In 1995, Fermilab and SAIC formed a collaboration with partners from the University of Washington (UW) and the Biomedical Research Foundation of Northwest Louisiana (BRF) to explore an innovative approach to the production of radioisotopes. The accelerator system that is being developed accelerates 3He to 10.5 MeV and then delivers this beam to the target to produce the short lived radioisotopes of interest to the PET community (18F, 15O, 13N, 11C). Research is being conducted to investigate the contribution that this promising approach can make to clinical and research PET.
The accelerator system has several very interesting
aspects. These innovations include multiple RFQ accelerators configured
in series, a gas stripper jet to doubly charge the low energy
(1 MeV) 3He beam, and
an isochronous matching section to manipulate the transverse and
maintain the longitudinal profile of the beam (without the use
of an RF buncher) in the charge doubler transition section between
RFQ's. This paper updates the progress of the PET 3He
RFQ accelerator, the current status of the design, and some of
the interesting ongoing research.
The idea of using 3He for the production of radioisotopes for PET is not new. Development work on this concept was conducted by SAIC and the University of Washington in the early 1990's.1 When the original program was being formulated, the PET environment in which it could make a contribution was significantly different than it is today. The original development was based on the belief that 18F labeled compounds would be favored by the PET community. Also important was a global shortage of 18O which made the standard approach of producing 18F using H218O expensive and potentially unpredictable.
Since that time things within the PET community have changed significantly. There is no longer a significant shortage of H218O. Also, FDA policy has changed regarding regulation of PET radiopharmaceuticals (Federal Register, February 27, 1995). The new policy no longer gives an advantage to 18F labeled compounds. This means that 11C agents are now no more trouble than 18F. Carbon opens up a much larger array of molecules to label. Furthermore, recent developments in radiochemistry for preparing the important precursor, 11CH3I, avoid the use of liquid solvents and LiAlH4 which are very air and moisture sensitive, and make the precursor directly in the gas phase and at substantially higher specific activity. This leads to a smaller yield (mCi) requirement of 11CO2. But, to take full advantage of this new technology, higher specific activity is necessary (i.e. new PET machines must be good producers of 11C as well as 18F). In these several ways the environment in which 3He RFQ technology can make a significant and meaningful contribution to the advance of PET has changed.
While there have been interesting developments of several new low energy accelerators over the last 2-3 years - the Cyclone 3D (IBA), the TR13 (EBCO), PETtrace (GE), the tandem cascades from SRI or the new deep valley machines (CTI Siemens) - all of these machines use essentially the same nuclear reactions and target chemistry. The RFQ using 3He, on the other hand, is a different approach and thus holds significant potential and research opportunities for advancing the state of the art in PET isotope generation.
Before the radiochemistry and targetry for 3He
could be investigated, an accelerator was needed that would supply
a beam with the desired characteristics and parameters. The accelerator
that had been developed by SAIC and the University of Washington
in the early 1990's was a good starting point but needed to be
upgraded to provide a more powerful tool for researching 3He
in light of current information. Analysis and a series of discussions
resulted in the baselining of new operating parameters as indicated
in Table 1.
Table 1 Accelerator Design Parameters
| Energy (MeV) | Ie (mA)
average | Rep. Rate (Hz) | PW (msec) | |
| Existing Sys | 8 | 300 | 360 | 55 |
| New Sys. | 10.5 | 200 | 360 | 70 |
Since the radiochemistry and targetry associated
with pulsed high intensity 3He
beams was to a large extent unknown, it was decided that the system
being developed needed to follow a conservative approach, i.e.
it needed to be flexible and powerful enough to accommodate a
wide range of targetry options. In particular it needed to be
able to produce large quantities of 11C
since this isotope is likely to be used increasingly in PET. Table
2 indicates the beam required as a function of the final energy
for the target quantities of the various PET isotopes.
Table 2: 3He Current Required for PET RFQ
| 18F | 600 | 333 | 207 | 177 |
| 11C (low SA) | 1000 | 202 | 140 | 125 |
| 11C (high SA) | 440 | 298 | 195 | 163 |
| 13N | 100 | 266 | 168 | 104 |
| 15O (low SA) | 800 | 517 | 360 | 318 |
| 15O (high SA) | 200 | 899 | 559 | 460 |
* 1 MeV energy loss in target window
** 1.5 MeV energy loss in target window
The requirements of Table 2 led to the accelerator
current and energy requirements stated in Table 1. With this baseline
information from our radiochemistry collaborators at UW and BRF,
the existing 8 MeV 3He
accelerator was redesigned to meet the new requirements. The results
of this redesign are shown in Figure 1. This layout makes the
most efficient use of the existing equipment while solving some
of the more challenging technical problems. Some interesting aspects
of this accelerator system are:
Since one can achieve a much more efficient acceleration (length and power) with a doubly charged beam, a very attractive approach would be to make use of a doubly charged ion source. Unfortunately, nature works against this goal. With the second electron being bound with an energy of about 54.4 eV, common ion sources do not produce sufficient quantities of the doubly charged ion (15 mA required). As an alternative, the singly charged beam can be accelerated to an energy where it can be efficiently stripped (1 MeV). It is this approach that has been taken.
At an energy of 1 MeV and a current of 400 mAavg (20 mApeak), carbon foil strippers could not survive the high power density. Both gas cells and gas jet strippers have been investigated. A jet stripper has been developed and tested with very promising results.
The most difficult aspect of this accelerator system is the matching element between the prestripper and the post stripper RFQ's. This transition section needs to accomplish several things. It must provide sufficient space to accommodate the gas stripper (gas containment) while maintaining the longitudinal bunching of the beam and transversely matching the beam into the second RFQ. To overcome experimental realities, tunable components are desired. The longitudinal phase space of the beam must be maintained in order to eliminate the buncher/shaper section from the second RFQ (which at this beam energy would add about 1.5 m to the length of the second RFQ). Previous attempts to utilize an RF buncher to contain the beam longitudinally had been unsuccessful due to the very tight space constraints and the large number of free electrons (due to the proximity of the charge doubler). Based on this, it was decided to build an isochronous beam transport system that maintains the longitudinal and manipulates the transverse phase space of the beam.
The accelerator that had been developed under the earlier program had been designed for a final energy of 8 MeV. In order to achieve the higher energy requirements of the new system, it was decided that the most direct approach would be to add a third RFQ (manufactured by SAIC) to the high energy section to go from 8 MeV to the final energy of 10.5 MeV. This resulted in three RFQ's operating in series. The RFQ cavities are not resonantly coupled. Each cavity must be synchronized to and resonant at the same frequency. To accomplish this the resonant frequency of each cavity is controlled through adjustment of the temperature of the cavity cooling water. No mechanical tuners are used. Tests on this tuning system at full (2.5%) duty factor have been successful.
The development of this system has taken place in
two phases. A 1 MeV test stand was assembled from the accelerator
components of the old system. Using this test stand, a number
of the more difficult aspects of the system were addressed. Among
the things that were studied are: low energy and medium energy
beam characterization, ion source operation with He, charge doubler
stripping efficiency, and charge doubler gas containment. The
results of these tests have been incorporated into the design
of the new components. Some of the information gained in the 1
MeV tests are summarized below.
He+ ions
are obtained from a fairly standard duoplasmatron ion source.
The source operates at 360 Hz with a pulse length of 70 ms.
It requires a gas consumption of 2-4 std cm3/min
(~ 1 liter / day of operation). Since 3He
is relatively expensive all attempts are made to minimize loss
by reducing the source button (aperture) and pressure. Also the
source is started and operated on 4He
except when 3He is necessary.
Because of the heavier ion and high duty factor, filament shielding
is critical to prevent overheating and fast erosion. The filament
is enclosed in a cylindrical shield with a sufficient opening
to extract electrons while minimizing back-streaming ions. Several
weeks of reliable and stable source operation have been obtained.
A 25 mA beam is extracted at 20 kV from an ~ 1 cm plasma cup through
a 0.8 cm grounded extraction electrode with an electron suppression
electrode. Slightly after extraction the ~90% normalized beam
emittance was measured to be 0.5 - 0.7 p
mm mrad. One magnetic solenoid is used to focus the 20 keV beam
into the RFQ. At the entrance of the RFQ, 0.7 m beyond the source,
75% of the beam is within ~0.5 p
mm mrad emittance (normalized)2.
After the Prestripper RFQ, at 1 MeV, the rms emittance
has been measured to be 0.2 mm mrad (or ~34 p
mm mrad unnormalized for 90% of the beam)2.
This was measured with 5.5 - 7 mA at 1 MeV from the RFQ. Better
matching and understanding of the RFQ transmission is needed.
A maximum beam of 11 - 13 mA has been observed from the RFQ and
appears to have similar characteristics. This was achieved with
a larger solenoid in the 20 keV transport line.
A prototype stripper cell based on a pulsed gas jet was built to determine efficiency and gas flow in a realistic geometry. A mechanical injector (Nissan fuel injector) provided gas pulses to a converging-diverging nozzle. A directed gas jet of line density approximately 3-6x1016 cm-2 was created at the nozzle. It passed across the beam and was directed into a vacuum pump. The flow rate of the gas jet was sufficient to prevent excessive heating of the gas by the beam, and the injected gas was pumped out between beam pulses.
A magnetic spectrometer that bends the 1-MeV He+
and He++ beam ions into Faraday collectors at
bend angles of 11.5_
and 23.6_
respectively, was used to test operation of the gas jet. Stripping
efficiency was determined by measuring the relative distribution
of beam current on the two collectors. Stripping efficiency for
several gases is shown in Figure 2 as a function of back pressure
on the injector.
The best performance was obtained with argon gas, which reached 80% stripping efficiency at a pressure of 25 psia. Pressure measured at the RFQ was 2.8x10-6 Torr for this operating point, at a repetition rate of 60 Hz. We expect to be able to operate at no less than 70% stripping efficiency at the design rate of 360 Hz by increasing pumping capacity. An operational version of the stripper cell is now in fabrication.
As part of the design effort for the new MEBT (Figure 3), a number of options were investigated. It was recognized that folding the machine would accomplish the goal of keeping the length of the system manageable and could also do the longitudinal dynamics. Designs were investigated that included 180 degree bends, two 90 degree bends, three 60 degree bends, four 45 degree bends, and 30-60-60-30 degree combinations. In each case where multiple dipole designs were tried, quadrupoles were placed between the dipoles and varied in strength and position. Various internal gradients and edge angles were also tried. While it is possible to make a 180 degree bend which has the proper transition energy, it has not yet been possible to have a 180 degree bend design which is isochronous. For this and other reasons it was decided to use 2 x 270 degree bending MEBT which could be made isochronous.
The beam optics of the MEBT are shown in figure 4.
The major magnets for this transport system have been fabricated
and are being tested. The installation and commissioning of the
transport system is scheduled to take place over the next two
months.
The modifications to the accelerator system are scheduled to be completed and tested in late 1996. Once completed, the accelerator will be run at Fermilab for 6 to 8 weeks in order to test shielding and do some initial targetry development. Following this run, the machine will be disassembled and shipped to the Biomedical Research Foundation. The accelerator system has been built in a modular fashion in order to facilitate moving. We anticipate that the move and commissioning will take about 8 weeks, after which the in-depth targetry and radiochemistry research will begin.
1. W. Hagan, et.al., "A Helium-3 RFQ Accelerator for PET Tracer Production", Proceedings of the Ivth International Workshop on Targetry and Target Chemistry, 1991, Pp 19
2. "Elliott McCrory, et.al. "Emittance
Measurement Techniques Used in the 1 MeV RFQ for the PET Isotope
Project at Fermilab". This Conference.
*Operated by the Universities Research Association
under contract with the U. S. Department of Energy.