UNIT 1 • WHAT IS A CHIP?

One of the hardest things people make, grown from sand

Every phone, car, and data center runs on chips. This course follows a chip from raw sand to a packaged part.

🏭

Unit 1 covers the first half of the journey: turning quartz sand into the flawless silicon wafers that fabs build on. Before a single transistor exists, the material has to be purer, flatter, and more perfect than almost anything else industry produces.

Key idea A chip starts long before the fab: as sand, then as an ultra-pure crystal, then as a polished wafer.

THE VOCABULARY

Wafer, die, package, transistor

💿

Wafer

A thin, polished disc of silicon, usually 300 mm across, that carries hundreds of chips

🔲

Die

One rectangular chip cut from a wafer, often about a centimeter on a side

📦

Package

The housing that protects a die and connects it to a circuit board

🔀

Transistor

The tiny switch that everything is built from; one die holds billions

People say 'chip' for both the bare die and the packaged part.

ZOOMING IN

From wafer to transistor

One wafer holds hundreds of identical dies. Each die holds billions of transistors, whose features are only tens of nanometers across.

The whole industry is about making the same tiny structures, billions of times, without mistakes.

HOW SMALL IS SMALL?

A scale ladder

Wafer About 30 cm across, roughly a large dinner plate

Die Around 1 cm on a side, about the size of a fingernail

Human hair About 70 µm wide, thousands of times wider than a transistor feature

Virus Around 100 nm, larger than many transistor features

Transistor features Tens of nanometers

From wafer to transistor feature, the size range spans about seven orders of magnitude.

NODE NAMES

What '3 nm' really means

'3 nm' ≠ 3 nm

Node names like 7 nm, 5 nm, and 3 nm began as gate lengths, but they stopped measuring any single feature years ago. Today they are generation labels, so two companies' '5 nm' chips are not identical.

Treat a node name as a marketing generation, not a ruler.

MOORE'S OBSERVATION

Why chips keep shrinking

2× per ~2 years

In 1965 Gordon Moore noticed that the number of components on a chip was doubling regularly, and in 1975 he revised the pace to about every two years. Shrinking features packed more transistors onto a chip, and that drove decades of progress.

Fabs exist to make the same structures smaller, cheaper, and more reliably.

WHAT A FAB DOES

Four verbs, repeated hundreds of times

Add Deposit or grow a thin film on the wafer

Pattern Use light to define where material goes

Remove Etch away what you don't need

Modify Change a material's properties with dopants or heat

Repeat that loop over dozens of layers and you have a chip.

UNIT 1 STUDY COMPLETE

💿

Ready for the Fab Challenge?

You can name the parts of a chip and explain what a fab does with them.