Oklo, the natural nuclear reactor in Gabon

A natural fission reactor needs no engineers. It only needs uranium concentrated enough, water to moderate the neutrons, and a geometry that does not let too many of them escape. Those three conditions came together at Oklo, in Gabon, about 2 billion years ago, and nowhere else that has been found on Earth. The key is uranium-235: today it makes up just 0.72% of natural uranium, too little to sustain a chain reaction without enriching it. But U-235 decays faster than U-238 — its half-life is about 700 million years, against 4.5 billion for U-238 — so the further back in time you go, the higher its share. Two billion years ago it stood at roughly 3 to 3.7%, close to the enrichment of fuel in a light-water reactor today. When groundwater soaked into a uranium seam rich and thick enough, at Oklo the uranium did, with no one designing it, what a PWR does now.

The phenomenon was found by accident. In June 1972, at the Pierrelatte enrichment plant in southern France, a routine mass-spectrometry measurement found that a batch of uranium hexafluoride carried 0.7171% U-235 instead of the expected 0.7202%. The gap was tiny, but it was enough for the French Atomic Energy Commission (CEA) to open an investigation. Once contamination and recycled spent fuel were ruled out, the trail led from Pierrelatte to the conversion plant at Malvési and from there to a mine run by the Compagnie des Mines d'Uranium de Franceville (COMUF), at Oklo, near Franceville. Some ore batches ran below 0.5% U-235, and between 1970 and 1972 the cumulative U-235 deficit in the 700 tonnes of uranium delivered to the concentration plant at Mounana topped 200 kilograms. The only explanation consistent with the fission-product spectrum found in the ore was that, at some point in the deposit's geological history, it had fissioned uranium on its own.

Follow-up surveys, coordinated by the CEA and later opened to the IAEA, identified several reaction zones inside the Oklo deposit itself, plus one more, the best preserved, some 30 km to the southeast at the Bangombé deposit. The studies do not agree on the exact total: a 2002 survey counts fourteen zones at Oklo plus one at Bangombé, fifteen in all, while a 2015 one counts sixteen plus one, seventeen. Each zone is a lens of uraninite from a few centimetres to half a metre thick and several metres long, at depths of 100 to 260 m at Oklo and just 12 m at Bangombé. Operation was not continuous: the heat of fission boiled off the water, the loss of moderator throttled the reaction, water seeped back in as the rock cooled, and the cycle repeated. An analysis of fissiogenic xenon trapped in phosphate crystals, published in Physical Review Letters in 2004, measured active pulses of about 30 minutes separated by roughly two and a half hours of dormancy. Kept up intermittently like that for 100,000 to 500,000 years, by the most cited estimates, the reactors together released a total energy on the order of 10,000 megawatt-years: an average power, spread across every zone, of only tens of kilowatts, tens of thousands of times less than the thermal power of a modern commercial reactor.

Oklo matters today mainly as a natural analogue for the geological disposal of radioactive waste. Inside the core of each reaction zone, plutonium, thorium, rare earths and elements such as zirconium, ruthenium, rhodium and palladium stayed almost entirely locked into the uraninite crystals and the surrounding clays, despite 2 billion years of groundwater circulation. Other fission products, such as caesium, rubidium, strontium and barium, were almost completely lost from the immediate surroundings because they are soluble or gaseous. The plutonium-239 formed by neutron capture in uranium-238 itself decayed away, with its 24,000-year half-life, long before it could migrate far. No serious body presents Oklo as proof that a geological repository will contain waste: the comparison has limits, because the reaction zones ran far hotter, for far longer, and without the engineered barriers designed into repositories today. But it offers an unusual, qualitative confidence that geology, unaided, can hold heavy elements in place for periods that exceed by several orders of magnitude the hundred thousand years a repository such as Onkalo is required to guarantee.

The reaction zones were not kept intact for science. COMUF, a company controlled by the French Cogema (now Orano) and the Gabonese state, mined the Oklo deposit by open pit from the 1970s and then underground from 1977 to 1997, alongside three other deposits in the Franceville basin. Between 1961 and 1999 it produced close to 28,000 tonnes of uranium in all, more than 14,000 from Oklo alone. Nearly every reaction zone was wholly or partly destroyed by mining, and the workings, now flooded, are no longer accessible. Uranium mining in Gabon ended in 1999 for lack of economic reserves, and the site was remediated in the years that followed. The exception is Bangombé, whose deposit was too small to mine: its reaction zone, the shallowest and the most studied, remains in place, and drill-core samples from the other zones are kept in laboratories to study how a nuclear fuel behaves after two billion years underground.