Last lesson gave you the shape in the abstract: a self-amplifying process has a threshold where its amplification factor k crosses 1, and the flip from “fizzles” to “runs away” is a phase change, not a ramp. That was the tour. Now we go back to the exact place the idea was forged — the inside of a lump of uranium — and pin down what makes a mass critical down to the atom. Because the nuclear version isn’t just the origin story; it’s the cleanest, most literal instance of the whole model, and it hands us one gorgeous surprise the abstract version can’t: whether k crosses 1 can depend on nothing but the size of the lump.
Before you read — take a guess
You have a sphere of pure bomb-grade uranium sitting just below critical mass — inert, safe to stand next to. Without adding any new material, changing its purity, or heating it, what could tip it over into a runaway chain reaction?
The fission chain reaction: the reinforcing loop
Picture a mousetrap-covered floor, each trap loaded with two ping-pong balls, packed wall to wall. Toss in one ball. It springs a trap, which flings two balls, which spring two more traps, which fling four balls — and in about a second the whole room is a blizzard. That runaway is the physical picture behind a nuclear chain reaction, and the real version is what physics handed the world in the 1940s.
Here are the three pieces of jargon, defined before we lean on them. An isotope is a version of an element with a particular number of neutrons in its nucleus (its heavy centre); uranium-235 (U-235) is the isotope of uranium whose nucleus has 143 neutrons, and it’s the star of the show because it’s fissile — easy to split. Fission is that splitting: a heavy nucleus breaks into two lighter ones, releasing a burst of energy. A neutron is one of the uncharged particles in a nucleus; on its own, drifting free, it’s the messenger that carries the reaction from atom to atom.
Now the loop. A free neutron slams into a U-235 nucleus. The nucleus wobbles, splits (that’s the fission), and in splitting it spits out a jolt of energy plus roughly 2 to 3 fresh neutrons. Each of those loose neutrons can now go find another U-235 nucleus and do the same thing. One split causes several neutrons, which cause more splits, which cause yet more neutrons — a textbook reinforcing feedback loop, the exact engine the whole course is about. The output of one round is the input to the next.
Why 2–3 neutrons is the whole ballgame
If each fission released exactly one neutron that went on to cause exactly one more fission, the reaction would just tick over forever at the same rate — steady, boring, useful. The fact that it releases 2 to 3 means there’s spare capacity to grow. Whether that spare capacity actually gets used, or leaks away, is the entire question — and it’s answered by a single number.
The k-factor: sub-critical, critical, super-critical
That single number is k, the neutron multiplication factor. Strip away the physics and it’s just the answer to one honest question:
On average, how many neutrons from one fission go on to cause another fission?
Not how many neutrons are released (that’s always 2–3), but how many actually land a hit and split something. The others miss, or escape, or get soaked up by impurities. So k is the net per-generation gain of the chain — and three sharp cases fall out of where it sits relative to 1:
- k < 1 — sub-critical. Each fission causes, on average, less than one further fission. Every generation is smaller than the last. The reaction limps along and dies. A lump of uranium sitting on a shelf is sub-critical: it’s not “off,” it’s losing.
- k = 1 — critical. Each fission causes exactly one further fission. The reaction holds perfectly steady — self-sustaining, neither growing nor shrinking. This is what a nuclear reactor deliberately sits at: a controlled, level burn.
- k > 1 — super-critical. Each fission causes more than one further fission. Every generation is bigger than the last, and because it compounds, the growth is explosive — literally. This is a bomb.
The words are worth memorising precisely, because “critical” in ordinary English sounds like “dangerous / about to blow.” In the physics it means the opposite of runaway: k = 1 is the calm case, the steady one. The dangerous one is super-critical.
k is the amplification factor from the introduction, wearing a lab coat
Everything the intro said about “the amplification factor crossing 1” is this. Neutrons are the events; a fission causing further fissions is one event causing more of itself; k is how many. Every other tipping point in this course — R0 in an epidemic, the viral coefficient of an app — is the same k in a different costume. This is just the version where you can point at the actual particles.
Worked example: why 0.9 and 1.1 live in different universes
Here’s the move people underrate: k doesn’t act once, it acts every generation, and generations in a real chain reaction tick over in tens of nanoseconds. Repeated multiplication is where a boring-looking gap becomes an abyss.
Start both runs with the same seed: 100 fissions in generation 0. Generation n has roughly (starting count) × kⁿ fissions. Watch what a mere 0.2 difference in k does over ten generations:
| Generation | k = 0.9 (sub-critical) | k = 1.1 (super-critical) |
|---|---|---|
| 0 | 100 | 100 |
| 1 | 90 | 110 |
| 2 | 81 | 121 |
| 3 | 73 | 133 |
| 4 | 66 | 146 |
| 5 | 59 | 161 |
| 10 | 35 | 259 |
| 50 | ≈ 0.5 | ≈ 11,700 |
| 100 | ≈ 0.003 | ≈ 1,378,000 |
Read the two columns side by side. They start identical. The left column bleeds out — halving every seven generations or so, until after 100 generations it’s rounding-error dust; the fire is out. The right column swells — after 100 generations it’s over a million times the seed, and remember each generation is nanoseconds, so “100 generations” is a rounding error of a second. The left lump is a warm paperweight. The right lump is a crater.
The point that should stick: the difference between “dies to nothing” and “consumes everything” was a change in k from 0.9 to 1.1. A 0.2 nudge. It matters so violently only because it’s the difference between shrinking-by-10%-repeatedly and growing-by-10%-repeatedly, and repeated a hundred times, 0.9¹⁰⁰ and 1.1¹⁰⁰ are separated by a factor of roughly four hundred million. That, in one table, is why threshold behaviour ambushes our intuitions — a subject the whole of lesson 4 is devoted to.
A chain reaction is running at k = 1.02 — only 2% above critical. A colleague waves it off: '2% is basically nothing, it'll behave almost exactly like the steady k = 1 case.' Where's the trap?
The deep insight: why there has to be a minimum mass
Now the twist that makes this the origin story of the whole model — and it’s a geometry argument, not a chemistry one.
Here’s the puzzle. If k is just “how many neutrons land a hit,” and every U-235 atom behaves the same, why should the amount of uranium matter at all? A neutron either finds another nucleus or it doesn’t — surely that’s a fact about the atoms, not about how big the pile is?
The answer is escape. Fissions happen throughout the volume of the lump — anywhere there’s fuel. But neutrons are lost mainly by flying out through the surface into the empty air outside, where they hit nothing and are gone forever. So the chain has a production term that scales with volume and a loss term that scales with surface area. And here is the crucial fact of geometry: as a sphere gets bigger, its volume grows as the cube of the radius (r³) while its surface grows only as the square (r²). The ratio of surface to volume falls as the sphere grows.
Translate that into neutrons. In a small sphere, almost every point is close to the surface, so a huge fraction of neutrons escape before they can cause a fission — losses dominate, k < 1, sub-critical. In a big sphere, most of the volume is buried deep in the interior, far from any surface; neutrons born there are surrounded by fuel and almost certainly hit something before they can reach the edge — losses are a small fraction, k > 1, super-critical. Somewhere in between is a single radius where escape and production exactly balance — k = 1 — and the amount of fuel at that radius is the critical mass.
This is literally why a bomb needs a critical mass
A bomb doesn’t need “critical fuel” — a gram of U-235 is chemically identical to fifty kilograms of it. It needs enough assembled together in one place that the interior outweighs the leaky surface. Below the critical radius, too many neutrons escape and the chain dies no matter how pure the fuel; at the critical radius the books balance; above it, it runs away. The threshold is a fact about geometry and size, which is exactly why the model is named for a mass.
Let’s put real numbers on it. A bare sphere of pure U-235 has a critical mass of about 52 kg — a ball roughly the size of a grapefruit, about 17 cm across. That’s it. Not a warehouse of the stuff: a grapefruit. Wrap that sphere in a neutron reflector (a shell that bounces escaping neutrons back in — effectively plugging the leak) and the critical mass drops dramatically, because you’ve cut the loss term. Same atoms, different critical mass, because you changed the escape.
You’ve actually already watched this happen. In the grid island from the introduction, cells on the edge fire some of their reactions off the boundary into nothing — that’s surface escape, in miniature. A tiny seed near the edge leaks away and fizzles; a big dense seed buried in the middle of the grid keeps its reactions in-bounds and sweeps. Go back and feel it in your hands, now framing the grid as fuel and each firing as a neutron finding a neighbour:
Below the threshold, above the threshold
Neutrons, fuel, and the leaky edge
Each square is a scrap of fuel; the coloured ones start the reaction. A reacting cell fires at its neighbours, triggering on average k new reactions before it burns out. Set k below, then step or run — hunt for the tipping point where “fizzles” flips to “runs away”.
Step 0: 9 cells reacting now; 9 ever activated (2% of the grid). Reacting…
Two things to try deliberately. First, park k just below 1.00 and press Run with a tiny seed — it fizzles, mostly through edge losses. Now, at the same k, crank the seed density up: more simultaneous chains, more of them buried away from the boundary, and the very same k can tip into a takeover. That’s critical mass appearing out of geometry while k the “per-atom” number never changed — the whole insight of this section, in your hands. Second, notice how sharp the flip is around 1.00: a smooth slider, an abrupt change of kind. Phase change, not ramp.
How humans steer k on purpose
If k = 1 is a knife-edge, the entire nuclear enterprise is about deciding which side of it to stand on — and staying there.
A bomb wants k well above 1, for as long as possible. The problem is that the instant a super-critical mass starts reacting, it heats up and blows itself apart, dropping back below critical in a flash — so you get a fizzle unless you assemble the super-critical geometry faster than the reaction can disassemble it. Two tricks do this. The gun method fires one sub-critical slug of uranium into another down a barrel, so that only when they slam together is the combined lump over critical mass (this is roughly the Hiroshima design). The implosion method takes a sub-critical sphere and crushes it inward with a shell of conventional explosives, squeezing it denser — which, as our r³-vs-r² argument predicts, shrinks the surface relative to the volume and shoves k over 1 without adding a single atom.
A reactor wants k pinned at exactly 1, indefinitely. That’s the steady, self-sustaining burn that boils water and spins turbines. But 1.000 is a single point on a continuum — leave it alone and it drifts. So a reactor is fitted with control rods, rods of a neutron-hungry material (like boron or cadmium) that absorb spare neutrons. Push the rods in and they eat neutrons, lowering k toward or below 1; pull them out and k rises. Operators nudge them constantly to hold k at 1.000 — surfing the razor’s edge on purpose, all day, every day. A reactor isn’t a slow bomb; it’s a system holding itself exactly at the threshold the bomb blows straight through.
Same physics, opposite goal
Bomb: get k as far above 1 as you can, as fast as you can, before it tears itself apart. Reactor: hold k at 1.000 forever, actively fighting every drift. The difference between a power plant and a weapon isn’t the atoms — it’s which side of the threshold you engineer for, and how tightly you control it.
The pitfall: “more fuel = more boom” (the linear trap)
The intuition to kill is the linear one: the quiet assumption that twice the uranium means twice the bang, and half the uranium means half the bang, as if output tracks input in a straight line. It’s the same instinct that expects twice the marketing to bring twice the users. It is wrong here in the most dramatic way possible.
Below critical mass, adding fuel does almost nothing — a 40 kg lump and a 50 kg lump are both duds, both sitting there sub-critical, both losing more neutrons than they make. Then you cross ~52 kg and the behaviour doesn’t increase, it changes kind: dud, dud, dud — crater. There is no proportional middle. The relationship between fuel and outcome isn’t a slope, it’s a cliff. This is the phase-change signature the whole course keeps returning to, and mistaking the cliff for a slope is the single most common error people make about every threshold system.
A subtler cousin of the same mistake: thinking k is a fixed property of uranium, a number you could look up in a table. It isn’t. k belongs to the whole configuration — the purity of the fuel (impurities and the wrong isotopes soak up neutrons), the shape (a sphere leaks least; a flat sheet or thin wire leaks like a sieve), the size, the density, and whether it’s wrapped in a reflector. The same kilograms of uranium can be safely sub-critical as a thin disc and lethally super-critical as a squeezed sphere. When you ask “which side of 1 is this on?”, the honest answer is always for this particular arrangement — never for the material alone.
Criticality accidents are made of this exact error
The deadly real-world version isn’t “too much material,” it’s “the same material in a worse shape.” Historically, criticality accidents have happened when technicians let a sub-critical amount of fissile solution slump into a rounder, more compact geometry, or brought two safe pieces too close — nudging k over 1 with no new fuel at all. The lesson has teeth: k is a property of the whole setup, and geometry alone can flip it.
Quick check
Neutrons, k, and escape
A reaction starts with 200 fissions in generation 0 and runs at k = 1.5. Roughly how many fissions are in generation 3?
Check your answer to continue.
Where this goes next
You now own the original, most literal version of the model: fission drives a reinforcing loop, k is its per-generation gain, and the phase line at k = 1 splits sub-critical (dies), critical (steady reactor), and super-critical (runaway bomb). You’ve seen why a 0.2 change in k decides everything (repeated multiplication), why the threshold is a mass at all (neutrons escape through the r²-surface while fissions fill the r³-volume), how humans deliberately steer k with implosion and control rods, and why “more fuel = more boom” is a cliff mistaken for a slope.
But nothing in that argument actually needs uranium. Swap “neutron finds a nucleus” for “infected person infects another,” or “one user invites a friend,” and every line of reasoning survives intact — the loop, the k, the threshold, even the escape term. Next up, lesson 2 — The General Threshold — lifts the whole model off the reactor floor and shows that any reinforcing feedback loop has a k, and critical mass is simply that k crossing 1. That’s the lesson that turns a fact about physics into a tool for reading the world.