Peaceful Uses of Fusion

By Edward Teller *

P/2410 USA

INTRODUCTORY REMARKS

It is a great pleasure to discuss with you the topicof the day and I should like to begin by expressingthe same sentiments with which the previous speakerfinished his paper. It is wonderful that over a largeand important area of research we can now all talkand work together freely. I hope that this spiritof cooperation will endure, that it will be generallyexercised throughout the world in this field and thatit will be extended also to other fields.

It is remarkable how closely parallel the develop-ments in the different countries are and this, of course,is due to the fact that we all live in the same worldand obey the same laws of nature. I was particularlyimpressed by the wealth of detail given by the previousspeaker. I should like to tell you a few generalitiesand, perhaps, a few little details which might help youor might amuse you. Additional details can, ofcourse, be found in other papers submitted to thisconference.

FUSION AND FISSION

The problem of how to release fusion energy in anexplosion process was solved several years ago.How to release this energy in a slow and controlledmanner has proved to be a much harder question.This situation is in sharp contrast with the history ofenergy release from fission. One year of intensivework had sufficed to produce the first nuclear reactorin the early winter of 1942, while several more yearswere needed to perform the first successful nucleardetonation.

Among the reasons for this difference I should liketo mention two which are simple and important.

Fission energy is released by the mechanism of achain reaction. The chain-carrying neutrons can beslowed easily. This facilitates a controlled reactionnot only because the time that the process takesbecomes longer but also because slow neutrons aremore easily controlled by specific absorbers and othermeans. On the other hand, release of fusion energyis accomplished by a heat explosion or, to use thetechnical term, a thermonuclear reaction. The par-

ticles involved in that explosion will necessarily movewith high velocity.

The second and more important difference is thefollowing: The fission chain reaction is regulatedby assembling the correct amounts of active materialsand absorbers. If the correct amount is exceeded,the energy release rises in a rather slow manner dueto the action of delayed neutrons. In contrast, themain regulating feature in a fusion reaction is thetemperature. When a certain ignition temperaturehas been exceeded, the reaction itself raises thetemperature and this leads to an exceedingly rapidacceleration of further energy release. Slow energyrelease is possible only at low densities ; but this raisesfurther problems.

HISTORY

The problem of controlled fusion is difficult but notnecessarily insoluble. Some of us discussed it in arather detailed manner during the war in Los Alamos.This group included Enrico Fermi, John von Neumann,James Tuck and Luis Alvarez.

Some elementary general facts were recognized atthat time : That deuterium gas could react t abovean ignition point of approximately 35 kilovolts; thatdeuterium-tritium mixtures could react î at a con-siderably lower temperature of a few kilovolts; thatthe gas should be introduced at an exceedingly lowdensity of approximately 1014 to 1015 particles/cm3

in order to make the reaction rates sufficiently slowand in order to avoid excessively high pressures;that at these high temperatures the gas will becompletely ionized (such an ionized gas is called aplasma) ; that with the presence of magnetic fields, theions in this plasma follow spiral paths and that byappropriate arrangement of the magnetic fields thelosses to the walls can be reduced; that the pressureof the plasma on the field leads to a thermal expansionof the plasma which tends to stabilize the reaction;that equilibrium with radiation is not established andthat the energy emitted as bremsstrahlung should betreated as a loss ; that for this reason atoms other than

Mev and* University of California Radiation Laboratory, Livermore,

California.

t The reactions areD + D - > T + HD + D -> He3 + n + 3.25 Mev.

t The reaction is D + T -* He3 + n + 18 Mev.

27

28 SESSION 4 P/2410 E. TELLER

hydrogen isotopes must be eliminated as completelyas possible; and that even the equilibrium betweenelectron and positive ion energies will not be completeat the highest temperatures. Very particularly, it wasalso noticed that in a simple closed field along a torus,the particles will not continue to spiral indefinitelyaround the same magnetic line of force but that theywill drift in a direction perpendicular to both the mag-netic field and the field gradient. This leads to smallerbut nevertheless prohibitive wall losses. For the timebeing, no experimental work was undertaken for thespecific purpose of carrying the theoretical considera-tions into practice.

In 1952 and 1953, under the stimulus of a successfulman-made release of thermonuclear energy, such anexperimental approach was started. To a very greatextent this was due to the initiative and confidenceof the chairman of the Atomic Energy Commission,Lewis L. Strauss. It was recognized that the prob-lems of heating and containing the plasma would beextremely difficult. For this reason, several parallelapproaches were undertaken which, between them,amounted to an attack on plasma research on a broadfront.

THE PRESENT APPROACHES

Two of these approaches, the pinch and the stella-rator, were based on the idea of containing the plasmain an annular region. The main difference betweenthese two approaches is that in the pinch an axialcurrent is established with magnetic lines of forceencircling the current, while in the stellarator the rolesof current and magnetic lines are reversed with theprincipal currents established outside the reactingregion. In the United States, the work on the pincheffect was started by James Tuck in Los Alamos.Later, similar work was undertaken in the Universityof California Radiation Laboratory by W. R. Bakerand S. A. Colgate and still later at other sites. Thisapproach has also been very strongly emphasized byworkers in the United Kingdom and in the USSR.Nicholas С Christofilos proposed an ingenious steady-state arrangement, the Astron, which bears somesimilarity to the pinch effect. The stellarator wasproposed by Lyman Spitzer at the very beginning ofthe thermonuclear program and work on it wascarried out in Princeton.

Taking an entirely different approach, the advan-tages of a confinement geometry based on externallygenerated, cylindrically symmetric magnetic fieldswere recognized by Richard Post and Herbert York.Starting with these concepts, Post evolved an approachwhich has come to be called the " Mirror Machine ",which uses various applications of the magneticmirror principle in its operation. In the magneticmirror machine the ions spiral along magnetic linesbetween two regions in which the magnetic field ismore intense and where the ions are reflected. Thiskind of machine is now under study at the RadiationLaboratory at Livermore, at the Oak Ridge National

Laboratory, at the Los Alamos Scientific Laboratoryand at the Naval Research Laboratory.

A COMPARISON OF THE APPROACHES

Each of these proposals has its peculiar advan-tages and its peculiar difficulties. The pinch effectis based on an earlier idea of Willard Bennett \ whoderived the relations according to which a violent gasdischarge will contract into a narrow filament underthe influence of its own magnetic field. The greatvirtue of this scheme lies in the fact that startingfrom a rather moderate plasma density one can pushthe system to very much higher densities and pressuresand at the same time expose the containing systemto much more moderate pressures. One price thatone has to pay is that the high densities are obtainedonly for short periods of time. In other words, weare dealing with a pulsed machine. In order to avoidinstability and breakup of the pinch, the originalsimplicity of the concept had to be compromised byintroducing magnetic fields parallel to the directionof the plasma current. The detailed theory of thepinch contraction, as well as the theory of pinchstabilization, was worked out by M. Rosenbluth atLos Alamos.

Rather early in the game it was noticed that thepinch effect in deuterium gives rise to copious neutronemission. The first public announcement of this factwas made by I. V. Kurchatov.2 Unfortunately, theseneutrons do not appear to be of thermonuclear origin.Considerable progress has been made by S. A. Colgatein explaining the mechanism by which these neutronsare produced. A general difficulty in the pinch effectseems to arise from the circumstance that during theformation of the high-current filament most of theoriginal energy output is lost to the walls. It seemsto be of the greatest importance to understand andprevent these losses.

The Astron is based on the following observation:In contrast to the instability of the pinch current, it isexpected that a current of relativistic electrons will bestable. If this proves to be correct then one can con-fine a plasma in the field which encircles this rela-tivistic current for a longer period than in an unstablepinch. This arrangement is under development atLivermore.

The thorough theoretical studies of the stellarator,carried out by L. Spitzer, M. Kruskal, E. Friemanand others, led to the recognition that the simplestarrangements of magnetic lines are unstable in thiscase also. In order to prevent these instabilities,additional magnetic fields are being introduced in adirection perpendicular to the original fields. Thusthe present pinch and stellarator designs are no longeras sharply different as they used to be. However, thestellarator continues to be a device in which steady-state operation will be attempted and in which themain magnetic fields will originate in a region outsidethe plasma. The problem of heating the plasma inthe stellarator has turned out to be a rather difficultone. The first step in the heating process is by

PEACEFUL USES OF FUSION 29

ohmic current, but soon the conductivity of theplasma becomes high and, instead of uniform heating,runaway electrons are produced with velocities greatlyexceeding the Maxwellian values. From this pointon, heating is accomplished by transferring energyfrom magnetohydrodynamic motions to the particlesin the plasma. This so-called " magnetic pumping "process bears a similarity to the acceleration of cosmicrays as described by, Fermi. Unfortunately, in thestellar at or, as in the pinch, a very considerablefraction of the energy is lost during the heating process.The heating derived from hydromagnetic shocks hasbeen investigated by H. Grad at New York University,M. Rosenbluth at Los Alamos and others.

In the evolution of the mirror machine the basicsimplicity of the original design has been maintained.In addition to this advantage, the straight geometrymakes it possible to operate with a relatively smallvolume. An obvious disadvantage of the magneticmirror machine is the loss of particles through themirrors. In fact, all particles get lost whose motionincludes a sufficiently small angle with the magneticlines. In the course of time, more and more particleswill be scattered into these orbits and eventually allof the plasma escapes. Detailed calculations haveshown that escape through the mirror is a seriousconsideration but that it does not rule out the mirrormachine as an effective thermonuclear energy genera-tor. Furthermore, the end losses can in principle bereduced by inducing a rapid rotation of the plasma.The open ends of the machine also lead to an advan-tage: a simple possibility of injection. If one canintroduce a sufficient quantity of high-velocity,ionized material through these ends, then, in contrastwith the other machines, one can circumvent theproblem of heating a plasma which is already present.The open ends have another advantage: they allowone to study the particles which escape and, therefore,to determine what energy particles must be added tothe plasma. In injecting fast ion beams, difficultieshave appeared due to space-charge limitation in theinjection mechanism.

As has been pointed out by the representative of theUSSR, we are particularly anxious to inject at highenergy because then the leak is not so serious. Aradically different system of injection is being tried atthe Oak Ridge National Laboratory. This system con-sists in bombarding the interior of the mirror machinewith D2+ ions. Inside the mirror machine the mole-cular ions are disassociated in a carbon arc (but wecould use a hydrogen arc) and the more stronglycurved D+ ions are trapped. Considerable progresshas been made in developing this technique and itmay lead to the establishment of a steady-state plasmawhich is suitable for detailed experimental investiga-tion. In any case, a completely satisfactory densityand temperature need not be obtained at once by theinjection into the mirror machine because simplystrengthening the magnetic fields or moving the mir-rors closer together will lead to an adiabatic compres-sion of the plasma which will increase both the densityand the temperature.

The homopolar, conceived by William Baker atBerkeley, is another machine based upon mirror con-tainment, but with an enhanced containment due tothe large induced rotation of the plasma. Similarwork on rotating plasmas (Ixion) is being done byKeith Boyer at Los Alamos. Generally speaking,there are many designs involving a rotating plasmatrapped in a mirror machine. This is accomplishedby using crossed electric and magnetic fields to inducea rotation, with consequent enhancement of the mirroraction.

There is one additional design which has receivedserious consideration but on which little experi-mental work has proceeded. The name of this designis the picket fence or cusp geometry considered byJ. Tuck and H. Grad. It consists of a series of mag-netic field systems so arranged that the magneticfields curve at each point away from the space whichis enclosed and which contains the plasma. At thejuncture of any two of these systems, the magneticfields have a cusp through which the plasma canescape. The difficulties connected with the rate ofthis escape have been the reason for not pursuing thisparticular design any further. It is by no means clear,however, that such a design could not be successful.

In later sessions a more detailed discussion will begiven of the several designs. Here I should like tomention a few general difficulties and to discuss theway and extent to which these difficulties have beenovercome.

SINGLE PARTICLE CONFINEMENT

One may consider the plasma from the point ofview of the individual charged particles in a fixed field.The first and simplest question is whether or not afield of magnetic lines can retain ions on their orbitsfor a sufficiently long time. Since the ions slowlydrift in a direction perpendicular to the magnetic linesof force and to the gradient of these magnetic lines, itis not clear that an effective confinement is actuallypossible. In a simple torus with a longitudinalmagnetic field, the charged particles spiral aroundclosed magnetic lines of force. The inhomogeneity ofthe field causes them to wander off steadily to thewalls. The stellarator is designed to avoid the diffi-culties associated with the drift of particles out of asimple torus. One such method, for instance, is to twistthe torus:

This was proposed as early as 1951.In the stellarator the lines get twisted, with the

result that a line no longer closes but instead returns,after a circuit, into the neighborhood of its startingpoint. If many circuits are executed, a single mag-netic line will have closely covered a toroidal surface.Now an ionized particle, in following this more com-

30 SESSION 4 P/2410 E. TELLER

plicated line of force, will be subject to displacementsdue to the inhomogeneity of the magnetic field. Onecan show, however, that these displacements willaverage to zero in good approximation.

In the mirror machine, an ion spiraling around amagnetic line of force performs a quasi-periodicmotion due to oscillations between reflecting points.The periodicity of the orbit is not strict because theguiding center drifts to neighboring lines of force, ashas been mentioned above. In a cylindrically sym-metric case, it is evident that the drift perpendicularto the line of force will map out a surface of cylindricalsymmetry. But it is not evident where the particleswill drift if there exists a non-symmetric perturbation.The following consideration shows, however, thatdespite such a perturbation the particles do return totheir original line of force with a high degree ofprecision.

During the periodic motion in a static magneticfield the energy will be strictly conserved. Two" adiabatic invariants " are also very accurately pre-served, even though they vary to a small extent.These are the magnetic moment of the spiraling par-ticle in a direction along the magnetic line and theaction jp y dq. Herein is the momentum of the ion(or the electron) along the magnetic line of forceand dq is an element along the magnetic line on whichthe guiding center of the spiraling particle happensto move. The essential character of the orbit of aparticle is determined if its energy and magneticmoment are known. From this, the action \p\\dqfollows uniquely. If now the particle were to shift toan arbitrary neighboring line of force, conserving itsenergy and magnetic moment, the action integral wouldin general be different. However, the action integralmust be conserved also. Therefore, a shift to anarbitrary neighboring line is not possible, but theguiding center of the particle must remain on a sheetconsisting of a single continuous set of magnetic linesof force. In this way, the particle will shift aroundwithin the machine, remaining always on a surface onwhich the action integral is conserved. When theparticle arrives back in the original neighborhood itwill, as a general rule, actually embrace the samemagnetic line from which it had started. While thisstatement is not absolutely correct, it holds with avery high precision for small Larmor radii. Un-fortunately, the conservation of the adiabatic in-variants is not valid if the field varies with a suffi-ciently high frequency. Under these conditions, ofcourse, the energy is not conserved either.

INSTABILITIES

Let us consider a time-independent distribution ofmagnetic lines of force in space, and let us also considera static distribution of a plasma with vanishingelectrical resistivity. The magnetic lines need not becurl-free because we may assume that there areelectric currents in the plasma to account for the curl.No electric fields will be present : if they were parallelto the magnetic fields they would give rise to a change

in the electric current ; while if they were perpendicularto the magnetic lines they would cause the plasma tomove, which is contrary to our assumption of a staticsituation.

Now the stability of this situation may be decidedby investigating infinitesimal virtual displacements.In such a displacement, the material of the plasmamay slide along magnetic lines of force but, in anymotion transverse to the lines, the plasma and thelines must move together. Furthermore, any expan-sion or compression of the plasma accompanying thedisplacement must be considered as adiabatic. Ifsome of these displacements lead to a linear changeof the energy of the plasma then the system is not inequilibrium. If all changes depend quadratically onthe displacement and if all infinitesimal displacementsare connected with positive energy changes, then theplasma is stable. If some displacements are con-nected with negative quadratic terms we expect anunstable equilibrium. Many such instabilities havebeen discussed, and the discussion has been carried toa detailed calculation of the way the particle distribu-tion, magnetic fields and currents change during thedisplacement. However, the hydromagnetic con-sideration given above appears to be a simple andgeneral description which in many cases is satis-factory.

The first models considered for the pinch machineand for the stellarator turned out to be unstable.Appropriate modifications of the field distributionhave given rise to reasonable hopes for circumventingvarious troubles. It appears, in a great measure, to bea question of whether these instabilities are of a hydro-magnetic origin or are connected with plasma vibrationsexcited by runaway processes. We are more doubtfulabout the plasma vibrations, which may be a very realand more lasting stumbling block. In this kind ofmotion, it turns out that the faster the electron thelonger is the wavelength of the plasma vibration thatcan be excited, and the longer the wavelength the moreenergy it can take up. To summarize the situation verybriefly: it has been suspected and partly found thatthese plasma vibrations have the disagreeable conse-quence that the plasma can rapidly diffuse and thereis a leak in the magnetic confinement. These are someof the difficulties in the stellarator and pinch devices.Work is proceeding and there are methods, partlyhoped for and partly realized, which will be discussedin other sessions.

The general question of whether a moving, steady-state, hydromagnetic system is stable or not has beendiscussed in some interesting special cases but has notbeen treated along similarly simple lines as has thecorresponding static problem. These dynamic effectsprobably play a very important role during theconstriction of the electric current in the pinchprocess and during the heating of the plasma in thestellarator. At the same time, the runaway electronsmentioned above probably give rise to plasma vibra-tions, which also might arise by coupling with thehydromagnetic motions. All these motions can gene-rate enough disorder to destroy conservation of energy

PEACEFUL USES OF FUSION 31

and conservation of the adiabatic invariants for thesingle particles which are supposed to be confined inthe system. As a result, both plasma and energy canbe lost to the wall.

We see, therefore, that heating, plasma injectionand plasma compression do represent sizeable addi-tional problems. These problems are, however,closely connected with the more general question ofconfinement. If the latter were solved completely,then we could take time with the other processes andprobably execute them sufficiently carefully so thatthe plasma could be brought to the appropriate con-ditions. The actual fact is, however, that the problemof plasma confinement is only partially solved and,therefore, this problem and that of the establish-ment of the right temperatures and densities have tobe handled together.

EFFECTS OF IMPURITIES

Another difficulty that is quite generally encounteredis connected with impurities. The presence of im-purities leads to increased bremsstrahlung radiation andthe consequent rapid loss of energy from the plasma. Attemperatures low compared with the ignition tempera-ture, these losses are often aggravated by radiation fromincompletely ionized atoms. These partially ionizedsystems are usually not in S aha equilibrium, since ofthe four possible processes, collisional ionization,radiative electron capture, recombination by triplecollisions and ionization by photons, only the first twooccur. This somewhat involved problem has beenthoroughly investigated by R. F. Post and has led tothe recognition that heating must be carried outrapidly, particularly in the low temperature regions.

If neutral atoms are knocked off the walls, thesecause an energy loss by an additional mechanism:They can neutralize fast plasma ions by chargeexchange and these fast neutral particles can cross themagnetic barrier and reach the wall. Here they canknock out more neutral particles and a loss mechanismcan be built up.

EXPERIMENTAL DEVELOPMENT

The above enumeration of difficulties and problemsis by no means complete. Neither does this theoreticaldiscussion do justice to the experimental problems ofmachine design and to the diagnostic work which hasbeen carried out. The development of extremelylarge, pulsed capacitor banks for creating the largevolume, high magnetic fields required, and the opera-tion of large " baked out " vacuum systems for highplasma purity at pressures as low as 10~~10 mm Hg aremajor achievements of the experimentalists. Theobservation of gross current and electric field behaviorand the determination of magnetic field and plasmapressure by means of probes introduced into a systemhave given valuable data. Plasma densities havebeen measured by experiments involving absorptionand transmission of electromagnetic waves in theradar range. Spectroscopic observations have been

carried out and the appearance and behavior ofneutrons have led to additional knowledge about thestate of the plasma. Thus, our knowledge of plasmaphysics is increasing considerably with each passingyear. In this way, important features of plasmatheory have been experimentally verified. The inter-action between magnetic pressure and hydrodynamicmotions have been observed and transport phenomenain the plasma have been investigated. These aregreatly complicated by the effects of collectivemotions but, when properly attacked, have yieldedsatisfactory verification of our expectations.

Hydromagnetic motions have been extensivelystudied, most particularly in observing the formationof a pinch. In connection with the same experiment,the evolution of instabilities has also been investigated.In the stellarator, the mirror machine and the pinchdischarge, the importance of impurities has beendemonstrated by studies of the influence that the wallmaterials have on the observed phenomena. In thecase of the stellarator, a very ingenious device, thedivertor, has been developed which harmlessly elimi-nates that part of the plasma which otherwise wouldtouch the wall and would give rise to the emissionof disturbing atomic species. The use of the divertorhas greatly improved the operation of the machine.

After proper precautions, temperatures of 100 elec-tron volts or more have been reached in the variousmachines. This should be compared with the fewthousand electron volts needed for the effectiveburning of a deuterium-tritium mixture. More than100 electron volts have been reached in the stellaratorand still higher temperatures in the mirror machine.Proved temperatures in pinch devices are a little lowerbut, on the other hand, greater densities have beenobtained.

Perhaps the most important observable variable isthe containment time. In the pinch effect, the timeduring which a high density prevails is only a few tensof microseconds. On the other hand, the operationof this machine is in least need of protracted confine-ment times. In the stellarator, confinement times ofa millisecond or more have been established. In themirror machine, confinement times of several milli-seconds are not unusual. This is particularly pleasingbecause of the small size (a few cubic centimeters)and high electron temperature (more than 10 kev)of the plasma. Furthermore, in this machine nodifficulties have arisen in connection with instabilitiesor disordered motion. I would therefore term thisquite a hopeful approach. These times have to becompared with a period of approximately one secondin which a successful thermonuclear reaction woulddouble the original input energy.

NEUTRONS

Throughout the development, it has been believedthat production of true thermonuclear neutrons wouldbe a milestone in the development of a controlledthermonuclear reactor. The observation of suchneutrons would not be merely encouraging but would,

32 SESSION 4 P/2410 E. TELLER

at the same time, give us a sensitive method of measur-ing temperature distributions within the machine:the slightest rise in temperature would give an enorm-ous rise of neutron production. In " thermonuclearmachines " neutrons have been repeatedly observed.Unfortunately, these neutrons may be due to someform of organized motion rather than to the randomthermal agitation. This shows in a most directmanner that we are still very far from a successfulthermonuclear machine. Furthermore, even whentrue thermonuclear neutrons are obtained, their effectscould at first be obscured by the presence of theneutrons due to collective motions.

THE FUTURE OF THERMONUCLEAR POWER

I should now like to ask: Where are we going?I believe that thermonuclear energy generation ispossible. Whether it will be in precisely seventeenyears, as our Chairman predicted three years ago, orat some other time is a matter that I think he will notargue with me and I shall not argue with him. How-ever, I will say this: The problem is not quite easy.I will also say that on the path there may be somelittle flowers to be picked. Plasma physics hasimportance in the cosmic arrangement of things, aswe heard in Professor Alfvén's paper during thissession. It may have important technical applica-tions other than energy production. I wish I couldknow and tell you more about this, and I hopesuggestions will come forward; because this needsimagination and I am lacking in imagination. If wewant to shoot for the jackpot, for energy production,I think that it can be done, but do not believe that inthis century it will be a thing of practical importance.This view may be much too conservative and may beproved wrong within the coming decade, but I shalltell you why I believe what I believe.

It is likely that we shall be dealing with an intricatemachine which is inaccessible to human hands becauseof radiation and on which all control and maintenancemust proceed by remote control. The irradiation ofmaterials by neutrons and gamma rays will cause theproperties of these materials to change. Surfacesbombarded by the bremsstrahlung radiation will getheated more fiercely than is the case in any portion ofour present nuclear reactors. You can operate themachine to the extent that this one surface can becooled, the rest of the machine being at a relativelylow temperature. These and other difficulties arelikely to make the released energy so costly that aneconomic exploitation of controlled thermonuclearreactions may not turn out to be possible before theend of the 20th century.

Nevertheless, the ultimate goals toward which weare working are apt to be highly rewarding. Wheneconomic thermonuclear energy production becomesfeasible we shall reap a number of important benefits.The fuel of the thermonuclear reactor is cheap andpractically inexhaustible so that, if I may put it thisway, we have deuterium to burn. Thermonuclearreactors produce Jess dangerous radioactive materials

and, when once brought under control, are not likelyto be subjected to dangerous excursions. Therefore,they can be operated more safely than fission reactors.Finally, the interaction of a hot plasma withmagnetic fields opens up the way to the direct pro-duction of electrical energy. This may be of greatpractical advantage since high-temperature heat ex-changers and many moving parts could be eliminated.

Now I have a question: Can all this be done?I think we are at a stage similar to the stage at whichflying was about one hundred years ago. There aresome wise people now, as there were at that time, whohave proved that it cannot be done. I should like to saythat those people were perhaps better off because atleast they saw the birds. All we can see are the sunand the stars. The sun produces thermonuclearenergy by brute force or, what is worse, by sheerinertia. Other people will say that the sun does itwith the help of infinite patience. I do not thinkany physicist wants to go along either of those direc-tions.

PLOWSHARE

This survey should not be concluded without men-tioning another way in which thermonuclear energycan be made to serve peaceful and constructivepurposes. Our recent investigations have led to theincreasing expectation that thermonuclear explosionscan be brought sufficiently under control to be of helpin earth-moving, in mining and also, conceivably,even in the production of energy. We can also under-take projects which I should like to designate as geo-graphical engineering. We can dig canals, we canmake harbors, we can open up water reservoirs forbetter water supply of arid or semi-arid regions. Inthis field, people from the USSR have made very earlystatements which I believe raise very early hope,hope which I believe, in principle, to be correct.

Recently, an underground detonation of a fissiondevice in Nevada has demonstrated the feasibility ofthe confinement of the energy of the device. Further-more, this experiment has shown that the bulk of theradioactive fission products were entrapped in theglass formed by the molten rock. The energy in thiscase was primarily dissipated at a low temperature invaporizing the relatively large water content of therock but, in addition, a large quantity of rock wascrushed. With the equivalent of one ton of highexplosive we can pulverize more than one hundredtons of rock. This could be useful in mining and forthe enhancement of the oil flow in large low-specific-yield formations. Therefore, as has been the case sooften in history, a method for destruction has turnedin a direction where it can be made useful for the mostimportant and, in the long run, cooperative effortbetween the nations.

In the peaceful use of nuclear explosives, the fusionreaction is likely to play a prominent role for tworeasons. One is that the fusion fuel is exceedinglycheap. The other reason is that fusion reactions giverise to less radioactive contaminations than do

PEACEFUL USES OF FUSION 33

fission reactions. In fact, a careful planning of thefusion explosion can reduce the activity to a pointwhere its effects are quite small compared with theeffects from an equivalent fission reaction. This isparticularly important in earth-moving applicationsdesigned to build harbors, canals, change the distribu-tion of water or lay bare mineral deposits.

There is no difficulty, in principle, of releasing largeamounts of fusion energy underground, in the con-version of this energy into high pressure steam and inthe use of this steam for the production of electricity.It is, however, important to find a cheap way of con-fining the explosion in order that the economicalexploitation of explosive fusion energy for the pro-duction of electricity may become possible. It is notunreasonable to expect that explosive fusion energycan be harnessed for the production of electricitybefore the same feat can be accomplished in the con-trolled fusion process. It will also be noticed thatboth the cheapness of the fuel and the reduction ofradioactive contamination are the same factors whichalso play an important role in our future expectationsconcerning controlled fusion reactions.

Even though the first application of thermonuclearpower may be by using controlled explosions—andthis would be a very great achievement indeed—in

the long run, the inherent advantages of controlledfusion in a steady-state system will probably dominateour planning for the far future. At any rate, in thedirection of controlled explosions, as well as in thegreat field of plasma physics, I see a very fruitfulfield of research for physicists, for groups of physicists,for scientists in a whole country and for scientists inthe whole world. I also believe that one thing whichscience and technology can do for us is to draw us allcloser together into mutual help and into better under-standing if we are guided by a search for the thingsthat can be accomplished by active work, which wecommunicate to one another and in which we cooperate.In any case, I am convinced that the extraordinaryand abundant power of fusion energy can be used inpeaceful pursuits for the benefit of mankind. Whenthis has been accomplished it may well turn outto be one of the most important advances of our age.

REFERENCES

1. W. H. Bennett, Magnetically Self-Focusing Streams,Phys. Rev., 45, 890-7 (1934).

2. I. V. Kurchatov, On the Possibility of Producing Thermo-nuclear Reactions in a Gas Discharge, Harwell Conference(1956).