By Jeff Hecht
This short history of fiber optics was originally published in Fiber Optics Technician's Handbook, by Jim Hayes, Delmar Publishers, Albany, New York. City of Light: The Story of Fiber Optics, tells the story in much greater depth.
Optical communication systems date back two centuries, to the "optical telegraph" that French engineer Claude Chappe invented in the 1790s. His system was a series of semaphores mounted on towers, where human operators relayed messages from one tower to the next. It beat hand-carried messages hands down, but by the mid-19th century was replaced by the electric telegraph, leaving a scattering of "Telegraph Hills" as its most visible legacy.
Alexander Graham Bell patented an optical telephone system, which he called the Photophone, in 1880, but his earlier invention, the telephone, proved far more practical. He dreamed of sending signals through the air, but the atmosphere didn't transmit light as reliably as wires carried electricity. In the decades that followed, light was used for a few special applications, such as signaling between ships, but otherwise optical communications, like the experimental Photophone Bell donated to the Smithsonian Institution, languished on the shelf.
In the intervening years, a new technology slowly took root that would ultimately solve the problem of optical transmission, although it was a long time before it was adapted for communications. It depended on the phenomenon of total internal reflection, which can confine light in a material surrounded by other materials with lower refractive index, such as glass in air. In the 1840s, Swiss physicist Daniel Collodon and French physicist Jacques Babinet showed that light could be guided along jets of water for fountain displays. British physicist John Tyndall popularized light guiding in a demonstration he first used in 1854, guiding light in a jet of water flowing from a tank. By the turn of the century, inventors realized that bent quartz rods could carry light, and patented them as dental illuminators. By the 1940s, many doctors used illuminated plexiglass tongue depressors.
Optical fibers went a step further. They are essentially transparent rods of glass or plastic stretched so they are long and flexible. During the 1920s, John Logie Baird in England and Clarence W. Hansell in the United States patented the idea of using arrays of hollow pipes or transparent rods to transmit images for television or facsimile systems. However, the first person known to have demonstrated image transmission through a bundle of optical fibers was Heinrich Lamm, than a medical student in Munich. His goal was to look inside inaccessible parts of the body, and in a 1930 paper he reported transmitting the image of a light bulb filament through a short bundle. However, the unclad fibers transmitted images poorly, and the rise of the Nazis forced Lamm, a Jew, to move to America and abandon his dreams of becoming a professor of medicine.
In 1951, Holger Møller [or Moeller, the o has a slash through it] Hansen applied for a Danish patent on fiber-optic imaging. However, the Danish patent office denied his application, citing the Baird and Hansell patents, and Møller Hansen was unable to interest companies in his invention. Nothing more was reported on fiber bundles until 1954, when Abraham van Heel of the Technical University of Delft in Holland and Harold. H. Hopkins and Narinder Kapany of Imperial College in London separately announced imaging bundles in the prestigious British journal Nature.
Neither van Heel nor Hopkins and Kapany made bundles that could carry light far, but their reports the fiber optics revolution. The crucial innovation was made by van Heel, stimulated by a conversation with the American optical physicist Brian O'Brien. All earlier fibers were "bare," with total internal reflection at a glass-air interface. van Heel covered a bare fiber or glass or plastic with a transparent cladding of lower refractive index. This protected the total-reflection surface from contamination, and greatly reduced crosstalk between fibers. The next key step was development of glass-clad fibers, by Lawrence Curtiss, then an undergraduate at the University of Michigan working part-time on a project to develop an endoscope to examine the inside of the stomach with physician Basil Hirschowitz, physicist C. Wilbur Peters. (Will Hicks, then working at the American Optical Co., made glass-clad fibers at about the same time, but his group lost a bitterly contested patent battle.) By 1960, glass-clad fibers had attenuation of about one decibel per meter, fine for medical imaging, but much too high for communications.
Meanwhile, telecommunications engineers were seeking more transmission bandwidth. Radio and microwave frequencies were in heavy use, so they looked to higher frequencies to carry loads they expected to continue increasing with the growth of television and telephone traffic. Telephone companies thought video telephones lurked just around the corner, and would escalate bandwidth demands even further. The cutting edge of communications research were millimeter-wave systems, in which hollow pipes served as waveguides to circumvent poor atmospheric transmission at tens of gigahertz, where wavelengths were in the millimeter range.
Even higher optical frequencies seemed a logical next step in 1958 to Alec Reeves, the forward-looking engineer at Britain's Standard Telecommunications Laboratories who invented digital pulse-code modulation before World War II. Other people climbed on the optical communications bandwagon when the laser was invented in 1960. The July 22, 1960 issue of Electronics magazine introduced its report on Theodore Maiman's demonstration of the first laser by saying "Usable communications channels in the electromagnetic spectrum may be extended by development of an experimental optical-frequency amplifier."
Serious work on optical communications had to wait for the continuouswave helium-neon laser. While air is far more transparent at optical wavelengths than to millimeter waves, researchers soon found that rain, haze, clouds, and atmospheric turbulence limited the reliability of long-distance atmospheric laser links. By 1965, it was clear that major technical barriers remained for both millimeter-wave and laser telecommunications. Millimeter waveguides had low loss, although only if they were kept precisely straight; developers thought the biggest problem was the lack of adequate repeaters. Optical waveguides were proving to be a problem. Stewart Miller's group at Bell Telephone Laboratories was working on a system of gas lenses to focus laser beams along hollow waveguides for long-distance telecommunications. However, most of the telecommunications industry thought the future belonged to millimeter waveguides.
Optical fibers had attracted some attention because they were analogous in theory to plastic dielectric waveguides used in certain microwave applications. In 1961, Elias Snitzer at American Optical, working with Hicks at Mosaic Fabrications (now Galileo Electro-Optics), demonstrated the similarity by drawing fibers with cores so small they carried light in only one waveguide mode. However virtually everyone considered fibers too lossy for communications; attenuation of a decibel per meter was fine for looking inside the body, but communications operated over much longer distances, and required loss no more than 10 or 20 decibels per kilometer.
One small group did not dismiss fibers so easily -- a team at Standard Telecommunications Laboratories initially headed by Antoni E. Karbowiak, which worked under Reeves to study optical waveguides for communications. Karbowiak soon was joined by a young engineer born in Shanghai, Charles K. Kao.
Kao took a long, hard look at fiber attenuation. He collected samples from fiber makers, and carefully investigated the properties of bulk glasses. His research convinced him that the high losses of early fibers were due to impurities, not to silica glass itself. In the midst of this research, in December 1964, Karbowiak left STL to become chair of electrical engineering at the University of New South Wales in Australia, and Kao succeeded him as manager of optical communications research. With George Hockham, another young STL engineer who specialized in antenna theory, Kao worked out a proposal for long-distance communications over single-mode fibers. Convinced that fiber loss should be reducible below 20 decibels per kilometer, they presented a paper at a London meeting of the Institution of Electrical Engineers. The April 1, 1966 issue of Laser Focus noted Kao's proposal:
"At the IEE meeting in London last month, Dr. C. K. Kao observed that short-distance runs have shown that the experimental optical waveguide developed by Standard Telecommunications Laboratories has an information-carrying capacity ... of one gigacycle, or equivalent to about 200 tv channels or more than 200,000 telephone channels. He described STL's device as consisting of a glass core about three or four microns in diameter, clad with a coaxial layer of another glass having a refractive index about one percent smaller than that of the core. Total diameter of the waveguide is between 300 and 400 microns. Surface optical waves are propagated along the interface between the two types of glass."
"According to Dr. Kao, the fiber is relatively strong and can be easily supported. Also, the guidance surface is protected from external influences. ... the waveguide has a mechanical bending radius low enough to make the fiber almost completely flexible. Despite the fact that the best readily available low-loss material has a loss of about 1000 dB/km, STL believes that materials having losses of only tens of decibels per kilometer will eventually be developed."
Kao and Hockham's detailed analysis was published in the July 1966 Proceedings of the Institution of Electrical Engineers. Their daring forecast that fiber loss could be reduced below 20 dB/km attracted the interest of the British Post Office, which then operated the British telephone network. F. F. Roberts, an engineering manager at the Post Office Research Laboratory (then at Dollis Hill in London), saw the possibilities, and persuaded others at the Post Office. His boss, Jack Tillman, tapped a new research fund of 12 million pounds to study ways to decrease fiber loss.
With Kao almost evangelically promoting the prospects of fiber communications, and the Post Office interested in applications, laboratories around the world began trying to reduce fiber loss. It took four years to reach Kao's goal of 20 dB/km, and the route to success proved different than many had expected. Most groups tried to purify the compound glasses used for standard optics, which are easy to melt and draw into fibers. At the Corning Glass Works (now Corning Inc.), Robert Maurer, Donald Keck and Peter Schultz started with fused silica, a material that can be made extremely pure, but has a high melting point and a low refractive index. They made cylindrical performs by depositing purified materials from the vapor phase, adding carefully controlled levels of dopants to make the refractive index of the core slightly higher than that of the cladding, without raising attenuation dramatically. In September 1970, they announced they had made single-mode fibers with attenuation at the 633-nanometer helium-neon line below 20 dB/km. The fibers were fragile, but tests at the new British Post Office Research Laboratories facility in Martlesham Heath confirmed the low loss.
The Corning breakthrough was among the most dramatic of many developments that
opened the door to fiber-optic communications. In the same year, separate teams
at the Ioffe Physical Institute in Leningrad (now St. Petersburg) and at Bell
Labs independently made the first semiconductor diode lasers able to emit continuouswave
at room temperature. Over the next several years, fiber losses dropped dramatically,
aided both by improved fabrication methods and by the shift to longer wavelengths
where fibers have inherently lower attenuation.
Early single-mode fibers had cores several micrometers in diameter, and in
the early 1970s that bothered developers. They doubted it would be possible
to achieve the micrometer-scale tolerances needed to couple light efficiently
into the tiny cores from light sources, or in splices or connectors. Not satisfied
with the low bandwidth of step-index multimode fiber, they concentrated on multi-mode
fibers with a refractive-index gradient between core and cladding, and core
diameters of 50 or 62.5 micrometers. The first generation of telephone field
trials in 1977 used such fibers to transmit light at 850 nanometers from gallium-aluminum-arsenide
laser diodes at data rates of 6.2 and 45 megabits per second.
Those first-generation systems could transmit light several kilometers without
repeaters, but signal transmission distance was limited by loss of about 2 dB/km
in the fiber. A second generation soon appeared, using new InGaAsP lasers which
emitted at 1300 nm, where fiber attenuation was about 0.4 dB/km. Pulse dispersion
in multimode fibers also was somewhat lower at 1300 nm than at 850 nm, allowing
data rates of 90 Mbit/s over about 20 km.
Those distances and data rates were fine for the trunk lines linking telephone switching offices in nearby communities, but developers working on a new generation of submarine cables wanted higher speeds and needed to span much longer distances. The submarine cable group at Bell Labs turned to single-mode fibers to meet that requirement. Simple step-index single-mode fibers had essentially zero dispersion at 1310 nm, so repeater spacing was limited only by the fiber attenuation. In 1980, Bell Labs announced plans for the first transatlantic fiber-optic cable, TAT-8, to carry a total of 565 Mbit/s through two operating fiber pairs starting in 1988.
Multimode fibers remained in use for interoffice trunk lines for a few years, but when deregulation opened the long-distance market to competitive carriers in the early 1980s, MCI turned to single-mode fiber systems that could transmit 400 Mbit/s at 1300 nm for 50 km between repeaters. Other long-distance carriers also built their national networks with single-mode fibers, and by the time TAT-8 was turned on in December 1988, single-mode fiber systems operating at 1300 nm were standard for telecommunications.
Developers had earlier recognized that the minimum attenuation of glass fibers was below 0.2 dB/km at a longer wavelength. 1550 nm. That wavelength did not come into widespread use quickly because lasers were harder to make and because single-mode fibers had higher chromatic dispersion. However, 1550-nm transmission was used to stretch repeater spacing in submarine cables starting with TAT-10, which began service in 1992.
The next big step was the advent of the erbium-doped fiber amplifier. In 1986, Dave Payne of the University of Southampton showed that an optical fiber with erbium added to its core could amplify light at wavelength around 1550 nm. Elias Snitzer had demonstrated the first optical amplifiers in the early 1960s, but they had found no practical applications. Payne's erbium-doped fiber amplifier turned out to be very good for amplifying the weak signals transmitted in fiber-optic communication systems. Such an all-optical amplifier was much simpler than an electro-optic repeater, which had to convert an optical signal into electronic form to amplify it, then convert it back into optical form to send through another length of fiber. It also operated across a range of wavelengths where fiber attenuation was at a minimum.
It took several years for Payne's group at Southampton, Emmanuel Desurvire's group at Bell Labs and others to develop practical erbium-doped fiber amplifiers. A key problem was finding the right way to excite the erbium atoms in the fiber core so that they could amplify the weak optical signal. They also found that erbium-doped fiber amplifiers could simultaneously amplify signals at two or more wavelengths transmitted through the same fiber, a concept called wavelength-division multiplexing.
It was well-known that optical fibers could carry signals at two or more wavelengths that did not interfere with each other. However, wavelength-division multiplexing had seemed impractical because the signals could not be amplified without separating them, converting them to electronic form, and passing them through separate electronic amplifiers before converting the signals back to light and recombining them to transmit through another length of fiber. Erbium-doped fiber amplifiers simply took the input signals in the form of light and amplified them in a single stage, without converting them into electronic form or having to separate them. That made wavelength-division multiplexing practical for the first time.
Optical amplifiers multiplied the potential transmission capacity of a fiber. The first transatlantic cable to use optical amplifiers, TAT-12/13 in 1996, was able to transmit 5 gigabits per second through a single pair of fibers at one wavelength. By 1998 it was followed by other submarine systems that could transmit 2.5 Gbit/s signals at four or eight wavelengths through a single pair of fibers - a total of 10 or 20 Gbit/s.
Terrestrial systems could carry even higher bandwidths by packing the wavelengths close together, a technology called dense wavelength-division multiplexing or DWDM. The technology arrived as the Internet boom was pumping up demand for long-distance transmission capacity. Fiber promised plenty of capacity, with engineers developing systems that could transmit dozens of channels simultaneously at 2.5 or 10 Gbit/s each. By 2001 some manufacturers were offering DWDM systems able to transmit 100 wavelengths at 10 Gbit/s each, a total of one terabit per second.
By then, dot.com companies had begun collapsing. Investors briefly jumped to telecommunications companies, figuring that fiber-optic transmission lines were more substantial than companies with few assets beyond their web sites. However, the tremendous capacity of fiber optics turned out to be too much of a good thing. Telecommunications companies had installed more fiber than they needed to meet near-term demand.
Construction costs push telecommunications companies to install excess capacity. Cables can easily accommodate more fibers, and it's much cheaper to add more fibers to a cable being laid now than to go back and lay a second cable later. DWDM could multiply the capacity of each installed fiber. And it turned out that projections of Internet traffic doubling every three months were wildly exaggerated. The telecommunications bubble collapsed, causing the bankruptcies of major telecommunications companies like Worldcom and Global Crossing.
The long-distance industry is recovering slowly. But a bright new hope has emerged. Consumers are signing up in increasing numbers for broadband Internet service provided by telephone companies using digital subscriber line and by cable-television companies using cable modems. Fibers often even more transmission capacity, and are now being installed to homes in large new developments, rural areas, and some other areas. The numbers are small but growing, and big phone companies -- notably Verizon -- have begun their own fiber to the home installations. Stay tuned.
Acknowledgment: Thanks to the Alfred P. Sloan Foundation for research support. This is a much expanded version of an article originally published in the November 1994