Semiconductors, the PN Junction and Diodes

Everything active in your station — the rectifier in the power supply, the mixer in the receiver, the transistor in the transmitter — rests on one idea: a crystal that is neither a conductor nor an insulator, persuaded by a trace of impurity to conduct in one direction and not the other. Section A returns to this material every year, and the questions are rarely about theory for its own sake. They ask what the depletion layer is, what 0.7 V means, which diode does which job. Learn the mechanism once and the answers stop being a list to memorise.

Conductors, insulators and semiconductors

What separates the three is the number of electrons in the outermost shell — the valence shell — of each atom.

Silicon and germanium sit exactly on the fence with four valence electrons each, and that single number is the most-asked fact in this lesson.

The pure crystal: covalent bonds and electron-hole pairs

In a crystal of pure silicon every atom shares each of its four valence electrons with one of its four neighbours. Each shared pair is a covalent bond, and with all bonds complete there are no loose electrons to carry a current. Silicon with nothing added to it is called intrinsic — pure — and at the absolute zero of temperature an intrinsic semiconductor is an insulator.

Warm it and thermal energy starts breaking bonds. Each broken bond releases a free electron and leaves behind a vacancy in the bond: a hole. The two are created together and are always counted together as an electron-hole pair. A neighbouring electron can hop into the hole, which fills that bond but opens a new hole where the electron came from. The electrons move one way; the hole appears to move the other way, and it behaves in every circuit calculation like a positive charge carrier. In intrinsic material the two are always equal in number.

Why a semiconductor has a negative temperature coefficient

This is the payoff. Heat a copper wire and its resistance rises: the number of carriers does not change, but the lattice vibrates harder and gets in their way. That is a positive temperature coefficient. Heat a semiconductor and heat manufactures carriers — more broken bonds, more electron-hole pairs — and the extra carriers swamp the extra scattering. Its resistance therefore falls as it warms. A semiconductor has a negative temperature coefficient of resistance, and so does carbon. This is not a curiosity: it is why a hot transistor passes more current and why the next lesson spends time on thermal runaway.

Doping: giving the material a preferred carrier

Intrinsic silicon is nearly useless as it stands — its conductivity depends on how warm the room is. Adding a controlled trace of impurity, about one atom in ten million, fixes the conductivity at a value you choose. That is doping, and the doped material is called extrinsic. The property being controlled is conductivity, which is the wording the examiner uses.

Two things about that table trip people up. First, the names are the opposite way round from what the letters suggest: a pentavalent impurity with five electrons gives N-type, and a trivalent impurity with three gives P-type. Second — and this is the favourite trap — both types are electrically neutral overall. N-type is not negatively charged. Every donor atom brought its own proton along with its spare electron, so the sample as a whole has exactly as much positive charge as negative. “N-type” describes which carrier does the conducting, not a net charge.

Heat still generates electron-hole pairs in doped material, so N-type silicon contains a few holes and P-type a few electrons. These are the minority carriers, and although they are vastly outnumbered they matter: they are the whole explanation of reverse leakage current a few paragraphs down.

The PN junction

Grow a single crystal with P-type material on one side and N-type on the other and something happens at the boundary with no battery connected at all. Free electrons on the N side diffuse across into the P side, where they are plentiful and holes are waiting; holes diffuse the other way. Wherever an electron meets a hole they recombine, and both vanish as carriers.

What is left behind is the important part. The donor atom that lost its electron is now a fixed positive ion, locked into the crystal lattice. The acceptor atom that gained one is a fixed negative ion. Neither can move. So a thin region either side of the junction ends up containing charged ions and no free carriers at all — it has been depleted of them. That is the depletion layer, also called the depletion region or the barrier.

Those fixed ions face each other across the gap, so there is a voltage across the depletion layer. It is called the barrier potential, and it is what stops any further diffusion — the process shuts itself down. Its size depends on the material:

• Silicon: about 0.7 V
• Germanium: about 0.3 V

The barrier potential falls slightly as the junction warms, by roughly 2 mV for every degree Celsius. The examiner asks this directly: for a forward-biased diode, the barrier potential decreases as temperature increases.

Forward bias and reverse bias

The P side is the anode, the N side the cathode. Learn the polarities as words, not as a picture that you may draw the wrong way up:

Conventional current in a diode flows from anode to cathode — in the direction the triangle in the symbol points. On the component itself, the printed band marks the cathode. That band is the only thing standing between you and a rectifier fitted backwards.

The characteristic curve and the knee

Plot current against applied voltage and you get the curve the exam calls the I-V characteristic. Below the barrier potential almost nothing flows. At the barrier potential the curve bends sharply upward — that bend is the knee, sometimes called the cut-in point — and beyond it the current climbs almost vertically for very little extra voltage. A diode is therefore non-ohmic: it has no single resistance, and Ohm's law does not describe it.

Use the explorer below to see that in numbers rather than in words. Drag the voltage marker along the forward curve and watch the calculated V ÷ I figure collapse from megohms to a few ohms across a couple of hundred millivolts, then switch the material tab from silicon to germanium and see the knee move from 0.7 V to 0.3 V.

Reverse leakage, breakdown and PIV

Reverse bias does not stop the current completely. Minority carriers — the few holes in the N material and the few electrons in the P material, generated by heat — find the reverse field pushing them the helpful way, and they cross. The result is the reverse saturation current, usually called leakage. It is a matter of microamps or less in silicon, it is almost independent of the applied voltage, and it roughly doubles for every 10 °C rise, which is precisely what you would expect from something made by heat.

Increase the reverse voltage far enough and the junction gives way altogether. In a lightly doped junction the mechanism is avalanche breakdown: carriers are accelerated hard enough to knock others out of their bonds, which knock out more again. In a heavily doped junction the depletion layer is so thin that the field rips electrons straight out of their bonds — zener breakdown — and that happens at a much lower voltage.

The rating that tells you how much reverse voltage a diode will stand is its PIV — Peak Inverse Voltage. Exceed it in an ordinary rectifier and the diode is destroyed, so the PIV rating must sit comfortably above the largest reverse voltage the circuit can apply. In a mains power supply that is a good deal higher than the DC output voltage, as Rectifiers, Smoothing and Regulation works through.

The diode family, and what each one is for

Two of those repay a second sentence. The varactor works because reverse bias widens the depletion layer, and a wider carrier-free gap between two conducting regions is exactly a capacitor with its plates moved further apart — so more reverse voltage means less capacitance. Put one across a tank circuit and a DC control voltage sets the frequency, which is how nearly every modern rig tunes and how AFC pulls a receiver back onto an FM signal.

The Schottky is fast because its junction is metal-to-semiconductor rather than P-to-N. There are no minority carriers stored in it, so there is nothing to sweep out when the voltage reverses and it stops conducting the instant it is asked to.

Finally, the plain diode itself does more than rectify. It is a switch, it is a detector or demodulator — an AM detector is a diode plus an RC filter, recovering the audio envelope — and it is a clipper. When a paper offers “rectifier / switch / detector / all of the above”, the answer is all of the above. Note also that a diode is a passive device: it has no gain and needs no supply of its own.

Worked examples

Example 1 — a series resistor for an LED

You want to run a panel LED at 15 mA from the 13.8 V bench supply. The LED drops about 2 V. What series resistor is needed?

R = (13.8 − 2) ÷ 0.015 = 11.8 ÷ 0.015 = 787 Ω. Fit the nearest standard value above it, 820 Ω, and check the dissipation: P = I²R = 0.015² × 820 = 0.18 W, so a quarter-watt resistor will do.

Example 2 — zeners in series

Two identical zener diodes in series break down at 12 V in total. Each one must therefore break down at 12 ÷ 2 = 6 V. Zener voltages add in series, which is the standard way of making a reference the drawer does not contain — three 6.2 V zeners in series give 18.6 V.

Example 3 — forward drop in a bridge

A silicon bridge rectifier passes current through two diodes in series on every half cycle, so the total forward drop is 2 × 0.7 = 1.4 V. Feed it 12 V peak and the reservoir capacitor can only reach about 10.6 V. That missing 1.4 V is why a “12 volt” transformer rarely gives 12 volts DC.