How a semiconductor actually works

A copper wire conducts electricity because its electrons are free to move. Glass does not conduct because its electrons are locked in place. Silicon sits between the two, and that is not a defect — it is the entire point. A material that conducts only under conditions you control is a material you can use to build a switch, and a switch you can build a billion times over on a chip the size of a fingernail is the foundation of every computer, phone and controller in use today.

Bands, gaps, and why silicon is interesting

Electrons in a solid cannot take just any energy; they occupy allowed ranges called energy bands. The band filled with the electrons that hold atoms together is the valence band. Above it sits the conduction band, where an electron is free to roam and carry current. Between them lies the band gap — an energy range no electron can occupy.

The size of that gap decides everything. In a metal the bands overlap, so electrons flow freely and the material conducts. In an insulator the gap is enormous and no realistic amount of energy pushes an electron across. In silicon the gap is about 1.1 electronvolts: large enough that pure silicon barely conducts at room temperature, small enough that heat, light or an applied voltage can lift electrons across it.

When an electron does jump to the conduction band it leaves behind a vacancy in the valence band. That vacancy, called a hole, behaves like a positive charge carrier as neighbouring electrons shuffle to fill it. Semiconductors therefore carry current two ways at once — negative electrons moving one direction, positive holes moving the other.

Doping: making silicon deliberately impure

Pure silicon is not much use. The breakthrough is doping — adding a trace of a different element, often just one atom in ten million.

• Add phosphorus, which has five outer electrons where silicon has four, and the extra electron is left over with nowhere to bond. It sits just below the conduction band and escapes easily. The result is n-type silicon, where electrons are the main carriers.
• Add boron, which has three outer electrons, and a bond is left incomplete. That missing electron is a hole, ready to accept an electron from a neighbour. The result is p-type silicon, where holes dominate.

Both types are electrically neutral overall — doping does not add charge, it adds mobile charge. What matters is that an engineer now controls, region by region, which kind of carrier a piece of silicon uses.

The p-n junction: a one-way street

Put n-type silicon directly against p-type silicon and something immediate happens. Electrons from the n-side diffuse across into the p-side and fill holes; holes diffuse the other way. Near the boundary the mobile carriers annihilate each other, leaving a depletion region — a thin zone stripped of carriers, with fixed charged atoms on either side that create a built-in electric field.

That field opposes further diffusion, and equilibrium is reached. Now apply an external voltage:

• Forward bias (positive to the p-side) pushes carriers toward the junction, shrinks the depletion region, and current flows easily.
• Reverse bias pulls carriers away, widens the depletion region, and almost no current flows.

A p-n junction is therefore a diode — a component that passes current in one direction and blocks it in the other. Rectifiers, solar cells and LEDs are all p-n junctions with different priorities: a solar cell uses incoming light to create carriers that the built-in field sweeps apart, and an LED runs the process backwards, with electrons and holes recombining across the junction and releasing the energy as photons.

A transistor is not a component that amplifies. It is a valve, and amplification is what a valve does when a small movement controls a large flow.

The MOSFET: the switch everything is built from

Almost every transistor in a modern chip is a MOSFET — metal-oxide-semiconductor field-effect transistor. Its structure is simple: two doped regions called source and drain sit in a substrate of the opposite type, and between them lies a gate electrode separated from the silicon by an ultra-thin insulating layer of oxide.

With no gate voltage, the source and drain are separated by silicon of the wrong type, and no current flows. Apply a voltage to the gate and its electric field reaches through the oxide and pulls minority carriers to the surface, forming a thin channel of the right type connecting source to drain. Current now flows. Remove the gate voltage and the channel disappears.

The gate draws essentially no current — it is insulated — so switching costs very little energy. That efficiency is why MOSFETs, and not earlier transistor designs, made dense digital logic practical. Combine an n-channel and a p-channel device so that one is always off, and you have CMOS, which consumes power mainly when it switches rather than continuously.

From that single switch everything else follows. A few transistors make a logic gate; a few gates make an adder or a memory cell; hundreds of millions make a processor core.

From sand to chip

Fabrication turns these ideas into hardware through repetition. A cylinder of ultra-pure single-crystal silicon is sliced into wafers and polished flat. Then, layer by layer, the pattern of a circuit is built up: an oxide is grown, a light-sensitive photoresist is applied, and photolithography projects the circuit pattern onto it. Exposed resist is developed away, the underlying material is etched, and dopants are driven in by ion implantation. Metal layers are deposited to wire the devices together.

This cycle repeats dozens of times, each layer aligned to the one beneath it within nanometres. A finished wafer is tested, cut into individual dies, and packaged. The relentless shrinking of feature sizes — the trend behind Moore's law — is what allowed transistor counts to climb from thousands to tens of billions, though it now runs into physical limits that push designers toward new device structures and three-dimensional stacking.

Why it matters for students and researchers

Semiconductor physics is where quantum mechanics stops being abstract and starts paying salaries. It links solid-state theory, materials chemistry, device engineering and manufacturing process control in one continuous chain, and a student who understands the band gap also understands why a solar cell has a theoretical efficiency limit and why a chip gets hot. The field is also in visible flux: wide-band-gap materials such as silicon carbide and gallium nitride are displacing silicon in power electronics, two-dimensional materials are under active investigation for channels a few atoms thick, and semiconductor supply chains have become a matter of national policy. Following the peer-reviewed literature is how researchers and engineering students keep up with device physics, fabrication advances and reliability work that textbooks cannot refresh quickly enough.

Frequently asked questions

What is a semiconductor?

A semiconductor is a material whose electrical conductivity lies between that of a conductor and an insulator, and which can be controlled by adding impurities, applying a voltage, or changing temperature or illumination. Silicon is the most widely used, with germanium, gallium arsenide, silicon carbide and gallium nitride serving specialised roles.

What does doping do to silicon?

Doping adds a trace of another element to supply mobile charge carriers. Phosphorus or arsenic donates spare electrons and creates n-type silicon; boron or gallium creates vacancies called holes and produces p-type silicon. The material stays electrically neutral — doping controls which carrier type dominates, not the net charge.

How does a transistor switch current?

In a MOSFET, a voltage on the insulated gate creates an electric field that pulls carriers to the silicon surface, forming a conducting channel between source and drain. With the channel present, current flows; without it, the device blocks. Because the gate is insulated, controlling the switch costs almost no current.

Why is the band gap important?

The band gap sets how much energy an electron needs to become mobile. It determines whether a material behaves as a semiconductor at all, what wavelengths of light it can absorb or emit, how much leakage current a device has at high temperature, and how much voltage it can withstand — which is why different applications call for different semiconductor materials.