by Chris Woodford. Last updated: April 18, 2016.
Have you ever heard of a 1940s computer called the ENIAC? It was about the same length and weight as three to four double-decker buses and contained 18,000 buzzing electronic switches known as vacuum tubes. Despite its gargantuan size, it was thousands of times less powerful than a modern laptop—a machine about 100 times smaller.
If the history of computing sounds like a magic trick—squeezing more and more power into less and less space—it is! What made it possible was the invention of the integrated circuit (IC) in 1958. It's a neat way of cramming hundreds, thousands, millions, or even billions of electronic components onto tiny chips of silicon no bigger than a fingernail. Let's take a closer look at ICs and how they work!
Photo: An integrated circuit from the outside. This is what an IC looks like when it's conveniently packaged inside a flash memory chip. Inside the black protective case, there's a tiny integrated circuit, with millions of transistors capable of storing millions of binary digits of information. You can see what the circuit itself looks like in the photograph below.
What is an integrated circuit?
Open up a television or a radio and you'll see it's built around a printed circuit board (PCB): a bit like an electric street-map with small electronic components (such as resistors and capacitors) in place of the buildings and printed copper connections linking them together like miniature metal streets. Circuit boards are fine in small appliances like this, but if you try to use the same technique to build a complex electronic machine, such as a computer, you quickly hit a snag. Even the simplest computer needs eight electronic switches to store a single byte (character) of information. So if you want to build a computer with just enough memory to store this paragraph, you're looking at about 750 characters times 8 or about 6000 switches—for a single paragraph! If you plump for switches like they had in the ENIAC—vacuum tubes about the size of an adult thumb—you soon end up with a whopping great big, power-hungry machine that needs its own mini electricity plant to keep it running.
Photo: An integrated circuit from the inside. If you could lift the cover off a typical microchip like the one in the top photo (and you can't very easily—believe me, I've tried!), this is what you'd find inside. The integrated circuit is the tiny square in the center. Connections run out from it to the terminals (metal pins or legs) around the edge. When you hook up something to one of these terminals, you're actually connecting into the circuit itself. You can just about see the pattern of electronic components on the surface of the chip itself. Photo by courtesy of NASA Glenn Research Center (NASA-GRC).
When three American physicists invented transistors in 1947, things improved somewhat. Transistors were a fraction the size of vacuum tubes and relays (the electromagnetic switches that had started to replace vacuum tubes in the mid-1940s), used much less power, and were far more reliable. But there was still the problem of linking all those transistors together in complex circuits. Even after transistors were invented, computers were still a tangled mass of wires.
Integrated circuits changed all that. The basic idea was to take a complete circuit, with all its many components and the connections between them, and recreate the whole thing in microscopically tiny form on the surface of a piece of silicon. It was an amazingly clever idea and it's made possible all kinds of "microelectronic" gadgets we now take for granted, from digital watches and pocket calculators to Moon-landing rockets and missiles with built-in satellite navigation.
Photo: A typical modern transistor mounted on a printed circuit board. Imagine having to wire hundreds of millions of these things onto a PCB!
How are integrated circuits made?
How do we make something like a memory or processor chip for a computer? It all starts with a raw chemical element such as silicon, which is chemically treated or doped to make it have different electrical properties...
If you've read our articles on diodes and transistors, you'll be familiar with the idea of semiconductors. Traditionally, people thought of materials fitting into two neat categories: those that allow electricity to flow through them quite readily (conductors) and those that don't (insulators). Metals make up most of the conductors, while nonmetals such as plastics, wood, and glass are the insulators. In fact, things are far more complex than this—especially when it comes to certain elements in the middle of the periodic table (in groups 14 and 15), notably silicon and germanium. Normally insulators, these elements can be made to behave more like conductors if we add small quantities of impurities to them in a process known as doping. If you add antimony to silicon, you give it slightly more electrons than it would normally have—and the power to conduct electricity. Silicon "doped" that way is called n-type. Add boron instead of antimony and you remove some of silicon's electrons, leaving behind "holes" that work as "negative electrons," carrying a positive electric current in the opposite way. That kind of silicon is called p-type. Putting areas of n-type and p-type silicon side by side creates junctions where electrons behave in very interesting ways—and that's how we create electronic, semiconductor-based components like diodes, transistors, and memories.
Photo: A traditional printed circuit board (PCB) like this has tracks linking together the terminals (metal connecting legs) from different electronic components. Think of the tracks as "streets" making paths between "buildings" where useful things are done (the components themselves). There's a miniaturized version of a circuit board inside an integrated circuit: the tracks are created in microscopic form on the surface of a silicon wafer. This is the reverse side of the flash memory chip in our top photo.
Inside a chip plant
The process of making an integrated circuit starts off with a big single crystal of silicon, shaped like a long solid pipe, which is "salami sliced" into thin discs (about the dimensions of a compact disc) called wafers. The wafers are marked out into many identical square or rectangular areas, each of which will make up a single silicon chip (sometimes called a microchip). Thousands, millions, or billions of components are then created on each chip by doping different areas of the surface to turn them into n-type or p-type silicon. Doping is done by a variety of different processes. In one of them, known as sputtering, ions of the doping material are fired at the silicon wafer like bullets from a gun. Another process called vapor deposition involves introducing the doping material as a gas and letting it condense so the impurity atoms create a thin film on the surface of the silicon wafer. Molecular beam epitaxy is a much more precise form of deposition.
Photo: A silicon wafer. Photo by courtesy of NASA Glenn Research Center (NASA-GRC).
Of course, making integrated circuits that pack hundreds, millions, or billions of components onto a fingernail-sized chip of silicon is all a bit more complex and involved than it sounds. Imagine the havoc even a speck of dirt could cause when you're working at the microscopic (or sometimes even the nanoscopic) scale. That's why semiconductors are made in spotless laboratory environments called clean rooms, where the air is meticulously filtered and workers have to pass in and out through airlocks wearing all kinds of protective clothing.