INESAP Information Bulletin No.23, April 20041

Alexander Glaser

The recent exposure of an operatingsupplier network around the Pakistaniscientist A.Q. Khan has propelled thefront-end of the nuclear fuel cycle backinto the center of the non-proliferationdebate. The ongoing debate is focusedupon measures that emphasize exportcontrols and technology denial as aprimary means to halt nuclear prolifer-ation.1 These are insufficient or eveninadequate proposals because they ad-dress only part of the problem. The gascentrifuge for uranium enrichment is ahighly proliferation-prone technology.Nonetheless, its large-scale commer-cial use has never been seriously dis-couraged. Indeed, the centrifuge playsan increasing role in the enrichment in-dustry and all facilities recently built,under construction, or planned arebased on centrifuge technology. In2001, for the first time, a larger enrich-ment capacity was provided by cen-trifuges than by the gaseous diffusionprocess.

As will be argued below, the factthat advanced centrifuge technologyhas been developed and deployed invarious regions of the world withouthaving a competitive technological al-ternative to it, may become a centraldilemma for future nuclear energy use– unless an internationally acceptableand non-discriminatory supply sys-tem can be devised and agreed upon.

Centrifuge Technology

The development of centrifuges forisotope separation began in the 1930sand had already been considered foruranium enrichment during the Man-hattan project. In the early days, thetechnology was not competitive withalternative enrichment processes, be-cause huge facilities based on thegaseous diffusion process were alreadyunder construction or operational andno further capacities were needed.Nonetheless, development of the gascentrifuge continued after World War

II, especially in the Soviet Union, butlater also in Western Europe and theU.S. With the installation of significantcapacities in the U.K., the Netherlands,and Germany since the 1970s and thetechnological advances that came alongwith the associated research and devel-opment programs, the modern gas cen-trifuge gradually became the work-horse of the international enrichmentindustry. Table 1 lists centrifuge facili-ties worldwide and their respective op-erational status.2

In essence, the gas centrifuge isa rapidly rotating, vertically orientedrotor, into which a uranium-contain-ing gas (UF6) is injected. Once fol-lowing the motion of the wall, thegas molecules experience an enor-mous centrifugal force and accumu-late on the inner surface of the rotor.Based on an elementary physical ef-fect, the relative abundance of thelighter uranium-235-containing mol-

ecules increases with distance fromthe wall due to the mass difference ofthe two relevant uranium isotopes.This effect can be used to extract aproduct from the machine that isslightly enriched in the desired iso-tope.3 In practice, several thousandmachines have to be connected inparallel to generate significant urani-um throughput and connected in se-ries to achieve adequate enrichmentof the product – a configurationcalled cascade.4

The Safeguards Approach toCentrifuge Facilities

The first enrichment facilities in theworld were built for military purpos-es and, hence, designed and used forHEU production. These facilitieswere unsafeguarded then and usuallystill are so today. Until the mid-1970s,supply of enrichment services for

Protecting the Law: Non-Proliferation and Disarmament

Beyond A.Q. KhanThe Gas Centrifuge, Nuclear Weapon Proliferation, and the NPT Regime

Table 1: Centrifuge enrichment facilities of the world.

Country Location/Name Status Start-Up Capacity

Brazil BRN Aramar Operational 1992 5 tSWU/yBRF Aramar Operational 1998 4 tSWU/y

Resende Under Construction 2004 120 tSWU/yChina Shaanxi Operational 1997 200 tSWU/y

Lanzhou 2 Under Construction 2005 500 tSWU/yFrance GBII Tricastin Planned ? 7,500 tSWU/yGermany Jülich Operational 1964 Laboratory

Gronau Operational 1985 1,800 tSWU/yIndia Rattehalli Operational 1990 3—10 tSWU/yIran Natanz Under Construction ? ?Iraq Al Furat Destroyed in 1991 — —Israel (unconfirmed) Dimona Operational 1980 Pilot scaleJapan Ningyo-Toge Shutdown in 2004 1979 75 tSWU/y

Ningyo-Toge Shutdown in 2004 1989 200 tSWU/yRokkasho Operational 1992 1,050 tSWU/y

Libya ? Abandoned in 2004Netherlands Almelo Operational 1973 2,200 tSWU/yNorth Korea ? Under Construction ?Pakistan Kahuta Operational 1984 5 tSWU/yRussia Sverdlovsk Operational 1949 7,000 tSWU/y

Seversk Operational 1950 4,000 tSWU/yAngarsk Operational 1954 1,000 tSWU/y

Krasnoyarsk Operational 1984 3,000 tSWU/yU.K. Capenhurst Operational 1972 2,300 tSWU/yUS Portsmouth Awaiting license ? Pilot scale

? Planned ? ?

Total centrifuge capacity operational in 2004 ~ 23,000 tSWU/yTotal enrichment capacity available in 2004 (all processes) ~ 53,500 tSWU/yTotal enrichment capacity required in 2004 ~ 35,000 tSWU/y

INESAP Information Bulletin No.23, April 2004 2

Alexander Glaser

commercial purposes was monopo-lized by a few nuclear-weapon states.Only then it became obvious thatnon-nuclear-weapon states, namelysome Western European countrieswith advanced nuclear programs,would aspire and acquire independentenrichment capacities, which wouldbe placed under safeguards.

For these reasons, work on safe-guards procedures on enrichment fa-cilities started relatively late and wasthen focused on centrifuge facilities,the main candidate to be covered bysafeguards. Furthermore, studies car-ried out under the auspices of theIAEA in the 1970s revealed that nosimple safeguards concept existed thatwould be adequate for centrifuge en-richment facilities.6 This is mainly dueto the fact that centrifuge facilitiesshow a high degree of operational flex-ibility, which complicates safeguardsprocedures in general. At the sametime, they do also involve sensitivetechnologies, which is why technologyholders have been generally concernedthat design information might be com-promised by visual access to the ma-chines. Unsurprisingly, the question ofwhether or not inspectors would haveaccess to the cascade halls at all wassubject of considerable debate duringnegotiations on safeguards conceptsfor centrifuge facilities. The originalsafeguards concept developed specifi-cally for centrifuge facilities under anINFCIRC/153-type agreementemerged from the discussion andanalysis of the Hexapartite SafeguardsProject (HSP) that convened from1980 to 1983. In essence, the HSP ap-proach, which became the de factostandard for centrifuge facilities sincethen, envisions two conceptually dif-ferent activities:■ Activities outside the cascadehalls are primarily based on “conven-tional” safeguards practices and arefocused on material accountancy todeter or detect diversion of declaredmaterial.■ Activities inside the cascade hallsare used to verify that no material be-yond the declared enrichment level,and in particular no HEU, is beingproduced. Access to the cascade areasis governed by the so-called LimitedFrequency Unannounced Access

(LFUA) concept, which regulates de-lay and maximum duration of the vis-its, as well as permitted activities ofthe inspectors.More recently, safeguards techniquesapplied in centrifuge facilities havebeen extended by EnvironmentalSampling (ES) or High PrecisionTrace Analysis (HPTA). Approved bythe IAEA in 1995 and used on a rou-tine basis since then, the method con-sists in the collection of depositedparticles with swipe samples, usuallytaken during inspections of the cas-cade areas along the agreed accessroutes. Subsequently, these samplesare analyzed off-site allowing ex-tremely accurate determination of thecomposition of the feed and productmaterial. This method is a formidabletool to identify traces of HEU in gen-eral and has been extremely successfulso far. As a result, clandestine produc-tion of HEU in a safeguarded de-clared facility has become an extreme-ly risky undertaking for a potentialproliferator as it is effectively doomedto be uncovered sooner or later.

Proliferation Challenges

Uranium enrichment technology hasalways been recognized as a highlysensitive technology.7 Indeed, it is theonly technology used in the civiliannuclear fuel cycle whose technicalitieswere not shared within the Atoms forPeace program – contrary to the detailsof plutonium reprocessing. Whilesome enrichment processes are moreproliferation-prone than others, themain barrier preventing a faster spreadin the past was the difficulty to masterenrichment technologies, a feature in-herent to all known processes. Possibleproliferation risks associated with en-richment technologies in general, and

with centrifuge technology in particu-lar, are listed in Table 2.

Only one category of prolifera-tion and diversion scenarios for en-riched uranium are addressed by tra-ditional safeguards measures appliedin declared facilities under an INF-CIRC/153-type agreement. Currentsafeguards standards are very effec-tive in verifying that no declared ma-terial has been diverted and that noHEU has been or is being producedin the facility – the latter aspectstrongly benefitting from the envi-ronmental sampling techniques in-troduced in the 1990s and supple-menting the somewhat limitedLFUA concept.

However, current safeguardsprocedures are based on the funda-mental assumption that no undeclaredmaterial is processed in the facility –and they are not designed to detectsuch an activity. To address this issue,one could require additional surveil-lance in the facility or install instru-ments monitoring uranium flow andenrichment on a continuous basis.8Such a strengthened safeguards ap-proach would require revision and anegotiated upgrade of the existing one.

Detection of Undeclared FacilitiesWith covert HEU production in asafeguarded facility becoming moredifficult to conceal effectively, theconstruction of undeclared facilitiesironically becomes more attractive or“inevitable” for a potential prolifera-tor. Clearly, detectability of such facil-ities would be highly desirable – butthe experience made recently in Iranpoints in a different direction.

The main concerns specific tocentrifuges relevant in this context aretheir low energy consumption and thesmall process area compared to alter-

Declared Facility Undeclared Facility

Covert Diversion of LEU(operation as declared) (abrupt or protracted) Production of HEU

(possibly using LEU feed)Covert Excess LEU production(with modifications) (or production of HEU)

Overt Break-out scenario

Table 2: Proliferation scenarios associated with enrichment technology

INESAP Information Bulletin No.23, April 20043

native processes in use today, namelygaseous diffusion: the footprint of acentrifuge facility large enough to pro-duce 25 kg of HEU annually is about600 square meters and its energy con-sumption would be less than 100 kW.9Both numbers indicate that detectionby satellite remote-sensing techniquesis virtually impossible. Another op-tion to detect an undeclared nuclearfacility would be to use wide area en-vironmental monitoring (WAEM) tocollect characteristic particle samplesemitted by the plant. Note, however,that an enrichment facility based onthe centrifuge process uses equipmentoperated under high-vacuum condi-tions. Leaks primarily lead to an in-flow of air into the centrifuge equip-ment, not to a significant release of gasmolecules from the system. Eventhough the presence of uranium-con-taining particles in any operational orpreviously operational facility can bedetected inside the building with theabove-mentioned sampling tech-niques, for plausible reasons, a cen-trifuge facility is no major emitter ofcharacteristic signatures that would bereadily detectable via WAEM.

Again, safeguards are obviouslynot designed for this category of pro-liferation scenarios. The provisions inthe IAEA Additional Protocol dohowever facilitate detection of unde-clared facilities substantially, althoughnot necessarily in a timely manner.

Breakout ScenarioFinally, an existing and declared en-richment facility may be used forovert HEU production in a breakoutscenario. The use of nuclear technolo-gies under national control for thisproliferation path can never be ex-cluded once a political decision hasbeen taken to violate a safeguardsagreement or related internationaltreaties, to which the country is a par-ty. Nonetheless, contrary to other en-richment processes, a centrifuge facili-ty can be reconfigured and used toproduce HEU without delay – a factthat is obviously of serious concern inthe present context.

Due to the high separation fac-tor of a modern gas centrifuge, only asmall number of stages is required toenrich uranium to significant urani-

um-235 fractions. The hold-up timeof the material in each stage reported-ly is in the order of 10-20 seconds and,as a consequence, the total mass ofuranium (as UF6 gas) present in thecascade is extremely low, typically be-tween several 100 grams and one kilo-gram.10 A low inventory is equivalentto a short period required to achieve anew enrichment level, which typicallyis in the order of one hour. It shouldbe emphasized that alternative enrich-ment processes, like the gaseous diffu-sion process or the chemical exchangeprocess – the latter studied in the1980s – have equilibrium times in theorder of several months or even years,which makes facilities based on theseprocesses rather unattractive assets torely upon in a breakout-scenariocompared to the alternative plutoni-um-extraction route.

What Can Be Done About It? Moving Beyond Traditional Frameworks

For the above reasons, in many cir-cumstances, even a comprehensivesafeguards approach may be consid-ered inadequate to address all prolif-eration risks associated centrifugetechnology. As a remedy, the idea ofmultinational operation of sensitivenuclear facilities – or even of an inter-nationalized nuclear fuel cycle –comes to mind. This idea is not new.In fact, the concept was invoked in theearliest days of post-war organizationof atomic energy.

To begin with, it is worthwhileto examine existing arrangements andassess their non-proliferation impact.Both existing multinational arrange-ments for uranium enrichment, Uren-co and Eurodif,11 were not designedwith non-proliferation criteria inmind, but mainly to minimize andshare economic risks involved in de-veloping enrichment technologiesable to compete with existing U.S.supplies. It is not surprising, there-fore, that these arrangements do nothold out very well against an analysisof non-proliferation criteria. Table 3compares both arrangements with re-spect to their non-proliferation attrib-utes using a set of criteria based on anearlier analysis.12

A detailed discussion of multi-national concepts is beyond the scopeof this article,13 but the single mostimportant criterion from the non-proliferation perspective is the re-quirement that multinationally oper-ated fuel cycle facilities have to be asubstitute of, rather than an additionto, corresponding facilities under na-tional control. Otherwise, the virtuesof multinational operation are almostentirely lost. This is obviously themost fundamental flaw of the Urencoarrangement since each participantcarries out national centrifuge re-search and development – and eachpartner ultimately built its own en-richment facility.14

Nonetheless, multinational facil-ities, even “poorly” designed arrange-ments, have indisputable and impor-tant virtues – and even though theyare no stand-alone solution to addressemerging proliferation concerns, theymay represent the only approach ca-pable of minimizing proliferationrisks in the long-term, especially if thecriteria of Table 3 are applied whensuch arrangements are set up.

Urgent Action Needed

Any future energy scenario partiallybased upon nuclear energy will re-quire large-scale operation of enrich-ment facilities. As shown above, themodern gas centrifuge, which is thefavored enrichment technology todayand already dominates the market, ishighly proliferation-prone and diffi-cult to safeguard and detect.

Nonetheless, the U.S., France,and China are gradually abandoningtheir gaseous diffusion plants andplan to replace them with centrifugefacilities. Capacities in the originalUrenco countries are being expandedand additional countries, often with-out significant domestic nuclear pro-grams (like Brazil or Iran), are inde-pendently pursuing centrifugedevelopment. It is therefore likelythat a growing number of countrieswill have access to centrifuge tech-nology in the near future. Unfortu-nately, no alternative enrichmentprocesses with more favorable non-proliferation characteristics havebeen seriously considered since the

Beyond A.Q. Kahn

INESAP Information Bulletin No.23, April 2004 4

Alexander Glaser

1970s and, hence, no alternative tech-nology can compete economicallywith the gas centrifuge today. Finally,we might also witness an erosion ofthe technological barriers thatslowed down the spread of centrifugetechnology in the past.

The response to this dilemmahas to be manifold – envisioning bothshort-term and long-term strategies.First, in addition to measures that aretaken as a response to the exposure ofthe A.Q. Khan network, the currentsafeguards standards should be re-vised and substantially upgraded. Re-luctance of those already operatingsafeguarded facilities to do so is pre-dictable, but it will be key to convinceall parties involved that the objectivesare of utmost importance – and also intheir own best interest.

France and the U.S., both plan-ning to build new enrichment facili-ties based on the centrifuge process,must set positive examples in this re-spect and design their facilities safe-guards-friendly from the ground up.

In the longer term, these meas-ures alone will not be sufficient. Withcentrifuge technology in widespreaduse, we are apparently reaching alimit where national control plus in-ternational safeguards becomes anunacceptable compromise. The Di-rector General of the IAEA, M. El-Baradei, in his Statement to the Fifty-Eighth Regular Session of the UnitedNations General Assembly on No-vember 3, 2003, emphasized this im-portant aspect: “In light of the in-creasing threat of proliferation, bothby States and by terrorists, one ideathat may now be worth serious con-sideration is the advisability of limit-ing the processing of weapon-usablematerial (separated plutonium andhigh enriched uranium) in civiliannuclear programmes – as well as theproduction of new material throughreprocessing and enrichment – byagreeing to restrict these operationsexclusively to facilities under multi-national control. These limitationswould naturally need to be accompa-nied by appropriate rules of assur-ance of supply for would-be users.”

Are such proposals realistic? –Most think that they are not.Nonetheless, previous fruitless dis-

cussions of multinational or interna-tional fuel cycle arrangements werefocused on reprocessing facilities, in-cluding plutonium and waste storage.Especially, the Western Europeancountries were suspicious about suchinitiatives in the 1970s and 1980s, at atime when they were expanding theirnuclear programs or exports.15

Today, there is certainly a higherlevel of convergence of internationalnuclear fuel cycle policies, less eco-nomic competition on the interna-tional export market for nuclear tech-nologies, and broader acceptance ofnuclear supplier guidelines. Moregenerally, in the political arena, thereis now nearly universal support of theNPT and a strong sense of urgencythat the regime has to be strength-ened. If the A.Q. Khan incidentserved as a wake-up call for the inter-national community to get such a dis-cussion started, something usefulcould ultimately emerge from the cur-rent crisis of the NPT regime.

The prospects of international-ization and progress in nuclear disar-mament are closely related. The recentrenewed interest of the U.S. in newnuclear weapons, the possibility of re-sumed nuclear testing, and the newnuclear posture are undermining anypossible progress in the area of reduc-ing the military relevance of nuclearweapons – and hence also, of a moreinternationalized structure of the nu-clear fuel cycle.

1 This strategy is exemplified in U.S. Presi-dent G.W. Bush’s speech given at the Na-tional Defense University on February 11,2004, articulated in the following excerpt:“The world’s leading nuclear exportersshould ensure that states have reliable accessat reasonable cost to fuel for civilian reac-tors, so long as those states renounce enrich-ment and reprocessing. Enrichment and re-processing are not necessary for nationsseeking to harness nuclear energy for peace-ful purposes. The 40 nations of the NuclearSuppliers Group should refuse to sell en-richment and reprocessing equipment andtechnologies to any state that does not al-ready possess full-scale functioning enrich-ment and reprocessing plants.”

2 Data for centrifuge facilities retrieved fromIAEA Integrated Nuclear Fuel Cycle Infor-mation System (iNFCIS) in March 2004, ex-cept for the case of India and the uncon-firmed case of Israel.

3 In addition to the radial effect, an axial flowof the gas along the rotor is established,which substantially increases the separativepower of the centrifuge and also allows forconvenient extraction of product and wasteat the ends of the rotor.

4 There exists an extensive literature on iso-tope separation and cascade theory, cf. forinstance: D. G. Avery and E. Davis, Urani-um Enrichment by Gas Centrifuge, Mills &Boon Limited, London, 1973; or S. Villani,Isotope Separation, American Nuclear Soci-ety, 1976.

5 See footnote 2.6 For a discussion, see: A. von Baeckmann,

Implementation of IAEA Safeguards inCentrifuge Enrichment Plants, Proceedingsof the Fourth International Conference onFacility Operations-Safeguards Interface,September 29 – October 4, 1991, Albu-querque, New Mexico, pp. 185-190.

7 A. S. Krass, P. Boskma, B. Elzen, and W. A.Smit, Uranium and Nuclear Weapon Prolif-eration, Stockholm International Peace Re-search Institute (SIPRI), Taylor & Francis

Urenco Eurodif

Only NPT parties Yes Yes (was: No)Governmental participation Yes No

International safeguards Yes Yes (was: No)Withdrawal exclusion No No

Prohibition of national control No NoMultinational R&D only No No

One facility No YesProhibition of technology transfer No No

Exclusion of internal technology sharing No Yes (in theory)HEU production exclusion No No

Proliferation-resistant process No No (was: Yes)

Table 3: Non-proliferation criteria applied to the Urenco and Eurodif frameworks

INESAP Information Bulletin No.23, April 20045

Ltd, London and New York, 1983. For acomparison of different enrichmentprocesses, see p. 19, Table 2.1.

8 For instance, flow rates and enrichment lev-els are monitored in the Chinese Shaanxi en-richment facility, which uses Russian cen-trifuge technology. For a detaileddiscussion, see: A. Panasyuk, et al., Tripar-tite Enrichment Project: Safeguards at En-richment Plants Equipped With RussianCentrifuges, IAEA-SM-367/8/02, 2001.

9 About 6000 separative work units (SWU)are needed to produce 25 kg of weapon-grade HEU from natural uranium. As canbe inferred from aerial views of existing cen-trifuge facilities, the specific capacity of typ-ical plants is 10-20 SWU/yr per square me-ter of the facility’s total footprint. Urencoquotes 40-50 kWh per SWU for its ad-vanced technology.

10 See Krass, op. cit., p. 45 and p. 133 (Table6.2), for exemplary numerical data.

11 The Urenco and Eurodif frameworks arediscussed in: L. Scheinman, MultinationalAlternatives and Nuclear Nonproliferation,International Organization, Vol. 35, No. 1,Nuclear Proliferation: Breaking the Chain,Winter 1981, pp. 77-102.

12 Krass, op. cit., p. 71 (Table 3.1).13 See for instance Scheinman, op. cit., for an

excellent discussion of virtues and limits ofmultinational frameworks.

14 The concept did not solve the problem ofdissemination of centrifuge design informa-tion. Urenco involuntarily became a sourceof centrifuge technology for proliferatingstates, namely directly in the cases of Pak-istan and Iraq – and from Pakistan to someother countries.

15 For a contemporary discussion, see: J. S.Nye, Maintaining a NonproliferationRegime; and P. Lellouche, Breaking theRules Without Quite Stopping the Bomb:European Views. Both in International Or-ganization, Vol. 35, No. 1, Nuclear Prolifer-ation: Breaking the Chain, Winter 1981, pp.15-58.

Alexander Glaser is a physicist and re-search associate with the InterdisciplinaryResearch Group in Science, Technology,and Security (IANUS) at DarmstadtUniversity of Technology, Germany;alexander.glaser@physik.tu-darmstadt.de.

Beyond A.Q. Kahn