--- author: Veritasium date: '2026-10-12' guest: '' layout: post.njk source: https://www.youtube.com/watch?v=NiCgcUTL9_Q speaker: Veritasium tags: - fiber-optics - optical-communication - telecommunications - total-internal-reflection title: 光纤通信革命：从束缚光线到构建全球互联网 summary: 文章详述了光纤通信技术的诞生与演进史。面对20世纪60年代铜线通信的物理带宽瓶颈与激光器发明后的光通信探索，高锟（Charles Kao）突破性地揭示了玻璃高损耗源于杂质吸收而非材料本征特性的物理本质。康宁公司（Corning）团队随后创新性地采用掺杂二氧化硅熔融技术攻克制造难关，将光损耗降至实用水平，彻底奠定了现代全球数字互联网的高速传输基石。 insight: '' draft: true series: '' category: tech-trends area: tech-engineering project: [] people: - 高锟 companies_orgs: - Corning - AT&T - Bell Labs products_models: [] media_books: [] status: evergreen --- ### 铜线通信的物理极限与光的无限潜力 1887年，一位年轻的物理学老师将细玻璃棒固定在自制十字弓的箭尾，用火炬将其端部加热至熔融成大头针大小的玻璃珠，随后扣动扳机射出飞箭。透过射入室内的明亮阳光，闪烁着一根极细的玻璃纤维——其最细处不足人类头发直径的十分之一。这位物理学者最初利用这些纤维来测量微小外力，但在接下来的几十年中，其他研究人员发现它们具备一种截然不同的用途：**传输光信号**。玻璃纤维即使在弯折甚至呈90度角时也能传导光线。 到了20世纪60年代，电信行业迫切寻求更高效的信息承载方式。一位美国工程师曾断言：“我们理应能够通过单根光丝，同时传输全球正在进行的所有对话。”然而当时的玻璃纤维性能极其糟糕，光信号在传输短短一公里后衰减便严重到无法解调，几乎无人相信它具备实用性。彼时电信公司对更优通信系统的渴望，甚至仅仅是为了应对爆炸式的电话业务需求：在1960年至1973年间，美国日均电话通话量从2.85亿次翻倍至5.76亿次。同期美国宇航局（NASA）耗费260亿美元将人类送上月球，而**美国电话电报公司**（AT&T）为了铺设传统电话基础设施，耗资接近NASA的三倍，却依然难以赶上需求增长。 当时电话主要通过铜线传输。其基本原理是：拿起听筒时，恒定电流流经线路为电话供电；讲话时，声音振动受话器内的微型振膜对电流进行调制，音量越大电流摆幅越大，音调高低则改变电流振荡频率；在接收端，电流驱动扬声器重新还原为声音。然而电话公司不可能为每家每户拉设专属专线，单根电缆必须复用承载多次通话。为此工程人员采用了**频分复用**（Frequency Division Multiplexing）技巧：尽管人耳可听范围高达20 kHz，但理解语音仅需300至3400 Hz频段，因此超出4 kHz的语音频带被滤除；通过调制搬移电信号频率，一通电话占用0-4 kHz，另一通占用4-8 kHz，接收端再利用带通滤波器分频并解调还原。 但在铜线上推高频率遇到了致命的电磁学物理障碍。传输信号的高频振荡电流会诱发磁场并产生**涡流**（Eddy currents）。由于其几何结构，这些电流环路会阻碍导线中心的电流变化，促使电流集中于表面流动；信号变化越快，涡流越强，最终将导线内部振荡完全抵消。当信号被挤压在导线表层的极小截面内传输时，有效电阻急剧增加，高频信号尚未传播多远即被衰减殆尽，即所谓的**趋肤效应**。当频率超过65 MHz时，高频信号在不足两公里内便彻底湮灭。早在1948年，**香农-哈特利定理**（Shannon-Hartley theorem: 描述信道在特定功率、噪声和带宽下的最大信息传输容量）便指明了提升容量的核心在于增加**信道带宽**（Bandwidth）。铜线的物理极限已至，而可见光拥有高达数百太赫兹（THz）的极高频率。正如贝尔实验室研究人员所感叹：“超高频率以及随之而来的极其宽广的频带，就像一座令人神往的黄金之国（El Dorado），时刻吸引着我们前行。”
Original English Source - In 1887, a young physics teacher attached a thin glass rod to the end of an arrow and loaded it into his homemade crossbow. He torched the rod's end until a molten glass bead. Roughly the size of a pin's head, it was barely hanging on. And then he fired the arrow. Bright sunlight streaming into the room glinted off a tiny glass fiber. At its thinnest, it was less than a tenth of the diameter of a human hair. Now legally we cannot shoot a crossbow indoors. Luckily I am pretty quick. That is cool. Look at that. Look how thin we got it. The physics teacher was using these fibers to measure very small forces, but in the decades that followed, other researchers found a different use for them: sending light signals. Glass fibers can carry light even through bends. Look at it. It's like that's 90 degrees. And during the 1960s, the telecoms industry was desperate for a better way to carry more information. As one American engineer said, "It should be possible to transmit all of the conversations going on anywhere in the world simultaneously over a single thread of light." But these glass fibers were terrible. The light faded so much that you couldn't decode a signal after just one kilometer. No one thought they'd be practical. Except for one man. His own boss called his dream "a wild goose chase," but if he was right, if he could prove that fiber optics were in fact viable, and if someone could then figure out how to make them at scale, well, eventually they could connect continents. How much is in this warehouse? - Oh, there's millions of kilometers here that we're pushing through here every month. - It's approximately three times as hot as the surface of the sun. - That's crazy. And one day, it could build the internet. All that was unimaginable in the 1960s, but even then telecom companies wanted a better communication system, just for phone calls. Between 1960 and 1973, the number of daily calls doubled in the U.S. from 285 million to 576 million. In that same time period, NASA sent people to the moon for a total cost of $26 billion. While AT&T had to spend nearly triple that amount just to install new telephone infrastructure. They just couldn't keep up with the demand. At the time, phone calls were mostly transmitted over copper wires. Now imagine I'm trying to phone my buddy Gregor. Now, when I pick up, a constant current immediately flows through the wire to supply power to the phone. When I speak, my voice vibrates a tiny diaphragm in the mouthpiece that modulates the current. The louder I am, the bigger the current swing, and my voice's pitch changes how fast the current oscillates. At Gregor's end, that current reaches a speaker that turns it back into sound. Now, this works for one call on one line, but phone companies aren't going to install wires between every single house. One cable needs to carry multiple calls. So phone companies figured out a neat little trick. Even though humans can hear frequencies up to 20 kilohertz, you really only need to hear the ones between 300 and 3400 hertz to understand what someone's saying. So phone companies drop frequencies above four kilohertz or anything outside the voice band. But if everyone used that zero to four kilohertz band on the same line, well, then all the conversations would be on top of each other and you wouldn't be able to hear the person that you're talking to. Luckily, we can shift those electrical signals, so my conversation can use zero to four kilohertz while another person's can use four to eight and so on all the way down the same copper wire. At the other end, a filter separates out the different frequency bands, similar to how a radio can tune into one station. Then the signal is shifted back down to the zero to four kilohertz range before it hits the speaker, so it sounds like a normal voice. But as phone companies started using higher and higher frequencies, they started running into problems. That oscillating current carrying the signal generates additional fields, a magnetic field shown here in red, and little loops of current called eddy currents, which are shown here in blue. Now, because of their shape, those loops resist changes to the current at the center of the wire, and they reinforce changes along the surface, but eddy currents get stronger the faster the signal changes. And if you want more calls, you need more bands, which use higher and higher frequencies. So soon the eddy currents are strong enough to cancel out the oscillation in the center of the wire. But if the signal can only travel along the wire's surface, it's limited to a smaller cross section. So resistance goes up and higher frequency signals can't travel very far before they're absorbed. For a while, telephone companies were able to work around the high loss by boosting signals every few kilometers. But past 65 megahertz, well, the high frequencies died off in less than two kilometers, so they were too impractical to use. And this trade-off between signal power and data rate was actually discovered back in 1948. The Shannon-Hartley theorem described the maximum capacity of a communication channel based on the signal power, noise, and bandwidth, or the difference between the highest and lowest frequencies. The best way to increase capacity is to increase the bandwidth. But they'd already reached the limit for copper. The telecoms industry didn't have another alternative, but then they saw the light. Visible light has frequencies in the hundreds of terahertz. As one researcher for Bell Labs put it, "the possibility of extremely high frequencies and accordingly vastly wider bands of frequencies appeared as a fabulous El Dorado, always beckoning us onward."
--- ### 光纤光损耗困局与高锟的理论突围 利用光波通信面临两大前置瓶颈：首先是缺乏高强度、可调制的发射光源；1960年**西奥多·梅曼**（Ted Maiman）发明了**激光器**（Laser），瞬间解决了光源问题并重燃了光通信希望。但第二道难关接踵而至——玻璃看似透明，实则光通量透过率极低。在激光问世仅一年后，贝尔实验室研究人员评估光纤的传输损耗（以分贝/公里衡量信号功率衰减）：若损耗低于20 dB/km（相当于输入100 kg沙子运送一公里后残留1 kg，即保留1%信号功率），光纤才具备替代铜线的商用价值。然而测试现有顶级光纤的结果令人绝望——高达 1000 dB/km。从 1000 dB/km 优化至 20 dB/km，意味着信号传输效率需要实现 $10^{98}$ 倍的提升——这相当于从可观测宇宙的总质量中最终只剩下一个原子的极端差距。面对这一看似工程天方夜谭的数字，**贝尔实验室**（Bell Labs）果断放弃了光纤路线。 然而，年轻的电气工程师**高锟**（Charles Kao）并未随大流。1960年他在英国标准电信实验室（STL）工作，面对上级将其研究斥为“徒劳无功的捕风捉影（a wild goose chase）”，高锟反复追问本质问题：“玻璃为什么会有如此高的损耗？这些损耗因素究竟是可以人为消除的杂质所致，还是玻璃材料本身的内禀属性？”如果属于内禀物理特性，光纤必死无疑；如果是杂质，则尚存生机。 首先被排除的是“光从纤芯侧壁泄漏”。光在玻璃内部能够高效约束，依托的是**全反射**（Total Internal Reflection: 光从光密介质射向光疏介质且入射角大于临界角时全部反射回原介质的现象）。依据**斯涅尔定律**（Snell's Law），光穿过介质交界面的折射程度由折射率决定；当入射角达到临界角乃至更大时，所有光能全部反射。由于空气接触受损或污染会破坏这一全反射边界，在此十年前研究人员便确立了**双层玻璃结构**：由高折射率的“纤芯”（Core）负责导光，包裹以低折射率的“包层”（Cladding），使光线牢固封闭于中心。 既然并非边界泄露，高锟转向分析物质内部机制：**散射**与**吸收**。二氧化硅（$\text{SiO}_2$）是非晶态固体（Amorphous solid），四面体原子结构在无序冷却时会残留微观密度波动，形成**瑞利散射**（Rayleigh scattering）。高锟通过查阅**康宁公司**（Corning Glass Works）科学家**罗伯特·莫勒**（Robert Maurer）关于玻璃散射的论文计算发现：即便在1000摄氏度高温下固化结构，玻璃的理论瑞利散射损耗极限也仅约为 1 dB/km，远不足以解释千分贝级的衰减。问题必然出在**吸收**。 从能带理论来看，固体中的电子能级形成能带。纯二氧化硅中硅原子与氧原子紧密成键，其**带隙**（Band gap）宽达约 9 电子伏特（eV），远高于可见光光子能量（1.63至3.26 eV），因此可见光无法激发电子跃迁，纯石英理应完全透明。常规玻璃的高损耗源自原材料（石英砂、纯碱、石灰等）中夹杂的微量过渡金属杂质，尤以**铁离子**（Iron ions）最为严重。铁离子在玻璃网状结构中形成正离子，强烈吸收紫光、蓝光及红外波段，使得厚玻璃边缘往往仅透出微弱的绿光。高锟据此推断：一旦彻底剔除这些金属离子杂质，玻璃的损耗完全能够降至 20 dB/km 以下。1966年，高锟团队发表了开创性论文，坚定指出“关键的材料问题极为困难，但并非无法实现”。随后三年间，由于缺乏长纤维样品，高锟与团队自主研制了极高精度的微量损耗测量仪；1969年，他对不足30厘米的高纯熔融石英样品测得其实际损耗低于 5 dB/km，震惊了全球电信学界。
Original English Source First, there wasn't a strong enough light source to send a signal. A light bulb couldn't produce enough power. There just weren't any options. - And along in 1960, Ted Maiman invented the laser. - [Reporter] A science fiction weapon that melts steel, a death ray that could be more deadly than a nuclear bomb. It's a beam of light. This is called a laser. - [Henry] There's a running joke that the laser was "a solution looking for a problem," but that's not really true. - [Donald] It rejuvenated the interest in possible optical communications. - As one Bell Labs engineer said, "Think of all this bandwidth!" But they couldn't figure out the second problem. Glass is not as clear as it looks. Only a small amount of light actually makes it through. In fact, only a year after the laser's invention, a Bell Labs scientist started looking into fiber optics. His boss told him to check its loss, or how much a light signal weakens as it travels through the glass. Now, signal loss is measured with decibels, similar to how you measure audio volume. Both are on a logarithmic scale, but while volume uses sound pressure, loss uses signal power. His boss said it would only be worthwhile if the loss was less than 20 decibels a kilometer, so it was comparable to copper. But when he checked the loss for the best fibers, a thousand decibels a kilometer. To understand just how bad that was, imagine a leaky truck carrying sand over one kilometer. If the truck starts with 100 kilograms, and it ends up with one, well, that represents a loss of 20 decibels per kilometer. That's only 1% of the signal left. It's what was happening with copper, bad but still manageable. A loss of a thousand decibels a kilometer is like starting with the mass of the observable universe and ending up with less than the mass of a single atom. To be practical, these fibers would need to improve by a factor of 10 the 98th. It sounded impossible. It would have to be one of the most absurd engineering improvements in history. So Bell Labs gave up. And fiber optics might have died right there if it wasn't for one man. Charles Kao. In 1960, he started working for an English telecoms lab that was looking for a way to communicate with higher frequencies, but not with fiber optics. Kao's boss said that it was "a wild goose chase" and the rest of the lab agreed. But Kao kept coming back to it, asking, "Why is glass so lossy and can we remove those factors or are they intrinsic to the material?" If it was the latter, then fiber would be dead in the water. But if not, then maybe it had a chance. So the question was: Why was the loss so high? Well, you might think it's because light is exiting through the sides of this fiber, but glass is really good at trapping light, even through bends. Here's how. I've got a bucket of propylene glycol here and a green laser. When I remove the tape, the laser bounces one, two, three times. And there are two things working together here that actually trap the light. See, light travels slower in some materials than others, and it travels as a wave shown here as a line of crests with the beam always pointing perpendicular to that line. When the wave meets a boundary at an angle, one end of the crest crosses into the new material first, and it changes speed before the rest. If light travels slower in that new material, then that end drags behind and the beam bends or refracts away from the boundary. How much it bends depends on the angle it hits the boundary and on how much each material slows light down, which is described by their index of refraction. This relationship is defined by Snell's Law. Now if light is moving from a material with a higher index of refraction into one with a lower index, like propylene glycol to air, well then one part of the wave crest speeds up and the beam bends toward the boundary. But as you increase the angle of the light, the refracted beam gets closer and closer to the boundary until it's running right along it. That's the critical angle. A little bit more and all the light reflects back into the glycol. That's total internal reflection, and you get the exact critical angle by setting the angle of refraction to 90 degrees and then rearranging Snell's Law. - The key of it is the bigger the difference in index of refraction, the more powerful this effect is. If I have a really subtle difference between these two, then the amount of bending that takes place will be smaller. - You can do the same demo with water, but propylene glycol works a bit better because it has a higher refractive index, 1.43 compared with 1.33. See, once the liquid is low enough for the angle of the stream to be much sharper, the light breaks through the water stream, but it stays trapped in the propylene glycol. And if you use something with an even higher refractive index, like the glass types Kao was investigating, well, then the critical angle decreases even more, which traps the light signals through even sharper turns. Okay, but the light has to move from a high refractive index to a low one. In air that works right up until something touches the glass. If you disturb that boundary, well, light can leak out the side. So about a decade before Kao, researchers realized that a fiber needed to have two types of glass. The middle, the core, has the higher refractive index, and around it goes a second glass, the cladding, with a lower refractive index. Together, they keep the light trapped. Now the fiber could be touched, coated in plastic, buried in a cable, bent around things, and the light would still travel down the core and come out the other end. So Kao knew that if the high loss wasn't from light leaking out the sides, well, it had to be from something else. He spoke with material scientists and glass experts, but no one seemed to know what was really causing it. He kept looking and eventually he landed on two potential sources: scattering and absorption. If you shine light through a perfect crystal that's transparent, there's going to be a little bit of reflection, but almost all of the electromagnetic energy will keep going down in the same direction that it started. But glass isn't a crystalline solid. The basic building block is silica or silicon dioxide, one silicon atom connected to four oxygen atoms in a tetrahedra or pyramid shape. In the glass, the pyramids are connected at the corners, so every oxygen is shared with another silicon. That's why it's called silicon dioxide. For every two oxygens, there's one silicon atom. This is an amorphous solid. Unlike a crystal, its atoms have no long-range order. Instead, its disordered liquid-like structure locks in as it cools, with the exact arrangement depending on the glass type and cooling speed. For example, if you quench the glass really fast, the structure locks at a higher temperature. The final solid has lots of dense regions, which re-radiate more light than the sparse ones. The light scatters, which weakens the signal. This is called Rayleigh scattering, and you can't get rid of it completely, but if you cool the glass slowly, those pyramids have more time to settle and pack together evenly. The structure locks in at a lower temperature and light scatters less. But Kao needed to know the minimum scattering. Was it 10 decibels a kilometer, a hundred, a thousand? He needed something concrete. Finally, he stumbled on a paper by Robert Maurer, a scientist from Corning Glass who measured the Rayleigh scattering across several glass types and cooling rates. Kao used his calculations to estimate that if the glass's structure was locked at a thousand degrees Celsius, well, its scattering loss would be around one decibel a kilometer. He was surprised. He expected scattering loss to increase with higher temperatures. That's what causes more density fluctuations, but a thousand degrees Celsius was pretty hot compared to other glass types, so then scattering couldn't be behind those crazy high losses. Absorption had to be the problem, but Kao didn't know what exactly was absorbing so much light. Was it the glass itself or something else? Well, absorption works like this. In a single atom, electrons sit at discrete energy levels, and there are usually some empty spaces in the upper ones, but pack atoms together into a solid and those levels smear up into wide bands separated by gaps. Now when light hits the surface, it can excite an electron to a higher band, but only if the photon has at least as much energy as this band gap. If it does, the photon is absorbed and eventually it releases that energy as heat. In something like silicon, that band gap is very small. It's only around 1.12 electron volts. Visible light is between 1.63 to 3.26. It's got plenty of energy, so it gets absorbed. That's why pure silicon looks opaque to us, but the silicon in glass is connected to oxygen atoms that pull more on the shared electrons. Now it takes a lot more energy to excite an electron. The band gap is much larger, around nine electron volts. Visible light doesn't have enough energy, so it passes right through, and that's why glass looked clear to us. But most glass types aren't pure silica. The raw materials for glass, from sand, soda, lime, and other ingredients, well, they contain impurities like trace metals, and these metals absorbed far more light than silica did. One of the worst impurities in glass is iron. Inside the glass structure, iron atoms form positive ions that strongly absorb violet, bluish, and infrared wavelengths. Of the visible light spectrum, impure glass doesn't absorb green light as much, and you can even see this in a regular glass pane. Looking at it from the edge, light travels down the entire width of the glass, so by the time it reaches your eye, only green light is left. So if you can see this attenuation after just a few centimeters, imagine a glass fiber trying to carry light for kilometers and kilometers. No wonder it had a crazy loss. Kao spoke with more experts who estimated that without those metal ions, the loss would drop to around 20 decibels a kilometer, comparable to copper, just by removing impurities. It was all theoretical at this point, but it was enough for Kao. In 1966, his team published a paper on their theories, concluding with "the crucial material problem appears to be one which is difficult but not impossible." But nobody really paid attention to his paper because it really seemed impossible. A 10 to the 98th times improvement was unprecedented, and Kao didn't actually measure the loss in glass. He didn't have anything concrete, so he needed proof. [Sponsor segment omitted] Now, let's go back to Kao. Kao needed to prove that his theoretical low-loss glass was really possible, so he decided to measure the purest glass anyone had ever made, fused silica. Now, the reason glass manufacturers had added all those extra ingredients was to bring sand's melting point down to around 1400 degrees Celsius, but fused silica was pure silicon dioxide melted at over 2,000 degrees Celsius, so it was a lot harder to make, and that meant that Kao couldn't get his hands on much of it. Measuring loss normally takes hundreds of meters of cable, if not kilometers, but he only had pieces less than 30 centimeters long. To measure the loss in something that small, he and his colleagues invented a new, very precise instrument. Kao spent the next three years working on it, but when he finally measured a sample, he found that fused silica had zero loss. That couldn't be right. In a panic, Kao ran his experiment again. It turned out that the loss was so low that it was within their margin of error, a mere plus or minus four decibels a kilometer. He spent months fine-tuning his equipment, and in 1969, he finally reported that fused silica had a loss of less than five decibels a kilometer. Now, that got everyone's attention. One scientist at Bell Labs said, "Kao gave everyone a jolt. That was the first practical measurement which said, 'Hey, you're not just whistling Dixie.'" Now everyone was interested. Telecom labs around the planet rushed to fund fiber optics groups, but Kao had only worked with small samples. Nobody knew whether kilometers of low-loss fiber were even possible.
--- ### 逆向工程思维与康宁团队的技术攻坚 高锟用微型样本证实了纯石英的理论潜力，但能否工程化制造出数公里长的高性能光纤依然是巨大的问号。此时，前述被高锟引用的康宁公司科学家罗伯特·莫勒接过了这一挑战。 摆在莫勒面前的有三大制造关卡。首先是材料选择：常规玻璃熔炼依赖添加剂将石英熔点从2000℃以上降至约1400℃，但添加剂引入了杂质；**熔融石英**（Fused silica）虽然纯净，但其折射率仅为 1.46，在所有玻璃中几乎最低。这就导致了严重的物理悖论：全反射要求纤芯折射率高于包层，若将纯石英作为纤芯，工程上几乎找不到折射率更低的材料来充当包层。行业其他实验室因此纷纷放弃熔融石英，莫勒却采取了逆向思维：“如果纯石英折射率最低，为何不用纯石英做外层的包层，而在纤芯中反向添加掺杂剂来略微提升折射率？” 莫勒选用了**二氧化钛**（Titania）作为掺杂剂（钛原子也能与四个氧原子配位，不会剧烈破坏石英晶体结构），并依托康宁在1930年代发明的**火焰水解法**（Flame hydrolysis: 通过四氯化硅气体与氧气火焰反应生成极纯的二氧化硅玻璃粉末，即“烟灰” soot）。随后，莫勒聘请了陶瓷工程师**彼得·舒尔茨**（Peter Schultz）和物理学家**唐纳德·凯克**（Donald Keck）组成攻关三人组。他们设计了**预制棒**（Preform）沉积工艺：取纯熔融石英管，利用吸尘器抽吸负压引导气流，将掺钛石英粉末均匀附着在石英管内壁，随后高温熔塌成实心预制棒，再通过高温拉丝炉拉制成细纤。 ``` [火焰水解反应] SiCl4 + 2H2O + O2 ---> SiO2 (超纯玻璃粉末/Soot) + 4HCl ``` 在1969年炎夏，团队在高温炉前尝试了数十种成分与拉丝方案。起初钛在极高温度下出现分解加剧了吸光，团队通过额外的氧气热处理逐步克服缺陷。1970年7月22日，一根损坏的预制棒拉出仅29米短纤维，凯克在对其进行热处理测试时，惊讶地发现激光打入纤芯时光芒直透远端并反射回视线，测得损耗仅为 17 dB/km。他在实验记录本上写下了著名的“Whoopie”（万岁）。1970年8月21日，凯克在空无一人的实验室里完成了对210米完整光纤的复测，确认损耗为 **16.9 dB/km**——人类历史上首次打破了 20 dB/km 的商用门槛。莫勒带着这卷光纤巡回展示于各大电信实验室，甚至寸步不离严加看管，以防被同行剪走研析。 两年后，团队以**二氧化锗**（Germanium dioxide）替代二氧化钛作为掺杂剂。锗不仅完美提升折射率，而且耐高温性能优异，免去了繁琐的二次热处理，制造出的光纤在拉丝过程中即可从头至尾通体透亮。该光纤损耗直接骤降至 **4 dB/km** 且具备极高机械抗拉强度，标志着首批真正兼具商业可行性与可靠性的低损耗光纤问世。
Original English Source The problem caught the attention of one Robert Maurer, Corning researcher and author of the scattering paper that Kao had referenced. To make a practical fiber, he saw three major problems to tackle. First, he needed to decide which glass to use. He could try purifying raw materials for normal glass, or he could try using fused silica, which Kao had already proven to be very pure. Most labs were avoiding fused silica because of how hard it was to work with. But luckily, Corning had a secret technique. Back in the 1930s, another Corning scientist had ran silicon tetrachloride gas through an oxygen flame. In the heat, water vapor from the flame would break silicon tetrachloride into hydrogen chloride gas and silicon dioxide. The latter would then start clicking together in a fine glassy silica powder. Then the powder would be pressed into shape and melted to make fused silica. Corning was one of the few companies in the world using this technique called flame hydrolysis to make ultra-pure glass. So they had the supply, but there was a bigger problem with fused silica. It has a refractive index of 1.46, which is the lowest of any glass, but for total internal reflection to work, the core needs to have a higher refractive index than the cladding. So if Maurer wanted to use a fused silica core, well then finding another glass with a lower refractive index for the cladding, well, that'd be a nightmare. It simply didn't exist. So everyone abandoned fused silica except for Maurer. He said, "I had no reason to think fused silica was any better, but nobody else was working on it. If you do something that's different from what everybody else is doing, you've got two advantages. One is you may succeed where they fail, of course, but even if you fail, you will gather information that they don't gather." All the other labs were thinking of using a fused silica core. After all, that's where the light travels, but Maurer decided to use fused silica for the cladding. Then he could easily raise the refractive index of the core by adding some impurities or dopants. - The only glass that we had with a slightly higher index than fused silica was this titanium-doped glass. - [Henry] But that would mean adding titanium right where light needed to travel. Kao worked hard to prove that glass could be pure enough. All the other labs were trying to get rid of impurities, and Maurer wanted to add some back in. But he wasn't completely crazy. Like silicon, the titanium in glass can easily share electrons with four oxygen atoms, forming the same pyramid shape, so it wouldn't disrupt the overall glass structure too much. Now that he had the composition, Maurer needed to tackle that second problem. How do you make a tiny fiber with that perfect core and cladding cross section? To help, he hired two more scientists, Peter Schultz, a ceramics engineer, and Donald Keck, a physicist. - Each time you find a problem, you put a development team on it and strike it down and make it happen. - To understand their technique, I went to the Bay Area Glass Institute for some hands-on demos. Here, our instructor, Erin is taking a colored rod of glass and then dipping it in a bowl of clear molten glass. In fiber optics, this large cylinder is called a preform. We have two colors of glass here. We have a color core and a clear outside. If you want your inner rod to, say, be a quarter of the diameter of your outer one, you just set that in the preform, and then all the length along your pole, the cross-section's going to be identical. And this is pretty similar to how hard candies with patterns are made. This sugar is an amorphous solid, just like glass. So the process is the same, just at a lower temperature. But there is one problem with this method. Resting glass on a surface like this introduces a ton of stress and impurities, but when I try and suspend it in the air, well, it sags. - [Erin] But yeah, from all that movement, it did create stress, so now it's a curved piece of, of glass. - Thankfully, there's another option. Now, gravity is pulling our glass for us, and the surface section is keeping it together. So now we get a much more straight and consistent fiber. That's awesome. - With their drawing technique sorted, the corning team could tackle the final problem. How do you actually make the preform? Well, instead of starting with a core and adding the cladding, Schultz and Keck decided to start from the cladding and work in. They used the flame hydrolysis technique to make titanium dope silica powder, or soot as they called it. Next, they took a pure fused silica tube and tried spreading soot down the inside. If they could get an even layer, well, then they could melt and collapse the tube into a solid rod, a perfect preform. But the soot collected at one end no matter how they sprayed it. - So one of the technicians grabbed a vacuum cleaner from the corner of the room, and he held it up to the backside of this tube. And it sucked a stream of titania dope silica glass down through the center of the tube, coated it just like we'd hoped it would. - [Henry] The vacuum did not survive that experiment, but they tinkered with a new setup until they got the gas flow just right. All through the summer of 1969, their furnace burned "hotter than Hades." They drew kilometers of fiber from preform after preform, testing various compositions, but none of it was good enough. Titanium was breaking down in the high heat and absorbing too much light. They ran the finished fibers through a second heat treatment in an oxygen environment before they finally started seeing results. A year later, on July 22, 1970, a preform slipped against the furnace wall after Keck only got a few meters of fiber. The preform was ruined, but he decided to test the short fiber anyway after its heat treatment. - When I slid it across and the laser light hit the core of the fiber, all of a sudden I had another beam coming back in my face. The core was brightly lit, and it took a while to realize that the light was going all the way to the other end of the fiber and reflecting off the far end and coming back into my eye. - [Henry] The loss was 17 decibels a kilometer. - The only word I could come up with for this auspicious moment in history was "Whoopie". So that's what's in my lab notebook, "Whoopie". - [Henry] But the fiber was much shorter than usual, only 29 meters to their typical 200. He added a careful "must remeasure" in his notebook. Team quickly remade the preform, but it took almost a month to get 210 meters of finished fiber. It was finally done by late afternoon of Friday, August 21st. It was a gorgeous summer day and almost everyone left early, but Keck didn't want to wait. The fiber was so fragile, small temperature changes might break it over the weekend. - It would break if you looked at it crossways. - So alone in the lab, he anxiously measured the loss, 16.9 decibels per kilometer. - I said "QED, quod erat demonstrandum" There was no, (chuckles) no eureka. - [Henry] Maurer published their results that fall. As Schultz described it, no one believed it. And telephone companies all demanded to see the fibers firsthand, so Maurer hand carried a spool of fiber to all the big telephone laboratories. He would literally guard to make sure that they didn't snip pieces off and keep them. For two years, they refined their methods. And to make the fiber stronger, they switched to a germanium doping. Like titanium it easily swaps with silicon while still raising the refractive index. But unlike titanium, germanium fibers didn't break down in the heat, so they could skip the second heat treatment. When they drew their first germanium fiber, the end lit up with the light from the furnace. Even when it reached the bottom and wound up around a large drum, the fiber stayed bright. - And we kept winding and winding and winding and thinking the light was going to go out, and it never went out. - The loss was only four decibels per kilometer with excellent strength. Keck considers that the very first truly practical low-loss fiber.
--- ### 数字通信跃迁与现代光纤产业规模化 低损耗光纤的技术突破迅速引发了全球通信网络的底层换代。1977年，AT&T在芝加哥地下完成了首个光纤系统测试；1979年，日本研发团队将光纤损耗降至 **0.2 dB/km** 的物理极限级别，并一直沿用至今。1988年，人类首条跨大西洋光缆（TAT-8）正式铺设投运。 光纤不仅改变了传输介质，更重构了信号传输体制。在铜线时代，音频以模拟连续波形式传输；而在光纤中，语音信号被**离散采样**并量化为由0和1组成的二进制数字信号，半导体激光器以每秒2.8亿次（280 MHz）的极高频率闪烁（亮代表1，灭代表0）。更为关键的是**时分复用**（Time-Division Multiplexing）技术的引入：多路通话的比特流像拉链咬合一样精密穿插交织传输，接收端依据严格的时序节拍即可无损还原。首条跨大西洋光纤电缆单次可承载40,000路并发通话，传输容量达到以往同类铜线电缆的十倍。这一飞跃彻底打破了电话呼叫的限制，文本、图像、视频的多媒体数字化传输顺理成章，直接催生了现代互联网。 在现代工业制造中，光纤制造已演变为高精尖的等离子体沉积极限工艺。以全球线缆巨头普睿司曼（Prysmian）位于北卡罗来纳州的工厂为例： 1. **等离子体内沉积**：微波源使混合气体激发为数千度的紫色等离子体，在石英管内壁往复扫描数千次，逐层沉积精确控制折射率梯度的超纯二氧化硅薄膜； 2. **表面张力熔塌**：石英管随后送入高温炉，无需外力施压或机械抽吸，仅凭借玻璃分子自身的表面张力即可平滑塌陷为实心棒； 3. **外包层等离子体喷涂**：利用高达上万摄氏度（约为太阳表面温度三倍）的等离子体炬将石英颗粒瞬间熔融并喷覆于外壁，使纤芯直径控制在最终成品直径的精确比例（约7.2%）； 4. **高塔垂直拉丝与退火**：制成的预制棒移送至七层楼高的拉丝塔顶部，底部加热至2000℃以上软化形成泪滴状液滴下坠，被拉伸至标准直属的 **125微米**（$\mu\text{m}$）；随后的六层楼空间用于严密受控的缓慢降温冷却，以彻底消除内应力与微观密度涨落引起的结构性光损耗；冷却后立刻浸涂两层高分子保护胶以隔绝微划痕。 制成的光纤在同等粗细下的抗拉强度甚至超过钢丝，但由于脆性断裂机制，其极其畏惧尖锐的弯折与打结，因此必须经由多层铠装复合为高强度光缆。目前全球已敷设超过 **80亿公里** 的光纤网络，配合后续发明的**光纤放大器**（Fiber Amplifier: 彻底免除长途光电转换并大幅增强越洋信号），共同构建了数字文明的物理大动脉。 2009年高锟被授予诺贝尔物理学奖。他在获奖感言中总结道：“始于20世纪90年代的信息革命，如果没有光纤是绝不可能发生的。这甚至远超1966年最初关于光通信的构想，那时我们甚至连1 GHz带宽都仅仅视作一种奢望。”回顾这一历程，高锟并非传统材料学家，而是电气工程师，恰恰是作为“领域局外人”的独特视角，让他敢于质询材料专家习以为常的固有成见；而莫勒与康宁团队则敢于通过“故意掺杂”这种反直觉的逆向技术路径攻克难关。对既定边界的追问与对工程直觉的颠覆，最终将一道微弱的光斑化作了贯通全球的信息洪流。
Original English Source By 1977, AT&T trialed a fiber system underneath Chicago. Two years later, a fiber team from Japan was able to get a loss of just 0.2 decibels a kilometer, the standard up to today. Fiber networks continued to grow with the first transatlantic fiber optic cable installed by 1988. And it transmitted those phone calls differently than in copper. The audio wave was split up or sampled into discrete values, and then converted into binary to get a long string of ones and zeros. Then a laser flashed on for one and off for zero, 280 million times a second. These pulses of light traveled down the fiber, but that wasn't all. Multiple conversations could now be weaved together, one bit from one conversation, then one from another and so on, slotting it together like a zipper. As long as the receiver knew the timing of each conversation and the sampling rate, well, they could reconstruct the original waves. So this first intercontinental cable could carry 40,000 simultaneous calls, 10 times more than the previous copper version. This insane capacity opened up a whole new world, and with digital signals people realized that they weren't limited to phone calls. It was easier than ever to send texts, images, videos. All building up to the internet, and since then demand has only increased. - This reel right here? - Yeah. - 50.477 meter kilometers, so 50 kilometers. So it's about 30 miles. - [Henry] That's crazy. Like how much is in this warehouse? - There's millions of kilometers here that we're pushing through here every month. - [Henry] To see how modern fibers are made, we visited Prysmian's North Carolina factory. It all starts with the preform. Each pass of this microwave source ignites a mixture of gases into a plasma which deposits a thin layer of glass on the inside of the tube. - And you see when it turns purple, that's when we're making a plasma with that gas. - So, how thick are these deposition legs? Like how many times are you doing this? How thick are you doing-- - Thousands of passes. - [Henry] "Thousands of passes." - Several thousand times it's doing this. Every time we may go back and forth, we could change the composition of that if we wanted to. - [Henry] Oh, okay. After all the inner layers are done, the tube has a thicker wall, but there's still a gap down the middle. - So we want to collapse that channel upon itself. So what we do is we heat it up. We have a furnace that goes back and forth, and just via surface tension, it collapses upon itself. - Interesting, I would've thought maybe you'd like pull it, or just-- - [Sujeet] And I asked this exact question. I said, "Are you pulling a vacuum in there or what's making that thing shrink?" - Exactly. - It's really just a glass phenomenon where the molecules are trying to pull together. You heat it up, you'll lower the viscosity of the glass. It just... (exhales) - [Henry] Now they need to add more silica on the outside to get the preform to the right proportions. The core is very small, just 7.2% of the total diameter. - And you have a several thousand C plume of plasma right there. There's grains of silica that's flowing into that plume and effectively melting it and spraying it onto that rod. - [Henry] Wow. - That plasma torch is approximately three times as hot as the surface of the sun. - Plasma always blows my mind. This is absurd. Once the preform's done, it goes to the top of the draw tower, seven stories up to start making fiber. It really looks like some sort of steampunk thing, you know? All these coils and then this hot glass going down. The bottom of the preform is tapered to a point and heated to over 2,000 degrees Celsius. It's like there's some heat coming off this. The end of the taper slowly drops down to the floor below. - [Man] We call it a gob, but it looks like a teardrop. - The fiber is a little thick, so it'll be thinned down to the final 125 microns before it passes to the next level. It's like a little flashlight coming down. I didn't expect it to be so violent. If the drawing, like the pull down to that final diameter happens in such a short amount of... - In a few meters, let's call it. - A few meters. Then why do you need six more floors? - You got to cool the glass before you can apply the coating. And if you cool the glass too quickly, it'll have residual stress and then have optical loss attenuation. - Once the fibers are cool enough, they are dipped in two protective plastic coatings. At the bottom of the draw tower, the fiber is wound onto massive spools, each capable of carrying hundreds of kilometers at over 50 meters a second. Okay. Next, the finished fiber is tested for loss and strength. Fiber is stronger than a steel wire of the same thickness. I guess if I try to do it like... - [Sujeet] If you just try to do it like that, it's really hard to break. - First of all, it just slides down my hand. - [Sujeet] And that's a glass fiber with two coatings on it. - Ah, it's just hurting my hand. Oh, my god. Yeah, not easy. Definitely not easy. Ow. But it's weak to sharp turns. - [Sujeet] Tie it in a knot. - Tie it in a knot. - And watch what happens and see how much you can get before it breaks. Yeah, because the angle's so great, you can break it. - But after testing, fiber is bundled and wrapped into thicker cables so that it's harder for them to knot and break. These cables will join the more than eight billion kilometers of fiber already installed that carry information around the world, powering the internet and letting you watch this video. But some experts argue that even with fiber, the internet would not be possible without another key invention. The amplifier, that's what boosts signals across long stretches, and it revolutionized fiber capacity on land and sea. So if you'd like to see a video about submarine cables and fiber amplifiers, well, let us know in the comments. As for Kao, his dream went further than he ever thought. When he received the Nobel Prize in 2009 for his work, he said, "The information revolution that started in the 1990s could not have happened without optical fibers. It is beyond the dreams of the first serious concept of optical communication in 1966, when even one gigahertz was only a hope." For me, one of the things I kept on coming back to in this story is that Kao was not a material scientist. He was an electrical engineer. And it was only because he was an outsider that he was able to ask the questions that nobody was asking. He was able to push the status quo, but he was not the only one. Maurer's team were also working with a very unintuitive idea that by making something seemingly worse, they actually were making it better. As Kao said, "Who knows what other forward thinkers will dream of in the future?" Granted, he was talking about bandwidth, but I think it applies anywhere where people are able to question why things are the way they are and whether that's really the limit.