2
Nuclear Weapons

Transparency and monitoring measures can be employed at several phases in the life cycle of nuclear weapons, which is illustrated schematically in Figure 2-1. As discussed in Box 2-1, the key components of a nuclear weapon are those containing nuclear-explosive materials (NEM).1 These components, together with high-explosive assemblies and various electrical and mechanical components, are assembled into nuclear weapons in specialized facilities; in the United States, this is done at the Pantex plant in Amarillo, Texas. Assembled weapons are transported to military storage facilities, where they are stored pending deployment with a delivery vehicle. Nuclear weapons are operationally deployed and ready for use when they are mated to ballistic missiles and placed in launchers, loaded onto aircraft, or stored at air bases for nuclear-capable aircraft.2 Nuclear weapons may be removed from operational deployment from time to time for inspection and routine maintenance of the weapon or its delivery vehicle and launcher. Weapons also may be kept in long-term storage as spares, as a source of parts for remanufacture or the manufacture of other weapons, or held in reserve as a responsive force that may augment deployed forces. When a decision is made to eliminate a nuclear weapon, it is disassembled and the NEM components are stored for reuse or final disposition.

Past and current arms control agreements kept most of the nuclear weapons life cycle closed to outside scrutiny. As discussed in Chapter 1, nuclear weapons were not considered suitable for direct monitoring and transparency because of their small size, the ease with which they could be concealed, and the secrecy that sur-

1  

NEM are defined in Footnote 3 of Chapter 1; see also Chapter 3 and Appendix A for a further technical discussion.

2  

This applies to gravity bombs and weapons for ballistic or cruise missiles, which account for the vast majority of the world’s operational nuclear arsenals. In the past, the United States and Russia deployed other types of nuclear weapons, including landmines, artillery shells, depth charges, and weapons for air-defense missiles and torpedoes. In 1991-92 both sides pledged to withdraw from deployment these types of weapons and to dismantle all nuclear landmines, artillery shells, and weapons for short-range missiles (see Chapter 1). The United States has dismantled all of these other types of weapons; Russia may have air-defense and antisubmarine weapons in storage. China may stockpile some of these other types of nuclear weapons.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials 2 Nuclear Weapons Transparency and monitoring measures can be employed at several phases in the life cycle of nuclear weapons, which is illustrated schematically in Figure 2-1. As discussed in Box 2-1, the key components of a nuclear weapon are those containing nuclear-explosive materials (NEM).1 These components, together with high-explosive assemblies and various electrical and mechanical components, are assembled into nuclear weapons in specialized facilities; in the United States, this is done at the Pantex plant in Amarillo, Texas. Assembled weapons are transported to military storage facilities, where they are stored pending deployment with a delivery vehicle. Nuclear weapons are operationally deployed and ready for use when they are mated to ballistic missiles and placed in launchers, loaded onto aircraft, or stored at air bases for nuclear-capable aircraft.2 Nuclear weapons may be removed from operational deployment from time to time for inspection and routine maintenance of the weapon or its delivery vehicle and launcher. Weapons also may be kept in long-term storage as spares, as a source of parts for remanufacture or the manufacture of other weapons, or held in reserve as a responsive force that may augment deployed forces. When a decision is made to eliminate a nuclear weapon, it is disassembled and the NEM components are stored for reuse or final disposition. Past and current arms control agreements kept most of the nuclear weapons life cycle closed to outside scrutiny. As discussed in Chapter 1, nuclear weapons were not considered suitable for direct monitoring and transparency because of their small size, the ease with which they could be concealed, and the secrecy that sur- 1   NEM are defined in Footnote 3 of Chapter 1; see also Chapter 3 and Appendix A for a further technical discussion. 2   This applies to gravity bombs and weapons for ballistic or cruise missiles, which account for the vast majority of the world’s operational nuclear arsenals. In the past, the United States and Russia deployed other types of nuclear weapons, including landmines, artillery shells, depth charges, and weapons for air-defense missiles and torpedoes. In 1991-92 both sides pledged to withdraw from deployment these types of weapons and to dismantle all nuclear landmines, artillery shells, and weapons for short-range missiles (see Chapter 1). The United States has dismantled all of these other types of weapons; Russia may have air-defense and antisubmarine weapons in storage. China may stockpile some of these other types of nuclear weapons.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials rounded them. For the most part, therefore, transparency and monitoring measures extended only to delivery systems (ballistic missiles and their silos, mobile and submarine launchers, strategic bombers, and cruise missiles). Agreed limits on the number of deployed strategic weapons were verified indirectly with counting rules that attributed a certain number of weapons to each deployed delivery system of a particular type. The first Strategic Arms Reduction Treaty (START I), for example, provides for a small number of inspections each year to confirm that the number of weapons on a selected ballistic missile does not exceed the number allowed for that type of missile. There have been no agreed limits on the number of strategic nuclear weapons that can be kept in storage for possible deployment, no declarations or measures to confirm the number of strategic or nonstrategic weapons in the stockpile, and the only measures to confirm that nuclear weapons have been eliminated are those being undertaken in connection with agreements concerning the disposition of highly enriched uranium from dismantled weapons. FIGURE 2-1 Life cycle of a nuclear weapon.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials This chapter considers several possible measures for transparency and monitoring of nuclear weapon stockpiles. Chapter 3 then considers measures to apply transparency and monitoring to stockpiles of NEM, including current and past production in military and civilian programs and for the disposition of existing and future stocks. The measures related to nuclear weapons stockpiles include declarations of weapon inventories at various levels of detail; on-site inspections and other methods to confirm the accuracy of declarations disaggregated by facility; continuous monitoring of certain weapon stocks, such as excess or retired weapons in storage; transparency measures at weapon assembly facilities to confirm the dismantling and permitted remanufacture of weapons; and declarations and transparency measures covering the storage and fabrication of NEM components (pits and secondary assemblies; see Box 2-1 for an explanation of these terms). Other confidence-building measures that could prove useful include exchanges of information on historical weapon inventories including past assembly and disassembly records, facility design information and operating records, and other records. The various technologies available to support these measures are described and assessed in boxes; the chapter text describes how they might be applied. The various measures are presented as an integrated package of progressively more demanding steps. We emphasize, however, that some are less sensitive and intrusive and could have value as part of broader confidence-building efforts, even if more formal and intrusive verification initiatives were not adopted. These forms of “cooperative transparency” could be augmented by information gathered unilaterally by individual states, which is current U.S. arms control and nonproliferation practice. The available methods include “National Technical Means” (NTM) such as satellite imagery collected over a range of wavelengths, imagery from reconnaissance aircraft, data from chemical and radiological detectors on aircraft overflying a country or patrolling just outside its borders, and intercepted electronic communications. They also include information gathered by human sources—clandestine operatives—and information obtained from defectors or “whistle blowers.” DECLARATIONS OF NUCLEAR WEAPON STOCKS The logical first step toward increased transparency would be for nuclear weapon states to exchange information on their inven-

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials tories of nuclear weapons. Alternatively, a declaration or data exchange could be part of a formal agreement. Most arms control agreements have included provisions for initial and subsequent declarations of treaty-limited items; examples include START I, the Treaty on the Elimination of Intermediate-Range Missiles (INF Treaty), the Treaty on Conventional Armed Forces in Europe, and the Chemical Weapons Convention (CWC). In each of these cases, the initial declaration defined a baseline inventory from which the agreed reductions proceeded. A formal agreement has the advantage of precision since it could define the items covered and specify exactly what information is to be exchanged; for example, an agreement might define what constitutes a “nuclear weapon” (see Box 2-1) and specify when a weapon would be considered dismantled or eliminated. Declarations could also be made unilaterally and informally, with or without the expectation of reciprocation by other nuclear weapon states. The main advantage of voluntary declarations is that they can be accomplished quickly, without lengthy and detailed negotiations. The United States could lead by example, declaring its inventories of weapons and inviting other states to do the same. Given the size of the U.S. nuclear arsenal, making a unilateral declaration would not incur a security risk, but the United States would lose bargaining leverage in negotiating greater transparency if Russia did not reciprocate. The major disadvantage of this approach is that there would be no agreed definitions of exactly what should be declared and therefore no basis for transparency measures to confirm the accuracy of the information provided. This proved to be a problem with the informal Presidential Nuclear Initiatives on nonstrategic nuclear weapons announced by the United States and Russia in 1991-92, where lack of a formal agreement, detailed exchanges of information, or transparency measures at one point led some to charge that Russia was redeploying weapons in violation of its previous pledge, despite Russian denials.3 Proposals for U.S.-Russian Declarations Proposals for weapon declarations and related transparency measures emerged in the wake of the breakup of the Soviet Union. 3   See Rose Gottemoeller, “Offense, Defense, and Unilateralism in Strategic Arms Control,” Arms Control Today 31 (September 2001), pp. 10-15. Bill Gertz, “Russia Transfers Nuclear Arms to Baltics,” The Washington Times, January 3, 2001, p. A1.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials In 1992 concern about the safety and security of nuclear weapons and NEM in the former Soviet Union led Senator Joseph Biden to introduce the following amendment to the resolution of ratification for the START I Treaty: Nuclear Stockpile Weapons Arrangement. Inasmuch as the prospect of a loss of control of nuclear weapons or fissile material in the former Soviet Union could pose a serious threat to the United States and to international peace and security, in connection with any further agreement reducing strategic offensive arms, the President shall seek an appropriate arrangement, including the use of reciprocal inspections, data exchanges, and other cooperative measures, to monitor (A) the numbers of nuclear stockpile weapons on the territory of the parties to this Treaty; and (B) the location and inventory of facilities on the territories of the parties to this treaty capable of producing or processing significant quantities of fissile materials.4 The amendment was interpreted to apply to a future START III treaty, since the START II negotiations were already moving to a conclusion at that time. In 1994 the United States began trying to gain Russian agreement to greater transparency; since most of these efforts included both weapons and NEM, they are described together here. The United States had several motives for these initiatives: to fulfill the requirements of the Biden Amendment in anticipation of a future START III treaty; to facilitate U.S. assistance to Russia under the Cooperative Threat Reduction (CTR) program; and to bolster international support for the indefinite extension of the Nuclear Non-Proliferation Treaty (NPT). As described elsewhere in this study (see Chapter 3), part of this effort included a significant amount of joint work between the U.S. and Russian nuclear weapons laboratories to explore transparency and monitoring measures that could make this increased openness possible while still protecting sensitive information about weapon designs. At their summit in September 1994 Presidents Clinton and Yeltsin agreed to “exchange detailed information…on aggregate stockpiles of nuclear warheads, on stocks of fissile materials and on their safety and security. The sides will develop a process for 4   The START Treaty, Executive Report (Washington, DC: U.S. Government Printing Office, September 18, 1992), pp. 102-153.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials exchanging this information on a regular basis.”5 Their May 1995 summit statement reaffirmed that the two governments would negotiate agreements on stockpile data exchanges, and called for a third agreement on “other cooperative measures, as necessary to enhance confidence in the reciprocal declarations on fissile material stockpiles.” The two presidents also agreed to “examine and seek to define” possibilities for “intergovernmental arrangements to extend cooperation to further phases of the process of eliminating nuclear weapons.”6 The United States formally presented the draft text of an agreement to Russia in June 1995. The proposal included a call for a confidential exchange of data on current total inventories of weapons and NEM, as well as on the total number of nuclear weapons dismantled each year since 1980 and the type and amount of NEM produced each year since 1970. The Russians declined to discuss the draft, however. According to James Goodby, U.S. chair of the joint working group that had been exploring the arrangements, some Russian members of the group “gave the impression that the scope of the data exchange went well beyond what they were prepared to consider.”7 The United States nonetheless continued to seek agreement on stockpile declarations. At their March 1997 summit in Helsinki, Presidents Clinton and Yeltsin agreed that a START III agreement, to be negotiated following ratification of START II, should include “measures relating to the transparency of strategic warhead inventories and destruction of strategic nuclear warheads” and that transparency measures related to nonstrategic nuclear weapons and to nuclear materials would also be explored.8 When the United States presented a draft protocol dealing with transparency and monitoring for weapons in early 2000 for consideration in connec- 5   Joint Statement on Strategic Stability and Nuclear Security by the Presidents of the United States of America and the Russian Federation (Washington, DC: The White House, Office of the Press Secretary, September 28, 1994), p. 3. 6   Joint Statement on the Transparency and Irreversibility of the Process of Reducing Nuclear Weapons (Washington, DC: The White House, Office of the Press Secretary, May 9-10, 1995). 7   James Goodby, “Transparency and Irreversibility in Nuclear Warhead Disarmament,” in Harold A. Feiveson, ed. The Nuclear Turning Point: A Blueprint for Deep Cuts and De-Alerting of Nuclear Weapons (Washington, DC: Brookings Institution Press, 1999), p. 186. 8   Joint Statement on Parameters on Future Reductions in Nuclear Forces (Washington, DC: The White House, Office of the Press Secretary, March 21, 1997).

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials tion with START III, however, the Russians were not interested in pursuing the idea.9 In the end, START II never entered into force, and START III negotiations were never initiated. The obstacles to more far-reaching transparency—in the form of deep-rooted mutual suspicion about motivations, desire to protect sensitive information, and bureaucratic and legal impediments on both sides—are formidable, not least because some of them reflect genuine dilemmas about the balance of risk and benefit.10 Although no formal agreements on stockpile declarations have been achieved, some data have been released on a voluntary basis. No nuclear weapon state has revealed the precise number of nuclear weapons in its current stockpile; however, the United States, the United Kingdom, and France have released some information. The United States released an official accounting of the total number of nuclear weapons in its stockpile each year from 1945 to 1961; the total yield of the stockpile and the number of weapons retired or dismantled each year from 1945 to 1994; and, for fully retired weapon types, the total number assembled each year.11 Although it did not provide such historical details, the United Kingdom has stated that in the future it would maintain fewer than 200 operationally available weapons of a single type.12 In the mid-1990s France announced that it would eliminate its land-based nuclear forces, reduce the number of strategic submarines it would deploy, and dismantle the facilities used to produce NEM for nuclear weapons.13 9   Rose Gottemoeller, “Parsing the Nuclear Posture Review,” A ACA Panel Discussion With Daryl G. Kimball, Janne E. Nolan, Rose Gottemoeller, and Morton H. Halperin, Arms Control Today 32 (March 2002), pp. 15-21. Available as of January 2005, at: http://www.armscontrol.org/act/2002_03/panelmarch02.asp 10   For a discussion of these issues from both U.S. and Russian perspectives, see National Research Council, Overcoming Impediments to U.S.-Russian Cooperation on Nuclear Nonproliferation: Report of a Joint Workshop (Washington, DC: The National Academies Press, 2004). 11   Declassification of Certain Characteristics of the United States Nuclear Weapon Stockpile. Fact Sheet (Washington, DC: U.S. Department of Energy, June 27, 1994). Available as of January 2005, at: http://www.osti.gov/html/osti/opennet/document/press/pc26.html. 12   British Ministry of Defence, “Strategic Defence Review: Deterrence and Disarmament” (London: Ministry of Defense, July 1998). Available as of January 2005, at: http://www.mod.uk/issues/sdr/deterrence.htm. 13   Robert S. Norris, William M. Arkin, Hans M. Kristensen, and Joshua Handler, “French Nuclear Forces,” Bulletin of the Atomic Scientists 57 (July/August 2001), pp. 70-71.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials Levels of Detail in Stockpile Declarations Information on nuclear weapon stockpiles could be exchanged at various levels of detail, from highly aggregated to data on individual weapons. Table 2-1 illustrates this progression using four levels of detail: total number of weapons; total inventories by weapon type and status; inventories by facility; and a complete itemized inventory. TABLE 2-1 Illustrative Levels of Detail for Declarations of Weapon Inventories Level Type of Information 1 Current total number of nuclear weapons of all types. Each year since first test: total number assembled, disassembled, and in the stockpile. For each of next five years: planned number assembled, disassembled, stockpiled. 2 Current total number of each weapon type, by status (e.g., operationally deployed, active reserve, inactive reserve, retired/awaiting dismantling). Delivery systems associated with each weapon type. Each year since first test: total number of each weapon type assembled, disassembled, and in the stockpile. For each of next five years: planned number of each type assembled, disassembled, and in the stockpile. 3 Name and location of all facilities at which nuclear weapons are currently deployed, stored, assembled, maintained, remanufactured, dismantled, or other otherwise handled. Facility descriptions and site maps indicating each launcher, storage bunker, building, or other site in which nuclear weapons are or may be located. Number of each weapon type at each facility. Name and location of facilities that previously contained weapons. 4 For each weapon: serial number, weapon type, status, and current location. Total Inventories The simplest declaration would give the total number of nuclear weapons currently possessed by each state. Even at this gen-

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials eral level there are technical issues to resolve, such as whether unassembled or partly assembled weapons, test or mock devices, and explosive devices not intended for military use would count as “weapons” for the purposes of the declaration. The total worldwide stockpile of nuclear weapons is probably around 30,000, of which the arsenals of the United States and Russia constitute 95 percent. China, France, and the United Kingdom collectively account for 3 to 4 percent of the worldwide total, while Israel, India, and Pakistan together account for less than 1 percent. Although actual weapon inventories remain state secrets in each of these countries, we do not find a persuasive security rationale for keeping secret the total number of weapons held by each of the five de jure nuclear weapon states (i.e., the five states recognized as such by the NPT). The potential confidence-building value of such declarations is underscored by the substantial uncertainties in some official U.S. estimates; for example, for Russia this may be as large as “plus or minus 5,000” weapons.14 Most of this is due to large uncertainties about Russia’s stock of nonstrategic weapons, which may account for more than half of the Russian stockpile. It could also be very useful to declare historical weapon inventories to help build confidence in the accuracy and completeness of declarations of current inventories. States willing to share current data could also be willing to share comparable historical data. Beginning with the year of their first nuclear test, states could give the total number of weapons assembled and disassembled each year and the total number in the stockpile at the end of the year. States also could share information on their future plans for the weapon stockpile, giving, for example, the projected number of weapons to be assembled and dismantled each year for the next five years. Inventories by Type and Status The next level of detail would disaggregate total inventories by weapon type and/or associated delivery vehicle. The inventory could be further disaggregated by status, to differentiate between “active” weapons (those ready for immediate military use) and 14   See, for example, Lawrence Gershwin, National Intelligence Officer for Strategic Programs, DOD Appropriations for FY93, testimony before the House Committee on Appropriations, Part 5, (May 6, 1992), p. 499; and Gen. Eugene Habiger, U.S. Air Force, Former Commander U.S. Strategic Command, The Moscow Treaty. “The Treaty between the United States of America and the Russian Federation on Strategic Offensive Reductions,” Testimony before the Senate Foreign Relations Committee, (July 23, 2002).

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials “inactive” weapons (those without tritium and other limited-life components installed). Active weapons could be divided into operationally deployed weapons (i.e., mated with a ballistic missile or ready for immediate aircraft delivery) and other weapons that could be deployed relatively quickly (i.e., spares and reserves). The inactive inventory might also be further divided into reserve weapons and weapons that have been permanently retired and are awaiting dismantling. As with total inventories, it could also be useful to release comparable historical data and share future plans. For each weapon type (including types not in the current stockpile), for example, states could declare the number of weapons assembled and disassembled in each year since the first nuclear test and the total number of each type in the stockpile. Today, inventories by type and status are state secrets. In the case of Russia and the United States, which have exchanged detailed information on deployed strategic forces, such secrecy only serves to protect information on the number of nonstrategic and reserve strategic weapons. We judge that the potential security benefits of exchanging these data outweigh the benefits of continued secrecy. Facility Inventories The next level of detail would disaggregate weapon inventories by facility, which would be necessary before states could consider the possibility of inspections or other measures to confirm the accuracy of declarations. At this level, states would declare the name and location of all facilities at which nuclear weapons exist, preferably by type and status. Facility descriptions and site diagrams also could be exchanged, indicating the location of all storage bunkers or other areas where nuclear weapons might be present. The United States and Russia have exchanged information at this level of detail for nuclear delivery systems under the INF and START treaties, in order to facilitate verification of agreed reductions and limits. Relevant facilities would include intercontinental ballistic missile (ICBM), submarine, and air bases; nuclear weapon storage facilities at other military bases; and facilities at which weapons are assembled, disassembled, or maintained. A declaration could be prepared for each such facility. For example, states could declare the total number of weapons of each type mounted on ICBMs or stored as spares at each ICBM base, the number stored or deployed on submarines based at each port, the number of active and inac-

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials tive weapons of each type in each storage facility, and so on. (For very small inventories, weapons in transit would need to be declared and controlled as well, since transport systems could in that case conceal a significant fraction of the stockpile.) Much of this information has already been exchanged by the United States and Russia, as part of the START Treaty. Going beyond START, they could share information on facilities where nonstrategic weapons are deployed, and the location and inventory of weapon storage, maintenance, and assembly-disassembly facilities. It might be argued that providing such information would allow these facilities to be targeted, but it is reasonable to assume that Russia and the United States have already identified and characterized each other’s nuclear weapon facilities, and the military significance of these facilities is much smaller than that of deployed forces in any case. Serious security concerns would arise if declarations significantly increased the vulnerability of a state’s nuclear forces to attack; for example, declarations that included the location of every nuclear weapon might prove destabilizing if potential adversaries became more confident in their ability to destroy opposing forces (and therefore be tempted to launch a preemptive attack during a crisis) or more afraid of preemptive attack (and therefore more likely to launch nuclear forces before they could be destroyed). This should not be a concern, at least for the United States and Russia, for several reasons. First, the United States, Russia, France, and the United Kingdom maintain submarines at sea (and, in the case of Russia, also mobile ICBMs) that cannot be targeted or confidently destroyed in a preemptive attack. Second, the United States and Russia also deploy silo-based ICBMs that could, at least in principle, be launched on warning of an attack, further inhibiting each other from contemplating a preemptive attack. Third, as discussed below (see “Secure Declarations”), the use of cryptographic tools could allow parties to maintain control over access to the information contained in the declarations, so that the location of every nuclear weapon need not be revealed in advance. Itemized Inventory The most detailed declaration would be an itemized inventory. This could take the form of a table with a row for each weapon and columns for the weapon’s serial number, type, date of assembly, status, and location. The table could be extended to include weap-

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials ficult to spoof than attributes, assuming that the reference objects are authentic. The attribute approach alone is unlikely to work well for nuclear weapons, due to the difficulty of specifying a set of attributes that would be displayed by all authentic weapons but not by any nonweapon objects. The attribute approach is particularly vulnerable to scenarios in which the inspected party creates a number of low-cost dummy weapons that display the selected set of attributes, which could be substituted for genuine weapons. The inspecting party might believe that the genuine weapons had been retired or dismantled, when only the dummy weapons had been dismantled and the genuine weapons were stored in a secret facility. Templates are better suited to the problem of identifying weapons. The primary disadvantages are the need to create a template for each weapon type, to have confidence that the reference object is an authentic weapon, and the need to securely store templates between inspections. The template approach could be vulnerable to scenarios in which a fake weapon is presented for templating, or in which the signatures of genuine weapons are modified or disguised so that they do not match any template. If parties wish to have high confidence that nuclear weapons are genuine and that other objects are not weapons, it probably will be necessary to combine the template and attribute approaches. Data gathered for a template, for example, could be analyzed to determine whether the object contains certain general characteristics or attributes of an authentic nuclear weapon, such as the presence of a certain minimum amount of weapon-grade plutonium metal and high explosive. In either approach, measurements performed for identification purposes may contain sensitive information about the design of the nuclear weapon or component; this is particularly true for gamma-ray measurements. For this reason, template and attribute identification systems are likely to require an “information barrier” to protect the measurement, storage, and analysis of sensitive data. Only the result of the analysis (e.g., “yes” or “no”) would be transmitted to the inspecting party. Information barriers are described in more detail in Box 2-4C. Before implementing an actual system employing templates or attributes, a critical review should be undertaken to confirm the extent of the real security concerns involved in order to avoid unnecessary complication in the equipment and procedures required. *   U.S. Department of Energy, Office of Nonproliferation Research and Engineering, “Technology R&D for Arms Control,” NNSA/NN/ACNT-SP01 (Livermore, CA: Arms Control and Nonproliferation Technologies, Spring 2001).

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials BOX 2-4A Template Identification The template approach works by measuring certain characteristics of an object and comparing them with the same set of measurements taken from a reference object (e.g., an authentic weapon of a particular type): the template. If the two sets of measurements match, one can conclude that the object is a weapon of that type. The set of characteristics included in the template could include various combinations of a weapon’s mechanical, thermal, electrical, acoustical, and nuclear properties, but most concepts have relied entirely on gamma-ray emissions. Template systems based on gamma-ray spectra have been used for decades by the United States and Russia to identify their own weapons and weapon components; in Russia these gamma-ray templates are called “radiation passports.” The key components of nuclear weapons contain NEM, which are radioactive. Most of these isotopes emit gamma rays at a particular set of energies, and the isotope can be positively identified by its gamma-ray spectrum. These gamma rays are scattered and absorbed as they pass through materials inside the weapon; the fraction scattered or absorbed depends on the composition and thickness of the material and the energy of the gamma ray. Some isotopes also emit neutrons that, when absorbed by certain materials, result in the emission of gamma rays at particular energies. These many interactions result in a spectrum of gamma rays of various energies and intensities outside the weapon. An example of such a gamma-ray spectrum, recorded with a high-resolution detector, is shown below. More than two dozen peaks can be identified at energies corresponding to the decay of particular plutonium and uranium isotopes or neutron interactions with particular materials. The intensity of the various peaks depends on the exact composition, geometry, and configuration of many weapon materials and components. Weapons and weapon components of different types produce distinguishably different spectra.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials Illustrative gamma-ray spectrum of a Soviet nuclear weapon. Adapted from: Steve Fetter et al., Science 248 (May 18, 1990), pp. 828-834. Using the template approach for inspections presents several challenges. The most fundamental challenge is establishing the authenticity of the template—that the template was produced using an authentic weapon or weapon component. If the inspecting party is allowed to select, from the complete list of declared weapons or components, one or a few of a particular type for the template, one could be fairly confident that the selected weapon or component is authentic. In the case of nuclear weapons, this is particularly true if the chosen weapon is operationally deployed. A less satisfactory but simpler procedure would have the inspected party present several weapons or components of a given type to inspectors, who would then choose one for the template. Either scheme could be subverted by manufacturing and substituting bogus devices in place of weapons of a given type (with the authentic weapons moved to secret facilities). Although this seems unlikely, one could protect against this scenario to some extent by using the attribute approach to ensure that the objects presented for templating meet certain criteria expected of all weapons. Another challenge is protecting the sensitive weapon design information that may be contained in the template. According to present security criteria, gamma-ray spectra would have to be pro-

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials tected at all times: when templates are produced, stored, and when they are used to identify objects during inspections. Sensitive data can be protected during template measurement and analysis using an information barrier (see Box 2-4C). Protecting templates between measurements is a special challenge. Templates most likely would be stored on a removable disk or computer chip. The disk or chip must be protected so that the inspected party could not alter the data without detection and the inspecting party could not access the data. This could be done by placing the disk in a safe that requires two combinations to open, one held by each side; additional protection could be provided by encrypting the data with a two-part cryptographic key, with one part held by each side. Alternatively, the inspected party could have sole possession of the template and the information barrier could provide the inspecting party with a digest or “secure hash” of the template when it is made and each time it is subsequently used (see Box 2-2). This digest would uniquely identify the template and unambiguously confirm its authenticity, but the inspected party could not derive any information about the template from the digest. The U.S. Department of Energy has sponsored the development of several template-measuring and information barrier systems. These have demonstrated that a system could be designed for monitoring purposes that would reliably identify weapons while preventing the release of sensitive information. Brookhaven National Laboratory developed the Controlled Intrusiveness Verification Technology (CIVET) system, which was demonstrated to Russian scientists in 1997. Sandia National Laboratories modified CIVET for use with the Radiation Identification System to produce the Trusted Radiation Identification System (TRIS). Tests conducted at the Pantex plant demonstrated that TRIS could reliably identify various types of weapons and weapon components (pits and canned subassemblies). A sample of the TRIS results for templates representing five weapon types appears in the table below. For each template/object combination, the table gives a statistic that measures the goodness of fit between the object’s spectrum and the template. If the object is the same type as that used for the template, the value of the statistic should be about one; a value below a threshold of two or three might be used to indicate a match, with a very low probability of a false negative. In tests the system correctly indicated a match every time the object was the same type of weapon used to

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials make the template, and correctly indicated no match in all other cases. Similar results were also obtained when identifying pits and canned subassemblies (CSAs), with one exception: the template system could not distinguish between components with very similar designs. This is not a problem in an arms control context, in which there is no need to distinguish between items that are almost identical. Active systems. Template systems based on intrinsic gamma-ray emissions—sometimes called “passive” systems—should work well for U.S. weapons, since they contain plutonium and uranium isotopes that emit penetrating gamma rays that are readily detectable and cannot easily be shielded. The penetrating gamma rays emitted by U.S. HEU are due to very small concentrations of uranium-232—an isotope that does not exist in nature but that is produced in nuclear reactors. U.S. HEU is contaminated with this isotope because uranium recovered from spent nuclear fuel was used to produce enriched uranium. It is believed that HEU in Russia and the other nuclear weapon states is similarly contaminated. It is possible, however, that weapons that contain uncontaminated weapon-grade HEU exist or might be assembled in the future. Because the low-energy gamma rays emitted by uranium-235 are readily absorbed by other weapon materials or easily shielded, the gamma-ray emissions from such weapons may be too weak to provide a useful template, if they do not include plutonium as well.   Template for Weapon Type Object A B C D E Weapon Type A, #1 0.8* 92 32 7.7 42 Weapon Type A, #2 0.9 90 31 8.2 45 Weapon Type A, #3 0.8 91 32 8.5 45 Weapon Type B 496 0.8 140 336 491 Weapon Type C 63 43 0.9 34 128 Weapon Type D 11 102 26 0.6 46 Weapon Type E 55 174 86 31 1.0 Pit, Type A 558 91 319 547 794 Pit, Type E 858 203 566 821 1071 CSA, Type A 52 118 88 64 66 CSA, Type E 27 156 77 22 6.4 * The “reduced chi-square” is a measure of the goodness-of-fit between the object’s spectrum and the template. The gamma-ray spectrum between 80 and 2,750 keV was divided into 16 groups (two of which are discarded) and the number of counts in each group for the object and the template was computed; the reduced chi-square is the sum over all groups of the squared difference in the number of counts for the object and template divided by the variance, divided by the number of degrees of freedom. Adapted from: D.J. Mitchell and K.M. Tolk, “Trusted Radiation Attribute Demonstration System,” Proceedings of the 41st Annual Meeting of the Institute of Nuclear Materials Management (Northbrook, IL: Institute of Nuclear Materials Management, 2000).

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials To deal with such problems and provide a more robust template, an external radiation source could be used to stimulate fission in the weapon’s plutonium and uranium components. An example of an “active” system is the Nuclear Material Identification System (NMIS) developed by Oak Ridge National Laboratory, which uses an external neutron source and time coincidence and correlation techniques to accurately characterize fissionable materials using the neutrons and gamma rays emitted during fission events. The external neutron source required is modest, with an emission rate only several times greater than that of a plutonium pit. In a blind experiment, NMIS correctly distinguished between 16 different types of weapons and components, demonstrating its usefulness for template identification. An information barrier would be needed, as the data would contain information about the mass and geometry of the nuclear components. The need for an external neutron source and the greater complexity of the required hardware and software, which would complicate authentication, makes it unlikely that such systems would be used for monitoring purposes unless templates based on intrinsic gamma-ray emissions were judged not to provide the required degree of confidence. Unclassified Templates. There may be several nonradiation types of measurements that could be used as templates to identify nuclear weapons. Templates that do not contain sensitive weapon design information would be useful, because they would eliminate the need for information barriers and would greatly simplify template storage and the certification and authentication of the measurement system. It may be possible, for example, to distinguish weapon types based on a combination of their acoustic, electromagnetic, and/or thermal signatures. Such alternatives have not received as much attention as radiation templates, probably because they are seen as easier to spoof.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials BOX 2-4B Attribute Identification The attribute approach works by measuring a set of characteristics or “attributes” that should be displayed by all items of a given general type. To satisfy a given attribute, measurements of a weapon or component would have to fall above an agreed threshold or within an agreed range of values. This approach works best for items with the same general composition and design; for example, all plutonium pits contain a certain minimum amount of weapon-grade plutonium metal in a symmetrical shape. The United States developed a prototype attribute system to confirm the authenticity of plutonium pits to be stored in a U.S.-funded facility at Mayak, Russia. This system used six attributes: The presence of plutonium; Weapon-grade plutonium (Pu-240:Pu-239 < 0.1) ; Plutonium age (separated prior to January 1,1997); Plutonium mass (> 0.5 kilogram); Symmetry of plutonium mass; and Absence of plutonium oxide (< 10 percent plutonium oxide). The first three attributes were measured with high-resolution gamma-ray spectrometry; for example, the ratio of the intensities of the 642.5 keV and 646.0 keV gamma rays emitted by Pu-240 and Pu-239, respectively, is directly proportional to the Pu-240:Pu-239 ratio. Similarly, the relative intensities of gamma rays emitted by the Am-241 decay products of 14-year half-life Pu-241 can be used to determine its age (the time elapsed since the plutonium was last chemically purified). Plutonium mass was estimated using the number of single, double, and triple neutron events from Pu-240 spontaneous fission recorded by a neutron multiplicity counter, together with the Pu-240:Pu-239 ratio determined by gamma-ray spectrometry. Plutonium symmetry was determined with a neutron multiplicity counter, by requiring that the number of counts in each of eight detectors was within 15 percent of the combined average. The absence of plutonium oxide (a surrogate for the presence of plutonium metal) was measured using data from both detectors. As with templates, an information barrier must be used to protect the sensitive information contained in these radiation meas-

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials urements (see Box 2-4C). In the system described above, the detectors and computers were housed inside shielded enclosures, all data were stored in volatile memory, and a “security watchdog” monitored the system for unauthorized access. The only output was a set of green or red lights to indicate whether the item being inspected satisfied each of the six attributes. This prototype system was demonstrated to a team of Russian scientists in the Fissile Material Transparency Technology Demonstration held at Los Alamos National Laboratory in August 2000. Although the system was developed to confirm the authenticity of pits, Russia now plans to store excess weapons plutonium in unclassified shapes at the Mayak facility. The attribute approach could be extended to other NEM components. Because HEU emits few high-energy gamma rays and almost no neutrons, passive radiation measurements are able to do little more than indicate the mere presence of HEU. As with templates, an “active” system would be necessary to provide a reasonable set of attributes for components that contain only HEU; for example, the Nuclear Material Identification System (NMIS) described in Box 2-4A could be used with a low-intensity neutron source to determine the mass and enrichment of HEU components. Applying the attribute approach to weapons is less straightforward because it is difficult to specify a single set of attributes that would be displayed by all possible types of nuclear weapons. Different classes of weapon may require different sets of attributes. For weapons that contain a plutonium pit, one might select attributes that indicate the presence of a certain amount of weapon-grade plutonium in a symmetrical shape surrounded by high explosive. The presence of high explosive could be indicated by the gamma rays that are emitted when neutrons are absorbed by nitrogen in the high explosive. As noted in the text, the attribute approach may be best applied to nuclear weapons as a complement to templates, to provide further confidence that items are genuine. A key difference between the attribute and template approach is that attributes cannot be used to identify the particular type of nuclear weapon or component. Whether this is an advantage or disadvantage depends on the nature of the transparency regime and whether parties wish to exchange and confirm declarations that include such details. A key advantage of the attribute approach over the template approach is that identification is based solely on measurements of the item under inspection and there is no need to store sensitive

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials data for later comparisons. On the other hand, the specification of each attribute must be known to all parties, which could touch on sensitive information, such as the minimum amount of plutonium in a pit. This requirement, together with the need to accommodate differences among the various types of weapons and components and the desire to prevent false negatives (i.e., an authentic weapon or component that fails to display a particular attribute due to measurement error), could lead to thresholds for attributes that are well below the average values for most weapons. The threshold for plutonium mass selected by the United States was set at 0.5 kilograms, for example, which is believed to be well below the average. These factors make the potential for false positives (i.e., a nonweapon object that displays the set of attributes for a weapon) a greater concern for the attribute approach.

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials BOX 2-4C Information Barriers A key challenge in implementing a template or attribute system is protecting any sensitive weapon design information that may be gathered during the identification process. This can be accomplished by automating the collection, storage, and analysis of data, and by making only the conclusions of the analysis available to the inspectors; for example, a computer could light a green lamp if an object’s gamma-ray spectrum matched the template or an attribute within specified tolerances, or it might report a summary statistic that measures the degree to which the object’s spectrum and the template matched. As illustrated in the figure below, the detector, computer, and template storage are housed inside an “information barrier” which prevents transmission of electronic signals or other surreptitious access to the sensitive data contained inside. Countries could build the systems used to inspect their own weapons, so they could be sure there were no hidden transmitters or storage devices or intentional flaws in the information barrier. Because an information barrier prevents access to the data and the analysis upon which the results of an inspection are based, the inspecting party must authenticate the system. The inspecting party

OCR for page 45
Monitoring Nuclear Weapons and Nuclear-Explosive Materials must be confident that the results produced by the system accurately describe inspected objects over the full range of possible objects and conditions, and that the system contains no hidden features that could interfere with or bypass the proper analysis or result (e.g., hardware or software changes that produce a green light during every inspection, or that allow the system to respond to remote commands from the inspected party). Authentication can be facilitated by cooperative design of measurement and information barrier systems; thorough documentation; the use of simple, commercially available hardware; and the documentation of all source code for system software. If these guidelines are followed, the system can be authenticated by thoroughly examining the hardware and software and confirming that they correspond to the documented design. The inspected party could build multiple identical units and allow the inspecting party to choose one for weapon inspections and another for detailed examination, including the removal of selected components for laboratory testing. After a system is authenticated, tamper-revealing seals can be placed in key locations to detect any attempt to alter the system. Proper operation of the system over a range of conditions can be checked using a variety of unclassified test objects, which could be provided by either party. As noted in Boxes 2-4A and 2-4B, prototype information barrier systems have been developed by the United States for template and attribute measurement systems, and their use was demonstrated to Russian scientists during the Fissile Material Transparency Technology Demonstration.