Heavy Ion Inertial Fusion STATUS AND PERSPECTIVES R. Bock GSI Darmstadt, Germany Associate Member of EPS Research during the past decade has To evaluate heavy ion ICF, three aspects strongly increased our confidence that have to be investigated : the pellet, the the heavy ion accelerator is the superior driver and the concept for a power plant. driver candidate for economic power ge­ Implosion of the pellet, as one of the most neration by inertial confinement fusion. critical issues for ICF, has been and will The European Community nonetheless remain over the next years a domain of concentrates nearly all its fusion re­ laser experiments. An upgrade of the search spending on thermonuclear fusion NOVA laser facility has been recently using magnetic confinement techniques. recommended. In summarising on-going Results achieved with high intensity laser investigations in the field of heavy ion beams and systematic theoretical studies driven inertial fusion and the status of have provided the necessary reference accelerator facilities we shall concentrate data for implosion of a fuel pellet, the on the two other areas, namely systems most crucial issue in inertial confinement studies and the driver. fusion. With respect to the conditions for ignition and continuous burning, experi­ Reactor ments with laser beams are about as close The HIBALL system study [1] started in to break-even as those using magnetic 1980 and upgraded in 1985 demonstrated confinement. For example, a 20 kJ beam for the first time the feasibility of a com­ at the NOVA facility at Livermore in the Fig. 1 — Conservative and optimistic pre­ mercial heavy ion driven power plant. It USA has obtained 150 times liquid density dictions of the pellet gain as a function of also identified key issues in a systematic for the deuterium (D)-tritium (T) fuel mix, the deposited energy obtained at Lawrence way. The study envisaged injecting fuel while Japan's GEKKO XII laser facility has Livermore National Laboratory (LLNL), USA, compared with calculations by Meyer-ter- pellets into a reactor chamber (9 m in dia­ achieved a compression factor of 600. meter x 10 m in height) where they are Experimental results for pellet implosions Vehn [31. The model parameters are the ra­ dius of the hot spot Rs, the fuel pressure at imploded using multiple heavy ion beams. in underground nuclear explosions con­ The chamber has a wetted first wall con­ firm theoretical predictions that a pulse ignition p, the isentrope of the compressed fuel α, and the hydrodynamic efficiency η. sisting of an eutectic mixture of PbLi used energy of about 5 MJ is necessary for as a coolant and breeder. The advantage implosion. of this mixture is its extremely low vapour Unfortunately, laser facilities as drivers, Inertial Confinement Fusion Inertial confinement fusion (ICF) in­ pressure which facilitates beam insertion while indispensible for future target re­ and focussing at a rapid cycling rate. The search, have serious shortcomings for volves confining a D-T mixture of a few milligrams in a pellet of up to 10 mm dia­ flow of the PbLi along the walls is guided electricity generating power plant, notably by an array of porous SiC tubes. The tri­ low efficiencies and repetition rates such meter, which is heated and compressed by ablation to ignition conditions: energy tium breeding ratio is 1.25, and the low that it is hard to imagine how the neces­ solubility of tritium in the liquid PbLi re­ sary orders of magnitude improvements can be supplied using either laser or heavy ion beams. The inertial forces of the implo­ sults in less than 100 g of tritium inventory could be obtained to meet commercial in the blanket and fuel processing system. requirements. The principle alternatives sion keep the compressed pellet together are heavy ion accelerators as both the effi­ for a few nanoseconds, long enough to ciency and the repetition rate are not criti­ burn a substantial fraction of the fuel. At Drivers cal and other requirements such as final an ignition temperature of about 5 to 10 A qualitative evaluation of various driver focussing and beam-target coupling ap­ keV, a level of high compression has to be candidates for electric power production pear satisfactory. Enormous progress in reached in order to fulfill the Lawson crite­ [2] shows a clear preference for heavy ion accelerator technology and experience rion ρR > 3 g/cm2 where p is the fuel accelerators. Most important are the ex­ with operational reliability at complex density and R the fuel radius at ignition. cellent properties with respect to repeti­ accelerator facilities also place heavy ion The criterion tells us that, e.g., for an tion rate, efficiency (about 25%), and drivers in a favourable light noting, of amount of fuel which can be handled in a beam-target coupling. An efficient driver course, that the required beam intensities reactor chamber (up to a few milligrams), is essential: with a 5% driver efficiency exceed by a wide margin those of existing a compression of about 1000 times liquid some 60 % of the energy would circulate accelerators. density (200 g/cm3) is needed. for an assumed power gain of 100 for the In view of the progress, our aim here is An ICF power plant consists of three ignited pellet ("pellet gain"), whereas to review the development of heavy ion major components : the driver, the reactor with a 25% efficiency only about 10 % drivers for inertial confinement fusion and cavity (including the conventional electri­ circulates. consider some perspectives. cal installations) and the pellet factory. A heavy ion pulse of 5 MJ has to be deli­ One of the great advantages of the ICF vered to a D-T filled pellet for about the Professor Rudolf Bock is the Programme scenario as compared to magnetic confi­ 10-20 ns required by the dynamics of pel­ Director of the heavy ion fusion programme nement is the separation of the driver and let implosion in order to achieve ignition at the Gesellschaft für Schwerionenfor- the reactor cavity, thus facilitating cons­ and a significant burn. Consequently, the schung mbH, Postfach 11 05 41, W-6100 truction and maintenance as well as ex­ pulse power is 250-500 TW. The pellet Darmstadt 11. traction of the energy generated. gain is plotted in Fig. 1 as a function of the

Europhys. News 23 (1992) 83 Fig. 2 — Accelerator drivers for heavy ion inertial fusion : induction linac (a), recircula­ ting induction linac (b), and RF linac/storage ring (c) scenarios. deposited energy, where for a heavy ion beam of up to several hundred mA. How­ shaping as well as fine focussing. Cons­ beam, the pulse power is a product of the ever, accumulator and buncher rings are truction would take 4-5 years and it may heavy ion kinetic energy and the beam required for current amplification; storage be necessary to have an intermediate step current [3]. Since the kinetic energy is rings must hold the pulses for several milli­ before designing and constructing a full- limited to about 50 MeV/nucleon by the seconds in order to create the final pulse scale driver. A new concept, the recircula­ ion range in the pellet (10 GeV for the structure and pulse currents before feed­ ting induction linac [7] first proposed in heaviest ions), the current in the heavy ion ing the reactor. 1988 (Fig. 2b) is being also considered in pulse must reach 25-50 kA. Based on a order to reduce the cost of a full-scale 5 MJ pulse and a pellet gain of 100 (which Induction linac driver. It combines elements of both the from theoretical considerations — Fig. 1 The pioneering research on heavy ion RF and induction approaches. — is realistic), the output energy of a induction linacs was carried out during single pulse would be 500 MJ. Assuming the last decade at the Lawrence Berkeley RF linac a pulse rate of 20 pulses per second, a Laboratory (LBL), USA [4]. The basic idea The RF linac/storage ring concept looks thermal power of 10 GW would be obtai­ is to inject a long beam bunch of high more conventional than the induction ned — equivalent to an electrical output intensity and to achieve current amplifica­ linac because nearly all the accelerator of about 3 GW. Other problems, notably tion by ramping the inductive acceleration and beam handling elements are known those connected to the beam-pellet inter­ fields as each bunch passes. Pulses are from operating facilities. However, there action, pulse shaping, range shortening compressed from 20 µs at injection down exists little experience of their operation in in a plasma, and pellet design have been to 10 ns at the target, the current being a regime of space charge dominated, non- studied but will not be discussed in detail. increased from amperes to kiloamperes. relativistic beam dynamics, where various One important conceptual improvement types of instabilities may occur and where Heavy Ion Accelerators was the splitting of a single high-intensity the beam transport and any beam manipu­ We shall now consider 1) whether an beam into a large number of parallel beam- lations may be associated with severe accelerator can meet the requirements lets, each being separately focussed in­ emittance growth and intensity loss. discussed above; and 2) the design of an side the same accelerating structure. This We undertook the HIBALL conceptual accelerator-driven power plant. Two types concept improves focussing because of a design study [1] to identify major pro­ of accelerators look promising, namely, smaller emittance: it has been shown to blems: the similar studies performed in the induction linac [4] and the RF-linac be cost-effective if the number of beam- the US, the former USSR and Japan tack­ with storage rings [1]. The latter was lets is in the range of 8-16. The latest con­ led the same issues. In HIBALL, a 5 km RF- investigated in our HIBALL concept study : cept for a driver starts with 64 beamlets at linac provides acceleration at constant other systems studies undertaken in injection which later, when space charge current using RFQ, Wideröe and Alvarez Japan, the US and the former USSR con­ forces decrease, are combined to a final accelerating structures (Fig. 2c). Because firm the favourable prospects. 16 beams. For a Bi3+, the whole length of space charge limitations, the charge In spite of some common features such of the accelerator is about 5 km (Fig. 2a). state of the ions is chosen to be 1 + . Bis­ as the transport and handling of space Recent work at LBL concentrated on a muth was chosen as a beam because it is charge dominated beams, the technical prototype experiment for the multi-beam mono-isotopic, but other heavy elements problems associated with both concepts concept (MBE-4) consisting of 4 beamlets could be used. The front-end starts with are quite different, especially with respect [5]. In spite of being limited to the front- eight parallel channels owing to space to the pulse structure. For an induction end components of the accelerator, the charge limits at low velocity. By funnelling linac, the pulse sequence of the linac can prototype will allow many critical issues to connected with frequency doubling, the be accommodated to the needs of the be investigated because the front-end and parallel channels are combined stepwise reactor right from the ion source : transfor­ the initial pulse formation represent the over a length of about 500 m. Sufficient mation of the pulse structure is not nee­ most critical aspects. A 30 M$ Induction intensity of Bi+ with a normalised emit­ ded. An extremely high intensity from a Linac System Experiment (ILSE) has been tance of 0.2 x 10-6 m rad has already pulsed ion source is necessary, with pulse proposed [6] as a next step that is inten­ been obtained using sources developed at compression during acceleration of many ded to address all key issues of a full-scale GSI to allow a design value of 165 mA for orders of magnitude. driver, including the transport of space the linac current. One of the remaining A high-current RF-linac, on the other charge dominated beams, combining and problems in this area is the question of hand, can deliver a continuous heavy ion bending of beams, compression and pulse emittance growth by funnelling. ►

84 Europhys. News 23 (1992) Transfer, storage and buncher rings are Fig. 3 — An example of an indirect-drive necessary for current multiplication (by a target for heavy ion ICF [10]. The kinetic factor of about 104) and for producing the energy of the beam is converted into soft required pulse length and pulse sequence X-rays which implode the inner shell filled for the reactor. The storage time of 4 ms with D-T fuel. Radiation confinement is achieved with a high-Z material. The con­ which was adopted is a compromise bet­ version efficiency η depends strongly on the ween the longitudinal microwave instabi­ specific deposition power (lower) such that lity in the storage rings (requiring a short a total efficiency of 5-10% requires 1016 storage time) and the beam intensity pro­ W/g. vided by the linac. If the microwave insta­ bility turns out to be less serious, longer bunching capacity, become an essential storage times could be permitted and, ingredient of any driver concept. For consequently, the beam intensity of the example, the specific deposition power linac could be further decreased. On the can be increased by more than an order other hand, because of the short storage of magnitude by using non-Liouvillian time, the driver can provide a much higher techniques. So if non-Liouvillian injection pulse frequency (≈ 100 Hz) than a single works as expected, the deposition power reactor chamber could make use of (5 seems not to be beyond what is feasible, Hz): a single driver accelerator therefore even for the enhanced requirements of can serve a number of reactor chambers. indirect drive. The HIBALL study assumes four cavities Since C. Rubbia first suggested non- but more would be possible, thus further Liouvillean injection by photo-ionisation in reducing the cost of the electricity output. 1988 quite a number of different injection A number of problems have been identi­ schemes and accelerator scenarios have fied for the ring system and for injection instabilities (cover illustration) and upper been discussed, both for a driver accele­ into the reactor. Many of them will be limits for the deviation from symmetry in rator and for a test facility based on the accessible to experimental investigation in terms of Legendre polynomials have been requirements for indirectly driven targets. the near future at the SIS/ESR two-ring determined. Heavy ion beams with suffi­ In addition to schemes proposed by synchrotron facility at GSI (see page 86). cient intensity will definitely not be avai­ Rubbia, non-Liouvillian bunch compres­ Most important are the longitudinal (mi­ lable in the near future to experimentally sion was suggested by I. Hofmann for a crowave) instabilities, the phase space confirm these calculations. However, laser driver accelerator by pre-bunching coast­ dilution in multi-turn stacking and in all beams may give useful information, so ing beams in a stack of storage rings and kinds of beam manipulations, the ion-ion high-power laser experiments are thought depositing these bunches on top of each charge exchange (intrabeam scattering), to be essential. other, by non-Liouvillian injection, into a and final focussing and bunching. Based on present knowledge, it is gene­ stack of 20 compression rings for final For beam stacking, the method propo­ rally believed that the symmetry require­ bunching. Moreover, instead of changing sed by Rubbia (see page 88) will be used ments for pellet implosion might not be the charge at injection, the mass can be to reduce phase space dilution and impro­ achieved with a reasonably small number changed by accelerating a molecular ion ve beam quality during beam accumula­ of heavy ion beams. If this is so, an indi­ and dissociating it at injection. This con­ tion : non-Liouvillian stacking is achieved rect drive would be a necessary ingredient cept was considered by Arnold and Muller by photo-ionisation at injection into a sto­ of a pellet implosion concept. Two addi­ for a test facility, the advantage being that rage ring [8]. A photon energy of 14.5 eV, tional issues are then absolutely crucial, a more convenient laser wavelength is ne­ the ionisation potential of Bi + , correspon­ namely, the conversion efficiency and eded for the photo-dissociation. Some ding to a wavelength of 84.8 nm, is nee­ radiation confinement. preliminary results of all these considerati­ ded to ionise Bi+ to Bi + + . Other ion spe­ Various types of indirect-drive pellets ons are encouraging, but much work has cies with higher ionisation cross-sections with two converters have been suggested to be done in order to prove the feasibility (e.g., Ba) are under consideration. A pho­ and investigated by computer simulation of proposed schemes. ► ton beam with the necessary power can by Meyer-ter-Vehn [10] and others: an only be provided by a free-electron laser example is shown in Fig. 3. To achieve (FEL). Such an FEL appears not to be high conversion efficiency of kinetic ener­ beyond present-day technology. gy into the soft X-rays (which generate radiation pressure to implode the fuel cap­ Perspectives sule) the required deposition power must The principle uncertainties for heavy ion be about 1016 W/g : the total efficiency of inertial fusion are : the pellet is found to be 5-10%, which - Pellet: radial symmetry of the implosion should be adequate for a high-gain target. drive, its consequences for the growth The required deposition power of 1016 rate of hydrodynamic instabilities and for W/g calls for more severe conditions for the beam quality; conversion efficiency in the final focussing and bunching capacity the case of indirect drive. of the driver. - Accelerator: the growth rate of longi­ However, the symmetry obtained from tudinal microwave instabilities in storage two converters might be insufficient. Re­ rings, final focussing and final bunching. Fig. 4 — Some indirectly driven targets for cent analyses have therefore investigated heavy ion ICF [11]. a) The multiple filamen­ Pellet various configurations of indirectly driven tary converters can be designed for very The symmetry of implosion drive has targets (some options are illustrated in Fig. high ion energies but the position of the fuel been investigated recently by Atzeni [9] 4 [11]). capsule (shaded) is critical. b) The hohlraum using a two-dimensional simulation code. casing is completely filled with converter Driving forces on the fuel capsule must be Non-Liouvillian techniques material, attractive from the implosion sym­ extraordinarily uniform owing to the deve­ Non-Liouvillian techniques, in helping metry point-of-view but restricted to low lopment of Rayleigh-Taylor hydrodynamic overcome the driver's final focussing and energy ions.

Europhys. News 23 (1992) 85 Concluding Remarks Regarding the most important imme­ lity might be considered. It should, how­ The heavy ion accelerator is the most diate research objectives: ever, be based on new technology. promising driver candidate for the produc­ - There is a serious lack of experimental tion of electrical energy by inertial confi­ data on many key issues, in particular, REFERENCES nement fusion. The HIBALL design study beam dynamics and the physics of dense plasmas. The SIS/ESR two-ring accelera­ [1] Badger B. et al., HIBALL: KfK Report showed for the first time that the concept 3202 (1981); HIBALL II: idem. 3840 (1989). tor will be a unique facility for such investi­ of an accelerator driven fusion reactor [2] Hogan W.J., LLNL Report UCRL 50021- gations, and related theoretical work must should be technically and economically 86 (1986). feasible. be continued. Any opportunity to investi­ [3] Meyer-ter-Vehn J., Nucl. Fusion 22 However, heavy ion beams represent gate non-Liouvillian beam manipulations (1982) 561. a challenging driver option for ICF. Two and to research all related techniques, e.g., [4] Proc. Conf. on Heavy Ion inertial Fusion, accelerator designs, an RF-linac with FEL development, should be a priority. Washington DC, Eds. M. Reiser, T. Godlove storage rings and an induction linac, both - A new conceptual design study, repla­ and R. Bangerter (AIP, New York, NY) 1986. investigated in the framework of na­ cing HIBALL, involving workers at CERN [5] Fessenden T., Nucl. Inst. & Methods tional programmes during the last decade, (accelerator aspects) and Sincrotrone A278 (1989) 13. are candidates. Two accelerator facilities, Trieste (FEL) has been launched recently [6] Keefe D., [5] p. 266. SIS/ESR and MBE-4/ILSE, will study key (see page 91) in order to include the novel [7] Godlove T. and Yu S., Symp. Heavy Ion issues of both driver concepts. concepts of indirect drive and non-Liouvil­ Inertial Fusion, Monterey, USA (1991). For the enhanced requirements of in­ lian techniques. [8] Rubbia C., [5] p. 253. direct drive, the heavy ion accelerator - A strategy for building an heavy ion [9] Atzeni S., Europhys. Lett. 11 (1990) 639; also continues to be the preferred can­ fusion demonstration accelerator has yet Laser Part. Beams 9 (1991) 232. didate if non-Liouvillian stacking is inclu­ to be developed to enable significant [10] Murakami M. and Meyer-ter-Vehn J., ded. New accelerator scenarios based beam-target experiments and to demons­ Nucl. Fusion 31 (1991) 1315. on non-Liouvillian beam manipulations trate the feasibility of accelerator techno­ [11] Meyer-ter-Vehn J. and Murakami M., have been proposed and promise greatly logy and non-Liouvillian stacking. Either a Particle Accelerators 37-38 (1992) 519. improved beam quality and intensity to dedicated test facility (e.g., with low repe­ [12] Hofmann I., Drivers for Inertial Fusion, meet the requirements. tition rate) or the first stage of a larger faci­ Vol. I (IAEA, Vienna) 1991.

SIS/ESR (Fig. 2) became operational in 1990 and accelerator experiments using the available low intensities started a few months later A Nuclear Physics Facility for Heavy Ion [1]. The electron cooling capability was Inertial Fusion Research demonstrated in May 1990. Installation of the high-current injector indispensible for studies with very heavy ions (about 1011 uranium ions will be injected per spill) has A German government funded program­ unfortunately been delayed. The facility me on the fundamental issues of heavy ion incorporates in principle all the accelerating inertial fusion (HIF) began with an explora­ structures of a complete driver accelerator tory phase in 1979 and was devoted mainly for HIF based on a RF linac/storage ring con­ to the HIBALL system feasibility study (see cept and offers excellent opportunities for page 83), to theoretical investigations of dedicated research on key HIF issues in the accelerator and target issues, and to re­ field of driver and target physics. Most search on some key HIF aspects such as de­ driver relevant issues should be accessible velopment of high-brilliance sources, RFQ's to experimental investigation, except non- and other accelerator relevant components. Liouvillean stacking. Gesellschaft für Schwerionenforschung (GSI) - Darmstadt, KfK - Karlsruhe, the Max- Experimental Programme Planck-Institut fur Quantenoptik in Gar- Experiments in beam dynamics will cover ching, and a number of German universities issues affecting the acceleration, storage participated in the programme which con­ and other manipulations of high intensity centrated on the front-end of a RF linac beams at high phase space density. Experi­ driver approach. mental data are of fundamental interest as The major achievements were a full-scale, well as being urgently needed to design high-brilliance ion source for Bi ions which future facilities. meets inertial confinement fusion (ICF) Fig. 1 — GSI Darmstadt's UNILAC RF-linac. Schottky noise measurements with Ar182 driver specifications, and a low-frequency beams have already shown [2, 3] that the (13.5 MHz) RFQ prototype injector for high- Two additional devices of particular rele­ beam current threshold (the Keil-Schnell current injection into GSI's existing UNILAC vance to HIF are : limit) for the longitudinal microwave instabi­ linear accelerator (Fig. 1). Based on this - a high-current injector for high intensity lity can be exceeded by a factor five without experience, a prototype of a new 27 MHz operation of the synchrotron; any loss of stability. The appearance of a injector for high beam currents is under - a storage ring for beam dynamics studies double-hump structure in the Schottky construction at Frankfurt University, with at space charge limits, with a beam cooling spectrum at the highest phase space densi­ delivery of a prototype planned for 1992. facility for generating low-emittance beams ties marks the transition from single particle The HIBALL study showed that the exis­ to create high energy density in a target, to collective behaviour. ting experimental database is too narrow for and to allow investigations of beam instabi­ For beam-target interactions, the objec­ deciding upon critical HIF issues. Conse­ lity at phase space densities beyond beam tives are the generation and investigation of quently, in the second half of the pro­ stability limits. dense (heavy ion) plasmas of solid state gramme, starting in 1983, a concept for an density produced with a well-defined geo­ experimental facility was developed, partly The SIS/ESR Facility metry by a heavy ion beam of high phase based on the nuclear physics community's The two-ring, heavy ion synchrotron space density. This new area of research plans for a heavy ion synchrotron at GSI. (SIS)/storage ring (ESR) accelerator facility has been opened up by the ESR's beam

8 6 Europhys. News 23 (1992)