INTRODUCTION
There is a long tradition of shortly pulsed, low repetition rate,
room temperature (r.t.) synchrotron injector linacs for protons and
H− with beam currents up to 300 mA. The development of
superconducting (s.c.) linac structures since the mid-1960s has been
strongly motivated by high duty cycle applications. Elliptical
multicell cavities for velocities β ≥ 0.5 as well as single-,
two-, and three-cell structures for low- and medium-β profiles are
available now (Padamsee, 2001). After the
successful development of input power couplers, up to 400 kW, RF power
can be fed into a s.c. cavity by one feeder in cw operation (Schmierer et al., 2001; Campisi, 2002).
Pulsed operation of s.c. multicell structures became possible through
the use of mechanically stabilized cavities equipped with fast tuners
and with powerful digital controls (Simrock,
2002; Vardanyan et al.,
2002).
Recently, the development of multicell s.c. activities for the
fusion-driver-relevant β-range has started (Eichhorn et al., 2002; Zaplatin et al., 2002). It may lead to a
very compact and power-efficient driver linac solution at beam energies
above 5 AMeV. The s.c. multicell CH-linac development at IAP is
reported in this article.
A very flexible cw ion linac design for ions ranging from 1-GeV
protons to 400-AMeV uranium is investigated at the Argonne National
Laboratory (ANL) and at Michigan State University (MSU) within the Rare
Isotope Accelerator (RIA) project. An interesting feature is the
simultaneous acceleration of multiply charged ions after stripping
processes (Ostroumov, 2001). Up to the heaviest
elements, a total transmission of 80% should be reached by accepting two
charge states from the Electron Cyclotron Resonance (ECR) ion source and
including two stripping processes at 12.3 AMeV and at 85.5 AMeV. Although
the application of such a concept on a future fusion driver scenario is not
obvious to date, it could have an impact on the design of shortly
pulsed high current heavy ion injectors.
The currently most powerful heavy ion linac, UNILAC, at GSI, has a
beam transmission of only around 1.5% for uranium through the inclusion
of two stripping processes at 1.4 AMeV and at 11.4 AMeV, and the
acceptance of one charge state per section only. In 1999, the 1.4-AMeV
Wideroe section was replaced by the 91-MV High Current Injector HSI to
increase the heavy ion beam current by more than two orders in
magnitude (Ratzinger, 2001; Barth, 2002). This 15-mA, U4+ injector
consists of a 120-AkeV RF Quadrupole (RFQ) followed by a 11-cell RFQ
“Superlens” matching the beam to the Interdigital H-type Drift
Tube Linac (IH-DTL). This article will describe the HSI status and planned
activities for further improvement.
The suitability of H-type structures to accelerate intense
Bi1+-ion beams was discussed earlier (Ratzinger & Tiede, 1998). A detailed beam
dynamics study with respect to the H-DTL front end shows that the key
elements are the transversely focusing lenses. The requested parameters
are quite similar to those of the induction linac approach, where
promising results have been obtained already with s.c. quadrupole
prototypes (Faltens et al.,
2002).
H-type structures offer attractive features for high current heavy
ion acceleration like the small transverse dimensions at a given
resonance frequency, the option to realize multi-aperture cavities, and
the chance for innovative RF powering concepts (Setzer
et al., 2000).
2. SUPERCONDUCTING CAVITY DEVELOPMENT AND
CONSEQUENCES FOR A HIF DRIVER LINAC
At beam velocities β > 0.5, several coupled cavity structures
have been developed successfully. Typically four to six elliptical
cells are coupled in the π-mode through the aperture and are tuned
to a constant voltage profile (Ciovati et
al., 2001). The 350-MHz linac within the CERN-LEP ring was
operated with great success in the cw mode. For a high beam load,
powerful RF input couplers are already available (Schmierer et al., 2001; Campisi, 2002), and the present power limit of
around 400 kW per coupler in cw operation does not seem to be a final
technical limit. After a careful analysis of mechanical cavity
resonances along with the increasing capabilities of digital controls,
it became possible to operate s.c. cavities reliably at well-defined
pulsed time structures (Simrock, 2002). At
the velocity range of interest for a heavy ion fusion driver with β
≤ 0.35, there are no well-tested multicell s.c. cavities existing so
far. Only two-, three- and four-gap s.c. cavities have been
successfully operated for many years at nuclear research facilities,
where variable beam energies are needed (Fortuna
et al., 1996). It seems unreasonable to base a fixed
velocity, 10-GV linac on about 12,000 low-β s.c. cavities. On the
other hand, by using multicell cavities, one could aim to keep the
total number of cavities below 2000. At present, two types of multicell
cavities for a β-range from about 0.1 to 0.5 have been developed:
the CH cavity and the multicell versions of the spoke cavity (Shepard et al., 1998; Zaplatin et al., 2002). IAP has designed a
1.05-m-long 350-MHz, 19-cell prototype CH cavity (see Fig. 1) in collaboration with industry (ACCEL
Company, Bergisch-Gladbach, Germany) and has equipped a cryolaboratory
with a vertical cryostat (φi ∼ 600
movement, H = 3 m) and with a class 100 laminar flow box to perform
test runs with that cavity. First results are expected in 2004.
Moreover IAP is designing s.c. 175-MHz CH cavities for high current
applications at the HIF-relevant velocity range (Sauer et al., 2002). Table 1 shows the characteristic parameters of the
prototype cavity, resulting from Microwave Studio calculations. The low
maximum magnetic field levels at a given effective voltage gain are
promising. The ratio between the magnetic and the electric peak field
Bp /Ep is
7.3 mT/(MV/m). This value is about one-third lower than for
typical spoke resonators. On the other side, the electric peak field is
relatively high (Ep
/Ea = 6.59). This means that the s.c.
CH cavity will most likely be limited by field emission. An effective
acceleration gradient of 3.2 MV/m leads to a peak field of 21
MV/m, which still seems to be a moderate value. Tests will show
whether the relatively high electric peak fields localized at the drift
tube ends are tolerable. Alternatively the drift tube geometry must be
modified in a next step.
Characteristic parameters of the s.c. prototype cavity
Superconducting prototype CH cavity, designed by IAP, University of
Frankfurt (supported by BMBF, contract no. 06F134I, and by IFMIF) and
under construction at ACCEL Instruments GmbH, Bergisch-Gladbach,
Germany.
3. STATUS OF THE GSI HIGH CURRENT INJECTOR
The new 1.4-AMeV, 36-MHz High Current Injector has replaced the
Wideroe section of UNILAC and has been in routine operation since
October 1999 (Ratzinger, 2001). The beam
current design limit of the linac was reached by delivering 10 mA of
Ar1+. From a total U4+ beam current of 11 mA in
front of the quadruplet focusing into the RFQ, 6 mA were accelerated by
the HSI and delivered to the gas stripper at the beginning of the year
2003. This was a factor of 2.5 below the U4+ current limit.
Improvements of the MEVVA-ion source, of the electrostatic acceleration
geometry, and of the low energy beam transport section as well as of
the RFQ front end are under investigation to reach the design current of 15
mA for U4+ (Barth, 2002). Based on the
newest emittance measurements on U4+ beams at 2.2 AkeV just behind
the high voltage gap, it seems realistic to achieve up to 11 mA at the
HSI exit after minor hardware changes.
3.1. Superlens and RFQ
The Superlens (Ratzinger, 2001), which is a
short, 11-cell, 36-MHz IH-RFQ, showed an intolerably high dark current
load, when operated at U4+ levels. During the 2001/02
winter shut down, the mechanically polished RFQ electrodes from bulk
Oxygen-Free High Conductivity (OFHC) copper were replaced by
galvanically copper-plated ones and the electrode distances towards the
end walls were increased from 7 mm to 12 mm, as the dark currents were
concentrated at the electrode ends. As a consequence, the dark currents
are substantially reduced now (see Fig. 2). A
voltage level of 212 kV at a minimum vane–vane distance of 8.3 mm
is reached easily. This corresponds to A/q = 65 or to 9 V in
arbitrary units as used in Figure 2. An input
RF power of 100 kW has been measured, which results in a dark current
contribution of 20 kW at a neglectable level of reflected power. This
is acceptable at low duty cycle operation.
Reduction of the dark current contributions to the total RF power
losses by enlarging the distance of the quadrupole electrode ends
toward the cavity and flanges. Additionally, the minivans are
electrolytically copperplated now. The nominal vane–vane voltage
for U4+ corresponds to 194 kV (8.2 V in arb.
units).
When compared to the main IH-RFQ (9.4 m long, 356 cells, mechanically
polished electrodes), the dark current contributions per unit length at
similar electrode surface fields are now lower at the Superlens. There
is some evidence that it is advantageous to copperplate (20 μm
thick, typically) the RFQ electrodes. Moreover, no significant
reduction of dark current contributions was seen at the RFQ during
several months of operation at a high RF voltage level, while the
Superlens after a few hours of operation reached the performance as
plotted in Figure 2.
3.2. IH-DTL
The 40-MV IH1 cavity contains three internal quadrupole triplet
lenses (Ratzinger, 2001). During shut-down
inspection in January 2002 serious damage at the lens housing of the
first triplet was found. It is suspected that in the case of special
two-beam operation modes (50-Hz operation with 5-ms pulses at reduced
RF level and 1-ms pulses at high RF level in parallel), a
well-localized electron current toward a neighboring drift tube caused
local overheating and, as a consequence, melted the copperplated
stainless steel end plate. This damage happened twice at exactly the
same position. As the maximum heat load without dark current
contributions was estimated to remain below 0.05 W/cm2,
direct water cooling was provided only for the cylindrical surfaces but
not for the end plates of the lens housings. Air cooling is applied to
the inner housing surfaces (including the beam tubes).
Meanwhile, a second generation of IH internal lenses with directly
water cooled lens end plates was fabricated and the first lens of IH1
was exchanged in January 2003.
The geometry of the damaged asymmetric cell in tank IH1 was modified
(Liebermann et al., 2002) by reducing
the surface peak field at the drift tube after the lens housing (Fig. 3, pos. B). Perturbation measurements show that
this was the absolute IH1 peak field so far (Ratzinger
et al., 1999).
Longitudinal electric field distribution along path AB in gap no. 10
of the HSI-IH1 cavity. The drift tube (bulk copper) behind of the
quadrupole triplet lens was shortened by 5 mm, the end curvature was
increased from 3 mm to 5 mm radius, and the electric capacity between
lens and drift tube stem was reduced by a diminished stem
geometry.
There are some indications now, that an RF-power-level-dependent
vacuum leak in the tank end flange water circuit caused that drastic
lens damage described above. Meanwhile, such a leak was detected and
eliminated.
In the case of U4+, the dark current contributions to the
RF power losses are about 25% for IH tank 1 and 15% for IH tank 2. Dark
current contributions are clearly detected at operation above 60% of
the U4+ power levels.
The operation experience after these changes is quite promising. The
HSI is providing again simultaneous two-beam operation at power levels
and duty cycles corresponding to U10+ (30% d.c.) and
U4+ (2% d.c.).
4. DEVELOPMENTS WITH RESPECT TO THE
Bi1+ LINAC FRONT END
The high A/q ratio in combination with the high beam intensity
makes the development of new concepts and components necessary,
especially along the linac front end and up to the last funnel
section.
4.1. Bi1+ test injector
It is obvious that a high terminal voltage eases the beam transport
from the ion source to the RFQ. It was envisaged to base the
Bi1+ ion source development at IAP on a 500-kV platform
(Ratzinger, 2001). As an intermediate step, a
300-kV platform is under construction at GSI. At that test stand, IAP
plans to continue the development of an extraction and acceleration
geometry fitting to the Bi1+ plasma generator. The plasma
generator itself will be further optimized at an IAP test stand. The
Bi1+ fraction of the full beam from the IAP volume source is
93%, which is favorable for a compound extraction and postaccelerator
setup. A powerful slit-grid emittance measurement device was
constructed (PET, Darmstadt) and has been successfully tested with beam
at IAP.
Low energy beam measurements by CCD cameras will be further improved
(Jakob et al., 2002). The aim is to
develop a destruction-free emittance measurement, which will no longer
influence the conditions of space charge compensation as conventional
detectors do.
4.2. Investigations on Gabor lenses
Gabor lenses (GLs) are very powerful in focusing low energy heavy ion
beams (Mobley et al., 1979; Meusel et al., 2000). The positive ion
beam is passing an electron space charge cloud, which is stabilized
within an insulated beam tube on positive voltage
Va by superposition of a longitudinal
magnetic field BZ. Gabor related the
transverse cloud confinement to the condition for Brillioun flow. This
results in the following equation for the electron density
n:
Longitudinal confinement is provided by the electrostatic potential.
The space charge limit is reached in that direction when the on-axis
potential within the tube is decreased to zero by adding the space
charge potential to Va. Assuming a tube
filled homogeneously with electrons up to the aperture radius
a leads to a corresponding electron density:
The longitudinal confinement in praxis depends on the longitudinal
electron temperature as well as on the real radial density distribution
of the cloud that defines the useful lens aperture. In the IAP-type GL,
the electron production is accomplished by a gas discharge (typical
current range 0.1 mA to 10 mA) and by beam collisions with the rest
gas. No heated cathode is used. A filling factor is defined by relating
the electron density prediction of Eq. (1) to the experimental
result:
At IAP, a test stand for GLs was constructed. A 14-keV, 10-mA
He+ beam is matched by two GLs
(BZ = 7.8 mT,
Va = 1850 V, l1,2 =
0.1 m, ra = .03 m) to a 108-MHz RFQ. The
measured focal length of 0.33 m for each of these lenses corresponds to
a filling factor of about 0.32.
After acceleration to 440 keV, the bunched He beam will again be
focused by a powerful Gabor lens with the design values
BZ = 0.2 T,
Va = 65 kV, l = 0.3 m,
ra variable up to about 80 mm (O. Meusel,
pers. comm.). One GL of that type was constructed (see Fig. 4) and will be commissioned with beam during
2004. A second one is under construction. It is envisaged to test this
lens also at the GSI high current test injector with 150 keV
U4+ beams. Figure 4 shows a
cross-sectional view of the lens and a photo of the completed setup.
Photo and scheme of the new Gabor lens at IAP.
Table 2 compares field levels of
conventional lenses with the GL in case of Bi1+ and
He1+ beams. The thin lens approximation was used (Wangler, 1998). In each case, the lens system has a
total axial length of 0.3 m. It should be noticed that the GL enables
the transport of space charge compensated beams: As mentioned above,
the beam axis is kept on zero potential, while off axis, a
decompensation at the lens end regions is strongly hindered by the
action of the magnetic fringing fields. In contrast to GLs and to the
magnetic lenses, the space charge compensation by electrons is heavily
distorted in the case of electric quadrupole focusing. The comparison
made in Table 2 clearly demonstrates the
efficiency of Gabor lenses—even at an assumed filling factor of
0.32.
Comparison of different lens systems with identical geometric and
focal lengths applied on 2.5-AkeV Bi1+ and 110-AkeV
He1+ beams.
4.3. IH-type structure for beam acceleration behind
the RFQ
The capabilities of H-type structures were described earlier (Ratzinger & Tiede, 1998). At beam energies up
to about 5 AMeV, r.t. IH-type cavities will be preferable to any s.c.
alternatives, because of their excellent shunt impedance and because of
the expected higher rate of particle losses within that energy
range.
The KONUS beam dynamics for 20-MHz IH-type cavities at injection
energies of 60 AkeV and at 200 AkeV have now been studied in more
detail (see Ratzinger & Tiede, 2002). It
is clearly recognized that the needed axial length of the transversely
focusing elements sets an upper limit for the beam current to be
accelerated at acceptable emittance growth rates.
The development of powerful focusing elements will become a key
issue. At Lawrence Livermore National Laboratory quite attractive s.c.
quadrupoles are being developed at present within the induction linac
approach (Faltens et al., 2002). The
capability of Gabor lenses to focus bunched low energy beams will be
investigated experimentally (see Sect. 4.2).
5. CONCLUSIONS
Superconducting cavities for low- and medium-energy beams are under
development in several laboratories. Pulsed operation of
superconducting structures has been developed successfully for β =
1 structures. The concepts and principles can be applied to the pulsed
operation of low- and medium-β structures. As a consequence, an RF
driver linac scenario without funnel sections and operated at a higher
duty factor becomes attractive: It is in combination with accumulator
ring schemes filling the longitudinal phase space between linac bunches
and including multiturn injection.
The GSI High Current Injector HSI has been operated successfully at
the nominal field levels with A/q values ranging up to 60. The
current limit of 0.25 A/q in units of milliamperes was demonstrated
successfully for Ar1+ beams. In the case of U4+,
further improvements including the ion source, LEBT, the whole linac,
and down to the synchrotron injection are needed. The 36-MHz HSI has
successfully demonstrated that averaged effective voltage gains of up
to 8 MV/m along drift tube sections can be reached at RF
frequencies as low as 36 MHz. This is again an indication that there is
a rather weak dependence of maximum RF field levels from the RF
frequency below 300 MHz. This is of great importance for an attractive
HIF RF driver design. In a next step, the HIF specific linac
development has to be focused on the design of powerful transversely
focusing elements along the full energy range. IAP will investigate the
Gabor lens capabilities in focusing intense low-energy d.c. and bunched
beams.
ACKNOWLEDGMENT
This work was supported by BMBI (contract number: 06F134I and
EU-IFMIF).
