6. Nuclear Power

No topic in energy generates more heat and less light than this one. So we will do it slowly, from zero, and we will be honest in both directions — about what nuclear power genuinely risks, and about what it genuinely delivers.

Start with the punchline, because it defuses most of the mystique: a nuclear power station is a coal power station with a different kettle.

Just enough atoms

Everything is made of atoms. An atom has a dense nucleus of protons and neutrons, surrounded by electrons — the same electrons that carried current in Module 2.

The number of protons decides which element you have: 92 protons is uranium, always. Neutrons can vary, and atoms of one element with different neutron counts are called isotopes. Natural uranium is a mixture:

• Uranium-238 — 99.3% of it. Will not sustain a chain reaction.
• Uranium-235 — about 0.7%. Will. Everything depends on this seven-in-a-thousand minority.

That is genuinely all the atomic physics required. Two isotopes, one useful.

Splitting, and the leash

Fire a neutron at a uranium-235 nucleus and it can split — fission. Three things come out:

1. Two lighter nuclei, flying apart at enormous speed. Their motion is heat. This is the product we want.
2. Two or three spare neutrons.
3. A little mass that has vanished — converted to energy, exactly as Module 1’s 🔥🔥 block calculated.

Those spare neutrons are the interesting part. Each can split another nucleus, releasing more neutrons, which split more nuclei: a chain reaction.

Left alone, that chain either dies out or runs away. A reactor’s entire art is holding it at exactly the balance point where each fission causes precisely one more — no growth, no decay. Engineers call this criticality, and “critical” here is not a warning word: a critical reactor is a reactor working normally.

Three things hold the leash:

Control rods are made of neutron-hungry materials like boron or cadmium. Push them in and they swallow neutrons, slowing the reaction; pull them out and it speeds up. This is the throttle, and in an emergency the rods drop all the way in under their own weight in seconds, shutting the reaction down. That emergency shutdown has a name you will meet everywhere in this subject: a scram.

The moderator does something that sounds backwards: it slows the neutrons down. Neutrons leave a fission far too fast to be efficiently captured by U-235. Bounce them off light nuclei a few dozen times and they slow to “thermal” speeds, where they are hundreds of times more likely to cause the next fission. Ordinary water works; heavy water works better; graphite also works.

The coolant carries the heat out to make steam — and, crucially, must never stop. Here is the part that surprises everyone: pulling the leash all the way in does not switch a reactor off like a gas hob. The broken fragments left over from earlier fissions are themselves radioactive, and they go on decaying — and warming — for a long time afterwards. A reactor shut down a moment ago still gives off several per cent of its full heat this way. That leftover warmth has a name, decay heat, and it is what melts cores when cooling fails. It is the single most important thing to understand about nuclear accidents: the danger is not the reaction you stopped, it is the heat you cannot stop.

The kettle

Now look at that diagram’s right-hand half and compare it to Module 3. Steam generator, turbine, generator, condenser. Identical.

‹ Step back ▶ Play Step on ›

Stage 1 of 4

1. Steam generator

Fission heat is carried out by water kept liquid under great pressure, which boils a separate loop into saturated steam — cooler than a coal plant manages, which is why the ceiling below is lower.

Working fluid: water → saturated steam

You will notice one honest difference from the coal version: nuclear steam is cooler — around 285 °C rather than 565 °C, because the water is kept liquid under great pressure rather than superheated. Carnot then does what Carnot does: a lower hot-side temperature means a lower ceiling, so a typical nuclear plant converts around 33% of its heat to electricity, against 40–45% for a modern coal plant.

For coal that would matter enormously, because fuel is the dominant cost. For nuclear it barely registers, because uranium is such a trivial share of the bill. Efficiency is only worth what the fuel costs.

India’s fleet, and why heavy water

India operates 24 reactors totalling about 7.9 GW, producing roughly 57 TWh — about 3% of national generation (FY 2024-25). Small in share; strategically enormous in intent.

Most are PHWRs — Pressurised Heavy Water Reactors, designed and built in India. The choice of heavy water was forced by history and is worth understanding.

Ordinary water moderates well but also absorbs a meaningful fraction of neutrons. Run a reactor on ordinary water and natural uranium’s 0.7% U-235 is not enough — you must first enrich the fuel, meaning physically raise that U-235 share by sifting out some of the U-238 around it. Ordinary reactor fuel is enriched to three to five per cent U-235; “weapons-grade” means around ninety. Sifting apart two isotopes of the same element is extraordinarily difficult, because chemically they are identical and the only thing to separate them by is a mass difference of about one per cent — which is why enrichment is expensive, tightly controlled, and closely associated with weapons programmes. Heavy water absorbs far fewer neutrons, so a heavy-water reactor skips that step entirely and runs on natural, unenriched uranium.

For an India excluded from international nuclear commerce for decades after 1974, that was not a technical preference but a survival strategy: a reactor fuelled by ore it could mine itself, moderated by water it could make itself.

The world’s more common designs take the other path — ordinary water, enriched fuel. They come in two flavours, and between them they make up most of the world’s reactors: PWRs, pressurised water reactors, which keep the core water liquid under great pressure and boil a separate loop of water to make steam; and BWRs, boiling water reactors, which let the core water boil directly and send that steam to the turbine. Because both use ordinary water, they are known collectively as light-water reactors — “light” only in the sense of not heavy water. India’s two operating VVER-1000 units at Kudankulam, Tamil Nadu (Russian-supplied, 917 MWe net each, grid-connected in 2013 and 2016) are of this family, with four more units under construction.

Safety, told honestly

Two accidents dominate public understanding. Both deserve accurate accounts.

Chernobyl, 1986

A Soviet RBMK reactor in Ukraine, during a badly designed safety test, at low power where the design was least stable. Two fatal characteristics combined.

The first was a positive void coefficient. A “void” here just means a bubble: when coolant water boils, the steam pockets it leaves behind are voids. The void coefficient says what those bubbles do to the chain reaction, and its sign is the whole story. In the RBMK it was positive — bubbles made the reaction go faster, which boiled more water, which made more bubbles. Overheating fed on itself.

The second was no proper containment building: no sealed shell of steel and thick concrete wrapped around the reactor, whose one job is to ensure that if the core fails, what escapes the core does not escape the site.

Power surged catastrophically, a steam explosion blew the roof off, and the graphite moderator burned for days, lofting radioactive material across Europe.

The toll, on the international scientific consensus of UNSCEAR, WHO and the IAEA:

• 31 direct deaths — two killed in the explosion, one from a heart attack, and 28 emergency workers who died of acute radiation syndrome in the following weeks. Of 134 people diagnosed with acute radiation syndrome, most survived.
• Around 19,000 thyroid cancers by 2015 among people who were children at the time, of which UNSCEAR attributes roughly a quarter to radiation — caused largely by contaminated milk that authorities failed to withdraw. 15 had been fatal as of 2005; thyroid cancer is usually treatable when caught.
• Beyond this, projections of additional statistical cancer deaths across a large exposed population are genuinely uncertain and scientifically contested, because they extrapolate tiny individual risks over millions of people.

Two facts matter for judging modern reactors, and the first is more interesting than the tidy version you often hear. Light-water reactors — the pressurised and boiling-water designs that make up most of the world’s fleet — have a negative void coefficient: if coolant boils away, the reaction slows down on its own. But heavy-water designs do not. CANDU, and the Indian PHWRs derived from it that make up most of India’s own fleet, have a small positive void coefficient, and handle it with fast shutdown systems rather than with the physics of the core. So the honest statement is not that positive coefficients were never built — India runs them — but that no modern reactor combines a large positive coefficient with Chernobyl’s other failing. Containment buildings, which Chernobyl lacked, are universal.

Fukushima Daiichi, 2011

A magnitude-9.0 earthquake, then a tsunami far larger than the seawall was designed for. The reactors shut down correctly on detecting the quake. But the tsunami destroyed the backup generators, cooling stopped, and decay heat melted three cores. Hydrogen generated by overheating fuel exploded, wrecking the reactor buildings — chemical explosions, not nuclear ones.

The toll is the part almost everyone gets wrong:

• Radiation killed almost nobody. Japan has officially recognised one worker’s later cancer death as radiation-related. UNSCEAR concluded that doses were low enough that a Chernobyl-style wave of thyroid cancers can be discounted.
• The evacuation killed more than two thousand people. Over 2,000 deaths in Fukushima Prefecture are attributed to the disaster’s disruption — elderly and hospitalised people moved hurriedly, care interrupted, communities dissolved for years.
• The earthquake and tsunami themselves killed around 18,000 people. Almost none of that is in most people’s mental image of “Fukushima”.

The lesson the industry actually took is uncomfortable and important: protective measures carry their own death toll, and must be weighed like any other risk. That is not an argument for complacency about radiation — it is an argument for honest arithmetic in both columns.

Radiation, in units you can feel

Radiation dose is measured in sieverts, and everyday numbers are tiny fractions of one:

Exposure	Roughly
Eating one banana (naturally radioactive potassium)	0.0001 mSv
Dental X-ray	~0.005 mSv
Chest X-ray	~0.02–0.1 mSv
Delhi–New York flight (cosmic rays)	~0.05 mSv
Natural background, world average, per year	~2.4 mSv
CT scan	~2–10 mSv
Annual limit for a nuclear worker	20 mSv
Single dose causing radiation sickness	~1,000 mSv

Some places on Earth have natural background many times the average — the monazite sands of Kerala are a well-studied example — without detectable excess cancer in resident populations. Radiation is not magic; it is a dose-dependent hazard, like most others.

Waste: the honest version

Spent fuel is genuinely dangerous — intensely radioactive when it comes out of the core, and some components remain hazardous for thousands of years. Nothing about that is overstated.

What is usually misunderstood is the quantity. Because of the energy density from Module 1, a reactor consumes very little fuel, and so produces very little waste. All the spent fuel ever produced by a large nuclear programme occupies a volume measured in warehouse-loads, not landscapes — the entire lifetime output of one big reactor would fit comfortably in a school hall. Compare that with the ash ponds of Module 3, which are measured in square kilometres, or with CO₂, which is simply released into the atmosphere.

The technical position is that deep geological disposal is workable — Finland’s Onkalo repository is the furthest along. The unresolved problems are largely political: no community volunteers easily, and guaranteeing institutional stability over ten thousand years is a genuinely novel problem. India additionally reprocesses its spent fuel, extracting plutonium — which is not a waste-disposal strategy so much as step two of the plan below.

The three-stage plan

Here is the strategic core of Indian nuclear policy, and it makes sense only once you know two numbers: India holds roughly 1–2% of the world’s uranium and about a quarter of its thorium.

Homi Bhabha’s answer, sketched in the 1950s, was to build a ladder from the fuel India lacks to the fuel it has:

Stage 1 — PHWRs on natural uranium. Generate power, and produce plutonium in the spent fuel as a by-product. This stage is mature; it is the current fleet.

Stage 2 — fast breeder reactors on that plutonium. A breeder produces more fissile material than it consumes, by converting otherwise-useless U-238 into plutonium — and, critically, can also be loaded with thorium to breed uranium-233. India’s Prototype Fast Breeder Reactor at Kalpakkam, a 500 MWe sodium-cooled design, was fuelled in October 2025 and achieved criticality in April 2026 — decades later than originally hoped, but a real crossing into stage two.

Stage 3 — thorium reactors running on the U-233 bred in stage 2. Thorium-232 is not itself fissile; it must absorb a neutron and transmute. This is why India cannot simply skip to thorium, and why a plan begun in the 1950s is still climbing its second rung.

Whether the ladder is ever fully climbed is an open question — solar has become extraordinarily cheap in the meantime, which changes the calculation that motivated the plan. But it explains why India’s nuclear programme has always been shaped less like a power procurement and more like a seventy-year science project.

🔥Fusion: the other nuclear, and the twenty-year jokeAdvanced · optional

Everything above is fission — splitting heavy nuclei. Fusion is the opposite: forcing light nuclei together. It is what powers the Sun, and it is a better deal in almost every way.

• Fuel is deuterium from seawater and tritium bred from lithium — effectively limitless.
• No possibility of a runaway chain reaction; if confinement fails, the reaction simply stops.
• Waste is far shorter-lived: activated structural materials rather than long-lived actinides.

So why isn’t it everywhere? Because nuclei are positively charged and violently repel each other. To fuse, they must be squeezed close enough for the strong nuclear force to take over — meaning temperatures above 100 million °C, hotter than the Sun’s core. (The Sun manages at “only” 15 million °C by using crushing gravitational pressure and quantum tunnelling, plus a reaction so slow it takes billions of years — neither of which helps an engineer in a hurry.)

Nothing solid can touch that plasma, so it is held in a magnetic bottle — a doughnut-shaped tokamak. ITER, under construction in France with India as one of its seven partners, aims to demonstrate sustained net energy gain at power-plant scale.

Hence the joke that fusion is always twenty years away. It is less funny than it sounds: the field has made real, measurable progress, including recent experiments achieving more fusion energy out than laser energy delivered to the target. But “scientific breakeven in a laboratory” and “a plant that reliably sells electricity” are separated by an enormous engineering gap — materials that survive decades of neutron bombardment, tritium breeding at scale, and doing it all economically against solar that was already clearing Indian auctions at ₹2.15–2.68 a unit in 2024 (the same figures Module 12 uses).

🔥🔥The multiplication factor k, delayed neutrons, and why reactors are controllable at allDeep · optional

The one equation. Define the neutron multiplication factor:

k = (neutrons in one generation) / (neutrons in the previous generation)

• k < 1: subcritical — the reaction dies out.
• k = 1: critical — steady power. Normal operation.
• k > 1: supercritical — power rises. Used briefly and deliberately to increase output.

Control rods, moderator temperature and fuel burn-up all nudge k. Operators are, in effect, steering a number around 1.000.

The problem this creates. A neutron’s lifetime between generations in a thermal reactor is around 10⁻⁴ seconds. If k were merely 1.001 and all neutrons were prompt, power would multiply by e^(0.001 × t/10⁻⁴) — a factor of e every 0.1 seconds, roughly 20,000× per second. No mechanical control system could possibly keep up. By that reasoning, reactors ought to be uncontrollable.

Why they aren’t: delayed neutrons. About 0.65% of fission neutrons in a U-235 reactor are not emitted instantly. They come from the radioactive decay of certain fission fragments, seconds to minutes later. That tiny fraction transforms everything.

Operators keep the reactor subcritical on prompt neutrons alone, relying on the delayed fraction to reach k = 1. The effective generation time then becomes not 10⁻⁴ seconds but of order 0.1 seconds, and power changes unfold over seconds and minutes — comfortably within human and mechanical reaction times.

The boundary this defines is the most important line in reactor physics. Add so much reactivity that the reactor is critical on prompt neutrons alone — prompt criticality — and control is lost in milliseconds. That is essentially what happened at Chernobyl: a power excursion driven by the positive void coefficient took the reactor prompt-critical, and no operator action could have mattered.

Why a bomb is impossible here. A weapon requires a prompt-supercritical assembly of highly enriched material held together long enough — microseconds — for many generations to elapse. Reactor fuel at 0.7–5% U-235 cannot achieve the required neutron economy at all; long before anything approaching that, the fuel would disassemble itself thermally. Power reactors can suffer steam explosions, hydrogen explosions and meltdowns, all genuinely serious. A nuclear detonation is not among the available failure modes.

What comes next

You now know the most concentrated energy source we have ever harnessed, its real risks, and the seventy-year plan India built around a mineral it happens to own. Module 7 breaks the pattern of this entire course: a power plant with no boiler, no turbine, no shaft and no magnet at all — where light hits a slab of silicon and electricity simply comes out.