IEEE Transactions on Nuclear Science, Vol. NS-28, No. 2, April 1981ION ACCELERATORS AS DRIVERS FOR INERTIAL CONFINEMENT FUSION

Andris Faltens, Denis Keefe, and Stephen S. RosenblumLawrence Berkeley LaboratoryBerkeley, Calif. 94720 U.S.A.

INTRODUCTION

During the past few years the possibility ofusing intense ion beams to ignite a pellet of fusionfuel has looked increasingly promising. Ion beamsranging in mass from protons up to uranium have beeninvestigated and several machines have been built atdifferent laboratories to investigate the requiredtechnology. Light ion drivers are based on the useof high current, high voltage diodes arranged arounda central target. These devices have the necessarypower and energy to initiate fusion burn but sufferfrom the inability to transport stably the necessaryhuge beam currents over long distances to a smalltarget. Heavy ion drivers are based either on theradio-frequency linac or the induction linac.Because heavy ions have a much shorter range thanlight ions of the same energy, one is able to raisethe beam voltage by a factor of one-thousand andlower the current correspondingly. The expectedparameters for a fusion driver will be delineated andthe present state of development of the technologyfor the candidate ion beam drivers will be describedin light of these desiderata.

ENERGY PRODUCTION VIA INERTIAL CONFINEMENT FUSION

For inertial confinement fusion (ICF) to lead tonet energy production one must as a minimum:

1. Compress the D-T Fuel from its initialdensity to about 103 times liquid density.

2. Heat the fuel to T > 20 keV. If one were toachieve complete burn of 1 mg of D-T fuel thiswould yield 350 MJ, the equivalent of 1/10 ton ofTNT.To carry out the compression one surrounds the

fuel by concentric shells of material chosen suchthat the driver energy is absorbed in the outer shelland the ablation of this shell drives the fuelinward. Many schemes can be examined with multipleshell targets to "push" and "tamp" the fuel.Computer calculation of the hydrodynamics of thesetargets produce gain curves for ion drivers. Such aset of curves calculated by Bangerter1l2 is shownin Fig. 1. To achieve a gain of 100 it would seemwise to arrange for an input energy of 2 - 10 MJ.

Fig. 1 - Gain curve for ion driver inertial confinementfusion calculated by Bangerter

Of the drivers to be considered we will onlyconcern ourselves with those using ion beams. Theseare divided into two classes - those using lightions, based on neutralized beam transport and thosebased on heavy ions using conventional acceleratortechnology and charged particle transport.

To examine the relative merits of light and heavyions one must examine the range - energy curve inview of the requirement that in order to generate theimplosion one must deposit it least 20 MJ/gm in athin layer on the surface of a pellet of radius about1 mm. Thus,

W = 4 > 20 gJ2'rtR gm

with Q = total beam energy = I V in MJI = beam chargeV = beam voltage = kinetic energy / charge stateof ionrt = pellet radius in cm.R = particle range in g/cm2

Using this relation and range curves calculated byBangerter for particles in dense Pb plasma we cancalculate the curves shown in fig. 2. The four heavycurves show the results of this calculation forprotons, carbon, cesium and uranium ion beams. Abovethe curves one has a sufficiently short range thatthe target will tend to implode rather than explode.In all cases the total beam energy must be greaterthan 1 MJ in order to get a significant pellet gain.It is clear that light ions must push to very highcurrents (tens of MA at a few MV) while heavy ionscan do the job at several GV and tens of kA. Alsoindicated on the curve is the location in thisparameter space of the present major ion driverprograms. These will be described in the followingsections beginning with the driver requirements shownin Table 1.

Table 1: Driver Reouirements for Power Production

Energy - 1 to 10 MJ

Power - 100 to 600 TW

Pulse shape - Control needed

Efficiency x gain > 10

Focusing - to a few mm at 5 to 10 m

Reliability - > 80 percent on-time

Lifetime - 30 years

Rep rate - 1 to 10/sec

Cost - (a few) x 108 $/GW of electricaloutput

The energy requirement has been discussed already.The power and pulse shaping arise in order to depositthis energy in a time short compared to the time forthe hydrodynamic disassembly of the pellet.

0018-9499/81/0400-1335$00.75© 1981 IEEE1335

1000

Ccm

<a 1000)PI

10'-1 10

Input energy - MJ

A minimum efficiency-times-gain product requirementarises from trying to keep the recirculating power inthe plant small in order to make the cost of powereconomically reasonable. This can be readily seen fromthe following argument. To produce a beam energy, Q, adriver of efficiency, n, requires Q/n input energy. Ifthe electric power generators are taken to be 33percent efficient, and the pellet gain is G, then theoutput energy is QG/3. Requiring the recirculatingpower to be small implies that nG > 3 eg. nG - 10. Ifwe are limited to G = 100 this requires that n > 10percent. For laser drivers, with n<10 percent, theacceptable G must be higher whereas, becauseaccelerators have a higher n, G can be lower. The reprate arises from the requirement that the plant produce

1I GWe from a pellet burn in the reactor which wouldnot damage the structure in an unreasonably short time(less than several years).

We will now examine the different drivers and thenindicate their relative promise in meeting thosecriteria.

HEAVY ION DRIVERS

1. RF Linac

The RF linac is the driver most familiar to theaccelerator community. The technology needs nodescription here except to point out the driverscenario and the areas of uncertainty. The program isbeing carried out principally at Argonne NationalLaboratory (ANL)3 with strong contributions from

06

-o

Ewo4 00I L..

0._

< a)W 73

CDIU-0)0L

10I

100

10,-

10-2

Brookhaven National Laboratory (Meqalac) and Los AlamosScientific Laboratory (rf quadrupoles).

The basic problem is to develop a high current beamfrom a low current rf linac by transverse andlongitudinal beam manipulations. The scheme proposedfor a 1 MJ driver involves funneling current from atree of linacs and then amplifying the current by afurther large factor by multiturn stacking of the beamin both transverse phase planes in a storage ringfollowed by further bunching after single turnextraction.The main question here is whether thecomplex beam manipulations required cause sufficientdilution of the emittance that the final focussing ofthe beam onto the pellet is badly compromised.

The present program at ANL is directed towards aninvestigation of the low a part of the machine using amulti-aperture Xe source on a 1.5 MV Dynamitron highvoltage source. The beam will be injected into a 12.5MHz Wideroe Linac and then matched into a 25 MHzWideroe Linac and further accelerated. The main aimhere is to examine the experimental difficulties andemittance dilution involved in the funneling scheme. Anew feature to be examined is the longitudinalphase-space matching as the beam is transferred fromone linac to another one operating at a differentfrequency. In the future the ANL program is proposingto build a Beam Development Facility, (BDF) which willinvolve adding a storage ring at the output of the 25MHz linac to examine the stacking and extraction in thestorage ring. The parameters of this facility areindicated as BDF on Figure 2.

I MeV l0 100 I GeV 10 100

BEAM KINETIC ENERGYFigure 2 - Ion beam fusion driver road map

1336

2.Induction Linac

The induction linac is probably less familiarhaving only been used up to now to accelerate highcurrent (1-100 kA) electron beams. The acceleratingmodules are non-resonant, consisting of large toroidsof ferromagnetic material driven by pulsed modulators.A schematic representation of such a machine is shownin Fig 3. The modules can be viewed as single turntransformers in which the drive line is the primary andthe particle beam is the secondary. This driverprogram is being pursued principally at LawrenceBerkeley Laboratory (LBL)4. In this system, in a 1MJdriver the entire bunch of charge is accelerated in asingle pass from the source to the end of theaccelerator through a large beam pipe. The stacking inboth transverse phase planes is accomplished right atthe source by using a very large emitting area. Next,by ramping the voltage pulses in a number of theearlier modules, a compression process is begun suchthat the pulse-duration becomes shorter and the currentis amplified continually along the length of theaccelerator. To be reasonably efficient, the inductionlinac operates just below the maximum current that canbe stably transported, so the design requires a preciseknowledge of the space-charge limit.

.1mr..:: i : :I... 1' -F.- .6 :F...L

XBL 8010-7351

Fig. 3 - Schematic representation of an induction linac

Being a pulsed power device the induction linac isinherently adapted to high current, short pulseoperation. Although the manipulations are fewer andthe residence times in the accelerator are less than inthe rf linac storage ring approach, there are,likewise, questions to be answered of whether the spacecharge limited beam transport will lead to possiblyunforeseen emittance growth or instability. Thepresent LBL program is involved with the low beta endof such an induction linac and is centered around a 2MV, 1A Cs injector. The next stage of the programenvisions the construction of a modest Cs ion inductionlinac (10 MV, 5A) to examine the questions of transportstability and emittance growth in such a structure and,above all, demonstration of current amplificationduring acceleration in a satisfactory way. This deviceis indicated as LBL Test Bed (proposed) on Fig 2.

LIGHT ION DRIVERS

Light ion drivers differ from heavy ion driversprincipally in that they make use of schemes toneutralize the particle beams. The main effort in thisarea is at Sandia Laboratories in Albuquerque withsupporting research principally at the Naval ResearchLaboratory and Cornell University. A comphrehensivereview of this entire field has recently been given byHumphries.5

1. PBFA

PBFA I (Particle Beam Fusion Accelerator) ispresently in operation at Sandia LabortoriesAlburquerque (SAN)6. It consists of 36 Marxgenerators and pulse forming networks each pulsed to 2MV and delivering 400 kA, 40 ns pulses. This systemdelivers its energy to a cylindrical diode array at thecenter, the total delivered energy to a target being1MJ.

The diode makes use of a magnetic fieldperpendicular to the beam direction produced byexternal coils to prevent electrons from crossing thediode gap. Otherwise, most of the current would appearin the electron beam. A schematic diagram showing thisprinciple is shown in Fig. 4. The cathode is designedto be fairly transparent to the ions.

<- d -- 1

Fig. 4 - Schematic representation of a magneticallyinsulated ion diode (courtesy of S.Humphries, Jr.)

Because these beams cannot be contained byachievable magnetic or electric fields one mustneutralize them via transport in strongly ionizedplasma channels using a vaporized wire or a laserpulse. The main difficulty here is to hit reliably andrepetitively a mm size target in a 10m diameter vesselwith these beams. Targets of several cm diameter arepresently being used. At present PBFA II is beingplanned - it will have 72 beams of 4 MV each, thusraising the total beam energy to 4 MJ. These devicesare indicated in Fig. 2 as PBFA I and PBFA II.

2. Pulselac

Pulselac, a device also presently being developedat Sandia Laboratories, Alburquerque,7 couplesneutralized beam transport with inductionacceleration. The aim of this device is to bring aboutthe ion acceleration in several stages rather than in asingle stage as in PBFA. This might lead to lessstringent demands on the switches and pulse formingnetworks. The transport is brought about via thecollective focussing effect of injected cold electrons.Magnetic insulation must be used at the acceleratinggaps to prevent the electrons from shorting out theaccelerating field. A schematic diagram of a Pulselacgap is shown in Fig. 5. The inner and outer coils aredriven such that the main component of the B field isin the radial direction. The beam is in the annularspace between the inner and outer coils. Theaccelerating potential is applied across the gaps inthe tubes by an induction module . At present PulselacB is operating at a final beam voltage of 600 keV ofcarbon at 3 kA. The carbon beam is produced by aplasma gun which is also magnetically insulated fromthe accelerator section.

This is a promising technology in that it canaccelerate somewhat heavier ions than PBFA in a moregradual way such that the demands on the switches andpulse forming network are much less stringent. Atpresent, however, the beams are not of sufficiently lowemittance that they could hit a 1 mm pellet in areactor chamber, and repetition rate capability has not

1337

been demonstrated. On Fig. 2 is shown the relevantpoint for the present Pulselac B as well as the futurePulselac C which will have a 4 MeV, 5kA beam of carbonin a 50 ns pulse.

Accelerating gap-

PULSELAC ACCELERATING MODULE

SchematicXBL 8010-7360

Fig. 5 - Schematic representation of a Pulselac gap

Parameter

Energy

Power

Pulse Shape

Efficiency - GainProduct

Focussing ontopellet

Emittance Growth

Rep. rate

Lifetime

Reliability

Legend { 1

E:

Driver

R F InductionPBFA Pulselac Linac Linact

0* ** * _=(2) (

I N. A.

_= 00_= 00

_ 0011.~/~

I

SUMMARY

A representation of the state of ion drivertechnology is shown in Table II. As in any suchsuperficial treatment there are some subjectiveconclusions but the clear generalization is that thelight ion machines have the necessary power and energybut lack rep rate and reliability. The heavy iondrivers, on the other hand, are based on developedaccelerator technology which is demonstrated to be ofhigh rep rate and reliability, but is still not at therequired energy or power levels.

ACKNOWLEDGMENT

This work was supported by the Division of InertialConfinement Fusion of the Department of Energy underContract W7405-ENG-48.

References

1. A. Faltens, E. Hoyer, D. Keefe, L. J. Laslett, L.Smith, R. 0. Bangerter, W. B. Herrmannsfeldt,Proceedings of the 8th Int. Atomic Energy AgencyConference on Plasma Physics and Controlled NuclearFusion Research, Brussels, Belgium, July 1-10, 1980(to be published).

2. R. 0. Bangerter, in Proceeding of the Heavy IonFusion Workshop held at Brookhaven NationalLaboratory, Oct. 17-21, 1972, BNL 50769.

3. Semi-Annual Progress Report, Hearthfire Heavy IonFusion, October 1, 1979 - March 30, 1980,ANL-HIF-80-2, (unpublished).

4. Heavy Ion Fusion Half Year Report, October 1, 1979- March 31, 1980, LBL 10914, (unpublished).

5. S. Humphries, Jr., Sandia Laboratories ReportSAND 80-0402 April 1980, (unpublished).

6. Glenn W. Kuswa, in Proc. of 8th Int. Atomic EnergyAgency Conference on Plasma Physics and ControlledNuclear Fusion Research, op cit.

7. S. Humphries, Jr., in Proceedings of the Heavy IonFusion Workshop, Berkeley, Ca., Oct. 29 - Nov. 9,1979, LBL 10301.

Goal substantially achieved.Goal a likely extrapolation from

existing hardware.More "experimentation' required.

ION BEAM DRIVER COMPARISONX8L 8010-7353

TABLE II

1338