The Silicon Bench

Why copper conducts and a windowpane doesn’t

12 minutes · two numbers, not one

You already know the school answer: metals have free electrons, insulators don’t. That answer is true and it is not enough, because it only has one moving part. Conduction has two.

Commit before you touch anything

Take a copper wire and heat it from room temperature to 200 °C. What happens to its resistance?

Answer: B, it rises. Nearly a third of engineering students pick A. Hold on to why that feels right — we are about to take it apart.

The two numbers

How well something conducts is the product of two completely separate things:

How many charge carriers are free to move — call it n, measured in carriers per cubic centimetre. This is set by what the material is.
How easily each one gets through — call it μ, the mobility. This is set by how much the material gets in their way.

Conductivity is σ = q · n · μ. Two independent knobs. Almost every mistake people make in this subject — including option A above — comes from having collapsed them into one.

Now the copper answer. In a metal, n is fixed. Every copper atom donates one electron whether it is at 0 °C or 200 °C, so there are always about 8.5 × 10²² free electrons per cm³. Heat cannot add any. What heat does do is make the copper ions vibrate harder, so electrons collide more often and μ falls. One knob is nailed down, the other gets worse: resistance goes up, about 0.39% per °C.

Silicon is the opposite story, and you can see it on the bench. In silicon n is pitifully small at room temperature and heat creates carriers. That effect is so violent that it swamps the mobility loss completely. Same physics, opposite result — because a different knob is doing the moving.

In the wild

The bulb that blows when you switch it on. An old filament bulb almost always fails at the moment you flip the switch, never in the middle of the evening. A cold tungsten filament has high mobility and therefore low resistance — roughly a tenth of its hot value — so the inrush current at switch-on is about ten times the running current. That surge is what kills it.

Why your fan regulator gets warm and your wiring doesn’t. House wiring is copper because n is enormous and no engineering effort is needed. The PVC sleeve around it has essentially n = 0. Everything interesting in this course happens in the impossible gap between those two.

Before you move on

A material has very few free carriers but each one moves through it beautifully. Another has enormous numbers of carriers that can barely shuffle. Which conducts better?

Exactly. σ = q·n·μ is a product, so a huge n and a tiny μ can land in the same place as the reverse. Carrier counts in this subject span twenty orders of magnitude while mobilities span about three — so n usually wins the argument, but you still have to check.

Bench 01 · three materials, one temperature

● Locked until you commit a prediction above.

Temperature 27 °C

−50 °C100 °C250 °C

mobile electron fixed ion core (never moves) dot speed ∝ mobility · dot count ∝ carrier density

The gap

15 minutes · one number that decides almost everything

Recall From lesson 1: what are the two numbers that multiply to give conductivity? show answer

A single silicon atom has electrons sitting at sharp, separate energy levels — the picture from your chemistry class. Now bring 10²² of those atoms together into a crystal. Every level splits, again and again, until what were sharp lines have smeared into continuous bands of allowed energy.

Between two of those bands there is a range of energies an electron simply cannot have. That is the band gap. It is not a physical space or a distance. It is a height — an energy debt an electron has to pay before it can move.

Commit before you touch anything

Silicon’s band gap is 1.12 eV. A blue photon carries about 2.7 eV. Can you make a blue LED out of silicon by driving it harder?

Answer: B. The gap sets the colour. And silicon has a second, worse problem you will meet in a moment.

Three materials, one difference

Metal — no gap

The two bands overlap. There is always somewhere for an electron to go. Nothing to pay, nothing to arrange. It conducts, and that is the end of it.

Insulator — huge gap

Diamond’s is 5.5 eV. Room-temperature thermal energy is 0.026 eV. Nothing gets across, ever.

Semiconductor — awkward gap

Silicon’s 1.12 eV is too big to cross easily and too small to be hopeless. That awkwardness is the entire industry.

Drag the gap slider on the bench through those three regimes and watch the carrier count. It does not slide — it falls off a cliff, because the gap sits inside an exponential.

The gap is a colour

When an electron falls back down across the gap, it can give up the energy as a photon. The energy of that photon is the gap, and photon energy is wavelength:

λ (nm) = 1240 / E_g (eV)

That one line explains every LED you have ever seen. Drag the gap on the bench and the colour swatch changes with it — not as decoration, but because it is the same number.

Silicon’s second problem. Even at the right gap, silicon would make a terrible LED. It has an indirect band gap: an electron cannot simply drop across, it also has to hand off momentum to a lattice vibration at the same instant. That coincidence is rare, so the energy comes out as heat instead of light. Gallium arsenide and gallium nitride have direct gaps — the electron just falls, and a photon comes out. This is why the LED in your remote is not made of sand.

In the wild

Your TV remote. GaAs, gap 1.42 eV, so λ = 1240/1.42 = 873 nm — infrared, invisible to you and perfectly visible to your phone’s camera. Point a remote at your front camera and press a button; you will see it flash. That flash is the band gap of gallium arsenide.

The 2014 Nobel Prize. Red and green LEDs existed from the 1960s. Blue took thirty more years, and blue was the one that mattered, because blue plus a yellow phosphor gives white light. The blue LED bulb in your room is indium gallium nitride, gap around 2.7 eV. What held it up for three decades was not the gap — it was that nobody could dope gallium nitride p-type. Hold that thought until lesson 4.

Solar cells and the perfect gap. Small gap: you absorb lots of photons but each gives you very little voltage. Big gap: high voltage, but infrared light passes straight through and is wasted. The two effects fight, and the optimum lands at about 1.34 eV — a theoretical ceiling near 33%. Silicon’s 1.12 eV is not the best possible number. It is just close enough, and unlike everything better, we already knew how to make it perfectly pure and cheap.

Before you move on

A new material has a band gap of 3.4 eV. What is it likely to be used for?

1240/3.4 = 365 nm, which is UV — this is gallium nitride. A wide gap also means it takes an enormous field to break down, which is why GaN is taking over fast phone chargers and electric-vehicle power electronics. Wide gap is not the same as insulator; 5.5 eV is an insulator, 3.4 eV is a very useful semiconductor.

Bench 02 · the band gap SEMICONDUCTOR

● Locked until you commit a prediction above.

Band gap 1.12 eV

0 eV (metal)3 eV6 eV (diamond)

Temperature 27 °C

−100 °C150 °C400 °C

Carriers / cm³—

Thermal energy kT—

Photon λ—

Pure silicon is nearly useless — and the hole

14 minutes · the idea most people get wrong

Recall From lesson 2: what does λ = 1240/E_g tell you? show answer

Silicon has four outer electrons. In a crystal each atom shares one with each of its four neighbours, so every electron is locked into a bond. A perfect, cold silicon crystal has no free carriers at all. It is an insulator.

Warm it up, and occasionally a bond gets enough thermal energy to break. One electron comes loose. And it leaves something behind.

Commit before you touch anything

At room temperature, how many silicon bonds are broken at any moment? There are about 5 × 10²² silicon atoms per cm³.

Answer: C. About 10¹⁰ free electrons per cm³ out of 5 × 10²² atoms — one bond broken in five trillion. Pure silicon is not a poor conductor. It is very nearly a perfect insulator.

That number, n_i = 10¹⁰ cm⁻³, is called the intrinsic carrier concentration, and it is worth feeling how small it is. Ten billion sounds enormous. Next to fifty thousand billion billion it is nothing. Compare copper’s 8.5 × 10²² and you can see the whole problem: silicon by itself is twelve orders of magnitude short of being useful.

Watch the bench and count. At −50 °C almost nothing breaks. Push the temperature up and generation events start popping everywhere. As a rule of thumb, n_i in silicon roughly doubles for every 8–10 °C.

Now the hole — carefully

When the electron leaves, the bond it was in has a vacancy. A neighbouring bond electron can slide across to fill it, which leaves a vacancy where that electron was. Repeat. The vacancy travels through the crystal in the opposite direction to the electrons filling it, and it behaves in every measurable way like a particle carrying positive charge. We call it a hole.

Read this twice, because almost half of engineering students get it wrong. A hole is not an object. It is not a positron. It is not a permanent empty seat on a particular atom. It is a name for what a hundred billion billion valence electrons are doing collectively when one of them is missing. Nothing physical is moving to the right; electrons are moving to the left, and “a hole moved right” is a shorter way to say so. The reason we bother is that the bookkeeping is exact — you can give the hole a charge, a mobility and an effective mass and every equation comes out right.

The bubble in a bottle of water is the honest analogy. Nothing called a bubble exists; water moves down and we describe it as a bubble moving up, because it is easier. Nobody thinks the bubble is a substance. Try to hold holes at exactly that level of reality.

And note what this gives you: in pure silicon, every broken bond makes a pair. One electron, one hole, both free to move, both carrying current in the same direction of conventional flow. So n = p = n_i. Hold on to that equation — the moment we dope the crystal in the next lesson, it stops being true, and that is the point.

In the wild

The thermistor in your phone. Every lithium battery pack contains an NTC thermistor — a lump of semiconductor whose resistance falls steeply as it warms, because heat is manufacturing carriers. Your phone reads it constantly and refuses to fast-charge when it is hot. Same part inside your inverter AC and your BLDC ceiling fan.

Why your laptop throttles. A hot chip leaks more current through junctions that are supposed to be off, for exactly the reason on this bench: more thermal energy, more carriers generated across the gap. More leakage makes more heat, which makes more leakage. Thermal throttling is the software that stops that loop.

Why satellites and lab instruments get cooled. Infrared sensors are made from very small-gap semiconductors, so at room temperature thermal generation drowns the signal entirely. Cool them to 77 K with liquid nitrogen and the noise disappears — you are simply shutting off the generation you can see on this bench.

Before you move on

In a piece of pure silicon at room temperature, which statement is true?

One broken bond gives exactly one free electron and exactly one hole, so n = p = n_i. This is the definition of an intrinsic semiconductor, and breaking that balance deliberately is what lesson 4 is about.

Bench 03 · a silicon crystal, magnified INTRINSIC

● Locked until you commit a prediction above.

Temperature 27 °C

−50 °C125 °C300 °C

n_i (per cm³)—

1 broken bond in—

Resistivity—

free electron hole (a missing electron, not a thing) the drawing shows 28 atoms; a real crystal needs 200 billion of these panels to hold one broken bond

Doping: one atom in five million

14 minutes · the single most valuable trick in engineering

Recall From lesson 3: in pure silicon, how do the numbers of electrons and holes compare? show answer

Silicon has four outer electrons. Phosphorus has five. Slip one phosphorus atom into the lattice and four of its electrons take up the bonds — the fifth has nothing to bond with and no vacancy to sit in. It takes almost no energy to shake it loose. At room temperature, essentially every one of them is already free.

Boron has three. Slip one in and there is a bond it cannot complete. A neighbouring electron fills the gap, and the vacancy — the hole — is off wandering.

Commit before you touch anything

You dope a block of silicon with phosphorus so it is full of free electrons. Take a voltmeter and measure the block against ground. Is it negatively charged?

Answer: B — exactly neutral. Around 40% of students get this wrong, and it is the mistake that makes everything after it impossible. The bench draws the reason.

Why it stays neutral

You did not add a free electron. You added a whole phosphorus atom — fifteen protons and fifteen electrons, neutral before you started. When its fifth electron wanders off, what is left behind is not nothing. It is a phosphorus ion with a net +1 charge, welded permanently into the crystal lattice. It cannot move. Ever.

So every free electron you gained is paired with a fixed positive core you also gained. Add ten to the sixteenth donors and you get ten to the sixteenth mobile electrons and ten to the sixteenth immobile ⊕ cores. Net charge: zero, precisely.

The bench draws those fixed cores, and most textbooks don’t. That single omission is where the confusion comes from — a diagram showing only mobile electrons makes n-type silicon look negatively charged. Watch the charge audit at the bottom of the bench as you turn the doping up: the mobile count climbs, the fixed-core count climbs identically, and the net stays nailed at zero. When we join two doped blocks together in the next lesson, those immobile cores are the entire story.

The names, and what they do and don’t mean

n-type

Doped with donors (P, As, Sb). Mostly electrons carry the current. The n is for negative carriers, not for the charge of the block. The block is neutral.

p-type

Doped with acceptors (B, Al, Ga). Mostly holes carry the current. Also neutral, with fixed ⊖ cores this time.

Majority / minority

In n-type, electrons are the majority carriers and holes the minority. They do not vanish: n × p = n_i² always. Push one up and the other goes down.

Now the payoff. Turn the doping slider from zero to 10¹⁶ atoms per cm³. That is one phosphorus atom per five million silicons — a purity change of 0.00002% — and the resistivity falls from about 3.6 × 10⁵ down to 0.56 ohm-centimetres. Six hundred thousand times more conductive. Nothing else in engineering gives you a lever like that.

In the wild

Every chip ever made. A processor is not built from components that are then wired together. It is one crystal of silicon with a pattern of doping printed into it. Where the doping is n-type and where it is p-type, and how sharply the boundaries fall, is the circuit. Everything else in the chip is wiring.

Back to the blue LED. Remember lesson 2. Gallium nitride had the right gap from the start. What blocked blue light for thirty years was that nobody could make GaN p-type — the acceptors kept getting neutralised by stray hydrogen. Akasaki, Amano and Nakamura found how to un-stick them, and shared the 2014 Nobel Prize. The problem was doping, not physics.

The LDR in a street light. A light-dependent resistor is a semiconductor whose carriers are generated by photons instead of heat. Same n, same σ = qnμ, different source of energy. Your phone dims its screen using the same idea in a photodiode.

Before you move on

A block of n-type silicon has 10¹⁶ donors per cm³. Roughly how many holes per cm³ does it contain?

n·p = (10¹⁰)² = 10²⁰, so with n = 10¹⁶ you get p = 10⁴. Doping does not just add electrons — it actively suppresses holes, because more electrons around means any hole is recombined away faster. Those ten thousand survivors sound negligible, and in this lesson they are. In a transistor they turn out to be the whole game.

Bench 04 · doping and the charge audit INTRINSIC

● Locked until you commit a prediction above.

Doping concentration 10¹⁶ cm⁻³

none10¹⁶10¹⁹

Electrons n—

Holes p—

Resistivity—

vs pure silicon—

—

Push them together: the junction at rest

15 minutes · a battery you are not allowed to use

Recall From lesson 4: what is the net electrical charge of a heavily doped n-type block? show answer

Take the p-type block and the n-type block from the last lesson and make them one crystal, with a boundary in the middle. Nothing else. No battery, no wires. Watch what happens on its own.

Diffusion. There are far more electrons on the n side than the p side, so electrons wander across the boundary — not pushed, just spreading out, the way a drop of ink spreads in water. Holes wander the other way.
Recombination. An electron that crosses into p-type territory is surrounded by holes. It falls into one and both disappear. A thin region either side of the boundary is swept clean of mobile carriers.
The cores are exposed. Those mobile carriers were hiding fixed ions. On the n side, ⊕ phosphorus cores are left uncovered. On the p side, ⊖ boron cores. They cannot move to follow.
A field appears. Bare positive charge on one side, bare negative on the other, is a capacitor. There is now an electric field across the middle, pointing from n to p — and it pushes electrons back towards n.
Stalemate. Diffusion pushes carriers across; the field it created pushes them back. They balance, and everything stops. The cleaned-out zone is the depletion region, and the voltage step across it is the built-in potential, about 0.7 V in silicon.

Commit before you touch anything

There is a real 0.7 V step sitting inside every silicon diode, for free, forever. Put a voltmeter across an ordinary diode lying on your desk. What does it read?

Answer: B, exactly zero. And working out why is the most useful ten minutes in this chapter.

Why the voltmeter reads zero

Your first instinct should be suspicion. If you could read 0.7 V off a lump of silicon sitting at room temperature, doing nothing, you would have a machine that makes energy out of ambient heat forever. Physics does not permit that, so something must cancel.

What cancels is this: to measure a diode you have to touch metal probes to both ends. That is not one junction, it is three — metal-to-p, p-to-n, and n-to-metal. Each metal contact has its own contact potential, and in thermal equilibrium the three add to exactly zero. Always. The instant you go round a complete loop at a uniform temperature, everything sums to nothing.

The idea worth taking away. Equilibrium does not mean nothing is happening. Inside that junction, electrons are diffusing across and being swept back at enormous rates every second. The net is zero, which is a completely different statement. Almost everything confusing in semiconductors is a balance of two large opposing flows, and once you look for that pattern you will keep finding it.

On the bench, press form the junction and watch the sequence run. Then turn the doping up and watch the depletion region get thinner — heavier doping means the field is generated by fewer micrometres of exposed cores, so less width is needed to build the same voltage step. Typical widths are a few tenths of a micrometre: about a thousandth the width of a human hair.

In the wild

Your phone camera is fifty million depletion regions. Each pixel is a junction held in reverse bias, which widens the depletion region on purpose. A photon absorbed inside that zone creates an electron–hole pair, the built-in field instantly sweeps the two apart before they can recombine, and the charge is counted. A photon that lands outside the depletion region is mostly wasted. The photo you took this morning is a map of depletion-region events.

Tuning a radio without moving anything. Two sheets of charge with an insulating gap between them is the definition of a capacitor — and reverse bias makes the gap wider. So a junction is a capacitor you can tune with a voltage. This is the varactor, and it is how your phone, your car radio and every Wi-Fi chip tune themselves electronically instead of with the mechanical dial your grandparents turned.

Before you move on

The depletion region has almost no mobile carriers in it. What is its net electric charge?

This is the whole point of drawing the fixed cores. The mobile carriers left, but the ions they were covering did not — and cannot. That exposed, immobile charge is where the built-in field comes from. Note the signs feel backwards at first: the n side goes positive.

Bench 05 · the pn junction at equilibrium EQUILIBRIUM

● Locked until you commit a prediction above.

Doping on both sides 10¹⁶ cm⁻³

10¹⁴10¹⁶10¹⁸

Built-in potential—

Depletion width—

Peak field—

Voltmeter reads0.00 V

mobile electron mobile hole fixed ⊕ donor core — cannot move fixed ⊖ acceptor core — cannot move

two separate blocks, each neutral

Bias: now you have a diode

15 minutes · lowering a barrier, not pushing charges

Recall From lesson 5: what is left behind in the depletion region once the mobile carriers leave? show answer

The junction is sitting at a stalemate with a 0.7 V hill in the middle. Now connect a battery and change the height of the hill.

Forward bias — plus to p

The applied voltage opposes the built-in one, so the barrier drops from 0.7 V to (0.7 − V). The depletion region narrows. Diffusion is no longer balanced, and the majority carriers that were already there come flooding across. Current.

Reverse bias — plus to n

The applied voltage adds to the built-in one. The barrier gets taller, the depletion region gets wider, and majority carriers are even more firmly stuck. A tiny trickle still flows, made of thermally generated minority carriers. Nanoamps.

Watch your language, because the words do the damage. The battery does not push electrons through the diode. There is no shortage of electrons at the junction — there were 10¹⁶ per cm³ sitting on the n side the whole time. What the battery changes is the height of the barrier they face. Get this one sentence right and forward bias, reverse bias, breakdown and the transistor in chapter 2 all follow. Get it wrong and you will be memorising rules forever.

Commit before you touch anything

A diode passes 1 mA at 0.65 V forward. You raise the forward voltage to 0.71 V — sixty millivolts more. What current now?

Answer: C — ten times. Sixty millivolts is one full decade of current. Drag the bias slider and watch the log plot climb in a dead straight line.

There is no threshold. There never was.

Everyone is taught that a silicon diode “turns on at 0.7 V.” It is the most damaging half-truth in the subject, because it makes you believe there is a switch inside. There isn’t. The current is

I = I_S ( e^V/V_T − 1 ), V_T = 26 mV at room temperature

and an exponential has no threshold anywhere along it. On the bench, look at the two plots side by side. On ordinary axes you see the famous knee and your brain says “switch.” On a log scale the knee vanishes completely and you get a straight line at 60 mV per decade. The knee was never in the silicon. It was in the graph paper.

So what is 0.7 V? It is simply what you happen to measure when the surrounding circuit pushes about a milliamp through. Change the current a thousandfold and the voltage shifts by 180 mV. Useful as a rule of thumb; disastrous as a belief.

Push reverse bias far enough and it gives way

Take the slider down past −5 or −6 V on a heavily doped junction and the reverse current suddenly shoots up. Two different mechanisms do this, and which one you get depends on the doping:

Zener breakdown, below ~6 V

The junction is so thin and the field so intense that electrons tunnel straight through the barrier. Quantum mechanics, no collisions. Gets weaker as it warms.

Avalanche breakdown, above ~6 V

A stray carrier is accelerated hard enough to knock another electron out of a bond, which knocks out two more. A chain reaction. Gets stronger as it warms.

Around 5.6 V the two temperature effects cancel, which is exactly why 5.6 V Zener diodes are the classic choice for a stable reference. Neither mechanism destroys the diode — heat does, if you let the current run away. Breakdown is a design tool, not a failure.

In the wild

The diodes protecting your USB-C port. Every data pin on your phone has a tiny diode to ground, built to break down at a safe voltage. Touch the connector after walking on a carpet and several thousand volts of static arrives; the diode breaks down in nanoseconds and dumps it to ground instead of into the processor. It is engineered to fail safely on purpose.

Your charger measures its own temperature with one. At a fixed current, a diode’s forward voltage falls by about 2 mV for every degree Celsius — steady, repeatable, and free. Nearly every chip in your laptop has a diode on the die used as a thermometer for exactly this reason.

LiDAR and optical fibre. Run a photodiode in avalanche breakdown and one absorbed photon triggers a chain reaction of hundreds of carriers. That built-in gain is how a self-driving car’s LiDAR and an undersea fibre receiver detect almost single photons.

Before you move on

A Schottky diode drops about 0.3 V forward instead of 0.7 V. What does that tell you about the 0.7 V figure?

Right. 0.7 V comes from silicon’s band gap and the doping either side. Germanium gives about 0.3 V, a Schottky metal–semiconductor barrier about 0.3 V, a blue LED about 3 V — and that last one is just the 3 eV gap of gallium nitride showing up again in a different disguise.

Bench 06 · bias and the diode curve FORWARD

● Locked until you commit a prediction above.

Applied bias +0.65 V

−10 V reverse0 V+0.8 V forward

Current—

Barrier height—

Depletion width—

mV per decade59.6 mV

mobile electron mobile hole fixed ⊕ donor core — cannot move fixed ⊖ acceptor core — cannot move

The diode doing a job — and four ways to think about it

15 minutes · inside every charger in your house

Recall From lesson 6: how many millivolts of forward bias multiply the current by ten? show answer

Mains electricity is a sine wave, swinging positive and negative fifty times a second. Everything you own runs on steady DC. The bridge between the two is four diodes and a capacitor, and it is worth building on the bench because it is the first circuit where a diode is doing something you can see.

Commit before you touch anything

A half-wave rectifier: one diode, one resistor, a 12 V peak sine wave in. Add a capacitor across the resistor. What does the output do?

Answer: B. The capacitor charges to the peak, and then the diode goes reverse-biased and disconnects it, so it can only discharge into the load. That sag is called ripple, and you are about to see it.

Work through the bench in this order. Start with half wave and no capacitor: the negative half of the wave is simply gone, because during it the diode is reverse-biased and passing nanoamps. Switch to full wave and the negative half is flipped up instead of thrown away — four diodes, two conducting at a time. Now add the capacitor and watch it become almost-DC. Then drag the load heavier and watch the ripple grow, because a bigger load drains the capacitor faster between peaks.

Two things to notice that catch people out. First, the output peak is lower than the input peak — by one diode drop in half-wave, and by two in a bridge, because current always passes through two diodes in series. That is why a 12 V transformer gives you about 10.6 V, not 12. Second, once the capacitor is on, the diode only conducts in short sharp bursts at the very top of each peak. It is off for most of the cycle.

Four models of one diode

You have now met the diode at four different levels of honesty. All four are correct. The skill is picking the laziest one that still answers your question — that is engineering judgement, not a shortcut.

1 · Ideal switch

On in one direction, off in the other, no voltage drop. Use it when you are sketching what a circuit does and the supply is tens of volts.

2 · Constant 0.7 V

A perfect switch in series with a 0.7 V battery. This is the workhorse. It is what the bench above uses, and it is what you will use for 90% of your working life.

3 · 0.7 V plus a resistance

Adds a few ohms to account for the curve tilting at high current. Use it for power circuits where amps are flowing.

4 · The exponential

I = I_S(e^V/V_T − 1). The truth. You need it whenever the diode’s exact voltage matters — temperature sensors, log amplifiers, current mirrors, and the whole of chapter 2.

Notice that model 2 is the one that creates the “0.7 V threshold” misconception. Nothing is wrong with the model. What goes wrong is forgetting it is one.

In the wild

The brick on your laptop charger. Mains in, bridge rectifier, capacitor, then a switching converter running at a hundred kilohertz, then another rectifier and capacitor on the output. The circuit you just built on this bench appears twice inside it.

Why chargers use Schottky diodes on the output. At 3 A of output current, 0.7 V per diode is 2.1 W burned as heat in each one. Swap to a Schottky at 0.3 V and you have thrown away two thirds of that loss. This is why fast chargers can be small: less heat means less metal.

The diode across every relay and motor. Switch off a coil and its collapsing magnetic field produces a huge reverse voltage spike that will destroy whatever was driving it. A single diode wired backwards across the coil gives that energy a harmless path. It is called a flyback diode, it costs one rupee, and its absence is one of the most common reasons a beginner’s motor project kills its own transistor.

Before you move on

Your bridge rectifier output has too much ripple. Which change reduces it?

Ripple is roughly I_load × t / C — how much charge drains out between peaks divided by how much the capacitor holds. Lower-drop diodes raise the whole output slightly but do nothing to the sag; a higher input voltage does the same. Confirm it on the bench by dragging the load and the capacitor separately.

Bench 07 · rectifier scope HALF WAVE

● Locked until you commit a prediction above.

Smoothing capacitor none

none100 µF4700 µF

Load 220 Ω

heavy · 22 Ω220 Ωlight · 2.2 kΩ

Output peak—

Ripple—

Average DC—

Load current—

input: 12 V peak, 50 Hz

→

What happens if you use two junctions

4 minutes · the door into chapter 2

You now have every idea the transistor is made of. Doped regions. Fixed cores. A depletion region. A barrier you lower with voltage. Majority and minority carriers. An exponential.

A bipolar transistor is n-type, p-type, n-type — two junctions sharing a middle layer. Forward-bias the first one and electrons pour from the first n region into the p region, exactly as in lesson 6. Here is the twist: make that middle p region extremely thin, and most of those electrons shoot straight through it before they can find a hole to recombine with. Reverse-bias the second junction and its field grabs every electron that survives the crossing and sweeps it out.

The result is that a small current into the thin middle layer controls a much larger current straight through the device. That ratio is β, and it is the reason you have amplifiers, radios, computers and everything else.

Which also tells you why you cannot make one from two separate diodes soldered together. Between two packaged diodes there is a metal contact and a bond wire — a thick, hostile middle region where every injected electron dies. The thinness is the device. That is chapter 2, lesson 1.

Before you go on: if a sentence in this chapter did not land, this is the moment to go back, not after chapter 2. Lessons 4 and 5 — the fixed cores and the depletion region — carry more weight than the rest put together.

✓

Chapter checkpoint

8 minutes · shuffled on purpose

Mixed up deliberately. Half the difficulty in this subject is not solving a problem — it is recognising which idea the problem is about before you start.

Question 1

Heating a copper wire raises its resistance. Heating a silicon block lowers it. Why the difference?

Both materials lose mobility when hot. Only one of them gains carriers, and it gains them exponentially. Two knobs, not one.

Question 2

What is the net charge of a block of p-type silicon?

Every mobile hole is matched by a fixed negative acceptor core. You added neutral boron atoms; nothing can make the block charged. p means the carriers are positive, not the block.

Question 3

A material has a band gap of 0.67 eV instead of silicon’s 1.12 eV. What follows?

This is germanium, and it is exactly why germanium lost to silicon. Carriers go as e^−Eg/2kT, so a smaller gap means orders of magnitude more of them — and a device that stops working when it gets warm.

Question 4

Reverse-bias a diode harder. What happens to the depletion region?

More reverse voltage means more exposed fixed cores are needed to support it, so the cleaned-out zone grows. Wider gap between two sheets of charge means less capacitance — which is the varactor, and the reason your radio can tune itself.

Question 5

Why is silicon hopeless for making an LED?

Two independent problems, and you need both to answer properly. 1240/1.12 = 1107 nm is invisible; and even at that wavelength the indirect gap makes radiative recombination hopelessly improbable.

Question 6

Someone tells you “the diode drops 0.7 V, so there is a 0.7 V threshold below which no current flows.” What is wrong?

At 0.4 V the same diode passes a few nanoamps. At 0.5 V, a microamp. The exponential never starts, because it never stopped. 0.7 V is a measurement outcome that got promoted to a rule.

Checkpoint score

0 / 6

Answer all six. Four or more and you are ready for the transistor.

Every word this chapter introduced

If any of these still feels like a word rather than a picture, go back to the lesson in brackets before starting chapter 2.

Carrier L1

Anything mobile that carries charge. In semiconductors: electrons and holes.

Mobility μ L1

How easily a carrier moves through a material. Separate from how many there are.

Band gap E_g L2

The energy an electron must gain to become free. Not a distance. Sets colour, leakage and breakdown.

Direct / indirect gap L2

Whether an electron can drop across the gap alone, or needs a lattice vibration to help. Decides whether you get light or heat.

Intrinsic, n_i L3

Undoped. Carriers come only from thermally broken bonds, so n = p = n_i ≈ 10¹⁰ cm⁻³ in silicon.

Hole L3

A missing valence electron, tracked as if it were a positive particle. Bookkeeping, not an object.

Doping L4

Adding donors (5 outer electrons) or acceptors (3) to set the carrier count deliberately.

Fixed core L4

The ionised dopant atom left behind, locked in the lattice. The reason doped silicon stays neutral, and the source of every field in a junction.

Majority / minority L4

Which carrier dominates. n × p = n_i² always holds, so raising one lowers the other.

Depletion region L5

The zone either side of a junction swept clean of mobile carriers, leaving exposed fixed cores. Charged, but with nothing free in it.

Built-in potential V_bi L5

The voltage step across that zone, ~0.7 V in silicon. Real, and impossible to measure with a voltmeter.

Forward / reverse bias L6

Lowering or raising the barrier with an external voltage. Never “pushing carriers through”.

Breakdown L6

Reverse current running away, by tunnelling below ~6 V or avalanche above it. A tool, not a fault.

Ripple L7

The sag on a rectified, capacitor-smoothed output between peaks. Roughly I·t/C.

Where this goes next

CH 02

The Transistor Bench

Two junctions and a very thin base. Everything in this chapter, arranged so that a small current controls a large one.

CH 03

The emitter follower

Gain of exactly one, which sounds useless and is the most-used transistor circuit there is.

CH 04

Bias, and DC versus signal

What sits still and what wiggles — the single most documented source of confusion in the subject.

CH 05–12

Gain, output stages, FETs, op-amps — then integrated circuits

The rest of the first year, in the order that makes each one obvious rather than arbitrary.

About the simulations. The band-gap, doping and junction benches use standard semiconductor relations rather than cartoons: n_i = 2.5×10¹⁹(T/300)^1.5e^−E_g/2kT, mobility from the Caughey–Thomas fit so it falls with doping as it really does, V_bi = V_T ln(N_AN_D/n_i²), and depletion width from the depletion approximation. The diode curve is the Shockley equation. The rectifier scope deliberately uses the constant-0.7 V model, because that is the model an engineer would actually reach for — and lesson 7 says so out loud. Two things are drawn for clarity rather than to scale: the lattice benches show a few dozen atoms where a real crystal needs trillions before one bond breaks, and carrier motion is drawn far slower than the real drift velocity. One deliberate choice throughout: every diagram draws the fixed ionised dopant cores. Most textbooks omit them, and that omission is the documented source of the belief that n-type silicon is negatively charged.