What is a semiconductor, and why is it called "semi"?

A semiconductor is a material that conducts electricity better than an insulator but worse than a conductor, and the most common one is silicon. That in-between is what "semi" means: it doesn't conduct as easily as a metal, and it doesn't block electricity almost completely the way glass or plastic does. But what makes it so important isn't "conducting halfway". It's that whether it conducts is something people can decide. The chips in phones and computers are built on exactly that.

What's the difference between a conductor, an insulator and a semiconductor?

Think of electricity as water in a pipe:

• A conductor is a pipe that's always open. Push, and the current flows through. The copper inside a power cord is a conductor.
• An insulator is a pipe that's blocked solid. However hard you push, nothing gets through. The plastic around the cord, glass and ceramics are insulators.
• A semiconductor is a pipe with a gate in it. Normally the gate is almost shut and only a trickle gets through. But it can be opened, and you can set how far.

The "water" is really electrons. A metal is full of electrons that are free to move, so they flow at the slightest push. In an insulator, every electron is held tight by its atom and can't move. A pure semiconductor is in between: most of its electrons are held, and only a very few are free.

So why is it called a "semi"-conductor?

The name comes from how well it conducts: "semi" means half or partly, and the word says its conduction falls between a conductor's and an insulator's. So "semi" doesn't mean it conducts at half strength, or that it's on half the time. The useful part is that people can decide where that gate opens and how far.

The most common semiconductor is silicon. Why?

Silicon is an element, the second most abundant in the Earth's crust after oxygen. The main ingredient of sand and quartz, silicon dioxide, is silicon combined with oxygen. Chips use silicon for a few main reasons:

• There's plenty of it: minerals that contain silicon are everywhere.
• It grows its own good insulator: heat silicon with oxygen and its surface grows an even layer of silicon dioxide, an excellent insulator. Chips need "this part conducts, the part next to it doesn't" all over, and silicon can grow that layer right in a furnace. This is one of its biggest advantages.
• It copes with heat: the very first transistor was made of another semiconductor, germanium. Germanium leaks current more easily as it warms up, and its oxide is unstable, so silicon replaced it.
• It grows into large, orderly crystals: silicon with its atoms lined up perfectly can be grown into a big ingot and sliced into wafers (Why are wafers round?).

Doping: add a few other atoms and conduction changes completely

The silicon for chips is first purified to an extreme: metal impurities have to come down to around one part in a billion. Then a tiny amount of a chosen element is added on purpose. This step is called doping.

The amounts are very small. At the low end, it's about one added atom for every hundred million silicon atoms; at the high end, about one in ten thousand, and in some places more. Yet that little bit can make silicon conduct thousands or tens of thousands of times better than pure silicon, and with more added, over a million times better.

Why such a big difference? Think of hands. Each silicon atom has four "hands" (its four outermost electrons), and each hand holds one neighboring silicon atom, so the atoms form a neat grid. With every hand busy holding on, the electrons can't move, and that's why pure silicon barely conducts.

• Add phosphorus, get n-type: phosphorus has five hands. It takes the place of a silicon atom, holds its four neighbors, and still has one hand left over. That extra electron has nothing to hold, so it's free to roam. What moves is the electron, which carries a negative charge, so this is n-type. Arsenic also has five hands and works the same way.
• Add boron, get p-type: boron has only three hands, so one neighbor's hand has nothing to hold. That leaves an empty spot (textbooks call it a "hole"). A nearby electron hops over to fill it, which moves the empty spot somewhere else, like one empty seat in a row: everyone shifts over by one, and the gap travels down the row. A spot that's missing an electron acts like a positive charge, so this is p-type.

The silicon as a whole isn't charged; n and p only say whether what moves is an electron or an empty spot. In one line: the kind of atom you add decides n-type or p-type; how much you add decides how well it conducts.

How does a fab put the atoms in?

These days it's mostly done by ion implantation: atoms such as boron and phosphorus are turned into charged ions, sped up by an electric field and shot straight into the silicon. How many go in sets how well it conducts; how fast they go sets how deep.

Places that shouldn't be implanted are covered first with photoresist, and the resist's pattern is printed by litho. That's how each tiny n-type or p-type region can be drawn so precisely. After implant, the wafer is heated once more (an anneal) to repair the disturbed crystal, and only then do the atoms settle into place and start to conduct. Atoms that went in by mistake can't be taken back out, so the thing this step fears most is a wrong setting.

Why does controlling conduction matter so much?

A conductor is always on and an insulator is always off, so neither can be a switch. A semiconductor can be on in one tiny region and off in the next, and it can be switched with electricity.

Put tiny n-type and p-type regions side by side and add an electrode to control them, and you have a transistor: a small switch controlled by electricity, which switches between on and off when the voltage on that electrode changes, with on as 1 and off as 0 (how it works: What is a transistor?).

Before transistors, electronic switches were vacuum tubes: each about the size of a small light bulb, hot, and prone to burning out. Switches made from semiconductors are tiny, use little power, and can switch over a hundred million times a second. One chip can hold hundreds of millions, even tens of billions of them; taking a photo, finding a route or running AI on your phone all comes down to them switching on and off.

What does the news mean by "semiconductors"?

Often not the material, but the whole industry that makes chips: designing them, making them in a fab, then assembly & test, all the way to delivery. Put simply: a semiconductor is a material, and also the name of that industry; a chip is a thing made from semiconductors.

In fab24hr

In fab24hr, the production fab's implant area has an ion implanter, IMP-01. As the implant engineer (production fab › process › implant), you check the sheet resistance of each monitor wafer to see whether the right amount went in; lots that are done get lined up for an anneal in the furnace; and on the evening shift you check three lots' recipes line by line: boron, phosphorus and arsenic, energy and dose. The company, people, tool numbers, lot numbers and figures in the game are fictional; what a semiconductor is and how doping works follow public, general sources. For other jargon, see Chip fab jargon in plain English.