Play

← fab24hr Blog

Chip news explained4 min read

The news says "2 nm". What exactly is 2 nm?

The news says things like "2 nm enters mass production" and "a 3 nm phone chip". It sounds as if something on the chip is 2 nm across. In fact the number doesn't measure anything: it is the name of a process generation. To see why, start with the smallest part on a chip.

What is a chip actually doing?

The most basic part on a chip is the transistor. It is a tiny switch controlled by electricity: on is 1, off is 0. Taking a photo, finding a route, running AI: in the end, it is all a huge number of switches turning on and off. One phone chip now holds more than ten billion of them (this article explains transistors in more detail).

The smaller the switch, the more of them fit in the same space, and the less power each one needs to flip. So for decades, people have kept finding ways to make it smaller. When the news says "so many nanometers", it is saying which generation of that effort we have reached.

What did "nanometer" originally measure?

A nanometer is a billionth of a meter. A human hair is tens of thousands of nanometers thick.

Long ago, until roughly the 1990s, a process's name more or less matched a real size on the transistor, such as how narrow the finest line of that generation could be. Back then, the name was the size.

So why did they stop matching?

The goal of each generation is roughly to fit twice as many switches in the same space. Shrink the length and the width each to 0.7 times, and the area becomes 0.7 × 0.7, about half. So each new name was simply the last one times 0.7:

90 → 65 → 45 → 32 → 22 → 14 → 10 → 7 → 5 → 3 → 2

But at some point, transistors could no longer get better just by shrinking everything in proportion. Engineers switched to other tricks: they turned the transistor from a flat slab into a thin fin standing upright (like replacing a bungalow with a tall building), then wrapped the gate around the channel the current flows through on all four sides; they changed materials; they packed the wiring tighter. The overall result is still "more in the same space each generation", but no single dimension actually shrank to the number in the name.

So the name now only means "which generation this is". For some generations, different makers picked slightly different names, and the same number does not always mean the same size from one maker to another, because there is no common standard.

So how small is it really?

One public international semiconductor roadmap estimates that in the "3 nm" generation, the spacing between transistors in a row is in the forties of nanometers, and the tightest metal wiring is spaced in the twenties. The 2 nm generation is a little smaller, but still at the scale of tens of nanometers.

nm100k10k1,000100101about ten times apartA hair: tens of thousandsA bacterium: about 1,000A virus: about 100Transistor spacing: 40-someTightest metal spacing: 20-some"2 nm": just a name
Each step is ten times. The two spacings on the chip follow the public estimate for the "3 nm" generation; the 2 in the name is ten times smaller than the tightest spacing on the chip.

These are all much bigger than the number in the name, but still more than a thousand times thinner than a hair. Some films on a chip really are only a few nanometers thick, but the name isn't measuring them either.

What does a smaller number actually get you?

A chip the same size with more switches can do more, such as running AI on a phone; or it can do the same work on a smaller chip that uses less power. For you, that means faster, longer battery life, and a phone that runs cooler.

The price is that chips get harder and more expensive to make. To print finer patterns, the scanner has to use light with a shorter wavelength; the most advanced is extreme ultraviolet (its wavelength is only 13.5 nm, and even air absorbs it, so the whole light path has to be in a vacuum). A chip also goes through more and more steps, and each step can go wrong, so when a new generation starts, the share of chips that come out good (the yield) is often lower.

That is why not every chip needs the newest generation. Many chips in cars, home appliances and chargers are made the way it was done several generations ago: good enough, reliable and cheap. Those are what the news calls "mature nodes".

Next time you see it in the news

  • "2 nm enters mass production": the newest generation's way of making chips has started large-scale production. It does not mean someone made a 2 nm object.
  • "A 3 nm phone chip": this chip was made the "3 nm" generation's way.
  • "Expanding mature-node capacity": opening more lines that make chips the way of older generations, the kind used in cars and appliances.

In fab24hr

The fab in the game is fictional, and the game never says what node it runs. The exposure you see in litho and the line width checks you see in etch are the things behind the name that really do have to shrink, and someone looks after them every day. The sizes in this article follow public, general sources; where there is no single agreed number, it gives only the scale. You can look up other fab jargon in "Chip fab jargon in plain English".

Where you'll meet it in the gameIn litho, the scanner prints the pattern onto the wafer; in etch, someone measures line width every day. The things that really have to shrink are all looked after in the game.

In fab24hr, the things behind the name that really do have to shrink are looked after every day in litho and etch.

Play as the litho engineer and guard every layer's pattern

More articles

More on this site