Why is everything made of silicon?
A conductor cannot be switched off. An insulator cannot be switched on. Silicon sits in the goldilocks zone where a small applied voltage can flip the material between the two — and is plentiful enough that civilization can afford to use it for everything.
The choice of base material constrains every downstream decision. Silicon won not because it is best at any one thing but because its electrical, thermal, and economic properties together are tolerable enough that no competitor has reached escape velocity.
Conductors, insulators, and the thing in between
Electrical conduction depends on whether electrons in a material have anywhere to go. In a copper wire, the outermost electrons are loosely bound and drift freely under a voltage — current flows the moment you apply a field. In glass, the electrons are stuck. They sit in their atomic orbitals and require a violent input of energy before any of them can move.
A semiconductor is the material that sits between these two extremes. At room temperature, almost no current flows through pure silicon. But applying a small voltage, or shining the right wavelength of light on it, or introducing the right contaminant in trace amounts can release enough electrons to make the material conductive. Crucially, you can switch this on and off.
That switchable conductivity is the only thing computing has ever cared about. A material that conducts only when commanded is the physical substrate of a Boolean operation.
The band gap is the number that matters
Quantum mechanics groups the energy levels available to electrons into bands. The valence band is full of electrons that are locked into bonds; the conduction band is empty space those electrons could move through if they had enough energy. The gap between the two — the band gap — is the price of admission, measured in electron-volts.
Metals have no band gap; their conduction band overlaps with the valence band, so electrons move freely. Insulators have band gaps so wide that nothing realistic short of a lightning strike can promote an electron across them. Semiconductors have band gaps in the goldilocks range: 0.1 to 4 electron-volts, depending on the material.
Silicon's band gap is 1.12 eV. That number is the reason your laptop is not on fire and the reason it works at all.
Doping is the lever
Pure silicon is not very useful. The trick that made the semiconductor industry possible is doping: deliberately contaminating the crystal with parts-per-million quantities of another element. Add a few phosphorus atoms (five outer electrons each) into a sea of silicon atoms (four outer electrons each) and the extra electrons become loosely bound carriers. The material is now n-type — it carries current as electrons.
Add boron atoms (three outer electrons) instead and the missing electrons leave behind positively charged holes. The material is now p-type — it carries current as the absence of electrons, which behaves mathematically like a positive charge moving in the opposite direction.
Every modern device is built from juxtaposed regions of n-type and p-type silicon. The diode in chapter 2, the transistor that follows, the memory cell, the SRAM, the floating-gate flash cell — all of them are configurations of these two regions sitting next to each other.
Why silicon beat the alternatives
Silicon is not the best semiconductor at anything in particular. Germanium has higher electron mobility and was the original choice for transistors in the 1940s. Gallium arsenide is faster at high frequencies and is used in cellular radios. Silicon carbide and gallium nitride have wider band gaps and handle the high voltages needed in EV inverters and data center power supplies.
But silicon has three properties no competitor matches simultaneously. First, it is the second-most-abundant element in the earth's crust, which means the input cost is essentially zero. Second, when exposed to oxygen at the right temperature, it grows a native oxide (SiO₂) that is one of the best electrical insulators known — this single coincidence is what makes the MOSFET, the transistor that runs almost all modern computing, physically possible. Third, the crystal grows in giant, defect-poor cylinders cheaply and reliably, and the industry has had seventy years to refine every step of the manufacturing process around it.
A material is not chosen on technical merit. It is chosen on the integral of technical-merit-times-supply-chain-maturity over twenty years, and silicon won that integral decisively in the 1960s. The competing materials each have niches that exploit a single property silicon does badly, but none of them have a path to displacing silicon for general-purpose computing.
Strategic read
When someone announces a competing material — a carbon nanotube transistor, a 2D material like molybdenum disulfide, an exotic photonic compute substrate — the right question is not whether the new material is better at one specific property. It is whether the entire ecosystem of oxide chemistry, deposition equipment, lithography, doping tools, packaging, and trained workforce can be rebuilt around it within an investment horizon anyone is willing to fund.
It almost never can be. The substitution cost is the bottleneck, not the physics. Silicon will remain the substrate for general computing for as long as the alternatives lack the supply-chain maturity to match it, which is on the order of decades. The materials that succeed at the edges of silicon — SiC for power, GaN for high-frequency, GaAs for radio — succeed only because they own a workload silicon does badly enough to justify the cost premium of a parallel supply chain.