- Email: [email protected]
Superconducting cyclotrons
224
Nuclear Instruments and Methods in Physics Research A287 (1990) 224-234 North-Holland
SUPERCONDUCTING CYCLOTRONS U. TRINKS
Physik-Department, Technische Universität München, D-8046 Garching, FRG
Superconducting cyclotrons with compact geometry and solenoidal coils are operating since several years, indicating the successful introduction of superconductivity in cyclotron design . Some of these facilities use or will use tandems as injectors, some operate as stand-alone machines . The technical problems of this concept are caused mainly by the compactness, which is the attractive feature of the design of course . In order to achieve specific energies above - 300 McV/u separated sector cyclotrons with superconducting magnets are still under consideration. Technical difficulties are related to the large stored magnetic field energy, to the strong electromagnetic forces and to the non-circular shape of the coils. A rather small separated orbit cyclotron with superconducting channel magnets and superconducting cavities is under construction in Munich . This type of cyclotron offers unique focusing properties at relaxed injection and extraction conditions. The large number of individual controllable channel magnets gives some flexibility. On the other hand it demands a high degree of relinhility of man v complex components, requiering simple solutions wherever possible. The present status of the Tritron-project is girtr iii detail .
1 . Introduction
There are three types of superconducting cyclotrons for accelerating heavy ions . First Superconducting Compact Cyclotrons (SCC) were developed, characterized by spiral pole pieces and solenoidal coils inside a circular yoke. Five machines of this type are operating successfully, three additional ones are under construction [1-8]. The maximum specific energy achieved up to now is about 100 MeV/u . Generally it is not limited by the lack of bending or focusing power of the superconducting magnets but by voltage problems in the electrostatic deflectors of the extraction system due to the compact design . To achieve specific energies above 300 MeV/u feasibility studies on Svperconducting Separated Sector Cyclotrons (SSSC) are underway at three laboratories . At the Joint Institute for Nuclear Research at Dubna a full wale sector magnet with superconducting sector coils as well as an accelerating resonator for the first stage DC1 of a neutron spallation source are under construction [9j. For the European Light Ion Medical Accelerator EULIMA for cancer therapy a cyclotron with a pair of rather rig superconducting ring coils and four separated iron sectors with the return yokes outside the coils is under investigation [10] . At Ganil recently a SSSC was discussed as a third stage after the two existing normal conducting separated sector cyclotrons to increase the ion energies from - 100 MeV/u by a factor of - 5 [1 i ] . Finally at Munich a small prototype of a Superconducting Separated Orbit Cyclotron (SSOC) - the Tri(1168-9002/9{1/Sä3 .5() ,- Ellsevier Science Publishers B .V . (N(>rth-Holland)
tron - is under construction [12]. The ion bunches are guided by individually adjustable, narrow superconducting channel magnets along the spiral orbit, and accelerated by six superconducting cavities . The Tritron can be considered a linac curled up, with focusing properties similar to a synchrotron, and some restrictions caused by the cyclotron condition. It represents a linac of - 1 .4.0 m total length inside: a cryostat of 3 .6 m diameter with only six cavities, each crossed by 20 parallel bunches at the same time. On the other hand the extraordinary focusing properties have had to be paid for by a rather large number of channel magnets.
2. Superconducting compact cyclotrons
In fig. 1 a survey on all superconducting compact cyclotron pr(Jects is given. On the lower scale the date of the design start is indicated, on the upper scale the date of the first extracted beam respectively the first internal bean' at design energy [13] . The development was started in late 1973 by H . Blosser at the Michigan State Univet ,ity with the MSU-îC520 machine [1] and at the same A me at the Chalk River Nuclear Laboratories with the TASCC (Tandem Accelerator Superconducting Cyclotron [21) . After several years of design, model tests and finally construction, H. Blossers group obtained the first external beams in August 1982, `r`vitit maximum energies of aVout 55 MeV/u for ions up to °`'O, and correspondingly less for heavier masses. In September 1985 external beams were observed at Chalk
U. Trinks / Superconducting cyclotrons River, too . This pioneering work has triggered several additional projects within a few years . In June 1988 a K520-SCC, which is essentially a copy of the MSU-K520 machine, came into operation at Texas A & M University [3] . In Milan a machine somewhat bigger than MSU-K520 is under construction [4]. It shall be a booster for the 16 MV Tandem of the Laboratorio Nazionale del Sud in Catania, however, it could be operated as a standalone machine with an ECR-source, too. At MSU a still larger SCC was built - MSU K1200 - which came into operation in June 1988, accelerating ions to -100 MeV/u till now [5] . So far the last SCC planned to supply external targets with ion beams is AGOR (K600), which has been under construction at the Institut de Physique Nucl6aire of Orsay since 1985 and which will be transferred to Groningen later on [6] . This cyclotron shall accelerate protons to a maximum energy of 200 MeV, light ions up to 100 MeV/u and heavy ions to somewhat lowtr energies . During the las' few years the main activity on the development of SCCs has shifted to rather small cyclotrons with internal targets for medical applications. The first small superconducting cyclotron for neutron production by 25 MeV/u deuterons on an internal target was developed at MSU again (K100) . It accelerated its first beam in February this year [7] . The cyclotron has a total weight of -- 25 ton only . It can be rotated about a horizontal axis, where the patient may be positioned for neutron therapy. Neither a beam extraction system nor beam transport and rotation systems are needed, nevertheless neutrons can be directed
at the patient from any angle . Only one shielded room is necessary . A still smaller medical cyclotron with a total weight of 3 ton is under construction at Oxford Instruments [8] . Proton beams for the production of short-lived radioisotopes for Positron Emission Tomography are produced by accelerating H - -ions up to 12 MeV and extracting them by stripping at a radius of 22 cm. The time from design start to first extracted beam of the first superconducting cyclotron - MSU-K520 - was 8.5 years . All the following projects (except very small medical accelerators) needed the same period or even more. This may appear rather long to those waiting for the beam. However, it has to be realized, that the superconducting technology requires the installation and operation of a helium refrigerator plant and a cryostat with many feed holes. Strong stresses due to quadratically increasing electromagnetic forces on the coils and due to different thermal contraction have to be overcome. Vacuum leaks, eventually appearing at low temperature only, as well as heat leaks have to be detected in presence of restricted experimental conditions. Each cooling down and warming up cycle of the big masses takes at least several days. In addition the high magnetic field in a compact geometry results in a new design of all other main components, especially of the accelerating system anti the extraction and injection elements . With this in mind the 8 year period appears to be an impressive result . It may be mentioned, that the two big Ganil cyclotrons with maximum energies comparable to those of the MSU-K1200 machine alone FIRST
1375 DESIGN START
1910
225
1985
(EXTERNA~)
BEAM
1990
line) The Fig. l . A survey on superconducting compact cyclotrons with dates of design start (below) and first beam on target (upper are given. maximum specific energies K-number indicates the bending power. Typical ions together with present Vi . BOOSTERS
226
U. Trinks / Superconducting cyclotrons R CONDUCTOR
TM ROD (13 D1 EACH Hilt
Fig. 2. Schematic view of the Chalk River cyclotron .
needed about the same period from the design start to first beam. By the way, the beam intensities at Ganil are about one order of magnitude larger than those at MSU-K1200 . The basic concept of all superconducting compact cyclotrons is essentially the same. Fig. 2 shows a schematic view of the Chalk River superconducting cyclotron, fig . 3 a midplane section. This cyclotron was specified to accelerate ions from lithium (50 MeV/u) to uranium (10 MeV/u), injected radially from outside from a 16 MV tandem by poststripping at an injection radius ranging from 14.5 cm to 22 cm, and extracted at a radius of 65 cm. The bending field is produced by a pair of split superconducting solenoids, operated at rather low current densities of < 35 A/mm2 in a cylindrical He bath cryostat . Thus the coils are cryostable, which guarantees safe operation. Each cylindrical pole is supplied with four spiral ion hills for axial focusing. The pole gap is 4 cm in the hills and 64 cm in the valleys, the maximum magnetic induction 6 T and 4.3
T, respectively . The isochronous field is achieved by adjusting the two currents in the split coils properly and by axial positioning of 104 trim rods located in the magnetic flutter poles . Four accelerating dee's are installed in the valleys forming a single resonant cavity with two coaxial line tuners several meters long on the axis of the machine. They shall produce 800 kV accelerating voltage per turn in the maximum with frequencies from 31 to 62 MHz, corresponding to the 2nd, 4th and 6th harmonics . Single turn resonant extraction from the cyclotron is initiated by means of the outermost trim rods. Then the beam is deflected by an electrostatic channel, positioned inside a dee . The steep radial decrease of the bending field at the rim of the poles causes strong radial defocusing forces, which are compensated by two saturated iron lenses on the wall of the coil cryostat . Finally the beam enters a narrow long magnetic channel consisting of 78 separate superconducting coils and some additional saturated iron gradient bars. The other superconducting compact cyclotrons differ
U Trinks / Superconducting cyclotrons
N V,
LY,
C'
ZO
I7
0
=
cß 0.
v e~
,
04
ad
2 oG aL t ,-
'AA
O N
a
V_
a
MN
N
~ ,
.
~, ö gBS t
ô U .40 E
caz~x-a x wr` M1` ~"V OM OM fV e-. 00 r- 00 N t- O> C4
âo uÔ N
aaG (: ?
c
00 c
~ W 0 a.
câ"X . . ,n,
a
N cV ~O N
NO
°
Crlô
E
Fig. 3. Midplane section of the Chalk River cyclotron : A. yoke, B, cryostat ; C, stripper foil; D, injection steering dipole ; E, electrostatic deflector ; F, magnetic channel; G, effective field boundary; H, magnetic lens.
ô
âi
00
from the Chalk River machine eventually by the axial injection from ECR-sources, by the number of sectors (3 instead of 4) by trim coils on the hill sectors instead of the trim rods and by details of the rf and extraction system. In table 1 some parameters of all SCCs are given. In principle, the highest specific energy for a projectile with specific charge QIA depends on the bending power k 1, and the focusing power k t of the magnetic field, on the accelerating voltage of the deer and on the Meld of fe electrostatic deflector of the extraction system. Generally the design values of the magnetic field properties were achieved or even exceeded. So the bending limit
00
02 Â 3 oâ
00.
0
Ô N c 0 .r C
.~é
0 u 00 ä ç aU. w a, O H w 0 aa
a c
N S N kn 5, W, N
00
~-.
c
0o c14 N 0 00 N N N ~ q V1
C
O O r- O v4 00 kn N
M
1i
M N II IW
M M -Tr M e- c n M
r°4
b ei M M r~ O ~t d' ~R ef n Cq CPR q ci a vi vi ~ ~ \ ~i
U r U
ca,
CM .2 k W
H
r.
n i
E 1- cc w .~
0 0 o ~ T .o U â
227
Q i i
Q
N 00 %n 00 00 00 00 c ß~
N M ,
S U aa Q
01
u E
u ,0
0
depends linearly on Q/A, the same is true for the dee voltage limit and for the electrostatic deflector limit . All ,à®cJ these ilmlts should be matched . Aciually contact dee stems and overheating of various joints in the causes the dee voltage limed to be more severe than the focusing limit. However, the most serious limitation for all SCCs but the medical accelerators stems from voltage problems at the first electrostatic deflector of the extraction line. Table 2 shows, that only about 70% of the design values were achieved in the real environment inside the cyclotrons. An additional problem of the extraction system is the rather poor transmission. typia,
01 00
0 . ß
N
N N
w
00 00
T < kb(QIA) 2 (1 ) restricts the maximum energy only of very heavy particles with low Q/A value . The focusing limit T < k i (Q/A) (2)
E
v 1.
BOOSTERS
U. Trinks / Superconducting cyclotrons
22 8
Table 2 Design and routinely achieved fields of the electrostatic deflectors of SCCs with external beams Project
[kV/cm]
140 140 140 125 140 105
MSU-K520 TASCC Texas MSU-K1200 Milan AGOR
Gap [mm]
Eroutine
design
[kV/cm]
7 7 7 6
84 86 93 92 -
rally 600 or even less, indicating modest beam quality considering the geometrical limitations on the extraction channel . This experience was taken into account at the design of the AGOR-project [6] . The maximum magnetic bending field is less than that of the previous projects, the extraction radius is correspondingly increased and the requirements on the voltage, of the electrostatic deflector are reduced . 3. Superconducting separated sector cyclotrons .he turn separation decreases drastically with increasing energy : Ar=
e Q 1 rU, moc2 A ß2 Y 3
where m cc 2 /e = 931 .5 MV and U is the accelerating voltage per turn . For nonrelativistic velocity ®r decreases approximately linearly with increasing energy . Above ß = 0.8 corresponding to T = 600 MeV/u it decreases dramatically, finally as the energy cubed, as fig . 4 demonstrates . To obtain a certain turn separation 20
1
T ( GeV/ul
.ô
1
0
Fig. 4. The quantity G = ß 2 Y3, which craters into eq . (3) for the turn separation . versus the velocity respectively the specific energy Tof ions .
Fig. 5. General layout of the EULIMA cyclotron.
both the extraction radius and the accelerating voltage have to he sufficiently large. Thus strong average bending fields become increasingly useless if high energies are wanted, while more and more space for the installation of an increasing number of accelerating gaps is needed . These requirements finally rule out the compact cyclotron design, favouring separated sector magnets instead. Superconductivity then may be used to improve the field flutter and by this the focusing power, avoiding spiral-shaped superconducting coils, which are more difficult to produce than simple sector coils. The EULIMA-project with big superconducting ring coils with inner diameter 4.4 m and outer diameter 5.2 m, four separated iron yokes outside and spiral pole pieces inside represents an intermediate step in the development from compact to separated magnet design [l0] . The general layout is shown in fig. 5. It is a feasibility study on an accelerator for cancer treatment, therefore C, 0 and hle-ions shall be accelerated to energies up to 400 MeV/u . The total iron weight of 4 x 155 ton would exceed only slightly that of a superconducting compact magnet with comparable bending and focusing power, which was proposed by H. Blosser [14]. Axial injection from an ECR-source, eventually combined with a booster RFQ, is planned . Though the EULIMA concept appears to be an attractive solution to obtain inn energies of about 400 MeV/u, a conventional synchrott on dedicated for medical applications could be a realistic alternative due to the easy energy variation. To achieve even higher energies totally separated sector magnets are needed . The first feasibility study on a superconducting sector coil for a SSSC was performed at Munich [15]. Presently a complete sector magnet with two superconducting coils is under construction at Dubna [9] (fig. 6) In order to produce the field effectively the overall current density should be as high as tolerable with respect to safe protection of the coil in case of a quench, when the stored magnetic energy of
U. Trinks / Superconducting cyclotrons
229
Fig. 8. Test winding of a subcoil of the r)Cl-magnet .
Fig. 6. The ¢,zst sector magnet of DCl/Dubna. Height : 2.4 m, weight : 35 ton .
several MJ is dissipated into heat . At the Munich coil a new protection mechanism was tested (cross quenching [161), which enables overall current densities of the potted coils of more than 100 A/mm2 compared to 35 A/mm2 of cryostable coils, which generally need not be de-energized in the case of a local quench of course . The Dubna coils are wound directly on sectorshaped iron cores (figs . 7 and 8). The iron contribution to the field is optimized by this design, especially in the
central corner of the sector, no space is lost by magnetically dead material . Furthermore the thermal contraction of the coil exceeds that of iron slightly, so that the coil will be prestressed automatically . Finally the steel core attracts the coil, so that the strong outward directed forces or, the sides of the sector coil, trying to make the coil more circular, are reduced drastically. A test of the complete magnet as well as a test of a prototype cavity with full power will be performed presumably within one year . Up to now none of these feasibility studies on big superconducting separated sector cyclotrons for heavy ion beams with energies of several 100 MeV/u has been realized. ®f course very recently a proposal for a third, superconducting sector cyclotron for GANIL to achieve energies up to - 500 MeV/u is under preparation. It appears not unlikely, that this project actualy will be realized .
e superconducting separated orbit cyclotron At the Munich Accelerator Laboratory a superconnyni-non the - 4'tl1'l tthe :'.r. e ~p~r s® A-1-Al w v74iV 3vaa booster for the existing MP under construction as a tandem [121 . It shall increase the ion energies by a factor of - 4.9, e .g. a '2C6 + beam will have a maximum energy of 20 MeV/u, protons about 43 MeV. The main purpose of the project is to demonstrate the feasibility off this type of cyclotron . The characteristic features of the Tritron concept are the strong transversal and rather strong longitudinal focusing forces, the flat superconducting magnet sectors and the big superconducting
~urtin_g separated orbiti. c :üi~al'vi .3 orbit -- i7 a
SS=_
Fig . 7 . Cross sections of a DC1-coil . In (c) : 1, superconducting, potted windings; 2, iron cores ; 3, stainless steel cages .
VI . 13D0 TERS
230
U. Trinks / Superconducting cyclotrons
accelerating cavities. The excellent beam qualities of the tandem injector are conserved . For a 43 MeV-proton beam for instance an energy width of 8T = ± 34 keV is expected at a time spread 0t = 25 ps. By mismatched injection into the Tritron and an additional debuncher behind it the energy width can be further reduced to ± 2 keV at the expense of the time spread and vice versa. A first proposal was presented in December 1983, the definitive funding followed in July 1987 . Fig. 9 shows cross sections of the Tritron, table 3 the
main parameters . The beam is led along a spiral orbit of almost 20 turns with constant turn separation A r = 4 cm. Each of the 12 magnet sectors consists of 20 or 19 superconducting channels of the window frame type with maximum fields of - 1.4 T. In order to guide the bunches on the central path through the magnets, each channel can be adjusted individually by means of superconducting switches across the magnet coils. The radial beam positions will be measured by beam profile monitors installed in each second intermediate sector . There
Fig. 9. Tritron cross sections. M, magnets; R, resonators ; V, vacuum vessel ; S, 80 I{. shield ; He, liquid helium reservoir; support. T,
U. Trinks / Superconducting cyclotrons
231
Table 3 Tritron design data Injector Max . energy H1 + (Q/A =1) S16+ (Q/A = 0.5) Ag32 (Q/A = 0.29) Injection/extractions radius Turn separation A r Harmonic numbers Magnet sector data: Number of magnet channels Sector angle Bending radius pl, P20 Maximum magnetic field Bmax Radial gradients 4 -1 SB/Sr Dimensions of the supercond . cable Number of strands Strand diameter Strand material Cu/NbTi Maximum cable current Imax Cavity data: Gap width at injection/extraction Rf-frequency Maximum accelerating field Emax Epeak /Emax
Maximum voltage at r 2 Stored energy Dissipated power P Quality factor (unloaded) Qa Beam power Geometry factor G Surface resistance R,., = G1Q0
13 MV tandem 43.5 MeV 20.1 MeV/u 7.6 MeV/u 66 cm/145 cm 4 cm 18-53 20(19) 20 ° 430 mm/942 mm 1 .4 T 3.6 m - 1- -5.2 m -1 0.7 x 2.9 mm2 14 0.4 mm 1.35 1400 A 60 mm/130 mm 170 MHz 4.7 MV/m < 1.5 530 kV - 1 .7 J 6W 3.6x108 _ 200 W 94 SZ 2.6 x 10-7
are alternating gradients from one sector to the other to get strong focusing in both transversal directions . Six superconducting resonators operating at 170 MHz shall accelerate the ion bunches along the 20 parallel turns . To get longitudinal focusing, the bunches have to cross the cavities at a rf-phase with increasing voltage. The beam is injected through three superconducting channel magnets. The whole machine hangs under a torus-like helium reservoir on the upper half of the cryostat. The cavities and magnets are cooled indirectly by pipes connected to the torus (then-mal siphon cooling). There is no special vacuum system for the heam or the cavities. The cryostat including the torus and all transfer lines exists . The stand-hy losses of the cryostat are less than 5 W. A refrigerator (150 W at 4.7 K) and all transfer lines are installed . 4.1 . The Tritron magnets
The magnet sectors consist of two steel sheets with slots every 4 cm (fig. 10) . Th,- sectors are mechanically identical up to the coils generating the alternating gradient. The coils are wound by a computer controlled
Fig. 10. Vertical cross section of a magnet sector. Fe, iron yoke; G, gradient windings ; Cu, copper shields ; D, dead insulating layers. winding machine (one channel/day) directly into the slots and then potted. They are shielded from beam losses by Cu profiles leaving a beam hole of diameter 11 mm. The coils are supported by the copper profiles and disc springs between . The dimensions of the Tritron magnets are about one order of magnitude less than the accelerator magnets constructed before . Correspondingly the requirements cn the tolerances are rather tight. Up to now seven channels for test purposes were produced. A maximum current of 1840 A was achieved without training, which is well above the design value (fig. 11) . The effective magnetic length measured by Hall probes is in good agreement with TOSCA calculations. The stray field 35 mrn in front of the end plates is less than 0.5 G, so no problems due to frozen-in-flux in the superconducting layers of the cavities are expected. To analyse higher order components in the radial field expansion narrow, bent superconducting induction loops were used, measuring the change of the magnetic length if the radial position x is varied . Three channels were investigated in detail with corresponding results, demonstrating the reproducibility of the production process . A typical measurement of the effective length related to the central len gth at x = 0 is shown in fig. 12. The
Test e
Design
OÔ Fig. 11. Critical current limits at 4.2 and 7 K versus field. VI . BOOSTERS
U. Trinks / Superconducting cyclotrons
23 2
Fig. 12 . Effective magnet length related to that at the magnet centre versus x . No current in the gradient coil . sextupole component, at x = f 4 mm, given by the deviation of the curve from the tangent at x = 0, is less than - 4 x 10-3, which is within the tolerable limit of 5 x 10 -3 . Contributions come from the entrance and exit edges, from a slit between the two halfs of the coil due to different thermal contraction of the coil and the steel, and from the gradient windings . It may be further reduced by adding dead spaces at the upper and lower edges of the coil, respectively, which will produce positive sextupole contributions (see fig. 10) . A new test unit of three channels with such dead spaces has just been wound . Presently the steel plates of all sectors are premilled, of them are finally milled and ready for being polished . The superconducting cable exists, yet without insulation . The big winding machine is ready for first tests. A separate cryostat for testing single sectors is under construction . 10 n J
4.1 . The superconducting cavities The 6 wedge-shaped reentrant type resonators consist of half shells which are connected in the horizontal plane, so no currents will cross the joint. They are fabricated by electroplating copper (-- 10 mm) onto a fibre glass mould, which is destroyed and removed afterwards. The inner surface is electroplated with - 10 li m thick PbSn (2-4% Sn) . The first cavity exists, the next will be delivered soon . Fig. 13 shows the radial voltage and accelerating field characteristics . The rf-superconductivity properties of PbSn were investigated systematically by means of a reentrant type R 10 6
RBCS
tô'~
n
11
0 .6 04 02
/
IQ] -~
i
/ 1/
/
/
-
Q`O / /CP a
T
t
lu . 0. . 66
90
1O
â0
Radius [cm]
Fig. 13 . Voltage and electrical field characteristic of the cavity in relative units and half of the horizontal cross section .
0.2
0.4
06
0.8
1
f 1GHzl
2
Fig. 14 . Surface resistance R s of the test cavity (O) at different modes above 490 MHz and of the Tritron cavity (6) at 170 MHz. The data on R BOs (e) were obtained from the temperature dependence of R s . Alloy : PbSn with -- 2% Sri .
U. Trinks / Superconducting cyclotrons
Q6
,,,
23 3
T-4 .2K T z_5 K
N 2
a)
ee e
ted* ®
%,
e®
e e e
e
e
®
=®
e
e
e .,C)
10-!8
N1
Fig. 16 . Stability diagram and working line. n 1 , n 2 field indices of axially and radially focusing channels . Included are stability limits for different sextupole contributions at x = t4 mm, calculated for transversal emittances of 6 iT mm mrad .
e
0
100
200
300
400
500 Umax NV l
Fig. 15 . Some measurements of the unloaded quality factor Qo of the Tritron cavity versus the maximum gap voltage . Curve (a), fresh PbSn-surface, voltage limited by multipacting, caused by a small piece of steel inside the cavity. Curves (b) and (c) : after removal of the steel piece. (b), before and after heliumconditioning of the cavity . Voltage limited by field emission.
cavity of rotational symmetry, 45 cm in diameter, and a frequency of 490 MHz in the fundamental mode . The Sri critical temperature of PbSn with - 2% is 7 .4 K 7 .2 of Pb. The surface compared to K resistance R s is R the BCS-theory and the residual the sum of scs from R resistance REs due to imperfect surfaces (independent
of temperature) . R s was measured at 4.2 K in the range from 490 to 1670 MHz by operating the cavity in different modes (fig. 14) . The R Bcs-values were obtained from the temperature dependence of R s . The fitted line gives a frequency dependence -f' -' compared to f' .74 for Pb. The PbSn-line is about a factor of
two below the Pb-line. The observed improvement can be explained by the influence of the electron mean free path on R Bcs . It is remarkable, that the residual resistances of the test cavity are not too far above the R Bcs-line, which may be due to the increased chemical stability of the alloy with respect to oxidation . The surface resistance at 170 MHz in fig. 14 was obtained for the Tritron cavity with the best surface up opared to now, prepared c~avu wish several rinsing courses and fast drying by N 2 -blast following the electroplating process . It fits rather well into the data taken at higher frequencies . The unloaded quality factor Q o for this cavity is almost independent of the voltage across the accelerating gap up to a limit, when field emissions starts (fig. 15) . The design value of Q = 3 .6 x 10 8 was achieved and ever., exceeded, however, not yet reaching for the design voltage U ,a, = 530 kV . The voltage was mea-
sured by observing the energy gain of electrons from a 207Bi source . Oscillations of the frequency caused by acoustic noise were less than 50 Hz . No ponderomotoric oscillations were observed . 4.2 . Beam dynamics The magnets represent a beam transport system guiding the bunches along the spiral orbit . Their fields will be adjusted to keep the beam centered . Due to the alternating gradients the betatron oscillation numbers range from Q x = 1 .2, Q,, = 0 .9 at injection to Q,t = 1 .6, Q,. = 1 .9 at extraction . . In order to get a steady acceleration, the injection energy must correspond to a revolution frequency at
injection, which is almost a subharmonic of the rffrequency . With a proper choice of the injection phase between 0 ° and 90' (increasing accelerating voltage), the bunch is longitudinally focused . Therefore, the revolution frequency starts to oscillate around the value for the subharmonic of the rf-frequency. This leads to incoherent synchrotron oscillations of the ions with respect to the bunch center (Qsi ,, c = 0 .2, typical radial 10-3) . In addition amplitude < 0.5 mm for A p/p < 2 x the bunches will execute coherent synchrotron oscillations with Qsc,, somewhat less than Qsinc . The radial amplitude of these azimuthally fixed coherent synchrotron oscillations is suppressed by the magnet field setting, which oscillates azimuthally corresponding to the energy automatically according to the radial voltage; characteristic of the resonators and dte riecdvd eener ;; gain . A typical central phase curve starts at about 65* at injection and ends at 55 ° passing a minimum of 45 ° . Due to the rather large accelerating voltage per turn, the longitudinal focusing is not very small compared to the radial focusing . As a consequence the coupling between the radial and the longitudinal motion cannot be neglected . To get a stable motion the radial focusing VI . BIJOSTERS
234
U. Trinks / Superconducting cyclotrons
strength has to be sufficiently strong, excluding a weak focusing scheme. To fix the working line with respect to the field indices n t and n2, the limits of the stability diagram were calculated (fig. 16). The working line is directed from right below to the left, if all radially focusing channels with n 2 are operated without current and all axially focusing channels with n t at the same current of -- 500 A (raising n I). Resonances are crossed too fast to become effective . Included in the diagram are limiting lines of constant tolerable sextupole contributions (assumed to be constant in all channels) . From this the upper limit of - 5 x 10-3 for the relative sextupole component at x = ± 4 mm can be deduced . The transversal defocusing of the cavities, caused mainly by the radially directed electrical field components at the entrance and exit of the almost circular beam holes, shifts the working line only slightly. i. Conclusions The successful development of superconducting compact cyclotrons for specific energies of ions up to 150 MeV/u is in an advanced stage, having given a new push to the cyclotron evolution, which started almost 60 years ago. While presently no new projects for basic physics research are in view, the development of rather small superconducting compact cyclotrons for medical applications has started and may lead to the more widespread use of cyclotrons in hospitals. The feasibility studies on superconducting separated sector cyclotrons for specific energies of several 100 MeV/u have not been realized so far . The reason for this has to be seen from the fact, that such an accelerator system could be installed and used in only a very few big laboratories. The main purpose of the development of a small superconducting separated orbit cyclotron in Munich, the Tritron, is to demonstrate the feasibility of this type of cyclotron . It is characterized by extraordinary focusing properties, by flexibility in handling special beam qualities, and by new developments of superconducting magnets and cavites . Future will show whether use can be made of these properties in practise.
Acknowledgements I am very grateful to those who have kindly provided information and illustrations for this paper. References [1] H. Blosser et al., Proc. 11th Int. Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 157. [2] C.B. Bigham et al., Nucl. Instr. and Meth. A254 (1987) 237; see also: H. Schmeing et al., Proc. 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin. [3] D.P. May et al., Proc. 11th Int . Conf . on Cyclotrons and their Applications, October 1986, Tokyo, p. 195, and contribution to the 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [4] E. Acerbi et al., Proc. 11th Int . Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 195, and contributions to the 12th Int. Conf. on Cyclotrons and their Applications, May 1989, Berlin. [5] J.A. Nolen, Proc. 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [6] S. Gales, Proc. 11th Int . Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 184; H.W . Schreuder, Proc. 12th Int. Conf. on Cyclotrons and their Applications, May 1989, Berlin . [7] H. Blosser et al., Proc. 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [8] R. Griffiths, Oxford Instruments Limited, private communication . [9] N.L. Zaplatin, private communication ; J. de Pays. Cl (1984) 893 . [10] P. Mandrillon et al., Proc. 11th Int . Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 203 : contributions to 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [11] ,T. Fermé, private communication . [12] U. Trinks, W. Assmann and G. Hinderer, Nucl. Instr . and Meth. A244 (1986) 273. [13] Proc. 11th Int. Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 779 ; 814 ; 878; 879 ; 880 ; 887 . [14] H.G . Blosser, Proc. 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [15] U. Trinks et a] ., J. de Phys. C1 (1984) 217. [16] U. Trinks and G. Hinderer, Proc. 9th Int . Conf. on MAanPt TP,hnn1AQv Cc+ntennber 1985 Zurich p Ql7
Nuclear Instruments and Methods in Physics Research A287 (1990) 224-234 North-Holland
SUPERCONDUCTING CYCLOTRONS U. TRINKS
Physik-Department, Technische Universität München, D-8046 Garching, FRG
Superconducting cyclotrons with compact geometry and solenoidal coils are operating since several years, indicating the successful introduction of superconductivity in cyclotron design . Some of these facilities use or will use tandems as injectors, some operate as stand-alone machines . The technical problems of this concept are caused mainly by the compactness, which is the attractive feature of the design of course . In order to achieve specific energies above - 300 McV/u separated sector cyclotrons with superconducting magnets are still under consideration. Technical difficulties are related to the large stored magnetic field energy, to the strong electromagnetic forces and to the non-circular shape of the coils. A rather small separated orbit cyclotron with superconducting channel magnets and superconducting cavities is under construction in Munich . This type of cyclotron offers unique focusing properties at relaxed injection and extraction conditions. The large number of individual controllable channel magnets gives some flexibility. On the other hand it demands a high degree of relinhility of man v complex components, requiering simple solutions wherever possible. The present status of the Tritron-project is girtr iii detail .
1 . Introduction
There are three types of superconducting cyclotrons for accelerating heavy ions . First Superconducting Compact Cyclotrons (SCC) were developed, characterized by spiral pole pieces and solenoidal coils inside a circular yoke. Five machines of this type are operating successfully, three additional ones are under construction [1-8]. The maximum specific energy achieved up to now is about 100 MeV/u . Generally it is not limited by the lack of bending or focusing power of the superconducting magnets but by voltage problems in the electrostatic deflectors of the extraction system due to the compact design . To achieve specific energies above 300 MeV/u feasibility studies on Svperconducting Separated Sector Cyclotrons (SSSC) are underway at three laboratories . At the Joint Institute for Nuclear Research at Dubna a full wale sector magnet with superconducting sector coils as well as an accelerating resonator for the first stage DC1 of a neutron spallation source are under construction [9j. For the European Light Ion Medical Accelerator EULIMA for cancer therapy a cyclotron with a pair of rather rig superconducting ring coils and four separated iron sectors with the return yokes outside the coils is under investigation [10] . At Ganil recently a SSSC was discussed as a third stage after the two existing normal conducting separated sector cyclotrons to increase the ion energies from - 100 MeV/u by a factor of - 5 [1 i ] . Finally at Munich a small prototype of a Superconducting Separated Orbit Cyclotron (SSOC) - the Tri(1168-9002/9{1/Sä3 .5() ,- Ellsevier Science Publishers B .V . (N(>rth-Holland)
tron - is under construction [12]. The ion bunches are guided by individually adjustable, narrow superconducting channel magnets along the spiral orbit, and accelerated by six superconducting cavities . The Tritron can be considered a linac curled up, with focusing properties similar to a synchrotron, and some restrictions caused by the cyclotron condition. It represents a linac of - 1 .4.0 m total length inside: a cryostat of 3 .6 m diameter with only six cavities, each crossed by 20 parallel bunches at the same time. On the other hand the extraordinary focusing properties have had to be paid for by a rather large number of channel magnets.
2. Superconducting compact cyclotrons
In fig. 1 a survey on all superconducting compact cyclotron pr(Jects is given. On the lower scale the date of the design start is indicated, on the upper scale the date of the first extracted beam respectively the first internal bean' at design energy [13] . The development was started in late 1973 by H . Blosser at the Michigan State Univet ,ity with the MSU-îC520 machine [1] and at the same A me at the Chalk River Nuclear Laboratories with the TASCC (Tandem Accelerator Superconducting Cyclotron [21) . After several years of design, model tests and finally construction, H. Blossers group obtained the first external beams in August 1982, `r`vitit maximum energies of aVout 55 MeV/u for ions up to °`'O, and correspondingly less for heavier masses. In September 1985 external beams were observed at Chalk
U. Trinks / Superconducting cyclotrons River, too . This pioneering work has triggered several additional projects within a few years . In June 1988 a K520-SCC, which is essentially a copy of the MSU-K520 machine, came into operation at Texas A & M University [3] . In Milan a machine somewhat bigger than MSU-K520 is under construction [4]. It shall be a booster for the 16 MV Tandem of the Laboratorio Nazionale del Sud in Catania, however, it could be operated as a standalone machine with an ECR-source, too. At MSU a still larger SCC was built - MSU K1200 - which came into operation in June 1988, accelerating ions to -100 MeV/u till now [5] . So far the last SCC planned to supply external targets with ion beams is AGOR (K600), which has been under construction at the Institut de Physique Nucl6aire of Orsay since 1985 and which will be transferred to Groningen later on [6] . This cyclotron shall accelerate protons to a maximum energy of 200 MeV, light ions up to 100 MeV/u and heavy ions to somewhat lowtr energies . During the las' few years the main activity on the development of SCCs has shifted to rather small cyclotrons with internal targets for medical applications. The first small superconducting cyclotron for neutron production by 25 MeV/u deuterons on an internal target was developed at MSU again (K100) . It accelerated its first beam in February this year [7] . The cyclotron has a total weight of -- 25 ton only . It can be rotated about a horizontal axis, where the patient may be positioned for neutron therapy. Neither a beam extraction system nor beam transport and rotation systems are needed, nevertheless neutrons can be directed
at the patient from any angle . Only one shielded room is necessary . A still smaller medical cyclotron with a total weight of 3 ton is under construction at Oxford Instruments [8] . Proton beams for the production of short-lived radioisotopes for Positron Emission Tomography are produced by accelerating H - -ions up to 12 MeV and extracting them by stripping at a radius of 22 cm. The time from design start to first extracted beam of the first superconducting cyclotron - MSU-K520 - was 8.5 years . All the following projects (except very small medical accelerators) needed the same period or even more. This may appear rather long to those waiting for the beam. However, it has to be realized, that the superconducting technology requires the installation and operation of a helium refrigerator plant and a cryostat with many feed holes. Strong stresses due to quadratically increasing electromagnetic forces on the coils and due to different thermal contraction have to be overcome. Vacuum leaks, eventually appearing at low temperature only, as well as heat leaks have to be detected in presence of restricted experimental conditions. Each cooling down and warming up cycle of the big masses takes at least several days. In addition the high magnetic field in a compact geometry results in a new design of all other main components, especially of the accelerating system anti the extraction and injection elements . With this in mind the 8 year period appears to be an impressive result . It may be mentioned, that the two big Ganil cyclotrons with maximum energies comparable to those of the MSU-K1200 machine alone FIRST
1375 DESIGN START
1910
225
1985
(EXTERNA~)
BEAM
1990
line) The Fig. l . A survey on superconducting compact cyclotrons with dates of design start (below) and first beam on target (upper are given. maximum specific energies K-number indicates the bending power. Typical ions together with present Vi . BOOSTERS
226
U. Trinks / Superconducting cyclotrons R CONDUCTOR
TM ROD (13 D1 EACH Hilt
Fig. 2. Schematic view of the Chalk River cyclotron .
needed about the same period from the design start to first beam. By the way, the beam intensities at Ganil are about one order of magnitude larger than those at MSU-K1200 . The basic concept of all superconducting compact cyclotrons is essentially the same. Fig. 2 shows a schematic view of the Chalk River superconducting cyclotron, fig . 3 a midplane section. This cyclotron was specified to accelerate ions from lithium (50 MeV/u) to uranium (10 MeV/u), injected radially from outside from a 16 MV tandem by poststripping at an injection radius ranging from 14.5 cm to 22 cm, and extracted at a radius of 65 cm. The bending field is produced by a pair of split superconducting solenoids, operated at rather low current densities of < 35 A/mm2 in a cylindrical He bath cryostat . Thus the coils are cryostable, which guarantees safe operation. Each cylindrical pole is supplied with four spiral ion hills for axial focusing. The pole gap is 4 cm in the hills and 64 cm in the valleys, the maximum magnetic induction 6 T and 4.3
T, respectively . The isochronous field is achieved by adjusting the two currents in the split coils properly and by axial positioning of 104 trim rods located in the magnetic flutter poles . Four accelerating dee's are installed in the valleys forming a single resonant cavity with two coaxial line tuners several meters long on the axis of the machine. They shall produce 800 kV accelerating voltage per turn in the maximum with frequencies from 31 to 62 MHz, corresponding to the 2nd, 4th and 6th harmonics . Single turn resonant extraction from the cyclotron is initiated by means of the outermost trim rods. Then the beam is deflected by an electrostatic channel, positioned inside a dee . The steep radial decrease of the bending field at the rim of the poles causes strong radial defocusing forces, which are compensated by two saturated iron lenses on the wall of the coil cryostat . Finally the beam enters a narrow long magnetic channel consisting of 78 separate superconducting coils and some additional saturated iron gradient bars. The other superconducting compact cyclotrons differ
U Trinks / Superconducting cyclotrons
N V,
LY,
C'
ZO
I7
0
=
cß 0.
v e~
,
04
ad
2 oG aL t ,-
'AA
O N
a
V_
a
MN
N
~ ,
.
~, ö gBS t
ô U .40 E
caz~x-a x wr` M1` ~"V OM OM fV e-. 00 r- 00 N t- O> C4
âo uÔ N
aaG (: ?
c
00 c
~ W 0 a.
câ"X . . ,n,
a
N cV ~O N
NO
°
Crlô
E
Fig. 3. Midplane section of the Chalk River cyclotron : A. yoke, B, cryostat ; C, stripper foil; D, injection steering dipole ; E, electrostatic deflector ; F, magnetic channel; G, effective field boundary; H, magnetic lens.
ô
âi
00
from the Chalk River machine eventually by the axial injection from ECR-sources, by the number of sectors (3 instead of 4) by trim coils on the hill sectors instead of the trim rods and by details of the rf and extraction system. In table 1 some parameters of all SCCs are given. In principle, the highest specific energy for a projectile with specific charge QIA depends on the bending power k 1, and the focusing power k t of the magnetic field, on the accelerating voltage of the deer and on the Meld of fe electrostatic deflector of the extraction system. Generally the design values of the magnetic field properties were achieved or even exceeded. So the bending limit
00
02 Â 3 oâ
00.
0
Ô N c 0 .r C
.~é
0 u 00 ä ç aU. w a, O H w 0 aa
a c
N S N kn 5, W, N
00
~-.
c
0o c14 N 0 00 N N N ~ q V1
C
O O r- O v4 00 kn N
M
1i
M N II IW
M M -Tr M e- c n M
r°4
b ei M M r~ O ~t d' ~R ef n Cq CPR q ci a vi vi ~ ~ \ ~i
U r U
ca,
CM .2 k W
H
r.
n i
E 1- cc w .~
0 0 o ~ T .o U â
227
Q i i
Q
N 00 %n 00 00 00 00 c ß~
N M ,
S U aa Q
01
u E
u ,0
0
depends linearly on Q/A, the same is true for the dee voltage limit and for the electrostatic deflector limit . All ,à®cJ these ilmlts should be matched . Aciually contact dee stems and overheating of various joints in the causes the dee voltage limed to be more severe than the focusing limit. However, the most serious limitation for all SCCs but the medical accelerators stems from voltage problems at the first electrostatic deflector of the extraction line. Table 2 shows, that only about 70% of the design values were achieved in the real environment inside the cyclotrons. An additional problem of the extraction system is the rather poor transmission. typia,
01 00
0 . ß
N
N N
w
00 00
T < kb(QIA) 2 (1 ) restricts the maximum energy only of very heavy particles with low Q/A value . The focusing limit T < k i (Q/A) (2)
E
v 1.
BOOSTERS
U. Trinks / Superconducting cyclotrons
22 8
Table 2 Design and routinely achieved fields of the electrostatic deflectors of SCCs with external beams Project
[kV/cm]
140 140 140 125 140 105
MSU-K520 TASCC Texas MSU-K1200 Milan AGOR
Gap [mm]
Eroutine
design
[kV/cm]
7 7 7 6
84 86 93 92 -
rally 600 or even less, indicating modest beam quality considering the geometrical limitations on the extraction channel . This experience was taken into account at the design of the AGOR-project [6] . The maximum magnetic bending field is less than that of the previous projects, the extraction radius is correspondingly increased and the requirements on the voltage, of the electrostatic deflector are reduced . 3. Superconducting separated sector cyclotrons .he turn separation decreases drastically with increasing energy : Ar=
e Q 1 rU, moc2 A ß2 Y 3
where m cc 2 /e = 931 .5 MV and U is the accelerating voltage per turn . For nonrelativistic velocity ®r decreases approximately linearly with increasing energy . Above ß = 0.8 corresponding to T = 600 MeV/u it decreases dramatically, finally as the energy cubed, as fig . 4 demonstrates . To obtain a certain turn separation 20
1
T ( GeV/ul
.ô
1
0
Fig. 4. The quantity G = ß 2 Y3, which craters into eq . (3) for the turn separation . versus the velocity respectively the specific energy Tof ions .
Fig. 5. General layout of the EULIMA cyclotron.
both the extraction radius and the accelerating voltage have to he sufficiently large. Thus strong average bending fields become increasingly useless if high energies are wanted, while more and more space for the installation of an increasing number of accelerating gaps is needed . These requirements finally rule out the compact cyclotron design, favouring separated sector magnets instead. Superconductivity then may be used to improve the field flutter and by this the focusing power, avoiding spiral-shaped superconducting coils, which are more difficult to produce than simple sector coils. The EULIMA-project with big superconducting ring coils with inner diameter 4.4 m and outer diameter 5.2 m, four separated iron yokes outside and spiral pole pieces inside represents an intermediate step in the development from compact to separated magnet design [l0] . The general layout is shown in fig. 5. It is a feasibility study on an accelerator for cancer treatment, therefore C, 0 and hle-ions shall be accelerated to energies up to 400 MeV/u . The total iron weight of 4 x 155 ton would exceed only slightly that of a superconducting compact magnet with comparable bending and focusing power, which was proposed by H. Blosser [14]. Axial injection from an ECR-source, eventually combined with a booster RFQ, is planned . Though the EULIMA concept appears to be an attractive solution to obtain inn energies of about 400 MeV/u, a conventional synchrott on dedicated for medical applications could be a realistic alternative due to the easy energy variation. To achieve even higher energies totally separated sector magnets are needed . The first feasibility study on a superconducting sector coil for a SSSC was performed at Munich [15]. Presently a complete sector magnet with two superconducting coils is under construction at Dubna [9] (fig. 6) In order to produce the field effectively the overall current density should be as high as tolerable with respect to safe protection of the coil in case of a quench, when the stored magnetic energy of
U. Trinks / Superconducting cyclotrons
229
Fig. 8. Test winding of a subcoil of the r)Cl-magnet .
Fig. 6. The ¢,zst sector magnet of DCl/Dubna. Height : 2.4 m, weight : 35 ton .
several MJ is dissipated into heat . At the Munich coil a new protection mechanism was tested (cross quenching [161), which enables overall current densities of the potted coils of more than 100 A/mm2 compared to 35 A/mm2 of cryostable coils, which generally need not be de-energized in the case of a local quench of course . The Dubna coils are wound directly on sectorshaped iron cores (figs . 7 and 8). The iron contribution to the field is optimized by this design, especially in the
central corner of the sector, no space is lost by magnetically dead material . Furthermore the thermal contraction of the coil exceeds that of iron slightly, so that the coil will be prestressed automatically . Finally the steel core attracts the coil, so that the strong outward directed forces or, the sides of the sector coil, trying to make the coil more circular, are reduced drastically. A test of the complete magnet as well as a test of a prototype cavity with full power will be performed presumably within one year . Up to now none of these feasibility studies on big superconducting separated sector cyclotrons for heavy ion beams with energies of several 100 MeV/u has been realized. ®f course very recently a proposal for a third, superconducting sector cyclotron for GANIL to achieve energies up to - 500 MeV/u is under preparation. It appears not unlikely, that this project actualy will be realized .
e superconducting separated orbit cyclotron At the Munich Accelerator Laboratory a superconnyni-non the - 4'tl1'l tthe :'.r. e ~p~r s® A-1-Al w v74iV 3vaa booster for the existing MP under construction as a tandem [121 . It shall increase the ion energies by a factor of - 4.9, e .g. a '2C6 + beam will have a maximum energy of 20 MeV/u, protons about 43 MeV. The main purpose of the project is to demonstrate the feasibility off this type of cyclotron . The characteristic features of the Tritron concept are the strong transversal and rather strong longitudinal focusing forces, the flat superconducting magnet sectors and the big superconducting
~urtin_g separated orbiti. c :üi~al'vi .3 orbit -- i7 a
SS=_
Fig . 7 . Cross sections of a DC1-coil . In (c) : 1, superconducting, potted windings; 2, iron cores ; 3, stainless steel cages .
VI . 13D0 TERS
230
U. Trinks / Superconducting cyclotrons
accelerating cavities. The excellent beam qualities of the tandem injector are conserved . For a 43 MeV-proton beam for instance an energy width of 8T = ± 34 keV is expected at a time spread 0t = 25 ps. By mismatched injection into the Tritron and an additional debuncher behind it the energy width can be further reduced to ± 2 keV at the expense of the time spread and vice versa. A first proposal was presented in December 1983, the definitive funding followed in July 1987 . Fig. 9 shows cross sections of the Tritron, table 3 the
main parameters . The beam is led along a spiral orbit of almost 20 turns with constant turn separation A r = 4 cm. Each of the 12 magnet sectors consists of 20 or 19 superconducting channels of the window frame type with maximum fields of - 1.4 T. In order to guide the bunches on the central path through the magnets, each channel can be adjusted individually by means of superconducting switches across the magnet coils. The radial beam positions will be measured by beam profile monitors installed in each second intermediate sector . There
Fig. 9. Tritron cross sections. M, magnets; R, resonators ; V, vacuum vessel ; S, 80 I{. shield ; He, liquid helium reservoir; support. T,
U. Trinks / Superconducting cyclotrons
231
Table 3 Tritron design data Injector Max . energy H1 + (Q/A =1) S16+ (Q/A = 0.5) Ag32 (Q/A = 0.29) Injection/extractions radius Turn separation A r Harmonic numbers Magnet sector data: Number of magnet channels Sector angle Bending radius pl, P20 Maximum magnetic field Bmax Radial gradients 4 -1 SB/Sr Dimensions of the supercond . cable Number of strands Strand diameter Strand material Cu/NbTi Maximum cable current Imax Cavity data: Gap width at injection/extraction Rf-frequency Maximum accelerating field Emax Epeak /Emax
Maximum voltage at r 2 Stored energy Dissipated power P Quality factor (unloaded) Qa Beam power Geometry factor G Surface resistance R,., = G1Q0
13 MV tandem 43.5 MeV 20.1 MeV/u 7.6 MeV/u 66 cm/145 cm 4 cm 18-53 20(19) 20 ° 430 mm/942 mm 1 .4 T 3.6 m - 1- -5.2 m -1 0.7 x 2.9 mm2 14 0.4 mm 1.35 1400 A 60 mm/130 mm 170 MHz 4.7 MV/m < 1.5 530 kV - 1 .7 J 6W 3.6x108 _ 200 W 94 SZ 2.6 x 10-7
are alternating gradients from one sector to the other to get strong focusing in both transversal directions . Six superconducting resonators operating at 170 MHz shall accelerate the ion bunches along the 20 parallel turns . To get longitudinal focusing, the bunches have to cross the cavities at a rf-phase with increasing voltage. The beam is injected through three superconducting channel magnets. The whole machine hangs under a torus-like helium reservoir on the upper half of the cryostat. The cavities and magnets are cooled indirectly by pipes connected to the torus (then-mal siphon cooling). There is no special vacuum system for the heam or the cavities. The cryostat including the torus and all transfer lines exists . The stand-hy losses of the cryostat are less than 5 W. A refrigerator (150 W at 4.7 K) and all transfer lines are installed . 4.1 . The Tritron magnets
The magnet sectors consist of two steel sheets with slots every 4 cm (fig. 10) . Th,- sectors are mechanically identical up to the coils generating the alternating gradient. The coils are wound by a computer controlled
Fig. 10. Vertical cross section of a magnet sector. Fe, iron yoke; G, gradient windings ; Cu, copper shields ; D, dead insulating layers. winding machine (one channel/day) directly into the slots and then potted. They are shielded from beam losses by Cu profiles leaving a beam hole of diameter 11 mm. The coils are supported by the copper profiles and disc springs between . The dimensions of the Tritron magnets are about one order of magnitude less than the accelerator magnets constructed before . Correspondingly the requirements cn the tolerances are rather tight. Up to now seven channels for test purposes were produced. A maximum current of 1840 A was achieved without training, which is well above the design value (fig. 11) . The effective magnetic length measured by Hall probes is in good agreement with TOSCA calculations. The stray field 35 mrn in front of the end plates is less than 0.5 G, so no problems due to frozen-in-flux in the superconducting layers of the cavities are expected. To analyse higher order components in the radial field expansion narrow, bent superconducting induction loops were used, measuring the change of the magnetic length if the radial position x is varied . Three channels were investigated in detail with corresponding results, demonstrating the reproducibility of the production process . A typical measurement of the effective length related to the central len gth at x = 0 is shown in fig. 12. The
Test e
Design
OÔ Fig. 11. Critical current limits at 4.2 and 7 K versus field. VI . BOOSTERS
U. Trinks / Superconducting cyclotrons
23 2
Fig. 12 . Effective magnet length related to that at the magnet centre versus x . No current in the gradient coil . sextupole component, at x = f 4 mm, given by the deviation of the curve from the tangent at x = 0, is less than - 4 x 10-3, which is within the tolerable limit of 5 x 10 -3 . Contributions come from the entrance and exit edges, from a slit between the two halfs of the coil due to different thermal contraction of the coil and the steel, and from the gradient windings . It may be further reduced by adding dead spaces at the upper and lower edges of the coil, respectively, which will produce positive sextupole contributions (see fig. 10) . A new test unit of three channels with such dead spaces has just been wound . Presently the steel plates of all sectors are premilled, of them are finally milled and ready for being polished . The superconducting cable exists, yet without insulation . The big winding machine is ready for first tests. A separate cryostat for testing single sectors is under construction . 10 n J
4.1 . The superconducting cavities The 6 wedge-shaped reentrant type resonators consist of half shells which are connected in the horizontal plane, so no currents will cross the joint. They are fabricated by electroplating copper (-- 10 mm) onto a fibre glass mould, which is destroyed and removed afterwards. The inner surface is electroplated with - 10 li m thick PbSn (2-4% Sn) . The first cavity exists, the next will be delivered soon . Fig. 13 shows the radial voltage and accelerating field characteristics . The rf-superconductivity properties of PbSn were investigated systematically by means of a reentrant type R 10 6
RBCS
tô'~
n
11
0 .6 04 02
/
IQ] -~
i
/ 1/
/
/
-
Q`O / /CP a
T
t
lu . 0. . 66
90
1O
â0
Radius [cm]
Fig. 13 . Voltage and electrical field characteristic of the cavity in relative units and half of the horizontal cross section .
0.2
0.4
06
0.8
1
f 1GHzl
2
Fig. 14 . Surface resistance R s of the test cavity (O) at different modes above 490 MHz and of the Tritron cavity (6) at 170 MHz. The data on R BOs (e) were obtained from the temperature dependence of R s . Alloy : PbSn with -- 2% Sri .
U. Trinks / Superconducting cyclotrons
Q6
,,,
23 3
T-4 .2K T z_5 K
N 2
a)
ee e
ted* ®
%,
e®
e e e
e
e
®
=®
e
e
e .,C)
10-!8
N1
Fig. 16 . Stability diagram and working line. n 1 , n 2 field indices of axially and radially focusing channels . Included are stability limits for different sextupole contributions at x = t4 mm, calculated for transversal emittances of 6 iT mm mrad .
e
0
100
200
300
400
500 Umax NV l
Fig. 15 . Some measurements of the unloaded quality factor Qo of the Tritron cavity versus the maximum gap voltage . Curve (a), fresh PbSn-surface, voltage limited by multipacting, caused by a small piece of steel inside the cavity. Curves (b) and (c) : after removal of the steel piece. (b), before and after heliumconditioning of the cavity . Voltage limited by field emission.
cavity of rotational symmetry, 45 cm in diameter, and a frequency of 490 MHz in the fundamental mode . The Sri critical temperature of PbSn with - 2% is 7 .4 K 7 .2 of Pb. The surface compared to K resistance R s is R the BCS-theory and the residual the sum of scs from R resistance REs due to imperfect surfaces (independent
of temperature) . R s was measured at 4.2 K in the range from 490 to 1670 MHz by operating the cavity in different modes (fig. 14) . The R Bcs-values were obtained from the temperature dependence of R s . The fitted line gives a frequency dependence -f' -' compared to f' .74 for Pb. The PbSn-line is about a factor of
two below the Pb-line. The observed improvement can be explained by the influence of the electron mean free path on R Bcs . It is remarkable, that the residual resistances of the test cavity are not too far above the R Bcs-line, which may be due to the increased chemical stability of the alloy with respect to oxidation . The surface resistance at 170 MHz in fig. 14 was obtained for the Tritron cavity with the best surface up opared to now, prepared c~avu wish several rinsing courses and fast drying by N 2 -blast following the electroplating process . It fits rather well into the data taken at higher frequencies . The unloaded quality factor Q o for this cavity is almost independent of the voltage across the accelerating gap up to a limit, when field emissions starts (fig. 15) . The design value of Q = 3 .6 x 10 8 was achieved and ever., exceeded, however, not yet reaching for the design voltage U ,a, = 530 kV . The voltage was mea-
sured by observing the energy gain of electrons from a 207Bi source . Oscillations of the frequency caused by acoustic noise were less than 50 Hz . No ponderomotoric oscillations were observed . 4.2 . Beam dynamics The magnets represent a beam transport system guiding the bunches along the spiral orbit . Their fields will be adjusted to keep the beam centered . Due to the alternating gradients the betatron oscillation numbers range from Q x = 1 .2, Q,, = 0 .9 at injection to Q,t = 1 .6, Q,. = 1 .9 at extraction . . In order to get a steady acceleration, the injection energy must correspond to a revolution frequency at
injection, which is almost a subharmonic of the rffrequency . With a proper choice of the injection phase between 0 ° and 90' (increasing accelerating voltage), the bunch is longitudinally focused . Therefore, the revolution frequency starts to oscillate around the value for the subharmonic of the rf-frequency. This leads to incoherent synchrotron oscillations of the ions with respect to the bunch center (Qsi ,, c = 0 .2, typical radial 10-3) . In addition amplitude < 0.5 mm for A p/p < 2 x the bunches will execute coherent synchrotron oscillations with Qsc,, somewhat less than Qsinc . The radial amplitude of these azimuthally fixed coherent synchrotron oscillations is suppressed by the magnet field setting, which oscillates azimuthally corresponding to the energy automatically according to the radial voltage; characteristic of the resonators and dte riecdvd eener ;; gain . A typical central phase curve starts at about 65* at injection and ends at 55 ° passing a minimum of 45 ° . Due to the rather large accelerating voltage per turn, the longitudinal focusing is not very small compared to the radial focusing . As a consequence the coupling between the radial and the longitudinal motion cannot be neglected . To get a stable motion the radial focusing VI . BIJOSTERS
234
U. Trinks / Superconducting cyclotrons
strength has to be sufficiently strong, excluding a weak focusing scheme. To fix the working line with respect to the field indices n t and n2, the limits of the stability diagram were calculated (fig. 16). The working line is directed from right below to the left, if all radially focusing channels with n 2 are operated without current and all axially focusing channels with n t at the same current of -- 500 A (raising n I). Resonances are crossed too fast to become effective . Included in the diagram are limiting lines of constant tolerable sextupole contributions (assumed to be constant in all channels) . From this the upper limit of - 5 x 10-3 for the relative sextupole component at x = ± 4 mm can be deduced . The transversal defocusing of the cavities, caused mainly by the radially directed electrical field components at the entrance and exit of the almost circular beam holes, shifts the working line only slightly. i. Conclusions The successful development of superconducting compact cyclotrons for specific energies of ions up to 150 MeV/u is in an advanced stage, having given a new push to the cyclotron evolution, which started almost 60 years ago. While presently no new projects for basic physics research are in view, the development of rather small superconducting compact cyclotrons for medical applications has started and may lead to the more widespread use of cyclotrons in hospitals. The feasibility studies on superconducting separated sector cyclotrons for specific energies of several 100 MeV/u have not been realized so far . The reason for this has to be seen from the fact, that such an accelerator system could be installed and used in only a very few big laboratories. The main purpose of the development of a small superconducting separated orbit cyclotron in Munich, the Tritron, is to demonstrate the feasibility of this type of cyclotron . It is characterized by extraordinary focusing properties, by flexibility in handling special beam qualities, and by new developments of superconducting magnets and cavites . Future will show whether use can be made of these properties in practise.
Acknowledgements I am very grateful to those who have kindly provided information and illustrations for this paper. References [1] H. Blosser et al., Proc. 11th Int. Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 157. [2] C.B. Bigham et al., Nucl. Instr. and Meth. A254 (1987) 237; see also: H. Schmeing et al., Proc. 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin. [3] D.P. May et al., Proc. 11th Int . Conf . on Cyclotrons and their Applications, October 1986, Tokyo, p. 195, and contribution to the 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [4] E. Acerbi et al., Proc. 11th Int . Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 195, and contributions to the 12th Int. Conf. on Cyclotrons and their Applications, May 1989, Berlin. [5] J.A. Nolen, Proc. 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [6] S. Gales, Proc. 11th Int . Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 184; H.W . Schreuder, Proc. 12th Int. Conf. on Cyclotrons and their Applications, May 1989, Berlin . [7] H. Blosser et al., Proc. 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [8] R. Griffiths, Oxford Instruments Limited, private communication . [9] N.L. Zaplatin, private communication ; J. de Pays. Cl (1984) 893 . [10] P. Mandrillon et al., Proc. 11th Int . Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 203 : contributions to 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [11] ,T. Fermé, private communication . [12] U. Trinks, W. Assmann and G. Hinderer, Nucl. Instr . and Meth. A244 (1986) 273. [13] Proc. 11th Int. Conf. on Cyclotrons and their Applications, October 1986, Tokyo, p. 779 ; 814 ; 878; 879 ; 880 ; 887 . [14] H.G . Blosser, Proc. 12th Int . Conf. on Cyclotrons and their Applications, May 1989, Berlin . [15] U. Trinks et a] ., J. de Phys. C1 (1984) 217. [16] U. Trinks and G. Hinderer, Proc. 9th Int . Conf. on MAanPt TP,hnn1AQv Cc+ntennber 1985 Zurich p Ql7