acceleration of plasma into vacuum

P/355 USA

Acceleration of Plasma into Vacuum

By John Marshall *

The first part of this paper is a discussion of themagnetic acceleration of plasma. The second partcontains a description of some experiments whichhave been performed.

Work similar in some respects to that describedhere has been reported by Thonemann, Cowling andDavenport,1 by Bostick 2 and by Kolb.3' 4 Thone-mann and his collaborators studied the effect on aplasma of a radio-frequency traveling wave in acapacitor-loaded helical delay line. They found anet negative electric current to flow in the directionof the traveling magnetic waves, presumably becausethe electrons in the plasma were dragged along moreefficiently than were the ions. Bostick has developeda source capable of propelling puñs of plasma intovacuum at speeds of the order of 107 cm/sec. Hisplasma gun employs a vacuum spark between gas-loaded titanium electrodes. The plasma, which con-sists of a mixture of ions of titanium and the gas withwhich the source is loaded, is driven away from thesource by the magnetic field of the current flowingthrough it. Kolb has used programmed externalmagnetic fields to accelerate shock waves. Since hisshock waves move through a gas which is initially atlow temperature, dissipât i ve processes are involvedand the result is the heating of plasma rather than itsacceleration to high velocities.

In the work reported here the intention is :1. To produce a burst of gas in vacuo.2. To ionize the gas and heat it to such an extent

that it becomes a good electrical conductor.3. To accelerate the plasma thus produced into

vacuum by the use of external time-varyingmagnetic fields.

MAGNETIC ACCELERATION OF PLASMA

Let us consider a cylindrical tube surrounded by asolenoid, carrying current so as to provide an axialmagnetic field, and containing a hot plasma. Theplasma might form a cylindrical body down the axisof the tube excluding by its diamagnetism and surfaceeddy currents, the magnetic field produced by the

* Los Alamos Scientific Laboratory, University of California,Los Alamos, New Mexico.

solenoid. Such a situation actually can exist justafter the passage of a sufficiently strong shock wavein an axial direction along the tube.5 Now let usimagine a wave of increased axial magnetic field tomove along the tube. The magnetic lines of force, asthe wave approaches a given position, are pushedinward toward the axis of the system, and expand totheir original position again as the wave passes. Thelines of force ahead of the magnetic wave diverge soas to produce a forward component of force on theplasma ahead. Inside the magnetic wave the mag-netic pressure is high and, in the case of equilibrium,so would be the plasma pressure. If the magneticfield is high enough it will exert enough pressure onthe plasma trapped at the center to drive the plasmaforward ahead of the field, and thus to carry along aconsiderable mass of plasma at the speed of the magneticwave. The situation is superficially quite similar to peri-stalsis in the gastro-intestinal tract of an animal. In thesame way that a contractile wave along the esophagusof a giraffe can propel a gulp of water from groundlevel up to its stomach, so can a wave of increasedmagnetic field propel a mass of plasma along the axisof our tube.

As an example of the pressure which can be exertedon a plasma by a magnetic field let us consider a fieldof 30,000 gauss. This is approximately the strengthof the magnetic piston employed in some of theapparatus to be described below. B2/8n for this fieldstrength is 3.58 X 107 dynes/cm2 or about 35.4 atmo-spheres. Ordinarily one might expect to acceleraterather small masses of plasma so that a 35-atmospherepressure, if exerted over a suitable distance, shouldlead to respectable velocities.

The most convenient method of producing a movingmagnetic piston appears to be the discharge of acapacitor bank into one end of an artificial delay linecomposed of a solenoid loaded with capacitors atintervals along its length. The velocity of such a lineis given by v — 1/V(Z/C') and its characteristicimpedance by Zo = V(L'C'). Here the velocity willbe in cm/sec and the impedance in ohms if L' and С,the series inductance and shunt capacitance per unitlength, are expressed respectively in henries/cm andfarads/cm. It is quite plain that the parameters L'and С can vary as a function of distance along theline, so as to produce an accelerating magnetic piston,exerting a constant force over a considerable distance.

341

342 SESSION A-6 P/355 J. MARSHALL

It is not inconceivable that really high plasma velo-cities could thus be achieved.

Let us again use our example of a 30,000-gausspiston, and compute the velocity which could, inprinciple, be given to a lump of plasma of reasonablemass. Assume that the piston velocity is so adjustedas to exert the full pressure of a 30,000-gauss fieldagainst the plasma over a distance of 50 cm. Assumethat the plasma to be driven corresponds to 0.02 cm3

of deuterium at normal temperature and pressure forevery square centimeter of area. This would amountto about 3.5 [jig of plasma per cm2. These particularnumbers are chosen because they correspond to anamount of plasma which has been driven in this work,although not under these ideal conditions. Remem-bering that the velocity attained under constantacceleration a through a distance s is given by v =V(2 as), and substituting an acceleration of 3.6 X 107

(dynes/cm2)/3.5 X 10~6 (g/cm2) = 1013 cm/sec2, wecompute a velocity of approximately 3 X 107 cm/sec.The kinetic energy of the plasma j et would be approxi-mately 180 joule/cm2, and its momentum about100 g cm/sec per cm2 of area.

The system of a magnetic piston driving a lump ofplasma has phase stability similar to that which anuclear particle accelerator, such as a synchrotronor a linac, must have in order to operate successfully.The plasma rides on the front of a wave as does asurf board. If the plasma at some point should bemoving more slowly than the wave, it will drift back-ward relative to it and find itself riding higher wherethere is more force available to drive it. If it shouldmove faster than the wave, it will get ahead; the forcewill decrease and the wave will catch up again. Thusthe speed of the piston does not have to be preciselyadjusted. The plasma will tend to adjust its speedand to stay ahead of the piston. However if theaccelerating process relies on phase stability, the fullpressure of the magnetic piston cannot be used. Theplasma must necessarily ride part way up the wave.If it should pass over the crest and find itself on theback of the wave, it will be decelerated somewhat andleft behind.

Somewhat more latitude for error in the matchingof piston and plasma speed can be provided by the useof a piston with a flat top. This can reduce the dangerof the slipping of the plasma on to the back slope ofthe wave. The resistance of the lump of plasma to thepenetration of magnetic field implies that if thepiston tends to catch up on the plasma, the plasmawill make a hole in the magnetic field and thus auto-matically adjust its rear boundary to move at themaximum possible acceleration. A flat-top piston,of course, requires more energy than does a peakedone of equal strength.

If the back end of the plasma mass rides imbeddedin the front of the piston, however, it will be com-pressed sideways by the magnetic pressure in thepiston and its cross-sectional area will be reduced, asfor instance has been observed in the Los Alamosexperiments on shock channeling.5 This will tendto limit the amount of plasma which can be driven.

With such an arrangement, really appreciable frac-tions of the energy of a magnetic piston might betransferred to the jet of plasma. Of course as energyis taken out of the piston, its strength will decreaseand, to produce the effect described, means would haveto be found to feed more energy into the piston as itdrives the plasma.

From the point of view of efficiency of energytransfer, the traveling wave magnetic piston is nearlyideal. The magnetic wave carries all the energy ofthe driving capacitor bank (assuming negligibleresistive losses) and in principle should be able totransfer a large fraction of it to the plasma. Thetraveling wave, however, is not the only method ofproducing such a piston. Another method would beto have a large number of separate coils arranged alongthe acceleration system and excited from separatecapacitor banks in a programmed manner. Also itwould be possible in principle to use a long coil,tapered in such a way as to produce a magnetic fieldlarger at one end than at the other. If the currentthrough the coil is raised slowly the effect is, to someextent, that of a moving piston. In both of thesealternate methods, the energy required to produce themagnetic field is much larger than in the travelingwave method.

In order to drive plasma electromagnetically it mustbe highly ionized and heated to a high enough temper-ature so that there will not be serious penetration offield during the driving process. What appears to bea satisfactory method of heating and ionizing is thepassage of an intense shock wave along the tube justbefore the piston circuit is energized. A convenientmethod of generating such a shock wave is to dischargea capacitor through a one-turn coil looped around theacceleration tube.6 The shock wave, as it passesthrough the gas, ionizes it and incidentally imparts toit an initial velocity in the direction in which it is tobe accelerated.

Quick-acting Valve

A mechanical valve has been developed for pro-ducing the burst of gas at the input end of the plasmaacceleration system (Fig. 1 ). Basically what the valvehas to do is to admit something of the order of 1 cm3

of gas to the system in a time of the order of 100 [xsec.In 100 [jisec, deuterium, which is the gas used in mostof the experiments described here, moves approxi-mately 10 cm at the molecular velocities character-istic of room temperature. Thus it becomes possibleto have a pressure of deuterium gas of about one milli-meter of mercury in the input end of the systemwhile the rest of the system remains temporarily athigh vacuum. Under these conditions the plasmaproduced by the ionization of the gas can be accele-rated unimpeded through the system.

The valve is very simple and diners very little froman ordinary mechanical valve such as might be usedas a water faucet. The main differences are that it isheld shut by a spring, that it uses a teflon (polyper-fluoroethylene) gasket, and that it is opened by ahammer blow. The spring tension on the valve

ACCELERATION OF PLASMA INTO VACUUM 343

GAS INLET

SOCKET DRIVE SET SCREW^fOR SLOW LEAK

TEFLON"GASKET

SPRING

80 cm

ANVIL BLOCK

355.1

Figure 1 . Quick-act ing valve. O n e of t h e many possiblear rangements f o r such a valve. Gas f lowing th rough th e leakinto t h e small p lenum in the valve stopper can be admi t ted t o thesystem by a l lowing the hammer to fall and str ike the anvil block

stopper is used so that the hammer blow can beabsorbed non-destructively by an elastic system.Teflon is used as a gasket material because of its non-sticky properties. In such a valve it is essential thatopening can be achieved with a very small thoughsudden motion of the stopper. Any gasket stickinesswould tend to increase the stopper motion necessaryto open the valve, and almost certainly would reducethe reproducibility of performance from one shot tothe next. For this reason teflon was chosen andworked very well in the first model. No other materialhas been tried, and no experimental information hasbeen obtained that it is actually better than any othermaterial. A hammer blow was chosen to open thevalve simply because it provides a sudden motionthat is easily controlled in strength up to and beyondthe elastic limits of the materials used. The originalhammer consisted simply of a brass cylinder slidingunder the action of gravity along a length of drill rod.The cylinder can be dropped from a variable heightso as to strike, with an adjustable impulse, against ananvil block attached to the end of the rod. Theimpulsive wave produced moves back along the rodat the speed of sound in steel (4500 m/sec) and servesto open the valve. With this arrangement all otherfunctions of the apparatus are timed to occur subse-quent to the hammer blow and are controlled by an

electronic delay unit started by an electric contactmade by the hammer in striking the anvil.

In some experiments it is inconvenient to have allfunctions of the apparatus timed to start after thehammer blow, and it becomes expedient to use somesubstitute for the falling weight which can be timedprecisely after other events. For this purpose anelectromagnetic hammer has been built (Fig. 2). Itconsists of a small coil in close proximity to an alu-minum anvil block. A capacitor is dischargedthrough the coil by means of an ignitrón, and themagnetic pressure of the field generated by the lowinductance coil produces an impulse similar to thatproduced by a hammer blow. The design of the coilis such that most of its inductance is due to the highconcentration of magnetic field where it is squeezedbetween the coil and the anvil block. Since this gapis originally very small, the inductance is also small atthe beginning of the current pulse and the completeimpulse can be delivered in less than 50 (¿sec.

Three methods are readily available for regulatingthe amount of gas admitted by the valve. They arecontrol of (1) the volume emptied by the valve (thisimplies that the valve is opened far enough to emptythe volume involved), (2) the pressure of the gasbehind the valve and (3) the impulse delivered to thevalve stopper and, therefore, the degree to which itopens. The first method, which involves an adjust-able plenum just behind the stopper, is an attractiveone in that the valve may bounce and open more thanonce without affecting the amount of gas admitted.For most purposes, however, a combination of thesecond and third methods is easier and appears to becapable of giving highly reproducible results from shotto shot. In most cases, it turns out to be unimportant

BRASS CUP

ALUMINUMANVIL BLOCK

STEEL ROD

355.2

Figure 2. Electromagnetic hammer. Spiral coil wound ofcopper r ibbon is potted solidly in epoxy resin so as not to bedistorted by its own blows. Magnetic lines of force are crowdedby skin effect so as t o pass through 0.3 mm gap between coilface and anvil block. H a m m e r blow effect in steel rod is producedby discharging 2-3 kv capacitor through coil by means of an

ignitrón


355.3

Figure 3. Typical construction of piston coil. In this case thesolenoid is tapered in pitch so as to produce a line with increas-ing velocity. Capacitors are connected at the pairs of tabs be-tween turns by coaxial cable paralleled in groups of four. Groundreturn conductor is held close to axial transition sections ofsolenoid so as to cancel asymmetrical field and reduce inductance.Coil is insulated with teflon and potted in glass-reinforced epoxy

resin to give it strength to withstand magnetic forces

whether or not the valve opens more than once.Everything of interest for one shot is over long beforethe valve has a chance to bounce open again.

Plasma Jets from Shock Wave Blow-off

It has been mentioned above that an intense shockwave provides a convenient method of preheating andionizing the gas. Actually the passage of a shockwave through a gas which decreases in density, so thatthe shock runs into vacuum, is in itself a method ofintroducing a jet of plasma into vacuum. Presumablythe plasma jet consists of the particles more or lesswithin the last mean free path of the gas. Since theseparticles have no more particles in front with which tocollide, they simply keep on going. If the shock waveand the blow-off plasma jet are arranged to moveparallel to a magnetic field, along the axis of a current-carrying solenoid for instance, any sideways motionof the ions is averaged out and the jet tends to movealong the lines of force. Neutral gas, however, whichmay accompany the jet or be produced therein byrecombination, is unaffected by the magnetic field andconsequently is lost through the sides, leaving the jetfully ionized.

The apparatus used in studying this phenomenon(Fig. 4) has been a glass tube approximately 125-cmlong connected at one end to a vacuum system andat the other to the quick-acting valve. Over most ofits length, the tube was of approximately 10-cm outsidediameter and was wound with a solenoid which couldbe connected to a capacitor bank through an ignitrónso as to produce an axial magnetic field. At each endthe tube was reduced through conical transition sec-tions to a 5-cm diameter, and a two-turn shock drivingcoil was located about halfway along the transitionat the end near the quick-acting valve. The shockdriving coil could be connected by means of a sparkgap to a 22.5 [xf capacitor bank, charged to 20 kv.The timing of this spark gap and of the ignitrón onthe axial field (Bz) bank could be accurately controlledby a variable electronic delay circuit triggered by thehammer blow of the quick-acting valve.

The shock wave is launched from the shock drivingcoil described above by magnetic interaction of theprimary current in the coil and secondary currentsinduced in the low pressure gas inside it. Experiment-ally it is observed that the gas will break down onlyat times at which the magnetic field inside the coilis passing through zero. At such a time the gas can

-PREIONIZING SHOCK COIL

-TRAVELING WAVE COILBz WINDING

TO PUMPINGSYSTEM

-—PHOTOMULTIPLIER[J TELESCOPE

GENERAL ARRANGEMENT

O.I Л TERMINATING RESISTOR

7.5uf CAPACITORSSPARK GAP

23/¿f, 20 Kv

PLASMA ACCELERATION SYSTEM SCHEMATIC

FALLING HAMMER STRIKES ANVIL BLOCKOF FAST ACTING VALVE

-IGNITRÓN EXCITES Bz WINDING

HAMMER SIGNAL ARRIVES AT VALVE, TRAVELING"AT SPEED OF SOUND IN STEEL VALVE STEM*,

VALVE OPENS

IGNITRÓN EXCITES"PREIONIZING SHOCK COIL

LSPARK GAP EXCITESTRAVELING WAVE COIL

О

355.4

200 300 400¿i SEC

TYPICAL TIMING SEQUENCE

Figure 4. Arrangement of equipment for accelerating plasmawith a moving magnetic piston. The arrangement depicted herehas been superseded by one employing much higher voltages onthe shock driving coil. Note that here the magnetic piston isproduced in a line which is tapered by varying the intervals at

which capacitors are connected


be ionized by the back emf produced by the rapidlychanging flux inside the coil. A ring gas currentresults, of such magnitude and direction as to shieldthe regions inside the discharge from the field of thecoil. The magnetic pressure external to the dischargedrives the plasma toward the axis of the system,picking up gas ahead of it, pinches it off and squirtsit out axially in both directions away from the coil.The effect is as if there had been an explosion, andshock waves are produced traveling along the tube inboth directions.

The shock driving coil and the capacitor bank whichdrives it constitute a resonant system which rings soas to produce a train of damped oscillations. Gener-ally a new shock wave is generated each time thecurrent in the coil passes through zero ; at least thisis true so long as there is gas present to be driven.At very high voltages applied to the coil, the gasappears to be swept out completely by the magneticfield and only one shock wave is generated, but withthe arrangement described here, a long series of shockwaves is produced. In order to make the apparatusgenerate a shock wave on the first half cycle of thetrain of oscillations, it is necessary to pre-ionize the gas.This is done with an electrode brought into the tubein the vicinity of the shock driving coil and connectedto a small capacitor through a current limitingresistor, the capacitor being charged to a potentialof 10 kv relative to ground. Without pre-ionizationthe first shock wave appears after one half cycle andon a number of successive half cycles thereafter.

The propagation of the shock wave and of the plasmajet which it produces is conveniently observed byphotomultiplier telescopes connected to a multibeamoscilloscope. Each photomultiplier is arranged witha lens and a slit so as to be exposed to light from justone point along the axis of the tube. With suchan arrangement, it is observed that the shock waveaccelerates as it moves into gas of lower density andthat it becomes less steep. Presumably the sharpnessof the shock front is related to the mean free path ofthe particles under the conditions prevailing at thatpoint and with that particular shock intensity. Whenthe shock runs out into vacuum, the jet that it pro-duces begins to separate into slow and fast plasma;the fast particles moving out in front, leaving the slowparticles behind. In the absence of particle-particlecollisions, this would have the effect of reducing thetemperature in the forward and backward degree offreedom, while leaving the sideways degrees of freedomunaffected.

Unfortunately very little data have been takenwhich can contribute to a quantitative study of thephenomenon. This is partly the result of pressingonward toward the problem of acceleration with amoving magnetic piston, and partly because of thenon-existence of instrumentation for the measurementof the rapidly changing gas densities and velocitiesinvolved.

Systems identical with that described in this sectionhave been adapted or are being adapted as plasmainjectors for thermonuclear machines. One has been

in operation for some months on Ixion, the crossedelectric and magnetic field spinning plasma machinesunder development at Los Alamos.7 Another is beinginstalled on a large linear pinch machine which is beingstudied currently.8

Magnetic Piston

As mentioned above, the most convenient methodof producing a moving magnetic piston appears to bethe use of an artificial delay line consisting of a coarsesolenoid loaded at intervals with capacitors. The linecurrently in use consists of a solenoid of 2.7-cm radius,with the turns spaced 3.8 cm. It is loaded with one7.5 (¿i capacitor for each turn. Such a line has aninductance of 2 x 10~8 henry/cm, a characteristicimpedance of 0.1 ohm and a phase velocity for lowfrequencies of 5 X 106 cm/sec. A low impedancesource of emf V applied suddenly to one end of thisline should produce a magnetic field rising in a stepto a value В = 3.30 V. This would mean that a20 kv capacitor discharged into the line would producea 6.6 X 104 gauss field falling off exponentially aftera sharp rise with a time constant given by т = RC,where R is the 0.1 ohm line impedance and С is thecapacity of the bank. Actually the rise time of thesignal is lengthened by the imperfections of the line,so that a 22.5 [xf capacitor bank produces only abouthalf the magnetic field in the wave front that onewould compute.

In order to simplify the effects observed and also tolengthen the life of capacitors and spark gap, the lineis terminated at its output end with a 0.1 ohm resistor.This reduces reflections from the end of the line andmakes it so that the current in the line can be describedas a single pulse moving at line velocity from one endto the other. Some difficulty was encountered indeveloping a resistor suitable for this service. It mustbe capable of carrying pulse currents of at least 105

amp, have low inductance and absorb the full energyof the capacitor bank each time the equipment ispulsed, 1500 joules for the present arrangement. Alsoit should be of thin enough material so that skin effectdoes not materially affect its resistance for the im-portant frequency components of the pulse. Asuccessful design consists of nichrome ribbon clampedbetween brass plates with teflon insulation. Theteflon appears to be able to stand the momentaryhigh temperatures and the brass plates provide goodmechanical support as well as a low inductance returncurrent path.

Each capacitor as well as the terminating resistoris connected to the line through 4 parallel 30 ohmcoaxial cables about 70 cm in length so as to achievelow inductance. The central wires of each group offour cables connect to one turn of the solenoid whilethe outer conductors connect to a ground returnstrap. The return strap is as close as is practical tothe connection in the axial direction between solenoidturns so as to reduce inductance and avoid distortionof the field. Insulation between turns and from theturns to the ground return strap is of teflon and thewhole coil is potted in epoxy resin with glass tape


reinforcement so that it will be strong enough towithstand the electromagnetic forces. The coil in useat present has 10 turns and has a total length of about38 cm.

Shock Driving Circuit

The shock wave which pre-ionizes and heats the gasto be driven by the magnetic piston is, in the presentarrangement, produced by a one turn coil connectedthrough a spark gap to a 0.88 [iî capacitor which hasbeen operated as high as 60 kv. The inductance of thecoil is approximately 0.065 [xh while that of the entirecircuit is computed from the ringing frequency to be0.295 [xh. Thus only 22% of the voltage applied tothe capacitor appears across the coil. A good fractionof the parasitic inductance of the circuit is in thespark gap, which is of the four electrode type and isoperated in air at atmospheric pressure. The sparkgap is triggered from a 0.1 [xf capacitor through 3ignitrons connected in series so as to be capable ofwithstanding 50 kv.

PERFORMANCE

The apparatus in its present form appears to becapable of driving appreciable masses of plasma at thespeed of the magnetic piston. The speed is measuredby photomultiplier telescopes focused at intervalsalong the axis of the drift tube into which the pistondrives the plasma. An accurate measure of speed issomewhat difficult using this technique because of thediffuse nature of the plasma jet. However, grossfeatures of the luminosity vs. time curve can be fol-lowed from telescope to telescope (Fig. 5), and itappears that within 20% or so the plasma moves withpiston speed.

The momentum acquired by the plasma has beenobserved by two methods. One of them employs acapacitor microphone and the other a ballistic pendu-lum. The capacitor microphone is placed on the axis

355.5

Figure 5. 4-beam oscilloscope record of data from typicalshot with magnetic piston. This shot used 9 kv two-turnshock coil preheating with piston line tapered by varying inter-val between capacitor connection positions. Trace 1 shows Bz

beyond end of guide field solenoid. Traces 2, 3 and 4 are photo-multiplier telescope records at positions 53.7, 77.4 and 97.3 cmfrom the shock driving coil. The small pips are 10 [isec timemarkers. Wi th each trace is a null trace, obtained in the sameway as the traces carrying the signals, but with no gas admitted

to the system

355.6

Time (msec) ' ' ^

Figure 6. Tracing of oscilloscope record of microphone signal.Upper trace shows early ringing signal due to plasma impact on

diaphragm. Lower trace shows neutral gas signal alone

of the system on a probe extending well into thesolenoid which provides the axial magnetic field in thedrift space. A grounded aluminum foil, stretched infront of a high impedance pick-up electrode chargedto 300 v, is exposed to the plasma and the electrodevoltage is registered on an oscilloscope by means of acathode follower. Since the ringing period of themicrophone foil is long compared to the blow given itby the plasma, its behavior is really that of a ballisticpendulum, the initial ringing velocity being propor-tional to the impulse. However, its period is shortcompared to the time for un-ionized gas to reach itfrom the quick acting valve, and so it can separatethe two impulses.

With the capacitor microphone (Fig. 6) it is observedthat the neutral gas admitted by the valve arrivesmore or less with a sharp front at the other end of thetube, and that it moves at a velocity of approximately1250 m/sec. The plasma is observable as a consider-ably larger impulse arriving long before the gas. Thecapacitor microphone is somewhat difficult to cali-brate accurately because of the complicated responseof its diaphragm to the plasma.

A ballistic pendulum method has been developed tomeasure momentum transfer from the plasma. Thependulum is suspended in vacuum with a bifilarsupport beyond the end of the Bz solenoid, but with adisc on an extension arm inside the solenoid toreceive the impulse from the plasma. The pendulumhas approximately a 1.6 second period and a mass of17.5 g. The disc which receives the impulse is 4 cm2

in area and is non-conducting so as not to be affectedby the electromagnetic forces of pulsed magneticfields.

Impulses of up to 80 g cm/sec have been delivered toeach unit area of the pendulum by a shot. The valvealone, with no electrical apparatus triggered, deliversabout 16 g cm/sec per unit area of the pendulum disc.This is very nearly equal to the momentum whichwould be carried if all of the 2.25 cm3 atmos of gasadmitted in one shot were to move at the 1250 m/secobserved gas front velocity. Actually, of course, allof the gas will not move at this speed but, on the otherhand, the molecules will recoil from the disc anddeliver probably twice their axial component ofmomentum to it.

Subtracting the effect of the neutral gas we are leftwith an impulse of 64 g cm/sec delivered by theplasma to each unit area of the disc. At the observed


speed of 5 X 106 cm/sec (which is also the speed ofthe magnetic piston) this would be the momentumcarried by about 13 [Jig of plasma or a little more than3 (jig/cm2. If the plasma jet were of the same cross-sectional area as the tube in which it moves, it wouldhave a total mass of about 50 ¡ig and would correspondto somewhat more than 0.25 cm3 of deuterium gas atnormal temperature and pressure.

It has been found necessary to use rather highvoltage on the preheating shock driving coil. Initialattempts with a low voltage system, like that used inthe shock wave blow-off experiments described above,were not very successful in driving large amounts ofplasma. A considerable amount of momentum isimparted to the plasma by the high-volt age shockalone and, if the present state of the apparatus werethe final one, there might be no point in the use ofthe magnetic piston. This report is written, however,at an intermediate stage in the development of thetechnique and it is hoped that it may be possible, inthe future, to accelerate plasma jets in this way toconsiderably higher velocities.

Preliminary observations have been made of aphenomenon which may be quite interesting. A

search coil was inserted into the end of the system ona movable probe so as to generate an emf proportionalto the time derivative of the z component of themagnetic field along the axis of the system. With theoutput from the search coil integrated electrically soas to display Bz on an oscilloscope, very complicatedbut highly reproducible patterns were obtained. Theeffect was strongest just outside the end of the Bzsolenoid, but was still observable approximately20 cm farther out. It appears to be due to some sortof magnetohydrodynamic oscillation induced whenthe plasma jet crosses the diverging lines of magneticforce at the end of the solenoid.

ACKNOWLEDGEMENTS

The writer is most grateful to William Basmannwho has co-operated closely in this work and who isresponsible for the design and construction of a largefraction of the apparatus. Thanks are also due toother members of the laboratory for many helpfuldiscussions, in particular Keith Boyer, W. C. Elmoreand James Tuck.

REFERENCES

1. P. C. Thonemann, W. T. Cowling and P. A. Davenport,Interaction of Traveling Magnetic Fields with IonizedGases, Letter m "Nature", 169, 34-5 (1952).

2 W. H. Bostick, Experimental Study of Ionized MatterProjected Across a Magnetic Field, Phys. Rev , 104, 292-99(1956).

3. A. C. Kolb, Production of High-energy Plasmas byMagnetically Driven Shock Waves, Phys. Rev , 107, 345-50(1957)

4 A. C. Kolb, Production of Strong Shock Waves in PulsedLongitudinal Magnetic Fields, Phys. Rev., 107, 1197-98(August 1957).

5. F. R Scott, W. P. Basmann, E. M. Little and D. ВThomson, Magnetic Channeling of a Strong Shock, m ThePlasma in a Magnetic Field, Stanford University Press,Palo Alto, Calif. (1958).

6. W. C. Elmore, E. M. Little and W. E. Qumn, Neutronsfrom Plasma Compressed by an Axial Magnetic Field(Scylla), P/356, Vol. 32, these Proceedings.

7. K. Boyer, J. E. Hammel, C. L. Longmire, D. Nagle, F. LRibe and W. B. Riesenfeld, Theoretical and ExperimentalDiscussion of Iмоп. A Possible Thermonuclear Device,P/2383, this Volume, these Proceedings.

8. D A. Baker, G. A Sawyer and T. F. Stratton, LowVoltage-Gradient Pinches in Metal-walled Systems, P/1025,Vol. 32, these Proceedings.

acceleration of plasma into vacuum

Documents