That’s a good article. I read it some time ago, and it led me to read The Radioactive Boy Scout. (Amazon.com link.) The author of the Harper’s article, Ken Silverstein, expanded his Harper’s article into a book, and has done a lot more research, interviewed David Hahn after all the events were over, spoke with Hahn’s friends and family, and explained how Hahn built a nuclear reactor in a garden shed. Great read, and worth looking into if you’re wondering just how a high-school kid can build a functioning nuclear reactor in a backyard garden shed.
Wow, fascinating. I just don’t get one thing: they say the excursion lasted 4 ms and the rod was withdrawn about 20". How did it get that far in 4 ms? I would expect the reaction to start once the rod started moving, and the place to blow up before it got that far. It’s not like it was in a 19.99" shield and the reaction only started at 20.00".
No, the rod was inserted all the way. It need to be withdrawn by a certain amount to cause the reactor to go critical (3 inches was still shut down), and then not much farther for it to go supercritical. I would guess that it probably went critical at 10-15 inches, and then progressively more critical as the rod was yanked past that position.
I want to get the book now. I read that article when it was first published, I was surprised at the detail in it. Hahn seems to be a little nuts. I suppose if you build a nuclear reactor in high school the rest of life seems pretty dull.
If you really want to see what inspired Hahn, you can find the entire Golden Book of Chemistry Experiments online, in a very large PDF.
Not to spoil the book (the best part is the description of how he refined commonly-available substances into fissionable materials, which I won’t describe), but after the nuclear events described in the book were cleaned up, Hahn went on to a career in the Navy. Though he was assigned to a nuclear aircraft carrier, he was, not so surprisingly, allowed nowhere near the ship’s reactor.
In a similar vein, here’s a story of another teen nuclear genius.
David Hahn was recently (well, “recently” as in Aug. 2007) arrested for stealing smoke detectors, presumably to collect materials for yet another experiment. Nobody’s really sure what he was up to, though.
Toshiba is building a 10MW micro-reactor. https://en.wikipedia.org/wiki/Toshiba_4S
Longer list of micro-reactors: Small modular reactor - Wikipedia
Several factors in play here.
As beowulff said, the reactor was shut down. The rod was inserted fully, and there was no sustained reaction: the rod was consuming enough of the free neutrons to inhibit a growing reaction.
While the rod was in place, there were still neutrons flying about, just not enough. All reactors have some quantity of source neutron flux present, just at an immeasurably low power.
As the rod was extracted, its absence added positive reactivity (or, rather a lot of negative reactivity was being removed), meaning that the absence of control rod added to the reaction rate. Now the whisper of a nuclear reaction began to grow, as each generation of neutrons became larger than the previous, and a supercritical condition existed (simply meaning that power was going up).
Now for a little bit of finesse: There are two major sources of neutrons in a typical reactor, prompt neutrons and delayed neutrons. Prompt neutrons are the ones coming directly from fission; delayed neutrons come from downstream decay of the fission products. The sum of prompt + delayed neutrons forms the total neutron flux. About 7% of neutrons are delayed neutrons.
Delayed neutrons take on the order of seconds or minutes to appear, while prompt neutrons appear virtually instantly. This has an important impact on the startup rate of your core: As long as the population of prompt neutrons alone is insufficient to sustain a reaction the start up rate is relatively controllable. That is, as long as the delayed neutrons play a key part in criticality, their delayed aspect will temper the reaction rate.
At some point, if you keep adding reactivity (or removing negative reactivity), the reactor goes beyond standard criticality and enters a new and unpleasant mode called “prompt criticality”, where the delayed neutrons are unnecessary to maintain the reaction: with prompt neutrons alone, the reaction rate jumps from a period of minutes to a period of milliseconds.
Reactors start up at an exponential rate, so this means that a normal startup might have power doubling every minute, while a prompt critical reactor may double in power every few milliseconds.
And the final piece of the puzzle: even if the reaction does grow exponentially, it was very very low at the beginning, so there were many doublings needed in order to reach those final milliseconds of visible power.
Put it all together and you can imagine a certain sluggishness of the reactor as the operator jerked out the control rod, but once it got going they were done. The steam explosion is what did the damage. The steam was so powerful that the bubble in the reactor tossed the top layer of water up so quickly that it lifted the reactor vessel several feet when it hit the top.
Ah, so I’m guessing it only went prompt critical in the last few mm of travel of the control rod.
Sorry to derail the thread, but then what’s supercritical? Is it just encompassing prompt and delayed critical? And the aim of nuclear bombs is just to go to and sustain prompt criticality?
A nuclear bomb only operates in the “prompt criticality” regime.
So, before the core of the bomb has been “assembled,” it is sub-critical. Once it has been assembled, it only stays in that configuration for a few microseconds, so the entire reaction needs to start and finish within that time. This is why there is a neutron source at the pit of an implosion-type plutonium bomb - it needs to be guaranteed that there will be a free neutron available to start the reaction once the core is assembled. Once the reaction starts, it proceeded inconceivably quickly - each stage in the chain reaction (called a “shake”) takes on the order of 10 nanoseconds. The entire reaction is pretty much done in 50 shakes or so, so the core goes from inert to “boom” in under a microsecond.
Some discussion here: Introduction to Nuclear Weapon Physics and Design
Well that guy’s no fun. He probably didn’t endanger anyone else much less himself. If things didn’t go wrong for Hahn he’d probably be controlling the world with a Weather Machine right now.
From reading your description, I take it a safe-ish power reactor needs to be designed so that under no circumstances can it reach “prompt criticality”?
Yes, but it isn’t simple at all.
One thing that we are brushing up against in this discussion is the purpose of a moderator. Standard uranium reactors fission when thermal neutrons–neutrons that are traveling at more or less the same speed as the objects they strike. Neutrons from fission are flying at great speeds, far from being thermal.
So one needs a moderator in order to slow down the neutrons to the point where they can engage with a fissionable nucleus. The moderator does this by providing many like-sized objects for the neutron to strike (throw a billiard ball at a bowling ball and the pool ball just bounces off with most of its energy; throw a billiard ball at another billiard ball and it gives up much of its energy).
Common moderators include water, heavy water, and graphite. Each has its pitfalls.
The point of this discussion is that the choice and geometry of moderator affects the reactivity in the core, and how easily you can achieve criticality. Considering that water makes such a great moderator and that it has the ability to slosh around and form steam bubbles, water can change its moderation properties in the blink of an eye.
Worse still, reactors are affected by everything, including random human beings who are nearby. There have been criticality accidents in laboratories and refinement facilities where the water in a person’s body served to boost the thermal neutron flux in fissile material just enough to push it over the edge and make it critical. External objects can act as moderators, neutron absorbers, or neutron reflectors, each with different effects.
Now, once the moderator is chosen, geometry of the core is just as important.
One criticality incident happened because a fellow turned on a stirrer in a vessel that contained concentrated fissile material in solvent: The (plutonium?) had stratified in an oily layer of solvent on top of a large quantity of water. The water began stirring first, creating a vortex. This made just enough of a dip in the surface that all of the plutonium solvent mixture came to the center and reached criticality, resulting in a flash of blue and the eventual death of the fellow who flipped the switch.
So you have reflectors, moderators, and core geometry to worry about. And the reaction needs to hover in the delicate balance between “critical” and “prompt critical”
No wonder these are so difficult to work with.
And that Toshiba reactor? That’s a fast breeder reactor with a liquid sodium coolant loop. It doesn’t need a moderator, so sodium is used for its great heat capacity. But there have been countless sodium explosions and fires over the years as engineers have struggled with the technical challenges of working with an opaque and extremely reactive metal as coolant. (Imagine how you would refuel, blindly groping with robot arms in a pool of sodium to grasp fuel assemblies).
Minor,
I haven’t heard of a particular type of reactor except as research reactor and I’m wondering why:
Has no one used natural uranium, a graphite moderator and liquid metal coolant? Seems like it would be simple, cheap and safe.
Can the graphite moderator not be used as cladding or as a functional equivalent?
Ever hear of Chernobyl?
It was a graphite-moderated reactor.
As was Windscale.
You ask short questions that have long answers…
Each combination of materials has problems. As a simple example, graphite seems like it would be a perfect moderator since it is a structural material and doesn’t move.
As it turns out, graphite tends to grow dimensionally under a neutron flux (sometimes in one direction only, other times in multiple directions). Imagine your reactor developed with channels for your control rods working fine, until the channels tighten up enough that the rods don’t move.
Graphite also needs to under an occasional tempering process in order to remove internal stresses that could cause fire. And once the graphite starts burning you are in fairly serious trouble.
Now, with liquid metals you need to worry about whether the metal erodes other parts, and whether the metal will alloy with any of the reactor components.
Liquid metal reactors have serious engineering challenges: how do you make a pump seal that holds in liquid sodium? One design used solid sodium as the seal, with a special non-reactive coolant keeping the solid sodium from melting. Another design simply let the sodium squirt out and kept the whole mess immersed in a pool of sodium, with the motor poking out at the top on a long shaft. They used high pressure argon to keep the level from getting up to the motor.
As an example of corrosion, lead-bismuth is an interesting coolant, but it is very corrosive.
You can rest assured that deep thinkers have considered the combination you mention.
Does the reactor I just describe sound like it would have a positive void coefficient?
Minor,
How about gas cooled, then? The limeys seem fond of the graphite moderator and gas coolant combination.
I’m kind of curious about that myself. The benefit of liquid metal is that it is a great conductor, capable of moving great amounts of heat quickly, but it is opaque, typically very reactive, and you can never let the plant cool down (one sodium plant left in political limbo for some years racked up several million dollars per month in electric bills keeping the sodium liquid). It would seem that a gas cooled graphite plant might do the trick.
Sadly, the devil is definitely in the details. Even trivial things like how the metal in the pressure vessel will behave change once the power is on: steel pressure vessels undergo neutron embrittlement over their lifespan, as neutrons jostle up the crystalline structure of the steel kind of like quenching red hot steel does.
Anyway, you have reached the extent of my knowledge on the subject. My formal training and experience is limited to naval pressurized water reactors.
Which are used because they produce a lot of power in a compact package? If so, is this because of the greater heat of the coolant (which remains liquid because of the pressure) which allows greater efficiency?
I am asking leading questions because I am testing my own understanding.
The greater the heat differential between the heat source and the turbine, the better reactors work, right?
I understand that it’s outside your specialty but I do wonder about using gas coolant that goes straight from the core to the turbine. There is pretty much no practical limit to the heat that can be imparted to carbon monoxide as a coolant, right? My limited understanding is that this would irradiate the turbine but that such radioactivity is quite short-lived.