Chapter Vll

I

Sources of Nuclear Material

Preceding page blank

Chapter VII

Sources of Nuclear Material

A nation planning the development of nuclear weapons has several optionsfor obtaining fissile material. Diversion from commercial nuclear power facilitieshas received the greatest attention recently: nuclear material could be obtainedthrough the evasion of safeguards or the use of unsafeguarded facilities, possiblyfollowing abrogation of safeguarding agreements. The other routes are the con-struction of dedicated facilities, such as a small plutonium production reactor ora weapons-grade enrichment plant, and purchase or theft of weapons material orcomplete weapons. Each of these routes is subject to constraints and each coun-try will weigh the options differently depending on its own resources,capabilities, political situation, and intentions.

DIVERSION FROM COMMERCIAL POWER SYSTEMS

Although none of the nations that havenuclear weapons have obtained them by thismeans, it is possible that a nation could extractthe fissile material needed for nuclearweapons from its commercial nuclear powersystems. This section will examine existingreactors and several under development,along with their complete fuel cycles. Withthis background, the relative difficulty ofdiversion from each system can be understoodand compared. In the past, resistance to diver-sion has not been a parameter in the design ofnuclear power systems. As diversion is in-creasingly seen as a problem, research isbeginning on reducing the vulnerability of ex-isting systems. Some preliminary conceptualwork has also been done on reactor systemsthat are inherently resistant to diversion,

The Fuel Cycle

The flow of nuclear material in a commer-cial power program—from the mine, through

the reactor, to disposal or reuse—is called thenuclear fuel cycle. The nuclear materials of in-terest for either an explosive or a powerplantare those that release extra neutrons andenergy when they fission, or split apart. Suchfissile isotopes are not abundant in nature,although some are produced as a byproduct ofpower production: neutrons striking certainnuclei will convert them, after a short decaychain, to fissile isotopes.

Two general fuel cycles exist, each based ona different element. In the uranium cycle, theisotope U238 does not fission easily but doesbreed a fissile isotope of plutonium, Pu239. Afissile isotope U235 is also present in naturaluranium. In the thorium cycle, the thoriumisotope Th232 breeds the fissile isotope, U233.Within each of these fuel cycles, quantities andconcentrations of various isotopes, and theprocedures for processing them, vary with theparticular reactor type.

The two types of nuclear power reactorsavailable on the world market today both use

151Preceding page blank

the uranium fuel cycle. These are light waterreactors (LWRS) developed by the UnitedStates and Canadian heavy water reactors(CANDU). Others which have been largelydeveloped and could be deployed in the nearfuture are the high-temperature gas reactor(HTGR), and the advanced gas-cooled reactor(AGR). Most development effort in severaladvanced countries is focused on the liquidmetal fast breeder reactor (LMFBR), but com-mercialization is not expected for at least 10years. Development of another breeder reac-tor, the light water breeder reactor (LWBR) isproceeding at a slower pace.

All these reactors, plus a few others thatcould become important, are described indetail (along with their fuel cycles) in volumeII, appendix V. Because they are of immediateinterest, the fuel cycles and diversion potentialof LWRS and CANDUS will be summarized inthis section. The LMFBR, LWBR, and thoriumcycle in general will also be examined brieflyin this chapter. Research plans on alternatefuel cycles are briefly summarized. Concep-tual studies on inherently nonproliferatingreactors are described at the end of this sec-tion. Safeguards to prevent and/or detectdiversion are discussed under “Safeguards” inchapter VIII.

Light Water Reactors

Technical Description

The common types of light water reactorsdiffer in the coolant they use-either boilingwater (as in BWRS) or pressurized water (as inPWRS). They present identical problems forproliferation prevention, and will be con-sidered together here. Key characteristics ofthese reactors and their fuel cycles are given infigure VII-1.

The first stage in the LWR fuel cycle is themining of ore, which contains about 1,500parts-per-million uranium. The millingoperation then concentrates the uranium bystraightforward chemical processes intoyellowcake (U3 O8). By far the largest percent-age of natural uranium (99.3 percent) con-sists of the isotope U238. Only 0.7 percent isU 235, the isotope that will fission in a LWR.

Because LWRS are designed to operate with aU235 concentration of about 3 percent, naturaluranium must be enriched. In preparation forenrichment, yellowcake is converted in aspecial plant to uranium hexafluoride (UF6),which is a gas at a sufficiently low tem-perature to permit easy handling.

Although several enrichment techniquesare known, the only full-scale plants built todate use the principle of gaseous diffusion.The separation achieved in one stage of adiffusion plant is small, so a series of stages,called a cascade is required to raise the U235

concentration to the desired level of enrich-ment. A gaseous diffusion plant must be verylarge to be economical. Commercial plants arebuilt to. serve at least 50 large (i.e., 1000MW(e)) reactors. Each plant costs severalbillion dollars and consumes a large amountof electrical power (about 3 percent to 5 per-cent of the energy produced by its enrichedproduct). Another enrichment technique-gascentrifuge-can achieve a greater separationfactor between the isotopes in each stage. Itappears to be economical on a smaller scaleand requires much less electrical power. Thistechnique has ‘not yet progressed beyond thepilot plant stage, but several new commercialplants of this type are in the planning stage.

Other enrichment methods are muchfurther from commercialization. Such newtechnologies should be watched as theydevelop, since they may become inexpensiveand simple enough to be attractive to manycountries.

After enrichment, UF 6 is converted touranium dioxide (U02) and fabricated intofuel assemblies. The fuel assemblies areshipped and loaded into the reactor, wherethey remain for several years. One third of thefuel assemblies are replaced each year in apressurized water reactor (PWR), and onefourth in a boiling water reactor (BWR).Refueling involves shutting down the reactor,allowing it to cool, removing the reactor head,and transferring the spent fuel underwater toa storage pool. The entire process takes 4 to 6w e e k s .

At present, spent fuel is simply stored at thepowerplant or in spent-fuel pools at otherlocations. The intention of the industry is to

152

Iw ’

,

A\

. .. — . . .

153

reprocess spent fuel to recover unfissionedU235 and plutonium that was generated fromU 238. plutonium can directly replace U235 a sfissile material in fuel, thus reducing demandfor uranium ore and the need for enrichment.If it is found undesirable for economic, politi-cal, or safety reasons to reprocess spent fuel,the cycle can terminate at this point. If this isthe case, long-term storage facilities or a per-manent disposal method for the spent fuelrods must be planned.

If reprocessing does occur, spent-fuel ele-ments are sent to the reprocessing plant inlarge, heavy shipping casks designed to pro-vide both shielding against intense radiationand cooling to remove decay heat. At thereprocessing plant, the fuel elements arechopped up and the contents dissolved inacid. Solvent extraction is then used to sepa-rate plutonium and uranium from the fissionproducts, which are stored for eventual dis-posal. The plutonium and uranium emerge inseparate streams. The uranium is converted toUF6 for reenrichment, and the plutonium toplutonium dioxide (PuO2 either for stockpil-ing or recycling. All operations in thereprocessing plant must be performed byremote control, because of the intense radioac-tivity of spent fuel and the toxic nature ofplutonium.

If the plutonium is to be recycled, the PU02

is shipped to a mixed-oxide fuel fabricationplant, where it is combined with U02 so thatthe final mixture will contain the desired frac-tion of fissile isotopes. A mixed-oxide fuelfabrication plant is more expensive than auranium fuel fabrication plant, becauseremote handling is required for plutonium.

Diversion From the LWR Fuel Cycle

Material convertible to weapons gradecould be diverted at any point of the LWR cy-cle, but the difficulty of conversion, and hencethe attractiveness of the diverted materialvaries markedly from point to point. This sec-tion will provide an estimate of the amount ofmaterial that must be diverted at each stage ofthe LWR fuel cycle in order to produce onenuclear explosive, give a summary of theoperations that must be performed upon the

material to convert it to a form that can bedirectly used in a nuclear weapon (processdetails are given in volume II, appendix V),and assess the feasibility of (a) a nation and(b) a non-national group performing theseoperations.

The safeguards that a diverter or thiefwould have to evade or surmount aredescribed under “Safeguards” in chapter VIII,and are only briefly mentioned in this section.The resources required to construct a nuclearweapon once weapons material has been ob-tained are discussed in chapter VI.

Yellowcake (i.e., U 30 8) from a uraniummill, after a few chemical steps, could beenriched to weapons-grade uranium or par-tially transformed to Pu239 in national dedi-cated facilities, as described in the “DedicatedFacilities” section of this chapter and involume II, appendix VI.

A p p r o x i m a t e l y 6 . 5 m e t r i c t o n s o fyellowcake would have to be fed to an enrich-ment plant to yield 30 kilograms of 90 percentU235 (enough for one or two explosives). Ap-proximately 75 metric tons of yellowcakewould be required to supply enough naturaluranium to fuel a dedicated production reac-tor that would produce 10 kilograms of Pu239

per year (enough for one or two explosives).In the latter case, it would not be necessary torefuel the dedicated production reactor morethan once every 10 years or so. However, thenation probably would prefer to refuel everyyear or two in order to obtain weaponsmaterial quickly and steadily.

The capital cost of a reactor and reprocess-ing plant that could produce one or two ex-plosive’s worth of plutonium per year, start-ing with yellowcake, is in the tens of millionsof dollars. This effort is within the capabilitiesof many (perhaps close to 50) nations, but isentirely impractical for a non-state adversary.The cost of a small enrichment facility is morecomplex to assess; it is discussed under “Dedi-cated Facilities” in chapter VII. It is also en-tirely impractical for a non-state adversary.

International Atomic Energy Agency(IAEA) and Euratom safeguards exist foryellowcake (in fact, Euratom safeguards startwith uranium ore), but a country in the

154

market for yellowcake to supply a dedicatedfacility would probably have little difficulty inclandestinely purchasing a sufficient amount.Moreover, many countries have considerableresources of uranium ore, and in these coun-tries a dedicated mine and mill could be usedto supply a dedicated facility. (See chapter X.)

An enrichment plant presents a more at-tractive target to the diverter. Although thedesign output of commercial enrichmentplants is only 3 percent to 4 percent U235, andcompletely impossible (not merely impractical)to use directly in a nuclear fission explosive,much of the work to raise the enrichment toweapons grade has been accomplished. For 30kg of 90 percent LWs, nearly 8000 kg ofnatural uranium hexafluoride feed and 6900separative work units (SWU) are required, butif 3 percent U235 is the feed, only about 1500kg of uranium hexafluoride and about 2500SWU are required.

Several options are possible for a nationwhich elects to divert from its own commer-cial enrichment plant. The components of theentire plant could be reassembled so that theproduct would be highly enriched uranium.The change is not difficult for a centrifugeplant, but is complicated and time consumingfor a diffusion plant. Nevertheless, theChinese appear to have followed this route inconverting a U.S.S.R. supplied diffusion plant.This change is too drastic to be done covertlyif the plant were safeguarded.

If the nation had a large, safeguardedenrichment plant, it might choose to convertone section of the plant to a high-enrichmentcascade. Again, this would be difficult to do ina diffusion plant and relatively easy in acentrifuge plant.

An alternate option would be to divert partof the low-enriched uranium product and feedit into a separate, small enrichment plant toboost it to highly enriched uranium. The addi-tional small plant could be either inside thelarge plant or at another site. Only about 400centrifuges of European design would be re-quired to produce 30 kg of 90 percent U235 peryear from 3 percent LWs feed. For com-parison, an enrichment plant of near-competi-tive commercial size to supply ten, 1000MW(e), LWRS with low-enriched uranium

would have a capacity of 1,300,000 SWU peryear and contain approximately 200,000centrifuges of European design. Enrichmentplant safeguards are discussed under“Safeguards” in chapter VIII, but it should benoted here that the scenarios sketched in thisparagraph are not implausible as long as in-spectors are limited to monitoring theperimeter of the facility and unmonitored in-put and output paths are permitted.

As already discussed, enrichment is not anoption for the non-state adversary. However,low-enriched uranium could be an attractivetarget for embezzlers if a criminal blackmarket in low-enriched uranium developed.The black market could conceivably supplylow-enriched uranium merely as a fuel forpower reactors (see chapter V “Non-State Ad-versaries and volume 11, appendix 111 for a dis-cussion of a case of low-enriched uraniumsmuggling), or more ominously, as feed for adedicated national enrichment plant designedto produce weapons material.

From the output of the enrichment plant tothe loading of the reactor the only target in theLWR fuel cycle (without plutonium recycle) isthe low-enriched uranium itself, which must,as discussed above, be boosted to highlyenriched uranium in a dedicated enrichmentplant to be useable in nuclear weapons.

Because of the long time required for refuel-ing a LWR, national diversion of irradiatedfuel could not take place without considerableeconomic and power penalties, except at anormal discharge and loading operation, orfrom the spent-fuel storage pool.

L ight water reactor fue l (wi thoutplutonium recycle) of typical burnup containsabout 0.8 percent plutonium, of which about25 percent is Pu240 plus Pu242. With plutoniumrecycle, high burnup LWR fuel would containabout 1 to 2 percent plutonium of which about3 5 p e r c e n t w o u l d b e P u240 p l u s P u2 4 2

(Detailed data is given in volume II, appendixv.)

A high Pu249 plus Pu242 content is widely—but incorrectly—believed to render plutoniumunsuitable for militarily effective weapons. Ahigh content of these isotopes is a complica-tion; given a free choice, a weapons designer

155

would prefer plutonium with a low Pu240 con-tent, but it should be realized that effectivenuclear explosives can be made withplutonium of the Pu 240 content describedabove. This point is discussed in greater detailin chapter VI.

With a plutonium content of 0.8 percent,approximately 1.4 metric tons of spent fuel inthe form of uranium dioxide (UO2 must bereprocessed (at 100 percent recovery) to ob-tain 10 kg of plutonium. In a PWR, this is con-tained in three fuel assemblies. Stated anotherway, 1 year’s fuel discharge from a PWR con-sists of about 31 metric tons of U02 containingabout 240 kg of plutonium.

As discussed in the “Dedicated Facilities”section of this chapter and in appendix VI,volume II, it is well within the capability ofmany developing countries to construct theirown reprocessing plant to extract plutoniumfrom spent fuel for use in weapons. However,it appears probable that the IAEA will developthe capability to safeguard LWRS and LWRstorage pools so that it will be very unlikelythat a diversion could take place undetected.Thus, national diversion from a safeguardedLWR or LWR storage pool would probably beovert.

If the operator arranges a series of plausiblereactor problems leading to extensivedowntime for the year preceding the diver-sion, low burnup fuel with a low Pu240 con-tent will result. For example, at 5000 MWdays/metric ton burnup, the plutonium con-tent of the last reload would be about 70 kg, ofwhich only 10 percent would be Pu240.

Theft of spent fuel by non-state adversariesis just barely credible. The theft itself and sub-sequent transportation of the highly radioac-tive fuel (which would have to be cooled andshielded in transit) would require a number ofarmed and highly organized adversaries,some of whom would have to be willing to ac-cept considerable, possibly lethal, radiationexposure. Reprocessing of spent fuel by non-state adversaries is also just barely credible,even if the group were very well financed andpossessed practical chemical engineering ex-perience. A crude but technically feasible sol-vent extraction or ion-exchange system can beimagined, but it would require several

months of process time for extracting 10 kg ofplutonium, During that time the group wouldbe immobile and vulnerable to an intensesearch.

From the point of view of both the nationaldiverter and the non-state adversary, a largecommercial reprocessing plant is an attractivetarget. Appendix V of volume 11 discusses thediversion points in a model reprocessingplant. Plutonium nitrate stolen from the ni-trate blending area would require only a sim-ple precipitation to be converted into weaponsmaterial; plutonium dioxide from the conver-sion area could be used directly in a nuclearexplosive. A national diverter would probablytake the further step in either case of conver-sion to plutonium metal. The safeguarding ofa reprocessing plant is discussed under“Safeguards” in chapter VIII, but the pointwill be noted here that materials accountancy,by itself, has neither the sensitivity nor thepromptness to assure timely detection ofcovert diversion from a large reprocessingplant, either by a nation or by non-state ad-versary. Other safeguard measures aretherefore employed, such as portal monitorswhich can detect gram amounts of plutonium.(See chapter VIII.)

A model mixed-oxide fuel fabrication plantis diagramed in appendix V of volume II. Theoutput of the model plant consists of fuel rodswith a mixture of uranium dioxide and up to3.5 percent plutonium dioxide. To obtain 10kg of plutonium, about 300 kg of mixed oxidewould have to be diverted or stolen. Thelogistical problem of removing so muchmaterial is a significant deterrent, but the big-gest obstacle to the non-state adversary ischemically separating plutonium fromuranium. Although conceptually simple, in-volving dissolution followed by ion exchange,the task would need someone with practicalchemical engineering experience and wouldrequire perhaps several weeks to severalmonths, depending on the details of the ad-versary’s separation facility. For the nationaldiverter, the chemical separation problemwould be minor and could probably be ac-complished in one to a few days.

In the portion of the fuel cycle between theoutput of the reprocessing plant, and the

156

Figure VII–2.

Summary of the Diversion Points in the LWR Fuel Cycle

IS THE MATERIALUSEFUL TO THE

NATIONAL DIVERTER?

IS THE MATERIALUSEFUL TO THE

NON-STATE ADVERSARY?FACILITY MATERIAL

Natural uranium(0.7 percent U235 as

ore (0.2 percent uranium)U 3 0 8

UF6

YES, but only as feed for adedicated facility (plutoniumproduction reactor or enrichment

NO (but criminals might engae inblack market in these materials)

MineMillConversion Facility

plant)

low enriched uranium(3 percent U235 as

UF6

U02 YES, but only as feed for adedicated enrichment plant

Nation would eventually havereplace fuel

NO (Criminals might engage inblack market in these materials)

Enrichment PlantUranium Fuel Fabri-cation PlantTransportation toReactorTemporary Storageat Reactor

U02 in fuel assemblies to

Pu

YES; dedicated reprocessingfacility required

NO except Yes for large, verywell financed, technically compe-tent group with a secure base ofoperations and a few memberswilling to risk radiation injury

Reactor SpentFuel Storage

Pu —about 0.8 percent inhighly radioactive spentfuel

YES; Nation would probablyconvert material to metallicplutonium

Reprocessing Plant*Transport to fuelfabrication plantInput area tofuel fabricationplant

Plutonium FuelFabrication Plant

Pure Pu(N03)4

orpure Pu02

YES; If Pu(N03)4, simple conver-Asion to PuO2 equmd. It

Pu02, material directly usable inexplosive

YES, Chemical separation offrom mixture only a minor

Yes, BUT chemical separation aPu02 (3 percent to 7percent) mixed with over90 percent U02

time consuming operation.obstacle. Logistics of diverting100 to 300 kg of material for oneexplosive troublesome.

YES, as above.(Nation would eventually have toreplace fuel)

Logistics of stealing or diverting100 to 300 kg of material for oneexplosive cause problems.

Transport to ReactorTemporary Storageat Reactor

About 1 percent Pu asPu02 mixed with U02 in

Yes, BUT chemical separation atime consuming operation.Logistics of stealing completefuel assembliesfuel assembles present signifi-cant obstacle.

“With copreclpttatlon, however, dwerslon potential at these pants would be slmllar to dwerslon potential at plutoruum fuel fabrication plant. I e considerably less for the non-stateadversary and somewhat less for the national diverter SOURCE OTA

mixed-oxide blending area of the mixed-oxidefuel fabrication plant, plutonium would existin the form of plutonium dioxide. Thismaterial is directly useable in the fabricationof nuclear weapons, although a nation wouldprobably convert it to plutonium metal. Thisportion of the fuel cycle, which includesstockpiled plutonium, presents the most con-centrated target for diversion. Although onecan conceive of very stringent safeguardsagainst covert diversion even in this exposedportion of the cycle, safeguards, by theirnature, cannot prevent a nation from seizing aplutonium stockpile attached to its ownreprocessing plant. As discussed in chapter VI,a modest national weapons developmentprogram can attain a high degree of confidence inthe performance of its weapon withoutnuclear testing. Once the political decision istaken to seize the stockpile, the nation canhave a reliable explosive in a matter of days toweeks, even using reactor-grade plutonium.

Summary of Diversion Points in the LVVR FuelCycle. —The preceding discussion hasdescribed the diversion points in the LWR fuelcycle, specified how much material wouldhave to be stolen or diverted at each point toyield material for one or two explosives, andhas evaluated the difficulty of chemical andphysical processing necessary to convert thediverted material into weapons material.Figure VII-2 briefly and qualitatively sum-marizes this discussion.

The Canadian Deuterium Reactor(CANDU)

Technical Description

The Canadian Deuterium Reactor (CAN-DU) is able to operate with natural uraniumbecause heavy water absorbs fewer neutronsthan does ordinary water, leaving more tocarry on the chain reaction. This eliminatesthe need for the entire enrichment process, in-cluding UF6 conversion. The mining and mill-ing processes are the same as for LWRS, butreactor operation is substantially different.The CANDU is designed for ondoad refueling.Instead of shutting down and opening thereactor to change a batch of fuel, a refueling

machine opens both ends of one of the manytubes throughout the reactor. These tubes con-tain several short fuel rods, A fresh rod is in-serted in one end and a spent rod removed atthe other. The tube is then resealed andrepressurized with cooling water.

There are no plans at present to reprocessCANDU spent fuel. More plutonium is pro-duced than in an LWR of the same powerlevel, but it is more dilute because of thegreater amount of U238. The fraction of U235 inthe spent fuel is very low (actually less than inthe tails from present enrichment plants), andreprocessing would be less likely to beeconomical than for the LWR cycle. The spentfuel is now being stored indefinitely, pendingdevelopment of a final waste disposal method.

Diversion From the CANDU Fuel Cycle

The CANDU fuel cycle presents considera-bly different opportunities for diversion thandoes the LWR cycle. Separated fissile materialis not exposed anywhere in the CANDU fuelcycle, in contrast to the LWR cycle withplutonium recycle. The enrichment andreprocessing facilities are totally absent. Theonly diversion points in the CANDU fuel cy-cle are the reactor itself and the spent fuelstorage pool.

As in the case of the LWR, non-state theft ofspent fuel from the storage pool followed byreprocessing is just barely credible.

As discussed for the LWR, reprocessing forweapons purposes, spent fuel that has beendiverted from a reactor or spent-fuel storagepool is within the capabilities of many na-tions. The quantity of fuel that must bediverted from a CANDU to yield 10 kg of l%,and the quality of the Pu obtained undervarious conditions, is discussed below.

CANDU fuel of normal burnup (about 7500MW days/metric ton) has a plutonium contentof about 0.4 percent of which about 25 percentis Pu240. As described in appendix V ofvolume II, CANDU is refueled continuouslyand some fuel bundles could be pushedthrough more rapidly for lower burnup andlower Pu240 content. At a burnup of 2500 MWdays/metric ton (one-third normal) the

158

plutonium content is about 0.2 percent, ofwhich only 10 percent is Pu240. To obtain 10kg of plutonium at least 5700 kg of low burn-up uranium-oxide fuel would have to bediverted, or about 260 fuel bundles, For nor-mal burnup fuel about 2800 kg, or 130 bun-dles, would have to be diverted, For com-parison, in the CANDU-600 model, about 12fuel bundles are normally pushed through thereactor per day.

In contrast to the LWR, production of lowpu240 plutonium in the CANDU does not in-volve a significant loss of power output.

Safeguard systems for a CANDU reactorand storage pool can probably be designedand implemented so that repeated covertdiversions of fuel assemblies cannot take placeundetected during either normal or acceler-ated refueling. Diversion from the CANDU istherefore also likely to be overt,

Liquid Metal Fast Breeder Reactor(LMFBR)

Technical Description

The LMFBR is expected by the industry inevery nuclear supplier nation except Canadato be the successor to the LWR reactor since itwould essent ia l ly e l iminate uraniumresources as a constraint. The reactor is some-what analogous to a PWR, except that it usesliquid sodium at low pressure as a coolant andhas no moderator. The fuel is mixedplutonium (10 percent to 20 percent) and U238

oxide. Radial and axial blankets of U238 sur-round the core to capture escaping neutronsand breed plutonium, The LMFBR is expectedto produce as much as 15 percent more fueleach year than it consumes. This excess (about250 kg per year) can be used to fuel other reac-tors or diverted to a weapons program withno impact on the fuel cycle. Refueling is simi-lar to LWRS, with about one-half the core andone-third the blanket replaced each year. Noenrichment is required, except possibly for theinitial core, because the plutonium that is bredcan be used in subsequent cycles. Reprocess-ing, however, is central to the LMFBR cycle.The bred plutonium cannot be recoveredwithout reprocessing, and the whole point ofthe LMFBR is that it can breed enough

plutonium to refuel itself and to start up newreactors.

Diversion From the LMFBR Cycle

The diversion points in the LMFBR cyclecan perhaps be best explained by comparingthem to those of LWR cycle with plutoniumrecycle.

The mining and milling stages can be vir-tually eliminated, because the depleteduranium contained in the tails from presentenrichment plants can be used. Enrichment issuperfluous, except possibly for the intialcore.

As in the case of LWR recycle, thereprocessing plant, fuel fabrication plant, andfresh-fuel storage area at the reactor, includ-ing the transportation links between them, arethe points most vulnerable to diversion.

Diversion from the reactor itself is not cred-ible and the material in the spent-fuelstorage pool, in transit to the reprocessingplant, and in the input stages of the reprocess-ing plant is highly unattractive to the diverterbecause of its fierce radioactivity. However, asin the case of the LWR, handling andreprocessing diverted spent fuel in a smallreprocessing plant dedicated to the task iswithin the capability of many nations.

The input to the fuel fabrication plantwould consist of depleted or natural uraniumdioxide and pure plutonium dioxide from theoutput of the LMFBR reprocessing plant, withpossibly an additional contribution from aLWR reprocessing plant or stockpile. Theuranium dioxide and plutonium dioxide willbe mixed at the fabrication plant and com-pressed into fuel pellets. The ratio ofplutonium to uranium in the fresh fuel varieswith the exact design proposed, but would bein the range of 1:10 to 1:5. At 1:5, the materialwould be of only marginal usefulness in anuclear explosive; at 1:10 the material couldnot be used directly in a practical nuclear ex-plosive. However, only 55 to 110 kg of fuelwould have to be stolen to obtain 10 kg ofplutonium. Fresh fuel for the LMFBR wouldbe a factor of 2- to 6-times more concentratedin plutonium than fresh fuel in the LWR cyclewith plutonium recycle, depending on thedetails of both schemes.

159

Details on LMFBR reprocessing are not firmas yet. In general, diversion opportunities atan LMFBR reprocessing plant would be simi-lar to those at an LWR reprocessing plant,heightened by the fact that the throughput ofplutonium per metric ton of fuel input wouldbe greater by a factor of approximately 10.

In general, approximately 5 times as muchplutonium would flow through the LMFBRcycle as through the LWR cycle, for the sameamount of electricity generated. (LMFBR fuelgives about twice as much electricity permetric ton as does the LWR.)

In addition to the quantitative differencesbetween the two cycles (there is m o r eplutonium in the LMFBR cycle and it is moreconcentrated), there is also a potential qualita-tive difference. A significant amount of theplutonium produced in the blankets will con-tain less than 5 percent Pu240, i.e., it will beweapons-grade plutonium in a normal fuelcycle. (See’ chapter VI.) In the LWR cycle,plutonium of this quality is produced only byoperating with frequent, very costly refuel-ing.

Thorium Fuel CyclesPower-reactor fuel cycles employing

thorium have received much less attentionthan uranium fuel cycles. The thorium fuel cy-cle uses U233 as the fissile isotope and Th232 asthe fertile isotope. Several reactors have beenproposed that might employ thorium. A high-temperature gas-cooled reactor (HTGR) isoperating, and a demonstration light-waterbreeder reactor (LWBR) is presently beingconstructed.

Thorium fuel cycles that have been studiedinclude:

. High-Temperature Gas Reactor (HTGR);

. Light-Water Breeder Reactor (LWBR);

. Light Water Reactor (LWR);● Heavy Water Reactor (HWR);. Molten Salt Breeder Reactor (MSBR); and. Thorium and mixed thorium/uranium

fuel cycles in fast breeder reactors (FBR).

The limited availability of uranium is oftencited as a major reason for considering thethorium cycle. However, although it isassumed that thorium is 3- to 5-times moreplentiful than uranium throughout the world,160

the actual quantity of thorium,concentration of the ores, are in

and the likelyfact uncertain.

In thermal reactors the thorium fuel cyclemay permit (1) a more efficient use ofresources, possibly including the operation ofa breeder, which is impossible with theuranium/plutonium cycle. (U233 produces, onthe average, 2.28 neutrons per thermalneutron capture, versus 2.11 for Pu239. Thisprovides just enough extra margin so thatbreeding may be possible.) ; (2) moreeconomic power generation than that fromLWRS (uranium cycle) if uranium costs con-tinue to increase (provided thorium costs arelow); and (3) a delay in the need for fastbreeder reactors (FBR) and a lower eventualdemand for them because the demand foruranium would not be as great with thoriumfuel cycles supplying some power.

In fast breeder reactors a thorium or mixed-fuel cycle may permit (1) a larger margin ofsafety in the control of the reactor; and (2)production of a fuel which could be employedfor both fast and thermal reactors. (Thermalthorium-based reactors have a breeding rationear one, so they produce little, if any, excessfuel.)

The thorium fuel cycle has both dangersand inherent safeguards from the prolifera-tion point of view. The fuel that it breeds, U233

is an excellent weapons material, with a criti-cal mass approximately one-third that of U235.It is comparable in weapons-material qualityto Pu239. However, some protection againstdiversion is offered by the unavoidable pro-duction of U232 when U233 is produced. U232 isthe first in a chain of radioactive decays whicheventually yields thallium-208, which emits apenetrating 2.6-MeV gamma ray. The fabri-cated fuel and fuel materials are radioactiveand present a definite health hazard a fewdays after separation. After several years, theradiation dose from kilogram quantities ofU233 becomes high enough to rapidly deliver alethal dose to anyone in direct contact.

Anyone diverting U233 fuel would have toovercome radiation hazards to obtain andtransport the material, and to fabricate aweapon. The radiation also results in two in-direct safeguards advantages. First, access tothe material is limited by the requirement for

remote handling behind radiation shielding.With little likelihood of any hands-on opera-tions, access to the material for diversion pur-poses is much more difficult. Secondly, thepenetrating 2.6-MeV gamma ray enables por-tal monitors to detect extremely small(milligram) quantities of U233.

However, the radioactivity of U233 fuel isprimarily a safeguard with respect to non-na-tional adversaries. A national diverter couldeasily provide the radiation shielding neces-sary to handle the material. Indeed the coun-try would have to provide the shielding toutilize the thorium fuel cycle in its powerreactors,

The radiation hazards of U232 unfortunatelycreate problems for safeguard inspectors aswell as potential diverters. The necessity forremote handling may limit the accuracy ofsafeguard ‘measurements.

A key feature of the thorium fuel cycle rela-tive to proliferation control is the fact thatU 2 3 3 can be denatured. That is , i t Can be mixed

with the abundant U238 in concentrations ofabout 12 percent or less in order to make itunuseable in a practical nuclear explosive. Bycontrast, Pu239 cannot be denatured, as thereare no plutonium isotopes that could be mixedwith Pu239 that would preclude its use as anuclear weapons material. (See chapter VI“Nuclear Fission Explosive Weapons”.)

The number of gas centrifuges necessary toenrich U233 that has been mixed with U238 issignificantly less than that required to enrich amixture of U235 and U238 to the same degree.As a practical matter, however, the enrich-ment of denatured U233 would be difficult dueto the significant radiation danger involved.Contact maintenance would be very hazard-ous. The costs and technology required forremote maintenance on a gas-centrifugeenrichment facility would be high.

The characteristics of reactors that mightuse thorium fuel cycles are not well definedbecause most have only been studied onpaper. High-temperature gas reactors are themost advanced of all these concepts, with asmall commercial plant (the 330 MW(e) FortSt. Vrain plant) in operation. However, as dis-cussed in the following section, HTGRs expose

highly enriched uranium throughout theirfuel cycle. An LWBR demonstration plant isnow being completed. A very small MSBR hasbeen operated successfully, The others are stilldesign concepts. High capital costs associatedwith HWRS, due to the use of pressure tubes,large cores, and heavy water, and with LWBRS(including the costly prebreeder), may be asignificant disadvantage.

Conclusions on the Thorium Cycle

Thorium cycles look attractive from a non-proliferation point of view, and they areespecially resistant to diversion by non-stateadversaries. Selected thorium cycles should befurther studied to better define theireconomic, technical, and safeguards promise(e.g., see section on “Alternate Fuel Cycles andNonproliferating Reactors” below).

High-Temperature, Gas-CooledReactor (HTGR)

A small (330 MW(e)) commercial HTGR isnow operating near Fort St. Vrain, Colo. WestGermany is constructing a 300 MW(e) plantbased on a variation of this concept. Both arecooled by helium and moderated by graphite.

The outstanding feature of the HTGR froma proliferation standpoint is its use of highly

enriched (93 percent U235) uranium fuel parti-cles. These fissile particles of uranium carbide,with a hard coating of carbon and silicon car-bide, are mixed with fertile particles ofthorium dioxide in the fuel elements. Thisfresh fuel would be attractive to a diverter.Separating the uranium from the manufac-tured fuel should be possible, even for a sub-national group, although their process wouldprobably be clumsy and inefficient.

The HTGR must be shut down for refueling,Recycling is required to recover the bred U233

and the remaining U 235. As discussed involume II, appendix V, the relative economicmerits of various HTGR reprocessing andrecycling programs have not been fully evalu-ated, but they may favor a one-time recycle.

Developers of the HTGR are studying alter-nate designs that would use lower enrichedfuel.

161

Light-Water Breeder Reactor(LWBR)

Technical Description

The light-water breeder reactor (LWBR)relies extensively upon LWR technology andhas the major purpose of producing as muchfissile material as it uses. The present conceptsare based on the pressurized water reactor(PWR), and maybe implemented by placing adifferent reactor core and control system inpresent PWR reactor plants. A demonstrationoperation in the Shippingport reactor isscheduled for the late 1970’s.

The LWBR is a thermal reactor whichwould convert thorium to U233. Because thebreeding (conversion) ratio is near one,prebreeders are required to produce enoughU233 for the first few breeder cores.

The basic core design utilizes the seed-blanket concept, in which each fuel modulecontains fissile regions (seeds) and a fertileblanket. A low-water content in the core is re-quired to minimize neutron capture in hy-drogen, so a water-to-metal ratio of aboutone-tenth that of the standard PWR has beenproposed. Safety problems are exacerbated bythis difference.

To avoid parasitic neutron capture in con-trol rods, control is achieved by axial move-ment of the fuel modules in relation to eachother. Fertile blankets increase the size of thecore but capture neutrons that would other-wise be lost to the system.

It is expected that the reactor will berefueled in a manner similar to the LWRS. Thereactor will be shut down for a period of up to30 days, and the pressure vessel head taken offand a portion of the fuel removed.

Diversion From the LWBR Fuel Cycle

For the prebreeder, the first point at whichthe diversion potential differs from the LWRcycle is at the enrichment plant. Prebreederfuel will contain 10 percent to 13 percent U235.Although this enrichment is too low to be

used directly in a nuclear explosive, it pro-vides excellent feed for a dedicated enrich-ment plant. About 440 kg of 10 percent U235

hexafluoride feed would be required to pro-duce 30 kg of 90 percent U235, and about 180centrifuges of European design could producethis quantity of highly enriched uranium peryear from 10 percent feed.

Fuel modules for the prebreeder will con-tain uranium dioxide rods and thorium diox-ide rods. No chemical separation of the freshfuel would have to be done to acquire 10 per-cent LWs.

A total fuel discharge from a 1000 MW(e)prebreeder would contain about 100 kg ofplutonium concentrated to about 1 percent inthe uranium dioxide rods, about 300 kg ofU 233 concentrated to about 1 percent in thethorium-dioxide rods, and about 800 kg ofU235 at an enrichment of nearly 8 percent. NO

isotopic separation would be required to obt-ain pure plutonium or pure U 233. Onlychemical reprocessing would be needed to ac-quire material that could be directly used innuclear weapons. As for the LWR, this task iswithin the capability of many nations, but im-possible for all but very technically competentand well financed non-state groups.

The above numbers should be regardedwith caution. Detailed data have not beenpublished for a commercial-sized plant.

Two separate reprocessing plants might beused to reprocess LWBR prebreeder fuel. Thediversion potential for the facility reprocess-ing the uranium-dioxide rods would be verysimilar to that for the LWR facility. In the caseof the thorium-U233 reprocessing plant, a ma-jor difference would be the intense andpenetrating gamma radiation from U232 (asdiscussed in the previous section), renderingdiversion more difficult.

Fuel going back into the prebreeder couldeither be reenriched uranium dioxide plusthorium dioxide, or, more likely, mixedplutonium dioxide plus thorium dioxide, plusuranium dioxide, In the second case the diver-sion potential for fuel fabrication and mixed-oxide fuel assemblies would be similar to thatfor the LWR.

162

The uranium fuel for the breeder is pres-ently seen as being 90 percent U233 and 10 per-cent U235, A reactor load for a 1000-MW(e)core would contain 2,000 kg of this 100 per-cent fissile fuel and 93,000 kg of thorium. Thefuel would consist of mixed uranium dioxideand thorium dioxide pellets. The mixed pelletswould contain about 5 percent uranium diox-ide and 95 percent thorium dioxide. Fresh fuelcould therefore not be directly used in nuclearexplosive weapons, but only chemical separa-tion would be required. This chemical separa-tion would be a time-consuming process forthe non-state adversary.

Pure fissile uranium would be available atthe reprocessing plant.

The LWBR differs from the LMFBR in animportant point. The LMFBR produces a dis-tinct surplus of plutonium over what is re-quired to refuel itself. The LWBR, with abreeding ratio of close to one, produces onlyenough to refuel itself. Thus, fissile materialdiverted by a nation from the LWBR cyclewould have to be replaced from prebreederoutput or stockpiles. The most likely penalty acountry with an expanding LWBR economywould have to pay for diverting from itsbreeder cycle is a slowdown of expansion. Fora country with a static LWBR system and noprebreeding, replacing the diverted fissilematerial would present a serious problem.

Comparison of Reactors

The discussion of diversion from thedifferent reactor fuel-cycle systems has shownlarge differences in the levels and locations ofvulnerability. The vast number of variables,varients, and unknowns make an attempt atquantifying these differences premature.Figure VII-3 presents a qualitative evaluationof opportunities presented by the systems dis-cussed above and in volume 11, appendix V.The ranking is on the basis of the usefulness ofthe fissile material as follows:

A—No significant diversion potential.B—Highly dilute AND substantially

radioactive material. Diversion isbarely credible for the non-stateadversary,

C—Concentrated material, but containssufficient radioactive isotopes torequire heavily shielded process-ing.

Highly dilutedlarge quantities

Not impossibleversary to steal and convert toweapons material, but difficultyprovides a substantial deterrent.

material, so thatmust be diverted.

for non-state ad-

D—As F, but substantial chemical and/ormechanical processing needed.Possible for non-state adversary toconvert to weapons material.

orAs F, except material required forcontinued operation of fuel cycle.

F—Material in concentrated form suitablefor straightforward conversion toweapons, with modest radioac-tivity. Easy for non-state adversaryto use as, or convert to, weaponsmaterial.

The relative value of the opportunities fordiversion as summarized in figure VII-3 de-pends on the intentions and capabilities of thediverters. Four general categories of prolifera-tors can be envisioned.

Nations Desiring a Major NuclearWeapons Force

A major nuclear force might be required byan industrialized or emerging country intenton becoming a world or regional power. Alarge and reliable supply of high-quality fissilematerial would be needed. Covert diversionfrom safeguarded facilities would probably beprecluded by these criteria and by the incom-patibility of this method with the goal of inter-national prestige. Some non-weapons states(such as Germany and Japan) are capable ofbuilding their own facilities with the dualpurpose of power and fissile material produc-tion. India is developing this capability, butfew others will if economic power is a require-ment (discussed in chapter X). Nations party

163

92-592 0- 77 -12

to the Non-Proliferation Treaty (NPT) or sub-ject to safeguards on imported reactors wouldhave to abrogate safeguard agreements afterthe necessary facilities were in place.

System characteristics that would beespecially important for this category ofproliferator are:

. a high-production rate of high-qualityfissile material;

. immunity to international embargos andsanctions; and

. minimum impact on the fuel cycle.

The specific paths which this type ofproliferator could follow to obtain thestrategic nuclear materials are at present:

(1)

(2)

Enrichment: A plant with more capacitythan needed for domestic LWRS couldbe built. The excess could be ra-tionalized as being for export if it werenecessary to keep the intentions secretduring construction. In fact, no LWRSare needed in countries (such as SouthAfrica and Australia) that could becomemajor uranium exporters and prefer tosupply enrichment services also. Theamount of highly enriched uraniumthat could be produced without impact-ing on the fuel cycle would depend onthe excess enrichment capacity. A largesupply of uranium, either domestic orfrom a secure source, would be neededto keep reactors and weapon programssupplied. The cost would primarily bethe loss of enrichment revenues fromthe previously exported low-enricheduranium. This would amount to ap-proximately $20,000 per kg highlyenriched uranium.

Reprocessed LWR Fuel: An entire LWRfuel cycle would probably be required toresist nuclear embargos. The output ofone reactor operated to optimize thequality of the plutonium would be suffi-cient for 30 to 40 weapons per year. Thefrequent shutdown would result in theloss of one-half to one-third of thepower output, which is a high penalty,but after several years a substantial ar-senal would be available and the reactor

(3)

could be returned to normal operation.The plutonium lost to the fuel cyclewould have to be replaced by enricheduraniums, but the cost would not behigh if the uranium is recycled.

Reprocessed CANDU Fuel: A reprocess-ing plant would have to be built, butthis could be done covertly prior to thesafeguards abrogation. If plutoniumoutput is maximized, about 40 to 60weapons could be derived from each600 MW(e) reactor. Feed would have tobe increased considerably, sinceuranium recycle would be less attractivethan for the LWR. Full-power produc-tion could be maintained. Even if thefeed is not increased, 20 to 30 weaponscould be produced annually. Access toheavy water would have to be main-tained in either case. About 10 metrictons are required per year for normaloperation, more if refueling is acceler-ated. A small, unsophisticated plantmight produce heavy water at aboutdouble the normal cost of about $130per kg. This cost increase could add$1,300,000 or more per year for thequantity required for the operation ofthe reactor.

Comparison. —The third route is clearlypreferred if heavy water is not a problem. Theplutonium production rate is high, andvulnerability to international restrictionsalmost nonexistent. The total cost of the fullCANDU cycle should be less, though the reac-tor is 10 percent more expensive, because aheavy water separation plant is cheaper thanan enrichment plant.

Future Developments.—The near-term futurereactors (HTGR and AGR) do not presentmarkedly different opportunities. The HTGRuses high-enriched fuel, which means that if anation has a full fuel-cycle capability it alsohas another direct route to weapons material.The fresh fuel itself would not be of interest, asthen the reactor would have to shut down.The HTGR breeds more fuel than the LWR,and recycle is a virtual requirement. TheHTGR has somewhat more potential fordiversion than the LWR, but probably less

164

Figure VII-3.Reactor Diversion Report Card

Fabrication Reactor, including Spent Fueland Transport Fuel Storage at Transport andof Fresh FUel the Reactor Storage

LWR No Re-processing A B

LWR, Reprocessing,No Pu Recycle A B

LWR, pu C (Onsite FreshRecycle c MOX)

B (Spent Fuel)

LWR, DenaturedU-Th

HWR (CANDU), NO

Reprocessing

Uranium gas cooledReactors (AGR)

HTGR

LMFBR and GCFR

LWBR

A A

A B

A B

D D (Fresh Fuel)C (Spent Fuel)

D D (Fresh Fuel)C (Spent Fuel)

D B

MSBR A A

See figure VII-2 for a summary discussion of diverson points m the LWR fuel cycle“Nonexistent

B

B

B

B

B

B

c

c

B

(A)*

Reprocessed StockpileFuel-Fabrication of Excess

Reprocessing (including transport)

(A)* (A)*

F A

F (if fuel not blended atF Repro. Plant)

C (if fuel blended atRepro. Plant)

D A

(A)* (A)*

F A

F c

F F

D (National diverter) D (National diverter)F (Non-state F (Non-state diverter)

diverter)

F (A)*

SNM

(A)*

F

(A)*

(A)*

(A)*

F

(A)*

F

(A)*

F

SOURCE: OTA

than the CANDU. The AGR appears less ap-propriate for proliferation than the LWR.Fresh fuel has a lower enrichment and thespent fuel contain relatively little plutonium.Recycle is not expected, even if the fuel isreprocessed.

Some of the more distant reactors presentmore difficult problems. The LMFBR and thesimilar gas cooled fast reactor (GCFR) willboth produce copious quantities of high-gradeplutonium, and both are relatively easy tomake independent of international inter-ference since the cycles are self-supporting ex-cept for a supply of depleted uranium. Thefuel-cycle impact of diversion is negligiblebecause of the excess of plutonium.

The LWBR is not attractive to this type ofproliferator since the entire production of U233

is required to continue operation. Theprebreeder cycle could be supported, but mostof the fissile material produced is U233 whichis diluted in U238. This cycle would probablybe considerably more expensive than the PWRcycle for weapons-material production. Themolten salt breeder reactor would be onlymarginally better in that the breeding ratio isslightly higher, thus producing an excess ofU233 which would be adequate for producing4-8 weapons per year.

A qualitative ranking of the resistance toproliferation for all these systems is shown infigure VII-4.

Nations Desiring a Small, Not NecessarilySophisticated, Nuclear Capability

In this case, covert diversion is a possibilitybut may not be a necessity. If the facilities arenot safeguarded, the important characteristicswould be as follows:

●

●

●

Immunity to international embargos andsanctions—this type of nation is lesslikely to have full-fuel cycle facilities.

Minimum impact on the fuel cycle-asubstantial power loss would be harderto absorb.

Initial cost—nuclear reactors are alreadyvery expensive, These nations may not be

●

●

If

able to afford a more sophisticated oneeven if it is more vulnerable to diversion.

Ease of conversion to weapons material—the lesser sophistication of this type ofnation makes major processing difficult.

Production rate and quality of fissilematerial—this is less important than forthe previous case. Little material isneeded and the yield of the weapon ismuch less important than the fact of itsexistence.

the facilities are safeguarded, a differentset of factors apply.

High rate of material flow—to makediversion less noticeable.

Many vulnerable points to makesafeguarding difficult.

Minimum impact on fuel cycle.

Initial cost.

Ease of conversion.

The enrichment option of the previous casewill be plausible only if techniques other thandiffusion become viable. Diffusion is simplytoo big and expensive for this type of nation.Covert diversion of low-enriched uraniumcould be improbable since the country mightnot have the capability of building a smalldedicated weapons-grade enrichment plant,even using low-enriched uranium as the feed.Therefore, part of the plant itself would prob-ably have to be modified to yield high-enriched uranium.

The LWR reprocessing route is particularlygood for the covert diverter because of thelarge number of vulnerable points. The impacton the fuel cycle need not be large, because thediverted plutonium can be replaced byslightly more enriched uranium or slightlyless power output. It would not be necessaryto possess an enrichment plant. If there is nocommercial reprocessing, however, spent fuelwould have to be diverted to a small dedicatedreprocessing plant. It would be difficult toevade safeguards for long, so this path is im-probable. The overt diverter would need a

166 -

Figure VII4.

Reactor Systems Resistance to Proliferation(Note that a high rank means the system is least susceptible to diversion.)

Reactorsystem

Light WaterReactor (enrichment)

Light WaterReactor (spent fuel)

Light Water Reactor(reprocessing & recycle)

CANDU

High TemperatureGas Reactor

Advanced GasReactor

Liquid Metal FastBreeder Reactor—

Gas Cooled FastReactor

Light WaterBreeder Reactor

Molten SaltBreeder Reactor

SmallForce1 .

(unsafeguardedAvailability Force facilities)

Present 5 6

Present 4 3

Present 6 5

Present 8 7

NearTerm 7 4

NearTerm 3 2

R&D(advanced) 9 9*

R&D 10 10*

R&D 1 1

R&D(presently inactive) 2 8*

SmallForce

(safeguardedfacilities) Option

7 1

1 4

8

2

5

2

6 6

3 3

9 9

10 10

4 7

5 8

Non-SateAdversaries

1

4

6

2

7

3

9

10

8

5

“May not bean Option for cost or technological reasons SOURCE: OTA

complete fuel cycle—including enrichment—in order to thwart embargos. This might be soexpensive as to be impractical.

The CANDU would be excellent for theovert diverter who would simply process thenormal spent fuel. The covert diverter wouldhave to smuggle out his own spent fuel. This isnot an impossible task, as up to 10 years ofspent fuel could be in the pool in the form ofthousands of bundles. Accounting for all ofthem will be a formidable, but not impossible,task.

Comparison. —The overt diverter will preferthe CANDU, again assuming access to heavywater. An LWR with Pu recycle would be bet-ter suited to the covert diverter because of thegreater number of vulnerable points. Thestatic nature of the source (spent-fuel rods) inthe CANDU or LWR without reprocessingtends to make eventual detection of covertdiversion quite probable.

Future Developments. —The AGR presents es-sentially the same opportunity as the LWR.Reprocessing is less important than for theLWR, and could be eliminated. If there isreprocessing but not recycle the overt diverterwould have a substantial stockpile at his dis-posal, just as there would be for the sameLWR cycle. The AGR’s lower enricheduranium would be slightly easier to procurethan that for the LWR in case of embargos,and there would be essentially no impact onthe power production since the recoveredplutonium is not being used. The HTGR withits high-enriched fresh fuel and requiredreprocessing presents more opportunities forboth overt and covert diverters. The fuel-cyclefacilities, however, are expensive and tech-nologically demanding and might never beavailable for export. This could eliminateovert diversion.

The R & D reactors could again enhance orlimit opportunities. The overt diverter wouldprefer the LMFBR and GCFR for the samereasons as would the major nation, but maynot be able to afford them. The LWBR wouldbe quite inappropriate. The MSBR might pro-duce sufficient strategic nuclear materials, butits intricate technology would be difficult tomanage. The MSBR concept, however, may be

readily adaptable to small sizes, which wouldmake it more attractive. The covert diverterwould also prefer fast breeders, possibly by awide margin, Thermal breeders provide fewopportunities.

Nations Desiring the Option of RapidDevelopment of Nuclear Weapons in theFuture Should That Appear Necessary

The important factors are:

. rapid access to strategic nuclear materials

. high production rate of fissile material● cost

The enrichment option is not particularlyinteresting because the process is too slow, ex-cept possibly for centrifuge designed for fastconversion to high enrichment.

The LWR reprocessing route is veryvulnerable, in that a significant stockpile ofplutonium can be legitimately maintained.This provides immediate access, and con-siderably more can be supplied from thebatches of spent fuel wai t ing to bereprocessed, Spent fuel with no commercialreprocessing would require a small reprocess-ing plant to be built ahead of time and held inreadiness. Even then, there would be nostockpile for immediate seizure,

The CANDU would be less appropriate forthis proliferate unless an appropriatereprocessing plant already exists for otherreactors.

The HTGR could be useful because freshfuel could be quickly processed even if noother fuel-cycle facilities were available. Thiswould mean loss of the reactor as it could notbe refueled, but national emergencies might beseen to override this factor. The fast breederswould provide substantial strategic nuclearmaterials both in the fresh fuel and inreprocessing plant stockpiles.

The LWBR would provide useful high-enriched fresh fuel as in the HTGR and easilyprocessed spent fuel if needed, The entire coreinventory could be made available quicklyand would provide a great many weapons.The MSBR contains 2300 kg U233 at all times

168

(enough for as many as 460 bombs). Thismaterial could be processed immediatelybecause its low-fission product inventoryeliminates the need for long-term cooling. Thesmall normal excess of U233 could also bestockpiled, and would provide immediate ac-cess to about eight weapons forstockpiling.

Non-State Adversaries

The prime requirements are:

. many vulnerable points

. high rate of material flow

. ease of conversion

The only present generationoffers a significant opportunity

every year’s

system thatis the LWR

with reprocessing. Plutonium recycle allowsthe reprocessing plant, plutonium shipments,mixed-oxide plant, and possibly even thefresh fuel to be targets for attack or diversion,

The AGR is as resistant as the LWR if noreprocessing takes place. The HTGR fresh fuelis a possibility, but considerable work must bedone to separate the high-enriched uraniumfrom the thorium. If the high-enricheduranium can be attacked before it is mixedwith thorium, the weapon preparation wouldbe easier.

The LWBR can be attacked at the fuelfabrication plant or at the reprocessing plant.The U233 is more easily separated from thethorium here than in the HTGR. The MSBR isalmost invisible to the non-state adversary.All operations are performed at the plant site,and only a small amount of U233 need be ex-posed. This could easily be denatured in U233

before shipping.

Figure VII-4 ranks these systems in order ofvulnerability to each of the diverters.

Research Reactors

There are many research reactors operatingthroughout the world, Appendix V in volume11 lists the research reactors outside the UnitedStates with a power rating of 1 MW(t) ormore. Examination of that list shows that

there are 18 countries which possess either (a)natural uranium or low-enriched uranium-fueled reactors that will have accumulated 10kg or more of Pu239 by 1984, or (b) reactorsfueled with 80 percent to 100 percent U235

with a power rating of 5 MW(t) or more (i.e.,an annual fueling requirement of 5 or more kgof 80 percent to 100 percent U235), or (c) bothof the above.

Examination of a list prepared by ERDAshows that, through December 31, 1976, theUnited States exported a total of 1,115 kg ofplutonium to 38 countries. Eight countrieshave received more than 5 kg of plutoniumfrom the United States (see list in volume II,appendix V). From January 1, 1968, throughDecember 31, 1976, the United States exportednearly 10,000 kg of uranium enriched to 20percent or more in U235 to 21 countries. Eightcountries have received substantial amountsof highly enriched uranium.

The exported plutonium is used largely incritical assemblies, that is, experimentalfacilities run at zero power. This plutonium isessentially uncontaminated by fission prod-ucts, and is of very high quality for use inweapons.

Thus, substantial diversion or theft poten-tial exists outside the commercial power in-dustry. India’s nuclear explosive was madewith plutonium produced in one of theresearch reactors mentioned above.

Alternate Fuel Cycles andNonproliferating Reactors

Present commercial and near-commercialfuel cycles have been conceived and developedwith essentially no thought given to their im-plications for prol i ferat ion or to thedifficulties of safeguarding them. Otherpossibilities exist, however, that are lessvulnerable to diversion.

Alternate Fuel Cycles

ERDA has recently set up a study in theOffice of Nuclear Energy Assessments, Divi-sion of Nuclear Research and Applications, to

169

investigate and evaluate alternative fuel cy-cles. The criteria for evaluation of the alternatecycles are: (a) proliferation risk potential, (b)safeguard potential, (c) technical feasibility,(d) economics and resource utilization, (e)commercial feasibility, and (f) introductiondate. In evaluating proliferation risk potential,emphasis will be placed on diversion or theftof nuclear material for the purpose of makingan explosive weapon. Both domestic andforeign applications will be considered.

The schedule calls for a final report in Octo-ber 1978, with a developed set of proliferationcriteria and an assessment of selected alternatefuel cycles. ERDA is requesting supplementalfunds of $4 million from Congress for FY 77,and has budgeted the program at $7 millionfor FY 78.

The program is currently in the phase ofcollecting proposals for alternate fuel cyclesand issuing some contracts for promising pro-posals already collected. Some work pre-viously contracted by ERDA has beenassembled under the aegis of this project. Ascreening for the most promising alternates isset for July 1977.

For the results of this program to be mostuseful, the alternates that are selected forfurther study ought to be balanced betweenrelatively short-term payoff on technicalmodifications of existing cycles and radicallynew approaches. The differences between na-t ional capabi l i t ies and non-nat ionalcapabilities should be kept in mind. An alter-nate such as coprocessing, for example, mightput a substantial obstacle in the way of a non-national group but provide much less of adeterrent to national proliferation.

A good deal of emphasis is apparentlybeing given to an effort to develop a quantita-tive methodology for evaluating proliferationpotential, The first phase of this criteria effortis due to be completed in June 1977. Such aneffort can be extremely useful in forcing thepeople involved to think through theproblems in detail. However, a set of numeri-cal criteria purporting to quantitatively evalu-ate proliferation risk should be regarded withskepticism.

The areas that the program is currentlylooking at can be grouped in the followingcategories:

(1) Reexamination of the LWR and theLWR fuel cycle

(2) Introduction of the CANDU into theUnited States

(3) Thorium fuel cycles(4) The fast-breeder fuel cycle

In the first category, reexamination of theLWR fuel cycle, a variety of concepts are beingconsidered, most of which are aimed at in-creasing the energy obtainable from LWR fuelwithout going to plutonium recycle.

Several possibilities exist for modifying thedesign of the LWR so that spent fuel will havea lower fissile content, approaching that of theCANDU, thus reducing resource-utilizationpressures for plutonium recycle. Preliminaryestimates indicate that a modified LWR couldextract an additional 20 percent of power outof a given amount of uranium, as compared toan additional 30 percent with plutonium recy-cle and present LWR design. Possible changesinclude opening up the lattice and decreasingperiods between refueling; increasing the ini-tial uranium enrichment; decreasing neutronabsorption in the coolant, moderator, andcontrol rods by either geometry changes ormaterial changes, including use of heavywater; or the use of uranium-metal fuel.(Some of the above design changes are incom-patible with others.) Many such design varia-tions have been considered in the past, whennonproliferation was not a consideration, andrejected because of technical or economicreasons.

An updated assessment of the use of metalfuel has recently been completed at ORNL.The study indicates that uranium enrichmentcould be reduced by up to one half (i.e. 1.5percent instead of 3 percent), and that thefissile content of the spent fuel would indeedbe very low, Metallurgical problems have inthe past precluded this option; however, re-cent development work is reported to look ex-tremely promising,

The adoption of a throwaway cycle (i.e., noreprocessing) would make the LWR cycle a

170

less-attractive target for diversion by nationsand a much less-attractive target for theft bynon-state adversaries. The central interna-tional issues would revolve around the dis-posal of the spent fuel, including questions oftransportation safety, disposal sites, long-termstorage security, and disposal costs. Pressuresto recycle might recur if uranium prices rosehigh enough.

Other LWR schemes under consideration(but apparently not funded as yet) includeseveral reprocessing variants. In one concept,only uranium would be recycled; plutoniumwould be either (a) partially decontaminatedand stored as highly radioactive plutoniumnitrate solution, or (b) purified and stored.Variant (a) would provide some deterrent forthe non-state adversary, but neither variantaddresses the question of national prolifera-tion. Indeed, variant (b) involves stockpilingplutonium. Plutonium stockpiles are the mostvulnerable target for the national diverter andrequire massive security against the non-stateadversary. Coprocessing is also on the list ofalternates, and also apparently as yet un-funded. This concept would recycle LWRspent fuel without separation of uranium andplutonium. Instead of pure PuO2 at the end ofthe reprocessing/conversion stream, therewould be approximately 1 percent PuO2 in 99percent U02. The economics are unclear. Fuelfabrication costs would increase because morefuel would contain toxic plutonium.Reprocessing costs would decrease. Claimshave been made for increased fuel utilization.

Coprocessing would present a substantialobstacle to the non-state diverter, Plutoniumwould never appear in highly purified form inthe fuel cycle. One thousand kg of mixed-ox-ide material would have to be stolen to obtain10 kg of plutonium. The separation of PuO2 insuch a dilute form would present a very time-consuming task to the non-state adversary.

Coprocessing, however, presents a muchless-significant hurdle to the overt nationaldiverter, A nation could keep a small PuO2-UO2 separation plant “on ice” until it madethe decision to go for a nuclear explosive; itwould then appropriate the mixed-oxidematerial, and separate it in a matter of days.

The implications of coprocessing for thecovert national diverter are less clear. In the casediscussed above, the equipment for separatinguranium and plutonium would be absent,forcing the nation to divert 1000 kg of mixed-oxide material for every contained 10 kg of Pu.This presents serious logistical problems,which, however, possibly could be surmount-able, even in a safeguarded plant. In a variantof the above process, where uranium andplutonium are only partially separated (togive about 5 percent PuO2 and 95 percentUO2, as the final product) covert national pro-duction of tens of kilograms of pure PU02, un-detected by the materials accountancy system,is credible given a commercial-sized plant.Whether or not this material could then beremoved from the plant without detectionwould depend on the efficiency of contain-ment and surveillance safeguards. (Seechapter VIII.)

ERDA is at present, actively looking at someof the problems of collocating reprocessingplants, fuel fabrication plants, and possiblyplutonium-burning reactors. Generic studiesof environmental effects and institutionalproblems are underway, as are technicalstudies of a possible nuclear energy center atHanford. Confining plutonium in fresh fuel toa small number of fixed sites has the potentialfor reducing the risk of non-state theft. Theresults of these studies will also be applicableto multinational fuel-cycle centers. (See “In-ternational Control of Proliferation”, chapterVIII.)

Another LWR option being investigated byERDA is the tandem fuel cycle. In this scheme,discharged LWR fuel is inserted into a heavywater reactor (HWR) to achieve an additional33 percent burnup. After discharge from theHWR, the fuel would be stored indefinitely.There are severe technical, economic, andlicensability questions to be resolved, as dis-cussed in appendix V of volume 11.

Other concepts to extend the use of spentfuel without recycle include bombardingspent fuel with neutrons from:

. a target bombarded byhigh-energy acceleratorbreeder idea);

protons from a(the accelerator

171

. controlled thermonuclear fusion (thespent fuel would be inserted into ablanket in the fusion reactor);

. laser or ion beams on fusion targets; and

. an LMFBR

Using fusion neutrons to produce fissilematerial (either plutonium or U233) could beeconomical before self-sustained fusion wasachieved, The accelerator-breeder and fu-sion/fission devices may well have an impor-tant role to play in extending resources offissile materials and possibly in cleaning upnuclear waste. Both devices might have ap-plicability in the international thorium cyclediscussed below.

However, such neutron irradiation schemescan clearly be regarded as antioroliferationmeasures only at an international center. A na-tion with a device designed to irradiate spentfuel could as easily irradiate clean uranium orthorium. A nation would ship highly radioac-tive spent fuel to an international irradiationcenter, where the plutonium or U233 contentof the spent fuel would be increased by a fac-tor of 2 or more. The spent fuel might thenhave to be refabricated. Finally, the stillradioactive spent fuel, with enhancedplutonium or U233 content, is shipped back tothe nation for reinsertion into the reactor.After one or more such round trips, the spentfuel is shipped back for disposal. Both thisconcept and the accelerator breeder are dis-cussed in appendix V of volume II. Many ofthe metallurgical problems discussed for thetandem fuel cycle would exist for these op-tions, Neutron irradiation of spent fuel ap-pears to be a somewhat contrived anti-proliferation measure.

The ERDA studies on introducing the CAN-DU into the United States appear to be focus-ing on economics, licensability in the UnitedStates and U.S. commercial feasibility. No cur-rent U.S. reactor vendors manufacture CAN-DUS, and presently there is little U.S. utilityinterest, The proliferation potential of CAN-DUS and LWRS was compared in the preced-ing section, where it was concluded thatsafeguarding against the national diverter wasa harder problem for the CANDU than theLWR without recycle. Thus, purely from a

172

nonproliferation point of view, LWR redesignappears more attractive.

Through another program, ERDA is in-vestigating the problems of commercializingthe HTGR. Because the HTGR contains ex-posed 93 percent U235 in its fuel cycle, it hasserious proliferation implications. However,the alternates program is investigating the useof less than 20 percent U235 in the HTGR cycle.The detailed assessment is underway but notyet completed. Earlier studies indicated that alow-enriched uranium cycle, possibly as lowas 6 percent U235, would be technically feasi-ble but at a distinct economic disadvantage tothe 93 percent U235 cycle. The low U235 HTGRwould have better fuel utilization than theLWR. Major redesign of the HTGR mightyield more favorable results.

The program is apparently planning an ex-tensive investigation into thorium fuel cycles,including work on the recently proposed in-ternational thorium cycle. In this system, na-tional reactors would operate on a fresh fuelmix of something like 1 part U233, 6 parts U238,and 10 to 60 parts thorium. As discussed inchapter IV, a U233/U238 ratio of 1:7 representsa lower limit of concentration, below whichthe mixture cannot be used in a practical fis-sion explosive. Preliminary calculations sug-gest that spent fuel from such a reactor wouldcontain only one-fifth to one-tenth as muchplutonium as spent fuel from present reactors,for the same amount of power output, Spentfuel would be sent to international fuel supportcenters which would reprocess the spent fuel,extracting the plutonium for burning onsite,possibly in fast breeders with thorium as thefertile element. The U233 produced in the fastbreeders would be denatured with U238 at thecenter, fabricated into fresh fuel, and shippedto the national reactors. Thus either enrich-ment (for fresh fuel) or reprocessing (forspent fuel) would be necessary to extractweapons material from fuel in national hands.Both routes are possible for the nationaldiverter, but both require the construction of adedicated facility.

The national diverter, if discovered, is veryvulnerable to fuel-supply cutoff in the inter-national thorium cycle. The internationalthorium cycle offers a very high degree of pro-tection against the non-state diverter.

A partial list of questions to ask about thedenatured thorium cycle includes:

1.

2.

3.

4.

5.

6.

7.

What is the concept for starting upthorium cycle reactors? Can startup fuelmaterial be generated without paying theeconomic penalty that appears to be re-quired for the LWBR?How does the rate of growth of nuclearpower affect the attractiveness of the cy-cle?The thorium concept requires reprocess-ing, whereas an optimized throw-awayLWR U-PU fuel cycle does not. What arethe relative safeguard, economic, anduranium utilization differences for eachof these concepts?How much redesign of LWR S ( a n dH W RS) is necessary to achieve an op-t imum thor ium fuel -managementprogram?What are problems and costs of produc-tion development of the thoriumreprocessing (Thorex) process?What is the increased safety/radiationrisk of a thorium fuel cycle duringa) normal operation?b) abnormal situation (e.g., sabotage at-tempt) ?How much development and explorationis required for a - large-scale supply ofreactor-grade thorium?

The project is also studying coprocessing offast-reactor fuel for either the U/Pu or theTh/U233 cycles. The emphasis would be onmetallic fuels for breeding in the core, ratherthan in the blanket. As pointed out before,coprocessing is a tactic of limited usefulnessagainst national proliferation.

Nonproliferating Reactors

One of the most intriguing concepts thatERDA is studying is being funded at $250,000for FY 77 by the Division of InternationalSecurity Affairs. This is the concept of non-proliferating reactors.

Through strict design requirements, this ap-proach attempts to eliminate the diversionpaths present in current and projected power-reactor systems and their associated fuel cy-cles. Several key design criteria are: (a) the

system shall contain only a small amount offissile material at any given time; (b) thereshall be no access to the fuel during thelifetime of the reactor; (c) any diversion offuel will cause the reactor to shut down; (d)the reactor shall be refueled by the addition offertile (i.e., non-fissile) material only; (e) thereactor shall not operate as a breeder, but as asustainer, producing just enough fissilematerial to keep itself running (i.e., the breed-ing ratio should be essentially one); (f)reprocessing shall be done onsite inside abiological shield.

In addition, the reactor is required to pro-duce economical power and be designed sothat accidents have minimum consequencesof fsite,

This last requirement suggests that it mightbe possible to site the reactors fairly close toload centers and use the waste heat locally,thereby markedly increasing overall efficien-cy. Finally, although the optimum power levelfor such reactors is not known, preliminarystudies suggest that the reactors may beeconomical on a small scale, i.e., 50 to 250MW(e).

Preliminary conceptual studies have beendone on three reactor systems.

. Gas core reactor ($100,000)

. Suspended particle bed reactor ($40,000)

. Modified molten salt reactor ($100,000assigned; $20,000 spent)

Conceptual and design studies on a gas corereactor have been carried out for a number ofyears at Los Alamos Scientific Laboratory(LASL) under NASA funding. Some experi-mental work has been done for NASA with azero power assembly. An experimental flow-ing gas system has started up at LASL re-cently, and has attained criticality.

The gas core reactor designed for the non-proliferating reactor study is a conservativevariant of the 6000oK plasma reactor beingdesigned for NASA use around the year 2000.This particular nonproliferating gas-coredesign has the following features: U233

F6

gaseous fuel; beryllium moderator andgraphite neutron reflector; molten thorium-salt breeding blanket; relatively low-operatingtemperature of approximately 1200 ‘K; power

173

level of 200 MW(t); entire plant-fissile inven-tory of 100 kg of U233 (i.e., this includes thematerial being reprocessed). Diversion of 4 kgof U233 will shut the reactor down, as wouldadding more thorium in an attempt to in-crease the breeding ratio.

Preliminary calculations indicate that if theoperating temperature of a nonproliferatinggas-core design is raised to the 4000oK range(i.e., the magneto-hydrodynamic range), thetotal inventory of U233 may decrease to a fewtens of kilograms. Moreover, the quantity ofU 233 that could be diverted without shuttingdown the reactor would probably alsodecrease markedly. This sensitivity of fissileinventory to operating temperature should beexplored more thoroughly.

Another aspect of the gas core design thatmerits further investigation is the possibilityof using denatured U233 fuel.

The suspended particle bed reactor featuresextremely small coated-fuel particles and agas-cooled, heavy water moderated, fluidized -bed design. Such high burnup is attained thatreprocessing is of no benefit. The reactor isrefueled online with fertile material only, buthas a high fissile inventory of 3000 kg of U233

for a 300 MW(e) system.

The molten salt reactor concepts are basedon the use of a circulating fluid fuel withonline continuous fuel reprocessing.

A detailed 300 MW(e) molten salt breederreactor design previously prepared foranother purpose was examined to determinethe feasibility of redesign for nonproliferationrequirements. Potential diversion paths were

identified and changes suggested which werequalitative in nature (there was insufficienttime to actually redesign the reactor).

The modified molten salt reactor as a non-proliferation reactor has many features whichmake it attractive. However, it appears thatthe system would have difficulty meeting therequirement for a breeding ratio of approx-imately one. It is not known how significantthe deviation from one would be. The systeminventory is high, on the order of 500 to 1000kg of U233, which at this time would be judgedexcessive. Finally, it is not clear that diversionof a significant quantity of U233 would causethe reactor to shut down.

For all the nonproliferating reactor designs,enough U233 to start the reactor up wouldhave to be supplied from an external source,probably a thorium cycle fast breeder. Onethorium cycle breeder could provide enoughstart-up U233 for many nonproliferating reac-tors. Start-up U233 would have to be pro-duced, reprocessed and shipped under guard,In this sense, nonproliferating reactors wouldnot totally eliminate diversion possibilities, butthe concept does hold forth the promise ofenormously limiting diversion and prolifera-tion paths.

Conclusion on Nonproliferating Reactors.—This small program is the first attempt todesign reactors specifically with nonprolifera-tion and nondiversion in mind. As such, itdeserves continued funding at an expandedscale, a wide hearing, a thorough assessment,and an open-minded comparison with otheralternatives.

DEDICATED FACILITIES

All nations now possessing nuclearweapons obtained fissile material fromfacilities specifically dedicated to its productionor separation. Therefore, a nation need notundertake a nuclear power program in orderto have a nuclear weapons program. In fact, anation determined tomay be able to do soin a shorter period

174

acquire nuclear weaponswith lower capital costs,of time, and with less

scrutiny from other nations by buildingfacilities specifically dedicated to the produc-tion of fissile material by itself (or with graymarket aid).1

Isee “purchase and Theft” section, this chapter andappendix VII of volume 11 for a discussion of black andgray markets.

Such a nation would have two basic op-tions:

(1)

(2)

Construct a plutonium-productionreactor plus a reprocessing plant to sep-arate the plutonium from the spent fuel;

Construct an enrichment plant to pro-duce highly-enriched uranium {remnatural uranium.

Variants on the above two options arepossible. For example, a nation might feed adedicated reprocessing plant with spent fuelobtained from an unsafeguarded power orresearch reactor. This is the route India took,removing fuel from the unsafeguarded Cana-dian-supplied Cirrus research reactor. Alter-natively, a nation might divert low-enricheduranium from a safeguarded facility or buylow-enriched uranium in a black or graymarket and boost it to highly enricheduranium in a dedicated enrichment plant. Nocase of the diversion or purchase-plus-boost-ing route is known to have occurred.

A major motivation for nations to builddedicated facilities is to have a reliable, possi-bly secret, and/or legal source of fissilematerial. As safeguards are improved and ex-tended over all imported nuclear facilities, andas greater restraints are placed on the sale ofenrichment and reprocessing plants, more na-tions may be inclined to develop their ownfacilities.

The construction of any facility dedicated tothe production of weapons material, which ofcourse is not safeguarded, would constitute aviolation of the NPT by parties to that treaty.The NPT nation must accept IAEA safeguardson all its peaceful nuclear materials, in all itspeaceful nuclear facilities, and must requireIAEA safeguards on its nuclear exports to allnon-nuclear weapons states. However,nothing in the NPT prohibits the transfer ofnuclear material or technology to nonpartiesto the NPT, even though such nations mayhave some unsafeguarded facilities. At thepresent time, the non-NPT nation, even whilereceiving safeguarded imports from NPT par-ties, may still indigenously build or obtainunsafeguarded nuclear facilities from anothernonparty to the NPT.

In spite of the above fact, even countries notparty to the NPT would usually have strongincentives to attempt to keep construction andoperation of dedicated facilities secret, at leastuntil they had built up a stockpile of weaponsmaterial. A nation that can suddenlydemonstrate the capability to explode anuclear device has a strengthened position. Atthe same time, a clandestine weaponsprogram avoids the recriminations and inter-national political pressures that the nationmight encounter if it pursued the programopenly.

Under some conditions, a nation might feelit had little to lose and perhaps some politicalprestige to gain by the open pursuit of anuclear weapons option. This section will thusinclude consideration of dedicated facilitiesthat would be difficult to keep secret.

Weapons Program Levels

The magnitude of the weapons program anation decides to undertake is a crucial factorin determining what kind of dedicated facilityit will choose to build.

A country interested in only a smallweapons program would look first at option(a), the plutonium production reactor. Asshown in appendix VI of volume II, the rate ofplutonium production is proportional to thereactor-power level. For example, a reactoroperating at 25 MW will produce between 9and 10 kg of plutonium per year, enough forone or two explosives. As outlined below,such a reactor can be built and operated atnominal cost, in a relatively short time, with asmall number of personnel, and there is atleast a fair chance that its existence could beconcealed for several years. This size will bereferred to as a Level I reactor.

A more ambitious program, one whichwould yield between 10 and 20 explosives peryear, would require a reactor operating atabout 400 MW. This is referred to as a Level IIreactor. Its construction would require a largeinvestment in capital and involve a largenumber of engineers and constructionworkers, Because of the magnitude of the task,

175

there is little chance that the project could bekept secret, either during construction or inoperation.

An alternative to a single Level II reactormight be the construction of several Level Ireactors that together would yield the sameplutonium output as the larger reactor. A na-tion with a limited technological base mightfind it easier to build several smaller reactors,each based on the experience gained with thefirst.

If a nation decided to build an enrichmentplant to feed its nuclear weapons program, itwould have to allow for 15 to 30 kg of highlyenriched uranium for one explosive. The mostlikely choice of enrichment technique at pres-ent (as discussed below) is the gas centrifuge.Because construction of an enrichment dedi-cated facility would be more expensive anddifficult than a Level I reactor it is unlikely tobe considered by a nation that wants only oneor two weapons per year. One exceptionmight be a nation that has either developed orpurchased a centrifuge enrichment plant for acommercial power program. In that case, thecomponents for a dedicated enrichment plantmight cost no more than add-ens to the exist-ing plant, The cost for a small dedicatedenrichment plant would then be low enoughfor a Level I weapons program. (See alsochapter VII “Diversion From CommercialPower Systems” for a discussion of thisroute.) Another important exception in thefuture might result if other enrichment tech-niques are found that are cheaper and tech-nologically simpler.

Assessment of the likelihood of a nationbuilding any of these dedicated facilities, andof the probability that its efforts can bedetected, requires an evaluation of the cost,time, and personnel required,

The numbers vary widely with the types ofassumptions made. If one assumes that thededicated facility will be essentially a scaled-down commercial facility, the cost, time, andpersonnel estimates are generally quite high.One might more realistically assume that adesigner would make considerable simplifica-tions if the facility were built specifically to

produce nuclear weapons material. In particu-lar, such plants can be subject to less stringentsafety and radiation-protection restrictions.

The estimates of cost, time, and personnelwill also depend quite heavily on the particu-lar nation building the facility. Important fac-tors are the available natural resources, thetechnological and industrial base, the numberof trained scientists and engineers, and thecost of labor.

Level I Plutonium ProductionReactor2

The most likely choice for a Level I produc-tion reactor would be one fueled with naturaluranium, moderated with graphite, andcooled by air. The uranium might either bemined and milled indigenously, since manynations have at least small uranium reserves(see appendix VI of volume II), or it might bepurchased on a gray or black market if com-mercial purchases would raise suspicions.Graphite and heavy water are the only practi-cal moderators to use with natural uranium.The heavy water is an improbable choicebecause it is expensive, available from only afew countries, and indicative of its purpose ifimported in large quantities. Air is selected asa coolant rather than water because itsimplifies the design, construction, and main-tenance of the reactor and the fabrication ofthe fuel elements.

One graphite-moderated, air-cooled,natural-uranium reactor that has operatedsuccessfully is now fully described in openliterature. It might well serve as a model reac-tor to guide the construction of a dedicatedfacility. This reactor is the Brookhavengraphite research reactor (BGRR), describedin appendix VI of volume II. The BGRR is a 30MW reactor which, when operated withnatural uranium (from 1948 to 1957) forresearch purposes produced about 9 kg of

Wluch of this section originally appeared in: John R.Lamarsh, “On the Construction of the Plutonium Pro-ducing Reactors by Smaller and/or Developing Na-tions,” Prepared by CRS, April 30, 1976. See also appen-dix VI of volume II.

176

nearly pure Pu239 annually (enough for one ortwo weapons per year). The cost of the BGRRand its related equipment was $16.7 millionwhen built in 1948. It is not necessary toduplicate the BGRR in detail in order to attainthe same rate of plutonium production.Simplifications in the BGRR design wouldpermit the building of a plutonium-produc-tion reactor that would be cheap and reliable,and that would require the talents of only asmall group of conventionally trainedengineers.

The design of a simplified BGRR is dis-cussed in detail in appendix VI of volume IIwith cost estimates for the various compo-nents. Costs are based on current U.S. prices,and as such they may have only the roughestapplicability to another nation. Moreover, thecosts in appendix VI refer to a bare-bonesprogram, with primitive conversion and fuel-fabrication facilities and perhaps somesacrifice of safety and environmental controls.The overall reactor cost estimated with theseassumptions is $10 million. Other estimateshave been made for a Level I reactor of thesame basic type which are considerably high-er.

A conservative estimate for the capital costof a Level I reactor of modified BGRR designproducing 9 kg of Pu239 per year, is, therefore,in the range of $15 million to $30 million.

The personnel requirements for the designand construction of the facility are modest, asall of the essential design parameters are inopen literature. High-level research anddevelopment personnel are not required. Onlya handful of experienced and competentprofessional engineers—possibly no morethan 10—would suffice to design and overseethe construction of the facility.

The reactor could be ready for productionapproximately 3 years from the beginning ofthe project.

Level I Reprocessing Plant3

To fabricate nuclear explosives as quickly aspossible, the fuel from a dedicated Level I pro-duction reactor would be removed after it hadbeen in the reactor for approximately 1 year.The concentration of plutonium would thenbe about 9 kg in 75 tons of fuel, or about 120grams per ton. The nation would have to builda reprocessing plant to separate the plutoniumfrom the spent fuel.

A plutonium recovery plant must bedesigned and operated with care. The rawfuel, when first discharged from the reactor, ishighly radioactive. Even if the fuel is allowedto cool for 120 days, during which time the ac-tivity decays by a factor of 100 or more, thetotal radioactivity is still about 45,000 curiesper ton or 0.05 curies per gram of fuel. Thismeans that the chemical processing of the fuelmust be carried out remotely, in a shieldedcell, at least up to the point where the fissionproducts are removed.

It should be noted, however, that theradioactivity of the BGRR fuel is much lowerthan that of a typical power reactor. The ac-tivity of power-reactor fuel after a cooling-offperiod of 120 days runs between 2 and 3million curies per ton, a factor of about 50times higher than BGRR fuel. Considerablymore precautions must therefore be taken inreprocessing power-reactor fuel than fuelfrom a BGRR.

Although the chemical steps required in theprocess are straight-forward and well-known,design and operation of the plant is compli-cated by the radioactivity of the spent fuel, thetoxicity of plutonium, and the potential criti-cality of the plutonium fuel. These problems

Slvfuch of this section originally appeared in: John R.Lamarsh, “On the Construction of the Plutonium Pro-ducing Reactors by Smaller and/or Developing Na-tions,” Prepared by CRS, April 30, 1976. See also appen-dix VI of volume II.

177

require remote control, concrete shielding,and careful procedures, but do not constitutemajor obstacles,

Virtually all reprocessing plants built sincethe 1950’s use the Purex solvent-extractionmethod. Both the chemical engineering tech-niques and the designs of actual reprocessingplants are well documented in open literature.For example, the plans for the Barnwell, S. C.,reprocessing plant recently constructed byAllied General Nuclear Services (AGNS) havebeen widely distributed to the public and areavailable in the NRC Public Document Room.Because AGNS is such a large plant, with athrough capacity of 5 tons of fuel per day(1,500 tons of fuel per year), considerable scal-ing down of this plant would be necessary forthe purpose of reprocessing fuel from a Level Ireactor.

Plans and specifications for a smaller plantare also available. In the late 1950’s, thePhillips Petroleum Company undertook afeasibility study of a small reprocessing plantdesigned to handle spent fuel from Common-wealth Edison’s Dresden-1 plant, thenscheduled for operation in 1960. Phillipsissued a report on this study in 1961, contain-ing detailed drawings of every component ofthis plant. Although some chemical/nuclearengineers have expressed skepticism about theworkability of the Phillips plant, because of itscompact design and high level of automation,it nevertheless can be viewed as an excellentstarting point for the design of a clandestinereprocessing facility in a small and/ordeveloping nation.

A number of simplifications, described inappendix VI of volume II, are possible whenthe plant is designed for the sole purpose ofrecovering plutonium from BGRR fuel.Several of these simplifications result becausethe fuel has a lower burnup than fuel from apower reactor as discussed above, and, lessshielding and fewer precautions are necessarywhen reprocessing the production-reactorfuel,

All of the equipment and supplies requiredto build and operate a plutonium recoveryplant are generally available on the worldmarkets. There is no single item that is so ex-

otic as to be obtainable from only a singlesource,

Estimating the cost of a reprocessing plantis difficult even for commercial operations.The discussion in appendix VI of volume 11 ar-rives at a figure of less than $25 million, whileestimates from other sources range from a fewmillion, to $10 million, to $70 million. Thehighest cost estimates do not appear to takeinto account the simplifications possible withrelaxation of safety and radiation protectionstandards and the use of lower burnup fuel.The lowest cost estimates correspond to ex-tremely crude and imperfectly shielded (buttechnically feasible) solvent extraction or ion-exchange facilities, not suitable for a sustainedprogram but which might be constructed toobtain material for a total of only a few ex-plosives.

In view of this range of assumptions andcosts, a reasonable estimate for the cost of afrugally designed reprocessing plant forBGRR fuel , based on the Purex sol -vent-extraction process, is less than $25million. If the plant were built to handle high-er burnup fuel (for example, spent fueldiverted from a power reactor), the costswould be somewhat higher.

Thus the total capital costs of a Level I reac-tor and associated reprocessing plant are inthe range of several tens of millions of dollars.

Many of the same technical personnel in-volved in the reactor project could be utilizedfor the plutonium recovery plant. Such a planmakes good sense because the recovery plantwould necessarily be located adjacent to thereactor and would probably be built duringthe same time frame. The total engineeringpersonnel for the two projects would be in therange of 10 to 20. Top-ranking research anddevelopment personnel are not required, asthe staff largely follow and/or modifyestablished designs. Nevertheless, the staffmust contain competent engineers with ap-plicable practical experience. A reactor andreprocessing plant cannot be built by readingbooks alone.

Many developing countries with a modesttechnical infrastructure would have thecapability to build and operate the Level I

178

reactor and reprocessing plant describedabove, The construction of Level II reactors(producing 100 kg of plutonium/year), dis-cussed below, would not be feasible for coun-tries without a fairly high level of in-dustrialization and a considerable nuclearbase upon which to build.

Level II Plutonium ProductionReactor

It is reasonable to assume that any dedi-cated plutonium production reactor would befueled with natural uranium, because iffacilities for enriching uranium were availableit would be more logical to base a weaponsprogram entirely on enriched uranium ratherthan reactor-produced plutonium.

In order to produce 100 kg of Pu239 per year(enough for 10 to 20 nuclear explosives), areactor operating at about 400 MW is neces-sary (a reasonable allowance of 30 percentdowntime is made).

Several different choices of moderator andcoolant are possible. The moderator for anatural uranium-fueled reactor can be onlyheavy water or graphite. The coolant can beeither ordinary or heavy water, or any one ofa number of gases. As discussed in appendixVI of volume II, the most practical choicewould probably be a graphite moderated,light-water cooled reactor.

Such a reactor would be similar to the firstreactors built at Hanford, Wash., in theManhattan Project. While a nominal 400 MWLevel 11 reactor wouId operate at only aboutone-fifth the power of an early Hanford reac-tor, the nuclear designs of the two systemswould be very similar. (The designs of theHanford reactors have recently beendeclassified.)

One estimate of the total capital costs of aLevel 11 reactor with associate reprocessingplant is in the range of $175 million to $350million. Roughly 50 to 75 engineers would beneeded in the design and construction phaseof this Level II program, supported by roughly150 to 200 skilled technicians. The length oftime required from the start of the design tothe first output of plutonium metal would be 5

to 7 years. As in the Level I reactor, the outputwould be nearly pure Pu239.

Level I and Level 11 EnrichmentPlants

Several methods might be considered forenriching uranium. To date the most suc-cessful method is the gaseous diffusion proc-ess, which was developed by the ManhattanProject in World War 11. This technique has re-mained essentially the only source of enricheduranium for military and civilian nuclearprograms since that time, both in the UnitedStates and abroad. However, gaseous diffu-sion plants are inherently large structures thatutilize a relatively sophisticated technology,much of which remains classified; they re-quire an enormous investment of capital; andthey consume large amounts of electric power.Finally, they cannot be concealed. Thegaseous-diffusion route to nuclear explosivesis not feasible for any but a handful of thelargest and most developed countries, and willnot be considered further in this report,

Another method for enriching uranium isthe Becker nozzle process. Such an enrichmentfacility is being sold to Brazil by Germany,and a variation of it is being developed in theUnion of South Africa. However, this methodrequires a large number of stages (see discus-sion of stages in appendix VI of volume 11)and consumes 2–1/2 times as much electricpower as gaseous diffusion and about 30 timesas much as centrifuges (see below). Althoughthe Becker method has fewer classified criticalaspects, it does not appear to be a reasonablechoice for any but an advanced nation.

Separation by means of high-speedcentrifuges was explored during the Manhat-tan Project but later abandoned. This tech-nique has reemerged in the last few years andhas reached an advanced stage of develop-ment, both in this country and abroad. It ap-pears likely that the centrifuge method ofenrichment will prove to be cheaper than anyother presently developed method of enrich-ing uranium,

An Anglo-German-Dutchgroup, Urenco, has successfully

enrichmentdemonstrated

179

92-592 0-77 . 13

the first cascades of two small centrifugeplants, each with a planned capacity of about200,000 kg separative work units (SWU) peryear at Capenhurst, England, and Almelo,Holland. Urenco has plans to expand one orboth plants to a total enrichment capacity of 2million kg SWU by 1982. A small test facilityis in operation at Oak Ridge, Term. OneAmerican firm has proposed building a majorcentrifuge uranium-enrichment plant to pro-vide fuel for nuclear powerplants.

One advantage of the centrifuge method fora dedicated facility is that a small number ofunits or groups of centrifuges can be placed inoperation as soon as they are built and tested.The separative operation need not wait uponthe completion of a large facility. Productionof weapons-grade uranium can begin at asmall level and gradually be increased as addi-tional centrifuges are installed.

The capacity of an enrichment plant neces-sary to produce 30 kg of highly enricheduranium (enough for one or two explosives) isshown in appendix VI of volume II to be be-tween 6000 and 7000 kg SWU/year, dependingon the tails assay. If each centrifuge has acapacity of 5 kg SWU/year this size plantwould require 1200 centrifuges. An enrich-ment plant for a Level II weapons programwould have to be about 10 times this size,with a capacity of 60,000 kg SWU/year.

The costs of a Level I or Level II centrifugeplant can only be based on estimates made bythose now planning commercial plants. Thosefigures are not only estimates themselves, butmost are for plants considerably larger than adedicated enrichment plant would be andcosts do not scale linearly with size. Urenco,which plans a plant whose capacity is severalmillion kg SWU/year, (i.e., hundreds of thou-sands of machines), has estimated its capitalcosts at $165/SWU. A U.S. estimate of capitalcosts for a 3 million kg SWU/year plant is$300/SWU. Another U.S. estimate for asmaller (300,000 SW U/year) plant is$700/SWU. Finally, Japan expects the cost of a50,000 kg SWU/year plant to be $3,300 /SWU.

The only one of these estimates to corre-spond closely to the size of a Level IIcentrifuge plant is the Japanese estimate. On

this basis, one might put the cost per SWU at$2,000-$4,000 and the total plant capital costat $120-$240 million. Because, as discussedearlier a Level I centrifuge-enrichment plant islikely to be built only as an “add-on” to an ex-isting plant, its costs may run the same perSWU as those of a larger plant. Taking therange of U.S. estimates of capital costs of$300 -$700/SWU, this assumption leads to acost estimate for a Level I add-on plant of be-tween $2 million and $5 million.

The costs discussed above do not includethose for research and development.Centrifuge separation is a difficult technologyonly recently developed by a few of the mostadvanced nations. The AEC classifiedcentrifuge technology in 1960, and Urencoalso maintains tight security. Althoughunclassified details of early centrifuge tech-nology are available, considerable develop-ment work would be necessary before even asmall operable enrichment plant could bebuilt.

Comparison

The centrifuge enrichment route calls forquite different resources and capabilities thandoes plutonium production reactors. In thelatter case not only are complete facility plansreadily available, but nuclear reactor andchemical engineers are being trained openlyaround the world.

For these reasons it is improbable thatcentrifuge enrichment would be the routetaken by a country with a limited industrialand scientific base interested in a Level Ifacility.

There do not appear to be major differencesin personnel requirements between the twotypes of Level II facilities—plutonium produc-tion and centrifuge enrichment-although thecentrifuge program might require somewhatmore manpower, The centrifuge programmight also take longer from inception tometallic-weapons material. The capital andoperating costs appear comparable.

Thus, an industrialized country desirous ofproducing significantly more than one bombper year might carefully weigh the centrifuge

180

enrichment plant against a large plutoniumreactor.

Advanced Isotope SeparationTechniques

Several enrichment processes are underdevelopment that may allow highly enricheduranium to be produced from naturaluranium (or even depleted uranium) in a verysmall number of stages. Two of the processes,laser isotope separation (LIS) and the ion-cyclotron resonance process (the Dawsonprocess), are under development on contractto ERDA. There are two variants of the LISprocess. One, the atomic LIS process, is underdevelopment at Lawrence LivermoreLaboratory (LLL). The other, the molecularprocess, is under development at Los AlamosScientific Laboratory (LASL). The atomicprocess is also being developed by a privateU.S. firm, Jersey Nuclear AVCO Isotopes(JNAI), a subsidiary of Exxon Nuclear andAvco Corporation. Research applicable to LISis also being conducted in a number of othercountries, notably the U. S. S. R., France, andWest Germany.

A third process, an advanced form ofelectromagnetic separation, is under concep-tual investigation by a private U.S. firm,Phrasor Technology, Inc., and research maybe underway in at least one other country. It isunclear how much actual laboratory researchand development has been done.

The three processes, LIS, advancedelectromagnetic, and Dawson, share severalkey features. All promise to extend uraniumresources, because low-tails assay should beeasily achievable. The present gaseous diffu-sion facilities produce tails of 0.2 percent to 0.3percent U235, and operation at lower tailsassay would be very expensive.

The advanced processes project a tails assayof 0.05 percent U235 or less, and an economicalextension of uranium resources of about 30percent could therefore be achieved fromlower tails assay. In addition, tails accumu-lated over the years from the gaseous-diffu-sion process could be run through an ad-vanced process to extract residual U235. ERDA

has estimated that by 1989, at an average of0.25 percent UZSS in accumulated tails, enoughextractable UZSS will be contained in the tailsfor the lifetime fueling of 40 to 50 reactors,each of 1000 MW(e).

The three processes also hold forth a prom-ise of lower cost enrichment. The goal of theERDA program is a 50 percent to 75 percentreduction in enrichment costs, but muchgreater cost reduction may also be possible. Ifthese approaches are economical on the largescale, all would be also economical in small-scale plants, in marked contrast to centrifugeprocesses and especially to gaseous diffusionprocesses. The reason for this is that the ad-vanced technologies will probably requirevery few stages (possibly only one) to go fromnatural uranium to low-enriched uranium forreactors. The gaseous-diffusion process re-quires over a thousand stages; the centrifugeprocess requires the order of ten stages, withmany centrifuges per stage.

The LIS processes and the Dawson processare still in the research stage, with solutions toseveral difficult problems still to be demon-strated. The proprietors of the advancedelectromagnetic process claim that they areready to begin pilot plant development, butthey have apparently done little laboratorydevelopment. (It should be noted that a ver-sion of the electromagnetic process, thecalutron, was used during the ManhattanProject to separate U235 for the first uraniumweapon. The calutron method is described inappendix VI of volume II.)

The EXXON LIS process, although closer tothe pilot-plant stage than the correspondingERDA process (perhaps partially because ofits less ambitious cost-reduction goals) alsohas technical problems to solve.

All three processes have built on a hightechnology base. LIS development in theUnited States depends heavily on theelectro-optical base developed by the Depart-ment of Defense. The electromagnetic processhas apparently built upon ion propulsionresearch in the space program.

All three processes have the potential forexacerbating the nuclear proliferation

181

problem. This is true in general of all enrich-ment processes which could produce highlyenriched uranium from natural uranium in afew steps, because such processes are highlyeconomical on a small scale once research anddevelopment have been completed.

This report has looked more closely at laserisotope separation (LIS) than the other twoprocesses, and has had access to classifiedmaterial, including ERDA-prepared responsesto a series of questions and a classified discus-sion meeting with representatives from LASLand LLL. In order to keep this documentunclassified, much of the detailed materialsupplied by ERDA has been omitted. As a con-sequence, the detailed state-of-the-art anddescription and evaluation of remaining tech-nical problems are not presented.

It appears unlikely, based on knowledge ofU.S. technology, that LIS could contribute tonuclear proliferation before the 1990’s. ERDAplans to reach a decision in 1979 on which ofthe approaches, atomic LIS, molecular LIS, orDawson, to fund to the pilot-plant stage. Pilot-plant operation is scheduled for 1984. Thisschedule depends on the successful solution ofa number of difficult technical problems.

Proliferation From Advanced IsotopicSeparation Techniques

Like any other enrichment technology, LIScould theoretically contribute to proliferationin the following ways:

1. The indigenous development of a dedi-cated facility;

2. Misuse of a commercial facility;

(a) Replication for the purpose of pro-ducing weapons material,

(b) Covert diversion, and(c) Seizure.

These routes are considered in turn below.

Once LIS is known to work on the pilot-plant scale, research and development can beexpected to intensify in several technically ad-vanced countries. Some of these countrieswould probably develop LIS 5 to 10 years aftera U.S. demonstration. Countries with only amoderate technological base would takelonger.

The above discussion presupposed that LIStechnology remained tightly and effectivelyclassified. Leaks of essential data or technicaldetails would speed-up development of LIS byother countries by eliminating the need forsome basic research. However, the design,construction, and operation of a workable LISsystem (even one that was not commerciallycompetitive) from source preparation toisotope extraction would still require a leng-thy and expensive development and learningprogram.

For these reasons, indigenous developmentof an LIS dedicated facility to produce highlyenriched uranium is unlikely to be a feasibleroute for nations with a low c r moderate tech-nological base.

A greater danger is that LIS technology willbe marketed by one or several advanced coun-tries. France and the U.S.S.R, in particularcould well succeed in LIS technology at aboutthe same time as the United States (again, itshould be noted that the eventual success ofLIS is not a certainty). As noted above, severalother countries would probably be only 5 or10 years behind. Because LIS is economical ona small scale, many countries with a smallnuclear power program could make a goodeconomic case for wanting an LIS enrichmentplant.

The spread, through sale, of commercial LIStechnology would teach many purchasing na-tions a technology that they probably couldnot have developed for themselves. Replica-tion of the technology in a small facility toproduce weapons material would not be easy,but would be possible for more nations thanindigenous development, The sale of commer-cial LIS technology could also result in manynations possessing a declared and safeguardedfacility that could be modified, covertly orovertly, to produce weapons material.

It would be the aim of safeguards to detectcovert production of weapons material in acommercial LIS facility. It is not possible toassess a nonexistent safeguards system on anonexistent plant containing a nonexistentprocess. However, several general statementscan be made. The most important obstacle toeffective safeguarding of a LIS plant againstcovert diversion could turn out to be the

182

obstacle that presently might hampersafeguarding of centrifuge enrichmentfacilities: the fact that inspectors do not haveaccess to the area where the actual enrichmentprocess is going on, but must rely on monitor-ing inputs and outputs at the perimeter of thefacility, with some input and output routes ex-empt from monitoring (i. e., perimetermonitoring with undeclared paths. See chap-ters VIII “Safeguards” and VII “DiversionFrom Commercial Power Systems”.) On theother hand, the intrinsic nature of the LISprocess, with relatively small pieces of equip-ment and a low-process inventory, couldmake LIS plants easier to safeguard againstcovert diversion than present enrichmentfacilities. In addition, many LIS plants wouldbe small, and small plants are intrinsicallyeasier to safeguard than large plants becausethe uncertainties in materials accountancy aresmaller in absolute terms of kg of enricheduranium. Therefore, LIS plants may not pre-sent uniquely difficult safeguarding problems.

A greater danger than covert diversion isovert divers ion, which internationalsafeguards, by their nature, cannot prevent.Some form of sanctions would be the onlyeffective response to overt diversion. A nationwith an enrichment facility is in a strong posi-tion to withstand international embargosaimed at LWR fuel, and LIS facilities couldprovide this immunity to countries that couldnot consider present enrichment technologies.(See chapters 111 and VIII for a discussion ofsanctions. )

The difficulty of modifying a commercialLIS plant designed for 3 percent U235 reactorfuel to produce highly enriched U 235 f o rweapons would depend on the engineeringdetails of the process. (It should be noted thatone need not go to 90 percent enrichment tohave useful weapons material: anythingabove about 50 percent U235 would be useful.)There do not appear to be any basic physicsreasons to preclude obtaining weapons-gradematerial in a few stages in either the atomic ormolecular LIS processes. Jersey Nuclear AvcoIsotopes (JNAI) has stated that their processappears to be unsuitable for the production ofhighly enriched uranium. Representatives ofthe Lawrence Livermore Laboratory (LLL) LIS

group have stated that they do not agree withthe JNAI statement, if it is meant to apply toall possible atomic vapor processes, although,LLL continues, it could be true for the particu-lar JNAI design. The concept of a “tamper-resistant” LIS process, atomic or molecular, isan attractive idea, but a good deal of tech-nological analysis would be necessary toestablish how tamper resistant any particulardesign was. Moreover, too much relianceshould not be placed on tamper-resistant LISdesigns. Even a very tamper-resistant designwould not be an absolute fix; what it woulddo is drive the nation towards the route ofreplication with modifications (a research anddevelopment program might be necessary toaccomplish this) rather than overt seizure.

Some observers have suggested a U.S.moratorium on LIS development, coupledwith strenuous U.S. diplomatic effort to ob-tain agreement from other countries to sus-pend work on LIS. Others express great doubtthat the United States could achieve interna-tional agreement to stop the development ofLIS or other advanced enrichment tech-nologies, in view of both the pressures inmany countries for independent and inexpen-sive enrichment and the worldwide marketfor enrichment services expected to develop inthe 1990’s.

ERDA predicts the worldwide market forenrichment to reach about 130 million SWUper year in the year 2000, based on their pro-jections of 1200 GW(e) for LWRS worldwideby the year 2000. These projections may proveto be too high, nevertheless present andplanned U.S. and foreign enrichment standsnow at about 60 million SWU per year, all of itthe expensive diffusion or centrifuge proc-esses (see figure X-18). The advanced enrich-ment technologies, promising much less ex-pensive enrichment, are thus extremely attrac-tive to countries wanting both to assure them-selves of self-sufficiency at a low cost in meet-ing their own enrichment needs and to profitfrom the sale of enrichment services.

Some observers have argued that theUnited States should develop an advancedenrichment technology and guarantee to sellenrichment services for a low fee or at cost. If

183

this were done, they maintain, the profit in-centive for other countries to develop suchtechnologies would be removed, and the in-centive for smaller countries to buy an ad-vanced enrichment facility would be muchreduced. Thus, these observers argue, U.S.development of these technologies would infact slow down their spread.

It would be unrealistic to expect, if this hap-pened, that no other countries would developadvanced enrichment technology. A few ad-vanced countries, with large nuclear programsand an avowed interest in LIS or another ad-vanced enrichment technology (notablyFrance and the U.S.S.R.), would almost cer-tainly prefer their own low-cost enrichmentfacilities, even at the cost of indigenousdevelopment, to reliance on U.S. guarantees.

The same argument of desire for independ-ence could be used by countries seeking topurchase an advanced enrichment facility,even if guaranteed services were availablefrom the United States and perhaps a fewother suppliers. Whether the independenceargument will be plausible, or will be per-ceived as only a mask for an unstatedweapons objective, would depend strongly onhow supplier-importer relationships developover the next decade.

In summary, the sale of LIS and other ad-vanced enrichment technologies presents agreater proliferation danger than indigenousdevelopment of the technologies. The presentcourse of formulating suppliers’ agreementsto end the sale of enrichment facilities istherefore particularly crucial in the case of theadvanced technologies. (Chapters III and VIIIdiscuss methods to restrict the spread ofenrichment and reprocessing. )

All enrichment technologies capable of pro-ducing highly enriched uranium from naturaluranium in a few stages should be closelywatched by the United States. At the time ofthe ERDA decision point in 1979, the compet-ing ERDA technologies should be evaluatedfor proliferation potential, in addition toeconomical and technical promise. In particu-lar, the ability to safeguard advanced enrich-ment facilities and the possibility of tamper-resistant processes should receive attention.

In evaluating the proliferation potential ofadvanced enrichment technologies, the effectthat their uranium-conserving propertiesmight have on the economics of the introduc-tion of plutonium recycle and fast breederreactors should also be considered.

Detection of Dedicated Facilities

This report has not had access to anyclassified intelligence information. Therefore,only a few general comments on the detectionof dedicated facilities can be offered.

Once the political decision has been made, itwould take up to 5 years to build a facilitydedicated to the production of weaponsmaterial and to obtain the material for the firstexplosive. As discussed in chapter VIII “Inter-national Control of Proliferation, ” a nationwould probably be at an advantage if itsweapons program were not detected until afterit had assembled its first explosive. Therefore,the question of the detection of dedicatedfacilities focuses on the probability of detec-tion within a time span of approximately 5years—between the time a nation beginsserious internal discussion of the possibility toa short but significant time before it has theweapons material in hand.

The likelihood of detection of a dedicatedfacility in a particular country depends onseveral factors. For example, it will berelatively easy to detect a clandestine nuclearfacility in a country which otherwise has avery limited nuclear program. It will berelatively easy to detect a clandestine nuclearfacility in a country which appears to havecause to want a nuclear weapons capability,because intelligence analysts will be morealert for early indications of a move towardsclandestine nuclear activity. It will also berelatively easy to detect a large Level 11 nuclearfacility.

One of the most important intelligencetechniques, especially for the first indicationsof a dedicated facility, is political reporting.The very first indications of a dedicatedfacility are unlikely to come from technologi-cal techniques, such as satellite photography.Visible photography from satellite or aircraft

184

would become an important tool only after anactive, coordinated surveillance program hasbegun.

A sustained effort, probably over a periodof several years, coordinating many elementsof the intelligence system—political reporting,visible photography, monitoring of the move-ment of materials and persons, sampling forchemical or isotopic indicators (such as Kr85

for a reprocessing plant) would be necessaryto build up familiarity with the target of sur-veillance and thus confidence in conclusions.

It appears unlikely that a Level II facilitycould long escape detection. Too many peoplewould be involved in its design, construction,and operation. Level I facilities probablywould present a detection problem in manycountries, especially if the country were notconsidered one of the five or six most likelyNth countries. Intelligence agencies cannotcontinually monitor the whole world for dedi-cated facilities, and must allocate theirresources according to priorities of problemsand priorities of targets.

PURCHASE AND THEFT

A third potential route to the acquisition ofnuclear weapons is the direct purchase ortheft of either the fissile material or theweapons themselves. The commodities mightbe purchased through an illegal nuclear blackmarket, bought or traded from a friendly na-tion in what is termed a gray market, or evenstolen directly from some national nuclear-weapons arsenal. These paths bypass the needfor the expensive and demanding technologiesrequired by either the commercial power ordedicated facilities route, Thus, if this type oftransaction emerges, the scope of proliferationcould be extended to technologically limitednations and non-state adversaries (NSAS)who would otherwise have found the taskdifficult and risky. The pace of proliferationcould be further accelerated by the relativeease of obtaining weapons, a general sensethat the nonproliferation regime was crum-bling, and a specific concern that one’senemies could be covertly arming. This sec-tion describes and evaluates the three ele-ments to this route: black market; graymarket; and theft. Appendix VII of volume IIprovides further detail.

Black Market

The term black market, as used here, meansthe illicit trade of goods where the commoditydoes not in general belong legitimately to theseller. The commodities traded in a nuclearblack market could be fissile material,

weapons designs, or actual weapons. Themost probable fissile material is plutoniumderived from commercial power cycles,because it can be directly used for weaponsfabrication. Only a very small fraction of theplutonium expected to be moving in a world-wide plutonium fuel cycle by the end of thiscentury would have to be diverted to producemany bombs annually. Research-reactor andbreeder-experiment fuel are other potentialsources. A detailed design of an effectivebomb would be an attractive commodity,especially for NSAS, because it would reducethe time and risk necessary to develop aneffective weapon. The third black market com -modity —weapons —might be stolen frommilitary stockpiles, particularly if prolifera-tion continues and security is lax in the newweapons states.

Participants in black markets can becategorized as buyers, suppliers, and inter-mediaries. Several potential participants canbe identified in each category, and the type oftransaction and motivation varies with theparticipants. Buyers might be nations or sub-national groups (terrorists, political or mili-tary factions, and criminals). The types of na-tions most likely to pursue a black marketroute are those technologically limited but in-ternationally ambitious or those confrontedwith a sudden dire emergency whichprecludes the more conventional but time-consuming routes. Demand for illicit weaponsor strategic nuclear materials could arise for

185

economic reasons. An approximate price forplutonium if freely traded could be about$9000/lb. ($20/gram).4 Ten kg for one or twobombs would at that price be $200,000, and asmall arsenal of 20 bombs would cost lessthan $4,000,000. The black market pricewould probably be several times higher, buteven so the total cost could still be much lessthan that of the construction and operation ofdedicated facilities. Subnational groups thatconsider terror to be a legitimate weaponcould be drawn to nuclear weapons asdescribed in chapter V, but might find pro-curement of the material otherwise toodifficult. A military faction might wantnuclear arms to facilitate a coup, or to hold inreserve for a national emergency if the civiliangovernment has forsworn their development.Criminal groups, conceivably even in-dividuals, might want to acquire arms for ex-tortion.

Different commodities require differentsuppliers, Fissile material (plutonium) mightbe diverted by an employee at a nuclearfacility such as a reprocessing plant. Motiva-tion could be money, coercion, or ideology.Alternatively, strategic nuclear materialscould be acquired by terrorists or criminalgroups staging an armed attack, probably onshipments. Military weapons might also beprocured by armed attack, but the tightersecurity would require even higher motiva-tion on the part of the attackers. Corrupt mili-tary elements in a nuclear weapons statemight steal their own bombs for profit,especially if security is casual. If intermedi-aries are involved they would most likely becriminal or international terrorist groups.

One constraint on a nuclear black market isthe difficulty of initiating transactions. Mostbuyers and suppliers are unlikely partners.Contact and trust may be difficult to establish,except possibly between terrorist groups. Sup-pliers can generally find buyers more easily

qBased on previous expectations and discussions withindustry representatives. Utilities presently assign zerovalue to their plutonium in the spent fuel, but if recycleis allowed, the value would depend on the cost of theenrichment which the plutonium replaces, the cost ofreprocessing, and the additional cost of mixed-oxidefuel fabrication.

than vice versa, since potential buyers arerelatively obvious. By contrast, a suppliermight be the only employee out of 500 at areprocessing plant with the motivation andthe ability to divert plutonium. The supplier,however, runs the greater risk since he entersinto the transaction with the illegally obtainedcommodity.

These transactions are more likely to occurif both the supply and demand are high. Thesupply of weapons designs and weaponsthemselves is likely to change only slowly(although access to them may increasefaster). The potential supply of fissile material,however, could increase dramatically if large-scale reprocessing and plutonium recycle areinitiated. If all the spent fuel from 1,000 LWRS(anticipated by 1995) is reprocessed, thendiversion of one-tenth of 1 percent of the an-nually produced plutonium would be suffi-cient for about 50 bombs. This supply mightbe limited by effective safeguards and physicalsecurity, which can sharply reduce oppor-tunities for illegal diversion, just as theyreduce opportunities for national diversion.Material accounting, containment, and sur-veillance will reduce employee theft, whilephysical security should deter and repelarmed attack, Physical security is especiallyimportant to protect weapons.

Given sufficient supply and demand, a sus-tained market could emerge from initial inter-mittent transactions. Thus, the market wouldbe transferred from an amateur to a profes-sional operation. The latter would be moredangerous because it would be continuallyseeking new suppliers and customers, andbecause the greater expertise of the operatorswould inhibit interference. A full-blownmarket could consist of many individualdiversion activities and continuing networks,with criminal organizations providing neces-sary middleman services. A sustained blackmarket requires a high demand, which wouldprobably come only from less developedcountries: more advanced countries wouldwant more and better bombs than a blackmarket could be expected to provide, andNSAS are unlikely to be able to afford morethan a few. The major source of supply mightbe a number of reprocessing plant employees.If each smuggled out just one gram of

186

plutonium per day (an amount probably toosmall for either material accounting or portalmonitors to detect) he should realize at least$5,000 per year and maybe several times that.This source could be supplemented by attackson shipments of plutonium, which could netseveral million dollars worth of material. Amarket of several hundred pounds of fissilematerial worth millions of dollars per yearseems credible. Although small by com-parison to the drug market, this is largeenough to interest criminal groups and tohave a major impact on proliferation.

Gray Market

A gray market falls between a black marketand normal commercial transactions. Thecommodity belongs legitimately to the sellerand the transaction is legal under the laws ofthe nations concerned but must be covertbecause it would be unacceptable if knownpublicly. The main reasons for secrecy ofnuclear transactions would be to avoid alert-ing an enemy and to avert domestic or inter-national reaction to furthering proliferation,especially if in violation of the NPT. Thetransaction could involve weapons, fissilematerial, or technical assistance.

The buyer in a nuclear gray market couldonly be a government, because purchase byany non-national group would be illegal. Thesupplier could be another government, a cor-poration, or an individual. Government-to-government transfer of nuclear arms could oc-cur if a close and valued ally was on the vergeof annihilation. Sale or barter of such weaponsunder more normal conditions is less likely.Fissile material is a more probable com-modity, and technical assistance the mostlikely. The latter could consist of design infor-mation for either weapons or plutonium pro-duction facilities, or the critical componentsfor either one, A supplier nation might enterinto gray market transactions either at the de-mand of a nation that provides a vital resource(e.g., oil) or by the desire to gain political sup-port (e.g., Pakistan and India both trying togain favor with Arab nations), Alternatively,some nations may engage in a joint develop-ment program to reduce costs and shortenschedules.

Corporations with a large investment orsubstantial business expectations in anothercountry could be subjected to considerablepressure to assist in a weapons program, par-ticularly the plutonium production aspects.Revelations of corporate bribing of foreignofficials gives a certain credence to thisspeculation, but the difference between a bribeand a contribution to proliferation will not belost on corporation executives. The impact ofexposure could also be much larger, Further-more, the nations with the most leveragewould the ones needing the least assistance.Hence, this type of transaction seems lesslikely than governmental assistance. If it doesemerge, however, the most likely supplierswould be reactor manufacturers, architect-engineers, and consulting companies. Theseare discussed in appendix IV of volume 11.Companies might be more susceptible toforeign overtures if their domestic nuclear ac-tivities are curtailed.

Individuals could contribute to a weaponsdevelopment program by becoming scientificmercenaries. A sizable pool of scientific man-power conversant with plutonium reprocess-ing, materials handling, and related fuel-cycletechnology already exists. Lack of demand fortheir skills at home might force a few to seekemployment elsewhere, and bitterness overtheir loss of careers could overcome their

‘scruples about contributing to proliferation. Aconstraint on this movement would be thedesire of most nations to keep their weaponsprogram secret. The nation may not wish torely on the loyalty of foreigners in this situa-tion, and may be unable to sequester themvoluntarily for the long duration of thedevelopment program.

It is possible that some examples of graymarketing have already occurred. It wasreported in 1975 that West Germany had beencovertly involved in South Africa’s uraniumenrichment development programs Thiscooperation was denied but some evidence in-dicates it may have existed. Nuclear mercen-aries have a precedence in the migration ofscientific manpower to the developed coun-tries in the brain drain of the 1950’s and1960’s.

5 T/le obserz)er (London), Oct. 5, 1975.

187

Countermeasures to Black andGray Market

An important step in combatting thesetransactions is to detect them. Intelligence-gathering operations can serve to identify par-ticipants, but the difficulty experienced withcracking the illegal drug market illustrates theproblems that will be encountered inpenetrating a nuclear black market. Isolatedtransactions would be even harder to detectunless the participants revealed themselves. Ifthe buyer in either a gray market or blackmarket is a government, then some aspects ofits weapons fabrication may emit unique in-telligence signals (as for other weaponsdevelopment programs). This is discussed inthe previous section, “Dedicated Facilities. ”Intelligence activities could also track migrat-ing manpower, but the difficulty of separatingthe critical cases from the legitimate move-ments will be great and conflicts with civilliberties may arise. International safeguardsshould be capable of at least detecting whensignificant diversion has occurred. With thatas a start, then intelligence can more easilytrack the material and determine the partici-pants.

International safeguards have been directedat national diverters, but the same methodswould be effective against black market diver-ters. Both intelligence and safeguards can beenhanced and reoriented towards this threat.Increased effectiveness in detection would be apotent deterrent to potential participants. Thefactor that would probably have the greatestimpact in controlling a black market in fissilematerial would be to limit plutonium recycle.The supply that does exist can be made less ac-cessible by enhanced physical security,

The willingness of participants to engage inthese transactions depends not only on per-ceived rewards and risks of detection, but alsoon the consequences of detection. Possibleresponses might include sanctions againstcountries engaged in nuclear gray marketing,police work to capture black marketeers, andcontrol of the activities of potential nuclearmercenaries and corporations abroad.

Theft of Nuclear Weapons

The most direct route to a nuclear weaponis the theft of someone else’s. This report doesnot analyze weapons security in detail.Nevertheless, certain observations can bemade. Fewer groups are capable of attacking anuclear weapons stockpile or transport thancould participate in a black market. Onlyhighly motivated, well-organized, and well-armed attackers would have much chance ofovercoming effective military security sur-rounding weapons.

U.S. nuclear weapons consist of bombs,missiles, artillery shells, depth charges, tor-pedoes, and demolition charges.6 All are pro-tected against unauthorized use by internalmechanisms. None of these can prevent theweapons from being used simply as a sourceof high quality fissile material, but the delaywould enhance the chances of recovery. Evenwithout rebuilding the weapons, however, thethief would achieve full psychological value ofpossession.

U.S. weapons are kept in Europe, the PacificOcean area on naval vessels, and at home.Storage sites are usually on military installa-tions. The protection provided is morestringent than that required for commercialfissile material, but the need for upgrading isrecognized and being addressed by theDepartment of Defense. Weapons storedabroad might become less secure if the hostgovernment suddenly changed hands.Transport for logistical purposes is probablythe most vulnerable link, but it is also infre-quent.

It is difficult to defend against a determined,effective, comando type of attack. Groups ofabout 8 to 20 attackers using an imaginativeplan and aided by one or more insiders wouldbe especially difficult to resist without rapidreinforcement. On the other hand, it wouldalso be difficult to mount this type of attack

bJoint Committee on Atomic Energy, Development,Use and Control of Nuclear Energy for the CommonDefense and Security and for Peaceful Purposes, 1976.

188

without giving some warning to ap-propriately oriented intelligence activities.Massive attacks such as the Israeli raid on theEntebbe Airport are least likely to be suc-cessfully resisted, but neither can they be ac-complished anonymously. Consequently,political and military responses, if activated,can be expected to ensure return or destruc-tion of stolen weapons.

Other present nuclear weapons states ap-pear to present about the same barriers totheft as the United States. New nuclear states,however, may be more vulnerable. Some po-tential Nth countries have experienced tur-bulent domestic politics, and factions couldseize weapons for their own use or for sale ona black market. This threat could be exacer-bated if some Nth countries are unconcernedabout physical security, or feel it is secondaryto the need for immediate operational readi-ness. Furthermore, such nations will probablynot have the sophisticated protective mecha-nisms built into their weapons.

ConclusionsThe emergence of a black market is pres-

ently constrained by the lack of supply offissile material. Widespread plutonium recyclewould remove this constraint. Some demandappears to exist, as already evidenced byLibya’s attempts to buy a bomb.7 This demand

7steven J. Rosen, Nuclear Proliferatiotz and the Near-Nuclear Countries, p. 178, Bullinger Publishing Co.,Cambridge, Mass., 1975.

could increase if more nations feel intensesecurity concerns or if they sense a continuingpattern of proliferation and feel they, too,should have a few nuclear weapons in reserve.The inherent lack of prestige of weapons at-tained by this route may deter some, butothers might feel no compunctions. Thus, ifsupply is not limited, the outcome is likely tobe at least intermittent black market transac-tions.

Gray market transactions appear at least aslikely as those on the black market. The sup-ply of some commodities already exists, theparticipants are more natural partners, andless risk would be involved. Gray markettransactions would be individually negoti-ated, and so present less danger of spreading.The existence of either black or gray marketswould be a serious blow to nonproliferation.They would themselves lower the barriers toweapons, and the feeling that nonprolifera-tion efforts had failed would spur other na-tions to procure their own weapons.

Theft of weapons is the hardest to evaluate.Largely unpredictable conjunctions of motiva-tion, ability, and opportunity would have tooccur. Unless the attack is overwhelming, suc-cess will depend to some extent on luck. Themilitary and psychological effectiveness of astolen weapon would probably be substan-tially greater than that of a homemade one,particularly for non-state adversaries. Hence,physical security of weapons must be suchthat the risk of losing them is very low.

189