Temps de lecture :10 minutes
Earlier this year, James Lovelock, the originator of the Gaia Hypothesis, argued in his book The Revenge of Gaia that the threat of climate change is so real, so advanced and potentially so catastrophic that the risks associated with nuclear power are trivial by comparison – and that there really is no alternative to its widespread use. Nuclear power, he insisted, is the only large-scale option: it is feasible and practical; a nuclear renaissance is needed without delay. He robustly dismissed the idea that the growth of nuclear power was likely to be constrained by depletion of its raw material. This is how he put it:
“Another flawed idea now circulating is that the world supply of uranium is so small that its use for energy would last only a few years. It is true that if the whole world chose to use uranium as its sole fuel, supplies of easily-mined uranium would soon be exhausted. But there is a superabundance of low-grade uranium ore: most granite, for example, contains enough uranium to make its fuel capacity five times that of an equal mass of coal. India is already preparing to use its abundant supplies of thorium, an alternative fuel, in place of uranium. (1) “
Lovelock added that we have a readily-available stock of fuel in the plutonium that has been accumulated from the reactors that are shortly to be decommissioned. And he might also have added that another candidate as a source of nuclear fuel is seawater. So, if we put the supposed alternatives to uranium ore in order, this is what we have: (1) granite; (2) fast-breeder reactors using (a) plutonium and (b) thorium; and (3) seawater.
It has already been explained above that granite with a uranium content of less than 200 parts per million (0.02%) cannot be used as a source of nuclear energy, because that is the borderline at which the energy needed to mill it and to separate the uranium oxide for enrichment is greater – and in the case of even poorer ores, much greater – than the energy that you get back. But Lovelock is so insistent and confident on this point that it is worth revisiting.
Storm van Leeuwen, basing his calculations on his joint published work with Smith on the extraction of uranium from granite, considers how much granite would be needed to supply a 1 GW nuclear reactor with the 160 tonnes of natural uranium it would need for a year’s full-power electricity production. Ordinary granite contains roughly 4 grams of uranium per tonne of granite. That’s four parts per million. One year’s supply of uranium extracted from this granite would require 40 million tonnes of granite. So, Lovelock’s granite could indeed be used to provide power for a nuclear reactor, but there are snags. The minor one is that it would leave a heap of granite tailings (if neatly stacked) 100 metres high, 100 metres wide and 3 kilometres long. The major snag is that the extraction process would require some 530 PJ (petajoules = 1,000,000 billion joules) energy to produce the 26 PJ electricity provided by the reactor. That is, it would use up some 20 times more energy that the reactor produced. (2)
2. Fast breeder reactors
Lovelock’s proposal that we should use plutonium as the fuel for the nuclear power stations of the future can be taken in either of two ways. He might be proposing that we could simply run the reactors on plutonium on the conventional “once-through” system which is standard, using light-water reactors. This can certainly be done, but it cannot be done on a very large scale. Plutonium does not exist in nature; it is a by-product of the use of uranium in reactors and, when uranium is no longer used, then in the normal course of things no more plutonium will be produced. There is enough reactor-grade plutonium in the world to provide fuel for about 80 reactors. That is just about realistic, but there are another two theoretical but highly unrealistic possibilities. The first is that all weapons-grade plutonium could be converted into enough fuel for about 60 more reactors; the second is that all the spent fuel produced by all nuclear power stations in the world could be successfully reprocessed (despite the substantial failure and redundancy of reprocessing technology at present) and used to provide the fuel for the reactors of the future. That would provide fuel for another 600 reactors – making a total of 740 operating with plutonium alone. (3)
But since we’re trying to be realistic here, let us concentrate on what could actually be done, and stay as close as we can to what Lovelock seems to be suggesting: we could, using the plutonium that we actually have, build 80 reactors worldwide. At the end of their life (say, 24 full-power years), the plutonium would have been used up, though supplemented by a little bit over from the final generation of ordinary uranium-fuelled reactors, but soon all reactors would be closed down and not replaced, because at that time there will be no uranium to fuel them with, either. This would scarcely be a useful strategy, so it is more sensible to suppose that Lovelock has in mind the second possibility: that the plutonium reactors should be breeder reactors, designed not just to produce electricity now, but to breed more plutonium for the future.
Breeders are in principle a very attractive technology. In uranium ore, a mere 0.7 percent of the uranium it contains consists of the useful isotope – the one that is fissile and produces energy – uranium-235. Most of the uranium consists of uranium-238, and most of that simply gets in the way and has to be dumped at the end; it is uranium-238 which is responsible for much of the awesome mixture of radioactive materials that causes the waste problem. And yet, uranium-238 does also have the property of being fertile. When bombarded by neutrons from a “start-up” fuel like uranium-235 or plutonium-239, it can absorb a neutron and eject an electron, becoming plutonium-239. That is, plutonium-239 can be used as a start-up fuel to produce more plutonium-239, more-or-less indefinitely. That’s where the claim that nuclear power would one day be too cheap to meter comes from.
But there is a catch. It is a complicated technology. It consists of three operations: breeding, reprocessing and fuel fabrication, all of which have to work concurrently and smoothly. First, breeding: this does not simply convert uranium-238 to plutonium-239; at the same time, it produces plutonium-241, americium, curium, rhodium, technetium, palladium and much else. This mixture tends to clog up and corrode the equipment. There are in principle ways round these problems, but a smoothly-running breeding process on a commercial scale has never yet been achieved. (4)
Secondly, reprocessing. The mixture of radioactive products that comes out of the breeding process has to be sorted, with the plutonium-239 being extracted. The mixture itself is highly radioactive, and tends to degrade the solvent, tributyl phosphate. Here, too, insoluble compounds form, clogging up the equipment; there is the danger of plutonium accumulating into a critical mass, setting off a nuclear explosion. The mixture gets hot and releases radioactive gases; and significant quantities of the plutonium and uranium are lost as waste. As in the case of the breeder operation itself, a smoothly-running reprocessing process on a commercial scale has never yet been achieved.
The third operation is to fabricate the recovered plutonium as fuel. The mixture gives off a great deal of gamma and alpha radiation, so the whole process of forming the fuel into rods which can then be put back into a reactor has to be done by remote control. This, too has yet to be achieved as a smoothly-running commercial operation.
And, of course, it follows from this, that the whole fast-breeder cycle, consisting of three processes none of which have ever worked as intended, has itself never worked. There are three fastbreeder rectors in the world: Beloyarsk-3 in Russia, Monju in Japan and Phénix in France; Monju and Phénix have long been out of operation; Beloyarsk is still operating, but it has never bred. But let us look on the bright side of all this. Suppose that, with 30 years of intensive research and development, the world nuclear power industry could find a use for all the reactor-grade plutonium in existence, fabricate it into fuel rods and insert it into newly-built fast-breeder reactors – 80 of them, plus a few more, perhaps, to soak up some of the plutonium that is being produced by the ordinary reactors now in operation. So: they start breeding in 2035. But the process is not as fast as the name suggests (“fast” refers to the speeds needed at the subatomic level, rather than to the speed of the process). Forty years later, each breeder reactor would have bred enough plutonium to replace itself and to start up another one. By 2075, we would have 160 breeder reactors in place. And that is all we would have, because the ordinary, uranium-235-based reactors would by then be out of fuel. (5)
The safety/cost trap
The complexity of in-depth defence against accident can make the system impossible
There is a systemic problem with the design of breeder reactors. The consequences of accidents are so severe that the possibility has to be practically ruled out under all circumstances. This means that the defence-in-depth systems have to be extremely complex, and this in turn means that the installation has to be large enough to derive economies of scale – otherwise it would be hopelessly uneconomic. However, that means that no confinement dome, on any acceptable design criterion, can be built on a scale and structural strength to withstand a major accident. And that in turn means that the defence-in-depth systems have to be even more complex, which in turn means that they becomes even more problem-prone than the device they were meant to protect. A study for the nuclear industry in Japan concludes: “A successful commercial breeder reactor must have three attributes: it must breed, it must be economical, and it must be safe. Although any one or two of these attributes can be achieved in isolation by proper design, the laws of physics apparently make it impossible to achieve all three simultaneously, no matter how clever the design.” (6)
The other way of breeding fuel is to use thorium. Thorium is a metal found in most rocks and soils, and there are some rich ores bearing as much as 10 percent thorium oxide. The relevant isotope is the slightly radioactive thorium-232. It has a half-life three times that of the earth, so that makes it useless as a direct source of energy, but it can be used as the starting-point from which to breed an efficient nuclear fuel. Here’s how:
* Start by irradiating the thorium-232, using a start-up fuel – plutonium-239 will do. Thorium-232 is slightly fertile, and absorbs a neutron to become thorium 233.
* The thorium-233, with a half-life of 22.2 minutes, decays to protactinium-233.
* The protactinium-233, with a half-life of 27 days, decays into uranium-233.
* The uranium-233 is highly fissile, and can be used not just as nuclear fuel, but as the start-up source of irradiation for a blanket of thorium-232, to keep the whole cycle going indefinitely. (7)
But, as is so often the case with nuclear power, it is not as good as it looks. The two-step sequence of plutonium breeding is, as we have seen, hard enough. The four-step sequence of thorium-breeding is worse. The uranium-233 which you get at the end of the process is contaminated with uranium-232 and with highlyradioactive thorium-228, both of which are neutron-emitters, reducing its effectiveness as a fuel; it also has the disadvantage that it can be used in nuclear weapons. The comparatively long half-life of protactinium-233 (27 days) makes for problems in the reactor, since substantial quantities linger on for up to a year. Some reactors – including Kakrapar-1 and -2 in India – have both achieved full power using some thorium in their operation, and it may well be that, if there is to be a very long-term future for nuclear fission, it will be thorium that drives it along. However, the full thorium breeding cycle, working on a scale which is large enough and reliable-enough to be commercial, is a long way away. (8)
For the foreseeable future, its contribution will be tiny. This is because the cycle needs some source of neutrons to begin. Plutonium could provide this but (a) there isn’t very much of it around; (b) what there is (especially if we are going to do what Lovelock urges) is going to be busy as the fuel for once-through reactors and/or or fast-breeder reactors, as explained above; and (c) it is advisable, wherever there is an alternative, to keep plutonium-239 and uranium-233 – an unpredictable and potentially incredibly dangerous mixture – as separate as possible. It follows that thorium reactors must breed their own start-up fuel from uranium-233. The problem here is that there is practically no uranum-233 anywhere in the world, and the only way to get it is to start with (say) plutonium-239 toget one reactor going. At the end of forty years, it will have bred enough uranium-233 both to get another reactor going, and to replace the fuel in the original reactor. So, as in the case of fastbreeders, we have an estimated 30 years before we can perfect the process enough to get it going on a commercial scale, followed by 40 years of breeding. Result: in 2075, we could have just two thorium reactors up and running. (9)
Seawater contains uranium in a concentration of about thirty parts per billion, and advocates of nuclear power are right to say that, if this could be used, then nuclear power could in principle supply us with the energy we need for a long time to come. Ways of extracting those minute quantities of uranium from seawater and concentrating them into uranium oxide have been worked out in some detail. First of all, uranium ions are attracted – “adsorbed” – onto adsorption attracted – “adsorbed” – onto adsorption beds consisting of a suitable material such as titanium hydroxide, and there are also some polymers with the right properties. These beds must be suspended in the sea in huge arrays, many kilometres in length, in places where there is a current to wash the seawater through them, and where the sea is sufficiently warm – at least 20°C. They must then be lifted out of the sea and taken on-shore, where, in the first stage of the process, they are cleansed to remove organic materials and organisms. Stage two consists of “desorption” – separating the adsorbed uranium ions from the beds. Thirdly, the solution that results form this must be purified, removing the other compounds that have accumulated in much higher concentration than the uranium ions. Fourthly, the solution is concentrated, and fifthly, a solvent is used to extract the uranium. The sixth stage is to concentrate the uranium and purify it into uranium oxide yellowcake, ready for enrichment in the usual way. (10)
But the operation is massive and takes a lot of energy. Very roughly, two cubic kilometres of sea water is needed to yield enough uranium to supply one tonne, prepared and ready for action in a reactor. A 1 GW reactor needs about 160 tonnes of natural uranium per annum, so each reactor requires some 324 cubic kilometres of seawater to be processed – that is, some 32,000 cubic kilometres of seawater being processed in order to keep a useful fleet of 100 nuclear reactors in business for one (full-power) year. (11)
And what is the energy balance of all this? One tonne of uranium, installed in a light water reactor, is taken as a rule-of-thumb also to produce approximately 162 TJ (1 terajoule = 1,000 billion joules), less the roughly 60-90 TJ needed for the whole of the remainder of the fuel cycle – enrichment, fuel fabrication, waste disposal, and the deconstruction and decommissioning of the reactor – giving a net electricity yield of some 70-90 TJ. The energy needed to supply the uranium from seawater, ready for entry into that fuel cycle, is in the region of 195-250 TJ. In other words, the energy required to operate a nuclear reactor using uranium derived from seawater would require some three times as much energy as it produced.
1. Lovelock (2006), p 103.
2. e.g. S. Huwyler, L Rybach and M Taube (1975), “Extraction of uranium and thorium and other metals from granite“, EIR-289, Technical Communications 123, Eidgenossische Technische Hochschule, Zurich, September, translated by Los Alamos Scientific Laboratory, LA-TR-77-42, 1977). Cited and discussed in Storm van Leeuwen (2006), “Uranium Resources and Nuclear Energy”, Appendix E, in WSL/IPCC
3. Storm van Leeuwen (2006), “Breeders”, Appendix C, in WSL/IPCC.
6. Lawrence M. Lidsky and Marvin M Miller (1998), “Nuclear Power and Energy Security“: A Revised Strategy for Japan”, at
7. Uranium Information Council (2004), Briefing Paper 67, “Thorium”, at
9. Storm van Leeuwen (2006), “Breeders”, Appendix C, in WSL/IPCC.
10. Storm van Leeuwen (2006), “Uranium from Seawater”, Appendix E2, in WSL/IPCC.