How Charles Kao Bet the Future on Pure Glass Fiber

Glass Goes From Bolt to Cable (0:00)

Hey, chibis! I'm Aiwee, and today we're talking about how glass went from crossbow projectile to ocean-spanning fiber. If you enjoy stories like this, hit the like button and subscribe if you haven't already — let's go! By nineteen eighty-eight, the first transatlantic fiber optic cable had been installed. It carried forty thousand telephone conversations at once, ten times the capacity of the copper cable it replaced.

The thread inside it was glass, the same material that makes up a windowpane, and yet pure enough that a pulse of laser light could travel across an ocean and still arrive strong enough to read. One hundred and one years earlier, glass had been demonstrated by firing it from a crossbow. Between those two moments, something extraordinary happened to the transparency of glass. The fibers would need to improve by a factor beyond imagination, a one followed by ninety-eight zeroes. The question is how, and why it took so long, and who believed it was possible when almost no one else did.

To understand why anyone cared about glass, you have to understand what copper could and could not do. Between nineteen sixty and nineteen seventy-three, the number of daily telephone calls in the United States roughly doubled, from two hundred and eighty-five million to five hundred and seventy-six million.

Copper Hits Its Bandwidth Wall (1:30)

In that same stretch, NASA spent about twenty-six billion dollars to put astronauts on the moon. AT&T, by one account, spent nearly three times that much just to upgrade the telephone network. The limits were physical. A copper wire carries an oscillating current, and a faster oscillation means more bandwidth, more conversations packed into the same line. Engineers had a clever trick.

They shifted each call to its own slice of the radio spectrum, then filtered them apart at the other end. It worked, up to a point. As frequencies climbed past sixty-five megahertz, signals in copper died in less than two kilometers. The metal itself turned against the signal, and there was no more headroom. Then light arrived as an option.

Visible light oscillates hundreds of trillions of times per second, hundreds of terahertz, orders of magnitude beyond anything copper could carry. The only problem was generating it. In nineteen sixty, Theodore Maiman built the first working laser. Suddenly engineers had a coherent, focused beam, and the possibility of using light for communication reopened. The bandwidth question was answered in principle.

The remaining question was the medium. Could a glass fiber carry that light across meaningful distances?

Fitting Calls Into Radio Slices (3:00)

In nineteen sixty-one, a Bell Labs team set out to find out, and what they found was discouraging. The best fibers of the early nineteen sixties lost about a thousand decibels per kilometer. Decibels are a logarithmic measure, and that number is catastrophic. To put it in scale, a loss of twenty decibels per kilometer is comparable to a truck starting with one hundred kilograms of sand and arriving with one kilogram, a tough but workable ninety-nine percent loss. A loss of a thousand decibels per kilometer is closer to starting with the mass of the observable universe and arriving with less than the mass of a single atom.

The Bell Labs team concluded the technology was hopeless, and the project was shelved. Fiber optics might have ended there. It did not, because of one man. Charles Kao started at an English telecoms laboratory in nineteen sixty. He was an electrical engineer, not a glass expert, and he kept returning to a single question.

Why was glass so lossy? Was the loss an intrinsic property of the material, in which case the whole idea was dead, or was it coming from impurities that could be removed? He consulted material scientists and reviewed the literature. Most of the absorption, he learned, was not from the silica itself, which has a band gap around nine electron volts, far above the energy of visible light.

The Laser and the Medium Problem (4:30)

The culprit was trace contamination, especially iron ions, which absorb strongly in the violet, blue, and infrared regions of the spectrum. Remove those, and the loss should fall dramatically. In nineteen sixty-six, Kao and his colleagues published a paper estimating that with sufficient purity, glass could reach about twenty decibels per kilometer, comparable to copper. They described the material problem as difficult but not impossible. Few in the field took the claim seriously.

A one followed by ninety-eight zeroes, a number with no measurement to back it up. Kao knew he needed proof. The purest glass anyone had made was fused silica, pure silicon dioxide melted above two thousand degrees Celsius, but it was expensive and came only in small samples. He needed to measure its loss. Normally, you would measure fiber loss over hundreds of meters to get a clean signal.

Kao had pieces under thirty centimeters, so he and his team built a new instrument sensitive enough to detect the tiny amount of light that made it through.

Charles Kao Asks Why (5:45)

After three years of refinement, in nineteen sixty-nine, he reported a loss of less than five decibels per kilometer, with a margin of error of plus or minus four. The result was the first concrete evidence that high-purity glass could carry a light signal across meaningful distances. The number was small enough to silence doubt, and big enough to launch a worldwide race. At Corning Glass Works, a scientist named Robert Maurer had already been thinking about fiber. He had written a key paper on Rayleigh scattering, the way density fluctuations in glass deflect light, and Kao had used those numbers to estimate a theoretical floor of about one decibel per kilometer from scattering alone.

Corning had a second advantage, a process called flame hydrolysis, developed in the nineteen thirties, which could produce ultra-pure fused silica by burning silicon tetrachloride in an oxygen flame. They had the raw material. The catch was refractive index. Fused silica has an index of about one point four six, the lowest of any common glass, and for light to be trapped by total internal reflection, the core must have a higher index than the cladding. Most labs wanted a pure core.

Maurer flipped the design, putting fused silica in the cladding and adding a small amount of dopant to raise the core's index.

The 1966 Impurity Paper (7:15)

The first dopant was titanium, which can sit in the silica structure without disrupting it. Titanium, however, broke down at the high drawing temperatures and absorbed too much light. Two Corning colleagues, Donald Keck and Peter Schultz, joined Maurer, and together they tried a different approach. They built the preform from the outside in, depositing doped silica soot onto the inside of a pure fused silica tube. To spread the soot evenly, a technician held a vacuum cleaner to the back of the tube.

The vacuum did not survive the experiment, but the technique did. On July twenty-two, nineteen seventy, Keck measured a short fiber at seventeen decibels per kilometer. The next month, on August twenty-one, a longer sample measured sixteen point nine decibels per kilometer. The team had crossed the threshold. The dopant was soon switched to germanium, which held up in the heat, and the first germanium-doped fiber reached four decibels per kilometer, the level Keck considers the first truly practical low-loss fiber.

By nineteen seventy-seven, AT&T ran a trial system under the streets of Chicago. Two years later, a Japanese team reported zero point two decibels per kilometer, a figure that has remained a working standard for silica fibers. With the medium solved, encoding took over. Voice waves were sampled into discrete values and turned into binary ones and zeros.

Measuring Loss in a Small Sample (8:45)

A laser flashed on and off, around two hundred and eighty million times a second. Multiple conversations were interleaved, one bit from each in turn, then unzipped at the other end. The nineteen eighty-eight transatlantic cable carried forty thousand simultaneous calls, and modern global fiber now runs to more than eight billion kilometers installed. The numbers are no longer exotic. They are the ordinary plumbing of the modern internet.

Two honest caveats matter. First, Rayleigh scattering is a fundamental physical limit. It cannot be engineered away, only minimized. The zero point two decibel figure is close to the floor for silica, which is why long-haul systems rely on amplifiers to keep signals alive across oceans. Second, the analogy between the mass of the observable universe and the mass of a single atom spans an almost incomprehensible range.

It is a useful way to feel the scale of the improvement, not a literal engineering specification. The story also shows a recurring pattern. Kao, an outsider to materials science, asked the question the field had avoided. Maurer, working against convention, made the cladding from the better material and added impurity to the core. Outsider perspectives and counterintuitive moves did not just help.

They were the breakthrough.

Corning's Pure Fused Silica (10:15)

Charles Kao received the Nobel Prize in two thousand nine. The internet, as it exists today, would not be possible without the glass he believed in. If you want to know how light signals are kept alive across the seven thousand kilometers of an ocean, that is another story, about a different invention. Tell us in the comments if you would like to see it.