The state and future of nuclear waste

Start with what the material is, because the public image is wrong in a way that matters. Used fuel comes out solid, not as liquid goo: uranium dioxide, a ceramic, which Shafer compares to a coffee mug. Roughly 95% of it is still uranium. The remaining 5% divides into about 4% fission products, the pieces left when a uranium nucleus splits, and about 1% actinides such as plutonium, neptunium and americium, formed when a uranium nucleus captures a neutron instead of splitting. That small actinide slice is what creates the long-term management burden. Strip it out and most of what remains is a several-hundred year problem rather than a multi-million-year one, with a few standouts she names, technetium-99 and an iodine isotope she does not pin down, that keep long clocks of their own. She is careful to say that the long timeline is partly a choice rather than a physical necessity. The rule that whoever generates the material manages it for the long haul is, in her words, a social construct, and because the US separates nothing, everything stays in the fuel together and the whole package inherits the longest clock in it.

That framing is what makes the recycling argument bite, and it is the one place where host and guest pull in different directions. Shayle puts the common position: recycling is a way to need less freshly enriched uranium, not a solution to waste, and reacting the recovered uranium just produces more waste products. Shafer partly disagrees. In the open cycle the US runs, each load of fuel is irradiated once and fresh material is fed in behind it, so you end up with roughly double the material you would have needed had you reprocessed the first load. Recycling reuses the same stock, which she argues does minimize the radioactive waste generated, while granting Shayle’s underlying point that you still have to manage whatever you make. The stronger version goes further: actinides can be put back into a reactor and split into fission products, which mostly have shorter half-lives, and enough passes through that loop substantially shrinks the long-lived inventory. She attaches two conditions immediately. The loop is most plausible with fast reactors, and it would be very difficult to run on the current light water fleet. If it works, the payoff is a change of category rather than a change of degree: instead of managing something on an inter-civilizational basis, you are managing something for a few hundred years, a span over which we have demonstrably built buildings, and possibly something that could go into surface disposal rather than deep underground.

New reactor designs complicate this, and the variable that matters is fuel type more than reactor type, though the two are coupled. The criterion is how good a waste form the fuel makes, meaning how well it holds its radioactive contents once it is sitting somewhere for a very long time. Uranium dioxide ceramic is excellent, and so is TRISO, for the same reason it is liked as a fuel: nothing gets out. Metallic fuels oxidize, and fuel dissolved in a molten salt is poor, because salt does not hold up to water. Sodium-bonded fuel carries an element covered by federal hazardous-waste rules, which you would rather not put in a repository at all. So high-temperature gas reactors and molten salt reactors that use TRISO present no waste problem; molten salt reactors with fuel dissolved in the core, and sodium fast reactors such as TerraPower and Oklo, need conditioning at minimum; and light water designs have established pathways. This loops back to recycling: if you have to treat and condition the fuel anyway, Shafer notes, you are a significant part of the way to reprocessing it, which is why companies facing conditioning often open the recycling conversation.

None of that is the reason the US has no repository. The country chose deep geological disposal in the 1970s over alternatives including launching the material into space and dropping it in the Mariana Trench, legislated Yucca Mountain, and then stopped funding it around 2009 amid lawsuits, many from Nevada, without repealing it, so Yucca remains the law of the land and the system sits in a holding pattern. Shafer’s explanation for the failure is a pattern she says recurs: local communities are frequently supportive and the federal government is supportive, and the state is not. Her generalization is that regulatory frameworks are an extrapolation and formalization of social license, which is why she treats consent as the thing to engineer. The evidence she offers is survey work finding that asking a state to host a repository alone gets a poor reception, adding a recycling facility improves it, and adding a national laboratory improves it further, because the state becomes an energy provider and a jobs story rather than a dump. Finland is her example of a country that built the social structures to get a yes. Behind all of this sits arithmetic that a build-out makes worse: Yucca’s legal limit was 70,000 metric tons against the 90,000 already accumulated, so it would be nominally full on opening, and at three or four times today’s fleet the country would generate six to eight thousand tons a year rather than two thousand. On a rough heuristic of 80,000 tons per repository, that is filling one every decade, which she expects would be socially very hard. Recycling is her route back to a one-repository regime, and deep boreholes are the wildcard that might let a state manage its own material instead.