[NCI Logo]

The IAEA's Inability to Detect Diversions of Bomb Quantities of Plutonium


Paul Leventhal

September 12, 1994


The International Atomic Energy Agency (IAEA) is the world's nuclear watchdog. The Nuclear Non-Proliferation Treaty (NPT) requires "safeguards" to be applied by the agency. If nuclear material is diverted from a peaceful nuclear program for weapons purposes, the IAEA is supposed to detect the diversion in time to permit an international response before the diverted material is made into bombs. Many nations, relying on the IAEA to deter diversions by means of such "timely detection," accept the use of weapons-quality nuclear fuel (plutonium and highly enriched uranium) as a legitimate application of "atoms for peace" in civilian power and research programs.

The IAEA, in its public statements, promotes this trust by offering what appear to be categorical assurances that nuclear material at specific facilities is under safeguards and has not been diverted. In May, when the Nuclear Control Institute raised questions about an undisclosed discrepancy involving 70 kilograms of plutonium at a Japanese fuel fabrication plant1, the IAEA issued a statement explaining that the material in question had stuck to the surfaces of the plant's process stages known as "glove boxes." According to the IAEA, "The nuclear material held-up in the glove boxes of the Tokai Nuclear Fuel Fabrication Plant is not missing and remains under full safeguards and is declared."2

Upon close examination, however, such public assurances by the IAEA are ambiguous and misleading. The plutonium held up in a plant is, by definition, "under full safeguards," "declared," and presumably "not missing" when the facility is subject to IAEA safeguards and the agency has found no evidence of a diversion. Not addressed, however, are (1) the precise amount of plutonium stuck in the glove boxes and elsewhere in the plant and (2) how much plutonium could be diverted from such a plant under cover of the uncertainty in measuring this material and the difficulty this uncertainty causes the IAEA in determining the overall plutonium inventory at any given time.

The IAEA artfully avoids this question. The question is prompted by the technological fact of life that plutonium powder sticks to surfaces in the process stages of fuel fabrication plants and is quite difficult and hazardous to clean out. These plants also generate considerable scrap that cannot be measured accurately. Even greater material-accounting difficulties and uncertainties are encountered by the IAEA in reprocessing plants where highly radioactive spent fuel from nuclear reactors is chopped up and dissolved in acid and plutonium is extracted and processed as a liquid by remote-control.

By not publicly acknowledging these severe accounting problems, the IAEA avoids the fundamental question about safeguarded facilities where plutonium is handled in bulk: Can the IAEA meet its basic goal of declaring with high confidence that a bomb-quantity or more of plutonium has not been diverted?

The IAEA acknowledges in its technical safeguards documents that, due to measurement uncertainties, its material-accounting system cannot with confidence detect the diversion of bomb quantities of nuclear material. However, these documents state the problems in complex, scientific terms that are not understood by most policymakers and the general public. If properly understood, the IAEA's safeguards shortcomings have urgent implications for national policies and for the NPT and other international arrangements that permit commerce in weapons- quality nuclear material for civil nuclear programs on the premise that the agency's safeguards are effective.

What follows is a critique, expressed to the extent possible in non-technical terms, of the IAEA's inability to account accurately for civilian, bomb-quality nuclear material in world commerce.

Accounting Goals: A Moving Target

The IAEA begins with a theoretical accounting goal that, if met, would establish a meaningful non-proliferation standard. But the IAEA relaxes its standard several times prior to the actual implementation of safeguards. When IAEA safeguards are actually applied to commercial-scale nuclear fuel facilities, the agency's accounting methods could miss the diversion of dozens of weapons' worth of nuclear material a year at these facilities.

"Verification of Non-diversion." This is the IAEA's theoretical goal that the public and most policymakers assume the Agency can actually implement. Under this standard, the agency's safeguards are supposed to enable "the IAEA to conclude for a given period that no significant quantity of nuclear material has been diverted or that no other items subject to safeguards have been misused by a State."3 A "significant quantity" is the amount of nuclear material for which "the possibility of manufacturing a nuclear explosive device cannot be excluded."4 For plutonium, a significant quantity is defined as eight kilograms, although it is well established that modern nuclear weapons can be made from half that amount, or less.5

In its technical safeguards documents, however, the IAEA admits it cannot meet this important detection standard. The Agency concedes that such terms as significant quantity "have to be 'translated' into detection goals so that safeguards can be applied in an effective manner."6 This is an evasive way of saying that since the IAEA cannot meet its original safeguards standard, it must relax the standard to one that can be met.

There are two main reasons the IAEA cannot meet its original standard. First, there are unavoidable uncertainties in measuring the inputs, outputs, and inventory of material at a nuclear fuel facility. Since the IAEA can never establish with certainty how much plutonium has gone into a facility or how much has come out, it cannot know if a small but significant percentage is missing. This problem is especially pronounced at reprocessing plants, where plutonium is chemically extracted from the spent fuel of nuclear power plants. At the present time, the IAEA uses plus or minus 1% as the international standard for the expected uncertainty in measuring the amount of plutonium in such a plant. Most of this uncertainty is due to the error in measuring the plutonium entering the plant.7

Second, as noted, during operation of a nuclear fuel facility a significant amount of material inevitably becomes stuck inside. In older facilities, this "hold-up" is simply estimated, with notorious inaccuracy. In newer facilities, such as the Japanese fuel fabrication plant cited above, the held-up plutonium is remotely measured, but again with significant inaccuracy---at least a 5-to-10% uncertainty according to the equipment's designers, but more likely 15%, according to Japanese experts, and possibly as high as 30%, according to a U.S. expert. A more accurate accounting of the hold-up can be achieved by painstakingly cleaning out the plant's processing equipment. However, during the clean-out operation, some of the plutonium ends up in the plant's radioactive waste stream, where plutonium measurements are again highly inaccurate. Even a meticulous clean-out does nothing to address the fundamental uncertainty of inputs and outputs, especially in reprocessing plants.

"Expected Accountancy Capability" (a.k.a. "Accountancy Verification Goal"). This is the IAEA's first relaxation of its safeguards goal and is defined as "the minimum loss of nuclear material which can be expected to be detected by material accountancy."8 For bulk-handling plutonium facilities the minimum detectable loss can be quite large---some 263 kilograms, or 33 significant quantities, a year in a reprocessing plant such as the one that Japan plans to build to process about 8 metric tons of plutonium from 800 metric tons of spent fuel annually.9

The expected accountancy capability is arrived at by means of statistical theory and is based on assumptions of a detection probability of 95%, a false-alarm rate of 5%, and a measurement error of plus or minus 1%. The false-alarm rate is the chance that if accountancy measures indicate a diversion of material, the indication will turn out to be a false alarm. The false-alarm rate depends on how high the plant operator and the IAEA agree to set the threshold for the amount of material that must be measured as missing before the IAEA can suspect a diversion. The higher the threshold, the lower the false alarm rate.

In the example of the large reprocessing plant cited above, a threshold of 263 kilograms of "missing" plutonium is tolerated to ensure a false-alarm rate of no more than 5% (in combination with a detection probability of 95%). In other words, to avoid a false alarm, the IAEA could measure a shortfall of plutonium equivalent to 33 nuclear weapons but still not suspect a diversion. It could not even consider the situation to be an "anomaly"---a polite term used by the agency to describe an unexplained loss that needs to be resolved.

"Inspection Goals." These are the actual goals as specified by the IAEA to be used by inspectors for detecting diversions at a given facility. They "reflect actual conditions at the facility, requirements prescribed by the safeguards agreement [with the plant operator] and technical capabilities of safeguards measures."10 These goals can be relaxed further from the "expected accountancy capability."

One reason is that, in practice, the expected measurement uncertainty often proves to be too optimistic, resulting in much more measurement error and potential for diversion. Another reason is that although the IAEA assumes a detection probability of 95% and a false alarm rate of 5%, nonetheless, in practice, plant operators insist upon, and the IAEA agrees to, a lower false-alarm rate to avoid interference with cost-effective operation. However, as noted, reducing the rate of false alarms requires raising the threshold amount of plutonium that must be measured as missing to indicate a diversion. Thus, even larger accounting shortfalls are tolerated by the IAEA---ignored, really, because they do not qualify as "anomalies" needing to be resolved to the agency's satisfaction.

"Actual Implementation." Even these relaxed inspection goals may not be achievable in practice. As the IAEA acknowledges, "The actual implementation and the extent to which inspection goals embodied in inspection procedures can be achieved depend on co-operation from the State and the operator, on the availability of manpower, safeguards equipment and inspection support services, and on the IAEA budget."11 In other words, the IAEA is heavily dependent on the good will of the inspected party and on the adequacy of the agency's resources. If an operator acted on behalf of a state to employ systematic techniques to conceal diversions, the amount of diverted plutonium that could escape detection would increase still further.

In the case of the Japanese fuel fabrication plant cited above, the plant operator had been resisting for a number of years IAEA requests for a prompt clean-out of the plant, presumably because of the substantial costs, inconvenience and hazards involved. Controversy caused by public disclosure of the plutonium hold-up problem has since prompted the operator to cooperate with the IAEA in preparing a clean-out schedule. But according to Japanese officials, the clean- out could take several years, and there is no requirement that the final results of the audit be made public. In the case of an operator engaged in a systematic diversion scheme, the operator could delay and complicate the clean-out process and then argue endlessly with the IAEA over the accuracy of its audit.

Containment and Surveillance No Substitute for Materials Accountancy

The IAEA claims that containment and surveillance techniques, including application of tamper-indicating seals, installation of video cameras, and on-site inspections can make up for the above-cited inadequacies in materials accounting. In its safeguards agreements with NPT- member states, however, the IAEA establishes "materials accountancy as a safeguards measure of fundamental importance, with containment and surveillance as important complementary measures" for timely detection of diversions.12 One reason is that operators sometimes resist the installation of the most modern and intrusive containment and surveillance measures. Another is that even the most advanced measures can be defeated or rendered non-conclusive.

Seals can be broken or cut and the view of cameras can be blocked or obscured during normal operations or in response to emergencies without necessarily indicating an attempt to conceal a diversion. The IAEA does not publicly report breakdowns in containment and surveillance. Nor, in the absence of conclusive evidence of a diversion, is the agency in a strong political or technical position to challenge an operator's claim that the breakdowns are innocent. Yet, such unreported and unchallenged breakdowns could be used to conceal diversions, especially if the plant operator were the diverter.

Unpoliced diversion pathways are available to plant employees who might act independently of the operator in carrying out a diversion scheme on behalf of outside criminal or terrorist elements or a state determined to acquire nuclear material for weapons. One such pathway is the trash---the so-called low-level radioactive waste stream in which lightly contaminated uniforms, wipes, tools and equipment are collected routinely and removed from the plant. This waste stream is not well policed for plutonium because no appreciable amounts of plutonium reach it in the normal operation of the plant. However, a knowledgeable employee would be able to place diverted plutonium in a low-level waste drum in a way that shielded the plutonium from environmental-monitoring equipment used to detect excessive contamination of the waste. Once the waste drum were removed from the plant, the employee could recover the plutonium. More sensitive monitoring devices, such as neutron coincidence counters and gamma scanners, generally are not used to police the low-level waste for diverted plutonium because plant managers, in the interest of maintaining cost-effective operation, cannot slow down the flow of waste to a rate at which such detectors might reliably detect such concealment.13

Timeliness Goal Not Met

Timeliness of detection, the sine qua non of safeguards, is another of the IAEA's theoretical goals. As stated in the IAEA's safeguards agreements with NPT-member states, the agency's objective is "the timely detection of diversion of significant quantities of nuclear material from peaceful nuclear activities to the manufacture of nuclear weapons or of other nuclear explosive devices or for purposes unknown, and deterrence of such diversion by the risk of early detection."14

In practice, the IAEA simply cannot ensure timely detection through its safeguards methods of materials accountancy, containment and surveillance. The agency estimates that the conversion time for plutonium oxide powder to a weapon is on the "order of weeks (1-3)." It sets a "timeliness goal" of "within 1 month for fresh fuel containing HEU, Pu, or MOX [mixed oxide fuel]." Even the validity of this goal is open to challenge because a nuclear industrial state engaged in a diversion scheme could certainly convert diverted fuel into prefabricated weapons in a matter of hours or days.

At reprocessing and fuel fabrication plants, clean-out inspections are usually performed only annually. If a large discrepancy is detected, the agency will have to spend months working with the plant operator trying to figure out a technical reason for the discrepancy, prior to officially declaring the discrepancy an anomaly that needs to be resolved. The process of resolving an anomaly to the point of determining whether a suspected diversion should be reported to the IAEA Board of Governors could take months more, as could the process of the board determining whether the matter needs to be referred back to the inspectors for further resolution or is of a magnitude to be referred to the UN Security Council.

Thus, there is no credible assurance that the IAEA could blow the whistle on a diversion before the diverted material is turned into a bomb. Even around-the-clock surveillance by a few resident inspectors can have little deterrent effect against a nation determined to divert. It is an uneven contest, at best, technically and politically; the agency consequently is disinclined to suspect a diversion, and its inspectors realize there is no reward for blowing the whistle.

Potential Diversions Kept Secret

The safeguards process is further undermined by the IAEA's highly secretive procedures, in which the results of audits and the reporting of discrepancies are all kept "safeguards confidential." This practice ostensibly is intended to protect "proprietary information," but its effect is to hide information about the inadequacy of safeguards that is an embarrassment to the IAEA and to the nuclear industry. Safeguards secrecy feeds suspicions among regional neighbors that "peaceful" plutonium (or highly enriched uranium) is being, or could be, moved into a clandestine weapons program without being detected by the IAEA.


In principle, civilian use of weapons-quality nuclear materials is premised on the IAEA's meeting its stated objective of detecting with confidence the diversion of a bomb-quantity or more of this material on a timely basis. However, an understanding of the IAEA's actual standards and procedures for safeguarding plutonium makes clear that the Agency does not come close to achieving this goal in practice. Although the IAEA has never reported a diversion from a safeguarded nuclear facility, there can be no assurance that diversions have not taken place, or will be detected in the future, given the large uncertainties in the agency's measurements, the limitations of its containment and surveillance measures, and the wall of secrecy that surrounds its operations.

Given the IAEA's inability to safeguard plutonium effectively, the agency has an obligation to admit its own limits and not continue to provide legitimacy for the plutonium industry by giving the impression that safeguards are providing more protection than is really the case. The prudent course, given the lack of economic need for plutonium fuel,15 is for peaceful nations to agree to bar civilian use of plutonium. This is a step that can be taken in conformity with the existing terms of the Nuclear Non-Proliferation Treaty.16 IAEA safeguards could then be restructured away from the overwhelming task of verifying peaceful use of plutonium to the more feasible task of verifying the absence of this weapons-usable material.

Return to Nuclear Terrorism Page NCI Home Page

End Notes

1. "Astounding Discrepancy of 70 Kilograms of Plutonium Warrants Shutdown of Troubled Nuclear Fuel Plant in Japan," press release, Nuclear Control Institute, Washington, D.C., May 9, 1994, issued with letter from the Institute to Secretary of State Warren Christopher, dated May 4, 1994. Back to document

2. "Japanese Nuclear Material Under Full Safeguards," press release, International Atomic Energy Agency, Vienna, Austria, May 25, 1994.Back to document

3. IAEA Safeguards Glossary, 1987 Edition, International Atomic Energy Agency, Vienna, 1987, p.3.Back to document

4. Ibid. p.23. Back to document

5. "The Amount of Plutonium and Highly Enriched Uranium Needed for Pure Fission Nuclear Weapons," by Thomas B. Cochran and Christopher E. Paine, Natural Resources Defense Council, Washington, D.C. August 22, 1994. Back to document

6. IAEA Safeguards Glossary, op. cit., p.21. Back to document

7. Plutonium input at a reprocessing plant is determined by measurements of plutonium dissolved in acid from chopped-up spent fuel. Some plutonium remains in the undissolved "hulls" of the spent fuel and is even more difficult to measure. Because of such difficulties, deviations as high as 20% from the IAEA's expected 1% measurement uncertainty at reprocessing plants have been observed in the past. Another guide to the plutonium content of spent fuel is a record, maintained by reactor operators, of the power levels achieved by the reactors in which the plutonium was originally produced---a crude measure that is not used by the IAEA to establish material balances. Back to document

8. IAEA Safeguards Glossary, op. cit., p. 27. Back to document

9. The formula for determining Expected Accountancy Capability (E)---that is, "the minimum loss of nuclear material which can be expected to be detected by material accountancy"---is E=3.29sigmaA, in which sigma is the relative uncertainty in measurements of the plant's inputs and outputs and A is the facility's plutonium throughput in between periodic physical inventories. The expected minimum detectable loss (E) at a reprocessing plant that extracts 8 metric tons of plutonium from 800 tons of spent fuel per year is determined, on the basis of annual physical inventories, as follows:

E = 3.29 x (measurement uncertainty) x (annual throughput)
= 3.29 x .01 x 8000 kgs. Pu
= 263 kgs. Pu
= 33 significant quantities of plutonium

For further elaboration of the difficulties of detecting diversions from large-throughput reprocessing plants, see MarvinM. Miller, "Are IAEA Safeguards on Plutonium Bulk-Handling Facilities Effective?", Nuclear Control Institute, Washington, D.C., August 1990. Back to document

10. IAEA Safeguards Glossary, op. cit., p.28. Back to document

11. Ibid. Back to document

12. INFCIRC/153 (Corrected), International Atomic Energy Agency, Vienna, February, 1983, p. 9. Back to document

13. Paul Leventhal, Milton Hoenig and Helen Hunt, "Nuclear No Man's Land: Low Level Radioactive Wastes as an Unpoliced Diversion Path for Thefts of Weapons-Usable Nuclear Materials," Nuclear Control Institute, Washington D.C., September 16, 1988. This special report was presented at a hearing of the 2nd Committee of Investigation of the Bundestag, Federal Republic of Germany, September 22, 1988. West German officials acknowledged to the committee that the low-level waste stream in German plutonium plants was not policed against diversions by means of neutron or gamma measurements.Back to document

14. Ibid. Back to document

15. Mixed-oxide, plutonium-uranium fuel is at least 4 to 6 times more expensive than low-enriched uranium, the conventional fuel for light-water reactors. See Paul Leventhal and Steven Dolley, "A Japanese Strategic Uranium Reserve: A Safe and Economic Alternative to Plutonium," the Nuclear Control Institute, April 9, 1993. Back to document

16. Article IV of the NPT, establishing the "inalienable right" of Treaty parties to develop nuclear energy for peaceful purposes, specifically requires that its terms be interpreted "in conformity with" the prohibitions in Articles I and II against the giving or receiving of nuclear assistance applicable to development of nuclear weapons. The reason weapons-usable plutonium and uranium are not interpreted to be prohibited by the NPT is that the application of IAEA safeguards, as required by Article III, is regarded as adequate to verify that these materials are not being diverted to make nuclear weapons. Since IAEA safeguards are clearly inadequate to provide such assurances, and since non-weapons-usable fuels are readily available for civilian nuclear programs, plutonium and highly enriched uranium can be banned under the existing terms of the Treaty. See Eldon Greenberg, "The NPT and Plutonium: Application of NPT Prohibitions to 'Civilian' Nuclear Equipment, Technology and Materials Associated with Reprocessing and Plutonium Use," Nuclear Control Institute, 1993. Also see Paul Leventhal, "Plutonium and the NPT," Carnegie Endowment for Peace, November, 1993. Back to document