You are here

The proliferation dangers of centrifuge technology

Nuclear Monitor Issue: 
Mikael Shirazi and Andreas Persbo

In early April 2011, a nondescript industrial plant 50km west of Tehran, named TABA, came under public scrutiny when it was revealed as being a significant centrifuge manufactur­ing site—apparently unbeknownst to the International Atomic Energy Agency (IAEA). As the technology involved has become ever more accessible, centrifuge-driven uranium enrichment has emerged as a significant proliferation risk. It is therefore worthwhile to consider the IAEA’s ability to monitor the construction of these specialized machines.

The ability to monitor the construction of centrifuge-driven uranium enrichment is especially illustrative of the added value of the IAEA’s Additional Protocol to the process of confirming the exclu­sively peaceful nature of countries’ nuclear energy program­s. The Additional Protocol is a powerful legal instru­ment developed in the 1990s to complement member states’ Comprehensive Safeguards Agreements (CSAs). This article considers the proliferation risks involved in centrifuge pro­duction and the merits of the Additional Protocol with respect to two countries, Iran and Brazil, neither of whom implement the updated safeguards techniques, but who both possess the ability and will to manufacture centrifuges.

It is often considered that the most difficult stage in the production of nuclear weapons is acquiring the necessary fissile material: either plutonium or highly enriched ura­nium (HEU). In the past, acquiring these materials usually involved building and running a nuclear reactor (to make plutonium), or a gaseous diffusion plant (for HEU). Both required very substantial industrial capabilities. However, with the spread of gas centrifuge technology in the past three decades, the potential route to HEU has become both significantly less challenging—and less conspicuous.

Like the diffusion method, the gas centrifuge technique separates the two isotopes that make up uranium, concen­trating the crucial U-235 from the very slightly heavier U-238. In nature, uranium consists almost entirely of U-238 (at around 99.3 per cent) and therefore requires processing in order for the weapons-usable U-235 to be separated out. To be useful in ‘light water’ reactors, the raw material must be converted into uranium hexafluoride gas and subse­quently ‘enriched’ in the separation process to consist of 3-5 per cent U-235 particles (known as low enriched uranium, or LEU). Natural uranium can be used in other reactor types after some processing. Nuclear weapons require HEU at about 90 per cent enrichment. Enriching with the centrifuge process involves injecting uranium hexafluoride gas into cylinders rotating tens of thousands of times per minute. The effect of centrifugal force pushes the U-238 closer to the outer wall of the machine, with U-235 particles tending towards the center, which is then siphoned off. Each ma­chine can only perform a very small amount of enrichment. An effective enrichment plant therefore requires large num­bers of centrifuges linked together in so-called ‘cascades’.

The older gaseous diffusion system requires thousands more painstaking steps, which take place in immense facilities using significant amounts of energy, and emitting large amounts of heat. In contrast, centrifuges on average perform the same amount of enrichment in significantly fewer steps, consuming smaller amounts of electricity. Centrifuge fa­cilities therefore tend to be less conspicuous. They are typically much more compact, without the easily identifi­able electrical and cooling systems associated with gaseous diffusion plants, or heat emissions detectable to infrared imaging systems. It may be possible to trace uranium hex­afluoride gas accidentally released from a centrifuge enrich­ment plant, but these emissions are normally very small.

The number of centrifuges required to produce enough fis­sile material for a weapon depends on the design and effi­ciency of the centrifuges themselves—measured in kilograms of ‘separative work units’ per year (kg SWU/yr). This can range from lower than two kg SWU/yr for less advanced models to machines (currently confined to Europe or the United States) operating at 100 kg SWU/yr and above.

Centrifuge production and the Additional Protocol
As a rule of thumb, it requires about 100,000-120,000 kg SWU to produce enough LEU per year for an average sized nuclear reactor. In contrast, it requires only 6,000 kg SWU to produce enough HEU for one weapon a year (known as one ‘significant quantity’, defined by the IAEA as 27.8 kg of 90 per cent enriched uranium).

The potency of gas centrifuge technology in terms of pro­liferation risks is therefore clear: these are machines capable of producing ‘significant quantities’ of fissile material in relatively low numbers and with a small footprint, thus making them a good bet for states wanting to develop nu­clear weapons-usable material without being detected. However, centrifuges are complicated machines, requiring very specialized technical capabilities. One of the major difficulties is that even the slowest centrifuges spin at rates requiring unusually durable materials—ranging from alu­minium alloys for older machines and maraging steel (a particularly strong type of steel) to modern ultra-strong carbon composites. These materials require precision ma­chine tools to shape and strengthen them. The high-speed motors and their variable-frequency power supplies (which adapt the electrical current available from the power grid into an output of much higher frequency) also need to be specifically adapted for use in centrifuges.

Centrifugal safeguards standards
Under the 1968 Nuclear Non-Proliferation Treaty (NPT), non-nuclear-weapon states’ obligations on centrifuge manufacturing fall under two IAEA safeguards regimes: those with Comprehensive Safeguards Agreements (CSAs), and those who further implement the strengthened measures of the Additional Protocol to their CSAs.

Though each non-nuclear-weapon state’s CSA is individual, all follow the form and content of a standard text, ‘IN­FCIRC/153’, which obliges a country to provide information on all nuclear material and facilities, and to allow agency inspectors to verify these declarations. The resulting verifi­cation regime focuses largely on nuclear material account­ancy to check the accuracy of declared materials in declared facilities. According to Article 8 of INFCIRC/153, this guarantees the IAEA information on only those facilities ‘relevant to safeguarding such material’. The definition of ‘facility’ is articulated in Article 106 to include reactors, conversion plants, fabrication plants, reprocessing plants, isotope separation plants, separate storage installations, or any location where significant amounts of nuclear material is customarily used. As such there are no requirements re­garding centrifuge production facilities. CSAs were designed in an age when centrifuge enrichment technology was still in its infancy. The underlying assumption was that the production of HEU through conspicuous gaseous diffusion plants would be readily detectable, and that the proliferation risk came instead from the diversion of material from de­clared facilities.

With the discovery of Saddam Hussein’s secret nuclear weapons program in the aftermath of the 1991 Gulf War, it became clear that it was necessary to address pos­sible clandestine uranium enrichment—with centrifuge production being an important component. Partly as a result of this discovery, the Additional Protocol was developed and opened for voluntary signature in 1997. It is a legal instru­ment that provides the IAEA with more information and wider access rights, thereby strengthening its ability to verify that a country is not producing material for nuclear weapon purposes.

The document ‘INFCIRC/540’ describes the standard ob­ligations required under an AP. In contrast with IN­FCIRC/153, this document specifies in Article 2.a.(iv) that the participating state must provide the IAEA with a descrip­tion of the scale of operations involved in centrifuge produc­tion. According to Annex I of INFCIRC/540, centrifuge production is described as the manufacture of centrifuge rotor tubes or the assembly of gas centrifuges. These ac­tivities are further detailed in Annex II, which describes the purpose, general design, and component set of gas centri­fuges. Such constituent parts include: rotor assemblies, rotor tubes, bellows, baffles, top and bottom caps, mag­netic suspension bearings, molecular pumps, motor stators, centrifuge housings, and scoops, among others.

As well as indigenous manufacturing capabilities, the pro­tocol also brings into focus the other way of acquiring centrifuges (or their constituent parts)—import from for­eign trade partners. Article 2.a.(ix) of INFCIRC/540 outlines the state’s responsibility, when requested, to provide infor­mation to the IAEA on the identity, quantity, and location of the intended use of all the materials and equipment listed in Annex II that have been acquired from abroad. The information generated by these requirements enables the IAEA to develop a fuller understanding of a member state’s uranium enrichment program. It thus becomes pos­sible to draw comparisons between centrifuge production rates and centrifuge deployment in declared facilities: for instance, if more centrifuges are manufactured than de­ployed, the IAEA will be able to flag the discrepancy for further investigation.

The CSA and the AP differ not only in terms of the infor­mation flow that they can generate but also in the level of access for inspectors. According to Article 76.a of the model CSA text (INFCIRC/153), the IAEA is guaranteed access only to ‘any location where the initial report or any inspections carried out in connection with it indicate that nuclear material is present.’ There is a provision in Article 73 of INFCIRC/153 for ‘special inspections’, which give the agency the right to visit ‘locations in addition to the access specified’—a vague definition which John Carlson, a mem­ber of VERTIC’s International Verification Consultants Network, interprets as ‘anywhere in the state’ if there are ‘circumstances giving rise to suspicion.’ This could conceiv­ably include certain centrifuge manufacturing plants. His­torically, though, the special inspection tool (which, accord­ing to Article 77, must be obtained in agreement with the inspected state party) has been of little value. It has only been invoked by the IAEA on one previous occasion. This was against North Korea in 1992, and access was then denied. INFCIRC/540 (the model Additional Protocol) makes an important contribution in this area by outlining a system of ‘Complementary Access’ to inspectors. This expands the rights of the Agency to make visits to centrifuge manufac­turing plants according to Article 4.a.(ii), for the purpose of resolving ‘a question relating to the correctness and com­pleteness of the information provided [...] or to resolve an inconsistency relating to that information.’ There is no need to obtain agreement from the party and notification of a visit can be as short as 24 hours.

Though INFCIRC/540 specifies that the IAEA ‘shall not mechanistically or systematically seek to verify’ information provided by the state, its ability to make informed judgments about a proliferation risk is substantially increased, and a state’s corresponding ability to shield important in­formation from it is substantially diminished. With respect to the monitoring of centrifuge production, the salient points of the Additional Protocol are Article 2.a.(iv)’s en­shrined principle of information provision as a matter of routine, and Article 4.a.(ii)’s enshrined principle of Com­plementary Access as of right.

This has important conse­quences, explored below, for states that produce centri­fuges, as is made clear by the examples of Iran and Brazil, both of whom possess the indigenous capacities to manu­facture these machines, but neither of which currently implement the Additional Protocol.

Iran: AP, the option-limiter
The controversy and uncertainties surrounding Iran’s ura­nium enrichment program are well-known and well-documented. The Islamic Republic has signed an AP, but has not yet ratified it. Nevertheless, Iran implemented the protocol on a voluntary basis between 2003 and 2006, but cut off cooperation in retaliation to the IAEA Board of Governors vote to report Iran to the UN Security Council. During this time the Agency learnt a great deal about the Iranian nuclear infrastructure; since then, however, relevant knowledge about centrifuge production capabilities has deteriorated markedly.

It is therefore not difficult to appreciate the interest gener­ated, when, at a press conference in Washington, DC, an Iranian opposition group announced the discovery of the previously-undocumented role of a facility named TABA in producing centrifuge parts for Iran’s controversial uranium enrichment program. TABA apparently manufactures ‘casing, magnets, molecular pumps, composite tubes, bel­lows, and centrifuge bases’ primarily for the current gen­eration of machines—but also for emerging next-generation centrifuges. Ali Asghar Soltanieh, Tehran’s envoy to the IAEA, refuted any allegations of concealment, pointing out that Iran’s safeguards obligations did not necessitate any provision of information about the plant to the IAEA. Rather, they required only the ‘inspection of centrifuge machines.’ This is indeed broadly in line with the require­ments of the CSA as described above, which strictly speak­ing concerns itself only with the nuclear materials flowing within the machines.

The disclosure, however, highlights the proliferation risk resulting from the limited reach of the CSA. TABA is lo­cated in a nondescript industrial park and offers few distin­guishing features. The facility’s generic name—a Farsi ab­breviation of ‘Towlid Abzar Boreshi Iran’, meaning ‘Iran Cutting Tools Company’—also gives little away. This lack of transparency and openness over their centrifuge manu­facturing capabilities offers the Iranian authorities the pos­sibility—should they so choose—of secretly sending cen­trifuges to a undeclared enrichment installation to produce weapons-grade fissile material, whilst appearing to fulfill their safeguards obligations.

Enrichment facilities can be relatively small and largely indistinguishable from other industrial plants, or outright hidden as in the case of Iran’s underground Qom enrichment facility. The Qom plant was uncovered in September 2009 as a result of Western intelligence-gathering operations; its existence was previously a secret. In an atmosphere so fun­damentally degraded by a lack of trust between the princi­pal actors, the possibility that any small and inconspicuous enrichment facility could be discretely producing weapons-usable material is a serious consideration.

It is a possibility that Iran’s 2007 decision to suspend an essential commitment to the IAEA regarding the declaration of new facilities has made concerns over undeclared facilities significantly more acute. The commitment in question is set out in the modified Code 3.1 of Iran’s Subsidiary Ar­rangements, to which it acceded in 2003 and which the CSA specifies cannot be unilaterally modified without the IAEA’s consent. The result of the suspension, which the IAEA re­portedly did not agree to, is that Iran has reverted to an outdated requirement that any new facility need only be declared six months prior to the introduction of nuclear material, rather than as soon as the decision to construct it is taken. The option therefore exists for Iranian authorities to begin construction on sites that can house centrifuge cascades, and even to outfit them with this equipment, without violating any of its safeguards obligations. Of course, if undeclared enrichment begins, this is no longer true. But many of the crucial steps taken to get to this point in op­erating a clandestine HEU-producing program (the undeclared industrial development of centrifuges and their deployment in undeclared enrichment plants) will have been taken with little risk.

The power of the AP is to close off such windows of op­portunity and thereby build confidence among countries. INFCIRC/540 states clearly the IAEA’s right to be supplied with information regarding centrifuge production facilities, and its right to access these facilities. The result is an im­portant reversal of responsibility, away from the IAEA hav­ing to press for data and onto the state itself to provide the information in a routine manner.

Brazil: AP, an option limited
The Brazilian centrifuge program began as a covert project in 1979 at the behest of the military government that dominated Brazilian political life until 1985. A research team, under the direction of the Brazilian navy, developed over the next decade a centrifuge technology in which rotors spin not on the usual metal pin bearings, but on electromag­netic bearings, allowing the rotating and fixed parts in the machine to operate without any point of contact. This is designed to eliminate sources of friction which reduce ef­ficiency and durability, and recent enrichment capacities have been placed at 10 kg SWU/yr. Construction of these machines takes place at the navy’s Aramar Experimental Center, outside São Paulo. Brazil has ambitious plans to attain an enrichment capacity at its main deployment site at Resende, near Rio de Janeiro, of 300,000 kg SWU/yr by 2014, and up to one million kg SWU/yr by 2030.

The military origins of the program, its secrecy before the advent of democratic government, a late accession to the NPT in 1998, and the 2005 admission by a former president that Brazil had previously sought to develop nu­clear weapons to counter competition from Argentina all point to the need for a robust verification regime that instills confidence in the peaceful ambitions of the program as it exists today. Currently, this work is done through the 1991 Quadripartite Safeguards Agreement, which joins together Brazil, Argentina, the IAEA and ABACC (the Brazilian-Argentine Agency for Accounting and Control of Nuclear Materials) to mandate the application of nuclear safeguards. Analogous to the CSA, this ad hoc arrangement does not offer the extended measures provided by the AP, as described above, with the exception of some provisions for unan­nounced inspections. Monitoring, performed by both ABACC and the IAEA, focuses on flows of nuclear mate­rial and provides access only to those facilities through which significant amounts of such material passes. It seems likely that another possible route to fissile material may be opened up with the Brazilian navy’s development of nuclear-pow­ered submarines, in which uranium enriched as high as 10 per cent by centrifuges at Aramar will power a reactor out­side the reach of safeguards. Although the US has made efforts to persuade Brazil to give up on these plans, it was not successful, and negotiations with the IAEA to establish appropriate verification measures are ongoing. This is an eventuality that neither the CSA nor AP address directly, and for which entirely new arrangements will need to be developed.

Despite these substantial capabilities and ambitious plans, implementation of the Additional Protocol has been reso­lutely dismissed by Brazil, with the country’s 2008 Na­tional Defense Strategy rejecting it until further progress in disarmament is made by the NPT nuclear weapons coun­tries. Brazilian officials have offered a variety of other rea­sons—revolving primarily around an unwillingness to allow inspectors access to the commercially sensitive electromag­netic bearing technology, and the fact that it is an unneces­sary measure in a country with a solid non-proliferation record which constitutionally prohibits nuclear weapons development (Brazil has also joined the Treaty of Tlatelolco, which establishes a Latin American nuclear-weapon-free zone, and ratified the Comprehensive Test Ban Treaty, which Iran has only signed). Analysts suspect the main reasons for opposition are military in nature, with the navy unwilling to grant extended access rights to the centrifuge manufactur­ing facilities in Aramar that are co-located with non-nucle­ar submarine R&D activities. This is despite the fact that Article 7 of the AP outlines clearly a state’s right to request ‘managed access’ to protect proprietary information, and that the IAEA Department of Safeguards (in charge of the practical application of safeguards) has had regular access to sensitive technologies throughout its history without leaking them.

Many of the same clandestine enrichment options are there­fore as open to Brazil as they are to Iran - without, how­ever, the associated IAEA reports, UN resolutions or Secu­rity Council sanctions. Most observers, such as Jeffrey Lewis, director of the East Asia Nonproliferation Program at the Monterey Institute of International Studies, or Con­doleeza Rice during her term as US Secretary of State, seem not to question Brazil’s commitment against nuclear pro­liferation. Ad hoc measures, such as the Quadripartite Agreement and a future system to monitor enriched ura­nium production for nuclear powered submarines, are deemed to be imperfect but adequate safeguards measures - despite the lack of scrutiny on centrifuge production at Aramar. Crucially though, this type of safeguards develop­ment can only occur in an atmosphere with a certain level of trust; such as that which generally characterizes the IAEA’s relationship with Brazil.

One of the most important benefits of AP implementation is to lessen the impact of the wider political atmosphere. Should relations take a turn for the worse, the principles of information provision as a matter of routine and Comple­mentary Access as of right allow for confident conclusions to be drawn over the use of centrifuge technology regardless of political context. With the IAEA thus somewhat shield­ed by the AP from the vagaries of international tensions over policy and intent, it is able to focus with greater free­dom on states’ technical centrifuge capabilities, allowing for more reliable judgments on proliferation risk to be made. The effectiveness of the IAEA’s verification regime is dimin­ished, however, by the selective and voluntary implementa­tion of AP requirements in ‘suspect states’—much as the theoretically powerful CSA Special Inspection tool is often rendered impotent in practice. Universalization of the Ad­ditional Protocol should therefore be a central goal in strengthening the global nuclear non-proliferation regime.

Source and contact: Mikael Shirazi and Andreas Persbo, Trust & Verify 133, April-June 2011.
VERTIC (the Verification Research, Training and Information Centre) is an independent, non-profit making charitable organization. Established in 1986, VERTIC supports the development, implementation and verification of international agreements as well as initiatives in related areas. It can be reached: Development House, 56-64 Leonard Street
London EC2A 4LT, United Kingdom.
Tel: +44 20 7065 0880