Uranium Enrichment

Gaseous diffusion process

Commercial uranium enrichment was first carried out by the diffusion process in the USA. It has since been used in Russia, UK, France, China and Argentina as well.  It is a very energy-intensive process, requiring about 2400 kWh per SWU*. USEC says that electricity accounts for 70% of the production cost at its Paducah plant.

* It has been estimated that 7% of total US electricity demand was from enrichment plants at the height of the cold war, when 90% U-235 was required, rather than the reactor grades of 3-4 percent for power generation.
 

In recent years only the USA and France used the process on any significant scale, Russia phased it out in 1992. The remaining large USEC plant in the USA was originally developed for weapons programs and has a capacity of some 8 million SWU per year. It will be used to enrich some high-assay tails to 2013 before shutdown.  At Tricastin, in southern France, a more modern diffusion plant with a capacity of 10.8 million kg SWU per year has been operating since 1979 (see photo above). This Georges Besse I plant could produce enough 3.7% enriched uranium a year to fuel some ninety 1000 MWe nuclear reactors. It was shut down in mid 2012, after 33 years continuous operation. Its replacement (GB II - see below) has commenced operation.

In recent years the gaseous diffusion process has accounted for about 25% of world enrichment capacity. However, though they have proved durable and reliable, most gaseous diffusion plants are now nearing the end of their design life and the focus is on centrifuge enrichment technology which is replacing them.

 

Centrifuge process

The gas centrifuge process was first demonstrated in the 1940s but was shelved in favour of the simpler diffusion process. It was then developed and brought on stream in the 1960s as the second-generation enrichment technology. It is economic on a smaller scale, eg under 2 million SWU/yr, which enables staged development of larger plants.  It is much more energy-efficient than diffusion, requiring only about 50-60 kWh per SWU. 

The centrifuge process has been deployed at a commercial level in Russia, and in Europe by Urenco, an industrial group formed by British, German and Dutch governments. Russia's four plants at Seversk, Zelenogorsk, Angarsk and Novouralsk account for some 40% of world capacity*. Urenco operates enrichment plants in UK, Netherlands and Germany and is building one in the USA.

* In 20912 Russia was commissioning 8th generation centrifuges with service life of up to 30 years. The last 6th & 7th generation ones were installed in 2005, and 8th generation equipment has been supplied since 2004 to replace 5th generation models with a service life of only 15 years.
 

In Japan, JNC and JNFL operate small centrifuge plants, the capacity of JNFL's at Rokkasho was planned to be 1.5 million SWU/yr. China has two small centrifuge plants imported from Russia.  One at Lanzhou is 0.5 million SWU/yr and the other main one at Hanzhun is operating at 0.5 million SWU/yr and is being doubled in size.  Brazil has a small plant which is being developed to 0.2 million SWU/yr.  Pakistan has developed centrifuge enrichment technology, and this appears to have been sold to North Korea.  Iran has sophisticated centrifuge technology which is operational, with estimated 9000 SWU/yr capacity.

In both France and the USA plants with 6th generation Urenco centrifuge technology are now being built to replace ageing diffusion plants, not least because they are more economical to operate. As noted, a centrifuge plant requires as little as 50 kWh/SWU power (Urenco at Capenhurst, UK, input 62.3 kWh/SWU for the whole plant in 2001-02, including infrastructure and capital works).

Areva's new EUR 3 billion French plant - Georges Besse II - started commercial operation in April 2011 and will ramp up to full capacity of 7.5 million SWU/yr in 2016.

Urenco's new $1.5 billion National Enrichment Facility in New Mexico, USA commenced production in June 2010.  Full initial capacity of 3 million SWU/yr is expected to be reached in 2013, and 5.7 million SWU/yr is planned for 2015 - enough for 10% of US electricty needs.

Following this, Areva is building a $2 billion, 3.3 million SWU/yr Eagle Rock plant at Idaho Falls, USA which it expects to commence operation in 2014, ramping up to full production in 2019.  In 2009 it applied for doubling in capacity to 6.6 million SWU/yr.

USEC has been building its American Centrifuge Plant  in Piketon, Ohio, on the same Portsmouth site where the DOE's experimental plant operated in the 1980s as the culmination of a very major R&D program. Operation from 2012 was envisaged, at a cost of $3.5 billion then estimated. It is designed to have an initial annual capacity of 3.8 million SWU, though its licence application is for 7 million SWU to allow for expansion. Authorisation for enrichment up to 10% was sought - most enrichment plants operate up to 5% U-235 product, which is becoming a serious constraint as reactor fuel burnup increases. A demonstration cascade started up in September 2007 with about 20 prototype machines, and a lead cascade of commercial centrifuges started operation in March 2010. These are very large machines, 13 m tall, each with about 350 SWU/yr capacity. However the whole project was largely halted in July 2009 pending further finance. A total of $1.95 billion had been spent from May 2007 to December 2010, and a further $2.8 billion cost was then projected. In March 2010 the DOE made $45 million available to USEC for continued development.

 

* USEC's American Centrifuges are more than 12 m tall and 40-50 cm diameter.  The Russian centrifuges are less than one metre tall. Chinese ones are larger, but shorter than Urenco's. 

The enriched gas forms part of the feed for the next stages while the depleted UF6 gas goes back to the previous stage. Eventually enriched and depleted uranium are drawn from the cascade at the desired assays.

To obtain efficient separation of the two isotopes, centrifuges rotate at very high speeds, with the outer wall of the spinning cylinder moving at between 400 and 500 metres per second to give a million times the acceleration of gravity.

Although the volume capacity of a single centrifuge is much smaller than that of a single diffusion stage, its capability to separate isotopes is much greater. Centrifuge stages normally consist of a large number of centrifuges in parallel. Such stages are then arranged in cascade similarly to those for diffusion. In the centrifuge process, however, the number of stages may only be 10 to 20, instead of a thousand or more for diffusion.

Laser processes

Laser enrichment processes have been the focus of interest for some time. They are a third-generation technology promising lower energy inputs, lower capital costs and lower tails assays, hence significant economic advantages. One of these processes is almost ready for commercial use. Laser processes are in two categories: atomic and molecular.

Development of the Atomic Vapour Laser Isotope Separation (AVLIS, and the French SILVA) began in the 1970s. In 1985 the US Government backed it as the new technology to replace its gaseous diffusion plants as they reached the end of their economic lives early in the 21st century. However, after some US$ 2 billion in R&D, it was abandoned in USA in favour of SILEX, a molecular process. French work on SILVA has now ceased, following a 4-year program to 2003 to prove the scientific and technical feasibility of the process. Some 200kg of 2.5% enriched uranium was produced in this.

Atomic vapour processes work on the principle of photo-ionisation, whereby a powerful laser is used to ionise particular atoms present in a vapour of uranium metal. (An electron can be ejected from an atom by light of a certain frequency. The laser techniques for uranium use frequencies which are tuned to ionise a U-235 atom but not a U-238 atom.) The positively-charged U-235 ions are then attracted to a negatively-charged plate and collected. Atomic laser techniques may also separate plutonium isotopes.

The main molecular processes which have been researched work on a principle of photo-dissociation of Uf6 to solid UF5+, using tuned laser radiation as above to break the molecular bond holding one of the six fluorine atoms to a U-235 atom. This then enables the ionized UF5 to be separated from the unaffected UF6 molecules containing U-238 atoms, hence achieving a separation of isotopes. Any process using UF6 fits more readily within the conventional fuel cycle than the atomic process.

The main molecular laser process to enrich uranium is SILEX, which utilises UF6 and is now known as Global Laser Enrichment (GLE). In 2006 GE Energy entered a partnership with Australia's Silex Systems to develop the third-generation SILEX process. It provided for GE (now GE-Hitachi) to construct in the USA an engineering-scale test loop, then a pilot plant or lead cascade, which could be operating in 2012, and expanded to a full commercial plant. Apart from US$ 20 million upfront and subsequent payments, the license agreement would yield 7-12% royalties, the precise amount depending on how low the cost of deploying the commercial technology. In mid 2008 Cameco bought into the GLE project, paying $124 million for 24% share, alongside GE (51%) and Hitachi (25%). (Earlier, in 1996 USEC had secured the rights to evaluate and develop SILEX for uranium but baled out of the project in 2003.)

GE referred to SILEX, which it rebadged as GLE, as "game-changing technology" with a "very high likelihood" of success. GE-Hitachi is completing the test loop program, the initial phase of which has already been successful in meeting performance criteria, and engineering design for a commercial facility has commenced. GEH is operating the GLE test loop at Global Nuclear Fuel's Wilmington, North Carolina fuel fabrication facility - GNF is a partnership of GE, Toshiba, and Hitachi.

In October 2007 the two largest US nuclear utilities, Exelon and Entergy, signed letters of intent to contract for uranium enrichment services Global Laser Enrichment LLC (GLE). The utilities may also provide GLE with support if needed for development of a commercial-scale GLE plant. In August 2010 TVA agreed to buy $400 million of enrichment services from GLE if the project proceeds.

In mid 2009 GEH submitted the last part of its licence application for this GLE plant, which was expected to take the NRC about 30 months to process. At the end of February 2012 NRC published a favourable environmental review of the project, and its safety evaluation found that its programs for the physical protection of special nuclear material and classified matter, material control and accounting provided an adequate basis for both safety and safeguards of facility operations. Following a July review by the NRC Atomic Safety and Licensing Board, a full licence to construct and operate a plant of up to 6 million SWU/yr was issued in September 2012. GLE will now decide in the light of commercial considerations on whether to proceed with a full-scale enrichment facility at Wilmington. The project, enriching up to 8% U-235, could be operational from 2014, and ramp up to annual capacity of 6 million separative work units (SWU) in 2020.

Applications to silicon and zirconium stable isotopes are also being developed by Silex Systems near Sydney.

CRISLA is another molecular laser isotope separation process which is the early stages of development. In this a gas is irradiated with a laser at a particular wavelength that would excite only one of the isotopes. The entire gas is subjected to low temperatures sufficient to cause condensation on a cold surface or coagulation in the gas. The excited molecules in the gas are not as likely to condense as the unexcited molecules. Hence in cold-wall condensation, gas drawn out of the system is enriched in the isotope that was laser-excited.

Electromagnetic process  

A very early endeavour was the electromagnetic isotope separation (EMIS) process using calutrons.  This was developed in the early 1940s in the Manhattan Project to make the highly enriched uranium used in the Hiroshima bomb, but was abandoned soon afterwards. However, it reappeared as the main thrust of Iraq's clandestine uranium enrichment program for weapons discovered in 1992. EMIS uses the same principles as a mass spectrometer (albeit on a much larger scale). Ions of uranium-238 and uranium-235 are separated because they describe arcs of different radii when they move through a magnetic field. The process is very energy-intensive - about ten times that of diffusion.

Aerodynamic processes 

Two aerodynamic processes were brought to demonstration stage around the 1970s. One is the jet nozzle process, with demonstration plant built in Brazil, and the other the Helikon vortex tube process developed in South Africa. Neither is in use now, though the latter is the forerunner of new R&D. They depend on a high-speed gas stream bearing the UF6 being made to turn through a very small radius, causing a pressure gradient similar to that in a centrifuge. The light fraction can be extracted towards the centre and the heavy fraction on the outside. Thousands of stages are required to produce enriched product for a reactor. Both processes are energy-intensive - over 3000 kWh/SWU.  The Helikon Z-plant in the early 1980s was not commercially oriented and had less than 500,000 SWU/yr capacity.  It required some 10,000 kWh/SWU.

The Aerodynamic Separation Process (ASP) being developed by Klydon in South Africa employs similar stationary-wall centrifuges with UF6 injected tangentially.  It is based on Helikon but pending regulatory authorisation it has not yet been tested on UF6 - only light isotopes such as silicon.  However, extrapolating from results there it is expected to have an enrichment factor in each unit of 1.10 (cf 1.03 in Helikon) with about 500 kWh/SWU and development of it is aiming for 1.15 enrichment factor and less than 500 kWh/SWU.  Projections give an enrichment cost under $100/SWU, with this split evenly among capital, operation and energy input.

One chemical process has been demonstrated to pilot plant stage but not used. The French Chemex process exploited a very slight difference in the two isotopes' propensity to change valency in oxidation/reduction, utilising aqueous (III valency) and organic (IV) phases. 

Enrichment of reprocessed uranium 

 In some countries used fuel is reprocessed to recover its uranium and plutonium, and to reduce the final volume of high-level wastes. The plutonium is normally recycled promptly into mixed-oxide (MOX) fuel, by mixing it with depleted uranium.

Where uranium recovered from reprocessing used nuclear fuel (RepU) is to be re-used, it needs to be converted and re-enriched.  This is complicated by the presence of impurities and two new isotopes in particular: U-232 and U-236, which are formed by or following neutron capture in the reactor, and increase with higher burn-up levels.  U-232 is largely a decay product of Pu-236, and increases with storage time in used fuel, peaking at about ten years.  Both decay much more rapidly than U-235 and U-238, and one of the daughter products of U-232 emits very strong gamma radiation, which means that shielding is necessary in any plant handling material with more than very small traces of it.  U-236 is a neutron absorber which impedes the chain reaction, and means that a higher level of U-235 enrichment is required in the product to compensate.  For the Dutch Borssele reactor which normally uses 4.4% enriched fuel, compensated enriched reprocessed uranium (c-ERU) is 4.6% enriched to compensate for U-236.  Being lighter, both isotopes tend to concentrate in the enriched (rather than depleted) output, so reprocessed uranium which is re-enriched for fuel must be segregated from enriched fresh uranium.  The presence of U-236 in particular means that most reprocessed uranium can be recycled only once - the main exception being in the UK with AGR fuel made from recycled Magnox uranium being reprocessed.

All these considerations mean that only RepU from low-enriched, low-burnup used fuel is normally recycled directly through an enrichment plant.  For instance, some 16,000 tonnes of RepU from Magnox reactors* in UK has been used to make about 1650 tonnes of enriched AGR fuel, via two enrichment plants.  Much smaller quantities have been used elsewhere, in France and Japan.  Some re-enrichment, eg for Swiss, German and Russian fuel, is actually done by blending RepU with HEU.

* since Magnox fuel was not enriched in the first place, this is actually known as Magnox depleted uranium (MDU).  It assayed about 0.4% U-235 and was converted to UF6, enriched to 0.7% at BNFL's Capenhurst diffusion plant and then to 2.6% to 3.4% at Urenco's centrifuge plant.  Until the mid 1990s some 60% of all AGR fuel was made from MDU and it amounted to about 1650 tonnes of LEU.  Recycling of MDU was discontinued in 1996 due to economic factors.

A laser process would theoretically be ideal for enriching RepU as it would ignore all but the desired U-235, but this remains to be demonstrated with reprocessed feed. 

Tails from enriching reprocessed uranium remain the property of the enricher. Some recycled uranium has been enriched by Tenex at Seversk for Areva, under a 1991 ten-year contract covering about 500 tonnes UF6. French media reports in 2009 alleging that wastes from French nuclear power plants were stored at Seversk evidently refer to tails from this.

Enrichment of depleted uranium tails

 Early enrichment activities often left depleted uranium tails with about 0.30% U-235, and there were tens of thousands of tonnes of these sitting around as the property of the enrichment companies. With the wind-down of military enrichment, particularly in Russia, there was a lot of spare capacity unused. Consequently, since the mid 1990s some of the highest-assay tails have been sent to Russia by Areva and Urenco for re-enrichment by Tenex. These arrangements however cease in 2010, though Tenex may continue to re-enrich Russian tails. Tenex now owns all the tails from that secondary re-enrichment, and they are said to comprise only about 0.10% U-235.

After enrichment 

The enriched UF₆ is converted to UO₂ and made into fuel pellets - ultimately a sintered ceramic, which are encased in metal tubes to form fuel rods, typically up to four metres long. A number of fuel rods make up a fuel assembly, which is ready to be loaded into the nuclear reactor.

Depleted uranium and deconversion

Depleted uranium (DU) is stored long-term as UF₆ or preferably, after deconversion, as U₃O₈, allowing HF to be recycled. . To early 2007, about one quarter of the world's 1.5 million tonnes of DU had been deconverted.  World deconversion capacity is about 60,000 t/yr at end of 2010.

The main deconversion plant is the 20,000 t/yr one run by Areva NC at Tricastin, France, though the technology has been sold to Russia and a 10,000 t/yr plant is now operational at Zelenogorsk in Siberia. Two plants have been built by Uranium Disposition Services (UDS) at Portsmouth and Paducah, USA, with capacities of 13,500 and 18,000 t/yr respectively. A 6500 t/yr plant is being built at New Mexico in the USA by International Isotopes (INIS).
 

These use essentially a dry process, with no liquid effluent.  It is the same as that used for the enriched portion, albeit at a scale of 20,000 tonnes per year in the one plant.

The UF₆ is first vapourised in autoclaves with steam, then the uranyl fluoride (UO2F2) is reacted with hydrogen at 700°C to yield an HF product for sale to converters and U₃O₈ powder which is packed into 10-tonne containers for storage.

 UF₆ + 2H₂O ==⇒ UO₂F₂ + 4HF 

3UO₂F₂ + 2H₂O + H₂ ===> U₃O₈ + 6HF 

The INIS plant in Idaho uses a slightly different deconversion followed by fluorine extraction process (FEP), on a toll basis. The deconversion plant had been used to produce DU metal for the military and was purchased by INIS. In this, the depleted UF6 is first vapourised in autoclaves and hydrogen is added to give depleted UO2 and anhydrous UF4 which is the main product for sale. The FEP then involves reacting some UF4 with silica to give silicon fluoride (SiF4) as a commercial co-product.

Ownership title is normally transferred to the enricher as part of the commercial deal.  It is sometimes considered as a waste, though only for legal or regulatory reasons and in the USA, but usually it is understood as a long-term strategic resource which can be used in a future generation of fast neutron reactors. Any much more efficient enrichment process would also make it into an immediately usable resource to supply more U-235. Enrichment companies with ownership of large amounts of depleted uranium are quite clear that their stocks are a significant asset.

Environmental Issuses

With the minor exception of reprocessed uranium, enrichment involves only natural, long-lived radioactive materials; there is no formation of fission products or irradiation of materials, as in a reactor. Feed, product, and depleted material are all in the form of UF6, though the depleted uranium may be stored long-term as the more stable U₃O₈.

Uranium is only weakly radioactive, and its chemical toxicity - especially as UF₆ - is more significant than its radiological toxicity. The protective measures required for an enrichment plant are therefore similar to those taken by other chemical industries concerned with the production of fluorinated chemicals.

Uranium hexafluoride forms a very corrosive material (HF - hydrofluoric acid) when exposed to moisture, therefore any leakage is undesirable. Hence:

• in almost all areas of a centrifuge plant the pressure of the UF₆ gas is maintained below atmospheric pressure and thus any leakage could only result in an inward flow;
• double containment is provided for those few areas where higher pressures are required;
• effluent and venting gases are collected and appropriately treated.

Sources:
Heriot, I.D. (1988). Uranium Enrichment by Centrifuge, Report EUR 11486, Commission of the European Communities, Brussels.
Kehoe, R.B. (2002). The Enriching Troika, a History of Urenco to the Year 2000. Urenco, Marlow UK.
Wilson, P.D. (ed)(1996). The Nuclear Fuel Cycle - from ore to wastes. Oxford University Press, Oxford UK.
IAEA 2007, Management of Reprocessed Uranium - current status and future prospects, Tecdoc 1529.