IEEE Transactions on Nuclear Science, Vol. NS-28, No. 2, April 1981
INTENSE NEUTRON SOURCE DEVELOPMENT FOR USE IN CANCER THERAPY*
F.M. Bacon, R.J. Walko, D.F. Cowgill and A.A. RiedelSandia National Laboratories, Albuquerque, NM 87185
Abstract
A neutron production rate of 1.1 x 1011/s from theD(d,n)3He reaction has been measured from a 200 kV, 200mA d.c. deuterium ion accelerator during the initialbombardment of a scandium deuteride target. Over aperiod of 40 h operation, the output decreased to 8 x1O01/s due to impurities in the ion beam being implantedin the target and displacing deuterium just below thetarget surface. Deuterium concentration depth profilesin the target have been measured which show the deu-terium depletion near the surface. The initial neutronoutput scales to >1 x 10l3/s from the T(d,n)4He reac-tion for a D,T ion beam on a ScDT target, which is theoutput required for a clinically useful cancer therapymachine. Efforts to reduce the ion beam impurity con-tent to acceptable levels will be evaluated in a newaccelerator tube that is described.
Introducti on
A program is underway at Sandia NationalLaboratories, Albuquerque, to develop a 14 MeV neutronsource with sufficient intensity, lifetime, and reli-ability for use in cancer therapy. The neutron sourceunder development is based on the T(d,n)4He reactionand utilizes the target that was developed in anearlier program.1
The principal requirements of the neutron sourceare a neutron output2 of 1-2 x 10l3/s, operationallifetime of >100 hours and a neutron source area of.10 cm2. Thick target neutron yield calculations3indicate the accelerator requirements are a 200 kVacceleration potential and a 200 mA beam of 50% deu-terium and 50% tritium ions if each species is 50%atomic ions and 50% diatomic ions with a neutron pro-ducing target of ScDT or ErDT. With thi's beam powerand the neutron source area requirement, the averageion beam power density dissipation at the target is 40MW/M2 if the target plane is perpendicular to the ionbeam axis. To maintain the ScDT or ErDT target in thedihydrid phase, the surface temperature must be s4500C.The main result of an earlier program' was the success-ful design of such a target consisting of a scandium orerbium film deposited on a water-cooled copper sub-strate. Problems identified during this programincluded high-voltage power supply reliability and ionbeam impurities interacting with the target to reduceneutron output. This paper will summarize the stepstaken to solve these problems and present recentexperimental results.
Experimental Apparatus
Accelerator
To conduct target experiments, a 200 kV, 200 mAdeuterium ion accelerator was constructed; deuteriumhas been used throughout the experiments to date toavoid the health hazards associated with the use oftritium.
The accelerator has been modified by adding an
experimental chamber below the system describedearlierl so that more sophisticated target experimentscould be performed. A schematic diagram of the
*This work supported by NCI Grant CA25156 and DOE.
modified system is shown in Fig. 1. In the earliersystem, the target was located 0.6 m from the ionsource exit aperture and has been replaced by thewater-cooled collimator in the new system. The targetis now located in the lower chamber at a distance of
GROUNDPLANE
I m
Fig. 1. Schematic diagram of modified intense neutronsource target test facility.
1.5 m from the ion source. To transport the beam theadditional 0.9 m to the target and maintain a beam spotsize at the target of about 35 im diameter, the beamneutralization point was moved closer to the accelera-tor gap as discussed earlier.1
Target
A schematic drawing of the target with a cut-awaysection is shown in Fig. 2. For details of its con-
struction, see Ref. 1. Thin films of scandium or
erbium are deposited on the target and loaded with deu-terium. Film thicknesses up to 50 pm have been testedand films up to 130 pm thick have been fabricated forfuture tests.
Target Analysis Apparatus
The neutron production rate from the D(d,n)3Hereaction was determined from the protons produced inthe competing D(d,p)T reaction using a gas-filled pro-portional counter (Proton Counter in Fig. 1). Ionbeam profiles at the target were deduced from target
surface radiance measurements made with an infraredtelevision camera as described earlier.1 In additionto these experimental techniques, deuterium concentra-tion depth profiles in the target have been deduced4from the energy distributions of the protons producedby the D(d,p)T reaction which were measured using thecollimated, proton detector in Fig. 1 and a multi-channel analyzer.
Results
Initial experiments involving the new target cham-ber showed that the ion beam profile at the target waslarger than expected. The gas pressure in the lower'chamber during acceleratpr operation was about 10-5Torr. Adding argon or hydrogen gas to the lower cham-ber reduced the ion beam diameter and increased thepeak surface temperature of the target due to ion beamheating, as measured using the infrared camera. A plotof peak surface temperature as a function'of hydrogenpressure in the lower chamber is shown in Fig. 3 forseveral operating conditions. These data indicate thatthe hydrogen pressure should be greater than 3 x 10-4Torr to provide sufficient electrons from ionization
10-5 3 X 10-5 10-4
PRESSURE, HYDROGEN TORR
3 x 10-4 10-3
Fig. 3. Plot of target peak surface temperature dueto ion beam bombardment vs pressure in lowerchamber for several accelerator operatingconditions.
of the background gas for space charge neutralizationof the ion beam. Similar data for argon backfillshowed that the argon pressure had to be greater than1 X 10-4 Torr. The difference in pressure require-ments for hydrogen and argon is presumably due to thesmaller electron producing cross sections for hydrogencompared to argon.5
Modifications to the high-voltage power supply bythe manufacturer have eliminated the problem associatedwith internal breakdowns that was previously reported.'The accelerator can now be operated reliably for hoursat a time at 200 kV, 200 mA beam power. Immediatelyafter a spark in the accelerator, the ion source isturned off for one second to allow the system torecover before the ion beam is turned back on. Thisprocedure appears to improve the accelerator reli-ability.
Other modifications to the system to improveoperational reliability included enlarging the aper-tures in the ion source to prevent secondary electrondamage to the ion source electrodes. A cooled, secon-dary electron trap at the top of the ion source, asshown in Fig. 1, satisfactorily stops the electronswithout damage to the source.
A more rigid mechanical support structure for theextractor electrode was found to eliminate ion beammoti'on during operation which was apparently caused byelectrode movement due to thermal and electrostaticstresses.
Dispenser cathodes are now being used instead ofthe dipped cathodes6 because the latter were found toflake leaving a powdery residue inside the ion sourceand on the extractor electrode which lead to high-voltage instabilities. The dispenser cathodes haveeliminated this problem; however, they do seem to havemore vari ability in their acti vati on characteristicswhen new, a slight difficulty in reactivation after anair exposure, and a tendency to sag after several hoursof operation.
Neutron rate measurements vs operation time areshown in Fig. 4 for a 50 pm thick scandium deuteride
1.2
a.0
I-
D 0.8
= 0.4z
0
z
l
10 20 30 40 50
TIME (HRS)
Fig. 4. Neutron output vs operating time for a 200 kV,200 mA deuterium ion beam on a 50 pm thickscandium deuteride target.
target film and 200 kV, 200 mA accelerator power.During the first 29 h of operation, hydrogen gas wasused in the lower chamber to aid in the space chargeneutralization of the ion beam. The neutron outputdropped from 1.1 x 1011/s to 7.4 x 1010/s in thisperiod, during which there were two air exposures ofthe target to make system modifications. Followingthis run, argon was used as the backfill gas in thelower chamber and the neutron output dropped
1903
50pm Sc TARGET200 kV, 200 mA
H22* *.4
-:- AIR EXPOSURE 0OF TARGET a
f
I
precipitously to 6.0 x 1010/s after one hour of argonbackfill operation. The backfill gas was then changedto deuterium and the neutron output recovered to8.1 x 1010/s at 40 h of total operation time.
Following these series of experiments, the turbo-pump on the lower chamber was removed; and the flangewas blanked off so that all pumping for the lower cham-ber was by the diffusion pump on the upper chamber.During accelerator operation, the pressure in the lowerchamber was sufficiently high from the ion beam gasload that no additional gas feed was required for ionbeam space charge neutralization. The neutron outputduring this three-hour period varied from 4.8 x 10 0/sto 4.9 x 1010/s (these data are not plotted in Fig. 4).
The turbopump was reinstalled on the system anddeuterium backfill gas was again used to aid in spacecharge neutralization of the ion beam. The neutronoutput did not recover from the 4.8 x 1010/s rate inthree hours of operation. The target was removed fromthe accelerator for further analysis even though thefilm had not been sputtered through to the coppersubstrate.
Deuterium concentration depth profiles on thetarget are shown in Fig. 5. Curves A and C weremeasured on a laboratory accelerator with a mass ana-lyzed D+ ion beam after the target was removed fromthe intense neutron source accelerator. Curve B was
2. 0
1.6
0._
m 1. 2
E0
0.8-0.u 0. 8
0. 4
n0.2 0.4 0.6
Depth (,mi)
Fig. 5. Deuterium concentration depth profiles intarget: (A) Initial condition, (B) After40 h operation, (C) After 50 h operation.
bombardment and explains why the neutron outputdecreased during operation.
Discussion
The measured initial neutron output of 1.1 x1011/s scales to a 14 MeV neutron rate of >1 x 1013/sfor a mixed deuterium-tritium ion beam and a ScDTtarget. The decrease in neutron output with time isdue to the depletion of deuterium near the surface ofthe target film. Earlier work indicated that thedepletion can be explained by ion beam impurities beingimplanted in the target and displacing the deuterium.7This conclusion is based on ion beam impurity measure-ments8 and the post-mortem analysis of a target forlow-Z impurities by nuclear reaction techniques on a1.4 MeV accelerator.7 The measured oxygen, carbon,and nitrogen impurities in the target accounted forthe detrapped deuterium within 10 percent.
An average deuterium-to-metal atomic ratio in thetarget film can be inferred from the neutron outputand theoretical thick target neutron yield calcula-tions. More recent calculations9 and data indicatethat the earlier predictions for scandium are about 30percent high. Based'on the later data, the deuterium-to-metal ratio in the film was about 1.6 initially anddropped to about 1.0 after 40 h operation, in closeagreement with the deuterium depth profiles shown inFig. 5.
The 50 pm-thick target film was operated at fullpower for 50 h without reaching end-of-life. Previousdata' indicate that the film should last for 75 h.Thicker films have been fabricated for future tests.It now appears that an operational lifetime on thetarget of at least 100 h will be achieved.
Future Work
To reduce ion beam impurities, a new acceleratorhas been built using ultra-high vacuum constructiontechniques. A schematic diagram of the metal andceramic ion source and accelerator is shown in Fig. 6.All joints are made by either brazing or welding andall seals are made with compressed copper gaskets. Theion source and accelerator are attached to a stainlesssteel vacuum chamber that is bakeable to 450°C, shownschematically in Fig. 7. The base pressure of thesystem is maintained with an ion pump. During opera-tion, deuterium is supplied to the ion source from auranium storage bed and is pumped by a getter pump.
Acknowl edgment
The authors are grateful to R. W. Bickes, Jr.,and J. B. O'Hagan for several helpful conversationsconcerning ion source operation; to J. M. Harris,E. P. Boespflug, and R. T. Westfall for post-mortemtarget analysis, and to J. L. Provo for his effortsin target construction.
measured dynamically during 200 kV, 200 mA operationat the 40 h operation time. Curve A was measured at apoint on the target outside the ion beam perimeter andthereby represents the condition of the target priorto intense ion beam bombardment. Curves B and C weremeasured at the center of the beam spot on the target.The results show that the deuterium was depleted nearthe surface of the target during intense ion beam
1904
ION SOURCE
ACCELERATOR N
I - HV INSULATOR
Fig. 6. Schematic diagram of new ion source andaccelerator for ultra-high vacuum neutrontube.
Fig. 7. Schematic drawing of intense neutron source.
3. L.A. Shope, "Theoretical Thick Target Yields forthe D-D, D-T, and T-D Reactions Using the MetalOccluders Ti and Er and Energies up to 300 keV,"Sandia Labs. Report, SC-TM-66-247 (1966).
4. D.F. Cowgill, Nucl. Instrum. and Methods 145, 507,(1977).
5. E.W. McDaniel, Collision Phenomena in Ionized Gasespp. 282-283 (John Wiley, New York, 1964).
