P/1182 France

High Intensity Discharges in Deuterium in aMetal Wall Torus

By J. Andreoletti, С Breton, J. Charon, P. Hubert, P. Jourdan and G. Vend ryes*

In two recent papers1- 2 we described the produc-tion and investigation of high intensity discharges in atoroidal chamber, the wans of which were made ofPyrex, externally covered by metal shielding.

It seemed worthwhile to us to repeat these experi-ments in a metal walled chamber as, in fact, theoreticalstudies3 have shown that the stabilizing role of such aconductive jacket, with respect to the over-all changesin the shape of the plasma column, are all the moreefficient as the radius of this envelope is closer to theinitial radius of the discharge. On the other hand, theexperiments of the Harwell team4 showed that apinched discharge can be initiated under suchconditions.

An important advantage of a metal walled chamberis found in its mechanical behavior, which raises farfewer problems than do other types of chambers.The metal chosen is aluminium, largely because of itslow atomic number.

Objectives

The research was undertaken, first, to investigateproblems having to do with very high current dis-charges (over 50,000 amp.) in an aluminium torus,particularly:

(«) initiation of discharge—longitudinal magneticfield Bz, pre-ionization, number of breaks, etc.;

(b) insertion of insulating sleeves and connectionsin the torus section, the behavior of which mustbe good mechanically, electrically and under avacuum.

(c) Electric coupling of primary circuit andplasma.

Secondly, it was desired to study the dischargesthemselves by various techniques, particularly:ultrarapid moving pictures, spectrography, magneticprobes, microwaves, neutron scintillators, etc.

The purpose of this paper is to describe the experi-mental setup in its present condition of completionand to outline some of the preliminary resultsobtained.

Original language: French.* Commissariat à l'Energie Atomique, Service de Neutron-

ique Expérimentale.

EXPERIMENTAL ARRANGEMENTS

The discharge chamber (which is called Equator I)is an aluminium torus with a mean major diameter of80 cm and a minor inside diameter of 7 cm. Thealuminium wall thickness is 4 mm.

In order to avoid any turn short-circuit effect, thetorus is made of two sections (Fig. 1), the two halvesconnected by two insulating sleeves or connections.Various lengths of insulating breaks, ranging from2 to 50 mm, have been used. The insulating materialsinvestigated to date are Pyrex and polytetrafluoro-ethylene. It is proposed to use quartz, ceramic (suchas mullite) and alumina-coated aluminium sleeves.

An evacuating tube having a cross section of 3 cmis connected to the vacuum equipment which includesan oil diffusion pump (separated from the chamberby a metal valve with an indium gasket and two liquidnitrogen traps) and a vane type pump. Deuteriumfilling is by the same channel as that used forpumping.

The periphery of the torus carries four samplingports, at which it is possible to insert magnetic orelectrical probes or arrange a number of suitabledevices for investigation, such as spectrographs orscintillators.

Around the torus are 36 coils designed to create thelongitudinal magnetic field, Вг. These coils are madeup of six layers of enamelled wire 2 mm in diameter.For 0.5 sec at the time of a discharge, they can beconnected to a dc rotary generator with a nominaloutput of 1 Mw. The current through the coils can beadjusted by means of a resistor bridge. The longitudi-nal magnetic field Вг, which may be created in thetorus chambers by this arrangement, can thus bevaried from 0 to 5000 gauss.

The discharge makes up the secondary of a trans-former, the primary of which is made up of a certainnumber of turns of inductive loops arranged about thetorus. Allowing for the fact that investigation of theinsulating sleeves requires frequent disassembly ofthe torus and of its electric circuits, the primary coils,until now, have been made up simply of circular turnsarranged in the plane of symmetry of the torus, asclose as possible to it as shown in Fig. 1. This primarycoil is made of copper wire, 2 mm in diameter, under a

100

DISCHARGES IN A METAL TORUS 101

Magnetic fielcLwindings V

Pumpingtube

.Primary inductivewinding

Testing ports

To the capacitors

Magnetic circuit-

Figure 1. Equator I, schematic

1182.1

polythene covering. Most discharges took place usingtwenty inductive turns of wire.

The primary-plasma coupling can be improved bymeans of magnetic circuits wound about both theprimary and the torus. In this fashion, coupling takesplace in part through the air and in part through theiron. The results given here were obtained with amakeshift circuit having a low maximum inductionand substantial leakage in the air gap. We propose toimprove the coupling in the near future by eliminatingthe air gap and using oriented grain sheet, 0.35 mmthick, with a maximum induction of 16,000 gauss.

The wiring diagram of the setup is shown in Fig. 2.The bank of condensers consists of 30 elements,of 2 jui each, which can be charged to a maximum dcvoltage of 50 kv, for a total of 75 kilojoules, maximumenergy, available for each discharge.

The condensers discharge in the primary of thetransformer through a sphere type spark gap. Dis-charge is controlled by means of a voltage pulseapplied to an intermediate plate located between thetwo spheres of the gap.

The primary current is oscillating. This createsmechanical stresses in the dielectric of the capacitors.In order to avoid their deteriorating too rapidly, it ispossible to short-circuit the primary by means of amercury ignitrón, when the voltage at the terminalsof the primary circuit first passes through zero. Suchan arrangement was used for all discharges in whichthe capacitor charging voltage exceeded 30 kv.

In order to facilitate the initiating of the dischargein the gas, the latter is made slightly conductive bymeans of a high-frequency discharge, which is appliedfor a few seconds before the pulse is supplied by the

Releasing

impulse

П82.2 vtrrr

Magnetic circuit

Г V-Plasma

\Primary i

capacitors. The high-frequency generator has a powerof 2 kw and a frequency of 27 Me. Both inductive andcapacitive high-frequency discharges have been tried.The most efficacious method appears to be to apply thehigh-frequency voltage between the two insulatedmetal shells which make up the aluminium torus. Thishf discharge is interrupted within 5 fisec after thedischarge pulse.

Figure 3 shows the general arrangement. A floorwas arranged over the capacitor bank to carry thetorus and its accessories.

The high-voltage units are enclosed in a wire-netting cage, preventing access to the experimentalsetup while tests are being run. The equipment neededfor the investigation of discharges is arranged aboutthe torus. Note, in particular, the spectrograph (inthe background) which is sighted on the dischargethrough one of the ports described above. The imageof the spectrum so obtained can be formed, by meansof an optical bench, on a horizontal slit. A timeanalysis of the light passing through this slit canbe made by means of a camera with a horizontal-axisrotary mirror.

Primary inductance

Figure 2. Circuit schematic

Figure 3. Equator I

An ultra-rapid Beckmann camera which can give25 photographs at 1-jusec intervals (close-up, on theleft) can also give 25 motion-picture exposures of theplasma column during discharge. We are currentlyworking on the development of liquid and plasticscintillators, as well as on BF3 counters, for thedetection of the fusion neutrons which may beemitted during the discharges.

A thin section of the aluminium torus serves as awindow for the detection of X-rays by means of anionization chamber or an Nal scintillât or.

The capacitor charging devices and the observa-tion oscillograph are outside the high-voltage en-closure. Once the capacitors have been charged to therequired voltage, all operations are automaticallysequenced by means of a camshaft and electronicrelays. A coincidence device, in particular, providesmeans of bringing the discharge about when themirrors of the two rotary-mirror cameras are in asuitable position.

102 SESSION A-7 P/1182 J. ANDREOLETTI et al.

ПШМ

Figure 4. Vacuum system

Figure 4 shows the pumping equipment for thetorus, on the right, as well as the ignitrón tubedescribed above (close-up, on the left).

PRELIMINARY FINDINGS

The parameters, for the range investigated, are:Pseudo-periodElectric field, Ez

Pressure •

Initial magnetic field, BZQCurrent •

700-2000 /¿sec1-5 v/cm10-3-1 mm Hg of

deuterium0-4400 gauss0-104 ka

Gap Problems

Although this investigation is still in a very earlystage, results to date seem to indicate that:

1. discharge initiation is made easier by an increasein the number of gaps;

2. Pyrex sleeves are out of the question, becausethey break after a few discharges;

3. the polytetrafluoroethylene joints, in contrast,are particularly rugged (some of them, used formore than 150 discharges with currents of more than50,000 amp, showed no trace of deterioration whendismantled. The creeping or de-gassing problemshaving to do with these joints or connectors seemed tobe less important than might first have been thought) ;and

4. during the discharge, a heavy arc passes betweenthe edges of the gap; there is an actual transfer ofmetal from one edge to the other. Preliminary observa-tions seem to indicate that the current in the alumi-nium shell, during the discharge, may amount to afull 7 to 10% of the current in the gas. Eliminatingthis arc appears likely to be the main problem inproducing discharges in metal chambers.

Initiating the DischargeThe initiation of the discharge is greatly facilitated

by the presence of a longitudinal magnetic field, evena very weak one (# 50 gauss). For the initiation ofdischarges at low pressures (10~3 mm Hg) we are

studying, together with hf ionization, the possibilityof making the gas slightly conducting by means of astream of electrons from an electron gun.

Current and Electromotive ForceThe intensity of the discharge is measured by means

of a current transformer which surrounds the torus(Rogowsky belt). A single measurement gives both thederivative, dI2/dt, of the current, and the current, /2 ,itself—the current being deduced from dI2\dt by meansof an electronic integrator.

The electromotive force, U, is measured by takingthe voltage at the terminals of a circular coil tangentto the external jacket of the torus along its largemean diameter.

Figure 5 is a photographic reproduction of oscillo-grams for the current, /2 , its derivative, dI2/dt, andthe voltage, U. The horizontal calibrating lines show180 ka for the current and 600 v for the voltage. Thehorizontal time axis shows a calibrating signal every250 /xsec on all three oscillograms; in other words,3 msec for the whole of the sweep. Filling pressurehere is 1.2 x 10~2 mm Hg of deuterium, and the longi-tudinal magnetic field, at the start, is 120 gauss.

12 <H 2/ d t

Figure 5. Characteristics of the discharge

p = 1 .2x10-2 mm Hg, Bz = 120 gauss, Vo = 30 kv

Figure 6 shows oscillograms for / i (primary cur-rent), 12, dl2\dt and U with a total scanning time ofonly 1000 /xsec. The ignitrón, here, has short-circuitedthe battery of capacitors at the time when the voltagedropped to zero at the terminals of the primaryinductor. Current I\ continues to drop exponentiallyin the circuit made up of the ignitrón and a seriesresistance, 0.5 ohm, and the inductance of the pri-mary circuit. The calibration, on Ib is 8250 ampbetween the two horizontal lines and, for I2, 66,000amp per square. Calibrations on dI2\dt and U are thesame as in Fig. 5.

These oscillograms are remarkably reproducible,which seems to indicate good discharge stability.

dl/dt

Figure 6. Effects of the ignitrón

2.5 x Ю - 3 mm Hg, Bz = 390 gauss, Vo = 40 kv

DISCHARGES IN A METAL TORUS 103

inn

50

n

P o = l

12:[Kilo -amp.)

1000

.2X 10"2

- * —

—"• .

— - * •

— J .

2000

mm

' 2

R2

Hg.

n

3000

R2

4000

1182.7

(10- 3 П)

"в

B z o (в<"

10

5

ss)

5000

Figure 7

Influence of Magnetic Field Bz

Figures 7 and 8 show the variation in the maximumcurrent (/гтах) of the first half-cycle of the secondarycurrent as a function of the magnetic field, for fillingpressures of 1.2xlO~2 and 2.5xlO~3 mm Hg, andcharging voltages of 30 kv. The graphs also show theplasma resistance which corresponds to current

For the pressure of 2.5x 10~3 mm, it has not beenpossible to start the discharge at zero longitudinalmagnetic field.

At a pressure of 1.2xlO~3mm Hg, we note a suddenvariation in the current for a magnetic field variationbetween zero and 50 gauss. Some slight optimum ofcurrent 1% would appear to have been noted for amagnetic field of about 3000 gauss.

Effect of the Capacitor Charging Voltage

Figure 9 gives the maximum current of the firsthalf-cycle of 1% for a capacitor charging voltagebetween 25 and 40 kv, i.e., for a stored energy rangingfrom 19 to 48 kilojoules, and the plasma resistancecorresponding to this maximum current for the samevariation of the charging voltage.

It will be noted that /гтах increases substantiallyas the charging voltage, i.e., as the square root ofthe stored energy, and that the resistance, Rz, of theplasma decreases with the charging voltage. Theelectromotive force, U = Rrfz, corresponding to

varies rather little with the charging voltage.

Effect of Filling Pressure

In the presence of a strong magnetic field, Bz, themaximum current, I^max, of the first half-cycle variesbut little with the initial filling pressure.

The results shown in Fig. 10 were obtained with acharging voltage of 30 kv and a magnetic field, Bz, of1000 gauss. Other investigations are under way forinitial magnetic fields Bzo much weaker than those.

Discharge Spectra

Figure 11 shows some spectra of the discharge,photographed at a distance of approximately 40 cmfrom a gap area, for a pressure of 1.2xlO~2 mm Hgand a charging voltage of 30 kv, as well as for variousmagnetic fields.

1182.8

10

1000 2000 3000 4000Figure 8

100 J

50

B z o = 390 gauss

PO = 2.5X 10-3mmHg

1182.9

V(lcv)

10

10 20 30

Figure 9

40

100

50

B z o = 1000 gauss

I j (kilo-amp.) , 1

J»R2

Ik ' l

P

1182.10

2 (io-3n)'

o (mm Hg)

10

5

ю-4 10-2 1

Figure 10

There is a very marked widening, due to the Starkeffect, of the Д and у lines of the Balmer series ofdeuterium. Computations are under way to evaluatethe electronic density corresponding to these linespreads.. Beside each spectrum of Fig. 11 is shown the initialmagnetic field BZQ. The widening of line Ну wasmeasured with a microdensitometer. Figure 12 showsthe variation of this width, measured at half-height,for various magnetic fields Bz. It will be noted thatmaximal widening of the /Î and у lines is foundfor a magnetic field B2o between 300 and 400 gauss.

Temperatures

The present status of the research does not permit agood evaluation of temperature. However, it can beindicated that crude application of the formula

104 SESSION A-7 P/1182 J. ANDREOLETTI étal.

(gauss)

320

760

440

220120

FigureU

/2 = 4 NkT would give Г = 1.3xlO6 °K for thetests made with Bz = 0, V = 30 kv, p = 1.2xlO"2

mm Hg and I = 48.5 ka.Some measurements are now being planned for

the purpose of evaluating the discharge temperatureregardless of magnetic field Bz.

CONCLUSION

The preliminary studies already conducted withEquator I indicate that it is possible, in an aluminiumtorus, to achieve very high current discharges indeuterium. The appearance of the current and voltagecurves is not incompatible with good stability of thedischarge so obtained. The currents already recordeddo not invalidate the assumption that a high tempera-ture has been achieved.

The main problem to be solved is the arcing be-tween the edges of the insulation break. It is importantto see what fraction of the current is removed from thedischarge in this fashion, and to determine to whatextent the contamination this causes contributes tothe cooling of the plasma.

ACKNOWLEDGEMENTS

We particularly wish to thank Messieurs Giveletand Dupuy, of the Ateliers de Mécanique de Saclay,

ДА (А)

20 -

10

who greatly helped us in building the torus and theancillary equipment in as short a time as possible.

The major part of this research was carried out atthe testing laboratories of the Electricité de France,at Fontenay-Aux-Roses. We wish here to convey ourthanks for the valuable help of the E.D.F. in thoseexperiments, and particularly to thank MessieursNasse, Arribes, Anres and Gary, whose constant helphas been most welcome, as well as all the technicianswhose enthusiasm and devotion to duty enabled us tosurmount a great many difficulties.

ADDENDUM!

Hydromagnetic WavesA spectrograph with a rotating mirror was set up

at the observation port equidistant from the two gapsin the metal torus (Fig. 1). The resulting time-resolved spectrograms (Fig. 13) show a succession ofbands in the continuous background, generally occur-ring at intervals. A microdensitometer has been used todetermine the length of the intervals. Such a pheno-menon could be explained by the existence of trans-verse hydromagnetic waves excited by the formationof arcs at the insulating gaps and propagated alongthe magnetic field lines in the plasma.

Table 1. Comparison of Observed and TheoreticalPlasma Wave Velocities

Dischargenumber

395397399440441

Alfvén speed, VQcm/sec

3335.44.5

Observed speedscmjsec

2.3, 2.32.3, 2.3, 2.33.14.6, 6, 11, П5, 5, 5, 5, 5.6

To check this hypothesis, the observed velocitieshave been compared with the Alfvén speed, vo, forthe non-constricted discharge:vo = Б2о(4тгро)~^ whereBz0 and po are the longitudinal field strengths andplasma density at the beginning of the discharge.General agreement is shown by the results given inTable 1.

In Discharge 440 the observed speed increasedappreciably during an increase in the plasma current.Letting S equal the cross section of the plasmacolumn, one can write

pS = PQSQ, Bzb ^ JDZQSO

whence

uQ¡a = (So/S)* ^ V/VQ

where a0 and a are the initial and pinched radii, andv is the Alfvén speed for the pinched discharge in afield Bz. The data for Discharge 440 indicates a linearcompression a§\a ^ 2.4.

™ Bzo (oersted)

Figure 12. Width of H2 line as a function of the axialmagnetic field

f Additional authors for the addendum were P. Anres andH. Arribes, Electricité de France, Centre d'Essais de Fontenay-aux-Roses.

DISCHARGES IN A METAL TORUS 105

Figure 13. Time-resolved spectrogram of Equator I discharge andmicrodensitometer trace of continuous background

Figure 14. Steatite coupling sleeve for Equator С torus1, torus shell; 2, sleeve; 3, teflon seal; 4, conducting ring for

preionization of the gas

Equator С

The new copper torus, Equator C, has the samedimensions as Equator I. Its principal distinctions arean internal coating of 0.3 mm of enamel and a Steatiteinsulating sleeve of the form shown in Fig. 14. Thisdesign shields the vacuum seal from the dischargeand provides a long path length for possible arcsbetween the ends of the metal shells. Both indiumand polytetrafluoroethylene (teflon) have been usedfor sealing the joint. With this arrangement it wasshown that the current flowed entirely in the plasmaand not at all in the metallic shell.

This experimental apparatus permits the use ofmagnetic circuits premagnetized to saturation beforethe discharge. Under these conditions, the product ofthe average induced emf, Ё, and the characteristic

\л—~ " • — _

Figure 15. Typical oscillograms for p — 4x10~2 mm Hg, £o = 4v-cm; (left) Bzo = 100 gauss, (right) Bzo = 1000 gauss: /2, secon-dary (gas) current; d/2/dt, its derivative; E2, induced voltage;

/1, primary current.

time, T, before saturation is $oTEdt = ET = 2.5 x 108

maxwells. The primary circuit is coaxial with thedischarge tube and provides 31 conductors. Themaximum electric field in the plasma can be variedbetween 2.5 and 5 v/cm by varying the condenserpotential from 20 to 40 kv. The discharge charac-teristics depend upon whether or not the iron saturatesand upon the strength of the magnetic field. Figure15 shows typical current and voltage oscillograms formagnetic fields of 100 and 1000 gauss. For the lowerfield the gas discharge current, /2, rises more rapidlyat first but falls off very quickly.

ConclusionThe Equator С apparatus, with its enamelled

copper torus and Steatite insulating sleeves thatshield the edges at the gaps from the discharge,appears to give a satisfactory solution to the majorproblem that arose in the operation of Equator I,viz., arc suppression.

REFERENCES

1. C. Breton et al., Production et études de décharges à forteintensité, Proceedings of the Third International Confer-ence on Ionization Phenomena in Gases, Venice, pp.145-162, June 1957.

2. C. Breton et al., Observation spectrographique de décharges àforte intensité, Proceedings of the Third International Con-

ference on Ionization Phenomena in Gases, Venice, pp.163-169, June 1957.

3. R. J. Tayler, AERE, T/R I. 859 (1957).

4. Bickerton, Proceedings of the Third International Con-ference on Ionization Phenomena in Gases, Venice, pp.101-110, June 1957.