Lesson 4 — From Sand to Transistor
How Do Chips Actually Work?
Learning Material
1 pagesLesson 4 — From Sand to Transistor
Understanding the Complex: How Do Chips Actually Work?
Anchor question: How does pure silicon become a switch that operates at nanometer scale?
The silicon wafer sitting at the beginning of the chip-making process is beautiful in an industrial way: a perfect disc, 300 millimeters across, mirror-smooth, shimmering with an iridescent sheen. Under magnification, its surface is a near-perfect crystal lattice — silicon atoms arranged in a regular, repeating diamond cubic structure, with almost no defects.
But a perfect silicon crystal is not particularly useful. What makes silicon useful is not purity alone but the precise, controlled introduction of impurity.
This seems paradoxical. We spent enormous effort achieving purity, only to ruin it deliberately. But the impurity isn't random — it's carefully chosen, precisely located, and it's what turns inert silicon into a transistor.
Doping: turning silicon into a semiconductor
In its pure form, silicon has four electrons in its outer shell, and it forms bonds with four neighboring atoms. This leaves no electrons free to carry electric current — pure silicon is essentially an insulator at room temperature.
Doping changes this. By introducing tiny amounts of other elements — measured in parts per billion — engineers can alter silicon's electrical properties dramatically.
The two key dopants are phosphorus (which has five outer electrons) and boron (which has three). When phosphorus atoms replace silicon atoms in the crystal lattice, they bring an extra electron that doesn't fit neatly into the bonding structure. This extra electron is loosely held and free to move — it can carry charge. Silicon doped with phosphorus is called n-type (negative), because the mobile charge carriers are electrons.
Boron does the opposite. With only three outer electrons, a boron atom creates a "hole" where an electron ought to be. This hole can move through the crystal as neighboring electrons jump to fill it, effectively moving the absence of an electron in the opposite direction. Silicon doped with boron is p-type (positive), because the effective charge carriers are these positively charged holes.
The p-n junction: where the transistor lives
When n-type and p-type silicon are brought together, something interesting happens at their boundary. Electrons from the n-type region diffuse across to fill holes in the p-type region, creating a zone — the depletion region — that is depleted of mobile charge carriers. This creates an electric field that resists further diffusion, and the system settles into equilibrium.
This junction has a remarkable property: it allows current to flow easily in one direction (forward bias) but strongly resists it in the other (reverse bias). This is the p-n junction, and it's the fundamental building block of virtually all semiconductor devices, from the simplest diode to the most complex processor.
A transistor is, in essence, two p-n junctions arranged back to back, with a thin region between them controlled by a third electrode. The classic MOSFET (Metal-Oxide-Semiconductor Field-Effect Transistor) places a thin layer of insulating oxide on top of the silicon, with a metal or polysilicon gate electrode on top. When a voltage is applied to the gate, it creates an electric field that attracts charge carriers into the channel below — effectively turning on the flow of current from source to drain.
Remove the gate voltage, and the channel depletes again. The transistor turns off.
The scale problem — and how it was solved
For decades, transistors were made larger than they needed to be, because smaller was harder to make. The pressure of Moore's Law drove the industry to miniaturize relentlessly.
The scale of modern transistors defies easy visualization. Consider: a 3-nanometer transistor has a gate length of 3nm. That's about 15 silicon atoms across. The gate oxide — the insulating layer that separates the gate from the channel — is just a few atoms thick. At this scale, quantum effects become significant: electrons can tunnel through barriers that classical physics would call impenetrable.
This is why, beyond a certain size, the traditional planar transistor stopped working well. The gate couldn't adequately control the channel, current would leak even when the transistor was "off," and devices wasted power. The solution, introduced by Intel in the early 2010s and now universal, was the FinFET — a transistor shaped like a fin protruding from the surface, allowing the gate to wrap around three sides of the channel rather than just one. This dramatically improved control.
The next evolution, now entering production, is the Gate-All-Around (GAA) transistor, where the gate wraps completely around a nanowire or nanosheet channel. Samsung introduced GAA transistors in their 3nm process in 2022; TSMC is expected to follow suit at 2nm.
From transistor to circuit
A single transistor is a switch. But the power of chips comes from combining billions of them into circuits that perform logic, store data, and shuttle information at incredible speed.
The basic logic gates — AND, OR, NOT — are built from two to six transistors each. From these gates, more complex circuits emerge: adders, multiplexers, memory cells, clocks. A modern processor core contains many millions of transistors, organized into functional blocks — arithmetic units, cache memory, control logic — that together execute billions of instructions per second.
All of this emerges from the controlled introduction of impurities into silicon, and from the physics of the p-n junction.
What makes it possible to build billions of these junctions on a chip the size of a thumbnail? That's the subject of the next lesson.
Next lesson: The Secret of Lithography — how the transistor pattern is printed onto silicon using light, and why extreme ultraviolet changed everything.
Reading time: approx. 9–10 minutes