Doping, diodes & transistors

Pure silicon is a mediocre conductor

On its own, silicon is a bit of a letdown. Every atom uses all four of its outer electrons to bond with four neighbours, so the lattice is a party where every guest's hands are full — nothing is free to move, and the crystal barely conducts. Useful as a starting point, useless as a device.

The industry's masterstroke is doping: deliberately sprinkling in a trace of a foreign element — roughly one atom in a million — chosen for having exactly one bonding electron too many, or one too few. That tiny, precisely placed impurity is what turns inert silicon into the working material of every chip.

Two flavours: N-type and P-type

Pick a dopant one column to the right of silicon in the periodic table — phosphorus — and it arrives with five bonding electrons. Four slot neatly into the lattice; the fifth has no bond to join, so it roams free. That spare, mobile electron is a negative carrier, and silicon doped this way is N-type (N for negative).

Pick a dopant one column to the left — boron — and it brings only three bonding electrons. That leaves a vacancy in the bonding network, a hole. Neighbouring electrons shuffle in to fill it, which makes the hole itself appear to drift through the crystal like a positive charge. Silicon doped this way is P-type (P for positive). Toggle between them below.

Two points worth holding onto. First, neither material is actually charged overall — the names describe who does the moving, not a net charge. Second, the reason a sprinkle of dopant matters so much is that pure silicon starts with almost no free carriers, so even a few added ones dominate. The idea of the free electron and the hole comes straight from the band gap: a carrier above, a hole below.

Press them together: the junction

Doping's real power shows up when you put the two flavours side by side. Press a slab of N against a slab of P and something remarkable happens at the boundary. Free electrons from the N side and holes from the P side meet and cancel near the join, leaving a thin depletion region that acts like a built-in gate.

The result is a diode: current can cross the junction easily in one direction but is blocked in the other — a one-way valve for electricity. Select Diode in the figure above to watch current pass one way only. Diodes alone are everywhere — in chargers, radios, and every device that converts alternating current to direct — but they're also the stepping stone to something bigger.

Sandwich them: the transistor

Now make a sandwich: N-P-N, two outer regions of one flavour around a thin middle layer of the other. That thin middle layer becomes a control gate. Apply a small voltage to it and it opens a conducting channel between the two outer regions, letting current flow. Remove the voltage and the channel closes. Select Transistor above and toggle the gate to feel it.

That controllable sandwich is a transistor — the exact switch that all of computing rests on. It is the physical realisation of the door: a small control signal decides whether current flows or not, flipping between on and off billions of times a second. Doping is what makes it buildable, because it gives engineers the two flavours of silicon the sandwich requires.

Why this scales to billions

The last piece is that doping is patternable. Using lithography, engineers define exactly where the N and P regions go, down to a few atoms wide, and repeat the pattern across a wafer. A modern chip is billions of these doped sandwiches — transistors — wired together with fine metal lines. The same trick that made one switch makes a trillion.

Step back and the whole industry snaps into focus: refine sand into flawless silicon, dope it into N and P, stack those into transistors, wire the transistors into logic, and wire the logic into chips. Every layer rests on the one below — and doping is the layer that turns a passive crystal into an active switch.