Nature’s Nuclear Reactor: How Uranium in Gabon Ran Its Own Chain Reaction Two Billion Years Ago

In May 1972, an analyst named H. Bouzigues was running a routine check at a French plant in Pierrelatte that prepared uranium for nuclear fuel.  His job was to measure how much of the uranium was the useful kind.  Natural uranium is a mix of two main isotopes, or versions of the same element.  Almost all of it is uranium-238.  Only a tiny share, 0.720 percent, is uranium-235, the isotope that can split and keep a chain reaction going.  That number is about as close to a constant as nature gets.  It is essentially the same in uranium ore all over the Earth, and even in samples from the Moon and from meteorites.

Bouzigues got 0.717 percent.  It sounds like nothing, a difference in the third decimal place.  But in the nuclear business, where every gram of fissile material is supposed to be accounted for, it was alarming.  Either someone had made a mistake, or some uranium-235 had gone missing.  As investigators traced the uranium back through the supply chain, the shortfall grew.  By mid-1972, roughly 200 kilograms of uranium-235 seemed to have vanished from ore shipments, enough by one estimate to build half a dozen nuclear bombs.

The trail led to a single source, the Oklo uranium mine near the town of Mounana in southeastern Gabon, in west-central Africa.  And the explanation turned out to be stranger than theft.  Nobody had taken the missing uranium-235.  It had been burned up, split apart in nuclear chain reactions that started on their own, deep in the rock, about two billion years ago.  Long before Enrico Fermi built the first human-made reactor in 1942, nature had already done it, and then left the evidence lying around for the world’s nuclear industry to dig up.

A missing-uranium mystery

The French Atomic Energy Commission, known by its French initials CEA, took the anomaly seriously.  The first suspicion was contamination, that uranium already stripped of some of its uranium-235 in an enrichment plant had somehow been mixed back in.  That idea did not hold up.  Samples had been kept from every batch of ore processed at the mill in Mounana, and they showed that slightly depleted uranium had been arriving since 1970.  Drill cores from the deposit, some stored in France, settled it.  A few were badly short of uranium-235.  One core from the part of the ore body then being mined held just 0.44 percent, close to 40 percent below normal.

Chemistry cannot do that.  Uranium-235 and uranium-238 behave almost identically in chemical reactions, so ordinary geology can concentrate uranium or scatter it, but it cannot easily sort one isotope from the other.  Something nuclear had happened.  And there was a simple test.  When a uranium-235 atom splits, it breaks into lighter elements in a recognizable pattern, so if the ore had hosted a chain reaction, those fission products should still be there.

They were.  CEA chemists measured rare earth elements in the ore, especially neodymium.  Natural neodymium is about 27 percent neodymium-142, an isotope that fission does not make.  In the Oklo samples, neodymium-142 was under 6 percent, swamped by the isotopes that fission does make, in just the proportions a reactor would produce.  Ruthenium told the same story.  On September 25, 1972, the CEA announced that self-sustaining nuclear chain reactions had taken place in Gabon, on their own, in the distant past.  Researchers eventually identified 16 reaction zones in the Oklo and neighboring Okélobondo deposits, plus another at Bangombé, about 35 kilometers away.

Someone had predicted it

The idea was not entirely new.  In 1953, George Wetherill of UCLA and Mark Inghram of the University of Chicago had suggested that some uranium deposits might once have worked like natural reactors.  In 1956, Paul Kuroda, a Japanese-born chemist at the University of Arkansas, spelled out in a short paper exactly what it would take.  He listed four conditions.

First, the uranium vein had to be big enough, thicker than the average distance a neutron travels before it splits another atom, about two-thirds of a meter, so the neutrons would not simply escape.  Second, the uranium had to be rich enough in uranium-235.  Third, there had to be a moderator, something to slow the neutrons down, because slow neutrons are much better at splitting uranium-235.  Water works well.  Fourth, the rock had to be mostly free of “poisons” such as boron and lithium, which soak up neutrons and smother a chain reaction.  At Oklo, all four conditions came together.  Kuroda lived to see his prediction confirmed, and he spent his later years in Las Vegas, where he died in 2001.

Why it could happen then, and not now

The second condition is the key to the whole story.  Uranium-235 is radioactive and decays faster than uranium-238.  Its half-life is about 700 million years, compared with about 4.5 billion for uranium-238.  So the further back in time you go, the larger the share of uranium-235 in natural uranium.  Today it is 0.72 percent, too low for ordinary water to keep a chain reaction going.  Two billion years ago, it was about 3 percent, roughly the level that engineers now reach by enriching uranium for the fuel in most of the world’s nuclear power plants.  In other words, nature’s uranium came pre-enriched.

That is also why the Oklo reactors cannot happen today.  There is simply not enough uranium-235 left in natural ore.  Today’s reactors that run on natural uranium, like Canada’s CANDU design, need heavy water, a rarer form of water with a heavier hydrogen, which absorbs fewer neutrons.  Fermi’s first pile in Chicago used ultra-pure graphite for the same reason.  Ordinary groundwater worked in Gabon only because the clock was set two billion years earlier.

Oxygen made the ore

There is another reason the timing worked, and it has to do with life.  Uranium only dissolves well in water that carries oxygen.  For roughly the first half of Earth’s history, the air had almost no free oxygen, so uranium mostly stayed locked in place in the rocks where it formed.  Then, starting around 2.4 billion years ago, microbes that made their energy from sunlight began pumping oxygen into the atmosphere and oceans, in what geologists call the Great Oxidation Event.

With oxygen in the groundwater, uranium could be dissolved out of rocks, carried along, and then dropped again wherever conditions changed, for example where the water met carbon-rich layers that stripped the oxygen back out.  Over time, that process piled uranium into rich, concentrated seams.  At Oklo, some pockets of ore held as much as 20 percent uranium, while the deposit as a whole averaged well under 1 percent.  So the world’s first known nuclear reactors were, in an indirect way, made possible by microbes.  The same Franceville Basin also holds stromatolites, layered mounds built by microbial mats, and some of the oldest fossils ever proposed as possible complex life, about 2.1 billion years old.

A reactor that ran like a geyser

Once the chain reactions started, a big question remained.  Why didn’t they run away and destroy themselves?  The answer, worked out over the next three decades, is that the reactors regulated themselves with water.  Groundwater soaked into the uranium seam and slowed the neutrons, and the chain reaction started.  The reaction heated the rock, up to a few hundred degrees Celsius, until the water boiled off or was driven away.  Without water to slow the neutrons, the reaction faded.  When the rock cooled, water seeped back in, and the cycle began again.

In 2004, physicist Alex Meshik and colleagues at Washington University in St.  Louis worked out the schedule from xenon, a heavy noble gas made by fission that can stay trapped in minerals for billions of years.  Using lasers to release gas from single grains of a tiny rock fragment, just four millimeters across, they found that one Oklo reactor ran for about 30 minutes, then shut down for at least two and a half hours, over and over.  Meshik compared it to a geyser that heats up, blows off its water, refills and repeats.  Surprisingly, most of the xenon was not in the uranium minerals at all, but in grains of aluminum phosphate that had grown in the hot water, the highest concentration of xenon ever found in a natural material.

Modern engineers build the same trick into reactors on purpose.  In a light-water reactor, if the water boils away or turns to steam, the chain reaction slows down, a safety feature called a negative void coefficient.  Today’s new small modular reactor projects lean on that kind of built-in physics, too.  At Oklo, nobody designed it.  The rock and the water simply found a balance and kept it for a very long time.

How big was it?

Not very, at least in power.  The Oklo reactors were not a nuclear explosion or even a big power plant.  By most estimates, their average output was under 100 kilowatts, which Meshik described as enough to run a few dozen toasters.  What was remarkable was how long they kept going, on and off for a few hundred thousand years.  Over that time they burned through around five tons of uranium-235 and released about 15,000 megawatt-years of energy.

They also made plutonium.  Some neutrons were captured by uranium-238, which turned into plutonium-239.  More than two tons of it formed inside the deposit.  Plutonium-239 has a half-life of about 24,000 years, so it is long gone now, most of it having decayed back into uranium-235.  Some of it split, too, and left its own fission products as fingerprints.

A two-billion-year test of nuclear waste storage

For scientists, Oklo turned out to be much more than a curiosity.  It is a natural experiment in one of the hardest problems in nuclear power: how to store radioactive waste for a very long time.  The Oklo reactors produced the same kinds of fission products as a modern reactor, and then nobody touched them for nearly two billion years.  When researchers looked at where those leftovers ended up, most of the non-volatile fission products, along with heavy elements such as plutonium, had moved only centimeters from where they formed.  Some more mobile elements did wander, which is useful to know, too.

That makes Oklo a kind of preview for deep geological repositories, the underground vaults that countries such as Finland are building to hold spent nuclear fuel.  (For a sense of what conditions are like deep in the Earth’s crust, our story on the Kola Superdeep Borehole is a good place to start.) Meshik’s team also pointed out that the aluminum phosphate grains at Oklo held on to xenon for billions of years, a hint at how radioactive gases might someday be locked away.

Checking whether the laws of physics change

Oklo has one more strange use.  Physicists have long wondered whether the so-called constants of nature are truly constant.  One of the most important is the fine-structure constant, a number that sets the strength of the electromagnetic force.  If it had been even slightly different two billion years ago, certain nuclear reactions at Oklo would have run at different rates.

The best clue is samarium.  One isotope, samarium-149, is very good at capturing slow neutrons, and the Oklo reactors ate most of it.  In 1976, Soviet physicist Alexander Shlyakhter used the amount of samarium-149 left in the ore to show that the strength of that reaction, and therefore the fine-structure constant, could hardly have changed since the reactors ran.  Most later studies agreed, though a 2004 analysis by researchers at Los Alamos argued for a small change.  For a pile of rock in Central Africa, it remains one of the tightest checks on the stability of physics over geological time.

The only one we know of

Oklo is still the only place on Earth where natural reactors have been found.  That may not be because they were rare.  Two billion years ago, the conditions may have come together in other uranium deposits, too.  But very little of Earth’s crust is that old, and most of what is has been squeezed, melted or eroded so much that any record would be erased.  Meshik has suggested that a few telltale traces of xenon might one day help find another one.

In 1975, scientists from around the world met in Gabon at a conference organized with the International Atomic Energy Agency to compare notes on what they called “the Oklo phenomenon.”  Today, the site’s name lives on in an unexpected place: a U.S. company building small, advanced nuclear reactors, Oklo Inc., is named after it.  Central Africa has produced more than one natural surprise.  In 1986, at Lake Nyos in neighboring Cameroon, a crater lake suddenly released a cloud of carbon dioxide.  But Oklo may be the strangest.  Humanity likes to think it invented the nuclear age in 1942.  A seam of uranium in the African rainforest, a little groundwater and some help from microbes had beaten us to it by about two billion years.

See uranium’s glow, and read its story