Working in the Dark: Secrets, Silicon, and Light

Speaker 0:21

Pick up your phone, tap the glass. It responds flawlessly. We communicate through them and they tell us where to drive. We use them as flashlights. They function so seamlessly that we stop wondering. We accept them as polished little black boxes. Generally, unless you like to take electronics apart, you don't ever see into your devices. But you've heard of Silicon Valley. You know there's silicon in there, somewhere. And silicon is in sand. Quartz sand. But you can't shovel sand off the beach and turn it into the hardware that runs your phone. The physical reality of the digital world requires us to turn sand into something flawless. It requires light that is barely possible. And it all started with a mistake. I'm Daina Bouquin. This is found in the machine. The year is 1916. The world outside is tearing itself apart in the trenches of the First World War. But inside a metallurgy laboratory in Berlin, it is quiet. Jan Czochralski is working late. A Polish man. He is a self-taught chemist. He has spent the entire day and most of the evening staring at metals, heating them, cooling them, trying to understand the exact speed at which they crystallize. The kind of repetitive observation that blurs the lines of your vision. On his desk there is an inkwell. Beside it, there is a small crucible left over from a previous experiment. It contains a cooling pool of molten tin. Jan reaches for his fountain pen. Without looking away from his papers in front of him, he dips the nib of the pen, but his hand misjudges the distance. He dips the pen into the crucible. It goes straight into the liquid tin. Realizing his mistake, he quickly pulls the pen upward. Then he stops because something catches the flickering light. A silvery, hair-thin, shimmering wire is hanging from the tip of his pen. He doesn't just wipe it away. He looks. He wonders. Then he does it again. The pen goes back into the tin and slowly, deliberately this time, he pulls it up again. The thread grows. The tiny slit in his pen's nib is acting like a capillary. The liquid metal is being pulled up and is cooling the moment it touches the air, clinging to itself as it rises. And this isn't just a wire. As Jan pulled the pen, the atoms in the pool of tin didn't scramble into a chaotic jumble as they cooled. They aligned. This wire is a single, unbroken, continuous crystal lattice. It is perfect. Jan Czochrolski has just invented a way to grow flawless crystals. He wrote a paper about it. He called it a new method for measuring the crystallization rate of metals. But he had no idea what he had actually done. He would never know. There are two streams of materials that must collide to make a microchip possible. First, there is the silicon that will become the chip itself. It begins its life as metallurgical grade silicon, forged by taking high-purity quartz, which is a rock mostly mined in deposits from China, Norway, and Brazil, and throwing it into an electric arc furnace with carbon. It is blasted at up to 2,000 degrees Celsius. Then it is chemically purified over and over again until it reaches what engineers call the 9N level. That's 99.9999999% pure. Out of one billion atoms, only one can be an impurity. But having the purest liquid silicon on earth is only the first step. You also need to hold that perfect liquid without contaminating it. You need a very, very special crucible. The crucible you need does not come from a global supply chain. Overwhelmingly, it comes from a single town in the Blue Ridge Mountains, Spruce Pine, North Carolina. Through a geological anomaly, Spruce Pine holds the purest natural quartz ever discovered on Earth. So you have the ultra-pure silicon gathered from across the globe. You have the perfect quartz crucible pulled from the mountains of North Carolina. The silicon is molten inside that pristine vessel. Now, how do you get it out? How do you turn it into a pristine canvas for microchips? You use the method that Jan Czochrolski accidentally invented when he was too tired to notice where he put his pen. You see, in the late 1940s and early 1950s, as researchers tried to make reliable transistors, they kept running into the same wall. They needed perfectly ordered semiconductor material. A single flaw, a single microscopic misalignment in the crystal lattice, and the electrical signals would scatter and fail. They needed a perfect canvas. And they did not know how to make one. Until they found an obscure paper from 1916. Today, nearly all silicon wafers on Earth are made using Czochrolski's method, often referred to as the C Z method. Inside those ultra-pure Spruce Pine quartz crucibles, we dip a tiny seed of crystal into the molten 9N silicon, and slowly we pull it upward, spinning as it cools. We grow giant, flawless crystal cylinders, hundreds of pounds heavy, atoms aligned in perfect lockstep. But Jan did not spend the rest of his life being celebrated for it. He left Germany in 1928, returning to a newly independent Poland. He became a professor at the Warsaw University of Technology. He was a brilliant, wealthy patron of the arts. Then came 1939. The invasion. The universities were violently shut down. Academics were arrested, sent to camps, executed. Jan, leaning on his pre-war connections in Germany, and his German wife, got permission from the occupying forces to keep his laboratory open. He rebranded it as a materials research facility. To the people on the street, to his former colleagues, he looked like a collaborator. A man who had traded his country's dignity for his own survival. But they did not know that the lab was a front. Jan was secretly employing members of the Polish resistance, providing them with documents to keep them out of the camps. He was using the lab's machinery to reverse engineer German V-1 and V2 rockets. He was manufacturing grenade shells for the underground home army. His work had the approval of the Polish underground state. There's evidence that he sheltered two Jewish women in his own home. But he looked like a collaborator. And when the war ended, the new communist regime arrested him and tried him for treason. He was acquitted. An investigation proved that he did not collaborate with the Nazis. He had never been a traitor. But it did not matter. The reputational damage was done. He was stripped of his professorship. His name was meticulously and deliberately erased from the university's records. A man who would later be listed among the fathers of modern electronics died in 1953, running a small operation in his hometown, making cosmetics and sneezing powder. Here is what he never saw. By the time Jan died, the canvas was solved. His crystals would become every microchip we would ever build. The brains in every supercomputer, satellite, and smart toaster. But you still have to draw the pathways for the electricity to flow. In the early days, we drew those pathways using visible light, shining it through a stencil-like mask onto the silicon to etch the circuits. But as we demanded smaller and faster machines, the stencils had to shrink. Today there are billions of transistors on a single chip. Those transistors are only a few nanometers wide. Five nanometers is about the length your fingernail grows in five seconds. You cannot etch something that small with ordinary light. The wavelength of visible light is too wide. It is like trying to paint a microscopic masterpiece with a heavy push broom. To work at that scale, you need light that is roughly 100 times smaller than visible light. You need extreme ultraviolet light. The machines that make and harness EUV light are among the most complex devices humanity has ever built. One machine costs hundreds of millions of dollars and contains more than 100,000 parts. Inside, EUV light bounces through a series of mirrors before passing through the stencil. More mirrors focus it onto the silicon. Those mirrors are polished so perfectly that if you expanded one to be the size of Germany, the tallest bump on its surface would be less than a millimeter high. Those mirrors a blueprint for billions of transistors onto the silicon that Jan's method grew with unthinkable precision. It is akin to standing on Earth and shooting an arrow through an apple on the surface of the moon. Then doing it again . And again. Thousands of chips can be etched onto a single silicon fever. And the light itself does not want to exist. Extreme ultraviolet light is swallowed by air. It does not occur naturally on Earth. So, deep within the EU V machine, there is a vacuum chamber. In that airless space, a generator ejects a microscopic droplet of liquid metal into the dark. Then another. 50,000 droplets , every single second. As each droplet falls, a high-powered laser fires a weak pulse, just enough to flatten the tiny sphere into a tiny pancake. A microsecond later, a second pulse fires with devastating force. The metal vaporizes. It becomes a plasma burning 40 times hotter than the surface of the sun. And in that violent, microscopic death, it flashes, throwing off a burst of extreme ultraviolet light. And the metal, the drop that dies in the dark to light the digital age. It is tin. Molten tin. In 1916, a tired man dipped his pen into a crucible of tin and pulled out the foundation of the digital world. More than a century later, we blast falling drops of that same metal to make light so fragile it can barely exist. That light is used to draw microchips onto the perfect crystals that power our devices. And we use those devices as flashlights. I'm Daina Bouquin. And this is Found in the Machine. Before you go, if you'd like to support the show, rate and review this podcast wherever you listen. Or better yet, share it with a friend. I'm really trying to turn this into something sustainable, and I'd love your help getting the word out. Another way you can support the show is by going to bookshop.org/shop /found in the machine. Your purchases support independent bookstores across the US, and they also help make this podcast possible. And if you'd like to dive deeper into any of these stories, you can sign up for my newsletter at notes.foundinthemachines.com. Thanks for listening.