Imagine you have an actual nuclear bomb sitting around in your office to play with. What is the most fun thing to do with it? If right now you are thinking of showing visitors how close you can get it to detonation without blowing yourself up, I salute your gallantry.
Of course, nuclear technicians in the 1940's already had that kind of steel balls (apparently along with lack of common sense or a will to live). You can probably imagine where this story goes...
What is a Nuclear Bomb anyway?
The idea is to make sure that when atoms decay inside your material (which in radioactive elements happens at a higher rate naturally), each resulting neutron is likely to hit another atom, triggering it to break up and in turn expel several new neutrons along with a large amount of energy, which then starts a new cycle.3 This can result in a runaway chain reaction, producing very large amounts of energy in very short time: a nuclear explosion.
The key here is to make sure that your amount of material is large enough to make it hard for neutrons to escape before hitting something. A block (or more likely a sphere) of material large enough to achieve this is considered to be supercritial - it will go up in a nuclear explosion spontaneously without any trigger. Conversely, an amount of material too small for a runaway chain reaction is called subcritial,with the label critical being reserved for material that is hovering just on the edge with an equilibrium neutron generation.
If you can't come up with enough material to achieve supercriticality, you can get around minimum size by, for example, compressing the material to higher density or by reflecting neutrons back inside when they escape.4
For real, droppable nuclear bombs there are basically two resulting concepts: either they contain two subcritical masses that are being crashed into each other to create one supercritical mass, or they contain a subcritical mass surrounded by explosives that compress it to make it go supercritical. The second and third nuclear bombs built to be dropped on Japan in WWII were of the later type.
Of course, the third bomb was never actually detonated over Japan, because the Japanese surrender came four days before the scheduled drop. As a result, the bombs plutonium core, a 14 lb (6.2 kg) 3.5 in (89 mm) sphere, remained at the Los Alamos National Laboratory5 for further study and testing. And this is where the interesting part begins.
As an implosion-type bomb, the core, nicknamed Rufus, was designed to be a subcritical sphere. But in order for the compression to be effective, the margin of criticality needed to be small. The core was design to be a mere 5% below critical mass.
In August 1945, during a test to verify the margin below criticality, physicist Harry Daghlian was stacking neutron-reflective bricks around the core in order to push it as close to critical state as possible.6 He accidentally dropped a brick onto the core, causing it to go supercritical and instantly starting a runaway fission chain reaction.
Although he managed the remove the brick immediately, stopping the chain reaction, he still received a high enough radiation dose to die from acute radiation poisoning three and half weeks later. Thankfully, the only other person in the vicinity was a security guard, who received a lower radiation dose, but still died from leukemia a few decades later.
Tickling the Dragon's Tail
Naturally, so narrowly escaping a bigger disaster resulted in stricter safety procedures for further handling of the core. The reflective bricks where replaced by two manually-operated half-spheres made from beryllium which could enclose the core and reflect neutrons back into it.
A complete enclosure of the core with the beryllium sphere would send it supercritical again, so as a safety procedure, shims where to be used between the half-spheres to secure against accidental complete closure.
Of course, as we all know: safety procedures are for pussies. Physicist Louis Slotin, in true cowboy fashion, was known for performing the test using just a screwdriver to hold the top half-sphere (instead of using the safety devices) to the consternation of other scientists.
Enrico Fermi, of Fermi Paradox fame, told him he'd be "dead within a year" if he continued the same test protocol and Richard Feynman compared the experiment to "tickling the tail of a sleeping dragon."
What do you think is going to happen...?
In May of 1946, Slotin executed the test for the last time, demonstrating the procedure to a room full of observers. Evidently, the cowboy-way of performing the test ended sub-optimally, when Slotin's screwdriver-shim slipped and the sphere closed for about half a second, resulting in a blue flash and intense heat emanating from the core. He managed to uncover the core and the chain reaction subsided after about another second. Unfortunately, he received enough radiation exposure to die nine days later from acute radiation poisoning.
Physicist Alvin Graves, observing the experiment over Slotin's shoulder, fared slightly better, only ending up in the hospital for several weeks. He did however develop permanent neurological and vision problems as a result of the exposure. Other physicists, a photographer, and a security guard in the laboratory suffered minor to significant radiation exposure. In total, eight people were treated, with Slotin being the only immediate fatality.
The Demon Core
The reputation of Rufus suffered a bit from the accidents surrounding it: scientist started referring to it as the "Demon Core".
It was originally intended to be part of the Operation Crossroads nuclear tests in the Bikini Atoll with a target yield of 23 kt (equivalent to 23,000 tons of TNT),7 but test Charlie was cancelled and the accidents had left the core in a somewhat undefined state concerning its fissile potency.
Eventually, the Demon Core, twice robbed of its grand moment, was melted down and the materials were used in other nuclear bomb cores. I find it comforting to think that its evil lives on in nuclear warheads somewhere out there.