The damaged Fukushima Daiichi power plant. Credit: AP Photo/AIR PHOTO SERVICE

It came as no surprise when the Tokyo Electric Power Company (TEPCO) admitted last week that it will scrap its stricken Fukushima Daiichi reactors. After explosions, copious radioisotope leaks and a liberal dousing with sea water, the reactors are a write-off. But what will workers encounter when they finally start decommissioning the shattered plant?

On 11 March, a tsunami knocked out backup generators, preventing cooling water from circulating around the hot cores of reactors 1–3. The fuel rods inside began to warp, split and at least partially melt. Steam reacted with the rods' outer sheath of zirconium, creating hydrogen gas that caused a sequence of explosions (see Nature 471, 417–418; 2011).

But data from Japanese regulators and TEPCO suggest to some researchers that conditions inside the core could be far worse than a partial meltdown. Some believe that molten fuel may have flowed into the outer concrete containment vessel, whereas others suggest that nuclear chain reactions are still happening inside the fuel.

The most worrisome evidence comes from water found in a building next to reactor 1. On 26 March, Japan's Nuclear and Industrial Safety Agency reported the presence of chlorine-38, a radioisotope with a half-life of just 37 minutes that forms when natural chlorine-37 is hit by neutrons from fission. This could be evidence that fuel has clumped together into sufficiently large chunks to briefly restart nuclear reactions, says Ferenc Dalnoki-Veress, a physicist at the Monterey Institute of International Studies in California. Such bursts could put workers at extreme risk of radiation exposure during clean-up, he warns, and seriously complicate work at the site.

But Paddy Regan, a nuclear physicist at the University of Surrey in Guildford, UK, is sceptical. Other radioisotopes have a similar γ-ray spectrum and could be mimicking chlorine-38, he says, and TEPCO has already retracted erroneous measurements of other isotopes. Dalnoki-Veress agrees that the evidence is circumstantial, adding that he is frustrated by the lack of clear data coming from the plant.

Other theories also rest on tentative evidence. Richard Lahey, an emeritus professor of nuclear engineering at Rensselaer Polytechnic Institute in Troy, New York, believes that the core of reactor 2 may have melted its control rods, which are designed to stop nuclear reactions. Provisional pressure readings and high levels of radioactivity suggest to him that molten fuel has flowed through the control-rod system like lava and dripped into the containment vessel below, creating a clean-up nightmare.

The confusion recalls the weeks that followed the partial meltdown of a reactor at Three Mile Island in Pennsylvania in 1979. In the immediate aftermath of that emergency, the state of the reactor core was subject to "an ongoing debate that went on for months", recalls Jack DeVine, an independent nuclear consultant who spent six years cleaning up that accident.

Many scientists believed that the fuel rods at Three Mile Island were more-or-less intact, on the basis of computer models and simulations, says DeVine. But when a camera was finally lowered into the core in 1982, the damage was far worse than anyone had predicted. "It looked like my gravel driveway — a mess," he says. Engineers hoping to remove fuel rods in a process akin to conventional decommissioning of a nuclear core had to rethink their strategies.

It took 14 years to clear most of the fuel out of the reactor at Three Mile Island. Based on what he has seen so far, DeVine believes that decommissioning Fukushima will probably take longer.