Diodes: the simplest semiconductor device

A diode is the simplest semiconductor device: a single PN junction wrapped in a package with two leads. The leads are named anode (the p-side) and cathode (the n-side, marked with a band on the body).

The symbol — a triangle pointing into a bar — is a one-way arrow for current. Current flows easily from anode to cathode when the diode is forward-biased; almost none flows the other way until breakdown.

Four facts every circuit using a diode depends on:

• It conducts only above ~0.7 V forward (silicon's knee voltage).
• The forward voltage is roughly constant once conducting (~0.7 V).
• Reverse current is tiny (microamps or less) until breakdown.
• Switching takes nonzero time — the reverse recovery problem in fast circuits.

Everything else about specific diode types is a tradeoff among these four.

The Shockley equation describes a diode's IV behavior:

$I = I_s \left(e^{V/n V_T} - 1\right)$

where $V_T \approx 26$ mV (thermal voltage at 300 K), $I_s$ is the saturation current, and $n$ is the ideality factor (1 for ideal diodes, 1.1–2 in practice).

Three regions matter:

• Reverse ($V < 0$): $I \approx -I_s$, tiny and roughly constant.
• Sub-threshold ($0 < V < 0.5$ V): $I$ rises exponentially but is still small.
• Forward ($V > 0.6$ V): the knee. Current rises sharply; voltage barely changes past this point.

In most practical circuits engineers approximate the curve as "0 V when off, exactly 0.7 V when on" — a piecewise model that gets you within 5% of reality and is much easier to reason about.

AC power flips polarity 50 or 60 times per second. Most electronics need DC. A diode (or four of them) converts AC to DC by chopping off the negative half-cycles — rectification.

Three shapes you'll see:

• Half-wave: one diode. Blocks negative half-cycles. Output is a series of positive bumps. Inefficient — half the power is thrown away.
• Center-tap full-wave: two diodes + a transformer with a center-tapped secondary. Both halves get used; output is twice as smooth.
• Bridge: four diodes in a square. Same result as center-tap but no special transformer needed. The default in nearly every consumer power supply.

The ripple coming out is still bumpy. A capacitor across the load smooths it: the cap charges on each pulse and discharges through the load between pulses. Bigger cap, smoother DC.

All variants start from one junction; the differences are in material, doping, and operating region.

A regular diode breaks down catastrophically in reverse. A Zener diode is engineered to break down cleanly at a chosen voltage (5.1 V, 12 V, 33 V — whatever the application needs).

The trick is heavy doping on both sides of the junction. This narrows the depletion region so much that two breakdown mechanisms dominate at low voltages:

• Zener tunneling (under ~5 V): electrons quantum-tunnel through the thin barrier.
• Avalanche breakdown (over ~5 V): the field is high enough to knock electrons off atoms in a chain reaction.

Both mechanisms produce a sharp knee in reverse. Once the Zener voltage is hit, the voltage across the diode stays roughly constant regardless of current. That makes Zeners the dirt-cheap voltage reference: feed any DC voltage above the Zener voltage through a current-limiting resistor and take the voltage across the Zener as your reference.

A Schottky diode uses a metal-semiconductor junction instead of a PN junction. The metal-Si interface has a built-in voltage of only ~0.2–0.4 V instead of 0.7 V, and there's no minority-carrier storage so switching is extremely fast (picoseconds versus nanoseconds for silicon PN diodes).

Why this matters:

• Low forward drop → less power wasted in rectifiers. 5 A through 0.3 V is 1.5 W; through 0.7 V it would be 3.5 W. Big deal in power supplies.
• Fast switching → usable at MHz to GHz frequencies. Switching power supplies and RF circuits both rely on this.

Trade-offs:

• Higher reverse leakage than a PN diode (~µA instead of nA).
• Lower reverse breakdown voltage (usually 20–100 V max).

Use a Schottky when you need speed or efficiency. Use a regular diode when you need a wide voltage rating.

In a forward-biased PN junction, electrons and holes recombine in the depletion region. The energy they release equals the band gap $E_g$. In silicon that energy turns into heat. In direct-band-gap semiconductors — GaAs, GaN, InP — it turns into a photon.

The photon's wavelength is set by the band gap:

$\lambda = \frac{hc}{E_g}$

• Red LEDs: GaAsP, $E_g \approx 1.9$ eV → 650 nm.
• Green: GaP, ~2.3 eV → 530 nm.
• Blue: GaN, ~2.7 eV → 460 nm (the Nobel Prize-winning invention that unlocked white LEDs).

Run the junction backward — shine a photon at it — and the photon excites an electron-hole pair, generating a tiny current. That's a photodiode. Same physics, different direction. Photodiodes are the eye of every camera sensor, fiber-optic receiver, and barcode scanner.

Real diodes break in ways the textbook IV curve doesn't show:

• Thermal runaway — forward voltage drops about 2 mV per °C. As the diode heats, it conducts more, heats more. Parallel diodes need ballast resistors or matched thermal coupling.
• Reverse recovery — when you switch a silicon PN diode from forward to reverse, it conducts in reverse briefly while the depletion region rebuilds. Fast circuits use Schottky or fast-recovery diodes to skip this.
• Reverse leakage — small at room temperature, but doubles every 10 °C. Matters for precision photodiodes and low-current circuits at high temperature.
• Surge current — most diodes handle 5–10× rated current for a few milliseconds (start-up inrush) but not steady-state. Read the datasheet's $I_{FSM}$ rating.
• Avalanche destruction — unlike Zeners, normal diodes are destroyed by reverse breakdown.

Most diode failures in real designs are one of these five.