6

Nonproliferation Implications and Security Risks

This chapter reviews the nonproliferation implications and security risks of the fuels and fuel cycles associated with advanced nuclear reactors. It responds to the charges in the statement of task that call for assessments of high-assay low-enriched uranium (HALEU), uranium-plutonium mixed oxide fuel, and advanced fuel cycles that require separating plutonium from spent fuel; examination of nuclear material accounting and control, as well as containment, surveillance, monitoring, and timeliness of detection of diversion; and suggestions for how these nonproliferation implications and security risks can be addressed by International Atomic Energy Agency safeguard activities.

In this chapter, the committee provides the summary, findings, and recommendations up front (Section 6.1); describes relevant aspects of nonproliferation, safeguards, and nuclear security, including transportation security (Section 6.2); and examines the nonproliferation implications and security risks of fuels and fuel cycles for various advanced reactors, with particular attention to HALEU because of the interest expressed by most advanced reactor developers in using this material in their fuels (Section 6.3).

6.1 CHAPTER 6 SUMMARY, FINDINGS, AND RECOMMENDATIONS

Deployment of advanced reactors and their supporting fuel cycles will involve the production, transportation, storage, and irradiation of nuclear materials with significantly different characteristics from those of the current U.S. light water reactor (LWR) fleet, which is fueled with low-enriched uranium. These differences could impact nuclear proliferation¹ and nuclear terrorism² concerns posed by advanced reactors. Evaluating advanced reactor fuel cycles with respect to their potential nuclear proliferation implications and nuclear terrorism risks (referred to as “safeguards” and “security risks”) is warranted. It is also important to assess the feasibility and efficacy of ways to mitigate these risks by improving and strengthening the technical and institutional measures needed for nuclear material safeguards and physical protection.

Addressing these issues first requires identifying the features of advanced reactor fuel cycles relevant to safeguards and security. These include the following considerations: Advanced reactors may variously use special nuclear materials, such as HALEU, plutonium, minor actinides, or thorium and uranium-233, either separately or in mixtures. As discussed in Chapter 4, most advanced reactor designs propose using HALEU fuel, which, given

___________________

¹Nuclear proliferation is defined as diversion or misuse of civil nuclear materials by a state to produce nuclear weapons.

²Nuclear terrorism is defined as acquisition of nuclear materials by substate actors seeking to build improvised nuclear explosive devices.

that its material attractiveness is higher than low-enriched fuel, would require additional security considerations throughout the fuel cycle and potentially a new safeguards framework to mitigate proliferation concerns. Fuels for advanced reactors may also have nuclear material concentrations and chemical and physical forms that differ from LWR fuel. Advanced reactors may use different refueling cycles than LWRs, ranging from online refueling to long-lived once-through cores, both of which present unique safeguards and security challenges. Deployment of small modular reactors or microreactors would increase the number of units and sites requiring safeguards and security measures for a given amount of power capacity compared with large reactors—which could increase the burden on international safeguards inspectors and domestic security resources. One of the most important factors is whether fuel cycles are open (using interim storage and eventual direct disposal of spent fuel) or closed (involving reprocessing and reuse of separated nuclear materials).

With respect to the main focus of this report, the deployment of advanced reactors and fuel cycles in the United States, direct proliferation by the United States is not a concern, as it is a nuclear weapon state. Nonetheless, safeguards obligations on domestic facilities that the United States could assume under its voluntary offer agreement with the International Atomic Energy Agency (IAEA) would have indirect benefits by providing test beds for IAEA safeguards approaches for new technologies that may be deployed in other countries and would set an example for international cooperation.

Nuclear terrorism, however, is a concern within the United States and therefore must be addressed in the physical protection and emergency response plans for domestic nuclear facilities. As the 9/11 attacks demonstrated, various elements of the U.S. critical infrastructure could be a target for attack by terrorist organizations and require a high level of physical protection. Today, such groups include a range of substate actors, both foreign and domestic in origin, as well as those receiving state support. These threats can be carried out by insiders and/or external adversary groups, and could involve both physical attacks and cyberattacks. Current security requirements with regard to both theft and radiological sabotage may need to be modified to address the specific characteristics of various advanced reactors and fuel cycles.³

The committee makes the following findings and recommendations regarding the nonproliferation implications and security risks of advanced reactors and fuel cycles:

Finding 19: Expanding the global use of high-assay low-enriched uranium (HALEU) would potentially exacerbate proliferation and security risks because of the potentially greater attractiveness of this material for nuclear weapons compared with the low-enriched uranium used in light water reactors. The increased number of sites using and states producing this material could provide more opportunity for diversion by state or nonstate actors.

Finding 20: All of the advanced reactor fuel cycles will require rigorous measures for safeguards and security commensurate with the potential risks they pose. Issues requiring special attention include the following:

• Material accountancy (i.e., tracking and quantification) is more difficult for molten salt and pebble-bed technologies than for reactor systems that use stationary solid fuels because of the technical challenges in performing measurements with online fuel and bulk-handling facilities. Containment and surveillance will also be more challenging to implement for these types of reactors. Thorium/uranium-233 fuel cycles require development of safeguards technology because of the large number of variants in their systems. Moreover, safeguards tailored to traditional uranium/plutonium fuel cycles are not applicable to these systems.

___________________

³ A detailed evaluation of radiological sabotage is beyond the scope of this report.

• Fuel cycles involving reprocessing and separation of fissile material that could be weapons usable pose greater proliferation and terrorism risks than the once-through uranium fuel cycle with direct disposal of spent fuel, as the separated fissile material would not be uniformly mixed with highly radioactive fission products. Separated, potentially weapons-usable materials could include fissionable materials other than the “traditional” special nuclear materials of highly enriched uranium, plutonium, and uranium-233. Thus, for these closed fuel cycles, specific safeguard technologies will likely be required to meet the International Atomic Energy Agency’s goal of timely detection.

6.2 BACKGROUND ON NONPROLIFERATION, NUCLEAR MATERIAL SAFEGUARDS, AND NUCLEAR SECURITY

6.2.1 Nuclear Nonproliferation

The Treaty on the Non-Proliferation of Nuclear Weapons (NPT), which was signed in 1968, entered into force in 1970, and was extended indefinitely in 1995, seeks to globally “prevent the spread of nuclear weapons, to promote cooperation in peaceful uses of nuclear energy and to further the goal of nuclear disarmament”⁴ (DOS, 2010, n.d.; IAEA, n.d.-a; NPT, 1968; UN, n.d.-a). The three goals of the NPT—nonproliferation of nuclear weapons, nuclear disarmament, and peaceful use of nuclear energy—are described in detail in such sources as IAEA (1970, n.d.-a) and UN (n.d.-a).

6.2.2 Safeguards

The goal of safeguards is verifying that states are in compliance with treaty obligations prohibiting the diversion of civil nuclear materials and technologies for nuclear weapons. After states sign Comprehensive Safeguard Agreements⁵ (CSAs) or Voluntary Offer Agreements⁶ (VOAs), the objective of the IAEA is to verify a state’s safeguards compliance in order to ensure that “the nuclear material, non-nuclear material, services, equipment, facilities and information specified and placed under safeguards are not used for the manufacture of nuclear weapons or any other nuclear explosive devices or to further any military purpose” (IAEA, 2002). In this regard, the technical objective is specified as “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” (IAEA, 2002).

___________________

⁴ There are 191 states partied to the NPT, with only 5 recognized as nuclear weapon states: the United States, the United Kingdom, China, France, and Russia. Only these states are legally allowed to possess nuclear weapons under the NPT, and they have agreed to make good-faith efforts to disarm in the future. The NPT is open to any UN member state, provided they sign as a non–nuclear weapon state. Israel, India, Pakistan, and South Sudan are the only 4 of the 193 UN member states that are not party to the treaty. Palestine and the Holy See are observer states and thus technically party to the NPT, although they are not UN member states. North Korea claims to have left the treaty in 2003, but this action has not been officially recognized by all states party to the NPT.

⁵ Each non–nuclear weapon state party to the NPT is required to sign a CSA with the IAEA.

⁶ Nuclear weapon states can voluntarily offer to place selected nuclear facilities under IAEA safeguards via VOAs.

Safeguards procedures focus on ensuring peaceful uses of nuclear materials and technologies and detecting the diversion of significant quantities of nuclear materials in a timely manner. General tools used for safeguards include material accountancy (using various nuclear materials measurement techniques and equipment); containment and surveillance measures; process measurements, inspections, and tamper-indicating marking; monitoring (unattended and/or remote); and environmental sampling.

IAEA inspectors conduct various visits and inspections at facilities or other locations to verify a state’s compliance with the CSA or VOA obligations or to determine a state’s noncompliance. Noncompliance regarding the CSA/VOA obligations might involve diversion of nuclear material and undeclared nuclear material or activities. Examples of diversion of nuclear material include “(1) the undeclared removal of declared nuclear material from a safeguarded facility; (2) the use of a safeguarded facility for the introduction, production, or processing of undeclared nuclear material, e.g. the undeclared production of highly enriched uranium in an enrichment plant; and (3) the undeclared production of plutonium in a reactor through irradiation and subsequent removal of undeclared uranium targets” (IAEA, 2002). The IAEA has produced additional safeguards reports and recommendations specific to particular types of nuclear facilities or particular parts of the nuclear fuel cycle; see, for example, International Safeguards in Nuclear Facility Design and Construction (IAEA, 2013a), International Safeguards in the Design of Enrichment Plants (IAEA, 2019a), International Safeguards in the Design of Fuel Fabrication Plants (IAEA, 2017a), and International Safeguards in the Design of Reprocessing Plants (IAEA, 2019b). Box 6.1 contains definitions relevant for nuclear nonproliferation and safeguards. Box 6.2 includes a discussion of implementation of IAEA safeguards at U.S. advanced reactor and fuel cycle facilities.

6.2.3 Nuclear Security

The goal of security⁷ (or physical protection) is to prevent malevolent acts using radioactive or nuclear materials by substate actors. There is a distinction between the threats of radiological terrorism and nuclear terrorism. Radiological terrorism is an act that would lead to dispersal of radioactive materials, such as sabotage of a nuclear reactor, whereas nuclear terrorism is the theft of a nuclear weapon or the fissionable materials that could be used in making improvised nuclear explosive devices. In general, physical protection measures are designed to implement a strategy of “detect, delay, and respond” to prevent or mitigate a security threat. A key component of any physical protection system is the threat assessment—a determination of the magnitude and scope of the threats that a nuclear facility may face, including external physical attacks, insiders, and cyberattacks. This so-called design-basis threat (DBT) is then used to determine protective measures, including security forces, facility hardening, access controls, personnel vetting, and cybersecurity programs. The DBT is also typically graded to take into account the potential consequences of attacks, such as radiological sabotage or theft of fissionable materials with different levels of attractiveness for use in nuclear weapons.

This section describes the international and domestic (U.S.) frameworks for physical security, followed by a discussion of security measures for transporting fissionable and other radioactive materials. Transportation security is given distinct attention because the radioactive or nuclear material is mobile, outside any fixed facility protection system, and traveling on or through areas accessible to the public.

___________________

⁷ The discussion of security in this report focuses on physical security. Cybersecurity, also a key aspect of the overall security of a nuclear facility, will be covered by the parallel National Academies study, Laying the Foundation for New and Advanced Reactors in the United States.

6.2.3.1 International Physical Security Framework

The chief international instrument addressing nuclear security is the Convention on the Physical Protection of Nuclear Material (CPPNM) (IAEA, 1980) and its amendment (IAEA, 2005b). The CPPNM establishes legal obligations for protection of nuclear material used for peaceful purposes during international transport, and its amendment expanded the scope to include physical protection of nuclear facilities, nuclear material used for peaceful domestic purposes, storage, and transport. Unlike the NPT, however, there are no binding provisions for international inspections and enforcement of compliance with the CPPNM by national physical protection regimes.

The IAEA, through its Division of Nuclear Security, supports states in establishing an effective nuclear security system by assisting them with “implementation of relevant international legal instruments related to information protection; physical protection; material accounting and control; detection of and response to trafficking in such material; national response plans; and contingency measures. However, each state carries the full responsibility for nuclear security” (IAEA, 2002). To assist member states with this task, the IAEA has developed documents

and guidelines related to physical security of nuclear and radioactive materials during all phases of the nuclear fuel cycle (IAEA, 1975, 2011b, 2018b, 2021e). The CPPNM and its guidance documents address physical security for plutonium, uranium-233, and enriched uranium only, and not for radioactive materials (with the exception of spent fuel) as defined in IAEA (1980) (see also IAEA, 1975, 2011b, 2018b, 2021e). Physical security related to radioactive materials is also covered extensively by the IAEA for all operations associated with peaceful uses of nuclear energy (IAEA, 2011c, 2020a). In 2006, security publications published by the IAEA were organized under the Nuclear Security Series’ four sets of publications (IAEA, n.d.-b). Member states then use this guidance as a basis for establishing their regulatory structure to ensure secure management and operations of civil nuclear activities.

Incumbent in this work is a continual assessment of the adequacy of the guidance documents with respect to the evolution of international technologies, operations, and threats. The IAEA periodically provides up-to-date guidance on all aspects associated with physical security of nuclear and radioactive materials used for peaceful purposes, as reflected in both revised and new documents, to capture current conditions (IAEA, 2015c, 2016a,b, 2019c,d, 2020a,c, 2021a,b).

This comprehensive and continuing assessment is important with respect to advanced reactors and fuel cycles. One significant cross-cutting development is the proposed use of HALEU (with uranium-235 enrichments between 10 and 20 percent) as fuel for many advanced fuel designs. Physical security implications arise for the entire operational fuel cycle supporting HALEU-fueled power reactors (e.g., fabrication, use, storage, and transportation) because they will use Category II quantities of HALEU, which will require more stringent physical protection than Category III low-enriched uranium (see Sidebar 4.1 and Table 6.1 for the definition of the security categories). Because multiple countries have expressed an interest in HALEU-fueled reactors, the IAEA will need to the address physical security implications of using this fuel, including the potential for international transport of much larger quantities of HALEU in various forms than is currently the case. For example, as discussed in Chapter 4, domestic production of HALEU sources will likely not meet demand for the foreseeable future, and HALEU may need to be transported from other countries to the United States in hexafluoride, oxide, or metallic form (Griffith, 2021; Regalbuto, 2021).

The IAEA may need to develop specific guidance on physical protection for the production and use of Category II quantities of HALEU. Although a recent IAEA report that addresses LWRs using fuel enriched to over 5 percent states that current physical security measures can be accommodated for enrichments up to 10 percent, it does not address physical security impacts for reactors using HALEU fuels enriched to 10 percent or greater (IAEA, 2020d).

Physical protection requirements for advanced reactors and fuels will become clearer as designs evolve through the demonstration and prototype stages of development. As discussed in Chapter 5, advanced reactors and fuel cycles will create unique secondary wastes not associated with spent fuel, but these wastes are not expected to need physical protection controls beyond the current guidance provided for in existing IAEA documents. In December 2018, the IAEA convened a group of external experts to review the existing physical security documents in the context of advanced reactors. In parallel, the IAEA performed an internal review of these documents. The conclusion of both reviews was that “the current Nuclear Security Guidance remains valid and sufficient to address the known concerns for the protection of advanced reactors” (Larsen, 2021). This is not to say, however, that challenges and costs will not arise in characterizing these wastes to assess how they will fit into the existing security framework.

6.2.3.2 Domestic (U.S.) Physical Security Framework

Physical security is a vital component of domestic use of the commercial nuclear fuel cycle. The U.S. Nuclear Regulatory Commission’s (U.S. NRC’s) regulations governing physical protection, 10 CFR 73, “Physical Protection of Plants and Materials,” provides for comprehensive oversight of physical protection for special nuclear materials, as well as spent nuclear fuel. 10 CFR 73 is supported by numerous U.S. NRC regulatory guides and studies that inform the technical bases for the updated requirements for nuclear facilities and materials and provide acceptable approaches to implementing them.⁸

___________________

⁸10 CFR 73 does not address the security of such radiological materials as radioactive sources, other than irradiated fuel. 10 CFR 37 addresses the security requirements for other radioactive materials, which are far less stringent.

The U.S. NRC uses a nuclear material categorization system similar to that of the IAEA for grading the security requirements for protection against theft based on the attractiveness of the material for nuclear weapons. As shown in Table 6.1, 2 kg or more of the strategic special nuclear materials plutonium or uranium-233, or 5 kg or more of uranium-235 contained in highly enriched uranium, constitute Category I quantities of material, requiring the most stringent security measures. Security measures for Category I facilities include a requirement that licensees must protect the material from a specially designated DBT for theft that is more challenging than the DBT for radiological sabotage. The U.S. NRC does not require protection against the DBT for Category II or III facilities. There are currently only two U.S. NRC–licensed Category I facilities in the United States, where highly enriched uranium fuels are fabricated for U.S. Navy reactors, and one Category II facility, as of June 2021, the Centrus American Centrifuge Plant, where a HALEU enrichment demonstration is planned.

The U.S. NRC classification of special nuclear materials does not allow for a reduction in security requirements by crediting dilution with other materials, such as uranium-238, unless they emit high levels of external radiation; only contamination with radioactive isotopes to the extent that the material meets the “highly irradiated” standard would allow a reduction in the categorization of plutonium (see Table 6.1). However, the material classification system by the U.S. Department of Energy (DOE) does reduce the attractiveness level of plutonium or uranium-233 when diluted to 10 percent or less in a matrix that cannot be separated from the special nuclear materials with simple mechanical removal (DOE, 2019).

Since the 9/11 attacks, the U.S. NRC has engaged in comprehensive studies to assess the state of its physical protection requirements, has identified gaps, and has issued regulations and provided additional guidance to address the physical protection regime governing physical protection requirements in the United States (10 CFR

37; U.S. NRC, 2013a, 2014c). 10 CFR 37 addresses only Category 1 and 2 radioactive materials (which are outside the scope of this study),⁹ and, by definition, excludes radioactive material contained in any nuclear reactor fuel assembly, subassembly, fuel rod, or fuel pellet. Such material remains governed by the physical protection requirements under 10 CFR 73.

In 2006, the U.S. NRC began an effort to develop a rulemaking for enhanced security of special nuclear material. A 2015 regulatory basis document on the draft rulemaking provides detailed background on the development of U.S. NRC physical security bases (U.S. NRC, 2015). Among other changes, the rulemaking would have introduced a graded security approach similar to that of DOE, which would have given credit for dilution of special nuclear materials to less than 10 percent and would have included updated requirements for Category II facilities.

In 2018, the U.S. NRC Commissioners directed their staff to proceed with an expedited rulemaking exclusively focused on codifying the requirements of the post-9/11 security orders for nuclear materials, and not including the other aspects that the staff had proposed. This resulted in a policy issue paper, SECY-19-0095, which gave options from the staff to the U.S. NRC Commissioners, with recommendations from the staff to discontinue the rulemaking on the basis that a mere codification of the orders was not cost-justified (U.S. NRC, 2019a). The U.S. NRC Commissioners’ response in August 2021, however, disapproved the staff’s recommendation to discontinue the rulemaking activity (U.S. NRC, 2021g) and instead recommended that the staff provide the Commissioners, in the form of a notation vote paper, an expanded options analysis to cover potential regulatory, resource, and timing impacts associated with the identified options.

Recently, the U.S. NRC issued an informational sheet recommending assessment of additional security measures that may be needed based on current threat assessments (U.S. NRC, 2021h). The informational sheet specifically discusses HALEU fabrication facilities as Category II facilities and the potential need for supplemental physical security measures. These facilities will still be regulated under 10 CFR 73.67 but will be evaluated on a case-by-case basis to assess adequate levels of physical security protection. Implications of increased physical protection, while unknown at this time, may be significant from a cost and operational standpoint.

Coordination between DOE and the U.S. NRC on the appropriate level of protection at domestic HALEU production facilities is vital to maintain current schedules for deployment of prototype and demonstration advanced reactors. Similarly, coordination of the two agencies with international partners is also needed in order to understand the implications of using foreign-sourced HALEU to meet DOE advanced reactor needs.

As with the IAEA guidance discussed above, secondary wastes generated from advanced reactors are not expected to require enhanced physical protection measures beyond the requirements defined in the current U.S. NRC regulations; however, there may be additional costs and challenges associated with characterizing these wastes.

6.2.3.3 Transportation Security

Transportation is the most vulnerable phase in protecting nuclear materials from a security standpoint, as the material is removed from the confines of the nuclear facility. Regulatory, technical, and operational considerations are part of an overall strategy for providing secure transport of fresh and spent nuclear fuel. This is particularly important when transporting across international boundaries, as countries have different regulations based on their own level of risk acceptance. From an international nuclear security perspective, the IAEA provides umbrella guidance of how member states can comply with their CPPNM obligations regarding physical protection of nuclear and radiological materials in transport.

Spent fuel security requirements for U.S. domestic transportation are defined in 10 CFR 73.37, which provides detailed requirements that include development of a security plan with the full range of protective components needed to properly protect such shipments during transport. This regulation references the self-protection threshold for irradiated spent fuel of a minimum radiation level of 1 Gy (gray)/hr at 1 m in an unshielded state.

For fresh fuels, current LWR designs enriched to less than 10 percent in uranium-235 falls into Category III (special nuclear material of low strategic significance). Security aspects are addressed from the graded approach

___________________

⁹ Note that Arabic numerals for category numbers are assigned based on the activity content of radioactive materials, and Roman numeral category numbers are assigned based on the mass of fissionable nuclear materials. This study’s scope is relevant for the latter and not the former.

and are covered in 10 CFR 73, Regulatory Guide 5.59 (U.S. NRC, 1983), and Regulatory Issue Summary 2005-22 (U.S. NRC, 2005). A security plan is required for these shipments. Requirements for physical security focus mostly on administrative aspects: advance notification, receiver confirmation, container, inspection, and responsibility for in-transit physical inspection. Shipment of HALEU source material and HALEU fresh fuel will likely fall into Category II, requiring additional physical security protection measures. As discussed in Section 6.2.3.2, physical protection of these materials will be assessed on a case-by-case basis.

The U.S. Department of Transportation (DOT) also has a role in physical security transport. 10 CFR 71.5 includes U.S. NRC citations to the DOT regulations; specifically, it references DOT regulations under 49 CFR 172(I), covering security plans. However, 49 CFR 172.804 provides for comparable security plans developed under other federal agencies.

For commercial spent nuclear fuel, the Nuclear Waste Policy Act (NWPA) (Public Law 97-425) and its Amendments (Public Law 100-203, Part E), place the authority on DOE for managing commercial spent nuclear fuel for disposal and transport from the reactor sites to the disposal facility. The NWPA further stipulates that U.S. NRC licensing shall be used to the greatest extent possible. This means that DOE will use U.S. NRC–licensed casks and other regulatory protocols, such as security requirements, as mandated by the U.S. NRC. DOE’s Office of Nuclear Energy (DOE-NE) manages the program that is planning for transportation and disposal of commercial spent fuel. DOE-NE also sponsors and manages all the advanced reactor and fuel cycle programs that are being considered for the United States. Since these new concepts are for commercial power generation, DOE-NE is expected to use the U.S. NRC certification process in licensing, operation, and physical protection of these new reactors and fuel cycles.

DOE does have significant experience shipping spent fuel, nuclear waste, and nuclear materials under U.S. NRC physical security regulations. Under the Atomic Energy Act and its amendments, DOE has the authority to regulate the transport of its own materials (DOE, n.d.-a). In practice, DOE uses both self-regulation and the U.S. NRC regulations to operate its multiple energy and security programs. Within its weapons program responsibilities, DOE’s National Nuclear Security Administration (NNSA) self-regulates because of the important security issues and challenges involved, coupled with the highly classified nature of much of this work. DOE has its own set of review processes to address the nature of the materials that it ships. Under DOE, the Office of Secure Transport manages the safe and secure transport of government-owned special nuclear materials. Armed federal agents provide security and incident command response in the event of an emergency (DOE-NNSA, n.d.-a).

Other examples of DOE shipment programs include naval reactor fresh and spent fuel shipments (Miles, 2016) and DOE’s Office of Packaging and Transportation, located within its Office of Environmental Management (DOE-EM, n.d.). Naval reactors have dual certification for their Type B shipping containers, the M-140 and M-290 spent fuel casks.

6.2.4 Roles and Responsibilities of the U.S. Department of Energy and the National Nuclear Security Administration Relevant for This Study

The NNSA’s roles and activities in nonproliferation and arms control range from securing nuclear materials and controlling dual-use technology to obstructing adversary weapons development and managing the effects of a nuclear incident. Relevant for this report are the NNSA objectives of “prevent” and “counter”:

• “Prevent would-be proliferant states from developing nuclear weapons or acquiring weapons-usable nuclear material, equipment, technology, and expertise, and prevent non-state actors from acquiring nuclear and radioactive materials that can be used for malicious purposes”; and
• “Counter the efforts of both would-be proliferant states and non-state actors to acquire, develop, disseminate, deliver, or use the materials, expertise, or components of a nuclear or radiological device.” (DOE-NNSA, n.d.-b)

While most of the advanced reactor and fuel cycle facilities will be regulated by the U.S. NRC, those that are owned by DOE and located at DOE and NNSA sites may not be regulated by the U.S. NRC. DOE security directives will apply to those facilities.

Box 6.3 details the roles of specific offices within NNSA related to nonproliferation, safeguards, and security that are pertinent to this study’s scope.

Separately, in 2020 the DOE-NE established the Advanced Reactor Safeguards program (within the Advanced Reactor Demonstration Program), focused on assisting advanced reactor vendors with meeting domestic materials accountancy and physical protection requirements to “help reduce roadblocks in the deployment of new and advanced reactors” (Cipiti, 2021a).

However, it is also important for this DOE program to be aligned with the objectives of NNSA’s Office of Defense Nuclear Nonproliferation and Office of International Nuclear Safeguards, and be informed by NNSA expertise in nonproliferation and counterterrorism, especially regarding such relevant aspects as the expanded use of HALEU and the continued development of closed fuel cycles.

6.2.5 Nuclear Material Attractiveness

An important metric for assessing the relative nuclear proliferation and nuclear terrorism risks posed by different nuclear fuel cycles is the “attractiveness” of the various nuclear materials within the fuel cycle for different adversaries seeking to acquire nuclear weapons. The material attractiveness of a specific type of nuclear material can be defined as its “relative utility … for an adversary in constructing a nuclear device” (Bathke, 2021). It is an intrinsic property based on the physical, chemical, nuclear, and radiological characteristics of the material. Generally, more attractive materials require more stringent extrinsic material protection, control, and accountancy measures to mitigate their proliferation and terrorism risks than do less attractive materials.

Determining material attractiveness requires knowing “the time and potential difficulties” of (1) acquiring the material; (2) processing the material into a form suitable for use in a nuclear explosive device (NED); (3) fabricating the NED; and (4) utilizing the NED (Bathke, 2021). However, since these considerations depend on

the capabilities of the adversary, material attractiveness is not a fixed property of the material but will be different for different adversaries (Bathke, 2021). Its determination is further complicated by the fact that it also depends on the adversary’s objectives.

Nuclear materials that can be used directly in a nuclear device can be attractive targets. Highly enriched uranium, plutonium (excluding nearly pure plutonium-238), and uranium-233, which are defined as “strategic special nuclear materials” by the U.S. NRC and “direct-use materials” by the IAEA, are the materials of greatest concern.¹⁰ Although some are more readily usable than others, any nuclear material with a bare critical mass can theoretically be used in an NED and requires some level of safeguards and security protection (Bathke et al., 2012). The bare critical mass is a crucial parameter because it plays a fundamental role in determining the size and weight of a nuclear weapon, and it also sets the scale of the material acquisition effort for adversaries seeking to build a weapon.

Additional materials of concern include neptunium and americium. Although these are not defined under the IAEA statute as materials requiring safeguards, the IAEA considers them “alternative nuclear materials” and tracks them under voluntary agreements with relevant states (IAEA, 2002). In addition, other attractive isotopes, such as curium-245, are not currently designated as special or alternative nuclear materials (U.S. NRC, 2009). These materials are particularly relevant for proliferation and terrorism risks in those advanced reactor fuel cycles that separate them in quantities of significance from spent fuel, either individually or in mixtures (see Chapter 4).

In addition to the bare critical mass, other material characteristics considered important for attractiveness include the rate of decay heat generation, which affects the stability of the high explosive in a nuclear weapon, and the external dose rate, which affects the ability to safely handle, transport, and process the material.

A fourth material characteristic, the spontaneous neutron generation rate, could reduce the reliability of a nuclear explosive to achieve its design yield by increasing the likelihood of preinitiation before the necessary number of fissions occur. The degree to which these characteristics may impede their use in a nuclear explosive device will vary depending on the sophistication and technical resources available to the adversary. These parameters were incorporated by a DOE laboratory team into “figure of merit” metrics for comparing the material attractiveness of different isotopes and mixtures (Bathke et al., 2012).¹¹

Other considerations that could affect the attractiveness of nuclear material include its chemical form (for instance, metal versus oxide) and the degree to which it is diluted in a non–weapons-usable matrix from which it is not readily separable by mechanical means. However, chemical means of reducing attractiveness cannot render a nuclear material entirely useless for a weapon but can only increase the time an adversary would need to convert the material into a more usable form, providing delay that would give law enforcement authorities more time to locate stolen material. This principle is the basis for NNSA’s “dilute-and-dispose” approach for surplus weapons plutonium reviewed by a National Academies committee (NASEM, 2020), as well as the one of the bases of DOE’s graded safeguards approach (DOE, 2019).

One key standard is the dose rate that is considered “self-protecting”—that is, a high-enough level that makes it extremely dangerous for an adversary to prepare the material for malevolent use (Coates et al., 2005). Historically, a dose rate of 100 rem (equivalent dose) (1 Sv [Sieverts]) per hour at a 1-foot distance (unshielded) was the standard for denoting “highly radioactive” materials in guidance documents (the IAEA uses 1 meter rather than 1 foot, and the U.S. NRC now uses 3 feet)—a designation that reduces the material accountability and physical protection requirements compared with special nuclear materials that are unirradiated or have dose rates below the threshold. However, this value may not be high enough to be an effective deterrent to theft (U.S. NRC, 2015). Given the need for expanded and extended spent nuclear fuel storage, reconsideration of the self-protection standard may be appropriate (U.S. NRC, 2015).

The external gamma dose rate from spent nuclear fuel is dominated by 30-year half-life cesium-137 after shorter-lived fission products decay away. Thus, the self-protecting nature of spent fuel will eventually disappear

___________________

¹⁰ Due to a growing international consensus that the civil use of highly enriched uranium poses unacceptable security risks, such use has been nearly phased out over the past few decades.

¹¹ Following release of the prepublication version of this report, an erroneous sentence referring to the applicability of “figure of merit” metrics was removed.

over an extended period of time, depending on fuel burnup and other factors. The resulting consequences for the long-term security of irradiated materials in surface storage will depend on the residual attractiveness of the nuclear material after the radiation barrier falls below the self-protection threshold. A standard LWR uranium oxide spent fuel assembly with a burnup of 45 GWd/tHM (gigawatt-days per metric ton of heavy metal), which contains about 4–5 kg of plutonium (more than a Category I quantity), will remain self-protecting for well over a century after discharge using the 100 rem/hour (1 Sv/hr) standard, but for only less than 50 years using the more conservative approach based on a 1,000 rem/hour (10 Sv/hr) threshold described by Robel et al. (2013). However, even non-self-protecting uranium oxide spent fuel will be somewhat less attractive than separated plutonium because of the low plutonium concentration (approximately 1 weight percent) (Robel et al., 2013).

The committee highlights the following points concerning attractiveness of materials:

1. Nearly all isotopic mixtures of plutonium have some level of attractiveness for use in an NED; the only exception is mixtures containing 80 percent or greater of plutonium-238, which has a relatively high heat rate (NAS, 1994).
2. Given that minor actinides, such as certain neptunium, americium, and curium isotopes, may be weapon usable, reprocessing approaches such as pyroprocessing that separate plutonium in combination with one or more minor actinides do not necessarily render the product unattractive for use in an NED. The carryover of high levels of lanthanide fission products in the products of some reprocessing variants could increase the external dose rate and render the material less attractive for theft, although this would not be a significant deterrent for a proliferant state with chemical separation technology (Bari et al., 2009).
3. Thorium fuel cycles can produce attractive nuclear materials such as uranium-233. Although the admixture of uranium-232 reduces the attractiveness of separated uranium-233 through the ingrowth of the gamma-emitting decay product thallium-208, the figure of merit analysis finds that uranium-233 “is highly attractive at any practical concentration of ²³²U” (Bathke et al., 2012).
4. The attractiveness of nuclear materials can be reduced by diluting them to a low concentration with unattractive materials. For example, isotopic dilution of uranium-235 by uranium-238 to below the highly enriched uranium threshold of 20 percent uranium-235 can only be reversed through reenrichment. In contrast, dilution with a chemically dissimilar material, such as dilution of plutonium with uranium-238, can be reversed by chemical separation, although recovery will increase the time delay before an adversary can acquire the requisite quantity of weapon-usable material (U.S. NRC, 2015).

Notably, the attractiveness of the materials throughout a given nuclear fuel cycle is one of several factors that determines its relative proliferation and theft resistance. Fuel cycles containing more attractive materials, such as those involving reprocessing, require more stringent material protection, control, and accounting measures than those with less attractive materials, such as the once-through LWR fuel cycle. This is generally referred to as “graded” safeguards and security. Techniques and approaches needed to safeguard and secure separated highly attractive materials have technical, cost, and practical limitations that might constrain their implementation, compared with the simpler measures needed for the more proliferation- and theft-resistant once-through LWR cycle. However, it is difficult to make cost comparisons because the U.S. commercial nuclear industry (as opposed to DOE) has little experience with safeguards and security measures for materials that are more attractive than low-enriched uranium for LWRs.

6.3 EVALUATION OF NONPROLIFERATION IMPLICATIONS AND SECURITY RISKS OF ADVANCED REACTORS’ FUELS AND FUEL CYCLES

This section provides the committee’s evaluation of nonproliferation and security risks regarding the aspects identified in the statement of task (see Sidebar 1.1 in Chapter 1) for the fuel cycles of advanced reactor types examined in this study. In particular, this section assesses HALEU, uranium-plutonium mixed oxide fuel, and advanced fuel cycles that require separating plutonium and other fissionable materials from spent fuel. The section also provides the committee’s examination of nuclear material accounting and control, as well as containment,

surveillance, monitoring, and timeliness of detection related to the assessed advanced reactors’ fuels and fuel cycles. The baseline reference fuel cycle is the once-through low-enriched uranium oxide cycle (as defined in Chapter 2).

In responding to the statement of task, the committee used a simplified assessment that focused on certain characteristics of advanced fuel cycle technologies that differ from the baseline uranium oxide LWR once-through cycle and could have implications for nonproliferation and security, such as types and quantities of material, as well as impacts on material accountancy and safeguards. However, a comprehensive assessment (such as the studies of the Generation IV International Forum’s Proliferation Resistance and Physical Protection Working Group) was beyond the committee’s scope. Such an analysis would fully consider material attractiveness and proliferation resistance features, all routes to acquisition of nuclear weapon-usable materials, and the broader technical capabilities and political context in the countries where these fuel cycles may be deployed.

6.3.1 Once-Through Fast Reactors Using HALEU

This category includes the Natrium and ARC-100 reactors (by TerraPower and ARC Clean Technology, respectively), as well as the Oklo Aurora microreactor, all descendants to varying degrees of the Experimental Breeder Reactor (EBR)-II, although with some novel features. Factors that affect their proliferation risks compared with the LWR once-through cycle are the types and quantities of nuclear material in the fresh and spent fuels and the potential diversion and misuse pathways for obtaining weapon-usable material throughout the fuel cycle. These factors also affect the security risks, with additional considerations related to the attractiveness of the nuclear materials throughout the fuel cycle, such as the physical size, chemical composition, and external radiation barriers.

6.3.1.1 Natrium (Batch-Refueled)

The Natrium reactor initially uses a metallic fuel containing an alloy of 90 weight percent uranium and 10 weight percent zirconium. TerraPower is planning a staged approach for development of the technology. The 345-MWe (megawatts electric) demonstration plant will initially use Type 1 fuel with an average enrichment of 18.5 percent HALEU and an average discharge burnup of 59,000 MWd/THM (megawatt-days/tons of heavy metal) (Hejzlar, 2021; Neider, 2021; TerraPower, 2021a). The cycle length will be 12 months—somewhat shorter than is the current practice for LWRs, which are refueled every 18 to 24 months. However, as discussed in TerraPower’s 2021 presentation to the committee, by the mid-2030s, TerraPower aims to qualify and use an advanced Type 1B fuel with an average burnup of 150,000 MWd/THM and a cycle length of 18 months. In the longer term (2050 time frame), TerraPower hopes to develop even higher-burnup fuels and larger reactors that would allow reducing fuel enrichments to below 10 percent, and ultimately make possible a breed-and-burn operation, in which case the cycle length would be increased to 22 months. For the latter option (formerly known as the Traveling Wave Reactor and now called Natrium-U), proliferation risk would be expected to decrease after the reactor starts up with an initial core loading of HALEU, because subsequent fuel reloads would use only depleted uranium and thus not require uranium enrichment (Hejzlar, 2021; Neider, 2021; TerraPower, 2021a).

None of these systems are being designed as breeders, and as such do not contain natural or depleted uranium blanket fuel that would be irradiated only to low burnups and reprocessed to recover plutonium. In general, HALEU-fueled fast reactors have less favorable characteristics for breeding than plutonium-fueled reactors in any event (Kim et al., 1999).¹² In the longer term, once-through high-conversion reactors such as the Natrium-U may be able to significantly improve uranium utilization without the proliferation concerns and safeguards challenges of conventional closed-cycle fast breeder reactor systems.

The refueling frequencies for different Natrium reactor options are important to consider because, for reactors under international safeguards, inspectors must be present whenever the reactor is shut down for refueling. In this regard, the suite of Natrium reactor designs would not appear to impose an additional inspection resource burden on a per-unit basis compared with large LWRs. However, as with all smaller reactors, on a per-megawatt basis, the

___________________

¹² The Natrium-U would achieve high internal conversion by irradiating blankets to high burnups and utilizing them as driver fuel without reprocessing but would not actually be a breeder, despite often being called a breed-and-burn reactor.

overall inspection resources could be greater than those for larger reactors (IAEA, 2014). In addition, the different nature of the fuel and reactor design from that of LWRs may require changes to safeguards procedures, requiring new approaches and inspection requirements. The fresh fuel—the most attractive material on site—will contain HALEU, and therefore could require more frequent inspections if the IAEA were to decide that shorter timeliness goals were appropriate, as discussed in Section 6.3.4.5. On the other hand, Natrium (1B) and pressurized water reactor fuel assemblies each contain about the same quantity of uranium-235 (20 kg), so (unless the IAEA reduces the SQ value for HALEU) the inspection resources needed for item counting would be similar for both types of fuel.

All pool-type sodium-cooled fast reactors pose safeguards challenges. The design features a spent fuel storage area within the reactor vessel, where spent fuel discharges are first sent for cooling for up to 3 years before they are removed from the reactor vessel, cleaned of sodium, and either transferred to a water-filled spent fuel storage pool outside of the vessel or loaded into a dry canister and stored. Under LWR procedures, safeguards inspectors would be able to directly observe the refueling and verify the presence of all spent fuel declared to be in the in-vessel storage area during the required annual physical inventory. However, with sodium pool–type reactors, the fuel’s location within the vessel and the opacity of the sodium limit direct visual inspection, so new approaches will be required. TerraPower told the committee that the spent fuel assemblies will be “tagged and easily identifiable,” but it did not make clear how inspectors would be able to verify the tags (Hejzlar, 2021; Neider, 2021; TerraPower, 2021a). On the other hand, according to TerraPower, it will be difficult to access fuel stored in the reactor vessel, since it is a sealed system. Also, there will be a one assembly in, one assembly out restriction on the fuel handling machine, which will also have the capability to read and display the assembly tags that can be easily monitored. After the spent fuel is transferred to the water-filled pool and eventually to dry storage, the IAEA could use safeguards approaches it has experience implementing, such as the containment and surveillance measures it applied at the spent fuel pools at the Joyo and Monju sodium-cooled fast reactors in Japan, where both fresh and spent fuel assemblies were stored (Bays et al., 2021).

The security risks posed by the Natrium reactor will depend on the nuclear material content, size, weight, dilution, and other properties of fuel assemblies that would affect its vulnerability to theft. However, the fresh fuel will contain Category II quantities of HALEU and will therefore require appropriate security measures during transportation, receipt, and storage prior to irradiation.¹³ Depending on the reactor design and fueling strategy, the Natrium reactor’s spent fuel could contain significant plutonium and residual uranium-235. This would not initially pose a security concern because the radiation barrier will be substantial for many decades. However, the physically smaller and denser metallic fuel assemblies for the initial operation of Natrium-DEMO with sodium-bonded fuel may have lower dose rates at 1 meter than LWR spent fuels for the same burnup. Over time the increasing attractiveness of the spent fuel as the radiation barrier wanes could necessitate strengthened security measures. The higher burnup of Type 1B fuel may offset this to some extent.

6.3.1.2 Long-Life Cores Using HALEU

Both the ARC-100 and the Aurora designs use HALEU. One primary difference of these designs compared with the Natrium reactor is that they plan to use a single-batch core with a 20-year cycle length, instead of periodic refueling cycles. This will have both advantages and disadvantages for safeguards. According to the IAEA, “reduced core access and reduced refueling frequency makes misuse of the facility and diversion of spent fuel much more difficult” at reactors with sealed, long-life cores (IAEA, 2014). This advantage is also cited by advanced reactor developers (Sackett and Arthur, 2021). Since in principle the reactor vessel will not be opened over its lifetime, inspectors will not need to visit to observe refueling operations, as is the case for conventional, batch-refueled reactors. However, long-life reactor concepts “need to be reconciled with the traditional IAEA annual physical inventory of each reactor core, performed when access to the core is possible” (IAEA, 2014).

___________________

¹³ The Natrium Type 1B fuel is a uranium microalloy. Each Type 1B fuel assembly would contain about 21 kg of uranium-235 in HALEU (a Category II quantity of special nuclear material), and each Type 1B fuel reload would contain about 500 kg of uranium-235 in HALEU. Thus, according to current U.S. NRC guidance, as discussed in Section 6.3.4.2, the reactor would require enhanced Category II security measures for prevention of a “gross theft” of HALEU containing 75 kg of uranium-235 (the quantity in four Type 1B fuel assemblies).

Thus, there will be no opportunity following start-up for inspectors to detect anomalies in the core inventory that may indicate a diversion or substitution. Additional containment and surveillance measures, as well as reactor monitoring, may be necessary to compensate for the reduced opportunity for direct verification.

Other factors will need to be accounted for in designing a safeguards approach for long-life cores. For example, if a long-life core reactor has the potential to experience fuel failures that would require shutdowns and fuel replacements, a reactor owner would need to plan for such contingencies either by storing additional fresh fuel on site as a backup or arranging for new fuel deliveries from off site—both activities that would require additional inspections. Also, maintenance issues may arise during plant operation that could necessitate plant shutdown and vessel access. IAEA inspector presence may be required for such off-normal events, which could present challenges for small modular reactors and microreactors being considered for deployment in remote locations.

Security considerations will also differ for the smaller, long-life core designs. Despite their small size, these reactors will require substantial quantities of HALEU to achieve criticality. For example, based on a planned burnup of 1 percent, the 1.5-MWe Aurora will require several MT (metric tons) of HALEU assemblies with enrichments of up to 19.75 percent—well over a Category II quantity. Both the ARC-100 and the Aurora will be derated significantly in order to achieve long core lives without exceeding burnup limits.

The ARC-100 reactor fuel will attain an average discharge burnup of 76,800 MWd/THM over a 20-year period—comparable to the burnup of the Natrium Type 1B fuel after 5 years (Sackett and Arthur, 2021). This discharge burnup will provide significant self-protection for the spent fuel, although it will be slower to accumulate than for Natrium fuel.

In contrast, the fuel for the first Aurora unit will only achieve a burnup of 1 percent, or less than 10,000 MWd/THM after 20 years, although Oklo hopes to later achieve 2 percent (less than 20,000 MWd/THM) and eventually exceed 60,000 MWd/MTHM (DeWitte, 2021). The concern here is that, at least for the initial units, the dose rate of the fuel will remain relatively low both during the operating cycle and at discharge, and will fall off rapidly afterward.¹⁴

Depending on the dose rate during irradiation, the Aurora fuel may require Category II security not only before the reactor starts operation, but also at times during operation and after shutdown. Furthermore, as discussed in Section 6.3.4.2, the reactor could require an on-site security force to ensure prompt response measures should adversaries attempt “gross theft” of HALEU—especially given plans for deployment in remote locations where off-site local law enforcement response may be slow or insufficient. The plutonium in the Aurora spent fuel may also require an enhanced level of protection. While the need for a prompt security response would not be an issue for a unit deployed at the Idaho National Laboratory site, which has its own armed response force, such a requirement may limit broader deployment and would generally conflict with the company’s intention to not have an on-site security force for the Aurora and for each reactor to have facilities nearby open to the public (Oklo, 2020; Stamp, 2020).

6.3.1.3 Metallic Fuel Fabrication

Another aspect of the fuel cycle for the PRISM-type metal-fueled fast reactors relevant for safeguards is that current fuel fabrication methods generate a large quantity of unrecoverable scrap (nearly 30 percent) (Moore and Severynse, 2020). This waste stream could present challenges for material accountancy, depending on its form, how it will be stored, and the accuracy of the measurement techniques used to assay it.

The throughput of an industrial-scale fuel fabrication facility capable of supplying 1 GWe (gigawatt electric) for Natrium reactors would be on the order of 6.4 MT of HALEU per year (taking into account total scrap generation), and would therefore require Category II security, including measures to address the risk of “gross theft” of low-enriched uranium (see Section 6.3.4.2).

___________________

¹⁴ Although the committee did not receive spent fuel dose rate information about the Aurora, a technical analysis of a conceptually similar once-through long-life (30-year) small fast reactor utilizing metallic HALEU fuel, modeled after the 10-MWe Toshiba 4S, found that the dose rate at 1 meter of spent fuel with a burnup of 34,000 MWd/THM is about 350 rem/hr at 5 years after discharge (Frieß et al., 2015). If the spent fuel is discharged halfway through the cycle after attaining a burnup of 17,000 MWd/THM, the dose rate would be 200 rem/hr after 5 years (Frieß et al., 2015). Extrapolating from these data points, the Aurora spent fuel, at a burnup of less than 10,000 MWd/THM, would likely have a dose rate 5 years after discharge not much higher than current self-protection threshold of 100 rem/hr at 1 meter.

6.3.2 Pebble-Bed Reactors Using HALEU

Pebble-bed reactors using HALEU include the Xe-100 high-temperature gas-cooled reactor (HTGR) and the Kairos fluoride-cooled high-temperature reactor. The fuel for these reactors are graphite pebbles containing TRISO fuel particles, as discussed in Chapters 3 and 4. The fuel kernels for both reactor designs consist of UCO (uranium-carbon-oxygen), with equilibrium average uranium enrichments of 15.5 percent for the Xe-100 (Mulder, 2021) and 19.55 percent for Kairos (Blandford and Peterson, 2021). Pebble-bed reactors do not necessarily require the use of HALEU, but can also use LEU+—the German Modul design, upon which the Xe-100 reactor is based, as well as the Chinese HTR-PM reactor, both use uranium dioxide fuel with enrichments below 10 percent. The use of higher enrichments, in concert with the use of UCO instead of uranium dioxide, allows for higher discharge burnups to be achieved (Mulder, 2021). However, the use of HALEU will impact both international security and domestic material accounting and security requirements. The risks will be partly offset by the large number of items needed to acquire weapons-relevant quantities of material, as well as the lack of industrial maturity of methods for reprocessing TRISO fuel.

Each 80-MWe Xe-100 reactor will require 1,540 kg of HALEU for the initial core and 430 kg per full-power year afterward, or 6.2 MT and 1.72 MT per year, respectively, for a 320-MWe 4-pack (Mulder, 2021). The Kairos Power 140 MWe reactor core will require about 525 kg of HALEU for the initial core and 600 kg per year afterward (Blandford and Peterson, 2021). On a per-unit basis, these quantities are not large in terms of the relevant low-enriched uranium SQs, which are on the order of several hundred kilograms; but as discussed below, the total material inventory at a multiunit site, including fresh and spent fuel storage, can be substantial.

Pebble-bed reactors are refueled online, which is a major defining characteristic for the safeguards approach. Online-refueled (also called “on load-refueled”) reactors, such as CANDUs (Canadian deuterium uranium reactors), may require greater safeguards resources than batch-refueled reactors such as LWRs, because fuel is loaded and unloaded during operation, and not only during discrete shutdown periods when inspectors typically conduct a physical inventory.

Pebble-bed reactors present greater challenges than CANDUs for material accountancy because of the larger number of items containing fissionable material, the portability of these individual items, and the nearly continuous fueling and refueling cycles (Kovacic et al., 2021). A national laboratory team proposed that pebble-bed reactors should not be considered online refueled reactors with regard to their safeguards characteristics but should be placed in a new category of “bulk-fuel” reactors (Durst et al., 2009).¹⁵

This material accountancy challenge is partially offset by the large number of items that would need to be diverted to acquire a significant quantity of special nuclear material. For example, each Xe-100 fresh fuel pebble contains 7 g of HALEU, so approximately 69,000 pebbles would need to be diverted to acquire 1 SQ of low-enriched uranium (485 kg)—which is 30 percent of the core, or more than 1 year’s throughput of a single Xe-100 reactor module.¹⁶ Also, since each spent pebble would contain about 0.13 g of plutonium at peak burnup (Mulder, 2021), diversion and processing of a comparable number of pebbles—61,500—would be needed to acquire 1 SQ of plutonium. The large number of pebbles that would be have to be removed all but rules out the possibility that abrupt diversions of 1 SQ of either low-enriched uranium or plutonium from an operating module could be accomplished without detection.

Nevertheless, the potential for undetected protracted diversion or misuse cannot be excluded if the uncertainties of the material accountancy system are large enough and if all nuclear material on site is considered. In addition to each reactor module, inspectors will need to verify the inventories of fresh fuel and spent fuel storage areas. Each Xe-100 core (which contains approximately 224,000 pebbles when fully fueled) is fully replaced approximately every 3.5 years, so that a nuclear plant containing four Reactor Modules (the Xe-100 standard design to produce

___________________

¹⁵ For comparison, a typical CANDU reactor contains 380 distinct fuel channels, with 12 fuel bundles per channel, for a total of 4,560, and loads about 17 bundles per day. In contrast, each Xe-100 module contains some 220,000 fuel pebbles and will load and circulate over 40 pebbles per hour. A model FHR that serves as the basis for the Kairos FHR will circulate 450 fuel and graphite moderator pebbles per hour, or one every eight seconds (Disser et al., 2016).

¹⁶ Here the committee assumes that the IAEA SQ will remain 75 kg of U-235 for HALEU; if it is decreased, as discussed in Section 6.3.4.5, this would reduce the number of fresh pebbles needed to obtain 1 SQ, as well as increase the nondetection probabilities for diversions of 1 SQ.

320 MWe) will receive 10 million fresh fuel pebbles over a 40-year plant lifetime (plus replacements for damaged pebbles) and have the same number of spent fuel pebbles to store and ultimately dispose of at the end of plant life. In that context, the likelihood of timely detection of a diversion of 1 SQ may be lower than one might surmise based on the large number of items needed, and will decrease over time as spent fuel pebbles accumulate. Currently, there is no plausible way for pebbles to maintain a unique identifier over their operational life, although technical approaches have been proposed (Gitau and Charlton, 2012).

Because of the near-continuous fueling and refueling cycle, as well as other material flows in the fuel handling system, the material control and accounting system for pebble-bed reactors is envisioned as a hybrid of approaches used for reactors, which use item accounting, and those for bulk-handling facilities, such as enrichment plants, where inspectors monitor in-process nuclear material flows (Kovacic et al., 2020). Inspectors will not be able to conduct physical inventory verification of the reactor core under normal circumstances. For such reactors, the IAEA requires “dual C/S”—that is, two independent containment and surveillance (C/S) measures for each plausible diversion path (Durst et al., 2009). However, even dual C/S does not fully compensate for the inability to conduct periodic physical inventory verification, as failure of a C/S measure would cause loss of continuity of knowledge that could only be rectified by a timely physical inventory of the entire site—which may not be possible for a pebble-bed plant. Moreover, online refueled reactors require significantly greater inspector resources for application of C/S measures than batch-refueled reactors (Boyer, 2021).

For safeguards, flow monitors will be needed to count pebbles at various transfer points and to distinguish between fresh fuel, graphite moderator pebbles, and irradiated pebbles at various burnups, including spent pebbles (IAEA, 2014). The error estimates of the counting rate for current systems range from 5 percent to as low as 0.1 percent per instrument (Kovacic et al., 2020). It will be difficult to assess the effectiveness of these systems without knowing their uncertainties more precisely, and further research and development is likely required.

In addition, pebble counting alone will be insufficient to accurately determine and verify nuclear material inventories at pebble-bed reactors. The Xe-100 pebble-bed reactors use online burnup measurement systems to estimate the burnups of pebbles discharged from the core, in order to determine whether they should be reloaded into the core (e.g., partially spent) or discarded to temporary dry storage (e.g., fully spent). Such systems use gamma spectroscopy to measure concentrations of fission products such as cesium-137 as a surrogate for burnup. However, although such indirect measurements may be sufficient to determine the number of times individual pebbles have passed through the reactor, they do not accurately measure the fissionable material content of each pebble, and would be of limited use for safeguards. First, there are uncertainties in the burnup determination itself, which one study estimated as in the range of 5–10 percent (Hawari and Chen, 2005). Second, there are uncertainties of similar or greater magnitude in the depletion codes used to estimate radionuclide concentrations as a function of burnup for pebble-bed reactors, as demonstrated by a recent benchmark exercise (NEA-OECD, 2019).

Even with uncertainties on the order of 5 percent, a preliminary safeguards study of a model fluoride-cooled high-temperature reactor (FHR) similar to Kairos found that diversion of 1 SQ of low-enriched uranium from a single module would be readily detectable (Disser et al., 2016). However, the study also concluded that “the overall system non-detection probability may be called into question” for a multimodule site (12 units in the case of the model FHR) if the facility had common fresh and spent fuel storage facilities (Disser et al., 2016). The challenges of detecting a protracted diversion of one SQ would be compounded in a state with numerous multimodule sites (Gitau and Charlton, 2012).

Overall, it appears that the development of an effective safeguards approach remains a work in progress. Careful attention to integrated safeguards and security by design principles could certainly help, but fundamental challenges are likely to remain because of the inherent measurement difficulties in pebble-bed systems. However, as stated above, the large quantities of fresh or spent fuel that would have to be diverted to acquire 1 SQ are mitigating factors that would have to be considered in assessing the proliferation risks posed by pebble-bed reactors.

6.3.2.1 Proliferation-Resistance Features

Despite these challenges for material accountancy, X-energy claims that their Xe-100 HTGR is “extremely proliferation-resistant” (Mulder, 2021). X-energy told the committee that the fuel cycle renders the plutonium

isotopic mixture “useless for nuclear proliferation” because of the high in situ utilization of the plutonium-239 and -241 fissile isotopes and respective destruction/buildup of even-numbered (fertile) isotopes (e.g., plutonium-240 and -242). However, as discussed in Section 6.2.5 (Material Attractiveness), all isotopic mixtures of plutonium, except those with 80 percent or more plutonium-238, may be attractive to some degree for nuclear weapons.

A second argument is that TRISO fuel is inherently proliferation resistant because it is difficult to reprocess and there is little experience with the chemical techniques that would be required to do so, in comparison with highly mature LWR fuel reprocessing. X-energy makes the point that “TRISO particles are very tough to crack mechanically” (Mulder, 2021). The committee heard a somewhat different perspective from a BWXT representative, who said that recycling TRISO fuel is not as simple as recycling LWR fuel, though it can be done in a lab, and the challenge is to do so on an industrial scale (Lommers, 2021).

The industry is working to overcome the challenge of industrial-scale TRISO particle recycling, at least for fresh fuel. X-energy told the committee that the company will need to recover HALEU from scrap generated at its planned TRISO-X fuel fabrication plant, and accordingly, is developing methods to extract uranium from off-specification TRISO particles in order to minimize waste and to reuse the uranium (Pappano, 2021). Thus, one may expect that if a state has the capability to produce its own TRISO fuel economically, it will also have the means to extract uranium from TRISO particles—at least before irradiation. The current lack of industrial experience in reprocessing TRISO irradiated fuel particles does not imply that a determined state could not develop such capability in the future to obtain fissionable material for use in weapons.

Moreover, as discussed in Chapter 5, the large volume of contaminated graphite in spent pebble-bed fuel makes managing the waste form cumbersome and may pose problems for geologic disposal. If so, there may be an incentive to develop industrial-scale methods for separating the graphite matrix from the TRISO particles, and possibly for reprocessing the particles themselves. This would further diminish in the long-term any proliferation-resistance advantage the fuel may have today.

6.3.2.2 Security Considerations

Evaluating the security risks posed by pebble-bed reactors will require taking into account factors that affect the vulnerability of materials within the reactor and fuel cycle to theft and acquisition of sufficient material for a nuclear weapon. The primary factor is the presence of HALEU in the fresh fuel, which will require that pebble-bed reactors meet Category II security requirements, adjusted for the current threat environment as discussed in Section 6.2.3.2. In addition, the dose rate of individual pebbles is below the regulatory self-protection standard of 100 rem/hour at 1 meter and therefore may require the higher self-protection standard of 500 rem/hour value, as proposed by Chung et al. (2012).

The very small amount of nuclear material in both fresh and spent pebbles would make abrupt theft of a single Category II quantity of material challenging, not to mention a quantity sufficiently large to constitute a gross theft containing 75 kg of ²³⁵U (see Section 6.3.4.2). At 7 grams of total uranium per pebble, more than 9,000 fresh pebbles would need to be stolen to obtain a Category II quantity of HALEU (64.5 kg of 15.5 percent–enriched uranium), and over seven times that number for a 485-kg “gross quantity” of HALEU. For spent fuel pebbles, which would no longer contain HALEU at the peak projected burnup but would contain 0.13 g of plutonium each (Bays et al., 2021), about 3,850 pebbles would be required for a Category II quantity of plutonium and about 15,400 for a Category I quantity.

How many transport or storage containers would an adversary need to steal to acquire these quantities of pebbles? The Versa-Pac, which is being considered for fresh fuel, is a 55-gallon drum which could hold about 350 pebbles (Kovacic et al., 2021). Over 25 Versa-Pacs would need to be stolen to obtain a Category II quantity of HALEU, and nearly 200 for a gross quantity. These quantities represent from about 4 to 14 percent of the approximately 700 containers of fresh fuel a 4-pack Xe-100 plant would require annually, making theft an unlikely scenario.

For spent fuel, the AVR-TLK container, currently in use to store fuel from the two shut-down German pebble-bed reactors, has a capacity of 950 pebbles (Bays et al., 2021). Relatively low numbers of containers would be needed to obtain a Category II quantity (5) or a Category I quantity (17) of plutonium. However, other factors

would need to be considered, including the external dose rate and overall weight of a fully loaded container. In addition, if the U.S. NRC decides to take dilution into account in revising its physical protection requirements (see Section 6.2.3.2), then the spent pebbles would likely be considered a highly diluted waste form that would require less intensive security than more concentrated items.

6.3.3 Once-Through Molten Salt–Fueled Reactors Using Low-Enriched Uranium

The material accountancy challenges discussed above for pebble-bed reactors will be even greater for molten salt–fueled reactors. While the fuel in a pebble-bed reactor may be difficult to track and count, it does consist of discrete items. In contrast, the special nuclear material in a salt-fueled reactor constantly flows through and outside of the core (Scott et al., 2021). For the purposes of safeguards, molten salt–fueled reactors can be considered bulk-handling facilities similar to reprocessing plants. This is true even for designs that are nominally once-through—that is, they do not plan to use online chemical separation facilities to extract fissionable materials or neutron-absorbing fission products from the fuel, nor to (at least initially) reprocess and recycle the spent fuel. Such reactors include the Terrestrial Energy IMSR (integral molten salt reactor) and ThorCon thermal-spectrum designs, and the fast-spectrum MCFR (molten chloride fast reactor).

Currently, it is difficult to conduct accurate and timely material accountancy at bulk-handling facilities, especially for in-process nuclear materials. The total in-process inventory cannot be measured directly during operation, but only extrapolated through such means as sampling and destructive assay, nondestructive assay, and process monitoring. The very large throughput of special nuclear material of an industrial-scale bulk-handling facility, coupled with technical limits on the accuracy and precision of measurement techniques, can lead to a large level of material unaccounted for (see Section 6.2.2), consisting of many significant quantities of special nuclear material. Even when the facilities are cleaned out and physical inventories are taken, the accumulation of difficult-to-measure residual holdup can make a significant contribution to the material unaccounted for. As a result, the IAEA cannot rely on material accountancy alone to meet its safeguards goals at bulk-handling plants and must also use containment and surveillance (C/S) and other methods.

Material accountancy at a salt-fueled reactor will be even more complex because, unlike fuel cycle facilities, the nuclear material inventory within the reactor changes with time. Because the depletion codes for salt-fueled reactors are likely to have large uncertainties, it may not be possible to precisely estimate the reactor inventory as a function of time, even if inputs and outputs are accurately measured. This poses challenges for material accountancy approaches that do not attempt to measure the nuclear material within the reactor, but rely only on C/S (Hogue et al., 2021). Moreover, it has been pointed out that it would be difficult to use C/S measures to cover the entire primary flow loop (Shoman and Higgins, 2021; Soares et al., 2020).

Consequently, a second approach that would use measurements of nuclear materials in salt-fueled reactors is being pursued, taking advantage of the systems that reactor operators would use to monitor salt composition for process control. These process monitoring approaches could include destructive and nondestructive assay of salt samples and in situ measurements of process streams (Hogue et al., 2021). Oak Ridge National Laboratory researchers are working to identify signatures that could provide indications of a diversion, as well as developing measurement techniques and instrumentation for detecting those signatures (Dion et al., 2020). However, these approaches are still immature and will require research and development under representative conditions by working with actual irradiated fuel in a reactor environment. The extreme environmental conditions within salt-fueled reactors will make “the ability to monitor and perform measurements during operation of an MSR … a severe technical challenge” (Dion et al., 2020). No research facilities are currently conducting experiments on irradiated salts in the United States, although China has recently begun to operate a small molten salt test reactor (Mallapaty, 2021; Scott et al., 2021; WNN, 2022b).¹⁷

A further complication is that, as for pebble-bed reactors, there are no scheduled periodic outages for refueling that would provide opportunities for safeguards inspectors to conduct physical inventory verifications. However, because of the corrosive nature of molten salts, periodic shutdowns for maintenance will likely be required, which

___________________

¹⁷ This sentence was updated after release of a prepublication version of the report to reflect China’s recent startup of this reactor.

would provide opportunities for a cleanout and inventory, although not necessarily at the frequency that inspectors would need (Hogue et al., 2021).

Establishing an effective material accountancy approach in salt-fueled reactors will be a complicated endeavor, requiring measurement of the types and quantities of nuclear materials in the fresh fuel, irradiated fuel, process stream, and waste streams. These measurement technologies have not yet been fully developed and demonstrated. An IAEA official said that molten salt-fueled reactors “raise the bar for safeguards” because there is much more of an opportunity for material to be diverted, so stronger and more elaborate extrinsic measures will be required (Boyer, 2021).

6.3.3.1 Terrestrial Energy IMSR

The Terrestrial Energy IMSR will use low-enriched uranium, with an enrichment below 5 percent; it is the only advanced non-LWR design being supported by DOE that does not use HALEU. The use of low-enriched uranium reduces the proliferation risk associated with fresh fuel production and storage relative to HALEU-fueled designs. The greater concerns are the production of plutonium in the reactor and the challenges in accounting for the material. One problematic characteristic of once-through, low-enriched uranium, salt-fueled reactors is the steady increase of the core inventory of plutonium over an operating cycle (Higgins et al., 2021). Over time, for a fixed overall measurement error, the uncertainty in material unaccounted for will increase, and the likelihood of detecting a diversion of plutonium will decrease to an unacceptable level, highlighting the need to develop improved safeguards approaches (Higgins et al., 2021) (see Figure 6.1).

6.3.3.2 ThorCon and MCFR

According to the initial design of ThorCon’s proposed reactor, as presented to the committee, each 500-MWe ThorCon plant, consisting of two reactor modules, uses an average of 1.93 MT of HALEU enriched at 19.7 percent and 3.3 MT of thorium annually to generate uranium-233 in situ (Jorgensen, 2021).¹⁸ (About one-third of the total amount of HALEU would be required for the initial core, and the remainder would be used for continuous fueling over an 8-year operating cycle.) Unlike molten salt, in uranium-233 breeders, such as that proposed by Flibe, there is no plan to separate protactinium-233 to maximize uranium-233 production, a process that would produce a separated stream of uranium-233, a direct-use material comparable to plutonium in its attractiveness for weapons. Consequently, the once-through ThorCon design poses lower-order proliferation concerns than thorium molten salt breeders. Nevertheless, the design will require a greater level of safeguards than LWRs because of the large quantity of HALEU that will be required for continuous fueling, as well as the ingrowth of both uranium-233 and, to a lesser extent, plutonium.¹⁹ Also, although the initial design will be once-through, ThorCon might use reprocessing in the future to recover residual uranium for recycle (Jorgensen, 2021). Additional safeguards will be required in a possible future in which uranium in the spent fuel is extracted and possibly reenriched. However, this would not be done at the power plant site but rather at a separate safeguarded reprocessing site.

ThorCon currently plans to install its reactors on barges and deploy them by sea to countries such as Indonesia. For such cases of transportable nuclear power plants, early provision of design information to the IAEA is critical (IAEA, 2013b). Since each plant would presumably include sufficient fuel for its entire operating lifetime, about 15.5 MT of HALEU would have to be shipped with the plant, or more than 40 SQs (again assuming the current value of 75 kg of contained uranium-235). This substantial inventory, coupled with the storage and continuous feeding of the fresh fuel, could make the application of safeguards challenging at such a facility.

___________________

¹⁸ In the event that HALEU is not available, the reactor can start with 4.75 percent–enriched uranium (Jorgensen, 2021a).

¹⁹ The committee notes that during fact checking, in June 2022, ThorCon stated that its design has changed subsequent to its presentation to the committee in January 2021. Specifically, due to concerns about HALEU availability from Russia, ThorCon provided information to the committee in spring 2022 stating that it has altered its plans to use low-enriched uranium and to refuel once a year. Because of these plans, there will also be no use of thorium. The committee was not able to review the new design information after the closure of public information gathering, and the chapter’s text is based on the previous information that the committee had time to review during public information gathering.

ThorCon’s spent fuel is likely to present less of a proliferation risk than its fresh fuel. Plutonium production in its reactor will be lower than in a reactor fueled only with uranium because some uranium-238 is displaced by thorium, although this will be offset to some extent by production of uranium-233. The total quantity of uranium-233 in the spent fuel of a 500-MWe plant after 8 years would be about 400 kg, or 50 SQs, compared with about 150 kg of plutonium, or 19 SQs (Jorgensen, 2021). However, the uranium-233 would be mixed with about 10.6 MT of uranium-238, making it comparable to low-enriched uranium in its unattractiveness for weapons.²⁰

The TerraPower MCFR will be initially fueled with 12 percent–enriched HALEU, and in principle, it will only require additions of depleted or natural uranium over its operating life, as its purpose is to be a high-conversion reactor (Latkowski, 2021). This would mitigate the proliferation risks to some extent compared with salt-fueled reactors requiring continuous additions of HALEU or other fissionable materials. However, the reactor would present similar challenges for material accountancy as other uranium salt-fueled reactors, since the core plutonium inventory is likely to increase to a substantial level (on the order of metric tons) as the reactor converts depleted uranium to plutonium. The committee did not receive sufficient information about the fuel cycle for this reactor to perform a detailed assessment.

6.3.4 Nonproliferation and Security Implications of the Use of HALEU

Nearly all of the advanced reactor designs currently being supported by DOE and private entities would require HALEU in the enrichment range of 13–19.75 percent, but the fuel cycle supporting commercial nuclear power production today is not designed or optimized to accommodate reactors requiring HALEU fuel. Annual HALEU needs for a single reactor could range from hundreds to thousands of kilograms, depending on the design and power

___________________

²⁰ There is no definition for low-enriched uranium-233 comparable with low-enriched uranium-235 in IAEA safeguards, although a proposal has been made for how to define such a quantity (Forsberg et al., 1998). However, in such a determination, the uranium-235 content would also be counted. Using the formula in Forsberg et al. (1998), the mixture would be classified as low-enriched uranium.

rating. If such reactors were to be deployed on a significant scale, the entire nuclear fuel cycle—from uranium enrichment and conversion to fuel fabrication, transport, and storage—would need to be reconfigured for the different safety and security attributes of HALEU. This is no small task. In particular, it would require security upgrades throughout the fuel cycle, because HALEU (in sufficient quantity) is classified as a security Category II material in both domestic physical protection regulations and international conventions (see Sidebar 4.1 for definitions).

Key questions include (1) whether the large-scale deployment of HALEU-fueled reactors and associated fuel cycle facilities around the world could lead to a significant increase in nuclear proliferation and nuclear terrorism risks compared with the current LWR-dominated fleet; (2) what measures would be needed to adequately mitigate those risks; and (3) what impacts would such measures have on cost, plant operations, and other factors.

It is also important to distinguish between the additional risks that the use of HALEU could pose for nuclear proliferation by states and those for nuclear terrorism by substate actors. This distinction requires understanding the extent to which access to HALEU would be more advantageous than access to lower-enriched grades of uranium for these different types of adversaries.

An adversary can use HALEU to produce an NED by three pathways: (1) direct use of HALEU in an NED; (2) further enrichment of HALEU to highly enriched uranium; and (3) use of HALEU fuel in a reactor to produce plutonium (or uranium-233, for fuel cycles using thorium-232). The latter two routes are primarily concerns for proliferation by states, since any material acquisition path that would involve further enrichment or irradiation and reprocessing is generally considered highly implausible, if not impossible, for substate actors to accomplish on their own (U.S. NRC, 2015). (However, substate groups working in concert with adversary states may have access to such capabilities.) Thus, the committee views that the only plausible option for substate actors working alone is to use stolen HALEU directly in an NED.

However, the historical motivation remains elusive for designating uranium enriched between 10 and 20 percent as Category II material, requiring more stringent physical protection than Category III uranium enriched to less than 10 percent. The initial 1975 version of the IAEA’s physical protection recommendations, INFCIRC/225, introduced the designation, and it was subsequently adopted by the U.S. NRC in 1979. But the U.S. NRC’s stated rationale was not concern for the direct use of the material by terrorists, but the possibility that “quantities of uranium enriched to less than 20 percent in the U-235 isotope could be diverted, without timely detection, to other countries for additional enrichment or for plutonium production” (U.S. NRC, 1978).

Thus, in 1979 the U.S. NRC presumed that low-enriched uranium would be attractive to a subnational adversary only if it were colluding with a state with enrichment or plutonium production capabilities. Consequently, the higher attractiveness of Category II enriched uranium would be due solely to the fact that less separative work would be required to enrich Category II uranium than Category III uranium to highly enriched uranium levels. However, since 1979, enrichment plants have become more compact and efficient as gas centrifuges supplanted gaseous diffusion. It is not clear that access to HALEU would still provide a major advantage relative to Category III low-enriched uranium for nations with access to modern gas centrifuge enrichment technology. Given this, with respect to the ease of further enrichment, the 10 percent lower bound would appear less important today (Boyer, 2021).

Even so, the U.S. NRC recently reaffirmed that Category II HALEU is more attractive and requires greater protection than Category III enriched uranium (U.S. NRC, 2019b, 2021f). (The complete U.S. NRC analysis is classified.)

6.3.4.1 Practicality of HALEU as a Weapons Material

Since HALEU has a uranium-235 enrichment below 20 percent and therefore is considered low-enriched uranium, the IAEA classifies it as indirect-use material, unlike the direct-use materials: highly enriched uranium, plutonium, and uranium-233. However, this classification does not mean it is impossible to use HALEU directly in an NED. In fact, any nuclear material with a finite bare critical mass can be used, in theory, to make a nuclear explosive device; however, less attractive materials present greater technical challenges (CGSR, 2000). The main question with regard to the nuclear terrorist threat posed by HALEU is whether a given quantity can be used by

a subnational group to build a sufficiently practical and deliverable NED to achieve the group’s desired nuclear yield and reliability. This depends on the technical sophistication of the group in question, as well as its objectives.

In material attractiveness studies conducted by researchers at Los Alamos National Laboratory, Lawrence Livermore National Laboratory, and Pacific Northwest National Laboratory, HALEU is assigned a figure of merit between 0 and 1 and is characterized variously as “unattractive” (but “may still be theoretically possible to build a nuclear weapon or explosive device”) (Bathke et al., 2012), or “impractical, but not impossible” (Ebbinghaus et al., 2013). A 2005 National Research Council study stated that “nuclear explosives can in principle be made with material containing somewhat less than 20 percent ²³⁵U, but the amount of material required at enrichments below 20 percent is very large” (National Research Council, 2005). In comparison, highly enriched uranium with enrichments above 20 percent and below about 75 percent are described in the figure of merit studies as “attractive,” and enrichments above about 75 percent are characterized as “preferred” (using bare critical masses calculated in Glaser [2006]).

A primary difference between uranium of different enrichments is the critical mass—that is, minimum quantity of material that can provide for a self-sustaining nuclear chain reaction for a given configuration. A solid sphere of uranium-235 metal has a “bare” critical mass (the critical mass in the absence of a neutron reflector material) of about 48 kg. For HALEU with an enrichment of 19.75 percent, the bare critical mass is about 780 kg—16 times greater than that of uranium-235 (Glaser, 2006). In a comprehensive review of nuclear proliferation in 1977, the Office of Technology Assessment (OTA) noted that a minimum enrichment of 20 percent for a “practical nuclear explosive” had been specified many years previously, and that any fissionable mixture with a bare critical mass greater than about 850 kg “could not be used to construct a nuclear explosive of any practical weight” (OTA, 1977), which would correspond to a uranium enrichment of about 17 percent (Glaser, 2006). This suggests that material in the higher HALEU enrichment range could be used in a nuclear explosive of practical weight, by the OTA standard at the time.

Little public information is available about the exact level of enrichment below which an impractical nuclear device would be precluded. The Atomic Energy Commission’s “Hafstad Memorandum” of 1954 states that “the minimum enrichment which is capable of supporting a nuclear explosion with an infinite mass of material has been estimated as about 5 percent. Information from Los Alamos indicates that 10 percent enriched uranium is not suitable for any practical weapon but no definite upper limit can be set” (Brown and Glaser, 2016).

It is likely that changes in the threat environment and the advancement of technology in the decades since these statements were made have increased the cause for concern today and are driving the need for enhanced security and safeguards requirements.

6.3.4.2 The Need for Updated HALEU Security Requirements

The U.S. NRC’s security regulations for Category II materials (including HALEU containing at least 10 kg of uranium-235) in 10 CFR 73.67 were originally introduced in 1979 and have not been updated. These requirements provide for a lower level of protection than the regulations for Category I facilities.

In particular, facilities possessing a Category I quantity of highly enriched uranium or other strategic special nuclear materials must be able to protect the material from theft by a DBT adversary—a violent attack by a sophisticated paramilitary group. This requirement drives the need for such facilities to have a dedicated and well-trained armed response force capable of neutralizing the DBT adversary. In contrast, the U.S. NRC does not currently require that facilities possessing Category II materials protect them from theft by a DBT adversary, nor does it generally require that members of the security organization at a Category II facility be “uniformed or armed with a firearm,” nor have any “formal or comprehensive training” (U.S. NRC, 1983). However, per the U.S. NRC’s Category II regulatory guidance, the agency expects the licensee must demonstrate that each security person understands the particular duties assigned to him/her and is fully qualified and trained to perform them (U.S. NRC, 1983). The function of the on-site security organization is limited to “early detection and assessment”

of security incidents, and communication (if warranted) to the U.S. NRC and local law enforcement authorities (LLEA), who would execute the response (i.e., prompt recovery of the material) (U.S. NRC, 1983).²¹

The U.S. NRC’s current regulatory guidance for protecting HALEU, which dates to 1983 (U.S. NRC, 1983), identifies “gross theft” as a scenario with more serious potential consequences to public health and safety than a “minor theft” of low-enriched uranium (U.S. NRC, 1983). The U.S. NRC defines “gross theft of LEU” as “theft in a sufficiently large quantity that it could yield upon further enrichment or other processing enough material … to construct a clandestine fission explosive device” (U.S. NRC, 1983). The guidance estimates such a quantity as containing about 75 kg of uranium-235—the same value as the IAEA “significant quantity” of low-enriched uranium. According to the guidance, the U.S. NRC interprets “early detection” of a gross theft of low-enriched uranium as “detection during the attempted theft” (U.S. NRC, 1983).

The U.S. NRC has determined that these Category II security requirements need updating. One reason is because after the 9/11 attacks, the U.S. NRC required security upgrades via orders to licensed Category I and Category III facilities to address the new threat environment but was unable to do so for Category II facilities since there were none licensed at the time. Another is that the risks are “better understood than when the existing regulations were promulgated” (U.S. NRC, 2021h). Consequently, the U.S. NRC recently stated that “supplemental security measures for the protection of Category II quantities of SNM [special nuclear materials] may be required to address the current threat environment and the changing understanding of the risks associated with facilities possessing Category II quantities of SRM” (U.S. NRC, 2021h). This view is also shared by some industry stakeholders. According to a 2021 assessment of security requirements for Category II fuel cycle facilities by a former U.S. NRC official and two X-energy personnel, the current rules “do not identify appropriate or adequate security for HALEU that would be necessary today” (Rivers et al., 2021).

As discussed in Section 6.2.3.2, in 2021, the U.S. NRC commissioners asked its staff to evaluate the need for a new security rulemaking that could incorporate updated requirements for Category II materials, among other things. In the absence of a new security rulemaking, the U.S. NRC will handle applications for possession and use of HALEU at reactors and fuel cycle facilities on a case-by-case basis, and if necessary, will require additional security measures as license conditions. This process could be burdensome if the U.S. NRC receives a large number of Category II facility applications.

The U.S. NRC states that the supplemental security measures aim to change the general objective of a Category II physical protection system from early detection and assessment of external intruders to “prompt” detection and assessment. This could require enhancements in such areas as access controls, security patrols, and communication and coordination with LLEA (U.S. NRC, 2021h).

Rivers et al. (2021) go further, arguing that the physical protection system for a HALEU fuel cycle facility should be designed with sufficient delay to prevent adversaries from leaving a site with stolen HALEU, rather than simply to facilitate prompt recovery of the material after it is stolen. Their paper discusses the need to determine adversary characteristics, such as numbers and equipment—a process similar to the development of a DBT. Furthermore, the security response requirements may also take into account material characteristics, such as dilution, that could increase adversary task time. Facilities possessing at least one Category II “goal quantity,” which the paper defines as 40 kg of contained uranium-235 (about one-half of the U.S. NRC’s “gross quantity,” corresponding to around 200 kg of 19.75 percent–enriched HALEU), would have to establish “high confidence” in a timely LLEA response. Although those authors believe that the primary responsibility for this interdiction capability could be provided by LLEA response, they conclude that “if sufficient delay cannot be incorporated into the security program to allow for timely LLEA response, it may be necessary for the site to have its own armed response force” (Rivers et al., 2021).

Another consideration under current U.S. NRC rules is that nuclear power reactors are required to have armed response forces to protect against the DBT of radiological sabotage and therefore would be able to readily comply with a similar requirement for protection against theft of HALEU, should the U.S. NRC decide to impose

___________________

²¹ It is important to note that U.S. NRC guidance represents one acceptable method for meeting regulatory requirements but is not itself a requirement. However, each licensee must submit a security plan, and the guidance document serves as the U.S. NRC basis for review and audit of the plan.

one. However, this may not be the case in the future, as the U.S. NRC is currently developing a proposed rule for small modular LWRs and advanced reactors that would allow owners of such reactors to reduce or even eliminate armed response forces if they could demonstrate a radiological sabotage attack would not cause unacceptably high radiation doses to members of the public (U.S. NRC, 2018b). Such a regulatory approach could significantly reduce personnel costs for small modular reactors or microreactors—but not if they were required to maintain an armed response force to protect HALEU.

6.3.4.3 Further Enrichment of HALEU to Highly Enriched Uranium

The second route that an adversary with access to HALEU can use to acquire a nuclear weapon is further enrichment of the material to HEU levels. Less separative work is required to generate HEU from HALEU than from LEU feed material, and thus access to HALEU could be advantageous to adversaries. For example, to acquire one kilogram of uranium enriched to 90 percent (commonly referred to as “weapons-grade” highly enriched uranium), about 1750 SWU (separative work unit) would be required with 4.5 percent–enriched feed, compared with about 550 SWU with 19.75 percent–enriched feed (roughly three times smaller).²²

Thus, access to HALEU could reduce the time or resources necessary to acquire an IAEA-defined significant quantity of highly enriched uranium. Given the three-fold reduction in SWU, a country seeking to covertly produce highly enriched uranium in a clandestine enrichment plant could obtain a given quantity in one-third the time by using HALEU feedstock rather than LWR fuel feedstock, or in the same time in an enrichment plant three times smaller. On the other hand, the difference in SWU requirements would be less significant for a state with a large commercial enrichment facility (with a capacity of millions of SWU per year) that decided to break out and overtly produce highly enriched uranium by reconfiguring the plant. In either case, the actual impact on proliferation risk would depend on specific circumstances, including the state’s chosen acquisition path, its overall capabilities and technological sophistication, and its reactor and fuel cycle infrastructure.

The committee heard differing views on this matter. One expert concluded that “widespread use of HALEU could have [a] significant impact [on safeguards] at reactors and other facilities,” while pointing out that the “impact will depend on the specific design of the reactor … storage plans, and refueling schedules” and how the IAEA chooses to address timeliness issues associated with the potential reenrichment of HALEU. According to this expert, while there are many aspects of advanced reactor designs that will have a greater impact on safeguards than the use of HALEU, the use of HALEU is one common factor of almost all designs being supported by the United States (Stern, 2021). However, an IAEA expert stated that HALEU would not give states “an extra edge,” based on the modest difference between the SWU required to enrich 5 percent feed to highly enriched uranium compared with HALEU feed (Boyer, 2021). In the case of diversion of a single or a small number of items, such as UF₆ cylinders, to a clandestine plant, access to HALEU might reduce the time to acquire highly enriched uranium from perhaps 2 months to 1 month, which this expert did not think would significantly impact the IAEA’s ability to meet its timeliness goal (Boyer, 2021).

6.3.4.4 The Current Safeguards Framework for HALEU

As discussed in Section 6.2.2, for safeguards implementation, the IAEA classifies all low-enriched uranium as “indirect-use” material with a significant quantity of 75 kg of (contained) uranium-235, and the IAEA timeliness detection goal is set at 1 year for all low enrichments. This differs from the IAEA’s physical protection framework discussed in Section 6.2.3.1, in that there is no analogue in safeguards for the 10 percent enrichment threshold that distinguishes Category II from Category III enriched uranium. The committee has been unable to determine whether the reason for this disparity in categorization has a technical basis or is simply a historical relic.

Nevertheless, 1 SQ can be obtained by diverting less HALEU than LWR-grade low-enriched uranium, so

___________________

²² This estimate assumes a tails assay of 0.25 percent in both cases. In practice, an adversary might choose a higher tails assay, depending on the amount of feedstock available and the adversary’s objectives. This would reduce the SWU requirements for either type of feedstock, but would also reduce the advantage of using HALEU.

HALEU safeguards approaches will have to be somewhat more intensive to meet the IAEA’s inspection goals. For example, 1 SQ of 4.5 percent–enriched uranium is 1,667 kg, whereas 1 SQ of 19.75 percent–enriched HALEU is 380 kg. The impact of these differences on safeguards would depend on specific facility and material characteristics.

HALEU production and use could also cause other safeguards issues. One expert also raised the concern that an enrichment plant producing HALEU with just under 20 percent enrichment might cause challenges for current safeguards approaches because of measurement uncertainties and the potential that small amounts of more highly enriched material could occur in process and cause false positives (Stern, 2021). Indeed, the committee heard from Centrus that such false positives could occur as the result of off-normal conditions in a centrifuge plant producing HALEU, which is why the company received approval for possession of uranium enriched up to 25 percent, although it is not authorized to draw off product greater than 20 percent (Poneman and Cutlip, 2021).

In addition, although the inspection goals do not depend on the low enrichment, the level of inspection effort actually does, through the use of a unit called the “effective kilogram” (ekg). For all uranium with an enrichment of at least 1 percent, the ekg is defined as the uranium weight multiplied by the square of the enrichment (IAEA, 2002). The number of kilograms of enriched uranium corresponding to 1 ekg decreases as the enrichment increases. One ekg is about 500 kg of 4.5 percent–enriched uranium but only about 25 kg of 19.9 percent–enriched uranium. The maximum routine inspection effort (MRIE) at a facility is a function of the quantity of enriched uranium expressed in terms of ekg (IAEA, 2002).²³ For uranium enriched to greater than 5 percent, the MRIE increases as the square root of the annual throughput (or inventory, whichever is larger) in terms of ekgs; thus, for a fixed throughput it increases linearly with the enrichment.²⁴ Under the current regime, the maximum routine inspection effort would be significantly greater on a per-MWd basis for the HALEU fuel facility.²⁵

However, the actual routine inspection effort at low-enriched uranium fuel fabrication plants, which is determined on a facility-specific basis, has decreased significantly over time and is in recent years much smaller than the MRIE (IAEA, 1985, 2020b).²⁶ No public information is available regarding the relationship between enrichment and actual inspection effort. However, one expert told the committee that the IAEA may need to modify its safeguards implementation in countries possessing large quantities of HALEU (Stern, 2021).

The historical and technical basis for the ekg unit is obscure. One author describes it as “a concept introduced to establish equivalencies of different levels of enriched uranium,” but does not explain what property of the uranium is being equated (Scheinman, 1987). However, the physical mass corresponding to 1 ekg, which varies inversely as the square of the enrichment E, is roughly comparable to the power-law dependence of the bare critical mass of a metal sphere as a function of enrichment (for enrichments above the approximately 6 percent threshold that have a finite bare critical mass), which is about E^–1.7 (derived from data in Glaser, 2006). So at least for enrichments above this threshold, the ekg could be regarded as a surrogate measure of the relative amount of material needed for direct use in a nuclear weapon as a function of enrichment—an interpretation that is not consistent with the IAEA’s definition of all low-enriched uranium as indirect-use material.

___________________

²³ In addition, the ekg is used to establish the threshold between “facilities” (for which state parties must submit design information to the IAEA and conclude a facility attachment) and “locations outside of facilities” for the purpose of application of safeguards: a facility is “any location where nuclear material in amounts greater than one effective kilogram is customarily used.” Also, states are only required to report nuclear material transfers of at least 1 ekg to the IAEA, and the IAEA only conducts more than one inspection per year at a facility or location outside a facility with a content or annual throughput of more than 5 ekg.

²⁴ For uranium with enrichments below 5 percent, the MRIE is determined by a different relationship.

²⁵ According to the formula, the MRIE for a 500-MT fuel fabrication plant fabricating 4.5 percent–enriched LWR fuel—which could fuel about 25 1,000-MWe reactors per year—would be about 50 person-days of inspection (PDI)/year. For a HALEU fuel fabrication plant providing fuel for X-energy’s Xe-100 reactors, the annual throughput to supply an equivalent number of reactors would be about 150 MT per year of 15.5 percent–enriched fuel, and the MRIE for this facility would be 1,800 PDI/year.

²⁶ For example, according to the IAEA Safeguards Implementation Report for 2019, there were a total of 804 PDI for 21 low-enriched uranium fuel fabrication plants in states with comprehensive safeguards agreements and additional protocols in place, averaging less than 40 PDIs per facility—an order of magnitude less than the MRIE for a 500 MT/y facility (IAEA, 2020b). Most of these facilities have throughputs of at least 500 MT/y.

6.3.4.5 Is a Revision of IAEA Inspection Goals Needed for HALEU?

The IAEA has successfully implemented safeguards at many research reactors using HALEU fuel enriched to just under 20 percent around the world, but quantities at each facility are typically on the order of tens of kilograms—well below 1 SQ (Stern, 2021). In contrast, even smaller advanced reactors require many SQs of HALEU per year. Also, the throughput of fuel cycle facilities needed to produce HALEU and fabricate fuel for a fleet of such reactors could be on the order of tens or hundreds of SQs. Thus, it is important to consider whether current safeguards criteria and inspection goals need to be tightened for advanced reactor fuel cycles using HALEU. One expert told the committee that this was a fundamental issue and pointed out that, in 1977, the IAEA included in its safeguards reporting a separate category for enriched uranium comparable to security Category II, but this was not continued (Stern, 2021).

One question is whether the safeguards goals for low-enriched uranium should be changed for HALEU, which would affect both the quantity and timeliness components. This is both a technical and a diplomatic question and would likely require IAEA Board of Governors approval (Stern, 2021). As an indirect-use material, the low-enriched uranium SQ is presumably the amount needed for feed into an enrichment plant in order to produce the 25 kg of uranium-235 contained in highly enriched uranium, taking into account process losses and other difficulties with the conversion (Krass et al., 1983).

However, even if the SQ is regarded as a rough measure of the HALEU quantity needed to build a nuclear weapon without further enrichment, the current low-enriched uranium SQ of 75 kg would still make sense for HALEU. Assuming a first-generation implosion device, 1 SQ of HALEU at just under 20 percent enrichment would be about 350 kg, containing 70 kg of uranium-235.

A stronger case could be made for reducing the timeliness detection goal for HALEU from the current period of 1 year, given the shorter time needed to produce highly enriched uranium from enriching HALEU compared with lower-enriched uranium feed.

6.3.5 Proliferation and Security Risks of Fuel Cycles Using Reprocessing and Recycling

In general, closed fuel cycles using reprocessing and recycling, as discussed in Chapters 2 and 4, pose greater nuclear proliferation and security risks than once-through fuel cycles. As a result, closed fuel cycles would require the application of more intensive international safeguards and domestic physical protection measures, which can be more costly in terms of the financial, technical, and human resources needed for their implementation, than are required for once-through cycles.

In fuel cycles that involve spent fuel reprocessing, materials that are nuclear weapon usable are separated from the highly radioactive fission products that provide the self-protecting radiation barrier of spent nuclear fuel, and they are then concentrated for storage or reuse. These reprocessing products are therefore more accessible for diversion or theft than weapon-usable materials that remain bound in spent fuel—at least for the many decades after discharge that the spent fuel remains self-protecting. Depending on the fuel cycle of interest, these weapon-usable materials, as discussed in Section 6.2.5, include plutonium, 233-uranium, or transuranic elements, such as neptunium, americium, and curium. Also noted in Section 6.2.5, the residual uranium in fuels that use HALEU may be attractive as well, depending on the fuel burnup.

As discussed in Chapters 2 and 4, France and a few other countries currently operate industrial-scale reprocessing plants using the PUREX process, which allows for the separation of plutonium in a pure form. The IAEA’s development of a safeguards approach for the only large-scale reprocessing plant in a non–nuclear weapon state, Rokkasho-mura in Japan (which is not yet operating), has proven to be very challenging. Because it assumes only a short time is needed for conversion of separated plutonium into a nuclear weapon, the IAEA must conduct more frequent inspections at reprocessing or mixed oxide fuel fabrication plants in order to meet its timeliness goals than it does at low-enriched uranium fuel fabrication plants or spent fuel storage facilities. In addition, the 8 kg SQ of plutonium is a small fraction of the typical annual throughput of a large reprocessing plant, which can be on the order of 8,000 kg of plutonium per year. Therefore, to meet the IAEA’s quantity inspection goal, very precise and frequent in-process measurements of plutonium are required. IAEA material accountancy goals cannot

be met at large bulk-handling facilities for actinide measurement uncertainties much higher than 1–2 percent (Cipiti and Shoman, 2018). Although many techniques do exist that are below these uncertainty levels for some of the key measurements, achieving such low uncertainties throughout the process on a plant-wide basis is not feasible, necessitating implementation of a “strengthened safeguards approach” with such supplemental measures as the process monitoring used at Rokkasho (Durst et al., 2007).

The PUREX process does allow for relatively precise measurements of process streams (on the order of 0.1 percent) by sampling and destructive analysis. Of particular note is the presence of an “input accountability tank” at the head-end of the plant where dissolved spent fuel is well-mixed, sampled, and assayed, allowing precise measurement of input quantities. This step is the only way the actinide contents of the input spent fuel can be measured directly; otherwise, operators and inspectors would have to rely solely on less-accurate burnup measurements and calculations. Similarly, at the end of the process, the high purity of product materials, such as plutonium oxide, facilitates precise measurements.

However, it is impractical for safeguards inspectors to conduct sampling and destructive analysis throughout a large plant to the degree necessary to achieve such low measurement uncertainties overall. At the Rokkasho plant, which has been designed for a throughput of 800 MTHM of LWR spent fuel per year, or a plutonium throughput of about 8 MT, the IAEA established an on-site laboratory to meet its measurement requirements, which include not only intrusive monthly interim inventory verifications but also short-interval verifications requiring sampling every 7–10 days (Durst et al., 2007).

Even so, in-process nondestructive assay measurements are also necessary, but the uncertainties associated with such techniques are typically far greater than those needed to meet IAEA accountancy goals. The difficulties in meeting accountancy goals are compounded by such factors as the accumulation of residual holdup, or nuclear material in process that is difficult to recover even when the plant is cleaned out for periodic physical inventories, as well as the generation of hard-to-assay scrap and waste streams, in which substantial quantities of safeguarded materials can accumulate.

The proliferation risks of spent fuel reprocessing and plutonium recycling have long been a major concern for many experts and were the reason the Ford and Carter administrations imposed a moratorium on domestic reprocessing “unless there is sound reason to conclude that the world community can effectively overcome the associated risks of proliferation,” which the Clinton administration reaffirmed in 1993 (Andrews, 2008). Nonetheless, for decades there have been numerous efforts to identify ways to modify closed fuel cycles to increase the intrinsic proliferation resistance—including the International Nuclear Fuel Cycle Evaluation and DOE’s Nonproliferation Alternative Systems Assessment Program in the 1970s and 1980s. These studies, as well as many subsequent ones, have generally concluded that “there is no ‘proliferation proof’ nuclear power cycle” and that “all nuclear fuel cycles and many fissionable isotopes (including all those of Pu) entail some risk” (CGSR, 2000).

6.3.5.1 Proliferation-Resistance of Alternative Closed Fuel Cycles

Some experts have continued to argue that alternatives to the current PUREX-based fuel cycle that can significantly reduce proliferation and security risks (Shafer, 2021) are primarily based on reprocessing flow sheets that do not produce pure plutonium (or uranium-233) streams. As discussed in Chapter 4, such approaches include aqueous-based variations on PUREX that would coextract plutonium with uranium (COEX) or with various combinations of minor actinides, including neptunium, americium, and curium (the UREX+ suite). Some UREX+ variants would also produce mixtures containing lanthanide fission products, some of which could provide a radiation barrier. Similarly, nonaqueous technologies such as pyroprocessing, as discussed in Chapter 4, have been put forward as more proliferation resistant than PUREX because the product of normally operating pyroprocessing would be a mixture of plutonium, minor actinides, some uranium, and certain lanthanide fission products (ARPA-E, 2021).

In the 2000s, DOE initiated the Advanced Fuel Cycle Initiative (AFCI), which was intended to develop more proliferation-resistant separations processes, and later the Global Nuclear Energy Partnership, which in part sought to develop international spent fuel reprocessing and recycling centers in countries that already possessed such facilities (Andrews, 2008). The AFCI included work on demonstrating the UREX+ suite of processes and remote

recycle fuel fabrication, as well as continued pyroprocessing development. As part of the effort, DOE tasked the national laboratories with evaluating the potential benefits of such alternatives (Bari et al., 2009; Bathke et al., 2012). As discussed in Section 6.2.5, these studies concluded that none of the alternative reprocessing schemes considered conferred significant proliferation resistance compared to PUREX, because the product streams remained attractive materials for use in nuclear weapons (largely confirming what was already known) (National Research Council, 2008).

The general conclusion of these reviews was that these alternative fuel cycles would still require levels of safeguards and security comparable to PUREX-based fuel cycles.

Despite these findings, reactor developers interested in pursuing reprocessing options continue to maintain that fuel cycles that do not separate pure plutonium—most notably, pyroprocessing—do provide significant proliferation resistance. For example, Moltex, which seeks to pyroprocess spent oxide fuel to extract plutonium and other actinides for use as fresh fuel in its Stable Salt Reactor, a fast molten salt reactor, says that “the main output of the [pyroprocessing] process is an impure extraction of the minor actinides (including plutonium) … [that] would be useless in weapons” (Moltex Energy, 2022). Similarly, Flibe Energy, which is developing a thermal uranium-233 molten salt breeder reactor, ascribes great importance to the presence of uranium-232, asserting that it renders uranium-233 “highly undesirable … as a weapons material” (Flibe Energy, 2022). While these technologies may provide some benefit in delaying direct use of the materials, there was consensus among the committee members that none provided significant proliferation resistance at this time.

6.3.5.2 Impact of Alternative Fuel Cycles on Safeguards

The impact of alternative separations and fabrication technologies on the safeguardability of fuel cycle facilities and materials is another important factor in assessing their proliferation risks. As noted above, it is not possible to meet safeguards inspection goals at industrial-scale PUREX reprocessing plants and other plutonium bulk-handling facilities with material accountancy alone, even though relatively high-precision measurements can be made of the inputs and outputs. The situation will be more problematic at advanced fuel cycle facilities that do not separate pure fissionable materials and may provide fewer opportunities for precision sampling and destructive assay of process and product streams.

Aqueous reprocessing plants employing alternative fuel cycles in which plutonium is separated as part of a group of actinides and lanthanides, such as the UREX+ suite, will be able to utilize safeguards approaches similar to PUREX plants of comparable throughput. However, additional complications will arise from the fact that assays of streams containing actinide and lanthanide mixtures will be more difficult and time consuming. This will be true, in particular, for nondestructive assay, as the gamma and neutron emissions from the individual components of the mixtures, as well as their decay heat rates, can interfere with one another, obscuring the radiation or thermal signatures used to identify and measure different isotopes. Even destructive assay would likely require additional preparation and measurement time for more complex mixtures.

6.3.5.3 Material Accountancy at Pyroprocessing Plants

For nonaqueous recycle systems such as pyroprocessing, the difficulties associated with measuring heterogeneous material streams are compounded by additional technical differences between aqueous reprocessing plants that further complicate material accountancy. A recent article by Hoyt et al. (2021) notes,

More specifically, Cipiti et al. (2021a) state the following:

• Completely flushing out an electrochemical facility is not as feasible an operation as at an aqueous facility. If the operator decides not to flushout material, inventory measurements can be used instead. However, although holdup measurements may be no more difficult with pyroprocessing than any bulk-handling facility, they will be of greater importance if the plant is not periodically flushed out.
• The input spent fuel and the in-process electrorefiner salt are heterogeneous, which introduces large sampling errors.
• The uranium and transuranic metallic products can be measured using sampling and destructive analysis; however, this measurement is different from aqueous techniques in regard to the need to sample a molten metal, and “may be difficult and costly for routine measurements” (Cipiti et al., 2021a). “More experimental work is necessary” to develop techniques for accurate nondestructive assay measurement of the actinide composition of the metallic uranium/transuranic product (Cipiti et al., 2021a).

The impact of these additional challenges is apparent from the results of the DOE study known as the “MPACT” (Materials Protection, Accounting, and Control Technologies) 2020 Milestone. This study entailed development of a Virtual Facility Distributed Test Bed for a model electrochemical reprocessing (pyroprocessing) plant to demonstrate the implementation of safeguards and security by design (Cipiti et al., 2021a,b). The throughput of the model plant is 100 MT/y of LWR spent fuel, or about 1 MT of plutonium per year. This is one-eighth the size of a large aqueous reprocessing plant such as Rokkasho and would only be able to supply enough plutonium (and other transuranic elements) to fuel about one PRISM-sized fast burner reactor²⁷ per year.

The study estimated the detection probabilities for a range of material-loss scenarios, evaluating both the IAEA’s safeguards goals and the U.S. NRC’s MC&A requirements for Category I facilities (see Tables 6.2 and 6.3).²⁸ For the IAEA, the study used the goal of detecting the diversion of 1 SQ of plutonium (8 kg) within one material balance period (here taken as 1 month) with a 95 percent probability of detection and a 5 percent false alarm rate.²⁹ The study also considered the U.S. NRC’s more stringent domestic material control and accounting requirements in 10 CFR 74.53(b) for detecting losses of one Category I quantity of strategic special nuclear material (2 kg for plutonium) from accessible process areas within 7 days, with a 95 percent probability.³⁰

The MPACT study evaluated several loss scenarios while varying the assumed total measurement uncertainty (from both random and systematic errors) from 1 to 5 percent at the critical measurement points (input fuel, electrorefiner salt inventory, and product) (Cipiti et al., 2021a). These results indicate that the IAEA’s goal can only be met for all scenarios if the total uncertainty is kept to 1 percent or less for the critical measurement points, and that the U.S. NRC’s 7-day goal (for Category IB material) cannot be met except for abrupt diversions assuming the lowest measurement uncertainty category.

The actual performance of the material accountancy system would likely be lower than indicated by these results, which are based on optimistic assumptions. For example, they assume short and likely impractical material balance periods—8 days for the U.S. NRC and 30 days for the IAEA. The impacts of holdup were not considered in detail, even though holdup has been found to be significant in actual pyroprocessing operations (Vaden, 2007). Waste streams were assumed to have very low actinide content, and high measurement uncertainties were tolerated, even though experience has shown that some waste streams, such as cladding hulls, still contain a significant percentage of undissolved spent fuel (about 5 percent of the initial uranium) (Westphal et al., 2013). Sampling

___________________

²⁷ PRISM (Power Reactor Innovative Small Module) is a small modular, sodium-cooled fast reactor developed by GE-Hitachi Nuclear Energy, with proposed power output of 840 MWth/311 MWe (Triplett et al., 2012).

²⁸ These requirements would not apply to a U.S. NRC–licensed reprocessing facility today because its regulations declare reprocessing facilities to be Category II facilities by fiat, even though they would possess Category I quantities of material. This regulatory inconsistency would have to be addressed if the U.S. NRC were to receive a reprocessing plant application in the future.

²⁹ This is only a subset of the IAEA’s Safeguards Criteria, which generally require that abrupt diversions of 8 kg of Pu be detected within 1 month and protracted diversions within 1 year (Johnson, 2009). The MPACT project’s use of “abrupt” and “protracted” here needs clarification.

³⁰ This is only one of several U.S. NRC loss-detection goal requirements for Category I facilities. The U.S. NRC also requires that loss of a formula quantity Category IA materials (generally strategic special nuclear material items that can be carried by one person inconspicuously) has to be detected within 3 days with 95 percent likelihood. Some items, such as uranium/transuranium ingots, would likely be classified at Category IA.

errors were not specifically evaluated, even though those have been recognized as major sources of uncertainty both for the input spent fuel and for the electrorefiner salt.³¹ In particular, the salt itself is heterogeneous, as it forms several different layers with dispersed insoluble particles (Croce et al., 2021). And finally, an industrial-scale pyroprocessing facility would likely have a throughput several times larger than the plant modeled in MPACT—making it even harder to detect small diversions.

Given the numerous difficulties identified above that a pyroprocessing plant poses for material accountancy, the MPACT project concluded that achievement of 1 percent measurement uncertainties is a “best-case” assumption and in most cases is “an extrapolation of experimental work and represent[s] a best estimate of what may be possible with the technology” (Cipiti et al., 2021a). A recent survey of candidate measurement techniques reveals that most are in an early stage of development, and none have yet been validated under fully representative conditions, given the absence of a production-scale facility where they could be field tested (Coble et al., 2020). The MPACT campaign’s Virtual Facility Distributed Test Bed included experimental work at the DOE laboratories to further develop pyroprocessing measurement approaches that could help overcome the main technical limitations, including microcalorimetry, high-dose-rate neutron detectors, and cyclic voltammetry (see Box 6.4).

6.3.5.4 Material Accountancy at Molten Salt–Fueled Reactors with Reprocessing

As discussed in Chapter 4, some molten salt–fueled reactor concepts include online reprocessing systems to separate actinides and neutron-absorbing fission products from the irradiated fuel. Such systems would enable such designs to take full advantage of the flexibility in fuel composition and potentially very high burnup that liquid fuels could provide. They would also be needed to increase natural resource utilization compared with once-through molten salt reactors. The most prominent example is the thorium/uranium-233 thermal breeder, of which Flibe is an example. In order to maximize breeding performance, protactinium-233 must be separated and allowed to decay to uranium-233 outside of the reactor to prevent parasitic absorption and production of uranium-234, and other neutron poisons must be removed to maximize the neutron economy. This requires frequent passes of the

___________________

³¹ The MPACT study simply assumed that sampling errors were included in the overall uncertainty estimates.

core through the reprocessing system. Fast salt-fueled reactors are somewhat more tolerant of neutron poisons than thermal reactors but still require their separation at a reduced rate.

The addition of online reprocessing systems would clearly increase the difficulty of achieving accurate material accountancy at molten salt–fueled reactors, compounding the already formidable challenges described above for once-through molten salt reactors due to the inability to accurately estimate and verify the core inventory. The presence of a salt processing system would provide more opportunities for diversion, thereby increasing the credibility of scenarios that take advantage of the uncertainties in accountancy measurement to conceal diversions. This is of particular concern in breeder systems that require separation of relatively pure streams of weapon-usable material, such as uranium-233.

Although the ingrowth of uranium-232 due to the decay of protactinium-232 is often cited as a proliferation resistance measure because of the intense gamma emissions of its decay products, the presence of uranium-232 is also a hindrance because it could interfere with material accountancy measurements if it requires items to be shielded (Evans-Worrall, 2021). Moreover, the uranium-232 contamination level of uranium-233 can be minimized by separating uranium-232 from protactinium-233 before protactinium-233 decays to uranium-233, taking advantage of the significantly shorter half-life (1.7 days) of protactinium-232, which might be desirable from an occupational safety perspective.

Given the chemical forms of their fuels, nonaqueous separations technologies would be needed for the online reprocessing systems at molten salt reactors. Chloride salt–based systems could use pyroprocessing technologies similar to the one discussed in Section 6.3.5.3, and the material accountancy challenges discussed there would generally apply to an on-site reprocessing unit at a molten salt reactor. Some aspects, such as the inability to accurately account for input into the processing line, would increase the difficulty. Designs that require quasicontinuous processing would introduce additional complications compared with the batch process analyzed in the MPACT study. And the steadily increasing in-core inventories for some systems illustrated in Section 6.3.3.1 would effectively increase the fissionable material throughput in the system, and hence increase the challenges of detecting small diversions over time.

For fluoride salts, analogous pyrochemical processes could be used (Fredrickson et al., 2018). Although the material accountancy aspects of such systems have not been studied in as much detail as the MPACT project’s study of chloride-based pyroprocessing, one could expect similar challenges. One additional problem presented in systems containing uranium-233 is that techniques for uranium-233 nondestructive assay do not currently exist (Evans-Worrall, 2021).

6.3.5.5 Security Aspects of Alternative Closed Fuel Cycles

The question of whether alternative closed fuel cycles can be designed to have significantly lower security risks than PUREX-based systems depends fundamentally on whether the materials that are produced, stored, processed, and transported are of low attractiveness for substate actors seeking to acquire nuclear weapons. This depends on characteristics such as their vulnerability to theft, their direct weapon usability, and the ease of converting those materials to more attractive materials. Depending on these features, alternative fuel cycle materials might have greater theft resistance even though they may not be more resistant to proliferation. The 2009 Brookhaven National Laboratory study of proliferation risk reduction concluded that alternative fuel cycles producing mixtures of plutonium, neptunium, americium, curium, and possibly lanthanide fission products, would provide “some advantage” over fuel cycles that produced separated plutonium or plutonium and neptunium because of “larger radiation and heat loads” that “would be more difficult to handle for health and safety reasons” (Bari et al., 2009).

Although most lanthanide fission products are relatively short-lived, they are the largest contributors to radiation dose and heat rates of the alternative fuel cycle mixtures. But they must be removed before fabrication of current-technology fast reactor fuels because of their propensity to migrate to the fuel periphery and contribute to fuel-cladding chemical interaction (FCCI). In contrast, molten salt fast reactors may be able to retain them, as FCCI is obviously not an issue.

The 2009 Brookhaven study also concluded that “even with the lanthanides present the total dose is not very

high and would be unlikely to deter an adversary who was willing to accept injury or self-sacrifice” (Bari et al., 2009). It also pointed out that all such mixtures would still require Category I security as defined by the U.S. NRC and IAEA (Bari et al., 2009).

Nevertheless, in more refined security categorizations, such as the DOE-graded safeguards table or the comparable approach that underlies the enhanced security rulemaking that the U.S. NRC may once again consider (see Section 6.2.3.2), factors such as significant dilution by uranium or high-dose-rate and decay-heat materials might be credited in reducing attractiveness, and hence might warrant somewhat less stringent security measures than would separated plutonium. However, generally high rates of dilution (<10 percent) or a high concentration of lanthanides (>10 weight percent) are required to reduce the attractiveness of the product by one level (Bathke et al., 2012). Thus, processes such as COEX (50-50 uranium-plutonium mixture) or pyroprocessing do not significantly reduce attractiveness.

If resumed, the U.S. NRC rulemaking on enhanced security would provide an opportunity to clarify the attractiveness of all materials that may be present in advanced reactor fuel cycles and the appropriate levels of security to apply to them. A comprehensive rule would include appropriate treatment of weapon-usable materials, such as neptunium, that the current framework does not address, consideration of physical and chemical properties of materials, and a reevaluation of the dose-rate threshold for effective self-protection.³²

__________________

³² In SRM-SECY-09-0123, the U.S. NRC commissioners disapproved a staff proposal to include americium and neptunium in a revised material categorization scheme. Given the potential that advanced reactor fuel cycles involving separated streams or mixtures that include these elements may be developed in the future, the U.S. NRC should revisit this decision in its consideration of resuming the enhanced security rulemaking.