Nuclear treaty verification at AWE, Aldermaston

Transcription

1 Abstract Nuclear treaty verification at AWE, Aldermaston A Richings 1, S McOmish 1, P Thompson 1, 1 AWE, Aldermaston, Reading, Berkshire, United Kingdom Atomic Weapons Establishment provides technical expertise to the UK Ministry of Defence (MoD) on the verification of nuclear-related treaties or arms control agreements. This Nuclear Treaty Verification programme is divided into two parts: an element that monitors for nuclear testing in support of the Comprehensive Test Ban Treaty Organisation (CTBTO); and an element that conducts research to provide advice on the verification of potential future nuclear-related treaties or agreements. Under the first programme element, AWE provides the UK s national capability to monitor, analyse and advise on possible nuclear explosions, using radiochemical analysis and forensic seismology. The former in particular is relevant to Safeguards applications: as a thermal ionisation mass spectroscopy (TIMS) laboratory, AWE has been a member of the IAEA Network of Analytical Laboratories (NWAL) since 1996; and more recent capability developments at AWE in the fields of secondary ion mass spectroscopy (SIMS) and scanning electron microscopy (SEM) may in the future lead to an expanded contribution to NWAL. The second element of the Nuclear Treaty Verification programme, which looks to possible future verification requirements, provides advice to MoD on the options for conducting verification activities in future, and develops methods and technologies to support those activities. As with AWE s radiochemistry capability, some or all of these methods and technologies may be transferrable to Safeguards applications. This paper outlines AWE s views on some the most significant challenges associated with nuclear treaty verification, along with potential solutions that might meet those challenges. It also highlights some of the research activities within the AWE programme that are aimed at delivering particularly high priority capabilities. Our view on necessary areas for future research is also discussed.

2 1. Introduction AWE has a long history of providing technical expertise to the UK Ministry of Defence (MoD) on the verification of nuclear-related treaties or arms control agreements. Its modern Nuclear Treaty Verification programme is much broader than its historic counterpart, the locus of which was AWE s forensic seismology capability and is divided into two parts: an element that monitors for nuclear testing in support of the Comprehensive Test Ban Treaty Organisation (CTBTO); and an element that conducts research to provide advice on the verification of potential future nuclear-related treaties or agreements. Under the former element, AWE provides a national capability to monitor, analyse and advise on explosions through both seismic and radiochemical means. This capability also underpins the MoD s role as the National Authority for the CTBT, to assist in advancing international scientific and technical capabilities to monitor the Treaty. AWE has the responsibility to ensure that some facilities expected or required under UK treaty obligations to the CTBT are operated to the required standards. Under the latter element, AWE s Arms Control Verification Research (ACVR) programme provides both advice on potential future verification of nuclear disarmament issues and develops methodologies and technologies to undertake such verification. The Safeguards community is always in need of improvements to the techniques currently available, and in the development and validation of new techniques. It is rare that techniques are developed solely for the Safeguards community though, and in many cases new techniques are adapted from those developed for other purposes. With that in mind, this paper indicates areas of technical development at AWE conducted in relation to either the CTBTO or verification of potential future nuclear arms control agreements that may be relevant to Safeguards. It also lays out the broader scope of AWE s work in the latter context. 2. Support to IAEA Safeguards and the CTBTO AWE has been a member of the IAEA Particle Analysis NWAL since 1996 when it started making measurements using the Fission Track Thermal Ionisation Mass Spectrometry (FT-TIMS) method. In this method an aliquot, which potentially contains fissile particles, is spread as a removable film on a plastic substrate. The loaded plastic is then irradiated by thermal neutrons in a nuclear reactor. The neutrons cause fissile nuclei to fission, which creates

3 fission tracks in the plastic. On removal from the reactor the particle-bearing film is peeled off the plastic, and the plastic is then etched to enlarge the fission tracks so they can be seen under a microscope. Any observed fission tracks can then be related to a location on the removed film and the associated particle identified. This particle is then analysed on a mass spectrometer to produce uranium or plutonium isotopic ratios. The method is capable of producing very accurate isotopic ratios for selected particles, even when the particles of interest within a sample are scarce, although it is not particularly rapid. More recently AWE has also been analysing samples using Secondary Ion Mass Spectrometry (SIMS). In this method the sample of interest is instead spread onto a carbon substrate. The disc is then scanned by an oxygen ion beam. Any ionised uranium atoms are measured to produce isotopic ratios. The method is capable of producing major isotope ratios for many particles rapidly in timescales of the order of days, rather than weeks or months as for FT-TIMS. The next stage of research at AWE is to be the commissioning of a new SIMS mass spectrometer that may be able to approach the accuracy of FT-TIMS for many particles while retaining the speed of SIMS. In addition to these mass spectrometry techniques, AWE has capabilities in Ion Coupled Plasma Mass Spectrometry, Scanning Electron Microscopy and other analytical techniques that could potentially be applied to Safeguards problems. AWE s support to the CTBTO is provided in our capacity as one of 16 National Laboratories that are accredited to confirm results from the 80 atmospheric monitoring stations spread around the world. We are currently investigating methods of actively reducing gamma detector backgrounds using coincidence suppression techniques. Recently the AWE gamma spectrometry capability participated in an IAEA Bulk Analysis NWAL Proficiency Test exercise, and provided an excellent set of results. AWE also has the capability to measure radioactive xenon isotopes in air. Both the gamma spectrometry and xenon measurement capabilities regularly take part in CTBTO proficiency exercises. 3. Future treaty verification Since 2000 the UK Atomic Weapons Establishment (AWE), has supported a dedicated programme of research into technologies and methodologies that have potential to be used during the verification of possible future nuclear arms control agreements. This programme is in partial fulfillment of the UK s commitments under the Non-proliferation Treaty (NPT) article VI.

4 The programme looks at hypothetical bi-lateral or multi-lateral scenarios involving nuclear treaty verification. The focus has been on developing representative scenarios which follow the nuclear warhead from the arrival at an initial storage facility, through various transport and dismantlement phases, through to the storage of fissile components; in other words, the overall context in which our research has been pursued to date has been one of verified warhead dismantlement (figure 1). In these scenarios, technical verification would be undertaken by an inspecting party with reference to processes and facilities overseen by a host party. Figure 1 The outline scenario used to frame arms control verification research at AWE to date; other scenarios exist, and may result in different levels of emphasis on the various areas of research. Verification, in this context, is the process by which a level of confidence is reached in any declarations that have been made; ultimately this contributes to the judgement of whether the parties have adhered to the terms of any agreement. As such, the declarations made will constrain the limits of technical verification activity: what is declared can be verified, but no more. Therefore the hypothetical declarations considered during this work use the term Treaty Accountable Item (TAI) to refer to objects and materials potentially subject to verification, as these items could take many forms: a complete warhead; a partially dismantled warhead; a fissile component or fissile material from component destruction; or perhaps an object or set of objects containing no fissile material at all. A key driver influencing the design of the final negotiated verification regime, and the selection of technologies, is the differing perspectives of the parties involved. This leads to a major challenge for any verification regime operating

5 within a nuclear weapon complex: to provide the inspectors with the opportunity to gather sufficient evidence, whilst protecting sensitive or proliferative information held by the host. This may be particularly pronounced if the inspecting party includes personnel from Non Nuclear Weapon States (NNWS). The final verification regime must balance these two viewpoints and take account of any resulting constraints that may be imposed. 4. Inspection studies Any future nuclear arms control agreement will likely require verification measures that can demonstrate to some degree of confidence that activity has taken place according to the agreed treaty protocol. In general, the deployment of a team of inspectors to witness an activity, or post-activity outcome, is widely regarded as the most comprehensive method of doing this. This aspect of verification is commonly referred to as On-Site Inspection (OSI). OSI has been used extensively by the United States and Russian Federation under the START (Strategic Arms Reduction Treaty) and INF (Intermediate Nuclear Forces) Treaty. Conducting an OSI will normally require the host State to facilitate access to sensitive facilities for a team of inspectors, who will likely be from a foreign State (or States in a multilateral treaty). AWE has therefore run a number of exercises designed to examine inspections issue in detail, both in collaboration with international partners (the US, Norway) and indigenously. The AWE Inspection Studies programme is focussed on understanding the issues associated with this managed access of inspectors to nuclear facilities in the following areas: Understanding risks Negotiation of access Managing expectations of inspectors Route planning Escorting methodology Emergency planning Security and safety procedures Shrouding Handling questions and answers (inspectors to hosts) In addition to the issues involved with conducting managed access in the first place, inspectors will probably seek to carry out some form of technical measurement or verification activity within a facility. This presents us with the

6 problem of ensuring that these activities, if legitimate and mutually agreed, can be conducted in a way that reveals only the information to which inspectors are entitled under the verification protocol. Again, we must also ensure that proliferative information is not revealed, and that sensitive information is only revealed in a way that is planned, controlled, and authorised by the national authority. We therefore need to understand in detail the nuclear signatures that might be detectable within each facility that inspectors visit, and what this might reveal about current and legacy activities conducted therein. As part of the Inspection Studies programme we undertake a programme of facility monitoring in order to characterise some of these signatures. This in turn allows us to assess potential risks to inform the MoD and other policy makers in the UK government. 5. Chain of custody Chain of custody is an integrated series of procedures and technologies designed to provide: Evidence that an uniquely identifiable item has been initialised into a verification regime; Evidence that the TAI is progressing through the stages of the process as anticipated by both parties; Evidence that the TAI has, if required by the treaty, been accepted into a long-term storage scheme; and Evidence that the TAI has not at any point, after initialisation into the verification regime, been subject to any nefarious act by either party. Multiple technology options are required to provide a flexible response to different scenarios (even within the overall verified warhead dismantlement setting that has guided our work to date), to provide layered defence against tampering and evidence of good faith, and to avoid the vulnerability of a single point of failure. There are commercial solutions to some of these issues and the IAEA Safeguards programme has itself developed many potentially applicable solutions, so there is clearly significant overlap between the solutions for nuclear treaty verification and for Safeguards applications. There are, however, some constraints which are unique to their respective applications and these unique qualities will have constrained and directed the implementation of technology. Therefore, not all technologies are mutually compatible with IAEA Safeguards and nuclear treaty verification applications. As science advances, and potentially creates more sophisticated modes of attack, existing technologies must be reviewed and additional technologies

7 considered. It is therefore clear that no single technology can solve the chain of custody problem in the long term. The chain of custody programme at AWE is looking to build capability to provide each of the four kinds of evidence indicated above, now and in future. Links with academia are critical in this context, as we see this as the best way to ensure that novel developments are identified and integrated with our programme in a timely manner. This also allows us to support research which will ultimately allow technology providers to respond effectively to potential future treaty scenarios. 6. Device and material monitoring As noted above, inspectors overseeing the verification process will be looking to obtain a level of confidence that the object presented is as described in a given declaration. It is likely that technologies could be deployed which measure properties associated with the TAI in order to do this. This might involve - at the initialisation stage - looking for the presence of fissile material and explosive material in a configuration consistent with a nuclear weapon. After dismantlement these measurement requirements would naturally change, as the resulting items would no longer be in such a configuration, However, the verification process must also account for the obligations of the host to protect sensitive and proliferative information, and of the inspector not to seek or receive such data. For example, information related to masses, fissile material isotopics and configuration of specific devices might be considered to be sensitive [1]. This will not only impact on the design of measurement technologies, but also on the amount of information that can be revealed within any declaration. Furthermore, it is unlikely that the inspectors would be allowed to directly observe sensitive items if they are allowed to observe any items at all and are more likely to be presented with containerised objects. The device and material monitoring programme at AWE therefore involves research designed to: Identify attributes that would allow for verification measurements to be made whilst protecting national security and proliferative information; Evaluate technologies which might be used to verify declared attributes of a TAI, particularly during initialisation; Explore techniques and methods for protecting sensitive and proliferative information during the monitoring process; Study the impact of potential content of declarations, related to the TAI, which might be made in a treaty scenario; and Develop methodologies for building confidence in the equipment deployed as part of the verification regime.

8 There are clear challenges to developing high confidence in verification measurements under the constraints outlined above, especially when discussion of specific details of objects or facilities might be impossible. Our current approach is to develop a limited set of unclassified attributes that are consistent with an unclassified, non-sensitive definition of a nuclear weapon. These attributes are then incorporated into measurement systems that could be used within a potential verification regime. There are additional complexities associated with deploying equipment in the context of nuclear treaty verification, particularly when making measurement on nuclear weapons. Both parties must be confident in the equipment, in terms of its functionality but also that it only operates as expected and agreed (in other words, that no undisclosed or unexpected functionality is present). They must be confident that equipment has not been tampered with. Some of this is achieved through robust chain of custody of the measurement equipment, but a more complete solution requires both parties to be confident in the design, build and assembled state of the equipment as well. This necessitates a process we term authentication, an analytical process comprising both non-destructive and destructive elements to ensure that equipment can be characterised to the component material level if required, and its full functionality completely understood by both parties. 7. Support and advice The task of bringing all of these elements inspection studies, chain of custody, and device and material monitoring together, to provide integrated advice to the UK government on nuclear verification matters, falls to the Support and Advice area of AWE s arms control verification research programme. By considering plausible treaty and protocol scenarios for nuclear-related verification, alongside the results of technical development and assessment conducted under the areas discussed above, this section of the programme also develops AWE s capability to provide such advice in future. This work on scenario development is important because it helps to set the context and constraints by which technically focussed work must be bound. To date, most of AWE s practical work has been set against the hypothetical context of monitored warhead dismantlement. This has also been the context for much of our collaboration with US and Norwegian partners. Other plausible scenarios might include verification of absence of warheads, for example, or direct warhead counting and tracking; each of these would bring some unique challenges, as well as some that were shared with other scenarios.

9 A further strand of this work involves research into decision support tools that might be used to improve the rigour of any advice that AWE provides. This research explores options for formalising and quantifying our ability to assess the value of verification regimes in terms of the benefits they might provide set against the risks they might incur (of release of proliferative information, for example, or of information relevant to national security). At a more granular scale these tools might also help us to predict or assess the effectiveness of different technological solutions to verification problems. After all, if the purpose of deploying a particular technology a tamper-proof CCTV camera, for example is to enhance our confidence that a treaty-relevant boundary remains intact, it would be useful to be able to compare the effectiveness of that technology with tamper-indicating seals, or process monitoring, or with no monitoring at all. This might allow us to develop measures of effectiveness that could be set against constraints such as cost and negotiability. Our current focus for research in this area involves the application of Bayesian methods, especially Bayesian Belief Networks, to treaty verification problems. These networks are made up of nodes that represent statements about the world (for example, the inspected object is a nuclear weapon ) and our degree of belief in those statements, while the links between them indicate some form of causal relationship. Bayesian inference is used to update the probabilities attached to various nodes, given a series of observations. This approach offers potential in several areas: To compare information from different sources in a quantitative way; To represent our assumptions explicitly and to help avoid cognitive bias; and To analyse the sensitivity of verification conclusions to different kinds of finding. These methods can become computationally complex, and the generation of prior probabilities (the degree of belief in a proposition before any evidence is received) and can be a source of difficulty. However, they are also intuitive, powerful, capable of incorporating objective and subjective data, and useful for both diagnostic and causal reasoning (i.e. from effects to causes and vice versa). Importantly, they are also theoretically applicable on all scales of interest, from the macro-scale problem of regime assessment to the microscale problem of technology assessment. Our ongoing programme of work in this area is designed to explore the feasibility of their application in practice. A cross-cutting issue for each of these elements of the research programme is what exactly it means to have confidence in a verification process, whatever

10 that process may be, and how much confidence is enough for each party involved. British and Norwegian experiences of managed access exercises suggest that the tone of interactions between participants could be at least as important in determining the success of a verification process or at least the subjective levels of confidence that participants report as their technical content. Therefore our current exercise programme is designed to explore just how important the human factor is. In collaboration with academics from King s College London, UK and Norwegian researchers put groups of students through carefully controlled exercises that allow us to track the factors that increase and decrease inspector confidence in a host party. The educational and outreach value of this element of the UKNI is significant, with groups from all over the world taking part and learning about some of the issues that might be involved in nuclear arms control verification; the research value lies in determining how best a verification interaction might be structured to develop the confidence of the parties involved. 8. Conclusions This paper has outlined areas in which AWE already provides technical support to multilateral verification organisations and mechanisms IAEA Safeguards and the CTBTO as well as its more future-oriented arms control verification research programme. Although the scope of existing capabilities at AWE is broad, it is clear that the spectrum of problems that might be posed by nuclear treaty verification in future is at least equally so; collaboration with others is therefore an important part of AWE s approach to these issues. This can involve academic collaboration, industrial collaboration, or inter-state collaborations. Both Nuclear Weapon States (NWS) and NNWS have a role to play, and the UK and Norway have demonstrated that it is possible for a collaboration to work between the two. For their part it would be useful for NNWS to understand and define the level of confidence that they require in future verification regimes of various types, and define more precisely the capabilities that they need to build in order to facilitate their participation. NWS parties, on the other hand, need to identify how this capability development can be supported while remaining conscious of proliferation risks. Technical collaborations may be most feasible on areas related to chain of custody, as this avoids the most sensitive and potentially proliferative issues associated with initialisation [2]. This is also an area in which Safeguards-related technology [3] may find application. After all, the continuity of knowledge problem is not so different

11 from the chain of custody problem. However, the challenges associated with authenticating data and equipment in the context of scenarios relating to nuclear weapons, not simply nuclear material, may be a step beyond those envisaged in a Safeguards-like scenario. Greater tailoring to facility constraints might be required, for example, and considerable thought would need to be given to data management (in particular for active systems) with respect to the constraints of the NPT and the risk of inadvertent proliferation. 9. Reference [1] Allen K (et al.), UK-Norway Initiative (UKNI) approach for the development of a Gamma Ray Attribute Measurement System with an integrated Information Barrier, Proceedings of the 34th ESARDA Symposium On Safeguards and Nuclear Material Management, May 2013 [2]UKNI, The United Kingdom-Norway Initiative: Research into the Verification of Nuclear Warhead Dismantlement, NPT Review Conference, 2010 [3] Tolk K, Authentication Issues in Safeguards, IAEA-CN-184/175