Intense Gaseous Discharges

By J. S. Luce

P/1790 USA

Much of the gas discharge research at the OakRidge National Laboratory had its beginnings in theion-source development work carried on under thewar-time Electromagnetic Isotope Separation Projectwhich was begun at the University of CaliforniaRadiation Laboratory.1 Since the summer of 1954many of the studies have been motivated by the needsof the thermonuclear program. These studies haveled to the evolution of several new types of arcs whichexhibit certain interesting characteristics. Thesedischarges, all of which require an axial magneticfield for stable operation, appear to possess propertieswhich are qualitatively different from those of pre-viously investigated arcs. This paper describes thedevelopment of these arcs, their use in the ORNLthermonuclear program, and possible applications ofmore general interest. A more thorough study ofsuch discharges may eventually lead to a better under-standing of plasma physics.

In February 1955, the observation was made atORNL that energetic deuterons could be trapped in amagnetic field by the injection and dissociation ofdeuterium molecular ions.2 The idea is illustrated inFig. 1.

The essential feature of this method is the injectionof a D2+ ion across a magnetic field and the subsequentdissociation of this ion into a D+ ion and a D° atomor into two D+ ions and an electron. In either case,each resulting deuteron has half the momentum ofthe original molecule and hence half the Larmorradius in the magnetic field. If dissociation occursover a certain range of the molecular ion orbit, theresultant D+ is trapped and does not return to strikethe injector. It has been shown by calculation 3> 4

that if a sufficient number of deuterons with energiesof the order of hundreds of kilo electron volts could betrapped, a thermonuclear plasma would be formed thatwould maintain a high temperature and could thenbe fed by much ]ess energetic fuel.

Much of the effort in the last three years has beendirected toward the development and exploitationof devices designed to produce intense beams of mole-cular ions and devices to dissociate these ions with highefficiency. This program has culminated in the

* Oak Ridge National Laboratory, operated by UnionCarbide Corporation for the U.S. Atomic Energy Commission.

ORNL experimental device, DCX,4 in which anattempt is being made to achieve a reacting plasmaby the means just outlined.

The ion-source development program has led to theconstruction of sources capable of producing 2.5 am-peres of D2+ ions and five amperes of D+ ions. Thisprogram is described briefly in the Appendix.

The first part of the paper is devoted to the dischargenow used for dissociating molecular ions, which hasbeen given the name energetic carbon arc. A briefdescription of the arc is followed by an account ofits evolution and development. The experimentswhich established its high efficiency for dissociationare discussed in some detail; some experimentalresults on arc properties are described; and detailsof techniques of operation complete the discussion.

The last part of the paper describes experimentson the newly-discovered energetic deuterium arcsand some speculations on their possible uses.

ENERGETIC CARBON ARC

The discovery which, more than any other, madethe high-energy molecular-ion injection and trappingscheme appear feasible was that of a high-current dcarc which was found to have a remarkably highefficiency for dissociating molecular ions. This arcis run between graphite electrodes in a magneticfield of several kilogauss and is supported by carbonfrom the electrodes with no supplementary gas input.

The arc is ordinarily struck as a dc gas arc with aradio-frequency voltage applied for initial breakdown.After the discharge has become established, the gasflow is completely stopped. Initially, there is highlocal pressure (1 to 5 microns Hg) resulting fromoutgassing, and a secondary plasma is observedaround the arc. When the pressure has droppedbelow about 0.1 micron Hg, the secondary plasmadisappears, the light emitted by the anode increases,and the discharge is observed to separate into brightblue filaments which move slowly across the magneticfield. These filaments terminate at bright cathodespots located on the cathode rim.

Most of the arc current appears to be carried in ahollow cylinder whose diameter is roughly that of thecathode, customarily about half an inch. Figure 2shows a set of wafers cut from thin graphite plates asthe arc burned through them. The photograph was

305

306 SESSION A-6 P/1790 J. S. LUCE

D+ -— D+ + D° MAGNETIC FIELD D* - * * D+ + D+ + e

Figure 1. The trapping of energetic deuterons in a magneticfield by dissociation of deuterium molecular ions

made shortly after the arc was extinguished; thecarbon pieces were still quite hot. In arcs which areseveral feet long, the current channel appears to con-verge before reaching the anode.

Associated with the production and disappearanceof filaments, there appear to be electrical or thermalgradients at the cathode which cause small pieces ofhot carbon to be ejected at high speed (10 to 20 metersper second). Many of these particles rebound fromsolid objects without disintegrating. A photograph ofan arc and some of these fragments is shown in Fig. 3.

It has recently been discovered that this arc isbut one of a broader class of arcs, all of which arecharacterized by electron energies of tens to hundredsof electron volts and, it is believed, by comparableion energies. These have been given the genericname energetic arcs ; some preliminary results on theirbehavior are given in a later section.

Developments Leading to its Discovery

The initial investigations in the program were madewith arcs of the type developed for the isotope sepa-rator (mass spectrometer) ion sources.1 It was withsuch an arc that the ORNL experiments bearing onthe diffusion of ions across a magnetic field5 weredone. A typical arrangement of such an arc is shownin Fig. 4.

The electrode which defined the arc column,although useful in the original application, provedto be a serious obstacle to achieving high arc currentsbecause of excessive electron drain between the fila-ment and the defining electrode. Accordingly, modi-fications were attempted which dispensed with thedefining electrode. Figure 5 shows an early arrange-ment of this kind.

A sufficient flow of gas is supplied through a holein the cathode to provide ionization and space-chargeneutralization at the cathode face with minimalincrease in pressure in the rest of the system. Thedischarge is defined by the boundary of the cathodeitself. In striking, a radio-frequency discharge iscustomarily used to provide initial breakdown of thegas. In operation, a sheath forms at the cathodeface, providing an electric field to accelerate the ioniz-ing electrons. Arcs of this type have been operatedwith currents up to 5000 amp, and many of theirproperties have been studied.

Their markedly different behavior under differentpressure conditions is worth noting. It was foundthat at 50 microns pressure and no magnetic fieldthe arc would blow up to a diameter of several feet(Fig. 6). At a pressure of 0.5 microns the arc wouldspiral in several patterns depending on the currentand voltage applied (Fig. 7).

Experiments designed to measure the efficiencyof gas arcs of this kind for dissociating molecular ions

Figure 2. Half-inch carbon wafers, cut from graphite plates bythe arc

Figure 3. Twelve-inch energetic carbon arc

were unsuccessful because of high gas pressure. Inthe course of attempts to operate these arcs at suc-cessively lower pressures, it was found that underproper conditions a discharge could be made to ope-rate between carbon electrodes without a sustaininggas input. It was later found possible to operate thistype of discharge in lengths of 6 feet or more. Thelow gas pressure associated with this carbon arc madeit immediately evident that this was a strong candi-date for use as a dissociating agent for molecular ions,and experiments to test its efficiency were undertaken.

Molecular-ion Dissociation ExperimentsOf most immediate interest to the thermonuclear

program has been the energetic carbon arc's efficiencyfor dissociating deuterium molecular ions D2+which pass through it. The initial experiments wereconducted with 20 kev D2+ ions in a uniform magneticfield of about three kilogauss. The ions were producedin a modified source of the type originally used forisotope separation. The exit aperture of the source

INTENSE GASEOUS DISCHARGES 307

- CATHODE SHIELD

Ik.

•^~- ARC

^ ^ — D E F I N I N G ELECTRODE

- HEATED CATHODE

Figure 4. Arc with defining electrode

was a slit oriented parallel to the magnetic field, whichproduced a ribbon-shaped ion beam. The carbonarc was placed at the opposite end of the orbit dia-meter, a location of first-order focus for the beam,and current collectors were located on the anticipatedpaths of the molecular beam and the dissociatedatomic beam. The configuration is shown in Fig. 8.

The current collectors were water-cooled and wereprovided with raised portions around their edges toprevent the loss of secondary electrons. The targetfor the atomic beam was well shielded from the arcby a water-cooled copper box. An average figure forthe molecular ion current during these experimentswas 60 ma. The pressure was always under 3 x 10~5

mm of Hg, and the arc current was 300 amp.The intensity of the atomic beam was determined

both electrically and calorimetrically. The latterdetermination was accomplished by measuring thetemperature rise of a metered flow of cooling water.The two sets of data agreed to within a few per cent,about as closely as would be expected with the knownintrinsic measurement errors.

The major result of the experiment was the observa-tion that 40% of the molecular ion beam was disso-ciated. Allowing for the beam's imperfect focus, ithas been calculated that 70% of the molecular ionswhich passed through the arc were dissociated. Asubsequent measurement made with 700 amp of arccurrent was made inexact by beam losses due to highpressure but indicated that approximately 90% disso-ciation was achieved.

No metered loss of ions was detected in the electricalmeasurements. The sum of the currents in the R/2and the residual D2+ beam (plates A and B, Fig. 8)always equalled the current reading of the D 2

+ beambefore dissociation.

ANODE-

-CATHODE SHIELDS

Figure 5. Cathode-controlled dc gas arc

Figure 6. Behavior of 500 amp argon arc at 50 microns pressurewith no magnetic field

That dissociation occurs only within the arc wasshown by the good focus of the dissociated beam.The existence of this focus is indicated by the followingobservations: (1) probe-scanning of the focal line ofthe D2+ beam (at the arc position) and the focal lineof the dissociated D+ (at the position of collector Вin Fig. 8) showed no appreciable difference in widthof focus; (2) stainless steel strips placed at the focalline of the dissociated beam were readily meltedthrough even when a beam current as small as 10 mawas used. Such a strip is shown in Fig. 9.

Only brief studies were made on the circulatingdeuteron beam. These were sufficient to demonstratethat the deuterons did remain in the system for asubstantial number of turns and to suggest that theydid not lose appreciable energy before leaving thesystem. An indication of the shape of the circulatingbeam was obtained by permitting it to burn throughthin aluminum foils. A comparison of the patternmade by a single-pass beam with that made by acirculating beam was obtained by making use of amovable intercepting plate, as shown in Fig. 10.

The respective foils are shown in Fig. 11; the oneon the right shows that all of the circulating particleshave the same radius of curvature, indicating negli-gible energy loss in the arc.

The extremely encouraging results of the dissocia-tion experiments led to the decision to incorporatethe energetic carbon arc in a device designed tocreate a steady-state high-temperature plasma bythe injection and trapping of energetic deuterons.This device, DCX, and the basis for its design aredescribed in another paper.4 One crucial point in

Figure 7. Behavior of 500 amp twelve-inch argon arc at0.5 microns pressure and two magnetic field values

308 SESSION A-6 P/1790 J. S. LUCE

ENERGETIC NEUTRALS

tíI ' I

I I I , A = 60 maВ = 0

A = 36 ma

В = 24 ma

ATOMIC ION BEAM, —

Figure 8. Experiment to measure dissociation efficiency of carbon arc

determining the feasibility of the 600 kev D2+ injec-tion energy chosen for DCX was that the dissociationefficiency of the arc remain substantial at that energy.Experiments 6 verified that this was so.

Properties of the Energetic Carbon Arc

The remarkably high dissociation efficiency of thecarbon arc and the prospect of forming a thermo-nuclear plasma adjacent to it have occasioned aconsiderable interest in its detailed properties. Themeasured dissociation efficiency for 600 kev D2+ ions 6

implies that the product of particle density and crosssection for the event is of the order of 0.1 cm-1. Thus,either both quantities are extremely high in termsof ordinarily expected values (w~1013 cm-3, a>—Ю-16

cm2), or one quantity is anomalously high. Studiesso far have failed to provide unambiguous answersto these and other key questions. Therefore, onlya list of experimental results will be given here; noattempt will be made to use these results to analyzethe arc.

1. The energetic carbon arc has been operated atlengths of from 6 inches to 6 feet. The arc vol-tage, with an arc current of 250 amp, was found tobe 55 v for a 12-inch length and to increase roughlylinearly to 120 v for the 6-foot length. It was foundto be a slowly-varying function of the arc current,the voltage increasing for increasing current. Forany configuration, there was a critical current belowwhich the arc would not continue to run. For a12-inch arc, this current was as low as 50 amp for a fewseconds.

2. Measurements were made of the arc's efficiencyfor stripping additional electrons from 20 kev N+and He+. The experimental configuration was thesame as that shown in Fig. 8. The change of chargeon the stripped ions reduced their radius of curvatureby a factor of 2. The results were: for nitrogen,12% stripping efficiency; for helium, almost zero.

3. Spectroscopic observations 7 showed lines requir-ing the presence of 43 ev electrons for their excitation.The dominant spectra were those of C+, C+ + , and

C+ + + . The relative intensities of these lines variedwith arc length, the spectra corresponding to higherstages of ionization increasing in intensity as the arclength became greater. The resonance line of neutralcarbon at 2478 Â could not be detected in the centerof a 6-foot arc. Doppler shifts observed in thelong arc indicated ion rotational energies as high as50 ev. Velocities parallel to the axis were foundto correspond to energies of only a few ev except mthe vicinity of the cathode.

4. An experiment was performed to measure theamount of energy radiated from a one-foot sectionat the center of a 6-foot arc. Figure 12 shows thearrangement of the apparatus for this measurement.A water-cooled cylinder was placed around the arcand shielded from the electrode radiation by baffleplates which fit closely around the arc. The powerabsorbed by the cylinder was determined from thetemperature rise of its cooling water, whose flowwas measured. With an arc current of 240 amp andan applied potential of 120 v, it was found that theenergy transferred to the cylinder varied between2.5 and 3 kw, depending somewhat upon pressure.If this number constitutes a good average figure forthe power radiated along the arc's length, abouthalf of the total arc power input is thus accounted for.Experiments6 suggest that most of this radiationcomes off in the form of hard ultraviolet light(A—'1200 Â). This is a plausible finding in view ofthe fact that the resonance radiation from C+, C+ + ,and C+ + + lies in the hard ultraviolet region. Radia-tion balance calculations yield a probable ion densityrange of 1013 to 1014 ions/cm3 depending on theelectron energy assumed.

5. Probe measurements and arc-current studiesshowed that the operation of the arc was accompaniedby considerable rf oscillation with frequencies at leastup to tens of megacycles.

6. Preliminary results of microwave experiments 8

indicated the arc to be strongly absorbing for four-millimeter and twelve-millimeter waves. The sug-

INTENSE GASEOUS DISCHARGES 309

1,11.1.11.1111111» шиииишиишиитишиташ^щащаштаййга^

Figure 9. A 0.005-inch stainiess steel strip burned through by abeam of deuterons produced by dissociation of D2+ in a carbon arc

gested density is above 7xlO1 3 electrons/cm3, if thetheory is adequate.

7. Studies of the vacuum properties of the arc 9> 10

indicated not only that it would be useful as an ionpump, but also that the deposited carbon adsorbsgases and thereby provides additional pumpingaction. This action was found to be strong enoughto make it possible to valve ofí the diffusion pumpsof apparatus in which the arc was operating. Thebase pressure under such circumstances was of theorder of 10-5 mm Hg. When the arc was run througha chamber from which the electrodes were excluded,a base pressure of 2 x 10 -7mm Hg was achieved in the(unbaked) chamber.

Techniques for Carbon Arc Operation

In the two years since its discovery, the energeticcarbon arc has been operated with several markedlydifferent magnetic field configurations. Initially,it was run in the uniform field of a mass separatorunit with its length limited to about 12 inches. There,the assembly was arranged roughly as shown in Fig. 5with the anode at ground potential. The majorauxiliary components were the floating baffle plateswhich prevented damaging strike-overs from thecathode assembly.

The feasibility of operating a 6-foot carbon arcwas first demonstrated in the long solenoid shown inFig. 13. Typical electrode assemblies are shown

in Fig. 14. These are inserted in opposite ends of thesolenoid. In the experiments it was found convenientto replace the floating cathode shields with pieces ofquartz tubing slipped over the cathode holder andwith a floating baffle plate to prevent the back-streaming ions that miss the cathode proper fromstriking the cathode holder.

In the magnetic mirror apparatus, DCX, the car-bon arc has been operated in both on-axis and off-axispositions (Fig. 15). The electrodes are customarilyplaced as much as 10 inches outside the magnetic fieldmaxima. With the arc off-axis it was found that thebackstreaming ions no longer struck the supportingassembly for the cathode; the need for the quartzcover was thus eliminated. This method of operationalso provided the first opportunity to study these ions.It was found that an insulated plate in their pathacquired a positive potential as high as 110 v withrespect to the cathode. When the plate was operatedat anode potential, it drew a current of 20 amperes.The total arc voltage was about 140 v.

In order to reduce the conductance for gas flowbetween the two parts of the DCX dual vacuumsystem, it was found necessary to pass the arc througha series of close-fitting baffles at each end of the innerchamber. This method of arc operation presentedtwo problems, one of mechanical alignment and theother of baffle insulation. The alignment problemwas solved by burning holes through the baffleswith the arc at low current and with some gas feed,then mechanically reaming the holes to a larger dia-meter. In order to achieve stable arc operationthrough the baffles it was found necessary to insulatethe baffle holders carefully, since inadequate insula-tion allowed a part of the arc current to be by-passedthrough the baffle assembly.

ENERGETIC DEUTERIUM ARCS

The discovery of a deuterium arc capable of disso-ciating 20 kev D2+ ions resulted from a proposal by

CARBON DISCHARGE

INTERCEPTION PLATE

ION SOURCE

Figure 10. Foil-burning experiment to demonstrate beam build-i

CIRCULATING ~ BEAM

(BALL OF YARN)

up

310 SESSION A-6 P/1790 J. S. LUCE

Figure 11. Foils burned through by deuteron beams in a uniform magnetic field. Width of single-pass beam is 2 inches

P. R. Bell. He suggested that a gas arc could bemade to operate in a manner similar to the carbonarc if some way were found to limit sharply the extra-neous gas emitted into the vacuum chamber. Heproposed to achieve this by improving the tubularanodes (which had formerly failed to produce ade-quate gas arcs) by making them longer and makingthem fit the arc closely. With this improvement,gas fed in at the anode base would have to passthrough a considerable length of arc before it couldbe emitted into the vacuum chamber and would thushave a high probability of being ionized.

Several attempts were made with a tungsten anode5 inches long and bored to an inner diameter offive-eighths of an inch. The outer diameter of thetungsten cathode was half an inch. Electronsfrom the cathode were thus expected to clear theanode wall by one-sixteenth inch. Eventually, alow pressure arc (5xlO~5 mm Hg) was establishedwith a current of 150 amp and a potential of 200 voltsapplied across a 4-inch gap. An arc of this typewas demonstrated to have a 10% efficiency for disso-ciating 20 kev deuterium ions.

A subsequent discovery, which will be described,led to an improved version that provided 25% disso-ciation, although in all these cases tungsten impuritiesmay have played a part. The electrode configurationis shown in Fig. 16. Investigations of this arc arestill in their initial stages, and very little is knownabout the mechanism of its operation. The archas, however, displayed some unexpected properties.Investigation of these has led to several variations inarc configuration and has suggested possible applica-tions of much broader interest than originally antici-pated.

One curious feature of these arcs is the fact that theycan be run from the inner surface of a hollow cathode,thus depending on emission across the magnetic field.

BAFFLE

CATHODE

BAFFLEANODE

MAGNETIC FIELD

Hollow cathode arcs are not uncommon in applicationsin which little or no magnetic field is present, but inthis case fields up to 6000 gauss have been used and nodecrease in the emission observed. One hypothesiswhich has been advanced to explain this fact is thefollowing. Electrons are accelerated radially fromthe inside wall of the cathode across a very thinsheath. Some of those that suffer collisions beforereturning to the wall are trapped and thereupon aredrawn out of the cathode by an axial potential gra-dient in the arc. The idea is demonstrated in Fig. 17.

The hypothesis is supported by the observationthat at low pressure the arc current appears to becarried in a thin hollow cylinder close to the insidecathode wall. Under certain conditions, the currentflow to the anode was found to take place in a well-defined ring, giving rise to intense local heating.Figure 18 shows the result of insufficient cooling ofa tungsten anode. Improved cooling was achievedfor both the cathode and anode by inserting theelectrodes into thick copper jackets, as shown inFig. 19. The hole in the copper was made undersized

Figure 12. Experiment for determination of power radiated bythe arc Figure 13. The ORNL long solenoid apparatus

INTENSE GASEOUS DISCHARGES 311

MAGNETICFIELD

CUP ANODE

CATHODE ASSEMBLY

Figure 14. High-current vacuum carbon arc electrode assemblies Figure 16. An energetic deuterium arc with a cup anode

to the extent that it was necessary to heat the copperto several hundred degrees in order to insert the tung-sten. During operation the copper was water-cooled;the differential expansion, which occurred when thetungsten was heated by the arc, produced a verytight fit and resulted in good heat transfer.

As might be expected from the cross-field emission,the deuterium arc was found to require a much higherapplied voltage than the carbon arc. The arc voltagewas found to vary inversely with the gas input. Theexact part played by the hollow anode is hard tounderstand, since the highest efficiency for molecularion dissociation was achieved when all of the support-ing gas flow was into the cathode. This might havebeen due to contamination (tungsten) from the anode.

Figure 15. Off-axis arc in DCX. About three feet of arc are visible

This last observation led to experiments with arcsrunning between hollow cathodes and solid anodes.For these experiments, anode heating was minimizedby the use of a rotating anode assembly, which effec-tively spread the electron bombardment over a largersurface area. Both carbon and copper anodes havebeen used, the former arranged to permit anode gasfeed. These experiments are still in progress andwill be described in a future report.

The need to reduce anode heating with its conse-quent emission of impurities led to energetic arcexperiments with a reflex (Phillips Ion Gauge, orPIG) geometry. In one arrangement, hollow cathodeswere used at each end of a long tubular anode. Aninsulated plate substituted for one of the cathodeswas found to charge negatively with respect to the

remaining cathode to a voltage as great as the anode-to-cathode voltage. This phenomenon is believedto be closely related to the "Mode II" arc behaviorstudied at ORNL.11 The arrangement is shown inFig. 20.

This arc has even higher resistance than that shownin Fig. 16. It has been operated at potentials up toone kilo volt. A power supply which will provide2.5 kv is now under construction.

Formerly, arcs with large cross-sectional areaswere almost impossible to obtain because of currentlimitations and uneven emission. Even in smallarcs, emission tends to run from cathode spots. Thecross-field emission described here provides a largesurface area from which electrons may enter thesystem. The total current required to run the arcis considerably reduced since the electron stream isconfined to a thin cylindrical sheet.

When the pressure is lowered below some criticalvalue, adequate space-charge neutralization is pro-vided inside the cathode but not in the main volume.In one test, a quarter-inch diameter hollow streamof electrons was "blown up" to more than 3inches in diameter in a distance of 5 inches. Themechanism for this spreading is not entirely under-stood, inasmuch as the arc was operating in a 3000gauss magnetic field, but it does seem clear that it

TUNGSTEN CATHODE

ELECTRON PATH

Figure 17. Possible mechanism for arc operation from insidea hollow cathode

312 SESSION A-6 P/1790 J. S. LUCE

Figure 18. Hollow tungsten anode burned through by anenergetic deuterium arc

was associated with imperfect space-charge neutrali-zation. The oscillatory electric fields produced whenstreaming electrons are present in a plasma often giverise to organized ion motion of considerable energy.This effect has been observed by several investigators.11

The high-voltage, high-resistance characteristics ofthe energetic PIG arc should enhance this process.

One of the first projects undertaken by the ORNLThermonuclear group was the development of a sys-tem whereby a high potential (50 kv) could be sus-tained between a solid gas arc and a surrounding hollowgas arc running coaxially, or between an arc and acoaxial cylinder.12 It is notable that in the high-voltage energetic PIG arc, there has been a return toa very similar geometry.

It is interesting to extrapolate to the future some-what and to speculate on possible applications of thisnew class of arcs. Since fairly high electron and ionenergies appear to be involved (of the order of thatwhich would be imparted by the arc potential), anattractive proposition would be to pinch off a sectionof arc between two magnetic mirrors and compress itso as to multiply the energy and perhaps therebyachieve ignition of a thermonuclear plasma. Thedesign of an experiment to test this concept is nowunder way. The energetic PIG arc concept is appli-cable also to toroidal geometry, with or withoutcompression (see Fig. 21). The most exciting possi-bility presented by the hollow energetic arcs, however,is that of surrounding a reacting plasma with an arcwall. Should this concept be brought to its ideal

Figure 19. Electrode and holder for energetic gas arc

fruition, the sputtering problem would be eliminated,because a thermonuclear plasma contained inside ahollow arc would be protected from cold neutral atomswhich would normally flood in from the walls of thevessel.

The applications will almost certainly not be limitedto the thermonuclear field. For instance, P. B. Moon13

has considered the breakup of H2+, using a gas target,as a method of injection into proton synchrotrons.The energetic arcs, capable of high dissociation effi-ciency, sharply localized dissociating action, and low-pressure operation would be admirably suited to sucha use. Heavy-ion accelerators could also make pro-fitable use of such devices for stripping successiveelectrons from the ions to be accelerated.

CONCLUSION

The ORNL concept of producing a high-tempera-ture plasma by the dissociation-trapping of energeticions was brought within the bounds of feasibility bythe development of high-current molecular ion sourcesand by the discovery of the energetic carbon arc.An experiment, DCX, is now under way to determine

ELECTRON REFLECTOR

GAS FEED\

MAGNETIC FIELD

-OSCILLATING VOLUME

7CATHODE

ASSEMBLY

CYLINDRICAL ANODE

SHIELDS

ELECTRON REFLECTOR

Figure 20. Energetic PIG arc experiment

INTENSE GASEOUS DISCHARGES 313

OSCILLATING

VOLUME

ELECTRON REFLECTOR

MAGNETIC FIELD

Figure 21. Energetic PIG arc in toroidal geometry

whether this trapping system is, in fact, capable ofproducing a thermonuclear plasma. None of theexperimental results to date has indicated the pre-sence of insurmountable obstacles to the achievementof this goal. However, if such an obstacle shouldbe found to be presented by the energetic carbon arc,there is now available for substitution the energeticdeuterium arc. In addition to the obvious advantagesof an arc which does not contain heavy, partially-stripped ions, the deuterium arc has been shown to becapable of a flexibility of operation which shouldpermit a considerable measure of control over itseffects on the plasma. In addition, the deuteriumarc appears to be susceptible to other, and perhaps

of amen-

even more interesting, applications than thatdissociating agent. A few of these have beentioned above.

The study of energetic arcs is clearly in its infancy.Some of these arcs may even now constitute plasmaswhich have extremely high temperatures.

APPENDIX

Development of Ion Sources yielding Ampere

Currents of Atomic or Molecular Deuterium Ions

The approach to the establishment of an energeticdeuterium plasma by molecular dissociation and itspromise in the thermonuclear program pre-supposethe existence of intense currents of D2+ ions for injec-tion purposes. In 1955, certain modifications ofcalutron hot-cathode sources were explored, andoperating conditions in hydrogen were sought, underwhich the molecular ion fraction of the output currentcould be maximized. It was soon found that, in suchan ion source, the output can be controlled at will soas to give an ion yield that is either predominantlyatomic or predominantly molecular, and that themost important determining factor is simply thepressure. The atomic ions are favored at high pres-sures (red color of discharge) and the molecular ions arefavored at low pressures (blue color). This behaviorcan be qualitatively explained on the basis of adifference in the mean energies of the electrons in theion source plasma. The minimum electron energies forthe various processes are as follows:

1. H2 + e->H + H

2. H+e->H++e

3. H2 + e->H + +e

11.4 ev

13.5 ev

15.6 ev

MAGNETIC FIELD

ION BEAMS.

ACCELERATING GRID

CATHODESHIELDS

HIGH VOLTAGESHIELD — *

CATHODE

¡I || | | jj | | || |i Д Y SI j\ V/\ ¡\ \//\ j\ V/\ j\ Y/i j\ У/1 j\ Y/I ¡\ V/l j\ VA j\ \/A j\ Y A j\ VA /*

/ \J \J \j \J \J U\J \J \J \j \J \Á JL

WATER-COOLEDTANTALUM

GRID

-ELECTRON REFLECTOR

-DEFINING SLOT •ELECTRON STREAM

ARC CHAMBER

У////////////////77Л Y///////////////////////////

i HIGH VOLTAGESHIELD

Figure 22. Grid ion source

314 SESSION A-6 P/1790 J. S. LUCE

from which it would appear that if multiple collisionsdegrade the energy of the electrons under conditionsof high pressure, a combination of the first two pro-cesses might dominate over the third process Acorollary of such an explanation would be that process3 is primarily responsible for the production of mole-cular ions rather than a succession of processes 1 and2 followed by recombination The physics of thesituation has been discussed m greater detail by Masseyand Burhop 1 4

The use of metal walls m an ion source to enhancethe output of the molecular species is well known,in the adaptation of the calutron source for this pur-pose, water-cooled tantalum has been found to besuitable, and indeed it is especially important in theexit slit of the device Figure 22 shows a grid"ion source m its adaptation for work with deuteriumThe over-all length of the device is about 8 inchesIn the output, H+ dominates at high pressure andcurrent, H3+ at intermediate pressure and current,and H2+ at low pressure and current The pressuresetting suffices to allow one to choose between thefollowing operating ranges

Voltage aCurrenta

Favorable to H+Favorable to H3+Favorable to H2+

80 v100150

15 amp105

a Measured between cathode and arc chamber

An output of 2 5 amp of ions with an H2+ contentof 85% has been achieved from a grid source of thiskind having an arc area of three square inchesUnder appropriately altered conditions, five amperesof ions have been extracted from one square inch ofarc area, of which 95% were atomic

ACKNOWLEDGEMENTS

The writer was assisted by members of the Thermo-nuclear Experimental Division of the Oak RidgeNational Laboratory and several consultants Powermeasurements were made by С W Blue, and spec-trographic measurements were made by the Spectro-

scopy Research Laboratory Construction work wasdone by a group led by O D Matlock Experimentaluse of facilities was coordinated by V J MeeceEditorial assistance was given by С E Bettis, R JMackin, Jr , Mozelle Rankin and A H Snell Theassistance and encouragement of С F Barnett,P R Bell, E D Shipley and many others, whocontributed to the program, are gratefully acknow-ledged

REFERENCES

I R K Wakerling and A Guthrie Electro MagneticSeparation of Isotopes in Commercial Quantities, U SAtomic Energy Comm Technical Information ServiceOak Ridge Tennessee

2 J S Luce Proposed Sherwood Experiment ORNL CF-55 4 73 (1955)

3 A Simon and T M Rankin Some Properties of a SteadyState High Energy Injection Device (DCX) ORNL 2354(August 1957)

4 P R Bell The Oak Ridge Thermonuclear Experiment,P/344 this Volume, these Proceedings

5 A Simon and R V Neidigh Diffusion of Ions in aPlasma Across a Magnetic Field Phys Rev 98 317(1955)

6 С F Barnett The Dissociation of Diatomic HydrogenIons P/1789 Vol 32 these Proceedings

7 J R McNally Jr private communication8 W В Ard Jr and H O Eason private communication9 C E Normand, О С Yonts and С W Blue Controlled

Thermonuclear Reactions A Conference held at BerkeleyCalifornia p 462 TID 7536 Part 2 (22-23 February1957)

10 R J Mackin Jr Performance of the Carbon Arc as anIon Pump ORNL CF 52 2 86 (3 March 1958) Proceedmgs of Conference on Thermonuclear ReactionsWashington D С (3 5 February 1958)

11 R V Neidigh The Effect of an Applied Pressure Gradienton a Magnetically Colhmated Arc P/2396 this Volumethese Proceedings

12 J S Luce Conference on Thermonuclear Reactionsp 73 UCRL Berkeley California WASH 289 (7-9February 1955)

13 P B Moon С em Symposium on High Energy Accelerators(June 1956)

14 H S W Massey and E H S Burhop Electronic andIonic Impact Phenomena p 247 Oxford UniversityPress (1956)