Proliferation Risks of Spallation Neutron Sources
Matthias Englert
The acquisition of Nuclear Weapon Usable Materials (NWUM) is, metaphorically speaking, the eye of a needle through which any state or sub-state actor must go before they can construct a nuclear weapon. The material side is therefore one of the most effective ways to restrict access to nuclear weapons or close it altogether. The relevant NWUM in the context of horizontal proliferation are uranium (U-233, U-235) and plutonium (all isotopes, in particular Pu-239). In advanced weapons programs (vertical proliferation), tritium is also important to build so-called “boosted” designs that enhance the efficiency of a nuclear bomb.
Two pathways to acquisition of NWUM can be identified. On the one hand, there are huge existing civilian and military stockpiles of highly enriched uranium (U-235) and plutonium. On the other hand, an actor could use already existing or potential future technologies to produce NWUM. Since plutonium, U-233, and tritium can be produced by neutron reactions, all neutron-producing technologies (including spallation neutron sources, SNS) are capable of producing these materials in principle.
One strategy to prevent access to sensitive production technologies would be international control of existing technologies and an early analysis of new or novel technologies which could facilitate or open new paths to access weapon materials. An example of the latter is an investigation of proliferation risks associated with SNS. The question is what quantities of NWUM can be produced within a specific time with a specific technology? And what would concrete production scenarios look like?
If the quantitative approach shows that a SNS is sensitive to the proliferation of NWUM, the ideal goal would be to develop alternative, proliferation-resistant SNS designs, which would permit only civilian use. Less ideally, the design would prolong the time needed to divert or produce NWUM considerably and increase the expenditure for an actor. It appears realistic to analyze spallation technology, sensitive plant parts and parameters, materials, as well as purposes of use, in order to build international control mechanisms for such facilities and export controls.
How much NWUM is needed?
To answer the question of whether a technology is sensitive to proliferation requires a quantitative definition. First of all, the mass which is necessary to build one nuclear warhead should be considered significant. This mass depends on the technological capability of the actor and on the explosive force of the weapon. An actor with a low technical capability would need 4-5 kg Pu-239 to build a small weapon with a yield of 5-10 kilotons.[1]
In order to assess the proliferation risks of SNS, it is useful to look at production rates, because not only the mass but also the time required for production plays a significant role. For this article, a maximum production time of 10 years and a minimum mass of 5 kg Pu-239 were chosen as reasonable extremes. Accordingly, facilities with a minimum production rate of 500 g/y Pu-239 should be considered sensitive.
It is obvious that this threshold is arbitrary and could be defined differently. Reactor Safeguards of the International Atomic Energy Agency (IAEA), for instance, are activated at a production rate of 100 g/y Pu. However, this would correspond to a production time of 50 years in order to produce enough plutonium for one nuclear weapon. In the opinion of the author the value of 500 g/y Pu-239 gives a conservative, practical, and adequate threshold for a first assessment.
Spallation Neutron Sources
In a SNS, protons are accelerated to energies of several hundreds of mega-electron-volts or several giga-electron-volts GeV (for protons, 1 GeV corresponds to 87 percent of the speed of light). At the end of the accelerator, the protons hit a heavy metal target. The reaction of the incident proton with the nucleons of an atomic nucleus in the target material can be described in terms of an elastic particle-particle collision (intranuclear cascade). The energy of the proton (~GeV) is high compared with the binding energy of one nucleon in the atomic nucleus (~8 MeV). An incident proton can therefore knock out or spall off several nucleons from the nucleus.
As a result, the atomic nucleus is highly excited and evaporates neutrons (and other particles) to relax to a less excited state. The nucleus cools down. The incident proton or a spalled particle can have enough energy to cause one or several spallation reactions in other nuclei (internuclear cascade).
In fissile materials, the evaporated neutrons cause additional nuclear fissions, which increase the number of produced neutrons further.
The neutron yield, that is the number of neutrons produced per proton, is one of the most important parameters of a SNS and depends on the energy which the incident particle projects on the material, the geometry of the target, and other parameters. A value of 40 neutrons per proton is typical (target is a cylinder, natural uranium, R=5 cm, L=60 cm, 1 GeV protons). The flux of the neutron source (neutrons/(cm2s)) an essential parameter of the performance of a SNS depends in turn directly on the neutron yield and on the current of the accelerator (the number of the protons in the beam per unit of time).
Linear accelerators (LINAC), cyclotrons, or synchrotrons are usually used for the acceleration of the protons. Typical accelerator parameters used in existing SNS are currents of 0.5-1.8 mA and energies of 0.5-1.5 GeV. The energy of the incident particle is directly connected to the dimension of the facility the higher the energy, the bigger the accelerator. A linear accelerator is typically several hundreds of meters long. Cyclotrons are considerably more compact and have a diameter of several meters up to 20 m. The current of an accelerator is (together with other parameters) an indication for the technical sophistication of the accelerator.
SNS the Neutron Source of the Future?
The proliferation risks of SNS have so far been considered marginal, therefore no SNS specific safeguards and control mechanisms are in force. Plutonium was and is produced almost exclusively in reactors, the proliferation risks of which are widely known and very well examined. Accordingly, whole range of safeguards as well as export controls exist for reactors. The production of plutonium in SNS, in contrast, was regarded as a technical challenge because of its use of high current, high energy accelerators. It is also a cost-intensive process as compared to reactor production. Presently, there are only about ten SNS facilities in operation,
I will argue that this picture of the proliferation risks of SNS can change because of four different but interconnected arguments regarding the future of accelerators generally and of SNS facilities specifically:
The technological dynamics of the accelerator evolution in the last 20 years led to an increase of accelerator currents by a factor of 10. While in the 1980s cutting-edge accelerators for SNS had currents of several 100 μA, today accelerators with currents of 1-2 mA are in operation and for the low-energy range already commercially available,[2] a further increase of accelerator currents up to 10 mA in the next 10-20 years can be assumed, and the operation of prototypes (LINAC-RFQ) with up to 100 mA has already been successfully demonstrated.[3]
In assessing the proliferation risks of SNS, a current of several hundred μAcan be considered a conservative assumption. 1-2 mA is the current of state-of-the-art machines and 2-10 mA would be a progressive assumption for an accelerator current.
Due to the dynamics in the accelerator development, SNS will complement and/or replace the role of reactors for the generation of neutrons for scientific purposes. This in turn affects positively the dynamics of accelerator development. The achievable neutron peak fluxes of SNS are already higher than those of reactors. Although the average neutron flux of reactors is still superior to spallation sources, SNS have considerable potential for future developments in neutron performance.
Since the 1990s, new plans for the generation of electricity with accelerator driven systems (subcritical reactors, etc.) and for the transmutation of radioactive waste experienced new impetus, although such interest is restricted to research and design so far. But if any state were to launch a program to build such a plant, this will surely have a heavy impact on the development and commercialization of spallation and accelerator technology.
There is an increasing interest in the development of accelerators for industrial and medical applications and for the commercialization of accelerator technology.
These four points indicate that there is potential for the spread of scientific and commercial accelerators and SNS facilities, as well as for their improvement. Additionally, the presence of nuclear safeguards on other relevant processes make the potential use of SNS to produce NWUM very attractive to an actor wanting access to NWUM. This is a path which is technically and economically challenging, but offers the possibility of producing NWUM without immediately alerting international attention.
Whether this picture of the dynamic development of accelerators and SNS becomes reality in the future remains yet to be seen, but it seems very appropriate to investigate the proliferation potential and risk of SNS. Before looking in more detail at possible scenarios and actors, we should answer some quantitative questions first.
Plutonium Production with SNS
Even with rough approximations it can be shown that a significant production of plutonium in the kilogram range is a possibility with a SNS.[4] However, in order to assess more concrete and realistic production scenarios, it is necessary to look in more detail at the impact of relevant SNS parameters on the production rate. These are, among others, the current of the accelerator, the energy of the incident particle, the target dimensions, the beam profile, the burnup of the target, and the energy deposition. A simple cylindrical geometry (respectively a sphere) was used for the investigation of the parameter dependencies and their influence on the plutonium production, as well as for the determination of the simulation error in comparison with experiments. To simulate the proton bombardment of natural uranium and the neutron transport, the MCNPX computer code was used.[5]
The results presented below are limited to a few selected parameters and should give an impression of what conclusions can be drawn from such a parameter study using mathematical methods. The following underlying assumptions were made for all calculations:
The influence of the burnup is neglected, except for the investigation of the burnup itself. This corresponds to the assumption that the produced plutonium is continuously extracted from the target.
A continuous current in the accelerator was assumed.
The production rates were calculated for a continuous current without beam interruptions through maintenance, target replacement, etc., which typically occur during the operation of a SNS (i.e. a duty factor of 100 % is assumed). Depending on the reliability and the maintenance periods of a specific SNS, the values must be adapted.
Fig. 1: MCNPX calculations (Englert, 2004)[6] of the Pu-239 [kg/y] production rate in a quasi-infinite sphere of natural uranium (R=500 cm) with an isotropic point source in the centre, in dependency of the proton energy. The solid line corresponds to a current of 5 mA, the dashed lines to a current of 1 and 10 mA, respectively.
Maximum production: Figure 1 shows the annual plutonium production in a quasi-infinite sphere of natural uranium with an isotropic proton point source in the centre in dependency of the proton energy. The simulation in an infinite sphere indicates the maximum attainable production rate by direct bombardment of a uranium target under the simplifying assumptions (above), since no neutron losses occur through the surfaces and every neutron will be captured. With a 5 MW (1 GeV, 5 mA) beam, a maximum production of 32.24 kg Pu-239 per year is possible. This corresponds to a very big SNS facility and a progressive assumption for the beam current. The dashed lines indicate the production for a proton current of 1 mA (moderate) and 10 mA (progressive). It can be seen that above 400 MeV the production rate depends linearly on the proton energy. As the current is directly proportional to the number of protons bombarding the target, the production rate depends linearly on the current as well.
Energy deposition: Energy deposition does not have any direct influence on the production rates, but it determines the geometry and dimensions of the target in large part, and therefore limits the accelerator beam usable with a certain geometry. This means the energy deposition limits the achievable production rates. In reality, the usage of a compact uranium cylinder is not possible, as the energy deposited by a beam of several hundreds of kilowatts (or even MW) will melt such a target. However, the problems of heat transportation are solvable in principle (with specific limits of course) by using more complex geometries with sophisticated cooling techniques. It can be shown that the error of using compact cylindrical geometries in the computer simulation is not very big in comparison with real target constructions.
Burnup: With regard to the production of NWUM, it has to pointed out that the explicit calculation of the burnup in the simulation shows that after 500 days burnup (1 GeV protons, 5 mA current, cylinder of natural uranium, R=60 cm, L=60 cm) the plutonium produced consists of 99 % pure Pu-239, which is ideal for building nuclear weapons.
| Cylinder, R=90 cm, L=90 cm, m=43.6 t U-238 | ||||
|---|---|---|---|---|
| 0.5 mA | 1 mA | 5 mA | 10 mA | |
| 500 g/y Pu-239 | 438 MeV | 342 MeV | 187 MeV | 150 MeV |
| 2 kg/y Pu-239 | 1071 MeV | 684 MeV | 312 MeV | 255 MeV |
| 5 kg/y Pu-239 | >1500 MeV | 1253 MeV | 483 MeV | 341 MeV |
| Cylinder, R=25 cm, L=60 cm, m=2.2 t U23-8 | ||||
|---|---|---|---|---|
| 0.5 mA | 1 mA | 5 mA | 10 mA | |
| 500 g/y Pu-239 | 609 MeV | 428 MeV | 216 MeV | 162 MeV |
| 2 kg/y Pu-239 | >1500 MeV | 924 MeV | 380 MeV | 283 MeV |
| 5 kg/y Pu-239 | >1500 MeV | >1500 MeV | 609 MeV | 428 MeV |
| Cylinder, R=10 cm, L=40 cm, m=0.24 t U-238 | ||||
|---|---|---|---|---|
| 0.5 mA | 1 mA | 5 mA | 10 mA | |
| 500 g/y Pu-239 | >1500 MeV | 1078 MeV | 396 MeV | 289 MeV |
| 2 kg/y Pu-239 | >1500 MeV | >1500 MeV | 910 MeV | 587 MeV |
| 5 kg/y Pu-239 | >1500 MeV | >1500 MeV | >1500 MeV | 1078 MeV |
Tab. 1: Required proton energies for the production of 500 g, 2 kg, or 5kg Pu-239 per year for different accelerator currents and cylinder dimensions according to calculations with MCNPX (Englert, 2004).
Energy, current and target dimensions: Energy, current, target geometry, and the dimensions of the target determine the production rates to a large extent. The energy of the protons determines the size of the accelerator. Typical sizes are several hundreds of meters length for linear accelerators (LINACs) and 2-20 m for circular accelerators (cyclotrons). As indicated above, the current at energies above 150 MeV can give an impression of the technical sophistication of the accelerator and of the target design.
Table 1 shows the necessary energies of the protons and the currents of the accelerator bombarding the base face of a natural uranium cylinder to produce Pu-239 at a constant rate of 500 g/y, 2 kg/y, and 5 kg/y Pu-239. The more current, the less energy is necessary to achieve the constant production rate and vice versa. Low current and low energy can be compensated through enlargement of the cylindrical target (and with that the minimization of neutron losses through the surfaces).[7]
Even with moderate currents of 1 mA (state-of-the-art technology) and protons with an energy of 428 MeV bombarding a mass of 2.2 t U-238, more than the significance threshold of 500 g/y Pu-239 is produced. It should be pointed out that a mass of 2.2 t natural uranium is low compared to the uranium inventory of a small reactor core.
If one takes into account that the currents of accelerators could reach 10 mA or several 10ths of mA in the next 10-20 years (progressive assumption), a significant production of NWUM could be reached with very small SNS facilities, even with an non-optimized cylindrical target.[8]
Realistic Scenarios
As mentioned above, the use of cylindrical geometries is suitable for analytical purposes, for the investigation of the dependencies of the relevant parameters, and for the validation of the simulation model. With relatively simple geometrical changes, the production rates could be increased. Investigating the direct proton bombardment of uranium does not cover those cases in which the neutron producing target does not consist of uranium but of another heavy metal and where the target is spatially separated from the breeding blanket in which the plutonium is produced. In such scenarios, the neutrons could also be moderated before reaching the plutonium-breeding uranium blanket. This possibility involves all kinds of subcritical reactor designs.
Realistic scenarios involving a SNS for the production of plutonium depend considerably on the intentions of an actor and the initial purpose of the facility. The actors could be states, e.g. members of the NPT, trying to cheat and get clandestine access to NWUM.
Classification of Production Scenarios
In order to categorize the great number of possibilities to produce plutonium with a SNS, a rough classification was developed following two simple criteria: The plant is
a converted research neutron source,
particularly designed for the production of plutonium.
The first point refers to a state that might build a civilian research facility which will not only be used for scientific research but also to covertly produce NWUM. A historic reference might be the first research reactor delivered to India in 1955, which led to the “peaceful” explosion in 1974. The second point refers to a state trying to produce NWUM in a completely clandestine program. A historic reference might be the calutron program of Iraq, which was discovered in 1991 after the first Gulf War. The second classification is of technical nature.
The neutron producing target is identical with the breeding blanket where plutonium is generated, or
the neutron producing target is spatially separated from the plutonium producing breeding blanket.
Detailed calculations were not carried out for all scenarios since some can easily be excluded with rough estimates or have already been investigated by other authors (see below).
The combination chosen for this study therefore leads to four main scenarios (1a, 1b, 2a, 2b):
A converted neutron source:
The target material is uranium and is used for the production of NWUM. Example:
1a-1 Research spallation neutron source with a uranium target.
1a-2 Research spallation neutron source with a uranium beam dump.
Uranium breeding blanket separated from the spallation target. Example:
1b-1 Research spallation neutron source with irradiation positions for a uranium blanket in the moderator or near the target.
1b-2 Research spallation neutron source with use of the beam hole neutrons to irradiate a uranium blanket.
SNS specifically designed for the production of NWUM:
Target and blanket identical. Example: Small (commercial) cyclotron accelerator facility, which can be camouflaged.
Target and blanket separated. Example:
2b-1 Spallation target to produce neutrons surrounded with a uranium mantle.
2b-2 All kinds of moderated facilities, especially subcritical reactors driven by a spallation target.
Scenario Likelihood
At this point it is necessary to explicitly state, that the author does not presume, investigate, or fear a military use of existing research facilities like the Swiss Spallation Neutron Source SINQ at the Paul Scherrer Institute, Switzerland, or the pulsed neutron and muon source ISIS at the Rutherford Appleton Institute in Great Britain. These facilities are as examples for the examination of hypothetical proliferation scenarios because there exists vast experience in the operation of these SNS and data is publicly available in abundance.
1a-1: In a research neutron source, a uranium target could be installed and used for plutonium production. Using the results for the cylindrical geometry, approximations for existing facilities give production rates of 90-360 g/y Pu-239 for SINQ (1.8 mA, 590 MeV, 17x17x50 cm, rod bundle target) and 16-64 g/y for ISIS (0.2 mA, 800 MeV, cm 11x11x32 plate target). Bigger target dimensions or a higher energy or current would increase the production. Scenarios of converted research SNS using a uranium target are relevant for high beam currents and/or energies, as the target dimensions are usually small to get high neutron leakage currents from the target surfaces for scientific research.
1a-2: In every accelerator there is a beam dump for the maintenance and construction of the plant which could be used for plutonium production. Detailed calculation of a MCNPX model for the planned 500 kW European Spallation Source (ESS) beam dump result in a production of 353 g Pu-239 per year. The beam dump is very relevant if covert production of NWUM in a research SNS is considered. The calculated 353 g/y 239-Pu of course is below the threshold , but a change in geometry to get higher production rates is within the realm of possibility, as well as the construction of a beam dump for higher beam powers or with better cooling mechanisms. A beam dump constructed from uranium for the full 5 MW beams of the ESS would produce 3.53 kg 239-Pu per year, and better designs are both possible and feasible.
1b-1: This scenario is unlikely as the irradiation positions are usually very small. As it is possible to build targets with big irradiation positions in the moderator this question should be investigated in more detail. But production rates will be small even for very large SNS, except if huge masses of materials can be irradiated.
1b-2: This scenario is very unlikely as neutron flux is orders of magnitude too low to reach the production threshold.[9]
2a: An actor could construct one or several small covered production facilities with a small accelerator, e.g. small cyclotrons. The production rates of the cylindrical geometry can be used in order to estimate production rates. Even with an accelerator with a current of 1 mA and an energy around 400 MeV a production rate of more than 500 g/y is possible. With simple measures like a reflector mantle and a beam hitting the centre of a target instead of the base, the production rate can be doubled as compared to the simple cylindrical geometry (i.e. roughly doubled production rates in Table 1 except for low energies).
2b-1: More sophisticated designs like the split-target concept could be used to address heat problems (e.g. neutrons produced by tantalum plates with good heat performance). Production rates will be similar to 2a.
2b-2: A variety of designs with a neutron producing spallation target and a subcritical reactor assembly surrounding the target is possible. A study by Oak Ridge National Laboratory shows that even with a moderate SNS (e.g. 1 mA, 400 MeV) and 10 t of uranium, subcritical reactor production rates of 2 kg/y Pu-239 are possible. The significant threshold of 500 g/y can be reached already with a conservative current of 0.25 mA and energies around 500 MeV.[10]
Conclusion
The investigation of the general case of a cylinder clearly showed that significant production of NWUM with SNS is possible and maximum production rates up to 30 kg or more are possible with the parameters of state-of-the-art SNS, which are however still very big and expensive. Lower, but still significant, production rates can be achieved even with moderate accelerator parameters.
The investigations of the cylindrical geometry also showed that the isotopic composition of the produced plutonium is very suitable for nuclear weapons. Additionally, the mass of uranium needed as source material is very low in comparison with reactors, and both natural uranium and depleted uranium can be used as source material.
Due to the high dependency of the production rate on all the discussed parameters it is difficult to make general statements for production in real facilities, except that a significant production is possible. To quantify the potential production, it is necessary to consider the concrete design of a facility.
The classification of scenarios in this study tried to consider as many scenarios as possible. But this classification is still very rough in this study. It caused difficulties and should be improved for future investigations.
From all scenarios, only scenario 1-b2, the utilization of a beam hole, is negligible. For all other scenarios, the possibility to produce nuclear weapon relevant material (Pu-239) in significant quantities (500g/y) exists, in some scenarios (2a, 2 b) even with relatively small accelerator facilities. These scenarios, with facilities designed for the purpose of plutonium production, are also those with the largest proliferation potential in the view of the author. These scenarios do not only have a great flexibility for different designs; the use of very small and cheap accelerators (possibly in greater number) is also possible and many of the components needed are already or will be commercially available.
Together with existing dynamics in the development of accelerator technologies and SNS which can be expected in the next years and decades, the quantitative results show that it is essential to closely monitor such future developments.
Recapitulating, the danger of proliferation from SNS is perhaps not an immediate but is an impending danger, depending on the future development of the technology. Since the technology is just beginning to become commercially available and the spread of this technology has only just started, it is possible, while we still have enough time, to implement control mechanisms (proliferation-resistant design, safeguards, export controls, etc.) in the early stages of such developments to assure exclusively peaceful uses of spallation neutron sources.
Acknowledgement: Work on this Project was generously funded by the German Foundation for Peace Research – Deutsche Stiftung Friedensforschung (DSF), Osnabrück.
