- Email: [email protected]
Neutron spallation sources in Europe
lll[l l ~'-'ItlII: g+,'l[l+,'l
l~:lSEVlI-:R
Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 134
PROCEEDINGS SUPPLEMENTS
N eutron spallation sources in Europe P.J. Bryant CERN, CH-1211 Geneva 23, Switzerland After a brief general and historical discussion, the main design features of spallation sources are described. At the present time, Europe not only has the world-leading pulsed neutron spallation source, the SNS-ISIS at RAL, UK, but it is on the point of commissioning a world-leading continuous cyclotron-driven source, the S INQ at PSI, Switzerland. Looking to the future, yet more powerful pulsed sources are actively under study and the difficult problem of high-power target design (>250 kW) is leading to a new technology for liquid targets. The accelerator designs, although basically classical, require custom-built solutions that are often at the limit of presentday accelerator technology. L INTRODUCTION Accelerator-driven spallation sources attract large numbers of multidisciplinary users. Firstly, this is true of the neutron-user community itself, which embraces physics, chemistry, biology, material science and engineering and welcomes the fidl range of users from industrialists to basic researchers. In addition, it is also usual to make use of the accelerator complex for muon physics, neutrino physics and possibly also as the base for a medical facility. These features make such laboratories more attractive to national and regional funding agencies. The users have in common the need for a high-performance accelerator and a professional accelerator staff. Working alone, they would not be able to justify the cost of such a facility. Multidisciplinary centres of this type also present the accelerator physicists with a challenge to provide an efficient service and to open the full potential of their machines to the various users. The first proposal for an accelerator-driven source with a thermalised neutron facility was in the 1960s (Intense Neutron Generator, ING)[1]. One notable feature in this proposal was the use of a eutectic mixture of lead and bismuth in a flowing column for the target. The project was, however, never built. There was earlier work on 'electrical' breeders [2] and the direct bombardment of a heavymetal target for spallation had already been foreseen by Lewis [3], but the first demonstration of the spallation source concept was at ANL in 1972 (ZGS Intense Neutron Generator Prototype, ZING-P). A spallation source uses a medium-energy accelerator
to excite the nuclei of a heavy-metal target (often Ta or W alloy), from which neutrons 'evaporate'. These 'fast' neutrons are slowed in moderators set around the target. Neutrons make excellent probes for condensed matter [4], as their neutrality ensures deep penetration, their magnetic moment reacts to magnetic structures and their weak interaction minimises radiation damage, allowing in vivo experiments. Neutron sources are widely used in biology, chemistry, materials science and basic physics, as well as for technological applications such as radiography and materials testing. Spallation sources are 'environmentally friendly' compared to reactors that currently furnish the majority of neutrons for research purposes. The time structure of accelerator beams offers some experimental advantages and peak neutron intensities can exceed those of reactors. Although the present demand for neutrons can be met by reactors, this situation is unlikely to continue due to the increasing severity of safety regulations and the declared policies of many countries to close down their reactors within the next decade or so. At the same time, the demand for neutrons is expected to grow and, consequently, there has been an increasing interest in accelerator-driven sources. 2. BASIC REQUIREMENTS Figure 1 shows the generic pulsed neutron source. The target receives short (~gs), tfighintensity, proton pulses with repetition rates of typically 10-60 Hz. The intensity should be as high as possible, as the appetite of the users is currently
0920-5632/96/$15.00 e 1996 Elsevier Science B.V. All rights reserved. HI: S0920-5632(96)00423-9
126
l?J B~'ant/Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 134
beyond what the accelerator and target designers can provide. The pulses are formed by accumulating protons from an H- linac by chargeexchange injection into a large-aperture synchrotron, or accumulator ring, that is filled to its space-charge limit. The final energy of the beam needs to be 500 MeV to 4 GeV, but these are soft limits. The synchrotron can be used to perform the major part of the acceleration (a synchrotron-driven source), or it can be used as an accumulator/buncher leaving all the acceleration to the linac (a linac-driven source). When designing a spallation source, there are a number of practical considerations to be taken into account. In general, it is cheaper to accelerate in a synchrotron than a linac. Fast-cycling synchrotrons typically work over a magnetic field range of less than 1 to 8. This keeps the top field below saturation and avoids tracking problems between the different magnetic circuits. By using two synchrotrons of 25 Hz interleaved to produce 50 Hz, many problems can be eased. The higher the injection energy into the synchrotron, or accumulator, the weaker the space-charge and the higher the intensity that can be stored. There is a soft threshold near 130MeV, below which the losses are more tolerable from the point of view of induced activity. At this energy, it is also natural to change from an Alvarez to a coupled-cavity linac and go straight to say 800 MeV. It is true that the injection losses are more potent at higher energies, but a static accumulator should be easier to operate than a rapid cycling synchrotron so that losses can be reduced. At high energies, charge-exchange injection also experiences problems with magneticfield stripping. Targets have a soft threshold at about 200 kW, above which surface cooling is no longer feasible and either the coolant becomes directly irradiated by the beam, or a more complicated dynamic (e.g. rotating) target is needed. At MW powers a new technology is needed (e.g. liquid target). Synchrotron-driven sources work welI up to 1 MW average power with final energies in the range 0.5-2.5GeV. Linac-driven sources are planned for 1-5 MW with final energies in the range 0.8-1.5 GeV. For higher powers, it would be possible to raise the energy of synchrotron-driven sources to say I0 GeV, or one can speculate on the
future use of superconducting linacs of 3 GeV or more for linac-driven sources.
~
I
~
F
source
A~,~.t~¢j.~¢....1" lt~ ~ Additional source chain
Q
Debuncher • ~ Stripping Linac(s) ~ ~
Synchrotron \ $ / t) or Accumulator~ S / Additional t
Target station
and neutron lines
_ 2 ~ - "~].
"
rmg ~ ~ Additional r-" - _ target " station
Figure 1. Generic complex for a pulsed source. 3. EXIS1TING AND FUTURE SOURCES Currently the highest intensity pulsed source is SNS-ISIS [5] that was approved for construction at RAL in June 1977 (see Figure 2). The design used the hall, services and some equipment from the older NIMROD, 7 GeV proton synchrotron that was closed a year later in June 1978. ISIS is a synchrotron-driven source designed to operate with an 800 MeV, 50 Hz beam of 2 x 1013 protons per cycle (160 gA average current) in two bunches. The linac, rebuilt from the NIMROD linac, injects H- ions over 250 turns by charge-exchange into the synchrotron at 70 MeV. The space charge limit of the machine was originally estimated to be 2.3× 1013 protons per pulse (184~tA average current), but performance has improved steadily and the power delivered to the target has reached 160 kW (2.5 × 1013 protons per pulse or 200 gA average current). The synchrotron lattice is based on ten equal cells giving a circumference of 163 m. This choice was imposed by the building and the recuperation of the resonant power supply from another earlier machine NINA. Operating at 50 Hz introduces a number of special design features. Vacuum chambers in the magnets have to be made from ceramic and RAL developed their own technology and equipment for fabricating chambers up to 5 m long. The ceramic chambers have to be shielded from the beam by so-called rf cages to keep the beam-to-environment coupling impedance
P j Bryant~Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 134
small. This is another special feature of this type of machine. RAL also developed their own technology for fabricating the aluminium oxide stripping foils (120 × 40 mm 2 with 50 ~tg/cm2) for injection. The extraction scheme is a classical full-aperture kicker followed by a current-wall septum The accelerator delivers its beam in a horizontal configuration to a single spallation target equipped with 14 neutron lines. On the way to the spallation target, the beam passes through a muon transmission target that supports two facilities. Finally, there is a heavily shielded area for neutrinos. ISIS was first equipped with a U238 target, which gave acceptable lifetimes up to 100 ~tA, but for beam currents close to 200 ~tA, Ta is preferred. The ISIS target blocks are monolithic whereas more recent designs favour split targets. During 1996, the continuous, cyclotron-driven spallation source at PSI [6] will come into operation (see Figure 3). The main accelerator is a separated sector cyclotron. The injected beam from a 72 MeV cyclotron is accelerated over 220 turns to the maximum energy of 595 MeV. With the operating current of 1.5 mA, the design power is 0.9 MW. As at RAL, the beam first passes via two meson production targets before being refocussed onto the SINQ target. Unlike RAL, the target configuration is vertical with the feed from below. This geometry maximises the space available for neutron experiments, but requires careful design for maintenance. Looking more to the future, there are two spallation source studies in Europe, the AUSTRON [7] (see Figure 4) and the European SpaUation Source (ESS) [8,9]. The former has been completed and is awaiting the consideration of the Austrian Government and its regional partners. The AUSTRON is a synchrotron-driven source similar to the SNS-ISIS machine, but in its final stage (AUSTRON III) it would be more powerful, delivering 0.4MW shared between two target stations. The beam would be injected from an Hlinac at 130 MeV into a 50Hz synchrotron that takes the final beam up to 1.6 GeV. The beam has two bunches, as in ISIS, and there are 3.2 × 1013 protons per pulse (128 ~tA average current). The
127
technologies for stripping foils, ceramic vacuum chambers and rf cages would follow closely the experience of RAL. The first target is a solid, edgecooled split target made of a tungsten rhenium alloy (5%). The target station is foreseen to have 18 neutron beam lines. Although the AUSTRON is principally a neutron spallation source, it also includes a muon physics facility, a test beam facility for detector development (for projects such as LHC) and a light-ion and proton cancer therapy unit. The multidisciplinary aspects of the centre are seen as an important factor in the reactivation of the Central European Region that has been held back from such initiatives for many decades by the presence of the Iron Curtain. While the AUSTRON would be a relatively quick response to the expected future demand for neutrons, the ESS is seen as a longer term development that would reach 5 MW. The ESS study was initiated by KFA, Jiilich and by RAL. Other laboratories have joined the study and additional funding has been obtained from the European Communities. A reference design [9] has been published for a linac-driven source of 5 MW delivering 1 ~ts pulses at 50 Hz shared between two targets (see Figure 5). One target would be limited to 1 MW, but the other would be able to take the full power. The H- linac injects into two accumulator rings that are both operating at the full 50Hz repetition rate. The problem of losses and induced activity is being taken very seriously and the report states that "A basic aim is the minimization of particle losses...". The most visible feature in the design connected with this policy is the 1800 collimation bend in the injection line. In June 1995, first priority was given to unsplit, liquid targets using mercury for both stations. The key advantages of the liquid target are the convective heat removal, the very long lifetime of the target material, its potential for higher beam power and the reduced neutron absorption, radiolysis and tritium production due to the absence of cooling water. However, apart from a stationary lead target operated at TRIUMF that partially melts under beam loading, there is no practical experience of
128
P.J. Bryant~Nuclear Physics B (Proc. Suppl.) 51A (1996) 125-134
\
DEVA
KARMEN MARl
eVS POLAFIIS
TEST
EC MUON FACILITY EMU ~(
SANDALS '-PRISMA ROTAX
RIKEN PROJECT Q TFXA 800 Mev SYNCROTRON
CRISP
IRIS U 0
REP Tett Beam
70 Mev H" Unac
Figure 2 SNS-ISIS at RAL (Courtesy of RAL). SINQ-Target Hall
SINQ- Neutron Guide Hall *'":''"'''':"
,.[
]
I
Operations Building
Cockrofl Walton
7 . . ~
_
~
~
!! Low E Protons'
" I,-~.~ ptis ,~--~-~-
0
Hot Cell • l-and
Muon
) Faci, lities
I*
(nucl. phys. i " a n d muSR) i [ Inj. ~' IL
--r---
Isotopes
t
~TT
I
!i
590 M e V
~_ Proton • ILNeutrons RF,Building
Proton Therapy
Figure 3 SINQ and other experimental facilities around the PSI accelerators (Courtesy of PSI).
I?.i Bryant~Nuclear Physics B (Proc. Suppl.) 51A (1996) I25-134
!
129
MEDICAL
""
STORAGE RIN
ll-4 -JllJ
i
i
LINAC LC~' ] N ] E N S I T Y TEST DEA~ FOR DETECTOR
m
" ! i0
>
iOOm
i
!
I i
SPALLATION TARGET A
Figure 4 Proposal for A U S T R O N site layout. ,.
~
7
?10
...
5CGL
European
--F:I
Spallation
Source (ESS)
Main design: 2 Source chains + DTL + CCL + 2 Accumulators+ 2 targets Injection to accumulators 1.334 GeV Energy on target 1.334 GeV No. of particles per cycle 4.68 x 1014 Repetition rate 50 Hz, 40+10 Hz Average beam power 5.1 MW
First alternative: 2 Source chains + DTL + 2 RCSs+ 2 targets Injection to RCSs 800 MeV Energy on target 3.0 GeV Repetition rate 50 Hz, 40+10 Hz Average beam power 5.1 MW
Figure 5 Proposal for European SpaUation Source (ESS) machine complex (Courtesy of ESS).
130
Pj B~ant/Nuclear Physics B (Proc. SuppL) 51A (1996) 125-134
flowing-liquid targets in the accelerator laboratories, although this was a feature of the very first design for the ING [1]. 4. INJECTION, TRAPPING, ACCELERATION The injection into the synchrotron is the natural starting point for the design of the complex. This is a critical point regarding space charge and it determines the final intensity, emittance, momentum spread, duty cycle, debunching and chopping.
states leading to distributed losses that cannot be collected so easily. For this reason, higher energy injection schemes use a single, long. weak dipole. It is desirable at high energies to bring the H ° beam right out of the machine before collecting it, but the small angle given by a weak dipole makes this difficult. In the ESS, a second stripping foil is proposed in the H ° and H- beams followed by a septum magnet, once they have left the main beam. Dipole
unstripped beam H
Dipole
H- injected beam
~
/
J
f
4.1. Charge-exchange injection Charge-exchange injection was pioneered in Novosibirsk [10] and is now the preferred method for injection into high-intensity machines. A beam of H- ions is steered onto a closed orbit in the synchrotron between two dipoles (see Figure 6). This is possible, since the dipoles will bend oppositely charged particles in opposite directions. In the drift space where the two beams coincide, both pass through a thin foil, wtfich strips away the weakly-bound electrons from the H- ions to create protons for the main beam. Those particles that are not stripped, or are stripped to the H ° charge state, wSll be separated from the circulating beam by the next dipole. The stripping foil can be made from aluminium oxide, It is a critical element and a good example of the specialised technologies that abound in accelerator engineering. The fringe field of the dipoles is used to guide the stripped electrons onto a collector. The beam could also be injected onto an outer orbit, but the scheme in Figure 2 is more natural, since the injection takes places just before the field minimum and the falling field causes the newly injected beam to spiral outwards from the foil so filling the machine aperture. Making the injected beam cross the main beam also represents a saving in the aperture of the dipoles. Typically the stripping foil will be 98% efficient with the 'waste' beams emerging as unstripped ions and neutral H ° particles (some of which may be in excited H o* states). Figure 6 suggests that the simple solution of collectors placed outside the main beam will lead to a very clean injection. At low energies (< 500 MeV), tiffs is possible, but at higher energies strong (and especially inhomogeneous) magnetic fields can strip H" ions and H° - excited
.°
StriE)ping foil Inside of ring
Figure 6. Schematic view of H- injection.
4.2. 'Painting' The transverse enlittances of the linac beam are small compared to the synchrotron acceptance and 'painting' is used to fill the machine acceptance to its space-charge limit. Ideally, a KapchinskijVladimirskij (K-V) distribution is created by injecting a zero-emittance beam over N turns, on the locus shown in Figure 7(a). The position of the nth injected pulse [x(n),z(n)] with respect to the closed orbit in the synchrotron follows a square-root law as shown in Figures 7(b) and (c). 'Painting' occurs over hundreds of turns (e.g. AUSTRON130, $NS-ISIS at RAL- 200 and ESS- 1000 turns). The sinusoidal fall of the main field is not ideal for 'painting' so the position of the beam's equilibrium orbit has to be controlled in another way. This can be done by fast bumpers in both planes, starting with the maximum vertical amplitude and minimum horizontal amplitude and sweeping through to the opposite situation. Two sets of bumpers may be difficult to fit into the synchrotron lattice, so one alternative is to sweep the beam vertically using a bumper in the transfer line. Another possibility is to use a cavity in the injection line to modulate the beam energy while injecting into a region of high dispersion in the
l?J B~yant/Nuclear Physics B (Proc. SuppL) 51A (l 996) 125 134
synchrotron. This is proposed for the ESS in conjunction with a 'corner' stripping foil (see next Section). z(n) Vertical emittance
t
~z
z(max)4(1-n/N) z(max) -- -- -I
special advantage with the 'strip' foil in Figure 8(d). The 'strip' foil will select a narrow momentum bite from the beam, while allowing the off-momentum halo to leave the machine along the same path as the unstripped ions. The alternative to this method of halo removal is a momentum scraping in the injection transfer line.
(b)
Positions of N pulses
Injected am
~ x ~_ Turn n, or time (n) : x(max)~/(n/N) h
_
_ ~
--
Horizontal emittance
131
gx
(¢)
, ~
f
/
(a)
aX)
i
i N I I b,
Turn n, or time Figure 7. 'Painting' a K-V distribution.
4.3. Stripping foil traversais Figure 8 illustrates some foil geometries. In Figure 8(a), the equilibrium orbit of the beam is moved to the right by a horizontal bump, so increasing the horizontal emittance, while the injection spot is scanned downwards by a bumper in the injection line. In Figure 8(b), the principle is the same, but the bottom half of the foil has been removed to reduce unwanted traversals. Figure 8(c) shows a 'corner' foil. In this case, bumps move the equilibrium orbit diagonally. More aperture is needed in the injection region, but this case does offer fewer foil traversals. More complicated schemes are possible. The ESS, for example, plan to use a 'corner' foil in conjuction with energy ramping and a vertical orbit bump. The use of energy ramping avoids the extra aperture mentioned above with the standard "corner" foil configuration. Injection usually occurs into a dispersive region of the lattice and the question arises as to whether the linac beam should be dispersion-free or matched to the ring. A dispersion-free beam is smaller and can be drawn off the foil faster, despite the fact that it suffers an emittance dilution due to the dispersion mismatch. On the other hand, the dispersionmatched beam (typically Dx=5 m, 13x=l m) offers a
~
U
Centreof chamber
7:
lz
Injec6on spot moves down as I Equilibrium orbit moves across Injected beam with dispersion (c)
Foil
~
Movements as in (a)
~.
z
/ I Equilibrium orbit moves diagonally
(b) x
(d)
z
x
X
I~1
l
Movements as in (a)
Figure 8. Stripping foil geometries (real space). 4.4. RF TRAPPING AND ACCELERATION In synchrotron-driven sources, the rate of energy rise is usually too high for the trapping to be fully adiabatic and a tracking program that includes space charge is needed. When designing the RF program, the optimisation is directed to reducing particle losses. Apart from the losses that occur directly from the RF bucket, limits have to be set upon the incoherent tune shift that affects losses on resonances and the momentum spread that affects the aperture. The description below is based on the design experience gained with the AUSTRON [11], but the basic features should be universal. The RF cycle splits naturally into four stages starting with the injection and trapping that takes place on the falling magnetic field just before the minimum. The RF voltage is applied smoothly from a very low value and rotation in the bucket is optimised for uniformity of filling. The leading particles rotate by -90 ° during injection and reach -180 ° at the field minimum. Too much voltage during this first stage will increase the momentum spread in the next step. The second stage starts at
132
l~J B~yant/Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 I34
the minimum field point and continues to where the momentum spread is maximum (the incoherent tune shift peaks a little earlier). This first critical point occurs typically 1 to 2 ms into the acceleration and, since the adiabatic damping has had little time to be felt, the beam reaches its maximum radial size in the dispersive regions. The smoothness and rate of rise of the RF voltage are optimised to reduce losses. Once this aperture limit is past, the third stage is entered, which runs until the middle of the acceleration cycle when the synchronous phase angle is maximum. During this stage, the RF voltage has to be increased smoothly and rapidly to maintain the bucket area. The fourth and final stage up to the top energy poses no particular problems. One way to reduce losses during trapping is to pre-bunch the beam by chopping at a much lower energy in the injection line. This scheme features in the proposals for most future machines. Gaps of 25-40% are usually considered.
distortions due to eddy currents. To reduce static charging of the chamber walls, define the beam potentials and reduce the longitudinal and transverse space-charge impedances, internal metallic RF shields are used that ideally follow the beam contours. A proven design for RF shields is the stainless steel cage of axial wires with vertical side plates used in the SNS-ISIS machine, but thin metallisation of the chamber walls has been discussed as an alternative. Metallisation is attractive at first sight, but it is easily damaged by high-intensity beams and it is not as easily applied in long chambers as might be expected. Recent analytical work on the fields excited by longitudinally and transversely modulated beams in a wire cage can be found in [14] with approximations for the g-factors in [7] (spacecharge impedances are traditionally expressed in terms of the g-factor). The g-factors for a beam with elliptical, cross-sectional variations in a circular, metallised chamber are also reported in [7].
5. L A T T I C E AND RESONANT SUPPLY In order to keep the voltages applied to the magnet coils within reasonable limits (i.e. <10 kV for safe operation), the lattice should be divided into as large a number as possible of identical cells. One solution is to select an asymmetric cell (to get long and short drift spaces) and to contrive to fit all functions into these cells. SNS-ISIS follows this philosophy with 10 identical cells. At higher energies, a custom-designed lattice with superperiods becomes essential. To keep the voltages down, the upper and lower magnet coils can be separated to increase the number of identical cells as 'seen' by the resonant supply. Magnet lengths are adjusted so that field levels and saturation effects are about equal in quadrupole and dipole circuits. At around 1 GeV, a threefold symmetry is popular (e.g. AUSTRON, ESS, see Figure 9). Fast-cycling machines use a resonant circuit with DC-biased, AC excitation known as the 'White circuit' [12] (original single-cell circuit), [ 13] (first multi-cell circuit).
Drawn on a
Horizontal Plan View [X-Y plane] 7.0000m square grid
I i
I
/
i\1
Ii
Figure 9. AUSTRON lattice. 7. T R A N S F E R LINES
6. R F - C A G E .
Fast-cycling machines use insulating (ceramic) vacuum chambers to avoid power loss and field
The transfer line from the linac to the ring will exhibit strong space-charge forces that will be of importance for the calculation of the debunching
f?,Z Blyant/Nuclear Physics B (Proc. Suppl.) 51A (1996) 125-134
cavity. The injection line may also require a bend and a long straight section for momentum and betatron collimation to limit losses at injection (e.g. ESS). Space-charge effects are minimal in the high-energy, large-aperture extraction lines. A transmission target (_< 5% interception) may be proposed for a muon experimental area and this will require a zero-dispersion, low-13 insertion in the line. The background from a muon-target will need shielding and a bend in the line before entering the neutron experimental area is advisable. In order to have the maximum space for neutron lines, the target can be entered vertically (e.g. SINQ). However, this does complicate maintenance in an area that is likely to have a high level of induced activity. 8. RADIATION MANAGEMENT
The high intensity and repetition rate of pulsed spallation sources makes radiation management of prime importance. The activation of the exhaust air and waste water must be monitored and kept below hmits agreed with licensing authorities. Ventilation systems need low replacement rates (< 2 per hour). High-loss areas can be sealed and the air slowly leaked to lower-loss areas. An under-pressure is needed to prevent out-leaks. Storage of active air may be needed. All exhaust air must be filtered to remove 7Be and other aerosols. Intermediate storage of waste water, shielding of ground water and secondary cooling circuits are all standard considerations. The degradation of materials such as the coil insulation needs to be estimated and radiation-hard elements used in critical places (e.g. the last quadrupole doublet before the spallation target). Remote handling will be needed for the spallation target and the stripping foil. Care must be taken concerning dust, especially from fractured stripping foils, and exhaust air from roughing pumps must be filtered. Tight tolerances and high efficiencies for the collimator and beam control systems will be needed and machine operation should be interlocked to a bean1 loss measurement system. Where losses are not continuous, ahiminium vacuum chambers show less induced activity than steel ones, but for steady losses over many years the reverse can be true due to the build up of 22Na. Similarly, 'heavy' concrete has a shielding advantage, but after many years it can
133
build up more long-lived isotopes than ordinary concrete. The problem of the target station is particularly acute. A helium atmosphere is used around the target void and secondary cooling circuits are standard. Shielding around such machines is typically several meters thick and is built up with different combinations of earth, concrete and iron according to the situation. Two basic philosophies are possible: a large hall using concrete and iron shielding blocks (typically 2 m concrete and 2 m iron), or wide, cut-and-fill tunnels (typically with 12 m concrete walls and of the order of 8 m earth. A hall would be a 'controlled' area (< 25 ~tSv/h average dose and film badges for those working there), but the exterior of a shielding mound over a cut-and-fill tunnel would be a 'surveyed' area (< 2.5 ~tSv/h average dose and film badges not required). The former philosophy is applied at SNS-ISIS and in several high-intensity facilities such as the AA and ISOLDE at CERN and the latter philosophy was adopted for the AUSTRON design and will possibly be adopted for the ESS. Often the choice depends on external factors, such the re-use of an existing hall, but in general, shielding blocks have the disadvantages of high cost and the time and space needed when demounting for access. The weight of shielding is also so large that ground movements of millimeters can occur and over periods of months the machine alignment can be badly affected. On the other hand, tunnels must be wide to be acceptable in a high radiation machine. The main ring tunnel, where most machine maintenance is to be expected, must have cranes. For example, in the AUSTRON design the tunnel was circular, 14 m wide and has full crane coverage, leaving room for addition mobile shielding for hot spots and the temporary stacking of elements during maintenance. Local shielding over hot spots is also necessary to cut down the path lengths of ionising particles in the air to reduce air activation. Transfer tunnels cannot in general be equipped with cranes because of their irregular changes in direction, but a wide tunnel and a suitable magnet transport vehicle should meet most needs. The target station and last section of the incoming beam line are always in a hall with crane coverage. The target hall will typically contain neutron beam lines that radiate outwards from the
134
ICJ. Bryant~Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 134
station for 20-40 m. There are longer lines of up to 100 m, but these usually extend beyond the target hall and are covered in an ad hoc way. To provide an efficient service, the machine reliability must be extremely high, the complex must be divided into shielded zones that allow access to some facilities while others are operating and there must be well thought-out policies for dealing with activated air, water and equipment. 9. COMMENTS ON HIGH INTENSITY In recent years, there has been a growing interest in high-intensity, high-power and highefficiency accelerators. Some of the main areas of interest are: (a) High-intensity, lepton colliders, or so-called 'factories' e.g. Beauty factory, Tau Charm factory. (b) High-intensity booster injectors for high-energy synchrotrons for particle physics. (c) High-intensity, rapid-cycling proton synchrotrons for neutron spallation sources and muon-muon colliders. (d) High-intensity, heavy-ion linacs for inertialfusion drivers. (e) High-power cyclotrons for continuous spallation sources and the 'energy amplifier'. The requirements of these different applications put different emphasis on the different problems. An ion beam in an inertial-fusion driver has limited penetration mad deposits its energy in a thin surface layer. Losses can cause vaporisation of the vacuum chamber surface, whereas a proton beam would penetrate deeply causing less immediate physical damage but widespread induced activity. Spallation sources work with relatively long bunches of the order of 50 ns. Muon-muon drivers and heavy-ion fusion drivers require much shorter bunches of tO ns, which are more difficult to produce and more difficult to maintain short. In applications requiring high efficiency, the cyclotron is better adapted than a fast-cycling synchrotron. Nature is not always against the designer. Pulsed spallation sources and muon-muon drivers can work with two bunches per cycle, which avoids bunch to bunch instabilities in the synchrotron.
REFERENCES 1 P. Tunnicliffe, G. Bartholomew, E.W. Vogt eds., AECL study for the Intense Neutron Generator, Atomic Energy of Canada, Report AECL-2600, (1966). 2 Livermore Research Laboratory, Status of the MTA Process, Livermore Research Laboratory, Report LRL-102 (Del) (1954). 3 W.B. Lewis, The significance of the yield of neutrons from heavy elements excited to high energies, Atomic Energy of Canada Ltd., Report AECL-968, (1952). 4 J.M. Carpenter, Pulsed spallation neutron sources for slow neutron scattering, Nucl. Instr. Meths., 145, 91, (1977). 5 I.S.K. Gardner, ISIS status report, Proc. EPAC 1994, June 27-July 1, London, (World Scientific),3-7. 6 H.C. Walter ed., PSI Users Guide, (PSI, Villigen, Switzerland, March 1994). 7 P. Bryant, M. Regler, M. Schuster eds., AUSTRON feasibility study, (AUSTRON Planning Office, c/o Atom-institut der ¢)sterreichischen Universit~iten, Vienna, 1994). 8 H. Lengeler, Proposals for spallation sources in Europe, Proc. EPAC 1994, June 27-July 1, London, (World Scientific), 249-53. 9 I.S.K. Gardner, H. Lengeler, G.H. Rees eds, Outline design of the European Spallation Source, ESS 95-3--M, (Sept. 1995) l0 G.I. Budker, G.1. Dimov, On the charge exchange injection of protons into ring accelerators, Proc. 5th Int. Conf. on HE Accelerators, Dubna, 1963, (translation Conf114, US, AEC, Div. ofTech. Info., 1372-7). 11 E. Griesmayer, RF trapping and acceleration in AUSTRON, Report CERN/PS 95-10 (DI), (May 1995). 12 M.G. White, F.C. Shoemaker, G.K. O'Neill, The 3 BeV High Intensity Synchrotron, CERN Symposium on High Energy Accelerators and Pion Physics, 1956, CERN 56-26, vol. 1,525. 13 J.A. Fox, Resonant magnet network and power supply for the 4 GeV electron synchrotron NINA, Proc. IEE 112, no. 6, June 1965. 14 T. Wang, Electrostatic field of a perturbed beam with RF-screening wires, CERN/PS 94-08(DI), April 1994.
l~:lSEVlI-:R
Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 134
PROCEEDINGS SUPPLEMENTS
N eutron spallation sources in Europe P.J. Bryant CERN, CH-1211 Geneva 23, Switzerland After a brief general and historical discussion, the main design features of spallation sources are described. At the present time, Europe not only has the world-leading pulsed neutron spallation source, the SNS-ISIS at RAL, UK, but it is on the point of commissioning a world-leading continuous cyclotron-driven source, the S INQ at PSI, Switzerland. Looking to the future, yet more powerful pulsed sources are actively under study and the difficult problem of high-power target design (>250 kW) is leading to a new technology for liquid targets. The accelerator designs, although basically classical, require custom-built solutions that are often at the limit of presentday accelerator technology. L INTRODUCTION Accelerator-driven spallation sources attract large numbers of multidisciplinary users. Firstly, this is true of the neutron-user community itself, which embraces physics, chemistry, biology, material science and engineering and welcomes the fidl range of users from industrialists to basic researchers. In addition, it is also usual to make use of the accelerator complex for muon physics, neutrino physics and possibly also as the base for a medical facility. These features make such laboratories more attractive to national and regional funding agencies. The users have in common the need for a high-performance accelerator and a professional accelerator staff. Working alone, they would not be able to justify the cost of such a facility. Multidisciplinary centres of this type also present the accelerator physicists with a challenge to provide an efficient service and to open the full potential of their machines to the various users. The first proposal for an accelerator-driven source with a thermalised neutron facility was in the 1960s (Intense Neutron Generator, ING)[1]. One notable feature in this proposal was the use of a eutectic mixture of lead and bismuth in a flowing column for the target. The project was, however, never built. There was earlier work on 'electrical' breeders [2] and the direct bombardment of a heavymetal target for spallation had already been foreseen by Lewis [3], but the first demonstration of the spallation source concept was at ANL in 1972 (ZGS Intense Neutron Generator Prototype, ZING-P). A spallation source uses a medium-energy accelerator
to excite the nuclei of a heavy-metal target (often Ta or W alloy), from which neutrons 'evaporate'. These 'fast' neutrons are slowed in moderators set around the target. Neutrons make excellent probes for condensed matter [4], as their neutrality ensures deep penetration, their magnetic moment reacts to magnetic structures and their weak interaction minimises radiation damage, allowing in vivo experiments. Neutron sources are widely used in biology, chemistry, materials science and basic physics, as well as for technological applications such as radiography and materials testing. Spallation sources are 'environmentally friendly' compared to reactors that currently furnish the majority of neutrons for research purposes. The time structure of accelerator beams offers some experimental advantages and peak neutron intensities can exceed those of reactors. Although the present demand for neutrons can be met by reactors, this situation is unlikely to continue due to the increasing severity of safety regulations and the declared policies of many countries to close down their reactors within the next decade or so. At the same time, the demand for neutrons is expected to grow and, consequently, there has been an increasing interest in accelerator-driven sources. 2. BASIC REQUIREMENTS Figure 1 shows the generic pulsed neutron source. The target receives short (~gs), tfighintensity, proton pulses with repetition rates of typically 10-60 Hz. The intensity should be as high as possible, as the appetite of the users is currently
0920-5632/96/$15.00 e 1996 Elsevier Science B.V. All rights reserved. HI: S0920-5632(96)00423-9
126
l?J B~'ant/Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 134
beyond what the accelerator and target designers can provide. The pulses are formed by accumulating protons from an H- linac by chargeexchange injection into a large-aperture synchrotron, or accumulator ring, that is filled to its space-charge limit. The final energy of the beam needs to be 500 MeV to 4 GeV, but these are soft limits. The synchrotron can be used to perform the major part of the acceleration (a synchrotron-driven source), or it can be used as an accumulator/buncher leaving all the acceleration to the linac (a linac-driven source). When designing a spallation source, there are a number of practical considerations to be taken into account. In general, it is cheaper to accelerate in a synchrotron than a linac. Fast-cycling synchrotrons typically work over a magnetic field range of less than 1 to 8. This keeps the top field below saturation and avoids tracking problems between the different magnetic circuits. By using two synchrotrons of 25 Hz interleaved to produce 50 Hz, many problems can be eased. The higher the injection energy into the synchrotron, or accumulator, the weaker the space-charge and the higher the intensity that can be stored. There is a soft threshold near 130MeV, below which the losses are more tolerable from the point of view of induced activity. At this energy, it is also natural to change from an Alvarez to a coupled-cavity linac and go straight to say 800 MeV. It is true that the injection losses are more potent at higher energies, but a static accumulator should be easier to operate than a rapid cycling synchrotron so that losses can be reduced. At high energies, charge-exchange injection also experiences problems with magneticfield stripping. Targets have a soft threshold at about 200 kW, above which surface cooling is no longer feasible and either the coolant becomes directly irradiated by the beam, or a more complicated dynamic (e.g. rotating) target is needed. At MW powers a new technology is needed (e.g. liquid target). Synchrotron-driven sources work welI up to 1 MW average power with final energies in the range 0.5-2.5GeV. Linac-driven sources are planned for 1-5 MW with final energies in the range 0.8-1.5 GeV. For higher powers, it would be possible to raise the energy of synchrotron-driven sources to say I0 GeV, or one can speculate on the
future use of superconducting linacs of 3 GeV or more for linac-driven sources.
~
I
~
F
source
A~,~.t~¢j.~¢....1" lt~ ~ Additional source chain
Q
Debuncher • ~ Stripping Linac(s) ~ ~
Synchrotron \ $ / t) or Accumulator~ S / Additional t
Target station
and neutron lines
_ 2 ~ - "~].
"
rmg ~ ~ Additional r-" - _ target " station
Figure 1. Generic complex for a pulsed source. 3. EXIS1TING AND FUTURE SOURCES Currently the highest intensity pulsed source is SNS-ISIS [5] that was approved for construction at RAL in June 1977 (see Figure 2). The design used the hall, services and some equipment from the older NIMROD, 7 GeV proton synchrotron that was closed a year later in June 1978. ISIS is a synchrotron-driven source designed to operate with an 800 MeV, 50 Hz beam of 2 x 1013 protons per cycle (160 gA average current) in two bunches. The linac, rebuilt from the NIMROD linac, injects H- ions over 250 turns by charge-exchange into the synchrotron at 70 MeV. The space charge limit of the machine was originally estimated to be 2.3× 1013 protons per pulse (184~tA average current), but performance has improved steadily and the power delivered to the target has reached 160 kW (2.5 × 1013 protons per pulse or 200 gA average current). The synchrotron lattice is based on ten equal cells giving a circumference of 163 m. This choice was imposed by the building and the recuperation of the resonant power supply from another earlier machine NINA. Operating at 50 Hz introduces a number of special design features. Vacuum chambers in the magnets have to be made from ceramic and RAL developed their own technology and equipment for fabricating chambers up to 5 m long. The ceramic chambers have to be shielded from the beam by so-called rf cages to keep the beam-to-environment coupling impedance
P j Bryant~Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 134
small. This is another special feature of this type of machine. RAL also developed their own technology for fabricating the aluminium oxide stripping foils (120 × 40 mm 2 with 50 ~tg/cm2) for injection. The extraction scheme is a classical full-aperture kicker followed by a current-wall septum The accelerator delivers its beam in a horizontal configuration to a single spallation target equipped with 14 neutron lines. On the way to the spallation target, the beam passes through a muon transmission target that supports two facilities. Finally, there is a heavily shielded area for neutrinos. ISIS was first equipped with a U238 target, which gave acceptable lifetimes up to 100 ~tA, but for beam currents close to 200 ~tA, Ta is preferred. The ISIS target blocks are monolithic whereas more recent designs favour split targets. During 1996, the continuous, cyclotron-driven spallation source at PSI [6] will come into operation (see Figure 3). The main accelerator is a separated sector cyclotron. The injected beam from a 72 MeV cyclotron is accelerated over 220 turns to the maximum energy of 595 MeV. With the operating current of 1.5 mA, the design power is 0.9 MW. As at RAL, the beam first passes via two meson production targets before being refocussed onto the SINQ target. Unlike RAL, the target configuration is vertical with the feed from below. This geometry maximises the space available for neutron experiments, but requires careful design for maintenance. Looking more to the future, there are two spallation source studies in Europe, the AUSTRON [7] (see Figure 4) and the European SpaUation Source (ESS) [8,9]. The former has been completed and is awaiting the consideration of the Austrian Government and its regional partners. The AUSTRON is a synchrotron-driven source similar to the SNS-ISIS machine, but in its final stage (AUSTRON III) it would be more powerful, delivering 0.4MW shared between two target stations. The beam would be injected from an Hlinac at 130 MeV into a 50Hz synchrotron that takes the final beam up to 1.6 GeV. The beam has two bunches, as in ISIS, and there are 3.2 × 1013 protons per pulse (128 ~tA average current). The
127
technologies for stripping foils, ceramic vacuum chambers and rf cages would follow closely the experience of RAL. The first target is a solid, edgecooled split target made of a tungsten rhenium alloy (5%). The target station is foreseen to have 18 neutron beam lines. Although the AUSTRON is principally a neutron spallation source, it also includes a muon physics facility, a test beam facility for detector development (for projects such as LHC) and a light-ion and proton cancer therapy unit. The multidisciplinary aspects of the centre are seen as an important factor in the reactivation of the Central European Region that has been held back from such initiatives for many decades by the presence of the Iron Curtain. While the AUSTRON would be a relatively quick response to the expected future demand for neutrons, the ESS is seen as a longer term development that would reach 5 MW. The ESS study was initiated by KFA, Jiilich and by RAL. Other laboratories have joined the study and additional funding has been obtained from the European Communities. A reference design [9] has been published for a linac-driven source of 5 MW delivering 1 ~ts pulses at 50 Hz shared between two targets (see Figure 5). One target would be limited to 1 MW, but the other would be able to take the full power. The H- linac injects into two accumulator rings that are both operating at the full 50Hz repetition rate. The problem of losses and induced activity is being taken very seriously and the report states that "A basic aim is the minimization of particle losses...". The most visible feature in the design connected with this policy is the 1800 collimation bend in the injection line. In June 1995, first priority was given to unsplit, liquid targets using mercury for both stations. The key advantages of the liquid target are the convective heat removal, the very long lifetime of the target material, its potential for higher beam power and the reduced neutron absorption, radiolysis and tritium production due to the absence of cooling water. However, apart from a stationary lead target operated at TRIUMF that partially melts under beam loading, there is no practical experience of
128
P.J. Bryant~Nuclear Physics B (Proc. Suppl.) 51A (1996) 125-134
\
DEVA
KARMEN MARl
eVS POLAFIIS
TEST
EC MUON FACILITY EMU ~(
SANDALS '-PRISMA ROTAX
RIKEN PROJECT Q TFXA 800 Mev SYNCROTRON
CRISP
IRIS U 0
REP Tett Beam
70 Mev H" Unac
Figure 2 SNS-ISIS at RAL (Courtesy of RAL). SINQ-Target Hall
SINQ- Neutron Guide Hall *'":''"'''':"
,.[
]
I
Operations Building
Cockrofl Walton
7 . . ~
_
~
~
!! Low E Protons'
" I,-~.~ ptis ,~--~-~-
0
Hot Cell • l-and
Muon
) Faci, lities
I*
(nucl. phys. i " a n d muSR) i [ Inj. ~' IL
--r---
Isotopes
t
~TT
I
!i
590 M e V
~_ Proton • ILNeutrons RF,Building
Proton Therapy
Figure 3 SINQ and other experimental facilities around the PSI accelerators (Courtesy of PSI).
I?.i Bryant~Nuclear Physics B (Proc. Suppl.) 51A (1996) I25-134
!
129
MEDICAL
""
STORAGE RIN
ll-4 -JllJ
i
i
LINAC LC~' ] N ] E N S I T Y TEST DEA~ FOR DETECTOR
m
" ! i0
>
iOOm
i
!
I i
SPALLATION TARGET A
Figure 4 Proposal for A U S T R O N site layout. ,.
~
7
?10
...
5CGL
European
--F:I
Spallation
Source (ESS)
Main design: 2 Source chains + DTL + CCL + 2 Accumulators+ 2 targets Injection to accumulators 1.334 GeV Energy on target 1.334 GeV No. of particles per cycle 4.68 x 1014 Repetition rate 50 Hz, 40+10 Hz Average beam power 5.1 MW
First alternative: 2 Source chains + DTL + 2 RCSs+ 2 targets Injection to RCSs 800 MeV Energy on target 3.0 GeV Repetition rate 50 Hz, 40+10 Hz Average beam power 5.1 MW
Figure 5 Proposal for European SpaUation Source (ESS) machine complex (Courtesy of ESS).
130
Pj B~ant/Nuclear Physics B (Proc. SuppL) 51A (1996) 125-134
flowing-liquid targets in the accelerator laboratories, although this was a feature of the very first design for the ING [1]. 4. INJECTION, TRAPPING, ACCELERATION The injection into the synchrotron is the natural starting point for the design of the complex. This is a critical point regarding space charge and it determines the final intensity, emittance, momentum spread, duty cycle, debunching and chopping.
states leading to distributed losses that cannot be collected so easily. For this reason, higher energy injection schemes use a single, long. weak dipole. It is desirable at high energies to bring the H ° beam right out of the machine before collecting it, but the small angle given by a weak dipole makes this difficult. In the ESS, a second stripping foil is proposed in the H ° and H- beams followed by a septum magnet, once they have left the main beam. Dipole
unstripped beam H
Dipole
H- injected beam
~
/
J
f
4.1. Charge-exchange injection Charge-exchange injection was pioneered in Novosibirsk [10] and is now the preferred method for injection into high-intensity machines. A beam of H- ions is steered onto a closed orbit in the synchrotron between two dipoles (see Figure 6). This is possible, since the dipoles will bend oppositely charged particles in opposite directions. In the drift space where the two beams coincide, both pass through a thin foil, wtfich strips away the weakly-bound electrons from the H- ions to create protons for the main beam. Those particles that are not stripped, or are stripped to the H ° charge state, wSll be separated from the circulating beam by the next dipole. The stripping foil can be made from aluminium oxide, It is a critical element and a good example of the specialised technologies that abound in accelerator engineering. The fringe field of the dipoles is used to guide the stripped electrons onto a collector. The beam could also be injected onto an outer orbit, but the scheme in Figure 2 is more natural, since the injection takes places just before the field minimum and the falling field causes the newly injected beam to spiral outwards from the foil so filling the machine aperture. Making the injected beam cross the main beam also represents a saving in the aperture of the dipoles. Typically the stripping foil will be 98% efficient with the 'waste' beams emerging as unstripped ions and neutral H ° particles (some of which may be in excited H o* states). Figure 6 suggests that the simple solution of collectors placed outside the main beam will lead to a very clean injection. At low energies (< 500 MeV), tiffs is possible, but at higher energies strong (and especially inhomogeneous) magnetic fields can strip H" ions and H° - excited
.°
StriE)ping foil Inside of ring
Figure 6. Schematic view of H- injection.
4.2. 'Painting' The transverse enlittances of the linac beam are small compared to the synchrotron acceptance and 'painting' is used to fill the machine acceptance to its space-charge limit. Ideally, a KapchinskijVladimirskij (K-V) distribution is created by injecting a zero-emittance beam over N turns, on the locus shown in Figure 7(a). The position of the nth injected pulse [x(n),z(n)] with respect to the closed orbit in the synchrotron follows a square-root law as shown in Figures 7(b) and (c). 'Painting' occurs over hundreds of turns (e.g. AUSTRON130, $NS-ISIS at RAL- 200 and ESS- 1000 turns). The sinusoidal fall of the main field is not ideal for 'painting' so the position of the beam's equilibrium orbit has to be controlled in another way. This can be done by fast bumpers in both planes, starting with the maximum vertical amplitude and minimum horizontal amplitude and sweeping through to the opposite situation. Two sets of bumpers may be difficult to fit into the synchrotron lattice, so one alternative is to sweep the beam vertically using a bumper in the transfer line. Another possibility is to use a cavity in the injection line to modulate the beam energy while injecting into a region of high dispersion in the
l?J B~yant/Nuclear Physics B (Proc. SuppL) 51A (l 996) 125 134
synchrotron. This is proposed for the ESS in conjunction with a 'corner' stripping foil (see next Section). z(n) Vertical emittance
t
~z
z(max)4(1-n/N) z(max) -- -- -I
special advantage with the 'strip' foil in Figure 8(d). The 'strip' foil will select a narrow momentum bite from the beam, while allowing the off-momentum halo to leave the machine along the same path as the unstripped ions. The alternative to this method of halo removal is a momentum scraping in the injection transfer line.
(b)
Positions of N pulses
Injected am
~ x ~_ Turn n, or time (n) : x(max)~/(n/N) h
_
_ ~
--
Horizontal emittance
131
gx
(¢)
, ~
f
/
(a)
aX)
i
i N I I b,
Turn n, or time Figure 7. 'Painting' a K-V distribution.
4.3. Stripping foil traversais Figure 8 illustrates some foil geometries. In Figure 8(a), the equilibrium orbit of the beam is moved to the right by a horizontal bump, so increasing the horizontal emittance, while the injection spot is scanned downwards by a bumper in the injection line. In Figure 8(b), the principle is the same, but the bottom half of the foil has been removed to reduce unwanted traversals. Figure 8(c) shows a 'corner' foil. In this case, bumps move the equilibrium orbit diagonally. More aperture is needed in the injection region, but this case does offer fewer foil traversals. More complicated schemes are possible. The ESS, for example, plan to use a 'corner' foil in conjuction with energy ramping and a vertical orbit bump. The use of energy ramping avoids the extra aperture mentioned above with the standard "corner" foil configuration. Injection usually occurs into a dispersive region of the lattice and the question arises as to whether the linac beam should be dispersion-free or matched to the ring. A dispersion-free beam is smaller and can be drawn off the foil faster, despite the fact that it suffers an emittance dilution due to the dispersion mismatch. On the other hand, the dispersionmatched beam (typically Dx=5 m, 13x=l m) offers a
~
U
Centreof chamber
7:
lz
Injec6on spot moves down as I Equilibrium orbit moves across Injected beam with dispersion (c)
Foil
~
Movements as in (a)
~.
z
/ I Equilibrium orbit moves diagonally
(b) x
(d)
z
x
X
I~1
l
Movements as in (a)
Figure 8. Stripping foil geometries (real space). 4.4. RF TRAPPING AND ACCELERATION In synchrotron-driven sources, the rate of energy rise is usually too high for the trapping to be fully adiabatic and a tracking program that includes space charge is needed. When designing the RF program, the optimisation is directed to reducing particle losses. Apart from the losses that occur directly from the RF bucket, limits have to be set upon the incoherent tune shift that affects losses on resonances and the momentum spread that affects the aperture. The description below is based on the design experience gained with the AUSTRON [11], but the basic features should be universal. The RF cycle splits naturally into four stages starting with the injection and trapping that takes place on the falling magnetic field just before the minimum. The RF voltage is applied smoothly from a very low value and rotation in the bucket is optimised for uniformity of filling. The leading particles rotate by -90 ° during injection and reach -180 ° at the field minimum. Too much voltage during this first stage will increase the momentum spread in the next step. The second stage starts at
132
l~J B~yant/Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 I34
the minimum field point and continues to where the momentum spread is maximum (the incoherent tune shift peaks a little earlier). This first critical point occurs typically 1 to 2 ms into the acceleration and, since the adiabatic damping has had little time to be felt, the beam reaches its maximum radial size in the dispersive regions. The smoothness and rate of rise of the RF voltage are optimised to reduce losses. Once this aperture limit is past, the third stage is entered, which runs until the middle of the acceleration cycle when the synchronous phase angle is maximum. During this stage, the RF voltage has to be increased smoothly and rapidly to maintain the bucket area. The fourth and final stage up to the top energy poses no particular problems. One way to reduce losses during trapping is to pre-bunch the beam by chopping at a much lower energy in the injection line. This scheme features in the proposals for most future machines. Gaps of 25-40% are usually considered.
distortions due to eddy currents. To reduce static charging of the chamber walls, define the beam potentials and reduce the longitudinal and transverse space-charge impedances, internal metallic RF shields are used that ideally follow the beam contours. A proven design for RF shields is the stainless steel cage of axial wires with vertical side plates used in the SNS-ISIS machine, but thin metallisation of the chamber walls has been discussed as an alternative. Metallisation is attractive at first sight, but it is easily damaged by high-intensity beams and it is not as easily applied in long chambers as might be expected. Recent analytical work on the fields excited by longitudinally and transversely modulated beams in a wire cage can be found in [14] with approximations for the g-factors in [7] (spacecharge impedances are traditionally expressed in terms of the g-factor). The g-factors for a beam with elliptical, cross-sectional variations in a circular, metallised chamber are also reported in [7].
5. L A T T I C E AND RESONANT SUPPLY In order to keep the voltages applied to the magnet coils within reasonable limits (i.e. <10 kV for safe operation), the lattice should be divided into as large a number as possible of identical cells. One solution is to select an asymmetric cell (to get long and short drift spaces) and to contrive to fit all functions into these cells. SNS-ISIS follows this philosophy with 10 identical cells. At higher energies, a custom-designed lattice with superperiods becomes essential. To keep the voltages down, the upper and lower magnet coils can be separated to increase the number of identical cells as 'seen' by the resonant supply. Magnet lengths are adjusted so that field levels and saturation effects are about equal in quadrupole and dipole circuits. At around 1 GeV, a threefold symmetry is popular (e.g. AUSTRON, ESS, see Figure 9). Fast-cycling machines use a resonant circuit with DC-biased, AC excitation known as the 'White circuit' [12] (original single-cell circuit), [ 13] (first multi-cell circuit).
Drawn on a
Horizontal Plan View [X-Y plane] 7.0000m square grid
I i
I
/
i\1
Ii
Figure 9. AUSTRON lattice. 7. T R A N S F E R LINES
6. R F - C A G E .
Fast-cycling machines use insulating (ceramic) vacuum chambers to avoid power loss and field
The transfer line from the linac to the ring will exhibit strong space-charge forces that will be of importance for the calculation of the debunching
f?,Z Blyant/Nuclear Physics B (Proc. Suppl.) 51A (1996) 125-134
cavity. The injection line may also require a bend and a long straight section for momentum and betatron collimation to limit losses at injection (e.g. ESS). Space-charge effects are minimal in the high-energy, large-aperture extraction lines. A transmission target (_< 5% interception) may be proposed for a muon experimental area and this will require a zero-dispersion, low-13 insertion in the line. The background from a muon-target will need shielding and a bend in the line before entering the neutron experimental area is advisable. In order to have the maximum space for neutron lines, the target can be entered vertically (e.g. SINQ). However, this does complicate maintenance in an area that is likely to have a high level of induced activity. 8. RADIATION MANAGEMENT
The high intensity and repetition rate of pulsed spallation sources makes radiation management of prime importance. The activation of the exhaust air and waste water must be monitored and kept below hmits agreed with licensing authorities. Ventilation systems need low replacement rates (< 2 per hour). High-loss areas can be sealed and the air slowly leaked to lower-loss areas. An under-pressure is needed to prevent out-leaks. Storage of active air may be needed. All exhaust air must be filtered to remove 7Be and other aerosols. Intermediate storage of waste water, shielding of ground water and secondary cooling circuits are all standard considerations. The degradation of materials such as the coil insulation needs to be estimated and radiation-hard elements used in critical places (e.g. the last quadrupole doublet before the spallation target). Remote handling will be needed for the spallation target and the stripping foil. Care must be taken concerning dust, especially from fractured stripping foils, and exhaust air from roughing pumps must be filtered. Tight tolerances and high efficiencies for the collimator and beam control systems will be needed and machine operation should be interlocked to a bean1 loss measurement system. Where losses are not continuous, ahiminium vacuum chambers show less induced activity than steel ones, but for steady losses over many years the reverse can be true due to the build up of 22Na. Similarly, 'heavy' concrete has a shielding advantage, but after many years it can
133
build up more long-lived isotopes than ordinary concrete. The problem of the target station is particularly acute. A helium atmosphere is used around the target void and secondary cooling circuits are standard. Shielding around such machines is typically several meters thick and is built up with different combinations of earth, concrete and iron according to the situation. Two basic philosophies are possible: a large hall using concrete and iron shielding blocks (typically 2 m concrete and 2 m iron), or wide, cut-and-fill tunnels (typically with 12 m concrete walls and of the order of 8 m earth. A hall would be a 'controlled' area (< 25 ~tSv/h average dose and film badges for those working there), but the exterior of a shielding mound over a cut-and-fill tunnel would be a 'surveyed' area (< 2.5 ~tSv/h average dose and film badges not required). The former philosophy is applied at SNS-ISIS and in several high-intensity facilities such as the AA and ISOLDE at CERN and the latter philosophy was adopted for the AUSTRON design and will possibly be adopted for the ESS. Often the choice depends on external factors, such the re-use of an existing hall, but in general, shielding blocks have the disadvantages of high cost and the time and space needed when demounting for access. The weight of shielding is also so large that ground movements of millimeters can occur and over periods of months the machine alignment can be badly affected. On the other hand, tunnels must be wide to be acceptable in a high radiation machine. The main ring tunnel, where most machine maintenance is to be expected, must have cranes. For example, in the AUSTRON design the tunnel was circular, 14 m wide and has full crane coverage, leaving room for addition mobile shielding for hot spots and the temporary stacking of elements during maintenance. Local shielding over hot spots is also necessary to cut down the path lengths of ionising particles in the air to reduce air activation. Transfer tunnels cannot in general be equipped with cranes because of their irregular changes in direction, but a wide tunnel and a suitable magnet transport vehicle should meet most needs. The target station and last section of the incoming beam line are always in a hall with crane coverage. The target hall will typically contain neutron beam lines that radiate outwards from the
134
ICJ. Bryant~Nuclear Physics B (Proc. Suppl.) 51A (1996) 125 134
station for 20-40 m. There are longer lines of up to 100 m, but these usually extend beyond the target hall and are covered in an ad hoc way. To provide an efficient service, the machine reliability must be extremely high, the complex must be divided into shielded zones that allow access to some facilities while others are operating and there must be well thought-out policies for dealing with activated air, water and equipment. 9. COMMENTS ON HIGH INTENSITY In recent years, there has been a growing interest in high-intensity, high-power and highefficiency accelerators. Some of the main areas of interest are: (a) High-intensity, lepton colliders, or so-called 'factories' e.g. Beauty factory, Tau Charm factory. (b) High-intensity booster injectors for high-energy synchrotrons for particle physics. (c) High-intensity, rapid-cycling proton synchrotrons for neutron spallation sources and muon-muon colliders. (d) High-intensity, heavy-ion linacs for inertialfusion drivers. (e) High-power cyclotrons for continuous spallation sources and the 'energy amplifier'. The requirements of these different applications put different emphasis on the different problems. An ion beam in an inertial-fusion driver has limited penetration mad deposits its energy in a thin surface layer. Losses can cause vaporisation of the vacuum chamber surface, whereas a proton beam would penetrate deeply causing less immediate physical damage but widespread induced activity. Spallation sources work with relatively long bunches of the order of 50 ns. Muon-muon drivers and heavy-ion fusion drivers require much shorter bunches of tO ns, which are more difficult to produce and more difficult to maintain short. In applications requiring high efficiency, the cyclotron is better adapted than a fast-cycling synchrotron. Nature is not always against the designer. Pulsed spallation sources and muon-muon drivers can work with two bunches per cycle, which avoids bunch to bunch instabilities in the synchrotron.
REFERENCES 1 P. Tunnicliffe, G. Bartholomew, E.W. Vogt eds., AECL study for the Intense Neutron Generator, Atomic Energy of Canada, Report AECL-2600, (1966). 2 Livermore Research Laboratory, Status of the MTA Process, Livermore Research Laboratory, Report LRL-102 (Del) (1954). 3 W.B. Lewis, The significance of the yield of neutrons from heavy elements excited to high energies, Atomic Energy of Canada Ltd., Report AECL-968, (1952). 4 J.M. Carpenter, Pulsed spallation neutron sources for slow neutron scattering, Nucl. Instr. Meths., 145, 91, (1977). 5 I.S.K. Gardner, ISIS status report, Proc. EPAC 1994, June 27-July 1, London, (World Scientific),3-7. 6 H.C. Walter ed., PSI Users Guide, (PSI, Villigen, Switzerland, March 1994). 7 P. Bryant, M. Regler, M. Schuster eds., AUSTRON feasibility study, (AUSTRON Planning Office, c/o Atom-institut der ¢)sterreichischen Universit~iten, Vienna, 1994). 8 H. Lengeler, Proposals for spallation sources in Europe, Proc. EPAC 1994, June 27-July 1, London, (World Scientific), 249-53. 9 I.S.K. Gardner, H. Lengeler, G.H. Rees eds, Outline design of the European Spallation Source, ESS 95-3--M, (Sept. 1995) l0 G.I. Budker, G.1. Dimov, On the charge exchange injection of protons into ring accelerators, Proc. 5th Int. Conf. on HE Accelerators, Dubna, 1963, (translation Conf114, US, AEC, Div. ofTech. Info., 1372-7). 11 E. Griesmayer, RF trapping and acceleration in AUSTRON, Report CERN/PS 95-10 (DI), (May 1995). 12 M.G. White, F.C. Shoemaker, G.K. O'Neill, The 3 BeV High Intensity Synchrotron, CERN Symposium on High Energy Accelerators and Pion Physics, 1956, CERN 56-26, vol. 1,525. 13 J.A. Fox, Resonant magnet network and power supply for the 4 GeV electron synchrotron NINA, Proc. IEE 112, no. 6, June 1965. 14 T. Wang, Electrostatic field of a perturbed beam with RF-screening wires, CERN/PS 94-08(DI), April 1994.