Table of Contents (click to expand)
- What Does A Nuclear Chain Reaction Need?
- Why Was Uranium Better Fuel 2 Billion Years Ago?
- How Did Oxygen Pile All That Uranium Into One Place?
- How Did Groundwater Switch The Oklo Reactor On And Off?
- How Do We Know Oklo Really Ran? Ask The Xenon
- How Long Will Nuclear Waste Stay Put, And Could Oklo Happen Again?
In 1972, a French uranium plant found ore from Oklo in Gabon that was short of the splittable kind of uranium by three thousandths of a percent. Two billion years ago that ore was about 3.6% splittable, which is modern reactor grade, and groundwater seeping through it slowed the neutrons enough to keep a chain reaction going. Sixteen natural reactors switched themselves on and off for hundreds of thousands of years at under 100 kilowatts, and none can form again, because the splittable uranium has since decayed away.
In May 1972, a technician at a uranium plant in Pierrelatte, France, ran a routine check on a batch of ore. The plant did this all day long.
The number came back wrong. Not dramatically wrong. It was short by three thousandths of a percent.
Natural uranium is one of the dullest, most consistent substances on Earth. Dig it up anywhere and 0.720% of it is uranium-235, the splittable kind. This batch, from a mine at Oklo in Gabon, read 0.717%. Most people would round that away without a thought.
Somebody did not. Chasing that gap took French scientists to a hillside in West Africa, and to a conclusion that sounds like a joke. The uranium was short because something had already used it. Two billion years before anyone built a reactor, Earth had built sixteen.
What Does A Nuclear Chain Reaction Need?
Start with the fuel. Uranium comes in two main kinds, and they behave nothing alike.
Uranium-238 is the common kind. Fire a neutron at it and mostly nothing happens. Uranium-235 is the rare kind, and it is the one that splits in two when a neutron hits it. The split releases energy, and it throws out two or three fresh neutrons. If each of those goes on to split another nucleus, the process keeps itself alive. That is a chain reaction.
There is a catch. Neutrons come out of a split fast, and a fast neutron tends to sail straight past the next nucleus. So you have to slow them down, by bouncing them off small, light atoms that soak up speed without swallowing the neutron. A material that does this job is called a moderator. Some reactors use heavy water, but most use plain water, because water is mostly hydrogen and hydrogen is about as light as an atom gets.
So a working reactor needs three things at once. Enough of the splittable kind packed close together. A moderator to slow the neutrons. And nothing nearby that swallows neutrons and kills the chain.
Now look at today's ore. Out of every thousand uranium atoms you dig up, roughly seven are the splittable kind. Neutrons from those seven spend nearly all their time bumping into the common kind, which swallows them. Water slows neutrons beautifully, but it swallows a fair share too. Seven atoms in a thousand leaves no margin for those losses, so the chain sputters and dies before it starts.
This is why enrichment plants exist. Their whole job is to nudge that seven-in-a-thousand upward, because power reactors need fuel that is 3% to 5% splittable.
So a pile of ordinary uranium ore sitting in wet rock cannot start a chain reaction. That is true. It is also only true right now.

Why Was Uranium Better Fuel 2 Billion Years Ago?
Both kinds of uranium are radioactive, but they do not decay at the same speed, and that changes everything.
Every half-life, half of what you had is gone. The splittable kind has a half-life of about 700 million years. The common kind takes about four and a half billion, so the splittable kind burns down more than six times faster.
Now run the clock backwards. Two billion years is nearly three half-lives for the splittable kind, so there was about seven times more of it than today. For the common kind it is less than half of one half-life, so there was less than one and a half times more. Put the two together and uranium of that age works out at about 3.6% splittable.
That is the enrichment range from the previous section. Two billion years ago, every scrap of uranium on the planet already sat at power-reactor grade. Nature never needed an enrichment plant. Nature was the enrichment plant, and it has been shutting itself down ever since.

How Did Oxygen Pile All That Uranium Into One Place?
Good fuel is not enough. It has to be gathered into one dense body of ore, and for most of Earth's history uranium refused to gather. Without oxygen around, uranium barely dissolves in water. It stays locked in the rock it is born in, scattered thin, going nowhere.
Then life changed the planet. Cyanobacteria had been making oxygen through photosynthesis, and the air filled up with it. This is the Great Oxidation Event. In an oxygen-rich world, uranium switches to a form that dissolves in water. Now groundwater could pick uranium up and drop it again wherever conditions changed. Over time it piled up in sandstone layers of the Franceville Basin, in what is now Gabon.
One more thing had to go right. Some elements, including boron and cadmium, swallow neutrons greedily. A trace of them in the ore and no chain reaction can start. The Oklo ore happened to be almost free of them.

How Did Groundwater Switch The Oklo Reactor On And Off?
Water was already seeping through that ore, and water is a moderator. It slowed the neutrons down, and the chain reaction caught.
If a reactor starts by accident, with nobody watching, what stops it running away? Most people picture Chernobyl here. It is the wrong picture. The right picture is a geyser.
A geyser fills with water, heats up, blows its water out, and then has to wait for the water to return before it can go again. Oklo did the same, running on neutrons instead of steam. The thing that started it was also the thing that stopped it:
- Water fills the pores in the rock. Neutrons slow down. The chain reaction begins.
- Splitting atoms makes heat.
- Heat boils the water off.
- No water means no moderator. Neutrons speed up, miss their targets, and the reaction dies.
- The rock cools. Water seeps back in. Go to step 1.
That is a negative feedback loop, and it is a wonderfully dumb one. No sensor. No control rod. No operator. Just water that leaves when things get hot. Alex Meshik and colleagues read the timing out of the rock itself: the zone they studied ran for about 30 minutes, then sat dark for at least two and a half hours, over and over.
The output was modest. Average power stayed under 100 kilowatts, enough to run a few dozen toasters. But it kept going. Sixteen reactor zones sit across the Oklo and Okelobondo deposits, and together they ran for hundreds of thousands of years. Our own reactors are far more powerful, but ours would not last a week without people tending them. Oklo ran unattended for a stretch comparable to the entire history of our species.

How Do We Know Oklo Really Ran? Ask The Xenon
The ore gave up three separate kinds of evidence.
The first is the missing fuel. Roughly 200 kilograms of the splittable kind was unaccounted for in the ore being analyzed. That is the shortfall the Pierrelatte technician stumbled into, and only a slice of the several tonnes the reactors got through over their lifetime.
The second is the ash. Splitting uranium leaves fragments, which physicists call fission products and which we will call ash. The ash carries isotope mixtures that look nothing like the same elements in ordinary rock. Neodymium at Oklo has the fingerprint of splitting atoms, not of geology.
The third is the strangest, and it is why we know about the on-and-off cycle at all. Meshik's team found xenon trapped in aluminum phosphate minerals inside the reactor zones, in grains that hold no uranium of their own. Xenon is a noble gas, so it does not bond with anything. It just sat there for two billion years, minding its own business.
Xenon comes in several isotopes, born from parent atoms with different half-lives. Some appear within minutes of a split, others take days. Reading which isotopes got trapped, and in what proportions, told the team how long the geyser ran and how long it rested each time. The gas that refuses to react turned out to be the one witness that could still testify.

How Long Will Nuclear Waste Stay Put, And Could Oklo Happen Again?
Burying radioactive waste deep underground is the leading plan almost everywhere. The standing objection is that nobody can test it. The waste must stay put for tens of thousands of years, and no experiment runs that long. So people keep floating wilder options, like firing it at the Sun or dropping it into a volcano.
Oklo is the experiment. It ran for a few hundred thousand years, and then its ash sat in that rock for two billion more, hot and irradiated, with groundwater moving through the whole time. Those are close to worst-case conditions.
The rock held. Much of the ash never left the mineral it formed in, and several other elements moved only a few tens of meters in two billion years. That is the strongest real-world evidence we have that deep geological storage can work.
One caution, though. The noble gases, xenon and krypton, had largely escaped the reactor zones. Oklo held nearly everything. It did not hold everything.

So line the pieces up. Reactor-grade fuel, because the splittable kind had not yet decayed away. Oxygen in the air, so water could gather uranium into thick ore. Ore that happened to be free of neutron-swallowing elements. And groundwater running through it, slowing neutrons enough to light the geyser. Every condition a reactor needs turned up by accident, all at once, in the same wet rock in Gabon.
It cannot happen again, and the reason is the arithmetic that made it possible. The splittable kind keeps decaying six times faster than the common kind. It has fallen from reactor grade to seven atoms in a thousand, and it is still falling. The planet has plenty of uranium left. None of it is rich enough to light itself.
Oklo ran for hundreds of thousands of years, went quiet, and waited two billion years for someone to notice. We noticed in 1972, because one technician decided three thousandths of a percent was worth explaining.
Somewhere in the routine data you are about to round off, there may be a hillside in Gabon.
References (click to expand)
- Meshik, A. P. — The Workings of an Ancient Nuclear Reactor (Scientific American, 2005)
- Nature's Nuclear Reactors: The 2-Billion-Year-Old Natural Fission Reactors in Gabon, Western Africa (Scientific American guest blog, 2011)
- Radionuclide Basics: Uranium — US Environmental Protection Agency
- The Nuclear Fuel Cycle — US Energy Information Administration
- Brookins, D. G. — Migration and retention of elements at the Oklo natural reactor (Environmental Geology, 1982; record on OSTI.GOV)







