer instead. When the professors went off to a conference in December, however, Curtiss tried his glass-cladding idea anyway, walking down a long hallway to stretch a thin strand of glass. “I was forty feet down the hall, and I could still see the glow of the furnace. I knew it was good,” Curtiss recalls. He had achieved a critical breakthrough, the best fibers yet made.
The project quickly went into high gear. Curtiss, Hirschowitz, and Peters drew 25 miles of fiber in two months and assembled it into a yard-long gastroscope five-sixteenths of an inch thick. It contained 40,000 fibers and had a light on the end to illuminate the stomach as well as optics to collect and focus the light. In February Hirschowitz tested the bundle on himself and on a woman with an ulcer. His supervisor, who had wanted to use it first, then confiscated the instrument, so Curtiss had to build a second one. Hirschowitz showed it off at a May meeting of the American Gastroscopic Society by using it to read the fine print in a telephone directory. The whole project cost a mere $5,500 over two years, nearly $4,000 of which went to pay Curtiss.
Fiber optics was taking off, and this time nothing would stop it. Physicians found the thin, flexible fiber-optic endoscope easy to use, and within a few years they were buying thousands of them. Hirschowitz, Curtiss, and Peters filed for patents and licensed the technology to American Cytoscope Makers Inc., which hired Curtiss, still an undergraduate, to help its engineers manufacture fibers.
American Optical tried to get a share of the proceeds, based on O’Brien’s patent, until a sharp-eyed American Cytoscope lawyer showed that it was invalid. The reason was van Heel’s Dutch paper. O’Brien had seen only a reprint, with the date 12/6/53 written on its cover. American patent law requires applications to be filed no more than a year after the idea is first disclosed in a publication, and American Optical lawyers had thought they were safe when they filed in November 1954. They didn’t remember that Europeans write the day of the month first, so 12/6/53 meant June 12, 1953.
As the 1950s went on, American Optical developed other fiberoptic instruments, some of which were demonstrated on Dave Garroway’s “Today” show. Around the same time, another development hit the headlines that would vastly expand the potential of fiber optics beyond niche applications of transmitting images a few feet. That development was the laser.
Lazers excited engineers because they generated waves—ones that marched along in phase with one another, like the waves from a radio oscillator. This coherence offered the possibility of using light waves efficiently for digital communications, with a series of pulses corresponding to 0’s and 1’s. Ordinary light can be used in this fashion, as with naval signals that blink on and off every few seconds. But for the much higher speeds needed for data transmissions, white light, with its chaotic assortment of wavelengths and phases, was far too messy to give a clean signal. Lasers, by providing a single sharp wavelength, eliminate the problems associated with white light, and a laser beam could be rapidly switched on and off electronically.
Radio and telephone engineers already knew quite a bit about using radiation to transmit information. One critical lesson was that the higher the frequency, the more information a signal could carry. By the 1950s microwave frequencies measured in gigahertz (billions of cycles per second) were carrying long-distance signals in 50-mile hops between radio towers. But the steady growth in traffic was clogging microwave channels, so engineers were already developing systems that would transmit at tens of gigahertz—called millimeter waves because their wavelengths are measured in millimeters.
Since light waves have frequencies thousands of times higher than millimeter waves, they promised tremendous improvements in transmission capacity. The main question in the 1960s was what medium could best transmit a laser signal. While radio waves and microwaves could be sent through the open air, through air-filled pipes called waveguides, or over wires, it soon became clear that light waves would require completely different equipment.
At first engineers didn’t seriously consider optical fibers. Curtiss’s glass-clad bundles were very effective at allowing a physician to see the inside of a patient’s stomach through a few feet of fiber. Yet that was a static display. Could the same device transmit a rapidly varying laser signal over many miles with the necessary precision? The biggest problem was the imperfect transparency of glass. Only 80 percent of the light that entered a yard-long fiber emerged from the other end; if that light passed through a second yard, only 64 percent would emerge, and if the fiber was a hundred yards long, only one ten-billionth of the light would be transmitted. Look out the window, and the air seems much clearer.
Engineers at Bell Telephone Laboratories, then the research arm of AT&T, were quick to experiment. They hauled an early laser up to the top of a microwave tower and fired red pulses toward another lab 25 miles away, but the pulses were rarely visible. In later experiments a steady beam from a laser spread over an area as wide as a dining-room table after passing through 1.6 miles of air. Even worse, the experiments worked only on clear days. Thick New Jersey fogs would stop the beam cold, as would rain, snow, sleet, and haze. Tests abroad were equally discouraging. By 1964 one British engineer had bluntly concluded, “The atmosphere is completely inimical to laser transmission systems.”
That didn’t stop Bell Labs. Its next step was to try waveguides. A waveguide is essentially a metal pipe filled with air or gas whose dimensions are carefully chosen according to the characteristics of the radiation being transmitted. It confines waves to its interior and gives them considerable protection from outside influences. Microwave waveguides are easy to make, work well, and can guide radiation around corners. Millimeter waveguides have much tighter tolerances and can be curved only gradually, with half a mile needed to make a 90-degree turn. Nonetheless, waveguides were needed to isolate millimeter waves from bad weather, so Bell Labs had developed them. Now it would try to build them for visible light as well.
Researchers experimented with a series of lenses spaced a few hundred feet apart to focus light from one lens to the next. To avoid the reflections and transmission losses associated with glass, they invented a gas lens, which focused light by using heat to create a density gradient in a tube of gas. Progress was slow, and the technology was expensive, but AT&T had money to burn on developing high-speed transmission systems.
Standard Telecommunication Laboratories (STL), in Harlow, England, did not, even though it was a subsidiary of the giant International Telephone and Telegraph Corporation. Its resident visionary, Alec Reeves, was optimistic about the future of optical communications and convinced that the demand for transmission capacity would grow. People in the industry listened to Reeves, for in 1937 he had invented a cornerstone of modern communications, a method of transmitting analog signals, such as sound, in digital form.
Reeves and STL’s management were not impressed by optical waveguides. They were very expensive, and even if they could be made to work in the field (which was doubtful), they would be attractive mainly for long-distance transmission across wide-open spaces, which were scarce in Britain. Not wanting to abandon laser communications completely, STL’s engineers went looking for other approaches. The only plausible choice, they concluded, was optical fibers.
Antoni Karbowiak, an engineer at STL, had an idea for solving the problem of transmission losses. On the basis of experience and theory, he knew that if he made his fibers very thin and left them uncoated, light waves would travel along their outer surfaces instead of through the glass itself. Unfortunately, calculations showed that such fibers would have to be about 0.0002 millimeter thick—too small to see, too fine to handle, and too narrow to collect much light. Unwilling to give up on his idea, Karbowiak decided instead to try a thin plastic film or ribbon, which would retain the advantages of his ultra-thin fibers but be easier to handle. He turned its testing over to two young STL engineers, Charles Kao and George Hockham.
Then the University of New South Wales offered Karbowiak a job he couldn’t refuse, the chairmanship of its electrical engineering department, and Kao found himself in charge of the optical communications project. An experiment showed that Karbowiak’s thin film leaked prodigiously, so Kao and Hockham turned to clad fibers, which Kao considered more promising because their light-guiding cores could be made 10 to 20 times as large as Karbowiak’s bare fibers.
At the, start of 1965 fiber optics was the darkest of dark horses in the telecommunications technology race. Industry leaders were gearing up for Picturephone, the video telephone that AT&T had demonstrated at the 1964 New York World’s Fair, which they expected to be widely adopted. They thought only millimeter waveguides could handle its enormous transmission needs along with the rest of the expanding global communications network. Kao would soon change all that.
His initial focus was on the British Post Office’s need for high-capacity cable to run a few miles between telephone switching offices. The use of fibers would make the cable flexible enough to be pulled through existing underground ducts without digging up the streets, an expense that would have been unavoidable with waveguides. The tough part, as always, was getting enough light through the fibers. The target Kao set was to make glass clear enough that 80 percent of the input light would remain after 50 meters, instead of after 1 meter.
Kao recalls, “I was seeking the answer to the question, ‘What are the loss mechanisms, and can these mechanisms be totally removed?’ It appeared that no one had really asked this question before.” Was there some unavoidable minimum amount of loss? Glass chemists blamed most absorption on impurities, but they didn’t know how pure glass could be made or how much residual absorption would remain. Since glass had rarely been used in thicknesses of more than a few inches, absorption had never been much of a problem before. Finally an expert from England’s Sheffield Institute of Glass Technology told Kao that stringent purification should make it possible to reduce absorption to his target level. Armed with this opinion, on January 27, 1966, Kao told a London meeting that short fiber systems could carry the equivalent of 200,000 telephone channels.
“I was trying to sell a dream,” says Kao, looking back. The tiny, flexible fiber tubes he envisioned were elegant compared with the brute-force technology of firehose-size millimeter waveguides, which had to be laid as straight as high-speed rail tracks and could be impaired by thermal effects or soil subsidence. Yet Kao knew that even more formidable problems remained for the fiber-optic solution. “We were talking about a system concept which required a light source which at that time was working intermittently in liquid nitrogen, and an order-of-magnitude improvement in fiber that was so far out that people could not believe it was an attainable goal.” It was a risky but attractive gamble to an ambitious young engineer.
The British Post Office decided to place a small bet on fiber as an alternative to what one top engineer called the “pipe dream” of millimeter waveguides. The project manager, F. F. Roberts, put his staff to work studying glass and tried to interest outside companies in developing clearer materials. One British glass firm snorted that fiber optics would amount to only “a few buckets of sand,” but a visiting scientist from the Corning Glass Works was intrigued and took the idea home to the company’s laboratories in Corning, New York. Alec Reeves persuaded STL to fund more fiber research while Roberts and Kao tried to arouse interest in other labs in Europe, America, and Japan. The British Post Office program received a boost when American consultants persuaded its top management to raise the annual research budget by £12 million, or about $29 million.
Yet Bell Labs, considered the world’s top industrial research laboratory, remained doubtful that glass could ever be made clear enough for communications. Kao responded to their skepticism by measuring light loss in fused silica, a synthetic glass that is essentially pure silicon dioxide with less than one part per million of light-absorbing impurities. (Common glass is mostly silicon dioxide but contains other ingredients, usually salts of calcium and sodium.) After months of intense analysis, Kao came up with an impressive result: The glass was so clear that 80 percent remained after 200 meters. “Kao gave everybody a jolt,” recalls David Pearson of Bell Labs. “That was the first practical measurement which said, ‘Hey, you’re not just whistling “Dixie.”’”
The next step was to make fibers from ultrapure glass. Pearson’s group and most others around the world were still trying to purify conventional glasses. But Robert Maurer, the manager of a small research group at Corning, sought instead to draw fibers from fused silica, although he knew the odds were against success. Fused silica melts at 2,900 degrees Fahrenheit, a higher temperature than can be achieved with a standard laboratory furnace. It also has a very low refractive index, so finding a cladding material with an even lower one is difficult. But Maurer was a contrarian by nature, and his company had experience with fused silica. Coming’s laboratory also contained one of the few furnaces hot enough to melt it.
After some encouraging initial tests, Maurer hired Donald Keck, who had just earned his Ph.D. in physics at Michigan State University, to work on fibers full-time. Maurer also enlisted Peter Schultz; he had joined Coming’s glass chemistry group a few months earlier, after receiving a doctorate from Rutgers University. The two spent long months learning new ways to make fibers. They burned silicon chloride in oxygen to deposit a fine silicon dioxide soot inside a quarter-inch tube. Unfortunately, the soot tended to collect at the front end. Then the pair spotted an old canister vacuum cleaner that Schultz kept on hand to handle the mess that was inevitable in a glass lab. Its suction was just what they needed to pull the silicon dioxide into the tube. The vacuum cleaner didn’t survive the corrosive burner exhaust, but it showed how to deposit glass all along the tube’s inside. Adding titanium to the inside layer raised the refractive index to create a light-guiding core for the fiber they drew by stretching out the tube.
Keck recalls that “it was hotter than Hades” drawing fibers during the steamy summer of 1969. Still, step by step he and Schultz refined their techniques. In early 1970 he pulled about 60 feet of very clear fiber, but the glass rod slipped and was ruined before he could draw more. He and Schultz made another glass rod using the same process and managed to draw more than 300 feet of fiber. The final processing stage finished on a Friday afternoon, and Keck decided to test part of it immediately because he worried about the fiber’s fragility. He pointed a red laser beam down the fiber and was jolted by the results. “When the laser spot hit the core, all of a sudden I got this flash of light,” he recalls, ”… a spot that was different than the laser spot.”
It took him a few minutes to figure out that the light had gone back and forth through some 660 feet of fiber. The distant end of the fiber had reflected 4 percent of the light back toward Keck, and that light was still visible when it reached his eye. Keck realized that he had before him the clearest glass ever made. Exhilarated, he went searching for someone to share his eureka moment with, but there wasn’t a soul left in the lab.
Careful measurements convinced Corning that they had made fibers clearer than Kao’s goal. After preparing patent applications, Maurer, Keck, and Schultz cautions reported their results at a London conference in September 1970 and were amazed to discover that no one else had come close to making such clear fibers. Few people even realized the significance of Corning’s achievement. Save for a brief mention in the British weekly New Scientist , the press ignored it. However, Bell Labs, fearful of being left behind, belatedly started a big fiber optics push.
Corning expanded its effort as well, because its first clear fiber was far too fragile to be practical. The key to increasing its flexibility lay in substituting germanium for titanium in the lightguiding core. By June 1972 Schultz and Keck had made fibers five times as clear, with 80 percent of the input light emerging after 250 meters. “This was to my mind the breakthrough for fiber optics, because it was now a practical fiber,” says Schultz. “You could make fibers right off the draw with these nice low losses.”
It took a few more years to assemble the rest of the equipment needed to make fiber-optic communications systems. In January 1976 Bell Labs engineers began a closely watched fiber-optic field trial beneath a company parking lot in suburban Atlanta. Except for some problems with the laser transmitters, the trial went flawlessly. AT&T finished field trials of its millimeter waveguide system at about the same time and declared them successful as well. But the fat metal pipes went on the shelf while AT&T prepared for a 1977 trial of optical fiber carrying live telephone traffic in Chicago.
GTE, Standard Telecommunication Labs, and the British Post Office also sent live telephone traffic through optical fibers in 1977. The trial systems all passed their tests with flying colors. Major telephone carriers around the world geared up to install fiber-optic links running several miles between local switching centers in urban and suburban areas. The systems they designed were just the sort that Kao had envisioned.
Over the ensuing two decades plus, researchers have continued to make impressive progress. Japanese engineers discovered that transmission losses could be reduced even further by shifting to longer wavelengths, and by 1978 they had made fiber that transmitted 80 percent of input light over a distance of three miles. They also developed new fibers and transmitters that could send signals at much higher speeds. AT&T’s experimental Chicago system had transmitted 45 million bits per second, the equivalent of 672 voice telephone lines, over a few miles. The new systems, in 1982, could carry 6,048 voice channels, or 400 million bits per second, up to 30 miles.
The breakup of AT&T and the spread of long-distance competition opened the way to new fiber technology. By 1983 the long-distance carrier MCI was installing 400-million-bit fiber-optic systems as the basis for its national long-distance telephone network. In 1988 the first transatlantic fiber-optic cable, TAT-8, began carrying telephone signals from America to Europe.
The technology has hardly stopped moving since. Today’s fiber-optic transmitters can send signals at 10 billion bits per second, the equivalent of 129,000 telephone conversations. Moreover, engineers have devised ways for fibers to carry signals at hundreds of different wavelengths, so that a single hair-thin fiber can carry more than a trillion bits per second. Continuous advances are helping the fiber-optic backbone of today’s telecommunications network keep up with the soaring growth of Internet traffic. After more than 30 years, says Donald Keck, who now heads optics and photonics research at Corning, “It’s not over.”