How Did Earth Run Its Own Nuclear Reactors 2 Billion Years Ago? (The Oklo Reactors)

In 1972, French technicians found uranium ore from Oklo in Gabon that was slightly short of the rare, splittable kind of uranium: 0.717% instead of the 0.720% found everywhere else on Earth. Something had already burned several tonnes of it, because two billion years ago that ore held about 3.6% of the splittable kind, and groundwater trickling through the rock was enough to keep a nuclear reaction running. Sixteen of these natural reactors switched themselves on and off for hundreds of thousands of years at an average under 100 kilowatts, and none can ever form again, because that 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. Uranium arrives, you measure it, you write down the number, you move on.

The number came back wrong. Not dramatically wrong. It was short by 0.003 percentage points.

Natural uranium is one of the dullest, most consistent substances on Earth. Dig it up in Canada, Australia, or Kazakhstan and 0.720% of it is uranium-235. This batch, from a mine at Oklo in Gabon, read 0.717%. Most people would round that away without a thought.

Somebody did not round it away. 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 Has To Be True For Nuclear Fission To Keep Going?

Start with the fuel. Uranium comes in two main flavors. They behave nothing alike.

Uranium-238 is the common one. Fire a neutron at it and mostly nothing happens. Uranium-235 is the rare one, and it is the useful one. Hit a U-235 nucleus with a neutron and it splits in two. That split releases energy, and it also throws out two or three fresh neutrons.

Those loose neutrons are the whole trick. If each of them goes on to split another U-235 nucleus, the process keeps itself alive. That is a chain reaction. Physicists call this ability to split "fissile." It is why U-235 gets all the attention, while other elements sit on the bench.

There is a catch, and it is a strange one. Neutrons come out of a fission fast. A fast neutron tends to sail straight past the next U-235 nucleus without doing a thing. Physics rewards the slow here. So you have to slow the neutrons down. You do that by bouncing them off small, light atoms, which soak up speed without swallowing the neutron.

A material that does this job is called a moderator. Some reactors use heavy water. Some use graphite. Most use plain water. A water molecule is mostly hydrogen, and hydrogen is about as light as an atom gets.

So a working reactor needs three things at once. Enough U-235 packed close together. Something to slow the neutrons down. And nothing nearby that eats neutrons and kills the chain. Hold on to those three. They are the entire article.

What Stops Uranium Ore From Starting A Reactor Today?

Now look again at that 0.720% figure. Out of every thousand uranium atoms you dig out of the ground, roughly seven are the kind that will split. The other 993 mostly will not.

That is far too thin for water to work with. Neutrons from those seven atoms spend nearly all their time bumping into U-238. They get absorbed, scattered, or wasted. Water slows neutrons beautifully. It also swallows a fair share of them, and seven splittable atoms in a thousand leaves no margin for that loss. The chain sputters and dies before it starts.

This is why uranium enrichment plants exist, and why they start arguments between governments. Their whole job is to nudge that 0.7% upward. The US Energy Information Administration gives the target: power reactors need fuel enriched to 3% to 5% U-235. Getting there takes centrifuges, enormous amounts of electricity, and international treaties.

Enrichment is not the only way around the problem, mind you. Change the moderator instead. Heavy water is far stingier about absorbing neutrons than ordinary water is. With heavy water doing the slowing, unenriched uranium works fine. Canada built an entire nuclear industry on that idea. Every operating power reactor in the country is a CANDU. They run on natural uranium, moderated and cooled by heavy water. But heavy water is a manufactured product, and a costly one. Wet rock is not.

So the physics seems settled. A pile of ordinary uranium ore sitting in wet rock cannot start a chain reaction. Nature does not sort its own laundry.

That is true. It is also only true right now.

Why Was Uranium Better Fuel 2 Billion Years Ago?

Here is the piece almost every retelling skips, and it is the piece that makes everything click.

Both kinds of uranium are radioactive. Both are decaying away right now. But they are not decaying at the same speed, and that changes everything.

The rule for radioactive decay is simple. Every half-life, half of what you had is gone. Written out, the amount left after a time t is:

N = N₀ × (½)^t/T½

where N₀ is what you started with, T½ is the half-life, and t is how long you waited. To look backwards instead of forwards, you flip the fraction:

N_past = N_now × 2^t/T½

Now put the numbers in. The US Environmental Protection Agency gives U-235 a half-life of 700 million years. U-238 gets 4.47 billion. U-235 burns down more than six times faster.

Wind the clock back two billion years, one step at a time:

1. Two billion years is about 2.9 half-lives for U-235. So there was 2^2.9, or about 7.2 times more U-235 than today.
2. Two billion years is only about 0.45 of a half-life for U-238. So there was 2^0.45, or about 1.4 times more U-238.
3. Divide one by the other. U-235 stood about 5.3 times higher against U-238 than it does now.
4. Today's mix is 0.72 parts U-235 to 99.28 parts U-238. Scale the U-235 side up by 5.3 and you get roughly 3.8 parts against 99.28. That is about 3.7% of the total.

Scientific American puts the figure at about 3.6% for two billion years ago. That is the same answer, reached with marginally different half-life values.

Read that number again. Then read the enrichment range in the previous section. They are the same number. 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.

The reason is chemistry. In its reduced form, 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. Around 2.4 billion years ago, the air filled up with it. This is the Great Oxidation Event, and it rewrote the chemistry of the whole surface of the world.

Oxygen changed uranium. In an oxygen-rich world, uranium switches to a form that dissolves in water. Now groundwater could pick uranium up, carry it, 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, and you only notice it by its absence. Some elements, including boron and cadmium, are ferocious neutron eaters. A trace of them in the ore and no chain reaction can start. The Oklo ore happened to be almost free of them.

So the chain runs like this. Bacteria learned to breathe out oxygen. Oxygen let uranium dissolve. Water gathered it up. And the ore it built happened to be clean. It took the invention of photosynthesis to make a nuclear reactor possible. That is a long supply chain.

How Did Groundwater Switch The Oklo Reactor On And Off?

Water was already seeping through that ore body. And water, as we established, is a moderator. It slowed the neutrons down, and the chain reaction caught.

Which raises the obvious worry. If a reactor starts by accident, with nobody watching, what stops it running away? Most people picture Chernobyl here. It is the wrong picture.

Oklo could not run away, because the thing that started it was also the thing that stopped it. Work through the loop:

1. Water fills the pores in the rock. Neutrons slow down. Fission begins.
2. Fission makes heat.
3. Heat boils the water off.
4. No water means no moderator. Neutrons speed up, miss their targets, and the reaction dies.
5. 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 it sat dark for at least two and a half hours. Over and over. Think of a geyser, running on neutrons instead of steam.

The output was modest. Average power stayed under 100 kilowatts, which Meshik describes as enough to run a few dozen toasters. But it kept going. Sixteen reactor zones sit across the Oklo and Okelobondo deposits. Together they released something like 15,000 megawatt-years of energy. It took them hundreds of thousands of years.

Set that against a modern plant. Our reactors are far more powerful, but they are also needy: ours would not last a week without people tending them. Oklo ran unattended for a stretch of time comparable to the entire history of our species.

How Do We Know Oklo Really Ran? Ask The Xenon

A story this odd needs hard evidence, and the ore gave up three separate kinds.

The first is the missing fuel. Roughly 200 kilograms of U-235 was unaccounted for in the ore being analyzed. That is the shortfall the Pierrelatte technician stumbled into. The reactors got through far more than that over their lifetime. The 15,000 megawatt-years they released works out to several tonnes of U-235, not a couple of hundred kilograms.

The second is the ash. Splitting uranium does not leave a clean hole. It leaves fission products. Those products carry isotope mixtures that look nothing like the same elements in ordinary rock. Neodymium at Oklo has the fingerprint of fission, not of geology.

The third is the strangest, and it is why we know about the pulsing at all. Meshik's team found xenon trapped in aluminum phosphate minerals inside the reactor zones. Those grains 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 isotopes come from parent atoms with different half-lives. Some appear within minutes of a fission. Others take days. The team read which isotopes got trapped, and in what proportions. From that they worked out how long the reactor had run and how long it had rested. The gas that refuses to react turned out to be the one witness that could still testify.

Computer models have since rebuilt all of this in detail. A 2014 review by Davis, Gould and Sharapov walks through those reconstructions of the reactor zones. It also covers a stranger use. Physicists now probe Oklo to test whether the constants of nature have shifted in two billion years.

How Long Will Nuclear Waste Stay Put? Oklo Is The Only Real Test

Now for the part that matters beyond trivia.

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. 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. Then its waste 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.

And uranium-235 was not the only fuel involved. Remember U-238, the common kind that mostly refuses to split? Fire a neutron at it and something else happens instead. It swallows the neutron. Two quick steps later it is plutonium-239, which splits as readily as U-235 does. Meshik reports that more than two tons of plutonium-239 were bred inside the Oklo deposit. Some of it went on to fission, and we know that because its own characteristic fission products are still sitting in the rock. So the deposit did not only burn fuel. It quietly made more, and burned some of that too. Almost none of that plutonium is left today. Its half-life is about 24,000 years, and two billion years is a great many half-lives.

The rock held. David Brookins reported that many fission products never left the host mineral. Several others 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.

Be careful with the good news, though. There is a real exception. Brookins also found that xenon and krypton, the noble gases, had largely escaped the reactor zones. That is not a footnote. It is the same escape that made the xenon dating possible. Oklo held nearly everything. It did not hold everything.

So How Did Earth Run Its Own Nuclear Reactors, And Could It Happen Again?

Line the pieces up and the impossible story becomes almost inevitable.

Two billion years ago, natural uranium was about 3.6% U-235. That is modern reactor fuel. Cyanobacteria had oxygenated the air. That let water dissolve uranium and pile it into thick ore bodies. Those ore bodies happened to be nearly free of neutron-eating elements. And groundwater ran through them, slowing neutrons enough to close the loop. 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. U-235 keeps decaying six times faster than U-238. It has fallen from 3.6% to 0.720%, and it is still falling. The fuel is not hidden or hard to reach. It is gone. Every uranium deposit on Earth now sits below the threshold for a natural chain reaction. That also answers a question people often ask the other way round. The planet has plenty of uranium left. None of it is rich enough to light itself.

That leaves the part that is easiest to forget. Oklo ran for hundreds of thousands of years, went quiet, and waited two billion years for someone to notice. We noticed in 1972. One technician saw 0.717% where 0.720% should have been. They 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.