A Radio Frequency Quadrupole (RFQ) for acceleration
of a 10 mA lead 18+ ion beam from 6.9 keV/u to 100 keV/u has been
designed, built and tested in the framework of the CERN Laser
Ion Source (LIS) study. The challenge of the RFQ design was to
deal with a lead ion beam that includes about 10 charge states
with an overall current of some 100 mA. A new RFQ design, intermediate
between the two standard high-intensity and low-intensity designs,
has been applied in order to have a compact structure giving small
longitudinal emittance and high transmission.
The transport and matching line from the source to
the RFQ is made of two solenoids. The unwanted charge states are
not filtered and will enter the RFQ mis-matched. In order to test
the RFQ performance proper it was decided to operate it with an
equivalent mono-species proton beam during the first stage of
the commissioning.
The design criteria for this intermediate current
RFQ, the problems involved in dealing with a mixture of different
charge states, as well as the results of the first test with an
equivalent proton beam are presented in this paper.
The general criteria for designing the RFQ were a
high transmission (>90%) and good output beam quality: in particular
the r.m.s. longitudinal emittance should not exceed the value
of 18 deg keV/u at 100 MHz. Other limitations were given by the
length (fixed to 250 cm) and the output energy, which had to be
in the range of 100 keV/u to match the downstream analysing magnet.
The beam coming from the laser ion source is composed
of a mixture of about ten charge states [1] of varying energies.
The overall current was estimated to be 100 mA, of which 10 mA
represent the design particle. The unwanted charge states are
not filtered out prior to entering the RFQ and will contribute
considerably to the degradation of the beam quality, even considering
that the beam will eventually be partially filtered by the RFQ
itself. The challenge, then, was to design a machine which could
stand a rather high beam current and preserve the beam quality
but without reaching prohibitive lengths. For RFQ design, two
standard "recipes" are normally used: the "high
current" design [2] and the "low current" design
[3]. However, neither of these approaches could be successfully
applied to the current problem: the first would result in an excessively
long structure and the latter would give poor beam quality and
low transmission rates. The solution which proved to be successful
was to treat the beam as low-current when it is continuous and
as high-current as soon as the longitudinal current density starts
to increase. This strategy gave the rough design for the intermediate
current RFQ; subsequent modifications were made on the low energy
section to meet the requirement of an output r.m.s. longitudinal
emittance inferior to 18 deg keV/u at 100 MHz. Three techniques
have been tried in our case for limiting the longitudinal emittance
from the start: 1) to have a very long shaper; 2) to prebunch
the beam with an independent RF cavity before injection in the
RFQ; and 3) to keep the synchronous phase at -90 deg. while increasing
the accelerating component for about 20 cm at the low energy end.
This last method is equivalent to having a series of about 30
bunchers with a voltage adiabatically increasing. This was also
the selected approach as it provided a resulting longitudinal
emittance remarkably small with respect to the others. In Table
1 the RFQ characteristics are summarised and in Fig.1 the main
RFQ parameters are represented as a function of length. In Fig.
2 the evolution of the beam emittance for a nominal matched beam
is reported.
Table 1 - RFQ Characteristics
| Design Ion : Charge 18+, Mass 208 a.m.u |
| Frequency : 101.28 MHz |
| Vane voltage : 71 kV (1.8 Kilp) |
| Power losses (at 90 k-m) : 70 kW |
| Maximum electric field : 27 MV/m |
| Vane Length : 253.18 cm |
| Number of cells : 287 |
| Minimum bore radius : 0.2 cm |
| Modulation factor (max) : 2.1 |
| Transmission : 94 % |
| Current : 10 mA |
| Input energy : 6.9 keV/u (extrac. volt. 80 kV) |
| Output energy : 100 keV/u |
| Input acceptance : 300 mm mrad (tot,unnorm) |
| Transv. emittance growth : 0 |
| Output long. emittance : 12 deg keV/u |
Owing to the 4-rod structure the RF field pattern
is different in the horizontal and vertical plane [4]. Longitudinally
the beam undergoes a voltage difference on axis equal to half
the operating voltage. The input region has been modelled with
MAFIA and the beam trajectories have been integrated in the MAFIA
calculated field in order to estimate the effect on the beam dynamics.
The effect on the transverse plane (asymmetry between the x and
y focusing field component in the RMS) turns out to be negligible
in our case, and the 1-2% energy spread resulting from the non-zero
longitudinal field does not have any effect on the final beam
quality.
Due to the complexity and the uncertainties of the
beam parameters at the output of the LIS, it was decided to test
the performance of the RFQ itself with an equivalent proton beam.
The equivalent proton beam has an energy equal to the energy per
nucleon of the lead ion beam and a current scaled like the ratio
of charge to mass (11.56 in our case). The RFQ is run at a voltage
equally scaled like the ratio charge to mass.
The source used for the equivalent proton beam test
is a duoplasmatron source designed to produce a 90 keV 250 mA
proton beam [5]. As the beam required for the test is a 6.9 keV
20 mA proton beam the source was operated in a regime very far
from that of standard operations, giving rise to some instabilities
as well as a reduced proton yield. More specifically in normal
operation protons represent 70% of the total current, composed
also of 
and 
; in the
low energy regime the protons are only 45%. Operating the source
with Helium 1+ provided a cleaner beam with only few percent of
unwanted particles.
Before testing the performance of the RFQ in the experimental set-up as in Fig. 3, various beam parameters were measured at the position corresponding to the RFQ input plane and the solenoid settings giving the match to the RFQ were found.
As there was margin to increase the RFQ power from
the nominal level, several tests in different equivalent "rescaled"
conditions were performed. These provided insight to the RFQ dynamics
in areas that could not be explored before. The configurations
tested are reported in Table 2; the values underlined are the
"nominal" values for that configuration (i.e. the value
that corresponds to the design operation for lead 18+).
Table 2 - Configurations tested
| charge/mass |
| |||
| proton | 1/1 | 7,14,21 | 6.2 to 19 | |
 | 1/2 | 7,14,21 | 6.2 to 19, 12.4 | |
 | 1/3 | 7,14,21 | 6.2 to 19 | |
| Helium | 1/4 | 28 | 25 | |
What is expected is that each configuration when operated with the nominal settings gives the same output beam.
The measuring line shown in Fig. 3 - two quads, a
beam transformer, an emittance measurement device and a spectrometer
- allows the measurement of the RFQ transmission, of the output
transverse emittances and of the output energy spread. The measured
performances of the RFQ for the nominal settings are summarised
and compared to the calculated ones in Table 3, Fig. 4 and Fig.
5.
Table 3 - RFQ performance, nominal setting
The difference between the calculated and measured transmission can be explained by a non-perfect matching due to the solenoid field limit.
After completion of the measurements for the equivalent
beam other measurements were performed. The goal here was to study
and verify the calculated performance when the RFQ is operated
in other-than-design conditions. In particular we will report
here the most interesting results:
and 
)
were not accelerated through the RFQ. However, when the extraction
voltage was set to the nominal values for 
or 
, protons were always transmitted with
a rate depending on the rf level. This result shows that, as calculated,
the energy acceptance of the RFQ is quite large due to the very
adiabatic design in the first part of the machine. In fact, particles
which enter the RFQ outside the longitudinal stability region
can be reabsorbed in the stable bucket after the bunching process
is completed.
An RFQ for acceleration of a 300pi mm mrad, 18+ lead
ion beam from 6.9 keV/u to 100 keV/u with high transmission and
excellent output beam quality has been designed and constructed
at CERN in the framework of the LIS study. The first beam test
with an equivalent proton beam confirmed the theoretical prediction
of transmission and output beam quality.
We feel very indebted to W. Pirkl for his support
and guidance and to F. Neddam for the many beam measurements he
has performed. We also thank H. Charmot, C.E. Hill and J.P. Romero
for their interest and support.
[1] M. Bourgeois, G. Hall, H. Haseroth, S. Kondrashev, H. Kugler, K. Langbein, A. Lombardi, K. Makarov, W. Pirkl, Y. Satov, R.M. Scrivens, B. Sharkov, A. Shumshurov, A. Ster, A.J. Tambini, E. Tanke, "High charge-state ion beam production from a laser ion source", these proceedings
[2] K. R. Crandall, R. H. Stokes, T.P. Wangler, " RF quadrupole beam dynamics design studies" , Proc. 1979 Linac Conf.
[3] S. Yamada, " Buncher section optimization of Heavy Ion RFQ Linacs", Proc. 1981 Linac Conf.
[4] P. Bourquin, W. Pirkl , H-H Umstätter, "RF and constructional issues in the CERN LIS RFQ", these proceedings
[5] C. E. Hill, "Ion and electron sources",CERN/PS
94-36 (HI)