To produce spallation neutrons effectively requires protons of energy ~ 1 GeV. At this energy each proton can produce about 30 neutrons, depending on the target geometry. Increasing the energy increases this number more or less proportionally, and as it is easier to produce high currents than high energies (though this can be argued) this would not be cost effective. At lower energies the law of diminishing returns sets in rapidly. There is some scope for somewhat lower energy, but not much. The actual energy will need to be optimised carefully.

Beam CurrentsEdit

The currents required will be of the order of tens of milliamps. This is a lot for an accelerator.

Beam ParticlesEdit

Beam particles other than protons are not impossible, but it appears unlikely they would give any advantage. Electrons are easier to accelerate but give a very low neutron yield. Heavier nuclei can give a few more neutrons, but not enough to repay the extra work involved.

Protons can be accelerated as bare H+ ions, as negative H- ions, or as H2 + pairs. The advantage of the latter two is that one of the extraction (or injection) stages required to get the beam into and out of an accelerator can be done by stripping the electrons.

The disadvantage is that the binding energy of the electrons is not large, and a large electric field (from an accelerating RF cavity) or a large magnetic field can remove the electrons at places where this is not desirable. The figure

H- lifetime in electric field

shows the H- lifetime vs electric field as measured at the Rutherford Lab during the design of the TRIUMF cyclotron. (taken from G.M. Stinson, W.C. Olsen, W.J. McDonald, P. Ford, D. Axen, and E.W. Blackmore, Electric Disassociation of H- Ions by Magnetic Fields, Nucl. Instrum. and Methods 74 (1969) 333.: thanks to Mike Craddock for drawing my attention to this.)

When the H2 + pair converts, the proton current is double the ion current, but the energy is halved.


The reliability that will be required of an accelerator in an ADSR is well beyond anything that has been achieved to date. Today's proton machines are designed, usually at the forefront of technology and with a very constrained budget, to serve a user community which can tolerate downtime. But a power station has to be up and running all day every day, and the safety mantra that when the beam stops, the reactor stops,now works against us. If a power station drops out of the grid it loses money - and if the dropout is unscheduled it loses a great deal of money.

Short glitches, of the order of seconds, may be absorbed by the thermal latency. But they also raise the possibility of thermal shock and stresses in some reactor components (particularly the window and the target) which could soon lead to fracture. Various numbers are quoted, ranging from 3 trips per year to a few thousand. A full study is urgently required. But even the easier figure will require a design that is fundamentally different from the traditional approach. There is no reason this cannot be done. Reliability is achievable - at a cost - through redundancy, under-rating, graceful failure, and planned maintenance, using a holistic analysis of the complete system. Other industries produce reliable equipment that is at least as complex as accelerators.

Accelerator TechnologiesEdit

There are 4 different types of accelerator that can be considered.

Linear AcceleratorsEdit

Linear Accelerators can provide the necessary energy and current, but they are expensive. The SNS at Oak Ridge has a 2 MW 1 GeV beam, and a similar design running at a higher current would clearly be possible. However the cost, of the order of $1Bn, is very high. (It would be nice to get more exact figure.) The accelerator is hundreds of metres long, and each metre is full of RF cavities, focussing quadrupoles, beam diagnostic equipment, etcetera, Without the economy of recycling that a circular design brings, this large cost in inevitable.

It may be that a Linac is the best option for a prototype proof-of-principle ADSR, but if many of these reactors are to be produced, as part of the next generation of power sources, a much cheaper alternative must be found.


A classical (Lawrence) cyclotron is restricted to nonrelativistic energies, and is certainly inadequate for the ~1 GeV protons needed for effective spallation. More advanced designs, such as the separated-sector PSI cyclotron,

The 590 MeV PSI cyclotron

can attain higher energies, in the hundreds of MeV range. AS well as the energy, a cycllotron is limited in current by the lack of strong focussing, which can overcome the space-charge repulsion in high current machines. There are proposals for cyclotrons for ADSRs, for instance as a stack, but these definitely face severe challenges


In a synchrotron the ring is filled with particle bunches. These are then accelerated, and as part of this process the fields of the bending dipole and focussing quadrupole magnets are increased. After the beam has attained its desired energy and been extracted, the magnet currents ramp back down again. This cycling of the magnets restricts the frequency with which the accelertor can deliver particles, and thus means that the current is low. Rapid Cycling Synchrotron (RCS) systes operate at 30+ Hz, and this has been put forward as a possible solution, but it is difficult to achieve.

Anecdotally, synchrotrons are held to be less reliable than cyclotrons. The AC magnets are more complex than the DC magnets of a cyclotron or FFAG, and there is much more to go wrong.


The FFAG can be considered as a hybrid between a cyclotron and a synchrotron. The magnets are DC, like those of a cyclotron, but as the mean field increases with radius the particles experience a field which changes with time, increasing as their energy increases with acceleration.  Furthermore the field gradient provides horizontal focussing. Counterbends are used to provide focussing in the vertical direction, hence the 'AG'  - for 'Alternating Gradient' in teh  name.

CW operation may be possible, depending on the design. The variation of orbit time with momentum is basically parabolic, and if the minimum of the parabola falls in the middle of the momentum range it may be possible to achieve an isochronicity which is sufficient for the RF to run at a fixed frequency.  

This enables the FFAG to provide high currents at high (GeV) energies with low losses. It is simpler than a synchrotron (due to the DC magnets) or a linac (there are only 1 or two RF cavities) which should enable high reliability. This makes it ideal for ADSR systems. The drawback is that a design of this type has not been successfully demonstrated yet. There are proton machines in Japan and an electron machine at Daresbury, but these all operate at very low currents. 

Utility Grade Proton Beam SwitcherEdit

Under development

Utility Grade Distribution and Storage RingEdit

Physics research proton beam storage rings are relatively mature, but usage for a mesh network of accelerators feeding multiple reactors is new, and tied to beam switching capabilities. A model based on particle bunches occupying storage time/space slots in a ring, with accelerators injecting particle bunches into accelerator specific slots, and extractors pulling particle bunches out of the ring to feed a specific reactor is possible, but requires effective coordination between injection and extraction to equally feed all active reactors while maintaining an even distribution of packet bunches in the ring. In a 3 reactor, 4 accelerator, 2 ring arrangement, you gain N+1 redundancy for accelerators, N+1 redundancy for rings, and a particle buffer supply stored in the rings to ride out 1 accelerator tripping while the other accelerators ramp up power.