The nuclear power industries of Europe and the US plan to invest at least $5 billion over the next decade in separation of uranium isotopes. David Fishlock looks at how the technology behind nuclear fuel has evolved.

The nuclear power industries of Europe and the US plan to invest at least $5 billion over the next decade in separation of uranium isotopes. David Fishlock looks at how the technology behind nuclear fuel has evolved.

One of the most chilling stories I have ever heard about nuclear technology running amok was told in the Washington DC office of the then US energy secretary. He led me to a framed photograph of lofty machines called gas centrifuges; reinforced plastic tubes about 13 metres tall, nicknamed ’redwoods’. They were used to spin hot, highly corrosive gas uranium hexafluoride at supersonic speed, in order to separate isotopes of uranium. Spinning at full speed, a centrifuge rotor - one of a small plantation - left its mounting and travelled down the length of an entire corridor before coming to rest against a wall. Had it blundered into another rotor it could have wreaked nuclear havoc across Tennessee.

Nuclear technology of every kind - weapons, power and medical - relies on separation of isotopes. This was first shown to be possible by Joseph John Thomson in 1913, using a mass spectrometer and the inert gas neon. Subsequently, Frederick Lindemann and Francis Aston identified other ways of separating the chemically identical constituents of most elements: by distillation or diffusion; by exploiting their different densities; and by using electromagnetic forces. All four of these principal methods reached the industrial development stage in the US during the second world war, under the tight secrecy of the Manhattan Project - the allied effort to make the first nuclear bomb. Until then, separation of uranium isotopes had been achieved only on microgram scale.

Gaseous diffusion
By 1944, gaseous diffusion had blossomed into a full-scale industrial process to separate fissile uranium-235 from several other, far more abundant isotopes. One US enrichment site occupied 44 acres. The procedure is a form of ultra-filtration, first demonstrated in principle in Oxford University’s Clarendon Laboratory using a kitchen colander. The Manhattan Project invested in two huge diffusion plants to enrich (or ’refine’) enough pure uranium-235 for its first nuclear weapon. After the war, the UK built such a plant at Capenhurst near Chester; smaller, but still big enough to need 11 cooling towers to disperse waste heat. The USSR, France and China all followed with big diffusion plants, and the US with its third. These early monsters, despite major refurbishment and upgrading, are now nearing the end of their useful lives. The UK plant and one in the US have already been dismantled. A conservative estimate suggests that Europe and the US will invest at least $5 billion over the next decade in new uranium enrichment facilities for nuclear fuel production.

Today, diffusion is seen as an unnecessarily expensive way of enriching uranium. In the US and France in particular, plans are laid to replace it with a more compact technology needing one-tenth or less of the electrical energy. Diffusion needs a lot of energy to force the gas through ultra-fine filters, then remove the heat of compression. Europe’s latest diffusion plant, the Eurodif facility at Tricastin in southern France, built in the 1970s, has a nameplate rating of 10,800 tonnes of separative work units per year (SWU/a). The SWU - pronounced ’swoo’ - is used to measure isotope separation, a process in which the product is not chemically different from the feedstock. The feedstock is uranium hexafluoride, usually called ’hex’, of exceptional purity, prepared from uranium tetrafluoride. Hex is a colourless solid at room temperature. It sublimes to form a highly reactive gas at 56 ?C. Although only mildly radioactive, it is a troublesome gas to handle industrially, but it is the only practical vehicle for most uranium isotope separation processes.

The filtration barrier or membrane through which the hex is pushed consists of micro-tubules of nickel about 0.01 ?m in diameter. Crucially, these are resistant to corrosion by hex at 85 ?C. Its manufacture has been one of the most closely guarded nuclear secrets. The idea is that the lighter isotope will travel very slightly faster through such a membrane. By recycling the gas through many stages, arranged in cascades, the concentration of uranium-235 can be raised from the naturally occurring 0.72 per cent up to five per cent, covering requirements for most commercial nuclear fuels. The Eurodif plant has 1,400 enrichment stages, of three different sizes, diminishing with the gas volume as the hex gets richer. On one visit to Capenhurst, the general manager likened control of his lumbering process to steering an elephant: ’you jump down and give it a kick’.

In the 1960s the UK planned to build a new and bigger diffusion plant at Capenhurst, to enrich fuel for a new generation of nuclear reactors. It was to have a capacity of 3,000 tonnes of SWU/a by 1980. At a late stage those plans were superseded by a technology which the UK had pursued in great secrecy from the late 1950s.

The US had explored centrifugation as a way of amplifying differences in isotope density from the early-1920s. First to succeed was Jesse Beams at the University of Virginia in 1934. Beams, a specialist in fast-rotating machinery, showed how the evaporative centrifuge principle proposed by Lindemann and Aston could be made to work using carbon tetrachloride vapour, separating the chlorine-35 and chlorine-37 isotopes. He used a horizontal steel rotor 28 cm long and 7.6 cm inside diameter.

Five years later Otto Hann and Fritz Strassmann demonstrated that uranium-235 was fissile, triggering the quest for a nuclear weapon. Beam’s research was one of several technologies hotly pursued by the Manhattan Project, but was abandoned in 1944 when diffusion was clearly well in the lead. Research continued in Germany, however, where it attracted a young Austrian engineer, Gernot Zippe. Zippe, working in the Soviet sector when Germany was partitioned after the war, was drafted to nuclear research and development in Georgia. His ideas for a much lighter and faster centrifuge transformed the technology.

In essence, Zippe’s centrifuge was a vertical cylinder spinning on a pin bearing and kept upright by a magnetic bearing. It was virtually friction-free. Zippe demonstrated it to the satisfaction of the Soviets - hitherto reliant on diffusion - who later authorised his repatriation to Austria. He then worked on it in Germany then with Beams in the US.

Early UK investigations of centrifugation included flipping matchsticks into the wall-of-death ride at Battersea Funfair, to try to understand how gas behaved in a fast-spinning rotor. Initially, Zippe’s design was not kept secret and became well-known to those at the forefront of isotope separation. But in 1960 the US prevailed upon the UK, Germany and Holland to classify their centrifuge technology. They believed it could become highly attractive to people seeking highly enriched uranium for nuclear weapons. These fears were to prove all too justified.

By 1967, however, Germany believed its centrifuge development was within sight of industrial scale-up, with a view to supporting an ambitious domestic nuclear-power programme. The German government knew that any attempt to go it alone would have serious international repercussions, so it talked to Holland about a bilateral agreement to build an enrichment plant on Dutch soil. The UK got wind of these talks and believed they could undermine its own plans to become an international supplier of fuel enrichment services. It proposed tripartite talks, which led to the 1970 Treaty of Almelo, under which the three nations pledged to collaborate in the commercial exploitation of the gas centrifuge.

Enter Urenco
And so Urenco was born. The commercial rationale for the Urenco venture was that the US had a near monopoly in enrichment supply which the three partners believed they could undercut in price. The technical rationale was that the three countries, having started research and development with Zippe’s ideas at the same time, would have made similar progress, albeit in different ways because of differing engineering cultures. (France desperately wanted to join but was deemed not to have the same technical qualifications.) Each of the three had a potential domestic market and export ambitions. The Urenco venture - soon nicknamed ’the troika’ - began life in 1971 as two companies: Urenco in England, the commercial HQ; and Centec in Germany, which developed, designed and built gas centrifuge enrichment capacity for Urenco. The tripartite Treaty of Almelo stipulated a high degree of secrecy on technical matters, particularly materials. This is easily explained. The separative power of a centrifuge increases as the fourth power of the peripheral speed of its rotor; ie. a 20 per cent increase in speed would more than double the output from a given rotor. The treaty restricted enrichment levels to five per cent.

In 1971, Dutch development centred on aluminium centrifuge rotors; and German development on the stronger maraging steels used by its aircraft industry. The UK was technically inferior with its glass-fibre reinforced plastic rotor but had its sights set on carbon-fibre reinforcement, which all recognised could leapfrog the others. This indeed proved the case - eventually. But it took much reorganisation of the original companies and much heated debate before ideas were pooled in a single centrifuge design. The first troika centrifuge, TC12, with its rotor of carbon fibre reinforced resin, took two decades to develop. Today, it accounts for over 60 per cent of Urenco’s enrichment capacity, on three sites. In a sound bite from one industry executive, these machines ’spin a thousand times in the blink of an eye’.

Urenco’s business is to provide enrichment services to companies operating nuclear reactors. These companies, or their agents, buy hex from five major suppliers, deliver it for enrichment, then pass enriched UF6 to fuel fabricators. The annual world market exceeds 60,000 tonnes of uranium and is forecast to exceed 65,000 tonnes by 2015. Uranium has no other significant market.

US fears of proliferation proved right when the secrecy of the project was breached in 1974. A Pakistani scientist employed in a Dutch research centre purloined design data on early centrifuges, and used it to enrich uranium for Pakistan’s nuclear weapons. Dr Abdul Qadeer Khan rose to head Pakistan’s nuclear project, and subsequently was found to have led the transfer of weapons technology to Iran, Libya and North Korea, in a massive breach of the international regime to deter the proliferation of nuclear weapons.

Urenco, meanwhile, has grown into a substantial and profitable enrichment company with a capacity exceeding 6,500t SWU/a on sites at Capenhurst, Almelo (Holland) and Gronau (Germany). It has delivered over 65 million SWU and has contracts with nuclear-power companies in 17 countries.

Centrifuges vs lasers
The US used its war-time gas diffusion plants to build an international business providing enrichment for nuclear reactors, but it also continued to explore the gas centrifuge, using fibre-reinforced designs of rotor. From an early stage American researchers aimed for a much larger and faster rotor than European research and development, albeit still using Zippe’s ideas. By the mid-1970s, ’jumbo’ centrifuges around 10 metres tall were in operation; by the 1980s machines with five times the capacity of Urenco’s TC12 design had been built.

In the 1980s, after a comparative study of two huge public investments in uranium enrichment, the US Government abruptly ended centrifuge funding and decided to focus on another technology: atomic vapour laser isotope separation (AVLIS). The idea was to use a powerful laser beam to separate isotopes of uranium in a single stage by selectively ionising them. Instead of hex, the feedstock was uranium vapour; at atomospheric pressure uranium melts at 1,150 ?C and boils at 3818 ?C. In the US AVLIS was heavily promoted by the Lawrence Livermore National Laboratory, a major Department of Energy research centre in California.

The Urenco partnership could not match the scale of these AVLIS research projects but needed to know the potential risk to its long-term (10-15 year) centrifuge plans. It launched a tripartite study in 1989. But by 1995 it was convinced that its centrifuge technology could hold its own against the formidable technical problems still facing AVLIS. Five years later the US Government reached the same conclusion, and reverted back to the centrifuge technology it had shelved in 1985. France, which had planned to replace gas diffusion with AVLIS, has recently negotiated a partnership with Urenco to use its gas centrifuge instead.

Current US plans are to use American centrifuge technology - essentially as it existed in 1985 - to build a commercial plant of 3,500t SWU/a capacity; costing somewhere between $1-1.5 billion. This would be preceded by a demonstrator cascade having up to 240 centrifuges standing 13 metres tall. They will spin faster than Urenco’s TC12 and are expected to have over seven times the capacity. Presently unknown, however, is the reliability of these machines and the cost of maintaining them. Over several decades, Urenco has demonstrated a level of reliability that allows the company simply to ignore its very rare breakdowns.

The only other uranium enrichment technology demonstrated on a substantial scale is the helikon process announced by South Africa in 1970; a derivation of the German jet-nozzle idea. It required even more power than diffusion and was shut down in 1995.

Worldwide, over 400 nuclear reactors require enriched fuel to generate electricity, and more are under construction. Major investments in uranium enrichment are planned today in the US, France, UK, Holland and Germany, all using gas centrifuge technology. At Portsmouth, Ohio, the US plans a $1.5 billion investment in the American Centrifuge, to bring 3,500 tonne SWU/a of capacity on-stream by 2010. In New Mexico, Urenco Inc. is seeking a licence to build a 3,000 tonne SWU/a facility to supply the US market.

In France, Urenco is discussing with AREVA, the French nuclear group, replacement of its gas diffusion capacity with centrifuge technology. Urenco is also continuing to expand capacity at its three enrichment plants at Capenhurst, Almelo and Gronau. The three totalled 6,500 tonnes of SWU/a at the end of 2003.


David Fishlock is a freelance scientific writer