development(u) no unclassified -(0; mohoemmohi ... · the ebcs is an opening switch concept that...
TRANSCRIPT
RO-RI43 233 PLRSNR SWITCH DEVELOPMENT(U) JAYCOR ALEXANDRIR R1/2CONISSO S9 JUN 84 JRYCOR-J26-4-O.622a
NO M4-2-C-2114UNCLASSIFIED F/0 20/7 NL
LIE -(0; mohoEmmohI
mhhhomhhmmmummhhhhEMmhhhmhhl
IIJI. L3
1111111A 2
1.25.
MICROCOPY RESOLUTION TEST* CHARTNATIONAL BUREAU OF STANDARDS- 1963-A
. . .- -. . . . ..-. . ..-. . . . ..-. .- -- - - .- -
i
I
I PLASMA SWITCH DEVELOPMENT
J206-84-008/6220
044
I' '-U ~J4MCOR
1 4_
LUJL .j 205 South Whiting Street
Alexandria, Virginia 22304
. . .. . .. . . . - ..
PLASMA SWITCH DEVELOPMENT
J206-84I-008/6220
Final Report
by
Robert J. Commisso
June 8, 1984
Prepared for:
Naval Research Laboratory ~I4555 Overlook Avenue, SW ~ J 2
Washington, DC 20375
Under:
Contract Number NOOO1 4-82-C-2114
This document ,,,:s been aD01fol public 1010e1,, and sale; itsdisixjbutiC1 is unlimnited-
MR=-
6220
* June 8, 1984 3
Mr. Ihor M. VitkovitskyCode 4770Naval Research Laboratory4555 Overlook Avenue, SWWashington, DC 20375
SUB'JECT: Final Report, Contract Number N00014-82-C-2114
Dear Mr. Vitkovitsky:
JAYCOR is pleased to submit this Final Report entitled,"Plasma Switch Development," in accordance with the subjectcontract, CDRL Item Number A003.
£ If the Final Report is acceptable, please sign and forwardthe enclosed DD Form 250.
Questions of a technical nature should be addressed to Dr.Robert J. Commisso while- questions of a contractual natureshould be addressed to Mr. Floyd C. Stilley, our Contracts
* Administrator.
Sincerely,
Martin C. Nielsen -Vice President and Counsel
ssh-b
Enclosures
cc: Code 4701
Code 4703Code 2627DTIC -
205 South Whiting Street Alexandria, Virginia 22304 (703) 823-1300
q9ECU&ITV C.ASS|FICATION Of TS141 PAGE (Wien Dec Entered)
REPORT DOCUMENTATION PAGE 3RED OPCTLrGuOrSDEFoRE" COMPLE'rNG FORM
I. REPORT NUMBER 2. GOVT ACCION NO. 3. RCIPIENT'S CATALOG NUMBER
4. TITLE (and Subtitle) 5. TYPE OF REPORT & PERIOD COVERED
PLASMA SWITCH DEVELOPMENT Final Report: 02/26/82 thru02/25/84
6. PERFORMING QRG. REPORT M4UMGCfR
J206-84-008/62207. AUTHOR(a) a. CONTRACT OR GRANT NUMUIR(s)mRobert J. Commisso N00014-82-C-2114
9. PERFORMING ORGANIZATION NAME ANO AOOAESS I0. PROGRAM ELEMENT. PROET, TASKJAYCOR AREA & WORK UNIT NUM4E[S
205 South Whiting Street A003Alexandria, VA 22304
11. CONTROLLING OFFICE NAME ANO AOORSS 12. REPORT DATENaval Research Laboratory June 8, 1984
4555 Overlook Avenue, SW 13. NUMBER OF PAGESWashington, DC 20375 182 pages
14. MONITORING AGENCY NAME & AOORESS( f dittetat from Cttrollinr Office) IS. SECURITY CLASS. (of tle repor)
UNCLASSIFIEDI"I. OECkASSI FICATIONi DOWNGNAOING
I. OISTRIBUTION STATEMENT (of this Rope )
3 copies - Code 4770
1 copy - Code 47011 copy - Code 47036 copies - Code 2627
12 copies - DTICI?. OISTRSmUTION STATEMENT (at the abstract enteued In Stock 20, II diffen m fre Reprt)
U *
I. SUPPLEMENTARY NOTES
I. KEY WOROS (Continue an reverweo dsde It neoeeemo, and Identify by bock niber)
Electron-beam controlled switch (EBC6), plasma dynamic switch (PDS)
20. ABSTRACT (Contime a revorse id. It necooee and Idenftify b block number)
DO , 73 1473 EOITION OF INOV IS OUSOLETENCLASIEAN$N 0102*LF414-6601 UNCLASSIFIED-- i-- SECURITY CL-ASIFlICATION OF TMIS PAGE (Whben Data Seeredf)
TABLE OF CONTENTS
I. INTRODUCTION.........................1
II. E-BEAM CONTROLLED SWITCH...............2
III. PLASMA DYNAMIC SWITCH ................... 5
IV. SUMMARY............................15
V. REFERENCES ........................... 16
APPENDIX I..........................18
APPENDIX II.........................165
I. INTRODUCTION
This document is the Final Report for Contract
Number N00014-82-C-2114 (JAYCOR Proposal Number 8206-48).
i the work etne u.dr4---n-ac relates to two opening
switch concepts that may be applied to inductive store/pulse
compression techniques for various pulsed power applications.)
of interest to the Plasma Technology Branch of the Pla
Physics Division at the Naval Research Laboratory.-
t The opening switch concepts that are the subject of
this final report are the electron-beam controlled switch
(&S- nd the plasma dynamic switch,
The EBCS work is well docuknted and will only be
briefly discussed here. The details of this work can be
found in Appendix I, which is a compilation of a peer re-
viewed journal article ' , a journal paper accepted for publi-
cation', an NRL report', conference papers - 7 (one of which
was an invited paper') and presentations at various meet-
Ings-'1. This work was performed in collaboration with NRL
and JAYCOR personnel under Contract Number N00014-82-C-2336. -
Because the second area of investigation, the PDS,
has only been documented in JAYCOR Final Reports J206-83-
006/6209 (Contract Number N00014-81-C-2152) and J206-83-
010/6224 (Contract Number N00014-82-C-2336) and in one con-
ference presentation'2 (see Appendix II), the progress to
date will be detailed here.
II° K-BEAM CONTROLLED SWITCH
The EBCS is an opening switch concept that has the
potential for fast (-10 kHz) repetitive operation' - 4 . The
work performed in this area includes:
* a systems study that incorporates load pulse
requirements and EBCS physics self-consistent-
ly1 ' ' £
* experimental study of the scaling of various
switch parameters with comparison to
theory'5_-,1,1
* investigations into optimizing gas mixtures for
opening switch applications 1 '"'"'''a'1 ; and
a preliminary study of the discharge stabili-
ty ,2 ,1 .0 1
In addition, a new experimental facility for gener-
ating a long (Olus) e-beam pulse was designed, fabricated,
assembled and successfully operated. This device was neces-
sary to demonstrate EBCS operation in the -1is conduction
regime and is illustrated in Figure 1. Figure 2 shows a
typical current and voltage waveform for the generator.
EBCS experiments using this facility are now being planned.
References 1-11 appear in Appendix I. The engi-
neering drawings for the EBCS to be used with the long pulse
generator can be found in JAYCOR Final Report J206-83-010/
*6224.
-2-
Maa
gal%
UUI3
* .1
Li~
-- 3-
Cl)V q r
I--
~~LO
!e 0 tI
< Z 41
X cr-
4'
.0 m
S., q64..441 4
'44
'4y
00 0 0 0LO 0 U') 0 fO11
N N
(AN qA
-4-
III. PLASMA DYNAMIC SWITCH
The plasma dynamic switch concept is illustrated in
Figure 3. A switch plasma produced by a plasma source moves
with speed v toward two electrodes held at a relative elec-
trical potential difference (Figure 3a). The switch circuit
conducts when the plasma is between the switch electrodes
(Figure 3b). Conduction ceases (the switch opens) when the
plasma leaves the interelectrode region (Figure 1c). The
crucial areas of investigation for this switch concept are:
* the nature of the switch plasma density gradients
at the plasma leading and trailing edges;
* the interaction between the plasma and switch
electrodes; and
o the formation of an arc discharge between the
electrodes and how this might be avoided.
Experimental investigations into this switching
concept have been carried out in fulfillment of this con-
tract. They are a continuation of work done under Contract
Number N00014-81-C-2152 and N00014-82-C-2336. The results
to date have been mixed. There is some indication that the
switch opens; however, the discharge soon goes into an arc
mode from which there is no recovery.
Some of the initial results have been reported at a
conference12 and the conference presentation comprises Ap-
pendix II. The remainder of this section deals with most
recent results.
The PDS is an exploratory concept which makes use
of the high directed energy acquired by a gun produced plas-
ma. An important aspect of this concept is the plasma
spatial distribution. Earlier measurements with Faraday
-5--- --
SWITCHPLASMA
* m Vp
PLASMA SWITCH ELECTRODESI SOURCE
Ftl
II
,VVP
Figure 3. Schematic depiction of plasma dynamic switch concept.
-6-
cups indicated that the plasma filled the interelectrode
region for times long compared to the desired conduction
time (-5ps)(cf. JAYCOR Final Report J206-83-006/6209). To
m ascertain the plasma distribution, several measurements were
made. Witness plates (aluminum disks) were placed at vari-
ous distances from the gun. Shots were taken both with the
without a fuse in series with the gun. Photographs of the
plates appear in Figure 4. There is a decrease in the dam- S
age when a fuse is used. Also, the azimuthally asymmetric
nature of the gun plasma, particularly at the longer dis-
tances from the gun (AZ) is apparent.
A detailed set of Faraday cup measurements was
taken with screen attenuators. Figure 5 is a schematic of
the geometry used. The Faraday cups were of the same design
as described in JAYCOR Final Report J206-82-009/ 6215,
modified by the addition of permanent magnets. The magnets
provided a magnetic field measured to be ; 500 G to help
prevent plasma electrons from entering the cup and
secondaries from leaving the cup. The aperture diameter of
the 7araday cups was 64um.
The results with no screens were difficult to
interpret because the gun plasma has such a high ion flux.
However, with two attenuating screens, one of which (the one
closer to the gun) is grounded, the signals can be
interpreted in some cases. In Figure 6 the data from the
Faraday cups both with and without a fuse wire used in
series with the gun are plotted. The fuse consisted of
eight 28-gauge wires in parallel, each 11.4 cm in length. 4
There appear to be two plasma "blobs" early in time, the ion
flux without fuse being -2-3 times higher than the case
without fuse. This is a situation close to what is desired
-7-
41J
C4 ctn
)04
4
r) .- n
-8-4
*3
z w
O~4.)
0 0 0 u. .
0 0
a) 0 b ol O
a.
I ~~--- -- - -- --- 111 *
to c'i
z NrzW wLz<
cn c
a. L4
5; 0 0 b
0
0
D 0 U
U- po
o 0~
LL. U. U.. i-cl -0
Cf)) in OL) Qn i 0
(zUf/V) -lNSSdflWAvaIvvj
Sl to
00
(n 0
U'U
w
0
0 0 0 0 0 00 0(zwJO/V) 1lVNDIS dflo Avavzdv.
-10-
for switching. Figure 7 has the same data plotted but on a
less sensitive scale and for a longer time. Late in time
the plasma ion flux for the case with no fuse is so high
*that only the RC decay of the Faraday cup is observed. Note
that the fuse opens after one haltf cycle of gun current.
The late time signals for the case with fuse have disappear-
ed. If no screens are present, the signals look like the no
fuse case of Figures 6 and 7, both with and without the fuse 4
and the early time portion is very irreproducible.
A new electrode geometry was designed, assembled
and successfully tested. New experiments were undertaken
*using the attenuating screens, the new switch geometry and a 9
fused gun, together. A schematic of this system is given in
Figure 8. The coaxial geometry was chosen for better defi-
nition of the switch electrode surface. The gun circuit was
standard with C G 55 PF, L G - 184 nH and C Gcharged to 12kV 9
Csee Figure 8). The capacitor driving the switch was made
UP of two 13.4 AiF capacitors in parallel charged to 10 kV
and the switch circuit inductance was 2 750 nH.
* The results of these experiments indicate that even
with attenuating screens and fuse, the switch always devel-
oped a short circuit. This behavior is most likely due to a
combination of improper plasma spatial and temporal profile
and electrode plasma resulting from gun plasma-electrode
interaction. Investigations'' carried out In connection
with the PEGS indicate the electrode plasma expansion veloc-
ities can be >1 cm/pis. This means the interelectrode region
would be filled with such plasma in <10 uis. To accurately
assess the viability of this scheme, more effort than what
was provided to date is required.
0
00
LL k
CCj2
E - w0 --
I0 0 0 0S. S.I. LO
Ic) to
'4
Cf) U)
U-r~ 0- 0- 00
0~g 0 00
LO to LO t
(zLO/) lVN)I do xovv 0II~r 0124
IL fu0
4-
0
40 cnZ
Q '4
0 - *.I- S
1 co '
44'. - - - - - -c .
If interest in this scheme is rekindled there are
several improvements that could be incorporated into the
experiment:
* use of plasma source with a lower density, more
uniform plasma that can be controlled more
easily;
e use more transparent screens made of a highly
refractory material for the electrodes;
* clean electrodes (perhaps heating);
* implement better diagnostics so that the plasma
distribution can be well characterized; and
9 reverse the polarity of the switch electrodes.
-14-
I
0
IV. SUMMARY
The tasks performed under this contract relate to
-- the development and understanding of opening switch concepts
that can be applied to inductive energy storage schemes for
pulsed power applications. These concepts are the EBCS and
PDS. For all of these switching schemes, JAYCOR has provid-
ed diagnostics for characterizing the switching plasma and
the opening process, provided technical guidance in the
design of experiments, carried out analysis aimed at under-
standing the physical mechanisms governing the switch behav-
*ior, and prepared all the written documentation. In addi-
tion, JAYCOR contributed to the establishment of a long
pulse ( 1 Ps) e-beam facility.
The EBCS appears to be a viable switch concept for
*moderately high power (-1 0 1°W) particularly when fast, re-
petitive operation (-1 kHz) is required. The PDS has not
shown great promise; however, there are several improvements
which could be attempted.
* S
-IS-
I i n i m I m is -- " " " I I I I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
V. REFERENCES
1. R. J. Commisso, R. F. Fernaler, V. E. Scherrer andI. M. Vitkovitsky, IEEE Trans. Plasma Sci. PS-10,241 (1982).
2. R. J. Couimisso, R. F. Fernsler, V. E. Scherrer andI. M. Vitkovitsky, accepted for publication, Rev.Sci. Instrum., (1984).
3. R. J. Commisso, R. F. Fernaler, V. E. Scherrer andI. M. Vitkovitsky, NRL Memo. Rep. 4975 (1982).
4. V. E. Scherrer, R. J. Commisso, R. F. Fernaler, L.Miles and I. M. Vitkovitsky, Gaseous DielectricsIII, L. G. Christophorou, ed., Pergamon Press, NewYork (1983), pp. 34-39.
5. V. E. Scherrer, R. J. Commisso, R. F. Fernsler andI. M. Vitkovitsky, 1982 Fifteenth Power ModulationSymposium, Baltimore, MD, IEEE Cat. No. 82CH1785-5,pp. 146 (1982).
6. R. J. Commisso, R. F. Fernsler, V. E. Scherrer andI. M. Vitkovitsky, Fourth IEEE Pulsed Power Confer-ence, Albuquerque, NM, IEEE Cat. No. 83CH1908-3, 87(1 983).
7. V. E. Scherrer, R. 3. Commisso, R. F. Fernsler andI. M. Vitkovitsky, Fourth International Symposiumon Gaseous Dielectrics, Knoxville, TN, (1984), tobe published, L. G. Christophorou, ed.
8. V. E. Scherrer, R. J. Commisso, R. F. Fernsler andI. M. Vitkovitsky, 1982 IEEE international Conifer-ence on Plasma Science, Ottawa, Ontario, IEEE Cat.-No. 82CH177O-7, 153 (1982).
9. V. E. Scherrer, R. J. Commisso, R. F. Fernsler andI. M. Vitkovitsky, Bull. Am. Phys, Soc. 27, 982(1982).
10. V. E. Scherrer, R. J. Commisso, R. F. Fernsler andI. M.* Vitkovitsky, 1983 IEEE International Confer-ence on Plasma Science, San Diego, CA, IEEE Cat.No. 83CH1847-3, 88 (1983).
11. V. E. Scherrer, R. J. Commisso, R. F. Fernsler andI. M. Vitkovitsky, Bull. Am. Phys. Soc. 28, 1148(1983).
12. R. J. Commisso and I. M. Vitkovitsky, 1982 IEEE SInternational Conference on Plasma Science, Ottawa,Ontario, IEEE Cat. No. 82CH1770-7, 154 (1982).
13. J. M. Neri, R. J. Commisso, Y. A. Goldstein, R. A.Meger, P. F. Ottinger, B. V. Weber and F. C. Young,1983 IEEE International Conference on Plasma Sci-ence, San Diego, CA, IEEE Cat. No. 83CH1847-3, 124(1983).
-17-
- ,m • m m l m . . . . .. . . . . . .. .. . .. - - " - - ' " "
APPENDIX I
PAPERS ON EBCS
REFERENCES 1-11
I!
-18-
IMM TRANSACTIONS ON LASMA SCMCI, VOLI PS-O NO.4, DFECIN R 1982 241
Electron-Beam Controlled Discharges.X. J. COMMISSO, R. F. FE SNLER, V. E. SCHERRER, AND I. M. VITKOVITSKY
Ahmwt-Wbm dem bea (4m) is aected into a ps of high resistivity very quickly once the source of ionization is .bessd beewo two ud-eaud, a volwur dbhcMp, w"ici an removed. Unlike an arc discharge, this transition can be ac- 0mou @6 In rnoaf1.m w,1 te hem, - b puut We pummamia of ti Brnmy Mi te e =edmb dma f We prvn a d complished rapidly under an applied voltage.wkwm a poedm "padom for hibpow swhtchl8. Ti data In this paper, we present some measurements of dischargemgotbt a opdern bal. belweam the comotodimy qaue- resistivity and opening time in a repime that can be scaled to* of low is"v y Md dta ort 8 dlne may be jedhid by high-power applications. These measurements are necessary
NJe chi t ot 'itk. to understand the relevant basic physical processes, to provide
1. INTRODUCTION a bench mark for comparison with theory for scaling purposes, 0and also to provide data that are important in the design con-
ADVANCES in electrical high-power (;01 o W) technol siderations of an actual device (e.g., switch size, ohmic heatingf/ ogy ar necessary for the successful development of such loass, ratio of applied electric field to gas pressure, etc.). Theares as charged particle beam production and propagation, fmental theoretical considerations ae reviewed in Sectionhigh-power lom and inertial confinement fusion. One task II and the experimental apparatus is described in Section M.to be addressed in the improvement of highpower technology In Section IV we present and discuss the results of our me* -is the realization of rapid (>1 kHz) repetitive switching that surements and in Section V the implications and conclusions 0is compatible with an inductive energy store. At present, the are summarizedstate-of-he-art of high-power repetitive opening or domingswitches is limited to repetition rates of ! I kHz. Several IL THzoRY
authors 1I] -(71 have reported on experiments and theoretical When an e-beam is injected into a chamber containing ainvestigations in which an electron beam (e.beam) was used to mixture of attaching and nonattaching gases, the ionizationcontrol the resistivity of a gas mixture between two switch associated with the e-beam competes with attachment and -dectrodes. This concept has been shown theoretically to have recombination processes, thereby controlling the conductivity
application as a repetitive opening and closing switch of high- of the gas. For heuristic purposes, and because of the avail-repetition rate (in a bust mode) and high-power level [5]. ability of data, we consider a mixture of 02 as an attaching
An important distinguishing feature of this switch is the gas and N2 as a nonattaching gas. Then the dominant atomicability to open (cease conduction) under high applied voltage, and molecular processes involved in our parameter regime areThis Is achieved by using an external agent-in this case, the outlined in what follows..besrn-to control the gas ionization. Initially, avalanche ion- Iizaton is avoided by keeping the applied voltage across theswitch below the f-sparking threshold. As the discharge The number of discharge plauta electrons produced perevolves, cummulative gas heating also must be constrained so cubic centimeter per second as a result of ionization associatedthat thermal ionization and, mote importantly, hydrodynamic with the beam can be expressed asreduction in gas density do not significantly lower the self-sakig threshold. A discharge is thus S5 .-u kIo,,,. (1) .*vented. Under these conditions, the fractional gas ionization, 1sad thus the switch resistivity, at any time is determined by Here is the e-beam current density, e the electronic charge,the competition between ionization provided by the beam and nA the neutral gas density of species A, and vlA the totalthe various recombination and attaching processes characteris- effective ionization crow section associated with the beam fortic of the specific gas mixture, pressure, and applied electric species A. This -s section includes the ionization caused byfield. A second important feature of this switch is the volume primary beam electrons, as well as that done by all secondaries.discharge property. This characteristic makes It possible to Such secondary procsses may account for a substantial frac- -
aoid excessive heting of elctrodes and the switch gas. These tion of the total ionization. Ionization by discharge plasmatwo features permit the discharge to return to its initial state electrons accelerated by the applied electric field (i.e., ava-
anapt receIvd Aptil 28, 1962; re- d July 2,M2.Thiswork lanche ionization) is constrained to be small compared to*a apprs by die Naval Surface Weapon Cte at Da VA ionization associated with the e-beam.22448.
R. J. Cammiuso and R. F. Fensluare withJAYCOR. lnc.Alexandria. RecombiniadonVA 22304.
V. E. Scherer ad L M. VltkowiUky are with the Nan Research The most rapid recombination process for our parameterLaboratory, Washlnpoa, DC 20375. regime is that of two-body dissociative recombination
0093.3813/82/12000241S00.7S 0 1982 IEEE-19-
!
I I | | . . . . . . . . .| l~ l| . . . . . . . . . .. . . . . .. . . . . . . . . . . . . . . . . . . . •
242 IE TRANSACTIONS ON PLASMA SCIENCE, VOL. P-10, NO. 4. DECEME R 1962
e+A* -A +A (2) ...2OOOz t~ --
whs A is, in enera, any positive molecular ion (simple or
cluster) of species A. We define the effective rate at which this 1= Ter XPRNprocess occurs as P(s-1) and note that 0 is proportional to thedensy of pitiv molecularions,whchinturnis related to ,,o
the density of discharge plasma electrons through the chargeneutrality condition (see (6)). Thus this process proceeds ata progressively iower rate as the plasma elecron density i a.
The recombination rate between electrons and atomnic ions a i (Adtri0 4W 500(created by the c-beam) is much slower than for recombina- Fla 1. Cwpab ofmardntcr in dlschar circtit 1,, andtion with molecular ions. For a situation wher re bination Si ltiis the dominant process in the recovery of a switch, the pres-ence of atomic ions my slow the recovery process. This is neutral collisions, and p is the electron mobility. IE is thean undesirable effect, for example, in a fast opening switch. applied electric field, then pA is defined byWe further note that process two can rapidly heat the gas as aresult of the relatively low dissociation energie (-7 eV) of I& (8)the molecules compared to their typical ionization energies £(1S eV). Such heating is generally undesirable for switch where v is the electron drift speed.applications. Equations (5) and (7) are closed by the circuit equation (seeArthm nt (10)) and the appropriate Ohm's law for the discharge
Two attachment processes must be considered: 1) two-body E (9)dissociative attachment
e+02 -. O+0 () where Jp n.n is the current density associated with thedischarge plasma electrons only. Note that Jp is never mes-
and 2) three-body attachment sured directly; rather J, the net current density flowing in the
C + 02 +. A, O + A2. (4) discharge circuit, is drectly meastrable. J. is related to J. byWe defusaan effective rate for these processes uaa(s")~and - J= 4. For switch applications, we desire thatp >>J;
however, this condition Is not satisfied In all cases.note that ct is proportional to the density of the attachinl gas. By experimentaly determining p or Jp for a specific choiceBecame the gs is only weakly lonized, a is essentially indo- of gas mixtures, geometry, and discharge circuit parameters,pendent of the discharge plasma electron density. Ths type of a comparison can be made with the calculated values of p orbehavior is desired for fast opening switch applications. Jp. We have assumed that on the time scale of the experiment
Asunng sptia gradients ar unimportant, the continuity the bulk of the ions remain immobile.equation along with processes two, three, and four and (1) Using the air chemistry code CHMAIR [8], the appropriateresult in a differential equation goeining the time history of beam and circuit parameters may be coupled with the airthe discharge plasma electron density np chemistry to provide a numerical simulation of the experiment
4for one representative gas mixture. In this case, many moredtP = A nA - (a + 0)nP. (5) atomic and molecular physics processes than those previouslydt e Aoutlined can be included self-consistently. Fig. 1 illustrates
(In ignoring spatial gradients, we have also assumed electrode such a comparison. The measured net current i. (geometricallyeffects are not important, which is the cue for applied volt- proportional to J.) as a function of time is compared to theages much greater than the sheath potential). Equation (5) is calculated value. Because J was not measured directly, butcoupled to a set of rate equations, one for the density of each rather inferred from other measurements, there is a degree ofionic species, so that the conservation of chrg is self-conis- uncertainty in the initial conditions used in the code. More-tently satisfied, Le, over, even for air, many of the rates used in the code are not
well known. Nevertheless, the comparison confims the gen-n, + n Pr; n; (6) eralvalidityofthephysicalmodelused.
A A Other calculations have been performed, for a variety of gaswhere n; and n; are the densities of positive and negative mixtures, in which a Boltzmann analysis was used to deriveIonic species, respectively, the required transport coefficients and reaction rates [9].
The electron density n. is related to the gas resistivity Simulations have also been performed incorporating axialthrough spatial variations to assess the effects of the cathode sheath _
and related phenomena [101. We note that considerably-(epI)-t (7) altered behavior of p can often be obtained by choosing gaseswhere p is the pa resistivity, which is dominated by electron- in which 1 decreases and a increases rapidly with applied field.
-20-
COUMUO etraL: ELEOhON-SEAM CONTROLLED DISCHARGES 243
This point is discussed again in Section IV. Although suchj changes tend to complicate our understanding of the basic
ionization and deionization processes and can additionallyresult in unstable behavior [11], they can significantly im-
* prove switch response, and therefore can be very useful in s CAT"=
cran applications, diode =:An important issue for long time conduction and repetititive L - '
switch applications is the boating of the gas during conduction. L I ' W *Thi heating has been addressed theoretically [51 and is in- ---- - '----
An inductively d e-beam diode system, in which twoezploding win fuses are used in sequence as opening switcha Fi& 2. Schematic of apputus for investigatia e-berm controlldto produce a high-voltage pulse across the diode, provides thee-bean for the gas discharge studies. This system has beendiscussed in detail elsewhere [121. A general schematicof the system appears in Fig. 2. Generation of a seconde-lm pulse can be accomplished by switching in a second/ iset of fuses and switches (not shown in the figure), which aresima to the set used in generating the fi pulse [121. The _
central 16 cm2 of the anode is provided with holes to allowthe beam to enter the discharge region at a specified averagecurrent density.
Fig. 3 depicts, with arbitrary amplitude and time scales, thecurrent and voltage waveforms associated with single pulseoperation of the inductively driven e-beam diode. The voltage to It :1pulse across the diode is generated by triggering spark gap -S, (Fig. 2) at time to. Current il then begins to flow in Fig. 3. Depiction of inductively driven pular cuirnt and voltaethe first stage of the circuit, charging the inductor Lo. Fuse behaviot.Ft acts as an opening switch that diverts the current it, intime short compared to the rise time of 1 , into the second current through and the voltage across it. The circuit is do-
stage of the system via the self-breakdown of spark gap S2 at scribed by
timet,. Becauseoftherapidrseofl,thefuseF 2 canbeof 1s mialer dieterthan F, and thus responds to the current 12 VoR0'-1b)+Lf+d -L i, dt (10)by opening in a time short compared to the opening of F,. d CV f
This genrates the high-voltage pulse across the diode after where i, is the net current flowing in the discharge circuit, Voself-breakdown of sharpening gap SD at time t2 . The fuse FD is the initial voltage on capacitor C, L is the circuit inductance,interrupts the diode current at time t3 allowing the diode to and i. is the e-beam current. Note that the beam currentrecover from the whisker and arc plasmas associated with the not only provides ionization, but also enters as a currentfirst pulse so that a second e-beam may be generated. Typi- source in (10). p, J., and lb are geometrically proportional 0cally, voltages of -180 kV, currents of =1 kA at densities to the gas resistance, R, i., and i, respectively.of =50 A/cm2 , and beam widths of =200 ns were obtained Representative data obtained from this apparatus with call.with a 3-cm diameter sawbiade cathode, an anode-cathode brated Rogowski loops and voltage dividers appear in Fig. 4.spacing of 1.5 cm, and an initial charging voltage on the capaci- Plotted as a function of time for 20-percent O2-80-percenttorCe (56 pF)of=9 kV. N. at 760 torr are: the diode voltage and current, VD and iD ,
Alo illustrated in Fig. 2 are the driving circuit and electrode respectively; the voltage across the discharge Vd; the net cur-structure for the ps discharge. The gas discharge vessel was rent in the discharge circuit i. ; and the e-beam current ib . Asoperated at a prasure of 760 torr with an electrode separation noted in the previous section, ib was not measured directly butof 13 cm. The applied voltage on the capacitor was =8 kV was inferred from other measurements and assumed to changebelow the static breakdown voltage for the various mixtures of proportionally to iD from shot to shot. Hence, there is somepin used. A 50-gi mylar window maintained the pressure uncertainty in the determination of ib, especially its detaileddifferential between the discharge vessel and the evacuated time history.c-beam diode while allowing the high-energy portion ofe-berm to enter the discharge chamber. The values of the IV. RESULTS AND DISCUSSIONdischarge circuit capacitance C and inductance L were 1 IAF When considering the design of an actual e-beam controlledand =l IAH, respectively, switch, the following parameters are important: the rise time,
The resistance of the cell is determined by measuring the jitter, efficiency or current gain (Jp/J ), resistivity, discharge
-21- 0
244 MEN TRANSACTIONS ON PLASMA SCIENCE, VOL. PS-10, NO. 4, DECEMBER 1982
20% 0t-% N2 2250700O
Wo - £00
1250 9
low 00
70
Wo 100
____________________________0 20 A..1i0 40 1 a0 I0 go go 00.1 1...... 1. ArTAHolm Sa"
M Fs5. Plot of zldvlt at peek got . 00a annction ofpr-=en attach"n pa for vance' Ps mixtre AttaebIng PM Mt CO:
,a- 1000And 02-
........................................... ..............................
sea
300750
-- A
TIME (adlI.~ ' %~
M~4. Representation of the meemnd diode voltage VD. diode cset0-t~0 20 30 40230 070 90 0 C00
a fianitia of thne. Also = eZ is the lafene =Mer come %n A. H GAeSthe dblls II- F*g 6. Plot of diacherg crim denay dae (opening then) TO. as a
ftoction of petct attwacing see fat various pa mimer.. Attach-current decay time and recovery characteristics, and the ef- In, sam ase C02 and 0:.fects of energy deposition on these parameters. Rise times of-2.5 its with jitter less than the response time of the instru- described in Section III. Neglecting the last tern in (10) (<S.mentatlon (:90.2 us) [41 and current gains [61 of -1000 percent correction) we have at the time of peak 4',Le., whenhave been reported under a wide variety of operating condi- d1/dt = 0, a switch resistanc ofdions. Little data have yet been accumulated, however, on thedischarge resistivity p and its scaling with gas mixture,i4. bealm A0 a - 11
energy, and E/P. For example, the ohmic heating power di- 0i. - )sipation in the switch, which may have deleterious effects on Po is then related to R0 through the switch geometry. In Fig.switch recovery [S] may be controlled by varying the switch 5 we prsent a summary of p. as a function of attaching gassim and thus its resistance, for apgven p. for 10 A -4' <30 A/an 2 . Over this range of 4, the data
For opening switch applicatons,It is also important to know vaies less than afactor of 2,typically -40 percent. The at-.the discharge current decay time (opening time) and the dii taching gase are %2 and CO2. These results illustrate the rela-charge recovery characteristics. Although one investigator (31] tive merit of the gas combinations and percentages for ptoduc-reports that a current of -150 kA was Interrupted in -I jus at ing low resistivity. The discharge current decay time To isl4 a 1.5 A/an2, no systematic empirical or theoretical scaling presented in Fig. 6 for the same gas combinations and for theof these parameters in an applicable regime for use a a comn- same incident aeamm current density us the previous figure.ponent in a high-power inductive store pulser exists (Le., for As a reference, the decay time of the e-beamn is also given. Thea parameterrmuse of i- IO'-l0' A, V - 04 V, opening decay timesand rise times may be governed bydcrcuit param t
time -! 100 n, conduction time Z. I in). etera, a well as the amplitude variation of the beam currentMeasurements of resistivity and opening time as a function and the attachment rate.
of percent attahing gas for a variety of gas mixtures at a total Note that, in general. one would expect the gas combinationspressure of 760 tonr have been performed using the apparatus which give the shortest To hae the highest pg (assuming a and
-22-
COMMISSO etL: ELECTRON-.A"M CONTROLLED DISCU AAGES 24
I& are only weak functions of E/P, where P is the ambient Is a highpowet switch include: 1) the unpac" o" energy depou.pressure). The curves in Fip. S and 6 illustrate this point. tion on switch performance. 3nd 2) the radeof between *This effect also has been observed for a recombination domi- current pm (switch effici.ncy) and other switO parameters.nated regime [6). However, a gas mixture whose attachmentrate is low for low applied field (during conducting phase) but ACKNOWLiEDGMENThigh for high applied field (during opening phase) and whose We wish to acknowledge the expert assitance of H. Hall andmobility behaves just oppositely, can partially offset this trade- L Miles in obtaining the data.off between re and Po [51-(71. We see some evidence that
U this situation may be realized in practice from our data. Al- Ri awmthough p for C% -At is comparable (or somewhat les) than 11 . O. Hur. i ft - IEE b , Pbo a "d Power Couj. Lab-
Pe for CO2 -N2 , the decay time for CO2-Ar is clearly smaller bock, TX. 1976, ICS: pp. 1-6.than for C02-N2. Note that Ar is known to exhibit the [21 J. P. OAin. Js N@o .EEE FIF Ins. Aaia Pow Conf.,
Labbock, TX. 1976. CS: pp. 1-6.Ramissuer-Townad effct [1], [131. [31 L M. Kovachuk wndG. A. Meyat. So,. Te.k Phy& Len.. vL.2.These data were obtained at up to 75 percent of the self- p. 252.1976.
breakdown voltage for some mixtures with no post-discharge 4) K. McDonalK e .a ZE NRm kmS. S voi. MS-4. p. 181,1980.breakdown observed. The gas recovers to a dielectric stasa 1t 1 L F. Ftsk, D. Con. ad . A. V . LFEE Ihm
of -7 kV/cm. The current gains observed varied between Rnms Sa., vol. PS-8, p. 176, 1980.I and 10. [61 P. Bulsinger, in Pice Wrkshop R eperiue Openfm Switcsa,
Tamauoa. CO, Texas Tech Univ. Rep.. Apr. 1931. p. 128.V. SUMMARY AND CONCLUSaONS [71 L L Kline, in Pin. Workshop Repeftre Opeft Switches,V. SUMARYAND ONCLMONSTamaan.t CO. Texas Tech Univ. Req., Apr. 1981, p. 121.
We have reviewed the fundamental theory and presented [181 IL F. FrnDa, A. W. Ali J. R. Gre, ad L M. 12v.kNam Rea. Lab Memo Rep. 4110, 1979.
experimental data for an e-beam controlled discharge from the (91 . W. zman Mead Lp. 4110.1, We m E Corp.
perspective of high-power switching application. The data Rep. DYD-S545-AA. 1979.Sest that an optimum balance between the contradictory [101 J. J. Lowkl and D. K. Davies, . AppL Phyr voL 48. p. 4991,
1977.requirenmnts of low resistivity and short opening time may be [111 D. H. Daltona S. A. Mani, AppL Py. Lat, voLrealized by proper choice of gas mixture. Comparison of 23, p. 508,1973.theory and expedment confirms the gepnerl validity of the (121 B. Fell . J. Commio, V. E. Scherrer, and L SE Vitkovitzky,
. AppL Phyt., voL. 53, p. 2818, 1982.m[ physicalmodelued. 131 &CBrown, Bad Da of ewa,,ys. 1966,2nd. Ed. rised.
Issmu yet to be addressed for the successful development of Cambddle, MA: MIT Presa, 1967, p. 27.
-23-
HIGH POWER ELECTRON-BEAR CONTROLLED SWITCHES
x.J. Commismo, Ro.. oernsler, v.1. Scherrerand Z.M. Vitkovitsky
Naval Research LaboratoryWashington, DC 20375
-24-
abstract
the application of an electron beam controlled diffuse discharge to high
power C0 10%). repetitive opening switches is analytically formulated under a
set of assumptions. Das physics consaemrations are combin" with energ
transfer r qurments to obtain analytical estimates of the e-bermn controlled S
mitch paramters for given circuit re qrements. The switch design is
optimized by minimizing the wvitch pressure subject to the constraint of
system efficiency. the result of this optimization is that each of the major
naerw losses - conduction, opening, and electron beam production - are
roughly equal to each other. 2his formlation is used to relate the switch
Pramters to the desired operating characteristic for an arbitrary number ofaS
pauses. As an example, the formalism is utilized in outlining the design of a
single pulse, high power (0 10'%) inductive storage system. IL judicisu
choice of gas or gas mixture results in desirable changes in the system design
* or tfficiency.
9
-25-
S
S
am3 lOWER XECT1RII-32M CUTRCLED SulTCHES
R.J. Commmso. R.F. Fe.nslar, V.Z. Scherreur and X.Z. VitkovitAsky
Eawal Research Laboratory
Wshington. D.C. 20375
zi List of Symbols
A - switch area
I - electric field across the switch
C- z during conduction
a " C when I is maximam
(Z/P) B - reduced breakdown field strength
d - electronic charge
to W asb U
- ratio of e-beam energy to load energy
9C - ratio of energy dissipated during switch conduction to-load
energy
so a, ratio of energy dissipated during switch opening to load
enrgy
Ib ft e-beam current
IL - load current
I° - maximum Z.LL
- switch plasma current
-26-
0
Z 0 ma=0dum Io, aim SW -0
Jb - e-beam curent density
- switch plasam current density
kb - fraction of total. e-bem anerc deposited in .witch gas fro
inel.astic: processes
k,=- ratio of average switch current during conduction to I o
ko - fraction of peak load poer ssipated by switch during 0
kp - Op
T m Inductance of storage inductor
£ = switch length
= positive ion density-
- attaching gas density
U - hon-attaching gas density
S- mmber of pulses
np = switch plaa electron density
n - a. at ttim e-bsm. ceases
p - switch gas pressure
(Vl1d /dx)b m ,reduced enersy lost per unit length by beam electrons
%W = switch resistance
b - e-bem ionization parameter
as = safety factor for static breakdown
- safety factor for heating 0
Vb - *-beam accelerating voltage
VL - voltage across the load
Vo - muamm Vr,L
VSW - voltage across the switch
-27-
V0 - iximm V
yaW SW
v , plawma drift velocity
(W/P) 3 reduced critical energy deposited per unit volume of eitch
a - effective attachment rate coefficient
B - effective recombination rate coefficient
M M exret gain
W energy of beam electrons
9, - energy required for ionization per electron-ion pair
S... energy transfer efficiency
is * electron obility
S- gain factor
p - switch resistivity
PC W p during conduction
TC - tIm Interval during which switch is conducting
- current comutation tieCCH
- charactearistic load pulse widthL
To , time interval during which switch changes from conducting to
T- characteristic loss time for plasma electronsp
%A characteristic lose time for plana electrons - attachmentp
dominated dAscharge
- characteristic loss time for plama electrons - recombination
doonated discharge
11. ZUTROOUCTION:
The successful implementation of inductive energy storage to pulsed power
applcations 1 , 2 (> 109W) depends critically on the development of an opening
-28-
switch.. 2his switch mt conduct current for the charging period of an
inductor and then become sufficiently resistive to divert the current flowing
in the mitch to a load. the comtation =let occur in a tim short compared
to the inductor charging time and in the presence of an applied slectrz field
which Increases during the comtation process. Some applications
additionally require the switch to be capable of repetitively pulsed operation
(see, for umple, Pf. 3). Such a repetitively pulsed opening switch will
note than satisfy the requirements for a repetitively pulsed closing switch of
high ( 10 kz) repetition rate. Several authors4 -9 have reported on
siperiments and theoretical investigations in which an electron bea Ce-bean)
Is used as an external agent to sustain a diffuse (volumetric) discharge in a
gas. The concept as it might be used in an opening switch application is
Illustrated In Fig. 1. Zn thi scheme the gs resistivity at any time is
determined only by a competition between ionization provided by the e-beam and
the various recombination and attaching processes characteristic of the
specific gas mixture, pressure, and applied electric field. This, along with
* the volume discharge property, allows the gas to retu'n to its original non-
conducting state very quickly once the source of ionization is removed.
A particular design for the e-bea controlled switch (mm-) most specify
the switch gas pressure, gas mixture, area, length and e-beau generatar 0
requirements necessary to provide the required electrical characteristics of
the output pulse(s) at the desired system efficiency. The design also must
Insure that an arc discharge is avoided, otherwise the switch will not ehibit 0
a rapid recovery.
Zn this paper we review the physics of the ESCS that is relevant to the
switch design. The enerwr transfer efficiency for an inductive storage system
employing an hUCS is defined and formulated In such a way that it can be
-29-
oo
combined with the switch physics to arrive at a set of equations that yield an
analytical estimate of EIBM design parameters. This analysis self-
consistently includes the switch " ics, circuit requirements, and system
efficiency. 2e switch design is optimized by minimizing the switch pressure
subject t the constraint of system efficiency. as a resuLt of this analysis,
the range of applAcability for the WC in term of the switch conduction time
and output pulse width for a given system efficiency can be identified. To
i trate this formaUlim we outline a design for an inductive storage system
capable of delivering a single 200 k', 100 us full width at half maximum
(FlNM), 35 k& pulse into a 6 9 load.
2he witch physics and energ transfer efficiency are discussed in Sacs.
and XV, respectively. Tese considerations are combined in Sec. V to
arrive at general expressions for the switch design parmeters. An
illustration of this deign procedure is presented in Sec. VX.
IXX. SWITCH M ZCS CSMZERA=M!OS
various aspects of the physics associated with controlled diffuse
discharges as it relates to INC have been discussed. 4 - 9 Here we review the
physical model for the switch and discuss the discharge energetics. Zn the
model a uniform e-beam ionization source produces a very weakly ionized gas
between two electrodes across which a potential is applied. Typically,
np < 10 - 5 V., where np is the switch plasm electron density and No is the gas
density. The electrodes are perfectly uniform and the applied voltage during
the conduction phase is much greater than the sheath potential 6
- 1 Mi). The e-beam accelerating potential must similarly be high enough
that e-beam electrons can traverse the entire switch region (Ref. 10 suggests
the e-beam range- should be about twice the switch length). Another assumption
is a constant e-beam current and accelerating voltage with decay times that
-30-
are very short compared to the switch opening time (Sec. XXX-B). That these
conditions be net is important because any circumstance that allows the
presence of nonaifoxities in the Switch (other than in the cathode and anode
Sheaths) mW result in a nonunifo= electric field and degraded switch11 S
]perfomance (breakdown). hus, the gas resistivity of the volumetric
dachargW is determined by electron-neutral cof.lsions, and the electrical
pott4i in the mitch is a linear function of the distance between the
setod". 10 We further assume for ti analysis that the electron mobility 3
and attacnent rates are independent of the electron temerature. For mutch
deign purposes this iS a conservative approximation and is made to simpLify
the analysis. 2he advantages of using twerature dependent mobilties and
attafmnt rates winl be discussed in later sectiAons. Because e6peri=ats12
bare shown that ZS closing times can be very short: (< 2ns), 1the closing
phase is not considered-here.
A. Z.ectron Densimt
We ago=* that the switch has a two component gas mixture made up of a
ncpr-ttac::sg base g of density a N a with a small (< 10%) fraction of
attching gas at dikity va. Assuming further that density gradients are
negligible, the continuity equation for n. gives
an- , a + N )u. (1)
Here Jb is the e-beam current density, a s the attachment rats oefficiut
(c 3 .e' 1 ). P .is the recombination rate coefficient (cm 3 -sec-1), P is the
total gas pressure, and N+ is the positive ion density. 8b, a beam ionization
parameter, IS gi.ven by 5 b - (eC-'(pidCb/dx)h where * Is the electronic
charge* c, 35 eV is the energy required for ionization per electron-ion
-31-
pair, and (Pd /dx)b k eV/€m- i th e reduced energy lost per unit
length for the beam electrons, assumed constant. Because of non-ideal
conditions, including reflected electrons at the switch electrodes, thi
callatn my underestlate the electron proaction. 9 , 1 3
The tim decay of switch plasm electrons can be readily estimated frm
zq. (1) for both attachment dominated (a N >> O + ) and recombination
domite (CL Va << 15 N ) switch recoveries when Jb - 0. For an attacheent
domiated case
P p
Whore un0 is the switch plais electron density at the tUm the e-bea asep
(t -0)andt4 a ). Lettin le - in q. (1) for the recombinat-onP a
dodinated case results -in
0
n(t) - * • (3)
p
save T) (+op p
Nate that nP decays exponentially with t4e in the attachment dominated
case and approximately Inversely with U in the rec bna dominAted
case. noreover, because i4 is inversely proportional to Na, it can bep
externally controlled by adding more attaching gas.7 The recombination rate
between electrons and atomic ions (created by the e-beem) is much slower than
for recombination with molecular ions. For a situation where recombination is
dominant, the presence of atomic ions may slow the recovery process. This Is
an undersirable effect in a fast opening switch. Note further that when
molecular dissoc iative recombination is dominant, the gas can rapidly heat as
-32-
a result of the relatively low dissociation energies (-7 eV) of the molecoles
compared to their typical ionization energies 15 eV). Such heating is
ganeraly undesirable for switching applcations.
a. switch resistivity and Ovening Time
The switch pl.ama density is related to the switch pla.ma oxrent
density, J through
xS
-TW A .p
aere 1I1 is the switch plasma Current, & In the switch area, and v is the
ealectroot drift speed,
Y -DIt (5)
IV
where 11 is the electron mobility 1 4 and Z is the electric field across the
witch. Thus the switch resistivity is given by
Z -1p - -Cenit) (6)
53 P
x useful expression for the switch resistivity drinJg condaction PC can
be obtained for an attachment dominated switch in equilibrium (dnp/dt - 0) by
combining Eq. (1) and Eq. (6). hUS
PC-e a (7)bbp
A similar expression can be derived for a recombination dominated switch.
The recovery tie for attac ten, and recombination dominated switches may
be estimated by using Eqs. (2) and (3), respectively, in Eq. (6) and
-33-
• - . . . . . . . . ... . .. . . . , , . .. .. " -4
deterning the ti.e for p to increase by some arbitrary factor, say 102 . For
an attachment dominated recovery, this switch recovery (or opening) time
in -o -- 51 . while fo recabinatin dominated recovery thi tim
in-o 1'Y . Thus, because Ta can be made less than TV and on the basis
of other considerations discussed in Sec. X-A, an attachment dominated switch
is generally desirable for these applications.7
C. Scaling Relation
Replacing the last term in Eq. (1) with a phenomnological
expression n /-C , where T is the characteristic loss timer for the swi.tch
pp -p__l18
plains electrons in the absence of ionization i substituting np from Eq. (6)
into Eq. C 1); and assuming equilibrium Eq. (1) becomes
0
where f .U eS 1W is essentiall.7y a constant for a gilven gas coeposi.tion ando b
%0 is the electric field across the switch druing conduction at the te when
I .M is -- 4-mm. or sMost gases fo - 10S - 106 c/V-s. We also define the
current gain as C 2 0 , where Ib is the es-beam current and :I is the
-yfmi switch plasma current. The time history of I= will be discussed in
Sec. TV. The switch efficiency will depend on C (Sec. =I). This analytic
e rm scaling law has particular relevance for fast UCS applications.
In these applications one desires large
C and short Tp. For a given gas composition and given ZZ°/P, Eq. (8)
indicates there is a utrade-offO between "iigh C and small T while suggestingp
that operation at high pressure is favorable.
-4
-34-
in D. Discharge Stability
1. Static Breakdown
If the electric field across the switch exceeds the static breakdoM&
fi.eld at the ambient switch pressure, the switch may go into an arc mode,
which is undeirab-l because it prevents the witch resistivity frM being
externally controlled. Assuming unifozmdty
0VSW - ( )Bl(P ) , 9)
whox v0 - V is the m - eMected voltage aross the switch, (3/p) is53 LaI
the r*dced breakdown field strength (typically 20 hI/ca-atm), s i s 1 s a
dm iclssafety factor, and A is the switch length. This condition can
be relaxed somehat for transient (< 1 us) pulses.
2. Cumlative Effects.
Ctivestd heating of the gas =est be sufficiently constrained so that
Any reduction in switch gas density does not allow the self-sparking threshold
*to be exceeded. Energ is deposited in the switch from resistive heating
during the conduction phase, resistive heating during the opening phase, and
heating of the gas from the electron beam itself. This deposited energ per
unit Volum of switch mst be less than sCme critical value, required to
change the instananeous value of the gas density such that the self-sparking
threshold is exceeded. Assuming this reduced critical energy deposited per
unit switch Volume is (W/p)B, a constant for a given gas, we have
,-° "sWvsW o + WS. ?C + kb"9b,,n c - .(- )BA(,) (1,0)
.,., -.... 10)
TC
-35-
Here n is the umber of pulses, CC in the switch conduction tim, swVsI C
is the average paoer dissipated La the switch during conuction, issxro
the average paver dissipated in tho switch during opening, Ib and Vb are the
s-bei -crrent and accelerating voltag, respectively, and a t 1 is a
dimensionless safety factor. The factor kb is the fraction of the eObam
energy deposited in the gas and is estimated as
1 deb PI
* Typically kb 01
Omulative hentng can also change the chadcal properties of the gas
which may result -in additional detrimental effects. Studies in N2 have shan
that if in Eq. (10) (W/P) 3 is taken to be the reduced thermal energy density
of the ambient gas (-0.1 J/,3-at), these alterations are unimportant. 1 5
Such effects caused directly by the e-berm (e.g., direct dissociation) are
ignoired in this analysis.
3. AddItional Effects
Assuming that balk heating is sufficiently constrained, other phenoena
may affect the switch discharge stability. For exa ple, when a is a
strongly varying function of E, an attachment instability has been observed to
lead to breakdown in an e-beam controlled discharge. 16- 18 Also,
inhomogeneities in the discharge or switch electrodes and resulting local
heating have been suggested as possible breakdown mechanisms. 8 ' 1 1 ' 19 These
last effects can be minimized if in Eq. (9) we use %B < 0.7.8 ',19
IV. ENEMY TRANSFER EFFZCIENCY
We define the energy transfer efficiency, i , as
-36-
L@2(1/2)L " •bVbC
Z.earso and V*are the pusk current and voltage delivered to the load,andT is the cMtja tstic load pulse width. Note that for T. and the
trise tm into the load = U cougared to L/^ where L is generally
the storage inductance and R is the system resistance after the Z= is open
(rIg. 3) # z (although not at exactly the sam tme).
The energy st red in the storage indctor may be written as
jI o2 C > ( (13)
IThu in may be expressed as
S- -'- (14)
whee the gain factor, , s given by
o00
-r ', o>VsCT +Zv? • (is)
Yor energy transfer efficiency comparable to, for example, capacitive
store pulsed power generators (i >0 0 5), we require 1. Therefore, each
energy loss tern in the denominator of Eq. (15) (energy cost for switching)
meit be sufficiently less than the nmerator (energy delivered to the load).
Thusvwith 0 0y 0z .0 ,and >c edfn)he%V - V . L n Sc - R'Lq(I Tnd we define thredimensionless factors ogc" and gb as follows:
-37-
(-) 1, 16,)
L9C s e c R (S (16 (16b)
restricts th switch openingtm To - (Thse factor.
Sec-. 1'. ) Equation (16b) retric..s t:he voltagce drLop acrToss the rJ.swi ch a"wing
conducio. (The factor <xep:. L 0 nrq. (16b) in esimte i Sc..7fo
dJ ~~~vriu applicatitons. ) %hi4s condiLti'on in equivalent t=o requiring the c=e'
~~deca tim of t:he stor=age infttc t.ougb te swtch resist.ance (L/RSM) be
long compared t:o the swi.tch condcti :on time. EquatiJon (16c) rest:ricts the
_ mud~switc curr-entm gain, c. At. tchis poilnt€ the values of th g fact ors8 ar'e
arbitxaxy, consiLstent= wit.h the reuie eff.iciency throug r,:1. (17) and
(14). The arbitr.ary nure:. of th g factor wiJll dilsappear when the switch
pressure is minimized subject to the effi.ciency constxaint, jq. (1"7) (Sea. V).
V. 8WMTCH DZGH
The desigcn developed in trhis section emphasizes the overall energy
theansfe efficiency Of the switch, he fundamental circuit requiremens and
-38-
the switch physics. We asume all the load pulse current and voltage
c -eet.e.Aitics, T, and the system efficiency q are specified.
seemse the results of the two previous sections can be combined to
develop a self-consistent ZECS design, som specification of the switch
application will be useful. Illustrated in Fig. 2 are two idealized g r
*zomples of how as EOCS might be used to produce a short, high voltage
pu1e. In Fig. 2(a) the SCS must r current for the entire charging tim
of the storage inductance L. In sa applications this charging time my be
too lang fo one mswitch to fulfill al the requirements of the sysem
(especially a fast opening tim). The example illustrated in Fig. 2(b)
employs the UnCS an the last stago in a pulse compression scheme. Here L is
Charged through iitch Si - sme opening switch that can provido a
sufficiently long conduction time. The opening tm for S1 is i < C for the
K uMS but is long compared with To for the ZBCS. 2hus, the IncS is employed
as a second stage to obtain the desired pulse width and voltage. (A
discassion of the relation between conduction time, and load pulse width is
* presented later 1" this section.) The current source in both cases may be a
low voltage capacitor bank. Switch SL is a closing switch that may be
necessary to isolate the load during the conduction phase.
Keeping Figs. 2(a) and 2(b) in mind, we define a factor
k, a (18)
a 0The earlier assumption that Z. - SWresults in I 0.5 (Fig. 2(a)] or
k, M 1.0 (Fig. 2(b)]. depending oan vhich case is applicable. It will also be
useful to define a factor kO for the opening phase, 0
-39-
! •
U0i (19)
During the time interal Top the switch current decreases to zero, the load
current increases to rj, and the switch resistance increases. The tim
stoxy of the voltage across the switch vill depend upon the load voltage
tim behavior. For loads where the load voltage rises to its peak valm in a
tife - ao (i.e., resstive or inductive loads), <(sw>o a 0.5 ,and
•. V so that k - 0.23. ,or a capacitive load, the load voltage reaches
its Imadm in the ti it takes to charge the capacitor
(a TL fo e), ). Assuing in thi- case that during T h urn i ei
the capacitor Is Linear and noting that vT ,(t) - IL(t)/c, where VL(t) and
L(t) are the i taneons voltage and current across the load capacitor C,
ve obtain kt m 1 /9 TL
inally we define the pea eteur kp as
k Sl? (20)p Op
and recall fr Sec. IM-S that k -5 for attachment dominated recovery andp
-10 2 for a reccuahntion dominated recovery.
ITe switch pressure my now be computed. Using Eqs. (16a-c) and (9), the
definition of the current gain C , scaling relation Eq. CS), Eqs. (18) - (20)
and recalling that the .elect ic field across the switch at the end of the -
conduction phase is C o r
once P is, known, z. (9) gives ue witch i ennjh, 1, and Eq. (10) gtives
-40-
- ____ 7 L |a
I
the switch area, A. Thus, with the help of Eqs. (16 a-c) and (19), we
obtin
m f -( 2 1 b )% (z/P)3
and 00Z -(o
(21
LL - + g + k --b g21
a(W/P) 3PA gC 9
2be switch dimensions thus m uted insure that the switch will be large
eaou that the breakdown threshold ll not be reached.
as long as Sq. (17) is satisfied the g factors are arbi-axy. The
arbia-iness of the g factors in .q. (21a) can be removed by requiring that
£ the switch pressure be minima, subject to the efficiency constraint, Eq.
(17). Ths r.eqirment is desirable for simplicity of co-mnstu o. The
method of Lagrange undetermined -ltipliers 2 0 can be used to do this
I ,, a,, o . T e result is g o gc " qb Thus, the switch with the l ,est
pressure for a given ssma efficiency Ls one for which each of the energy
diss:Lpatin trm (each term in the demoninator of Eq. 15)) is equal to each
Other.
The amount of attaching gas necessary to obtain the required opening tim
cM be estimated from the arguments of Sees. ZX-A and 11-B. Por an attacnt
dominated recovery (T a STa)0 P
a (21d)a .?
The e-beam generator requireents are determined from Vb O 1 b, ' and
A. b cs computed from Eq. (16c) and the definition of c. The e-beam
-41-
generator mat actually provide a higher current than rb to account for
naryent lost to the structure support.ng the vacuum-high pressure interface.
The value chosen for Vb in Zq. (21a) can be combined with P and A to oomute
kr of Eq. (11). Note that for the beam to traverse the switch lengt.h Vb wil
depemd on the product PI an does VL. Therefore, one can show that
tpically Vb I L'
2he perfomance of an UCS is primarily limited by the desired switch
efficiency and by the ambient smitch Sgas pressure, as indicated by Zqs. (17)
and (21a). hoosing ko - 0.25. k p- S, and (Z/P)n - 20kV/cn-ata using
Zq. (14) and (17) with gOno-gb d ad assuming Vb - VL, Eq . (21a) can be
rewritten as
C (22)
Uq. (22) is plotted in Fig. 3 for n - 0.5 and 0.8, fo W WlO cm/v7s, and
p - 10 atm. The values of f. and P chosen are representative upper limits for
these parameters. igur_ 3 gives the effective range of applicabilj.ty of the
UCS in term of the ratio of conduction time to load pulse width for a
deired load pulse width.
V. KaWLZ ,-
Jlthough the results of Sec. IV can be applied for an arbitra-y number of
pulses, suppose we desire to generate only a single high voltage pulse in the
manner illustrated in Fig. 2(a), using a capacitor bank of capacitance C for
the current source. The desired pulse characteristics are 0 - 35 kh,
V. a 200 W, TL - 100 us, with 6 a) . This choice of these parameters is
relevant to several pulsed power applications.3 ,2 1 For this problem we choose
a conduction ti for the switch of 1 us. The situation depi ted in Fig.
2(a) has k * 0.5. Also, for an attachment dominated recovery kp - 5.
-42-
Referring to suggested gases5'2 2 we take for the svitch gas a 19 N2 -
Ar Mixture. This Mixture exhibits the Rassaurer-Townsend collision cross
section behavior. 5 , 22* 23 The addition of U2 cools the switch plain
electros throu* excitation of N2 vibrational levels, thereby keeping then
near the ar-send minima. We take for fo an reage value of
m2.s x 105 cn/v-sec and assm it is constant. Taking - 6.25, -0.7S v, ° V , 0U2) 0 - 0 Vo -a /i , - 0.5, (W/) - 0.1 /on =-at,
T - 1 lom, and q - 0.6, Eqs. (21a-c) give: p a 25 at=, L -0.6 on, and2
A a 335 - , rpectively. Equation (1Sa) gives To - 60 as and Eq. (16)
given a -70. Tus the current densities are j 100 /
and rb .5 /c 2 . The fraction f a attaching gas can be obtained from Eq.
(21d) and the (3-body) attacment rate for 02; f 0 0.2%. Finally the mitch
resistance during conduction cmse fram either q. (7) or Eq.
(16b),R S m 0.350.
With this information one can design the e-bem generator and calculate
the required values of C and L. Note that this is an LT: circuit. & finite
a mount of current (or, equivalently, magnetic flux) is lost during both thie
conduction and opening phases as a result of the finite switch and load
resistances, respectively. This is a direct consequence of having 11 < 1.
Therefore, the choice of L, C, and the initial voltage on C must be such as to
compensate for these losses.
The circuit of Fig. 2(a) can be solved analytically. Considering the
peroid during and after opening, using Zqs. (2) and (6) to give the time
dependence of the switch resistance Rw(t), assuming Ob - 0 at t - 0, and
taking C - 0.5 PF and L - 0.8 iPH we obtain the plots shown in Fig. 4 for the
system just described. Zn this came R (0) - RSW - 0.350 and C is charged to* SW S
-43-
. . . . .. _ _- _ _.. . .. ... ._ _ __._ __.. .. . . . .. . ._
50 W.
The use of a gas or gas mixture whose mobility and attachment rate vary
rapidly as function of "/Ro (I.e., electron temperature) has not been
discussed. Gases that have high mobility for low /HO and low mobility for
high Z/Wo in combination with gase whose attachaeat rates behave Just
oppositely have been suggested 7 ' 24 Ae * 2 m ued in the example
exhibits this tendency to some degree (factor -2): howver, C 4 shows
strougly increased mobility at low "o values vhile having comparable
mobAilty at high /0o.23 ow example, if a mixture of CX4 and C3 P 2 4 were
used in the example then average values of fo and a give a pressure of a 8 atm
with f a 1% for the same efficiency. This is a factor of -3 decrease in
pressure compazed with Ar-V 2 and is directly atrbtale to the factor
of - 3 increase in average f. for CH4 at the same /W . Because C34 " -s
polyatadIc it serves to cool the electrons itself lnd no additional gas is
needed to keep the electrn energr near the Rammauer-Townsend minimu. Xf the
presswe is left at the original 25 at, then 4 will increase from 0.6
to a 0.75. Xn thu case of C84 , however, the procedure of sec. XV is not
nearly as applicable as for f a constant. The problem should be done by
numrical simulation, which is beyond the scope of thi paper.
This wov was supported by the Naval Surface Weapos Center., Dah.grea
VIL.
References
1. R.D. rord, D. Jenkins, W.H. Lupton, and X.M. Vitkovitsky, Rev. Sci. Inst.
52, 694 (1981).
-44-
I2. 3.1. Nasar and N.H. Ioodsoft, Proceedings of the Sixth Symposium on
agineering Vroblm of Fusion Research, =333 Pub. No. 75c=1097-5-tMs
(1975), p. 316.
3. N.A. Barletta, Lawrence Lvezee, LTaboratoy Report UCRL-07288 (1991).
4. 3.3. Co=msso, a.7. Femsler, v.3. Schawrer, and x.L. ViJ kovitsky, Z
Tans. Planes Sc.- PS-10 241 (1962) and refereces therein.
5. 3.3- Kine, Z1 Trans. Pama Sci.., PS-10, 224 (1982).
6. M6 it. Hallada, P. Bletzinger, and W. F. Bailey, I= Trans. Plasma ScL.,
TSi--10, 218 (1982).
7. R.1. rernsler, D. Conta, and .M. Vitkovitsky, IzR Trans. Plasma ScL.
84 176 (1980).
S. 3..?. Corssor, 3.F. Fernsler, V.Z. Scherrer and I.Mo Vitkovitaiky,
I proceedings of 4th = Pulsed Power Conference, Albuquerque, m, x
eat. No. 83CS1908-3 (1983), p. 87. *
9. J.7. Lowl, L.K. ]line and r.V.3. eberlien, ibid. p. 94
* P. Bltzinger, ibid, p. 37.
10. A. M. Orishich,. A. G. Ponomaremo and V. G. Posulch, J. appl. Mech. and
Tech. Ihys. 1o.1, XX (1979).
11. Yu, D. rolev and A. P. Xhmzeev, Nigh Tamp. Phys. 13, 779 (1975).
12. K. McDonald, N. Newton, Z. 3. KMudhardt, M. Kristia en and A. r.
Gruenther, ZE Trans. Plsma Sci. PS-8, 181 (1980).
13. Richard Cecil Smith, Appl. Phys. Lett. 25 292 (1974).
14. S.C. Brown, Basic Data of Plasma Physics, (NIT Press, Cambridge, 1967),
pp. 87-94.
15. T.H. Lee, Physics and Engineering of High Power Devices, (MIT Press,
Cambridge, MA. 1975), p. 305.
16. 9.3. Douglas-Hamlton and Siva A. Mani, Appl. Phys. Letter. 23, 503
-45-
(1973).
17. D*E. Dcglas-Hamlton and 5.1. Mani, -7. Appl. Phys. 45, 4406 (1974).
13. 3.3. Adrews, X.G. Versiantsev, V.D. Pismnnyi. V.14. lolushkin,
A. 2. Kakhimov, M.A. Tlif sew, and E.G. Tzoeva, Soy. J. plaa Whys.3
(6), 770 (1977).
19. S.a. Geakin, Yuz. D. Kaeolev, V.G. Rahtkin, mid A.p. xhuzaew, Soy. J.
Plaw Phys. 7 (3), 327 (19a1).
20. a. Goldstein. Classical dMhnics, (Adison-W160l67, Peading, 1950), p. 41.
21. R. %T. Cctmsso, R. r. Fezusler, V. V. Schevr, and X. M. Vitkovitsky, NM
3~so Rep. 4975 (1982).
22. Dsluianhki and r. R. Yline, Air Force Wight Aeronautical Laboratoxy
Report ArWAL-TR-80-2041 (1960).
23. J. a. Basted, Physics of Atomic Collisions (American Elsevier, Now vork,
1972), pp. 306-310.
24. K.. G. Christophorome S. R. H-Uter, J. G. Carter, and IL. A. Mathis, App1.
Phys. Lott. 41p 147 (1982).
SFigure Captions
Fig. 1. Illustation of opening switch application for an e-bea controlled
S.diffuse discharge.
rig. 2. Two idealized generic 11"eraes of high-voltage short-yuLse generation
udg am ZS. xn case Ca) the UCm i used to charge the storage
Sndote. LO and to provide the fast opening. In case (b) the functions ,
of charging the inductor and fast opening av. accmlse usingf two
switches, the CS providing the fast opening in a pulse copressioa
* - scseme. "?CXX is the time for current to be diverted f rm SI to the mCS.
Fig. 3. atio of load pulse width, TL" to ZBCS condution t4, C, as a
function of load pulse width for system efficiencies of
n1-0.5 and 0.8.
Fig. 4. Plots of instantanseos switch resistance R(t] currents
lisO LM/ x - xsw + i, and load voltage V, normalized to the switch
resistance durLg condction RW; peak load current a - and peak load
voltage %x 0 ,-Tpectv*Iy-
-47-
I 0 taJ
X00
ik
-JL
-Mahiw
wzI
Figure
-46-00 P a
1-01
-. 4.
Iz I0 -
uA fAF
U)-49-
SI
€00
oc d
0 )~
a
L t I I I
0
C11
50
K) "
0 0i0
-50
eI
II
1. 200-
0
cr- in
w IL
< 0.4
z 0
'04
~100.
Figure 4
f4enra'auni Rtep it'97
:.7.
F-Ww
*1n I- Ic - -
0UEktdi BamC6r e w
~ #tf~s~4~pove rotpubic eleae. tsu~uno unlmit~?
V- 5
______________PO T.4sA
-52-,
SSCURITY CLASSIFICATION OF THIS WAGS& (371. 0..0 IM1,,.EA.INTRCTONREPORT DOCUMENTATION PAGE BEFORE COMPLETING FORM
I. aIPoNT" Plumes" GOVT ACCCSSION NO. 11. RECIPICT'S CATALOG NUMMEA
NRL Memorandum Report 4975_____________4. TITLE (a"d S-61itij A. TYPE OFr REPORT a 1111011O COV11:11o
Interim report on a continuingDESIGN OF ELECTRON-BEAM CONTROLLED SWITCHE~S - NRL Problem.
4. PICFOAING ORO. IMPORT NUNSCR
7- AQTNOR(oj 4. -CONTRACT OR GRANT kUMaSEt.)
R. J. Comiiso," R. F. Pernsler'" V. E. Scherer, andL U. Vitkovitsky
S. PSCRFORMING ORGANIZATION NAME AMC AQORESSN 10. PROGRAW LEMEN T. OE CT. TA~
Naval Research Laboratory AREA a ORIC UNIT NUNSIERS
Washington, DC 20375 61153N, RR0l 1094147-0878-0-3
it. conrioLliNso office NAME AND AG00555is OIL RP4RT OATS
Nava Reearh LaoraoryNovember 24, 1982Is. NoUldeRot ofAGEIS
Washington, DC 20375 2914LO4NIT0IN00 AGjRN MA4 NA ACONEs(it d#Jff@M. &M C41000010115 office) IS. 5ECURITY CLASS, (of hi Aleft
Unclassified-'I ECASFCATION OCOWNGNAOING
Its. W5turiUUTI STATEMEtNT (ad *~to Notion)
Approved f or Public release, distribution unlimited.
17. DISTRI11UTION $?TIN T (of the obeftetw iEee I .8". it Wieren.t hickn Reoe)
IL. SUPPLEMEN91TARY NOTCS
"Present address: JAYCOR, Inc.. '205 &. Whiting Street, Alexandria, VA 22302
Io. scay WORDS (CAWOo r0~00 d i1 fteo.for OW #~Etip bybl Meb erO)
opening SwitchRepetitively Pulsed OperationElectron BeamDiffus Discharges
IL AGSTNACT (Coothm. ON regaro 01d It M4960V Md litodftr &F 6.Ar.~
This Ppae reviews the priciples of operation of the electron-beam controlled switch (EBCS)and presnts a proceduue for its design. The ECS is compared to other switches which have pu-tentia application to repetitively pulsed, high power systems, and is found to have some substantialadvantages at high (> 10 kliz) repetition rates. arcuit requirements for the application of an EBCSto ETA/ATA like devices are outlined. A self-consistent formalism for optimum switch desi isderived. Ite formalism is applied to the previously outlined Circuit requirements using capacitive andcapacitive-inductive hybrid energy storage schemes. The required switches are readily designed.
DD I "' 1473 aOITINm of INO movs is onsSOLTa
6ECURITY CLASSIFICATION Of TWIS SAGIZ (111Iiv Date SAO~)
-53-
CONTENTS
1. INTRODUCTION .................................................. 1
UI. PRINCIPLES OF OPERATION.......................................................
IH. COMPARISON TO OTHER SWITCHES ............................................... 2
IV. DESIGN CRITERIA ...................................................................... 3
A. Capacitive Systems- Application of the Repetitive ClosingSwitch Mode........................................................................ 4
1. Electrical Characteristics for Closing Switch .................................. 42. Electron Beam Requirements................................................... 6
B. Inductive Systems- Application of the Repetitive OpeningSwitch Mode ........................................................................ 6
1. Hybrid Pulser .................................................................... 62. Inductive Pulser ................................................................. 63. Electrical Characteristics for Opening Switch-Hybrid Pulser .............. 9
V. SWITCH DESIGNS....................................................................... 10
A. Switch Parameters ................................................................... 10
1. Breakdown ....................................................................... 102. Efficiency .............................. .......................................... 113. Resistance........................................................................ 124. Closing and Opening Time ..................................................... 12
B. Switch Physics ........................................................................ 16
C. Design Procedure....................................................................IS
D. Design for Repetitive Closing Switch-Capacitive System ...................... 18
E. Design for Repetitive Opening Switch-Hybrid System......................... 20
VI. CONCLUDNG REMARKS .............................................................. 20-
VII. REFERENCES............................................................................. 24
VIII. TABLE OF SYMBOLS ................................................................... 26
-54-
DESIGN OF ELECTRON-BEAM CONTROLLED SWITCHES
I. INTRODUCTION
Recent investigations 4 into the phenomena associated with electron-beam (e-beam) controlleddiffuse discharges indicate potential applications for repetitive (> 10 kHz in a 0burst" mode), high power(-101 W) switching. Applications include use in both the opening and closing switch mode. Theswitch requirements may vary considerably depending upon the specific mode used.
There now exists a need for the development of these switches in several areas. One area offocus is the generation and propagation of intense charged particle beams.9 Here, the repetitive capabil-ity of the switch in the closing mode under high power operation is of critical importance. Anothercategory of general interest is the development of a compact, high power inductive energy storage sys-tem to replace convestional capacitive systems. t ° '1 2 This application has universal impact on manyareas where pulsed powei is required, including that of beam generation and propagation. In this casethe generation of a high voltage pulse (>200 kV) as a result of the fast increase of an opening switchresistance while the switch is being stressed by high electric fielis associated with the pulse is crucial.Depending upon the application, this opening switch may be repetitively pulsed. This would involve amore stringent set of requirements on both the switch gas and the switch e-beam driver than is the casefor single pulse switching. 4
In this report we review the principles of operation of the e-beam controlled switch (EUCS),highlighting its capability for rapid recovery-the essential requirement for repetitively pulsed systems.whether inductively or capacitively driven. Detailed explanations of the switch physics can be found inRefs. 5, 8, 13, and 14. The EBCS is compared to other candidate switches for repetitive, high powerapplication. Design requirements for two applications of this switch to devices with characteristics simi-lar to either the Experimental Test Accelerator (ETA) or Advanced Test Accelerator (ATA) 1 are thenoutlined. We next develop a formalism for switch design that combines the circuit requirements withthe switch physics. Using this formalism, which is backed by our present data base, 5. 8-13. 14 the neces-sary e-beam controlled switches are designed. The reader is cautioned that the design examples chosendo not necessarly make optimum use of the potential capabilities of the EBCS switch concept. A fullsystem study in which the EBCS is included as an integral part of the design process from inceptionwould be required for such a task. Rather. the examples chosen serve to illustrate the practicalengineering aspects of the switches, define some of the technical requirements necessary for their
A.. implementation, and indicate the steps that are necessary for a complete system design.
The results of this study indicate that EBCS's can be readily designed for repetitive (> 10 kHz),high power (200 kV, 20 kA) applications. Further research and development concerning gas chemistryand atomic physics, cumulative heating in the switch, and switch e-beam driver under repetitive, long 0conduction time (with respect to the load pulse width) conditions are needed in varying degreesdepending upon the specific switch application. The authors believe that such studies will lead tofurther significant gain in the practicality of EBCS's.
l1. PRINCIPLES Of OPERATION
When an electron beam is injected into a chamber containing a mixture of an attaching and anonattaching as the ionization of the gas produced by the e-beam pulse competes with attachment and
Manuomp submitted October S. 1 92.
-55- .S
recombination processes controlling the conductivity of the gas. The conductivity, and hence thedischarge current, is turned on and off in association with the e-beam pulse.
An important distinguishing feature of a switch based on this concept is the ability of the switchto open (cease conduction) under high applied voltage. This is achieved by using the electron beam tocontrol the gas ionization. To avoid avalanche ionization the switch must be designed so that the max-imum expected voltage across the switch is below the static self-sparking threshold (for sufficientlytransient voltages this requirement may be relaxed). As the discharge evolves, cumulative gas heatingalso must be constrained so that thermal ionization and. more importantly, local hydrodynamic reduc-tion in gas density do not significantly lower the self-sparking threshold. A self-sustained discharge isthus prevented. Under these conditions, the fractional gas ionization, and thus the switch resistivity, atany time is determined by the competition between ionization provided by the beam and the variousrecombination and attaching processes characteristic of the specific gas mixture, pressure, and appliedelectric field. A second important feature of this switch is the volume discharge property. This charac-teristic makes it possible to avoid excessive heating of electrodes and the switch gas (as well as lessenmechanical shock and minimize the switch inductance). All these features combine to permit thedischarge to return to its initial state of high resistivity very quickly once the source of ionization isremoved. Unlike an arc discharge, this transition can be accomplished rapidly in the presence of anapplied voltage. Some details or the gas chemistry and atomic physics associated with the switch opera--tion and their coupling to the switch circuitry can be found in Ref. 8.
Looked at from a different perspective, the EBCS behaves as a current amplifier. That is, thesmall (:S 1 kW electron beam current can control a large (-10 kA) switch plasma current. The ratioof these two currents is called the current gain, e. For e > I a substantial energy gain (i.e.. energydelivered to the load normalized to the energy dissipated) can be achieved, as discussed in Sec. V.
III. COMPARISON TO OTHER SWITCHES
Several summaries of high power closing and opening switches with potential application to repeti-tively pulsed systems exist.'"1 There are basically six switch types which emerge as candidates for ahigh power repetitively pulsed switch:9 (1) the low-pressure gas switch 17 (2) the surface flashoverswitch,18 (3) the thyrtron,'"1 (4) the high pressure spark gap, (5) the magnetic switch.' 9 20 and (6)the ECS.
The ongoing research for both the low pressure gas and surface flashover closing- switches hasyielded some encouraging results. The technology appears to be simple. At present, however, recoverytimes of only 100 lss have been observed for both devices with no applied voltage. Jitter may also be aproblem. Under repetitive burst operation, both switches may have to be cooled. For the surface flash-over switch, the insulator may additionally require cooling and the insulator lifetime is likely to be lim-ited.
The thyratron is a well developed device, however, the present limitations in voltage, current, andrisetime probably make it unsuitable for implementation in the systems of interest here. Further, thepower consumption and long warm up period for the cathode heater are disadvantageous.
The high pressure spark gap is the traditional choice for a closing switch of the pulsed power com-munity. It cannot be used as an opening switch. It is very simple mechanically and has associated withit an extensive data base. The problem areas are the substantial gas flow requirements and energy dissi-pation associated with this device when used in high frequency repetitive high power applications.Also, a trigger pulse amplitude comparable to the applied switch voltage is necessary for low jitter.
Magnetic switches are now a part of the ATA design. The potential advantages of such a switchare its simplicity, ruggedness, and lifetime. Questions still remain concerning overall size and weight,
-56-
evaluation of core materials, limiting output pulse duration, pulse compression ratios, trade-offs in 0number of stages in cascade, and core bias and/or resetting methods.
An understanding of the physics of the e-beam switch operation along with the results of recentNRL experiments leads one to conclude that the principal advantages of the EBCS are: (1) rapidrecovery (opening) of the switch when the electron beam ceases, and the consequent opening andclosing repetitive capability' '4 at high repetition rates; (2) negligible switch jitter,' (3) volume
* discharge resulting in low inductance and reduced switch current density (limiting electrode wear,switch gas heating, and mechanical shock); and (4) high power switching capabilitys (for limited pulsebursts).
Potential problem areas for this switch concept include the switch e-beam driver complexity, theeffect of cumulative heating on the switch's repetitive capability, and switch packaging. The e-beam 0
driver has modest peak power requirements on a per-pulse basis (200 kV, < 1 kA, < 108 W); however,it must provide a beam with the pulse shape and repetition rate required for the desired switchingcharacteristics. Thermionic e-beam generators, developed for excitation of high power lasers, now pro-vide 21"22 1-5 ;ss e-beam pulses with -100 ns rise and decay times at a 25 Hz repetition and at 10A/cm 2 . The repetition rate can probably be improved, e.g., if a lower e-beam current density is desired.With some further development, these devices may indeed provide the desired beam modulationcharacteristics. Thin film field emission cathodes with molybdenum cones 23 appear attractive because oftheir low control voltage (100-300 V), high current densities (-10 A/cm 2), and continuous operationcapability. An inductively driven electron beam system has successfully demonstrated two-pulse'burs" operation, generating two -I kA pulses with a 150-200 IS (limited by diode closure) interpulseseparation.2 In regard to the EBCS and the role of the e-beam source we quote from Ref. 9 (pg. 439),*The main application of this switch concept would seem to be as an opening switch. Since it has few
* rivals in this role, the complexity of the electron beam source becomes less formidable.'
The effects of cumulative heating in the switch gas are not well known. Preliminary experi-ments14 indicate that at least two-pulse operation is possible at a deposited energy density of -0.1J/cm3. These experiments also show a favorable (approximately linear) scaling of the maximum energydeposited before switch breakdown occurs with increasing switch ambient pressure. However, moreresearch in this area is needed.
Although a stand-alone switch packaging scheme has been oudined,2 a complete system designhas not yet been addressed. Such a design would be heavily dependent on the specific application andwould have to incorporate the switch as an integral part of the system ab initio. At present we see nofundamental limitation resulting from packaging considerations that would prevent integration of thisswitch into pulse power systems. For example, a compact, high pressure, a-beam controlled laser sys-ten with an e-beam generator (single pulse) compatible with severe optical requirements has been suc-cessfully assembled and operated. 2'
Much work is still needed to develop these concepts into viable repetitively operated switches. Insome areas (e.g., recovery) the EBCS has some very substantial advantages when compared to theother switch candidates. It is the only switch discussed here that can open under an applied voltageand, therefore, it is especially promising in applications which require high repetition rate (> 10 kHz)opening switches.
IV. DESIGN CRITERIA
In this section we outline switch performance characteristics upon which a switch can be designed.These characteristics strongly depend upon the specific application of the switch. Optimum switch per- Sformance requires that the switch be incorporated into a system design at its inception. However, toillustrate the practical design considerations for this switch concept we have chosen as an example the
-57-
requirements of the ETA/ATA inductive electron beam accelerator.' s In this context we describe therequirements for application of this switch to three energy storage schemes: capacitive, hybrid, andinductive. Depending upon the specific requirements, each scheme may involve, respectively, a morestringent set of switch performance and e-beam driver characteristics and a progressively greater extra-potation from the present data and technology base.
A. Capacitive Systems- Application of the Repetitive Closing Switch Mode
The e-beam controlled switch based on the principles outlined in Sec. II can be used at high powerlevels (-10I W) with capacitive energy storage as a closing switch, in applications where very fastrise-time pulses of short duration (< 100 ns) must be generated at high repetition frequency (> 10kHz).
1. Switch Electrical Characteristics for Closing Switch
One example of such an application is an output switch of the ETA/ATA. For definition pur-poses, the toad characteristics are assumed to be those summarized in Table I15
The characteristics displayed in Table I imply that the switch e-beam driver has sufficiently fastrisetime and that the working gas can respond sufficiently fast to the injected beam.
Two circuits which provide the required output for the load are shown in Fig. la and b. Both cir-cuits depend on use of switch S1 for charging. Typically, direct Marx charging as given in Fig. la canprovide a charge time for capacitor C,, of 'cH - I isec. Using the voltage step-up transformer (Fig.lb) has no deleterious effect other than to possibly increase Tc, to several microseconds.
Electrically, the switch must deliver a peak power P - 4 x 10' W to the load for each pulse.With > 10 kHz repetition rate (< 100 /sec pulse-to-pulse separation), the average power output of thepulse is P,v - P(rL/r,,) 1 x 100 W, where 'L is the load pulse width and r,, is the time betweenpulses. (For those scenarios where large numbers of accelerating modules are needed, the total averageoutput power can be as high as I x 109 W.) The power transport through the switch leads to some dis-sipation of energy in the switch. The amount that is dissipated depends on the pulse duration, thecurrent, and the switch voltage drop during conduction and is constrained to be less than the energydelivered to the load (see Sec. V.A.2). The opened state of the switch has negligible conduction in allcases considered in this paper.
Finally, the switch inductance must be limited to Ls* << VLR/IL ' 00 nH, as suggested bythe parameters in Table I. A schematic representation of the time history of the switch resistance, R,,.is shown in Fig. Ic.
TABLE I: SUMMARY OF ASSUMEDLOAD CHARACTERISTICS
(Closing Switch, Capacitive System)
Type of Load Electron beam diode
Load Voltage, V 200 kV
Load Current, IL 20 kA
Equivalent Impedance, RL 10 nPulse Duration, T L 40 ns
Pulse Rise Time, 'R 10 ns
Pulse-to-Pulse Time, r. < 100 js
Pulses per burst, n > 5
-58-
CAPACITIVE (CLOSING)
H.v. R0 S1 EC
C0 RL
(a)
H.V S ECxii
CO ] RL
TRANSFORMER
(b ) .
U, I_ .1
|-p
4 / IIs t PULSE. 2ndPULSE
c0 '--'iI , I
TIME 0(d)
Fig. I - Circuit diagram illusrating the application of an EBCS in the closing mode to a capacitive storage system. The circuit isshown both without (a) and with (b a step-up transformer. Also shown is a represnuttion of the switch resistance timebehavior (c).
-59-
"lll I I ... . . .. ... ... . .: . .. . . . . . .. ± _ _ .. .. .. ,_ 1. ,_.- .. .. ..
2. Eectmn Beam Requiremens
To obtain the necessary time variation of the gas conductivity, the electron beam injected into thegas must satisfy pulse shape, repetition rate, and current requirements. The ebeam current is set by anumber of factors, as discussed in Sec. V. The beam modulation was discussed in Sec. Il1. We assumefor the remainder of the discussion that an appropriate e-beam generator exists, or can be developed.
B. Inductive Systems-Application of Repetitive Opening Switch Mode
In projecting charged particle beam technology to deployable systems, practical size and weight ofthe system is one of the major considerations in the system design. The power source represents amajor component of any of the proposed systems. Because inductive storage offers a potential formuch more compact designs than those that would be possible with capacitive systems, there exists astrong incentive to develop the necessary inductive storage components. A repetitive opening switch isa fundamental component that must yet be developed. As in the repetitive closing switch development,e-beam control of gas conductivity offers a method that can be employed for repetitive opening switch-ing for high power pulse train production.
1. Hybrid Iuwr
Continuing with the scenario of an accelerator based on modular accelerating sections requiring 20kA, 200 kV, and 40 nsec pulses at a burst rate of > 10 kHz, the power supply circuits, shown in Figs. 2and 3, can be employed.
The circuit in Fig. 2a utilizes a pulse-forming line, represented as a capacitor C, to form a specificpulse shape required by the accelerator (i.e., rise time of 10 nsec and 40 nsec pulse duration). Theinductor, L,, is initially charged by a current, i,, through, in this example, an explosively actuated
switch 27"n (denoted by E) for a time r'oj- Note that some current generators, e.g., a homopolar, mayrequire earlier, additional stages of pulse compression. When switch E opens, the current is commu-tated (in a time rCoM) to the next stage which contains an EBCS oterating in the repetitive openingmode (denoted by the letter 0 in parenthesis). EBCS(O) is closed and conducting during rcoM. Afterthe commutation time, EBCS(O) is commanded to open. The opening generates a resistive voltage,and the pulse line C is charged for the charging time, rci - -rL. At the end of 'ca. EBCS(O) againcloses.
The output for the pulse line is an e-beam controlled device already described in the previous sec-tion, i.e., an EBCS operating in the repetitive closing mode (denoted by letter C in parenthesis). Fig-ure 2b is a schematic representation of the time histories of the resistance of EBCS(O) and EBCS(C) ofFig. 2a.
Thus, the circuit in Fig. 2a employs two repetitive EBCS's. The EBCS(O) has a long conductiontime, equal to r c - (ro, - rcq), and a short nonconduction time, T NC - rCq, during which the capaci-tor C is charged. The EBCS(C) is identical in its operation to the switch discussed in Sec. IV. A.
2. Inductdve Pulser
A more compact and simpler pulser would result if a circuit in Fig. 3a could be used. The pulserrepresented by this circuit is purely inductive, eliminating any need for capacitive storage. However,this scheme represents the farthest extrapolation of the present data and technology base thus far con-sidered. Figure 3b is a schematic representation of the time history of the opening switch resistance fora purely inductive system of Fig. 3a. The EBCS(O) in Fig. 3a must conduct for the period betweenpulses, ta,, and open repetitively for a much shorter period, -1 L. The output pulse shape, however, - -
imposes stringent performance requirements on EBCS(O), in terms of pulse risetime and duration.
-60-
- - ---- -
HYBRID (OPENING/CLOSING)
LO LSW
E EBCS(O) 11111 C EBCS(C) 11111
JRL
(a)
S
z rHr0 M+-rCH--izIL,a. / I ! ", p'£ 0
3TIME
E OPENz I- l'pp '
U) I
0 V T IM E -
Fig. 2 - Circuit diagram illustrating the alpplication of an EBCS in the opening mode. ESC$4O). anld closing mode.EBC$(C). ror aI hyb~rid (capecitive-inductive) inductive storage system (a-). Also shown are representations or" the timebehalvior or" both switch resistances (bl.
-61-
INDUCTIVE (OPENING)
(a)LLSW
(a)
(b)
0. -- I.,I-, p
o I
I PULSE 2 nd PULSE
t TIMEE OPEN
(bD)
Fit. 3- Circuit diagram illusurating the applicton of an EDCS in the opening mode for a purely inductive energy storagesystem (a). Also shown is a representation of the switch resistance time behavior (b).
-62-
I
Although this scenario is conceptually the simplest, it may be too speculative to assume that a sin- Sgle switch can achieve an opening time of less than 10 ns and a conduction time of greater than 10 As.Thus, this design is not considered for detailed engineering at present. The option of using a purelyinductive system is attractive and, therefore, suggests that future effort should be invested in develop-ment of switches with opening times of -10 nsec. We note that response times of < 10 ns are theoret-ically possible5; however, an e-beam driver with the necessary waveform capability needs to bedeveloped.
3. Electrical Characteristics.for Opening Switch-Hybrid System
The energy associated with a single pulse for a single module delivering 20 kA, at 200 kV, with a40 nsec pulse width, is very small: E, - 160 J. Considering a 10-pulse burst, -2 kJ is required as aminimum to be stored by the inductor. Taking 30% efficiency of conversion from stored to pulseenergy, -6 kJ must be handled by a storage system for charging one pulse line. This projects to - 107
J for a 1000 module accelerator. In considering the engineering of a switching system, one must knowthe number of modules to be powered by one switch. At this stage of switch development, where littleexperience with a practical switching device exists, the choice is dictated as much by the need to main-tain a reasonable experimental set-up as by the lack of certainty related to the ultimate application. Areasonable first attempt, that would uncover most of the problems of a design, would be a switch designfor a 10-module pulse train generator limited to 5-pulses. Circuit, current, and voltage parameters arederived below for the EBCS(O) switch of Fig. 2a.
Using the same final load pulse parameters as in the preceding section, the single pulse line capa-city is C - 2E11 V2 - 8 nF; i.e., in 10-module operation 80 nF must be charged to 200 kV. Choosing a5-burst pulse train for an initial switch design (and for experiment design to test the switch), the totalstored energy in the ten pulse lines is E2 - 9 kJ. We assume a 15 jAs interpulse time (-70 kHz) and a 42 -9s charge time. The energy stored in the inductor must be about E3 - 3 x E2 - 24 k, to accountfor the inefficiencies (-70%) associated with the circuit and fuses shown in Fig. 2a. This amount ofenergy suggests that low voltage capacitors can be conveniently used as the source of current (shown inFig. 2a) to charge the inductor in small laboratory experiments. As previously stated, the time requiredto pulse-charge the capacity C, is taken to be rci' - 2 1sec. This is consistent with compact water-dielectric pulse line requirements. The current needed to charge a capacitance to a given voltage, V is Si. - dQ/dt - CV/1rH. For C - 80 nF, V - 200 kV and rcj - 2 /s, !. - 8 kA. The storage induc-tor L. - 2 E3. - 750 1sH. This choice of inductor presents no structural or electrical design prob-lems.
To summarize, the EBCS(O) of Fig. 2a charging the ten pulse-lines must accommodate the circuitperformance characteristics shown in Table II.
TABLE II: Summary of AssumedCircuit Characteristics
(Opening Switch. Hybrid System)
Type of Load Capacitive
Peak voltage, VL 200 kV -
Load current, IL -. 8 kA
Conduction time, rc - (r,, - tci,) 13 /AS
Nonconduction time, ,Nc - rCK 2 ljs
*. Storage Inductance, L, 750/iH 0
Pulses per burst, n 5
-63-
V. SWITCH DESIGNS
In this section We apply the results of experimental and theoretical research at NRL and theswitch performance criteria described previously to obtain specific switch designs. First we review theimportant switch characteristics and their relationship to the system. A simple quantitative model forthe switch physics along with scaling relations is then presented. A design procedure is outlined thatself-consistently incorporates the switch electrical characteristics with switch physics. Finally, the values
fl of the switch parameters are obtained and the closing and opening switches are designed.
A. Switch Parameters
The following switch parameter considerations are of importance for opening and closing switchdesigns.
1. Breakdown
If the electric field across the switch exceeds the static breakdown field at the ambient switch pres-sure, the switch may go into an arc mode, which is undesirable because it prevents the switch fromopening. This leads to the constraint that
VL -. SB -a1)I)
where VL is the maximum expected voltage across the switch (i.e., the load voltage), Ego is the staticbreakdown field at atmospheric pressure P., S8 < 1 is a dimensionless safety factor, and I is the switchlength. We have assumed that Eg - El (PIP,), where E,9 is the static breakdown field at pressure P.This condition can be relaxed somewhat for transient. <1I 1&s, pulses.
Additionally, cumulative heating of the gas must be sufficiently constrained so that any localreduction in switch gas density does not significantly lower the self-sparking threshold. Energy is depo-sited in the switch from the following processes:
(1) During the time the switch changes from a conducting to a nonconducting state, i.e., duringj the opening time ro. the current in the switch is finite while the switch resistance is large. Thus, the
resi tive heating during this time may be significant. This cumulative heating is estimated by
H0 - kols. VLnro. (2)where Js* is the maximum switch current, Y1, is the maximum switch voltage (load voltage), M is thenumber of pulses, ro~ is the opening time, and ko 4 0.5 is a dimensionless constant which appropri-ately averages Us*~ VL) during ro (see Sec.V.C). Because ro > rSR, where the switch rise time, rsR,is the time for the switch to change from nonconducting to conducting, we will neglect the energy lossduring the switch closing phase.
(2) During the total time the switch is conducting, ni'c, there will be some resistive heating.
The total energy deposited in the switch as a result of this process is approximated by
HC - IJ*Rs~n'rC. (3)
Here Rs* is the switch resistance during conduction.
(3) When the beam is injected, a fraction of the beam energy will be directly deposited in theswitch gas as a result of inelastic collision processes. This energy deposition is estimated by
Hb - kh~b VbnrC, (4)
-64-
5 where b and Vb are the beam current and voltage respectively, and kb 4 1 is the fraction of the beamenergy deposited in the switch (determined from Vb, P, and 1). The conduction time rc is also the e-beam pulse duration.
To properly constrain cumulative heating in the switch, we must have
Ho + Hc + Hb < WSAI. (5)1 Here Wg is the deposited energy per unit volume at which the self-sparking threshold is altered and A
is the switch area. As shown in Fig. 4, W, scales linearly with ambient gas pressure p. 4 We thereforetake Wa - (PIP,) W1, where Ws - 0.15 J/cm3 is the breakdown energy density at atmospheric pres-sure. S.14.2 Equations (2)-(5) can thus be combined to give
koIsD VL + II RsO, + kb4 VbJ n c - sN[ A(PI), (6)
where sH 4 1 is a dimensionless safety factor.
2. Efficiency
The switch energy gain, f, is defined as the ratio of the energy delivered to the load to the totalenergy used in making the switch conduct. Thus, applying the same arguments used in obtaining Eqs.(2)-(4) we have
ILVL L (7)
kols VL + I.R,. + b Cwhere IL is the load current and T L is the load pulse width e.g., rL - CM for the opening switch ofthe hybrid system in Figs. 2a and 3b, while for closing switch TL - 7c. see Figs. I and 3a. In all cases.Is*. - IL (although not always at the same time).
The switch efficiency, -q, is related to f by,, - f/(C + 1). Thus, to attain a high efficiency,
Se> 1. (8)Upon substitution of IL - Is* and using the definition of current gain, e = Is*/4. Eq. (7) becomes
(?L/,"c) * 9
((vo1,r c)ko + (Js,.Rs/VL)] a + 1 (9)
Recently obtained measurements of a as a function of percent 02 in N, for 1, 2, and 3 atm at anapplied EIP - 10.5 V/cm-torr with an e-beam current density of 5 A/cm are illustrated in Fig. 5.Because of the short duration (200 ns) and long rise time (100 ns) of the e-bear,, 'ulse used in thesemeasurements, the values of c < 15 at low 02 concentrations (<20%) should be Lonsidered a lowerbound) 4
Equation (7) may be rearranged to yield
-'I/L VLrL - kols* VLro + Is*1 YLC + l1 Vrc. (1)To realize an energy gin, i.e., for f > 1, each term on the right hand side of Eq. ( 1) must besufficiently less than 1L VLrL. Thus we set
kolsu VLIO 9 0L VLIL, (12a)Is,.Rsc gcIVL" rL, (12b)
-65-
and 4 cI b VhTc - fh'L VL1L,(2)
where go, &c. gb < 1, such that
r' oI + S + gc+ <1. (13)
Equations (12a)-(12c) can be rewritten (with Is* - IL) as:
TO- (go/ko)rL, (14a)EC- .C S ( E/po) "L (14 )P TC
and
. fr c) (14c)VLI TL I.
where EC - 1si Rs/I is the electric field across the switch during conduction.
3. Resistance
The switch resistance during conduction is related to the switch gas resistivity during conduction,P., by
IRs- p, '. (15)
Plotted in Fig. 6 is the resistivity at peak switch current as a function of percent 02 in N2 for 1. 2and 3 atm at an applied E/P - 10.5 V/cm-tort for an e-beam current density of 5 A/cm2. Values of300.400 fl-cm are typical for <20% 02. We reiterate that because of the short duration (200 ns) andrelatively long rise time (100 ns) of the electron beam used in the experiment, the values of p. in Fig.6 for low concentrations of 02 can be considered as an upper bound (see Ref. 14).
By recalling that Fc - Is*Rsa]Iand using Eq. (1) to eliminate I we may rewrite Eq. (14b) as
Rso M-9 IL VI TL (16)
for the switch resistance during conduction. This equation and the condition gc < 1 is essentiallyequivalent to the requirement that the characteristic LIR time of the system be long compared to theswitch conduction time, so that the current will not resistively decay from the system.
4. Closing. and Opening 77mes
The characteristic time scale for the switch to change from nonconducting to conducting is theswitch closing time, -rsj. This time is important in closing switch designs (Fig. 1) and for our experi-ments has been limited by the beam rise time (-100 ns) and the circuit parameters. When fast rising(<$ ns) beams were used, rise times as short as 2 ns have been observed. 4
The characteristic time scale for the switch to change from a conducting to a nonconducting stateis the opening time, vo. On this time scale the switch current decreases and switch resistance increasesby orders of magnitude.
-66-
_ - . . . . • - _ . .. . _ . _ . . . . . . . . . . . . . . . .
0.4 il f7 1 1I1, 1J l
E/P 11 V/crn-torr.
0.310% 0 - 90% N2
In
~0.2-
00
GENKIN, etol.
0 I PP(AT M) 23Fig. 4 - Plot or deposited energy density required ror breakdown, WO, as a function of ambient pressure. P fr Air and a
10% 02-90 N, mixtture. Genkin. et &L is given in Ref. 29.
-67-
E/P~ 10.5 V/cm-torr
20. Jbu 5A/cmZ-1 atmn-U 2otm
A3atm
zwI
0 10 20 30 40 50. 60 70 80 90 1000% OIN N2
Fig. 5 - Plot of CUMren Pain, 4. as a runcion of(h concentration in N2 at LIP -10.S V/cm-tory and
4-5 A/cm2 ror ambient pressure P-1. 2 and 3 aim.
-68-
2500 - E/P-s 10.5 V/cm-tort- J1, 5A/cm2
I aft2000 I
S1500 T
1000 /
500-
0 10 20 30 40 50 60 70 80 90 100% 02 IN Na
Fia.6 - Plot ofresisivity atpeak switch current, ., a unction of0concntmtion in N2 at E/P- 10.5V/cm-torr and J14 5 A/an2 rot ambient Pressure P - 1, 2 and 3 atm.0
_7
Values of To obtained from single pulse experiments " 13.14 are plotted in Fig. 7 as a function of 02
concentration in N2 at 1, 2 and 3 atm with an applied E/P - 10.5 V/cm-torr for an e-beam currentdensity of 5 A/cm2. For this plot, rO was estimated from the inductively generated voltage, V, byO = LAI/ V,. where Al is the change in the system current and L is the system inductance. Typically,
values of rc 0 300 ns, limited by the beam decay time (-100 ns) are observed.
B. Switch Physics
The continuity equation for the switch plasma electron density, ri., can be expressed simply as14
--n . S. P - n, (17)
where J, ME Ib/A is the e-beam current density and T, is the characteristic loss time for the switchplasma electron density. Several to very many r,, periods are necessary for the switch to open, depend-ing upon the dominant mechanism responsible for switch plasma electron loss. Thus, we have
o M kpT. (18)where k, 5 for an attachment dominated switch or k- 102 - 103 for a recombination dominatedswitch.$
S. is a beam ionization parameter given by
dE. (19)
where e is the electronic charge, e.- 35 eV is the energy required for ionization per electron-ion pair.and (dE/dX). =a 3 keV/cm is the energy lost per unit length for the beam electrons at I atm.
The plasma density is related to the switch plasma current density through
Js sa / JsA - nev, (20)
where v is the electron drift velocity,
v - E. (21)
Here I is the electron mobility30 and E is the electric field across the switch. Thus the resistivity duringconduction is given by
O " Fs _ (en-A)-'" (22)
Substituting n. from Eq. (22) into Eq. (17), and noting that at equilibrium dnldt - 0, Eq. (17)becomes
Jbp,?p - J', (23)
where f, a eSP is essentially a constant for a given gas composition. For most uases f0 105 - 106cm/V-s. Note that Eq. (23) indicates that for a given beam current and gas with a constant f., there isa *trade-of between resistivity and opening time.
Finally, using the definition of a, Js, and Jb, Eq. (23) can be expressed as
.. - Ecfo. (24)
The relations derived in this section are used to relate the switch physics to the switch circuit charac-teristics outlined in Sec. V.A.
-70-
m. . .. -, h. , . .. . . ..... ... ... .. . . ... .. .. ... ... . . .. . ... ... .. . ... ... . . .
Soo
700E/P a 10.5 V/cm-torr
60Jb x5A/cm2
-0 latina- 2atm
500 3ot
300~
200'
0 10 20 30 40 50 60 70 80 90 100%I 02 in N2
Fig. 7 - Plot of switch opening time, r,. as a function Of 02 concentration in N2 at ElIP 10.5 V/cm-toffAnd 4. S A/cm2 ror ambient pressure P -1. 2 and 3 aum.
-71-
C. Design Procedure
In this section we combine the switch circuit requirements with the switch physics to develop aself-consistent procedure for obtaining the switch gas composition and pressure, the switch length, andswitch area (radius) for a given switch gain.
First we obtain the factor ko in Eq. (2), which when multiplied by the power delivered to the load -,gives the average power dissipated by the switch during the opening phase. During the time the switchis undergoing a transition from conducting to a nonconducting state (io), the switch current is decreas-ing and the switch resistance is increasing. During 7L, the switch current changes by 'sw (-IL) andswitch voltage changes by VL; thus we define
ko <IS VL > (25)
where < Is* > and < Vs > are the average values of the switch current and voltage during To.
The time history of the voltage across the switch will depend upon the load voltage time behavior(see Fig. I and 2).. For loads where the load voltage rises to its peak value in a time =ro (i.e., resis-tive or inductive loads), <Is*> =- 1/2 Is* and < Vs5 > =, 1/2 VL so that
ko =z 1/4. (26)
For a capacitive load (Fag. 2a), the load voltage reaches its maximum in a time rt. - T CH. In this case,we still have < is* > 1/2 Is*. However, < V3s > = < is* > ro/2C - VL7O/ 4 "L. Thus, for acapacitive load,
/o = 1/8 (ro/71). (27)
Once ko is known, the opening time. "o can be computed from Eq. (14a) for a choice of go.Knowing rro, Eq. (18) can be used along with data on attachment and recombination rates, to make ajudgement on whether the switch should be attachment or recombination dominated, thus suggesting aspecific gas composition. The value obtained for r o must be also consistent with the circuit require-ments. If not, the choice of go must be modified. l
We now set out to compute the switch pressure. Beginning with Eq. (14b) and substituting forAC from Eq. (24), for * from Eq. (14c), for r from Eq. (18), and for ro from Eq. (14a) we obtain
F - SC9b0f TL El I VL. L (28)" kok, P, ~t'cJ'
The factors go, :, and gi are chosen so that P can be made small consistent with the constraintof Eq. (13). An optimum choice for go, Sc, and gb can be obtained by using the method of Lagrangeundetermined multipliers,3
1 with the result that go gc = g6 - g (note that in some cases ko and k,can depend on the g factors). For example, if a f of 3 is chosen then g = 0.1 and - 0.75.
Once P is known, Eq. (1) gives the switch lengthi- . (29)
sJJP(E;/,,.)"
The value chosen for Vb in Eq. (28) can be combined with P and Ito compute kb of Eq. (4). Note thatin order for the beam to traverse the switch length V will depend on the product P4 as does VL.Therefore, one can show that typically Vb = VL, with kb " 0.3.
-72-
Sunni
The electron beam current to be injected into the switch can be obtained from Eq. (14c) and thedefinition of current gain, a:
Vb TC (30a ls*/4" -J
By substituting Rs# from Eq. (16) and 4 from Eq. (30) into Eq. (6) we arrive at the required switcharear m
A sW/P*pI 1+ - 1- + ki, I. (31)
The area thus computed insures that the switch will be large enough that the total energy per unitvolume deposited in the switch is less than the deposited energy density required to lower the break-down threshold. "
The e-beam generator requirements are determined from Vb, 4, and A. 16 is computed from Eq.(30). The e.beam generator must actually provide a somewhat higher current than 4 to account forcurrent lost to the structure supporting the vacuum-high pressure interface. The e-beam generatormun " supply a bem of area A.
D. Deskn for Repdtivo Clsing Switcb-Capcldve System
Taking the circuit parameters described in Table I. we apply the results of Sec. V.C to arrive at aswitch design for the repetitive closing switch of Fig. 1.
For this case we choose ., " I5 js, safety factors sg - 0.75, sm - 0.5, and Vh - 200 kV. Aspreviously stated (Table!!) Is* - IL - 20 kA, n -S. VL - 200 kV, and 5.L - 40 nS. Choosing - 2ives so = S - 0.17. With rc -rL - 40 ns, Eq. (30) requires e = 6. Because the load is notcapactive we substitute kO from Eq. (26) into Eq. (14a) to compute to:
to =1 4 o5.L = 30 as. (32)
It is not disturbing that to - TL because the pulse line does most of the load pulse shaping; therefore,the major requirements are that t 0
< 5 and that r5s ;S r.#. The second requirement can be easilymet for an .. beam risetime < to.' We choose k, in Eq. (18) to be -5 (attachment dominated) so thatthe plasma electron decay time rp - 6 ns. This can be achieved with a N2-0 2 gas mixture of 4:1.5
The switch pressure is given by Eq. (28) with ES = 20 kV/cm, f. - 2 x 105 cm/V-s for N2-02and ko - 0.25 (Eq. (26)):
P = 1700 Tof = 2.3 atm. (33)
The switch length is then (Eq. (29))
I M5 .9 cm. (34)
These. P, and V values are consistent with kb = 0.3.
Equation (31) is then used to obtain the switch area (radius)
A m 2590 cm 2
(r = 29 cm). (35)
giving a switch current density of Js* = 8 A/cm2. The -beam current density is thus Jb = 1.3A/cm2 . From Eq. (16) we compute Rs* = 1.7 (1 and from Eq. (IS) we obtain P. - 745 (1-cm. 0
-73- 0
Table II is a summary of the values of the switch parameters which, along with the requiredswitch inductance (4 100 nH), completely characterize the switch.
The e-beamn generator is required to deliver >20 J/pulse with a very fast rise and decay time.
The e-beam current density, however, is modest, - A/cm 2.
E. Design for Repetitive Opening Switch-Hybrid System
Taking the circuit parameters described in Table II, we apply the results of Sec. V.C to arrive at aswitch design for the repetitive opening switch EBCS(O) of Fig. 2.
For this case: i, - 15 1s, TNC TL - 2 , Is* - IL - . - kA, n- 5, and VL - 200 kV(Table I1). We choose S8 - 0.75, S - 0.5, and Vh = VL - 200 kV. Setting f - 4 gives (Eq. (13))so =I c = b - 0.08. With c =a rM - T ., Eq. (30) requires e =a 80. Because of the capacitive load,we substitute ko from Eq. (27) into Eq. (14a) to compute 'o:
'"o - TL == 1.6 ls. (36)
Since r - TL, a substantial fraction of the final charge voltage for the capacitor will be attained duringthe opening time. This presents no serious problems, because the major requirement is that the capaci-tor be charged in a time < r... We choose k, in Eq. (18) to be -5 (attachment dominated) so that-P -z 300 ns. which can be readily attained using N 2 with a small (-1%) admixture of 02.
The switch pressure is given by Eq. (29) with f, - 2 x 10' cm/V-s for N 2 and ko - 0.1 (Eq.
(27)):
P - 5230 Torr 7 atm. (37)The switch length is then (Eq. (29))
I - 1.9cm. (38)
Equation (31) is then used to obtain the switch area (radius):
A - 3000 cm2
(r, 31 cm). (39)
giving a switch current density of Js* == 2.6 A/cm2. The e-beam current density is thus Jb = 0.03A/cm2. At this point the switch resistance can be computed as in Sec. V.D.
Table IV summarizes the values of the switch parameters, which, along with a requ;red switchinductance (-5 ILH), completely characterize the switch.
The e-beam generator requirements are quite modest with only 2.6 J/pulse needed. This reducesthe complexity of the e-beam pulse modulation. The pressure requirement may be relaxed by as muchas an order of magnitude by choosing a gas with a higher mobility during conduction. e.g., a mixture ofAr with 02 or CH 4 with an attaching gas (C0 2). Alternately, use of these gases will increase the switchefficiency if the switch pressure can remain high, as is evident from Eq. (28). This is an area wheremore research is needed.
VI. CONCLUDING REAR6
We had three principal objectives in the present paper. The first was to review the principles ofoperation of electron-beam controlled switches, emphasizing the ability of these switches to recover andopen rapidly under high applied voltage. This rather unique capability makes the EBCS attractive, par- 4ticularly when compared with other switch candidates, as either a single pulse opening switch or as aswitch for repetitively pulsed systems with high repetition rates.
-74-
LI
S Table III - Characteristics of Closing Switchfor a Capacitive Pulser
A. Switch Parameters
gas N2:02 - 4:1* pressure, P 2.3 atm 0
length, 1 5.9 cmAre (radius). A (W' 2590 cm2 (29 cm)Current Density, If*. a A/cm2
energy gain (current gain). 14i) 2(6)resistance during conduction 1.7 nvoltage drop during conduction 34 kV
B. E-Beam Parameters
voltage, Va1 200 kV
current. 1 3.5 ktArisetime <rp <2 m;
decay time ;SrOns
Table IV - Characteristics of Opening Switchfor a Hybrid Pulser
A. Switch Parameters
* Gas N2:02 - 991MPressure P 7 atmnlength, 1 1.9 cmArea (radius), A (W' 3000 cm2' (31 cm)current density, J4w 2.6 A/cm2
Gain (current gain). 1(4) 4(80)Resistace during conduction, Rse 0.3 a1Voltage drop during conduction 2.4 kV
B3. E-Dean Parameters
Voltage, V1 200 kV
current.!4 100 Arise and dey tme< r, :IOOns
-75-
The second objective was to present a formalism for switch design. The formalism givenemphasizes the overall energy-transfer efficiency of the switch, the fundamental circuit requirements,and the switch physics. Our analysis indicates that efficiencies of - 80% should be achievable for theexamples chosen. We point out that these efficiencies are conservative in that the switch designs utilizethe well known but nonoptimum gases, N2 and 02. Significant improvements in switch energy gainshould result if N2 were replaced by a gas such as CH 4 with high electron mobility, or if 02 werereplaced by a gas such as C2F6 in which the attachment rate increases rapidly with applied field.
The third objective was to illustrate the capabilities of the EBCS in a parameter regime of interest,both as an opening switch for inductive energy store, and as a fast closing switch. In the exampleschosen, the opening switch would conduct for -15 s and open in -1 Ass; the closing switch wouldclose in -5 ns and conduct for -50 ng. The switch size in both cases was roughly one-half meter indiameter by a few cm in length.
Although single pulse operation has not been stressed in this report, it is important to note thatthe EBCS is capable of providing opening times substantially shorter than what can be achieved withwire fuses, the only available alternative for fast opening applications. New developments in erosionswitches32 may provide a fast opening time, but the conduction times are limited. Therefore, from thestandpoint of the general development of inductive storage, the EBCS may make a significant contribu-tion.
We stress that the performance of any EBCS is primarily limited by the desired switch efficiencyand by the ambient switch gas pressure, as indicated by Eqs. (13) and (28). Choosing nominal valuesfor ko , Ac, ss, E5/P,, recalling the switch efficiency q - f/(f + 1), and assuming Vb =Z VL, Eq. (28)can be rewritten as
I - 71O.6 P fL. (40)
The reader is reminded that f, is loportional to the electron mobility. Equation (40) indicates that asi is made large, 7c/rL becomes smaller. This type of relationship will exist for any opening switch,because there are always losses associated with the conduction phase.
As an example, we choose an energy efficiency of 80% (7 - 0.8), fo - 2 x 106 cm/V-s, andassume a maximum practical gas pressure of 10 atm. Equation (40) then becomes
11"_1 < LAl~ ns). (41)
This condition precludes sub-nanosecond load pulse widths. That is, for a closing switch (TL -- Tc),L Z I ns. In the case of an opening switch this condition additionally limits the maximum allowed
conduction time, 'c. For instance, for a load (open) time of 7T - 1 .s, the switch conduction time is
limited to rc 4 30 As. By reducing the switch efficiency to 50%, however, a conduction time of7, 4 250 As could be realized. Equation (41), along with the similar equation for 50% efficiency, isdisplayed graphically in Fig. 8.
This report has not addressed all issues of concern to EBCS design and operation. We have notdiscussed, for example, the impact of electrode sheath effects, current conduction by ions, or vacuuminterface problems. Nonetheless, none of these additional issues should seriously compromise thedesign performance. This statement is supported by experiments performed at NRL and else-where. 1
-4
,64 . 13. 14 These experiments, which cover a wide parameter space, demonstrate that the EBCS
does indeed open under high applied voltage and that the simple models used are adequate to explainthe observed switch behavior.
-76-
-i 4
- i03...1, I 1,
(r- r O7= 0.5
P I0 atm S
0 fo x 106 cm/V-s
TC 17=0.8.rc
10*o -
10-3 0-? 10-1 0
LOAD TIME,L(/Ls)
Fit. 8 - Plot of r'CL as a function of rL based on Eq. (28) for switch efficiences. vp or 0.5 and 0.3
-
-77-
The authors wish to thank J. M. Cameron and H. Hall for their expert technical assistance in thedesign and operation of the experimental apparatus. This work was supported by the Naval SurfaceWeapons Center, Dahlgren, VA.
VII. REFERENCES
1. R.O. Hunter, Proc. International Pulsed Power Conference, Lubbock, TX (1976), 1C8:1-6.
2. J.P. O'Louglin, Proc. International Pulsed Power Conference, Lubbock. TX (1976), 111C5"1-6.
3. B.M. Kovachuk and G.A. Mesyats. Sov. Tech. Phys. Lett.. 2, 252 (1976).
4. K. McDonald, M. Newton, E.E. Kunhardt, M. Kristiansen. and A.H. Guenther, IEEE Trans.Plasma Sci. PS-8, 181 (1980).
5. R.F. Fernsler, D. Conte and I.M. Vitkovitsky, IEEE Trans. Plasma Sci. PS-a, 176 (1980).
6. P. Bletzinger, Proc. of Workshop on Repetitive Opening Switches, Tamarron, Colorado, TexasTech University Rept. (April 1981), p. 128.
7. Lawrence E. Kline, Proc. of Workshop on Repetitive Opening Switches. Tamarron, Colorado,Texas Tech University Rept. (April 1981), p. 121.
S. R.J. Commisso, R.F. Fernsler, V.E. Scherrer, and I.M. Vitkovitsky, IEEE Trans. Plasma Sci., tobe published (1982).
9. J. Benford, Physics International Co., San Leandro, CA, private communication (1980).
10. J. Salge. V. Braunsberger. and V. Schwartz, Ist International Conference on Energy Compressionand Switching, Torino (1974).
11. I.M. Vitkovitsky, D. Conte, R.D. Ford, and W.H. Lupton, NRL Memorandum Report 4168(1980).
12. L.D. Ford, D. Jenkins, W.H. Lupton, and I.M. Vitkovitsky, Rev. Sci. Inst. 52, 694 (1981).
13. V.E. Scherrer, R.J. Commisso, R.F. Fernsler, L. Miles, and [.M. Vitkovitsky, Third InternationalSymposium on Gaseous Dielectrics, Knoxville, TN (1982).
14. V.E. Scherrer, R.J. Commisso, R.F. Fernsler, and I.M. Vitkovitsky, 1982 Fifteenth Power Modu-lator Symposium, Baltimore, MD (1982).
15. W.A. Barletta. Lawrence Livermore Laboratory Report, UCRL-87288 (1981).
16. T.R. Burkes, J.P. Craig, M.O. Hagler, M. Kristiansen, and W.M. Portnoy, IEEE Trans. Elect.Dev. ED-26, 1401 (1979).
17. E.J. Lauer and D.L. Birx. Third IEEE International Pulsed Power Conference. Albuquerque, NMIEEE Cat. No. 81CH1662-6 (1981), pg. 380.
18. I.D. Smith. G. Lauer, and M. Levine, 1982 Fifteenth Pulse Power Modulator Symposium. Ral-timore. MD (1982).
-78-
19. D.L. Birx, et al., 1982 Fifteenth Pulse Power Modulator Symposium. Baltimore, MD (1982).
20. D.L. Birx, E.J. Laurer, L.L. Reginato, D. Rogers, Jr., M.W. Smith, and T. Zimmerman, ThirdIEEE International Pulsed Power Conference, Albuquerque, NM, IEEE Cat. No. 81CH1662-6(1981), p. 262.
21. J.E. Eninger, Third IEEE International Pulsed Power Conference. Albuquerque, NM, IEEE Cat.M No. 81CH1662-6 (1981). p. 499.
22. A.W. Freidman and .E. Eninger, Third IEEE International Pulsed Power Conference, Albu-querque, NM, IEEE Cat. No. 81CH1662-6 (1981), p. 519.
23. C.A. Spindt, 1. Brodie, L. Humphrey, and E.R. Westerberg, J. Appi. Phys. 47, 5248 (1976).
24. B. Fell, R.J. Commisso, V.E. Scherrer, and i.M. Vitkovitsky, J. Appl. Physics 53(4). 2818 (1982).
25. L.A. Miles, E.E. Nolting, I.M. Vitkovitsky, and D. Conte, IEEE Conference Record, 1980 Four-teenth Pulse Power Modulator Symposium. Orlando, FL (1980). p. 68.
26. N.W. Harris, F. O'Neill. and W.T. Whitney, Rev. Sci. Inst. 48, 1042 (1977).
27. R.D. Ford and I.M. Vitkovitsky. NRL Memo Report 3561 (1977).
28. D. Conte, R.D. Ford, W.H. Lupton. and I.M. Vitkovitsky, Proc. 7th Symposium on EngineeringProblems in Fusion Research, Knoxville, TN, IEEE Pub. 77CH1267-4-NPS (1977), p. 1066.
29. S.A. Genkin, Yu. D. Kosolev, V.G. Rabothin. and A.P. Khuzeiv, Sov. J. Plasma Physics 7(3), S327 (1981).
30. S.C. Brown, Basc Data of Plasma Physks (MIT Press, Cambridge, 1967), pp. 87-94.
31. H. Goldstein, Classical Mechanics (Addison-Wesley, Reading. 1950), p. 41.
32. R.A. Meger, R.J. Commisso, G. Cooperstein, A.T. Drobot, and Shyke A. Goldstein, 1982 IEEEInternational Conference on Plasma Science, Ottawa. Canada, IEEE Cat. No. 82CH1770-7 (1982),p. 4.
-79-
List of Symbols
A - switch area
E - electric field across the switch
E - electric field required for breakdown
Ec - electric field across switch during conduction
• - electronic charge
f. - eS. p.
- ratio of beam energy to load energy
ic - ratio of energy dissipated during switch conduction to load energy
go - ratio of energy dissipated during switch opening to load energy
Hb - energy deposited by e-beam in switch gas by inelastic processes
HC - energy deposited in switch during conduction
Ho - energy deposited in switch during opening
4 - e-beam current
IL - load current (- Isn)
Is* - switch plasma current
Jb - e-beam current density
Jsa - switch plasma current density
kb - fraction of total e-beam energy deposited in switch gas from inelastic processes
ko - fraction of load energy dissipated by switch during opening
k, - ratio of opening time to characteristic loss time for switch plasma
L - inductance of storage inductor _ -
Ls, - inductance of switch
a - number of pulses
n, - switch plasma density
P - switch pressure
P. - atmospheric pressure
r - switch radius
sB - safety factor for static breakdown
sm - safety factor for heating
S. - ionization parameter
v - plasma drift velocity
V - -beam voltage
VL - voltage across the load
V51 - voltage across the switch
W - energy per unit volume deposited in switch
-80-
Ar
Wa - energy per unit volume deposited in switch required for breakdown at pressure P
WS - energy per unit volume deposited in switch required for breakdown at atmospheric pressure- current pin
di -energy required for combination per electron-ion pair
71 - switch efficiency
0, - electron mobility
- gain factor
p, switch resistivity at peak current
Tc - time interval during which switch is conducting
'c - capacitor charging time
r& - inductor charging time
rcm - commutation time
TL - load pulse width (zwc for closing switch; =w.NC for opening switch)
rC- time interval during which switch is not conducting
- characteristic loss time for plasma electrons
to - time interval during which switch changes from conducting to nonconducting
W - time interval between pulses
jt - rise time of load pulse 4
rM - time interval during which switch change from nonconducting to conducting
_o
- S
-81-
( -0
THE CONTROL UF BREAKDOWN AND RECOVERY IN GASES BYPULSED ELECTRON BEAMS*
V.E. Scherrer**, R.J. Coimfisso+, R.F. Fernsler+,L. Miles++, and I. M. Vltkovitsky**
*Work supported by the Office of Naval Research and Naval Surface Weapons Center**Naval Research Laboratory, Washington, DC 20375+Jaycor, Inc., Alexandria, VA 22304
++Naval Surface Weapons Center, White Oak, D 20910
ABSTRACT
Experiments in which a 150 keV electron-beam pulse is injected into test gasmixtures at atmospheric pressure are described. Electron attaching gases, includin
C09, CH4 and SF ,. are mixed with non-at~aching base gases, P.2 and Ar. 'Theionizing electron- em current (1-100 A/cm ) provides a volume discharge betweentwo electrodes. The discharge is used to determine the temporal behavior of gasresistivity as a function of beam current. Theoretical results for 02 :N2 mixturesagree with experimental values.
KEY WORDS
Gaseous Dielectrics, Air Chemistry Code, Volume Discharge.
INTRODUCTION
There is presently much scientific interest in, and a number of possibleapplications for, gaseous dielectrics in which the conductivity is controlled by anexternal source of ionization. When an electron beam (e-beam) is injected into achamber containing a mixture of an attaching and a non-attaching gas, the ionizatioof the gas produced by the e-beam pulse competes with attachment and recombinationprocesses controlling the conductivity of the gas. Thus, the discharge current isturned on and off in association with the e-beam pulse. This concept hasapplication as a high power switch for fusion experiments and charged particle -beafrproduction and has relevance to laser development and beam propagation. Use ofelectron beams to control gas conductivity was demonstrated by Koval 'chuk andesyats (1 976), Hunter (1976), and O' Loughlin (1976).
The objective of this study is to obtain basic data for volume discharges caused b3e-beams and to compare this data with a theoretical model. In the experiment, a 1!kV pulse is applied to an e-beam diode producing an e-beam pulse that is injectedinto test gas mixtures. An ancillary capacitor connected to the test chamber drivethe Xolume discharges. The ionizing e-beam current density was varied from 1 to 1(A/cml. For 0 *N mixtures, resistivity and discharge current risetime and decaytime agree wih heoretical values.
-82-
. ... ,NOF THE EXPERIMENT
A schematic of the facility is shown in FMg. 1. The system contains three principalparts: a high voltage pulser [Fell and colleagues (1982)3, an e-beam diode electronaccelerator, and a gas cell test chamber. In the pulser, a 20 kA current source isused to charge an inductive store with exploding wire fuses, F1 and F2, serving asopening switches that produce a 200 ns(FWHM) 150 kV pulse to generate the e-beam.
ANODE L-TF iS
CA71-IME
LOf EAM DISCMRGEMt\YLAR EETOFz MEMBRANEI
SREGION S(-m6 ToTw)
L iJ
Fig..1. Schematic diagram of experimental facility..
The test chamber is normally filled with a mixture of two gases at a pressure of 760Torr. The resistance of the cell is determined by measuring the current through andthe-voltage across it. The circuit voltage is sustained by the capacitor C. Thecircuit is described by
-V. aR (t) Ci-i(e)3+ L 7+ - I dt, (I
where I is the net current flowing in the discharge circuit, V is the initialvoltage on capacitor C, L Is the circuit inductance, and i(e) ?s the electron beam-current. The gas resistivity, p, is geometrically proportional to the resistance,R(t).
THEORY
Outlined here is a brief reformulation of the theory presented by Fernsler andcolleagues (1980). Consider a volume discharge in which gas conduction is regulated S
solely by an externally applied ionizing electron beam and by the relevant gaschemistry. In the present experiments, where attachment is typically the dominantelectron loss, the electron density evolves according to
dn JaN (2) aTt e
where J is the beam current density, a is the effective cross section for ionization
by the electron beam, N is the neutral gas density, n is the electron density
-83-
and a is the electron attachment rate. If J is applied as a square pulse ofduration T , the electron density n(t) can, readily be determined. Equivalently,for constant electron mobility u and constant attachment rate a , the switchresistivity 1/(enu) is given by
p(t) - (3)
a (e \4'
The minimum switch resistivity,
- , (4)Pimln ='OLiz .G? -I
is thus limited byt(f rN "-t Pmi n tadlll")
4In the present experiments, a is controlled by mixing a specified fraction of anattaching gas (e.g., 02, SF6 , .... ) with a non-attaching base gas (N2 , Ar).
Eqs. (2)-(4) provide a convenient basis for comparison to experiment. Alternately,numerical simulations coupling gas chemistry to the appropriate circuit and beamparameters may be used. Fig. 2 illustrates such a simulation using the air
I :
I(A)
II% i
Fig. 2. Comparison of experimental and theoretical results.
chemistry code CHMAIR [Fernsler and others (1979)). Similar calculations have beenperformed by Dzlmianski and Kline (1979) for a variety of gas mixtures using aBoltzmann analysis to derive required transport coefficients and reaction rates.Simulations have also been performed by Lowke and Davies (1977) incorporating axialspatial variations to assess the effects of cathode fall and related phenomena.
We note that considerably altered behavior of p can often be obtained by choosinggases in which u decreases and a increases with applied field.Such changes tend toobscure, however, the basic ionization and deionization processes, and canadditionally result in unstable behavior as discovered by Douglas-Hamilton and Mani(1 973).
-84-
-S
"VL1'I4;.TAL RESULTS
The preceding section describes the time dependence of the gas resistivity for thecase of a square pulse electron beam injected uniformly into the gas. The analytic -form of the resistivity is given by Eq. 3 for time intervals during and after beamInjection. The derivation of Eq. 3 and 4 requires several assumptions that may bedifficult to realize in practice. For these reasons we need to accumulate a database for comparing experiment with a more realistic numerical calculation. We havecarried out a systematic study of a number of gases. The resistivity, p, andfalltime, T were determined for each gas mixture and were plotted as a function ofpercent attaching gas, This was repeated for different values of electron currentdensity from 1-100 A/cm2. In Fig. 3 we present a summary of p (evaluated at peakdischarge current) as a function of percent of attaching gas for a narrow range ofincident
So I I I I I f 1
700 le- 17-34 A/CM2
m .2~oo •-
500 02-N2
i S
400
300
200
0 10 20 30 40 50 60 70 80 90 100% ATTACHING GAS
Fig. 3. Effect of gas mixtures on resistivity. _
electron current densities (17-34 A/cm 2). These results provide a comparison of therelative merit of gas combinations and percentages for producing low resistivity.
The current falltlme, T , is presented in Fig. 4 for the same gas combinations andfor the same incident electron current density as the previous figure. As areference, T of the e-beam is also given. The measured falltimes as well as thedischarge current risetime are governed not only by the magnitude and time variationof the beam current and the attachment rate, but also by circuit parameters.
-85-
--0
I]
In Fig. 4, only 02-Ar (above 50%) and 02-N2 (above 30%) appear to be limited by thec-beam failtime. All other values of T are longer than the c-beam falltime and aredetermined by the gas properties. Beca~si the voltage applied across the dischargeelectrodes is well below the level needed to support a self-sustained discharge, thegas recovers to the pre-discharge dielectric stress of - 10 kY/cm.
900
700
600 C0-N 2
500,
400
300 4-Ar
02-A
% ATTACHING GAS
Fig. 4. Eff ect of gas mixture on current falitime.
REFERENCES
Douglas-Hamilton, 0.1., and S. A. Mani (1973). An electron attachment plasmaInstability. P hl fs. Lett. 23 508-510.DzminsiJ.W.0 K ne( T'. High voltage switch using externallyionized plasmas. Westinghouse Electric Corp., Report DYD-555-85-AA.
Fell, B., R.J. Conunisso, V.E. Scherrer, and I.M. Vltkovltsky (1982). Repetitiveoperation of an inductively-driven electron-beam diode. J. Appl. Phys., to bepublished (May).
Fernsler, R.F., A.W. Ali, J.R. Greig, and I.M. Vitkovitsky (1979). The NRL CIIMAIRcode, Naval Research Lab. Memo. Report 4110.
Fernsler, R.F., 0. Conte, and I.M. Vltkovltsky (1980). Repetitive electron-beamcontrolled switching. IEEE Trans. Plasma Si., PS-8, 176-180.
Koval'chuck, B.M. and G.A. Mesyats (1976). Rapid cu~toTof a high current in anelectron-beam-excited discharge. Soy. Tech. Phys. Lett. 2, 252-253.
Hunter, R.0. (1976). Electron beam controlled switching. Proc. IEEE First Int.Pulsed Power Conf., 1C8:1-6.
Lowke, J.J. and O.K. L-uvles (1977). Properties of an electric discharge sustainedby a uniform source of ionization. A L. hs.4,49-00
O'Loughlin, J.P. (1976). PFN design interface wit e..beii sustained discharge.Proc. IEEE First Int. Pulsed Power Conf., IIIC5:l-6.
-86-
V.S. Scharcer. 3.J. Ccmsui"oS,R.!. Furn l.lz ar.d I.M. Vitkovitsky
Naval PAsacach Laboratory
Washington., DC 20375
Summary shunted to the load P . After ths pulse, SW 2 opens
e lectro-bum (&-beam) contolled switeh Is and the process Is repeated for a second pulse by
a promisity opening switch candidat 1 for Inductive aking the e-be1es witch conduct again. 7he switch
storave pulsed power applicationes where repetitivly SW-2 Ia generally not requLied eX sep for specific
pUise4 ) 1 kN2) output por of - 1010 W 1a applications. ThO most sILificant problem
.e.-as. Sev eral 4uthoral:1 1
have reported associated with the use of ouch a switch Is the
thetr-tic. enalysis and aperimental results for e- tor W Laos by joue heating in the switch. To
beam controlled Aw&tches. 'he develoet of this minimize this heating, the switch resistance mt be
s..itcAl roiq.res a number of saperImentso be as low ma possible while the witch Is conducting.
perfor ne to determine the gas conductivity, rate of Such heating can also lead to late tUe breakdown
Changs o cendctivity, Switch efficiency. and the making repetitive puls* operation imposaible. On
off-eta '-! ooorg deposition I- the gas on the the other head, this switch promises to provides a
switt. raco4ery. Such emperimonts, which wuld repetitive opening cab&bility with a very feat fall
e-t-blin the feasibilitiy of the i-bern switch for time of the switch urrent. Potentially, the
'2. y aer epp~lcetLoa, are being carried ou. with opening time of the switch can be substantially
s,-=e -.As %siturse over a range of ambient gas faster than that achieved with fuse arraye.
-revs=** and electron bem current densities.
ceioaare made between *apartmet ad
;.o ndrslaad the basic mechanim IWN-2OUCON LOUTPUT
SIn b.itz-t operation. Single Pulse iWMurni SWITCW
.aerLso-ts are also used to provide a data ban" for cu e€i, LOAD
11etwg the rs ettive capability of the CMRIT
iw trh. A tjrICAc design goal invoives Chargng of esnMa-a *tutevm Inimccar over a period of everal W TCH
a otg nd current levels at the load I1 0tIG ONPLEof '1 -500 V and 10-100 MA. with - pulee-to-pulas vs.
5~4ai1of -100 m3,4. Initial *apartmnt*
charac%.:'.:&ag the performance of a sewond pus A-W-sean Switch a.tacept
The principles z! an e-beam controlled switch fisgur 1. Schsatic of an Inductively driven palesr
fir repeti va op.rtiun are discussed in f. (). using an i-boam controlled switch.
A Induactive stczrau system using an s-beamcontrollid switch to generate a pulsed high power
outpUt is shown sohtioally in Fig. (1). Shen the a measure of the switch performance that
-bes s tLrneJ on, the s itch resistance dps to dtet e fnes how practical soh a switch can be in the
a a l, value. resulting in energy transfer from the current gain. t. it is defined as the ratio
2ur e to the inducor L. Whn the -beam is - )/., where i0 is the total current flowing
tuaA c!! (and Sw-2 Lis closed) the switch in the switch, to io the Injested s-bern current,
resistance becumee large and a.wItch current to and the bar indicates peek values. A curent gain
-87-
Ie
p'actical syse.'t the diode. 7he fuse TO is chosen to open after 200
ni. limit.ng ,latse generation In the diode. TwoTheoreticSl pfelLalions
! I ndicate that the pulse-foxming networks can be coupled to the diode
reistivity . &-itch current fall time., r# and for double-pulse studies of the e-bsm switch.13
The
current Srin c .*pen4 on the aimbient gas pressure. test chamber in isolated from the diode by a
i=Ltin.n values of pressure dend upon the 0.005-cm mylar window and is normally filled with a
properties of the thin window betwen the evacuated mixture of two gases at a pressure of 1-3 atm. The_S
(-10 ozrs e-bean diode and the high pressre resistance of the cell during discharge Is
1- 103 Tor) switch, electron energy requirementa, determined by measuring the current through and
and the strength of the switch chamber. A switch voltage across it. The circuit voltage is sustained
pressure of 10 atm appears feasible, by the cepacitor C. The circuit is described by:
The upper limit of current density that can be di
switride in the repetitive pulse mode depos on V - Rfin- ) * L - idt, (1)
ener deposltion in the switch by joule heating.
caesaivo energy depeLmtion is nif sted in delayed
breaklow1 2 which would preclude maltiple pulse were A is the switch resistance., . i the Initial
operation, as discussed later in detal, voltage on capacitor C. and L is the circuit
Inductance. we can neglect the last term in E. (I)
(45. correction). At the time of peak in (i.e.Deecripnlon of the Zaperlnt when din/dt - 0 the switch resistance is
'No experiment, shown asciUally is
vr.g. 131 . contains three principal perta: a * ... 2....*(2)high voltage polaer. oeisti tng of a low
volta" s 4( I, kv) source and a pulse-formi
nsbat?9 an s*-beas dlod1 and 9as ooll test The rsijtivity at peak net current, po, is related
chamber. X& the pulse:.1 3
a 26 kh u rent to R. through the switch gomstry. The fall time of
the current through the switch in esti ated from;
LL
CAT"WC
where V 2 is the transient voltage peak associated:KACM: with the rise and fall of switch current and is{EAMn i1AA LcfOCDS indicated in Fig. (3).
C" 1 - The diagnostics used in the experiment includeRITON) calibrated Magowski loops and voltage dividers for
e measuring the dieo voltage and current, Vo and ibrespectively$ and the voltage across and not current
Figue 2 fthcatia of the experiment through the discharge, Vd and in respectively. Slue
cellophane and film techniques weore used to verify
the tim integrated bem uniformity.samre" cnarges An L.dCt1VS store, with explodingwire fuses yt end T2 serving as opening switches
-88-
T *3. plt*t sr. tao r-~ of
For t.h switch coacept that we have described, varying j. over a wide range, using the aom
the '1)1o.uit. paramters are importants curr ent 204 02 - g0t N, i.axture discussed earlier. '%e data
jai. resistivity. sjith curUnt fal.l time and in Fig. (S). demonstrate that m* ust be below
recoavry characteristics. Also of rpotatnce Is the XOA/ca2
to approach c - 10 for this gas mixture at
. ect- o r gy depostLtion on these parameters. p 2 a&t.
S-..cn p e . nc. is exctad to lprovs as the Fig. (G) illubtrates the eftect on the net-
prusag. .S inereaced. Therefore, the eiperiments current in the discharge wten the pressure is
were focused an the accumulation of data to increased from I a&tm to 3 atm for a 20% 02
determin and 01T'" VaZY with gas cooPosItc concentration in N2 . and f/P - 10.5 V/cu-2brr.and P:-,aur . these measureamnts are also necessary
to ..-. d.sat-,d the basic physical processes involved
an t's swt-h anl to provide input for theoretical I2D% 02IN8% 2Analysis Zad scaling. U report primarily results 2NJ" "2cmfor m- .xlturas because the atomic and moIecular
,late are ltttr esblished than for other gas
sla re.5 5, 50
* An exa-ple o'. result@ for one discharge of the I
swi.ch usir;7 a 20% O0 2 - g0t M2 mixture at L
I •&.. preara, is shown in rig. (31. The incident i %
,-nes C.:rnt density, 3. , -as 24 W/cm2 .
2 0 top 507 // % 5Mtwo tracts Show dl.06. vol.tage. V 0 ,ad current I.
,.* vs iterred a-imam current into the switch,
toi. 14 . the bottoM section of the figure. The
* dferstne Is 4ecay Time between ID and Is is doe to 200
Filaaa form.- ton in the diode and the enerw
.ULsen.OAt iranInj,ion of electrons through the 1000
cylar vLndgi. Plane& current thrugh the switch. .+
- n . voltg across it, are sho %.i,• ...... ...- v
0the cae two -races. p is plotted as a function
A tie i. the bottOm Octlion. The fast talle ow
U £r•±s..n resietr f - 30 Ca) followed ky a fast 5' '0rise. rvs.a~t I~n gnad switching perfofsancas.
The quarntIttios 0, I a r - A4 C e plotted an a
fxtctlon of percent 02 in H2
tn rig. (4) for Jo in
the range 2.5-6.7 Wee2. Shown also for reference/40
see the beam rise and fall tin.
res..t&e !nJicats that 0.concentrations of 10-20% ~30Yie'A rcsistiyi'tiee of 203-30 0-cm. a current ~gan Over M, . , outequal to thatofthe
bern. Lip, where C is the applied electric field 250 100
and p is ambient gas pressure, was 10.5 V/c-larr ink
this case. 11b 2aze-Crim electrica, breabdoin was__________________
observ'd. 0 100 200 300 40 500•ie % as;. . of 0a and c with electron bas t Ins)
carrefnt .er.sity is illustrated in Fig (5). A P AS rigure 3. A emxaple of ae&aur nSe made on a
WII~ &A 23 .C.cn observel. buat at the Sacrifice single shot.
-89-
. . . . . . . . . . . . . . . . ... . . . = -. . . . - . . . . . .. .. . , . , = , , . . . . . q , .
&,pt-nema .7,rrwet .- * 2- i. both. c-.~e
issonin Fig. (7) where pa to plotted against P
wihconstant K/p - 10.5 V/co-Tort andWO MN
concentrationsupt20.pchnelilewh
pressure. This is mest likely because the duration F - 3AIM
J. - 13& Alsm2P.I T- pF. I, elm
%a
E sa a we
*... F. , , igure 6. zfsat of ambient gas pressure on switch
V, as u urrsn ttiwhistory.pigats, 4? Dependence f% t-. ndCfor iarious
percenes~ of0-P
we illustrate the dependence Of current gain an
presuq with the data presented in rig. (9) for SIof the s-been pulse Is too short fat the discharge 0,in 02. Z(P wasn l0.S V/cm-ftcr and Jwas
to reach Nqullibrium at the lower percentages of th- S.0. 2 0.3 A/on2 as before. e Jecroesee from 7.6 to5 ttatllifl ga9. 55 empiaisted I.n what fellow*. 11.IP when preesuce is increased fr, I to 3 atm.oD changeS substanti~ally with P for higher
roacentrOUGaa. Tn rig. (a) IF is plotted versus P
for 0, conicentrations of 10. 20. end 309 in N2 . Z/p
ws 18.3 V/cm-Torri J. wee S.0 2 0.3 A/cm2 .
to% oo% eO Na MIXTURESgwsj~a/mm-em) vsPoo Tarr) /
En*A/5 10. / .1P 15*5 .s
E I
Z./
-- - - - - . .
Figure S. -.lutrto at ___ __ ( 0 __arr
resitivtyve.ardcurentgai. Cas fuctin Fgur 7.C~fct o ami*-t sitc prssue a Pool be curent dnsit, I,
/90/
i. -I s~te tz. ato d-.nt~v Pgama e..ctran
lose rate. The switca resistivity is given byOt-Na MIXTURES
za Tf rInsl vs P(ATM)
E .. 0.5 Volts , (n) V (5)P ~m-Tarr
.................. *--1
1-"°°*°'° ". where P is the electron mobility. Min'-m
I.L am- resistivity, pa. is attoined when the left-hand
side of Zq. (4) goed to 3ero. This yield@ uon
substitution in Eq. (5)
SI . . P~n' - (eSjpup) - - (j/p) (6)
P(ATM) where the product (up) is a function of gas
Flga.r. 8. cFfect of ambient switch pressure cempsition and the field parameter 2/P. Expressingthe switch plasa current density as
T(; .()
leads to
"p " f(E/P) (8)e we P IATMI Min
,. a, I-% N,
1 Vola andP 0. cm- Ter'---
- t • ;f-f(Z/P )'19
where switch current gain C Br j eaksdomn
limitations contr in the field paraister to X/P<.
10 V/cu-Torr.
P(ATM) Ca -NZ MIXTURES N xI
.. 7are 9. Vtmct a! ambient switch pressure an PT,' 103 flcm-st) ews% 02IN N2
Current S.n. E Volts
he rLgD. (0-0) Indl."te, trade-offsa exist
between obtaanqng low resistivity PO, short fall a 1 2 I0 155V
t imes. r ,ad h.qh current "ins, C. these trade- Is
* aim can be wuderstood by expressing the continuity
e*gu atin for tnoe 4 chsrge plama electron density, 4L
Rp. Go K / |V
G P AeP - ( 4 ) , 1 - *. . . . . . . . . . .. .... . .A, s o i n . . . .
% Ot IN NZ
rigors 10. Wfet of varying percent 02 in K2 and
.t met th i.* tian p.ra--ter 9 depends an the gas pressure an ths product P
s-bos e.er snI gas composition and where
-91-
t0
NR L IN 233 ,R J COMISSO 0 JUN 4 JYCOR-J26-4-09/6220
UNLASSIFIED F/O 20/7 ML
*mmmuz omini
1111.0 L- L28 12.5
ILIj 1.82
ALI
l~ll F~m .__-' Hill1111.25 11111 1.4 1
MICROCOPY RESOLUTION TEST CHARTNATIONAL BUREAU OF STANDARDS-1963-A
" (6) Id), 1i ai vt) ia .Arib the qu it IV. 'n on. cutw or P&4. 1 iI -he a at er /P i !
"lavir Of e-ow05 controlled Si.tchua. Toe plotted versus time, beginning with the ambient
scaling relationships are valid. however. only if C/P - 13 V/cm-Torr. Z/V decreae@. during the rise
the s-beas pulse persists for a time longer than of curen in the switch, to a sl.iiu of 2.S V/c-
1, and shuts Off in a irm shotr tha o T i.t Tr. It theu revervee. crosses the ambient line
ar-nct of this restriction can be seeln by and peaks at 21.$ Y/cm-orr, before declining to the
contrstLnk.4 .. (6) wIth Fig. (1) in which the ambient of 13 V/cm-Torr at 440 nt. At a ouch later . ..
p turTie is _ottG4 Cur three different gas time, 720 no, the switch b:aks down discharging the
pressures as a function O 0 a cncantratio. since syatam, The second curve is a plot of resistIve
electron mbility in N2 and 0 Ic similar. V4. (6) energy deposited in the switch per unit volume. W.
predcts that pin 'r ic nearly constant At the tim of breakdown. W attains a value of U'.
injie.pdent of gas pressure, . Or 02 The Occurrence of breakdown at late times, i.e.,
conetratio. At low concentrations of the long after R/P peASke suKgSts that breakdown is
attach.Ing gas 02* Ty beoms long coreroed with the influenced more by deposited energy than transiea -
bae &aratLoa and hence as never reaches P, Zl/P.
lIuilazly• * t high concetretion end/or high Thie ffect is Soon in Fig. 12 in which a plotpress.*, becomes mller tha the beam f&]I of ,as function of P for X/P - 11 V/cml-Trr is
tiLae as a result the estimated switch fall tim.
*~, appruachee the bess fail time rather shown for air and for a mixture of lo O in N2 .
then Thi.'s effect is lco ses, in the data in
ti..W. -hoe Te r approaches at ih 02 .
concontretions. W.( va PIATM)
" P cm-Tot,t-i me k w 0 FROM REF 12 me!0
ne possible failre mde ar the e-bea switch .1
Wha Operated repetitively is breakdown involving a C
eel- -sustaining are from which the switch could hot
recover. We observe that in a cass breakdowns *a",
do semer lato In tim, well after the 5-heem has
bees togne Of. This phenomena is desuribe with mi
the aid Of rIq%.(llI a4n (ill. .
PIATM)
Figure 12. aroakdown energy per unit volme for 10. ............................. ain W
2 and aLr.
- [N Io t,- 90% NIPs2A1MShown also on the carve is one point from Tat. 12
Sobtained in an experiment where P - 0.13 stew This
S.71E point lis on a straight line extension of our
/ data. The exact mechanism responsible for late.I /
time breakdown will require further study.
-Double-Pulse XxperLimnts
Fliure 11t lo:i Of I,'P at, einergy absorbed by theSam of the problems associated with repetitive
uttan 9as, as a function of time.
-92-
jCM-jnqm o~uvo usauo~ ;o Ajuaqri
-Mj 'S sauiu 9*lwAlwa pue !U!j3eGT ;a Sori1zSU o4 Aq C26T UP13!AdOO -Pwagsonqs '3'II fl 11001 AN 63WA maN '3oa3S 'PLt =a CtE 1731 "l1~jmS Sup~4qld '133 100263 'UDPRSSiWJd UOPInqqndba Jiupadaia 'hqdoo nimpo .ao *a*; vwtpPi mw wooxmp pjuauo ao; sappa pasuosl Adooaqd 03 pouluumd wt sicamilaq*soaad ;o asm puaewmo~u 'eu~id a.; n ao; anmp& javi iqspAdoo rnl ;o 31wf a3 PuoAaumiCAs i !op"Adoamoq at paaumod an sijeri *aoxm st e o a~w mwIM Peaujwiad 5i gup~swusv uowunAOD Tfln .&p JO &WnofdueA aoo~n TwAvN 'pumwuio~ swakASA wn tWAN 'putuwo) 4uoaA.osA pus qzmnasad =uW3jt=T Away -rn 'kuatJaspnN amapaop4 M' t !13q ;o a'Uo A~ia ;o suouniedeo 'k.%ts3y L'*IJda 4=Mdua POWAPy eumPo '901HO ~amAutay 'ipnwd *UpueS p Is Wo aoajoj jiV 'Aja3.aqu- u~~nO1d-oGsV eSaO . ' uon*&S wnbmnWj Mi' M i33M 'Aaarin wmatnp u=Ws put impnN. M4 NOp ;o 'pT~dtwi jo psssadze iapp 'suawasaopu. jo sapuod Tepto smt IuW=a-sads A£jmus s pea~m aeq am1 ppmpq pur snqane &4p o samp an 3uomzWno sTp ul pauj=~o zwpnpuo= pus SaSA suL6
jou~no s'odv*AL &=I;MS 1WAUNpirwwoo gua3sAS WrS MAIN
pummuw swsdrsam pus £Pnsa mioa AuWy rnAmuav JU5Fi'N a-Iso
uwin WWuaq p slrno AiJSI2 ;o uuweudAwav~ Z201wd ipJW5d POWUsPV &"PC5
wT3Io IpinwN AuAyV
A-viujoqr1 uos.ulndoad-oay &=jo jjV* UOP3-as erabamnbqjy 334
9 uojlstg 123Aarpos MOM'PS 'US'ld PUB AO~nN 23M
IU&M0MN azUaipxo o"iWW UMjIwd 1'nupoP5 Iww wJUoS.!JO3JOqI T9iDPV WIPI5S .aai uavda 4 mJMm5 vMMWAWN
upia1" '1-9 mow*ar
Cl=l MON 'anbjsnbqjy
aSO.MWWO J"o P"Ddi I=u 9
~fl~d V~IO acd .LIMIG
-(C961) BGC 'CS *34W1 *Td4df V 'A1312TAJ(2TA
N11? POV *23ZJsq05 ZA -a 7=0 ru 'ilea a *tt AxnTaLTA -Wi Put eauco a xsr!cuj*A5**
'LEC rrM GOTMA111 VU r *' AD *ATOORIM -d-
P. "2141110411 *D*A 'AWT@d 'Q *1'VIO "VIN~ i t V ocer vA wT3PuvxTv -ou vU1U0 r-t
mainOUTUs", ITS~ ,I 9UU 1TUUrT KIv tT 18AW 30 OD01 POO 'OA 'U*36IIIUI SS2UOWD ID04tIWU3ZU1 JO&UI Vito Act poisoddno. af W IC I24p
*qGU OR Il1U'I1Q0c UT u03UUV3 *UL'r put VOT1% n~ *tT'f
IN" POO "b1 .s *wr I*rr w ti :.yuu *j-u *ox .zsf; 1 ;G
awT2 uo;2izpuco %ozjs v 6ulu-.Y2c~ (p ;v 'Iw.!,vTW
-M Bd I (uvG ITsdv) 2~dua AaTusgaluf ipi sWxo& DO~ uvwydo bu~xo.I (C -uv,:,z~dv ~,:w.Ufft'0puZ0g, UO3*Igl.m S'064M" buToo4 *AITpvdii .9grWd Do vuoou p £5zu.e ;a 2**;;*Z. ku~p~
*(@us) Tu 6-6 *%;I '~OM sufti it.& Iq 4 e1 zU..*'
eqA (un umsys A.**vpozxuoz TvaoCp vpe..q
-(LIn '121 - *-a" uAq, couwq wwy~do DO ws 4 ellupuj win~ eq;L *I-.Y
"Pea *AeS 'aAuw *o Put 11p21a05 *UC l wuosrodk Act -T M~zocd. put =%: awvq-v
*AT21260d*v Sol ideoa 9 PaSu9AA .Tu Oft
*9-185i 1(9467) 34 q2Oqqv -;Me~ 28.04 S@5!U
posmri *Aux 10STS 1232 *oozA U!tr12.o .*r .9
sorts - NIll DO SIllt 01.p -
9-tU ( tL6TX IL 21*Oqqwl -JUC0 IhOu. ;@ 939I23OduT O@42 azlsvqtn L 1A4 u
2004 P03234il nl 292T. 2231 -*Dad zslUqq -0-1 -S .31d Sol a247nwoz a*Twqo zaeda. o. ez.z sweet
1W ;3o Pug go p~ujzw% OR DOD %13^ *auI SI1*1pul
*ztsn nl-L ':14 6ucY;vAj*..q*OW vott *..d iMoq 3o;
'uCI STIAxogo 1*zaaW2*T8 vatmow go. onTdaUi g 042 A23211lxf~ddW IIITA J;* Pug 90 uv~l el:
l"13UwV0UU rVZT0 'M(Ifl3AOIIA -14-1 PUS 66174 POAZIOqo ISA WITAG cig, *.uuo :.a .5~ 6SA:*Wqo
"I124OM Ze .11 *v ogglumo ruv X.z~og A t02 90311 OGAS. £TanOTA*Sd p~qjs5... C.312eOuVTp
SV0APill ZUSSZ OIL. *JJOtc.7iA S101 . 4/2 P0-f
*9gj 'id I(T9e2 jjdv) iso461 -ATun 121 L - 4 1114A 2V~ %GS-1,) %cs 4.47 P-1113 11. i4*3-
0^11271.dSi go doQM~Ov.. ;0 .*2* 1*6ujz32.V '4 *C 2OUT1:wrp CAL ar1Ted 024 -P*P WV,_~* .%r. r-. pr.i-i'4v
*Sam Js. cc Aq Lo.1.v*.d. 166&bd tm, -1 1(761 *AISOldxO a-rzlv ri-,** 9 %;Fic3m4 poldnei*
o~qOvlAS worel go onaSE 7110oft .*z W4 Ie *IYow ... 8 &A*n Poo* * ea .*W.g~2vV 6vicao; *w-wd
2" 1vsp~vflqud "o; psidecas 1A"32 AollTA *1., onz q4Ik Down ge-l (t) *2Tt 30 l21"215 0.4..g*%jrA
POO **1:-4o3 -2-A SJOTIILAW -1JU '0663;.m -r-1 -Z 0112.423 mISwe 0143 &Jv 9*111hiJO11342l %P 1 44
azT1u0o 02 tip ema12UWo idv.-id .. *wmtb.,
=77
APPUCATION OF ELECTRON-BEAM CONTROLLED DIFFUSE DISCHARGES TO FAST SWITCHING+
R.J. Camelsso a R.F. Fernsler' . V.E. Scherer, and I.M. Vitkovitsky
Naval Research LaboratoryWashington, OC 20375
Smm once the source of ionization Is removed. An arcdischarge is avoided by proper switch design.8
Recent Investigations into the phenomena The switch length, pressure and area areassociated with electron-beam controlled ditfuse maintained In such a way that self-sparking is
n discharges indicate a potential application forrepetitive1 10 kHz, in a 'burst' mode), high avoided.power (-10 -) switching. In such discharges theconductivity of the gas is regulated by thecompetition between beam ionization and attachmentand recoination processes In the gas. Short ML 0 Mopening times, limited in practice to the decaytime of the electron beam, can be realized while OMAN--high voltage Is being applied to the switch. Wereport on one series of experimonts using a N2-0 2
.combination In which the time dependence of tie -"cRndischarge current at two pressures has beendetermined. The results are compared to a zero-dimensional numerical simlation where the gascheistry is coupled to the discharge circuit.Experiments exploring the advantages of gases withdifferent collision cross sections and the "Wbehavior of the overall discharge stability have -- mns$also been performed. A formalism for switchdesign In which the switch gas mixture and I/
pressure, switch area and length are estimatedself-consistently for a given system efficiency isreviewed. The formalism is used to design asingle pulse, 200 kV, 30 kA (6 0) ; 100 ns FWKIMinductive storage generator.
Introduction
The successful Implementation of inductive Fig. 1. Illustration of opening switchenerw storage and Ils Inherent benefits to pulse application for an *-beam controlledpower applicationsi* depends critically on the diffuse discharge.development of opening switches. These switchesmust be capable of conducting current during the 0charging period of an inductor and then become In this paper we present some theoreticalsufficently resistive to divert the current analyses and data from an e-beam controlledflowing in the Inductor to a load. The current diffuse discharge experiment that are relevant tocommtation must occur on a time scale which 1s understanding the physics associated with theshort compared to the inductor charging time and behavior and optimization of an a-beam controlledin the presence of an electric field across the switch (EKS). Included in this are generalswitch that increases during the commutation equilibrium scaling laws, the consequence of usingprocess. Som applications require the switch to gases which exhibit unusual collision crossbe capable of repetitively pulsed operation (see, section behavior, and stability of thefor example, Ref. 3). Such a repetitively pulsed discharge. Because of the reliability andopening switch automatically satisfies the accessibility of various cross section data, mostconditions required for operation as a of the experiments were performed with mixtures ofrepetitively pulsed closing switch. 04 and N . Also reviewed is a design formalism
4-at self-consistently estimates the switch gasSeveral authors 4 7 have reported on mixture, pressure, area, length and e-beam
experiments and theoretical investigations in generator characteristics for a given system 0which an electron beam (e-beam) is used as an energ transfer efficiency, output pulse andionizing agent to sustain a diffuse (volumtric) number of pulses. Using this formalism we outlinedischarge in a gas placed betwoe two a design for an inductive storage system capableelectrodes. The concept, as it might be used in a of delivering a single- 200 kV, - 100 ns fullswitching application, is Illustrated in Fig..1. width at half maximum (FWH4), - 30 kA pulse intoIn this scheme the gas resistivity at any time is a - 6 a load.determined by a coupetition between ionizationprovided by the beam and the various recombination Oescrption of Experiment 0and attaching processes characteristic of thespecific gas mixture, pressure, and applied A schematic of the experimental apparatus Iselectric field. This. along with the volume shown In Fig. 2. An Inductively driven *-beamdischarge property, allows the gas to return to diode system, In which two exploding wire fusesits original non-conducting state very quickly are used in sequence as opening switches to
-95-
produce a high voltage pulse across the diode, where I is the discharge plasma current. for aprovides the e-beum for the switch experiments. 20% 02 - 80% N2 mixture with a ratio of appliedThis systev has been described in detail electric field to gas pressure of 10.5 V/cs-elsewhere. Gneration of a second s-bee. pulse torr. Because the polarity of the discharge wascan be accomplished by switching in a second set such as to accelerate plasma and e-bea. electronsof fuses and switches (not shown in Fig. 2). which in the same direction, Ip is related to in and
t bare similar to the set used in generating the byfirst pulse. The 3-cm dia. cathode is constructedof sa blades. The anode plate is drilled with0.5-ca dia. holes over an 8-cm dia. with a tp - ibgeometrical ratio of open area to total area of0.68. At the typical charging voltageson C (240 PF) of a 9 kV and with LO 7 Ol, pulses 30 * * , I . . ... I * 0i 'i,,Of ;0180 kV with a 200 as FNINH are produced. Thebern current d insties can be varied from EP ans O !u0.2 to S A/cm by varying the anode-cathode 002 W%% WOseparation. Because the e-beam pulse is so short,the discharge does not always have time to come to 4Wequilibrium. However, as long as the time ) 0histories of the beam current and voltage are O00
known, investigations Into how the basic physics ' 100of these discharges relate to EBCS applications , ,ahdcoparison to theory and numerical simulations 0can be carried out, as will be discussed in later 0O.sections.
- .. - ---- WsLc~nmoo
I
so so ..
Y0A -A
P. ( -I vl 'a
I- lO- TOM) .4
Fig. 2. Schemattc of experimental apparatus. til ,ci ! l ,, , 1,I0 2WO 3O 400 WOO OW
TIME IsOeAlso schematically illustrated in Fig. 2 are Fig. 3. Representative time histories of
the driving circuit and electrode structure for disa e r merso ied f
the gas discharge region. The current source discharge parameters obtained frmdriving the discharge in these experiments was a standard diagnostics.capacitor, C(1 if), whose initial charging voltageVd(O) could be varied from 0-30 KV. The systeminductance was measured to be L - 430 nil. Thedischarge electrode separation was I cm for allwork discussed here. A copper screen The circuit equation for the discharge isof v 60% measured current transmission was used asthe ground electrode. The 50-m mylar window tmaintained a pressure differential betwoeq the V R(L d i + I . lidt, (2)discharge vessel and the evacuated (- 10" torr) d(O) * 1b) +Is-beam diode region while allowing the high energyportion of the s-beam to enter the discharge where R is the discharge resistance. 1hechamber. The apparatus has been operated at discharge resistivity. o, is related to R throughpressures of 1-7 atm. A Rogowski loop, located the discharge geometry only. Neglecting the lastbetween the screen electrode and mylar window, was term in Eq. (2) (( S-percent correction) we haveused to measure ib. the a-beam current flowing at the time of peak In a discharge resistivity ofinto the discharge region. Additional diagnosticsfor this experiment include a voltage divider to Vd(0)measure the voltage across the diode, V0 ; Pgowski - () A)loops to measure the diode current, iD; and a O_ (3)comercial current monitor located at the drivingcapacitor to measure the net current in the where A is the discharge area and I the dischargedischarge circuit, In . Shown In Fig. 3 are electrode separation.representative time histories of V I t and Ip
96-
LV,.
The resistivity of the discharge duringconduction is one important characteristic. 1Another important parmete is the current C (8(en))(8gain, c, defined asdp
S- (4) here is the electric field across thedischarge during conduction.
where 10 and 10 are the peak values of It and tb We simulate the experiment by numericallyrespecttely. j a high energy transfer solving a coupled set of rate equations, one forefficiency. a ))1 . The current gain for the data each ionic species, along with Eqs. (2) and (S).in Fig. 3 is a 7. In this way np can be self-consistently obtained
Theoretical Model as a function of time for a given J and beamenergy. It in turn may be used witR Eq. (2) inEqs. (6) and (8) to predict macroscopicVarious aspects of the theoretical concepts observables %the experiment. An air-chodstry
associated with controlled diffuse discharges as code, CMIR, 1 is used in solving Eq. (S). Thisthey relate to the EBCS have been discussed., allows iny more atmic and mlecular processesIn this section we give a cursory review of the than those ontlned here to be included. sphysical model for the discharge. We describe a Co artsons beten measured and calculatedsystm in which a volumetric ionization souce par isn be mesured an calculateproduces~a very weakly ionized gas. Typically, parmters will be discussed in the next sectn.n - 10' N , where n Is the plasma electrondBnsity associated with the ionizing e-beam and Na We replace the last term in Eq. (5) with a
phenomnological expression. n /T , where isis the Initial gas density. We assume electrode the characteristic loss time fAr he discMPgeeffects are not important, which is the case for plasma electrons in the absence of Ionization.applied voltages mwah greater than the sheath Subtituting nn from Eq. (8) Into Eq. (5), and 0potential (< I k). Under these conditions the noting that fn quilbrium dn/dt 0, Eq. (5)gas resistivity is determined by electron-neutral becomescollisions, and the discharge potential is alinear function of the distance from the dischargeelectrodes. _' (..)f P(g)
We assume a two component gas mixture made upof a non-attaching base gas with some fraction of where f a eS P is essentially a constant for a
uattaching gas. Because an attachment dominsale B gi veg ca For most gases .recovery is desirble for EBCS applications, '
fop- - 10 cm/V-s. Also. Ec/P vtll bewe will ignore recombination. Assuming further acroxmstely independent of pressure (sethat density grdients are negligible the Sect. V). his analytic equilibrium scaling lawcontinuity equation for n. gives has particular relevance for fast ECS
applictions. In these applications onen ndesires c to be large and v to be short. For aet SolbP anp given gas composition, Eq. 9) indicates there is
a "trade-off = between high e and mall T whileHMere Sis abeam ionization parameter,
8 thsuggesting that operation at high pressure andJ (a IA) is the a-bem current density , i mobility is favorable.
etfect ve attacment rate coefficlent(cae-s ), Pis the total gas pressure and Na is the density of Overvie of E-Ues Controlled Discharge Resultsthe attaching gas. The first term on the righthand side of Eq. (5) represents the source of to this section w reviw several aspects ofdischarge plasm electrons. 7he mchanism for the experimental results and theoretical analysesdischarge plasm electron loss Is represented by that are particularly relevant to the design of anthe second term on the right hand side of Eq. (5). E[CS.
The discharge plasma density Is related to Figure 4 shows a comparison of the measuredthe discharge plasma current density through and predicted net discharge current I as a
function of time at pressures of I an 3 atm for a20 24- 80% N, mixture. The code uses the
J~/A n ev (6) meare ap p p ,u tna diode voltage, VD, to predict the Scircuit behavior. The measured e-beam current wasmodified so that only those beam electrons withwhere v is the electron drift sped. sufficient energy to traverse the switch were usedIn the calculations. Because the code is zero-dimensional, It can not accurately predict what
V a (7) happens when beam electrons do not traverse the
entire switch length as a result of eitherHere u is the electron mand E is the insufficient initial directed energy or excess
oobility10 stransverse energy. This effect is particularlyelectric field across the discharge. Thus the evident in the 3 at case. To improve thisdischarge resistivity during conduction is given comparison the a-bem electron energy must beby Increased. In making this comparison the screen
electrode transmission of 60% and the timeresponse of the Rogowski loop measuring ib weretaken into account. The code result was also
-97- 0
shifted 5-10 ns in time. The uncertainties in the the opposite effect occurs with the resistivitycode and measurement are estimated to be - IS% for Ar-O 2 about 5 times higher than that for N2 -each. As can be seen from Fig. 4, the code 02.accurately tracks the measured net current towithin these uncertainties and the limits of the A thorough analysis of this observationmodel. By using the results presented in Fig. 4, requires knowledge of the electron energythe scaling relation Eq. (9) can be checked. The distribution, which we do not have at present. Wevalues of - obtained from the code are a 50 nsand v 20 nsPfor l and 3 atm respectively. 7hus ' 'the ratio of (/T ) at 3 atm to (/T ) at I atis - 4, compared tO 3 from Eq. (9). Ifedifference can be ascribed to the lack ofequilibrium in the I atn. case. mOi
E/P .10 IOviesnlw
S iSMWATlONam.-IM0 P-l 00%00i .su ..DP, " \ 4" &-aom2
4618 S
-
". t*:". tla TO So • •
Fig. S. Comiarison of discharge resistivity.
p ,- p.for Ar-0 2 and N2.0, mixtures as a"W fuiction of initial vollage across
C ,, discharge at I atm.
,.SPAULAY., estimate13 for the Ar case that the electrons have/3 a mean energy ranging from 1-S eV for the applied2olo-06O%N voltages in Fig. S. This energy will be
. J,6Ad/M considerably lower in the M2 case because of the
'' so; ma.o& 3 o40 6 tacesibiit ofd at aetontolvl in the.lAttthe
.,U accessibility of vibrationaT levels In N2 . iAt the
,*,S. P*.5-20.a pressure indicated (I atm). two body dissociativeV IPalos attaclhmnt will dominate three body attachment for
electronInergies of - 5 eV. Also the Ramaurer-1.0 . Townsendi cross section for Ar is rapidly varying
for electron energies between O.S-S eV. 1hus the1 observed effect must be related to the higher mean
-/ -- -- electron energies in the Ar case, resulting in0 1-3 rapidly varying cross sections for momentumo 5o 100 15 soO 250 300 35040 transfer and attachment. Control of the electron
1ru(ml energy distribution function is thereforedesirable in switch applications, as has beI2
Fig. 4. Comarison of measured and calculated emphasized by Christophorou and co-workers.net discharge current for I and 3 atm.
Discharge stability is of concern in eitherthe repetitive or single pulse mode. Thedischarge stability can be affected by severalprocesses. Changes in elect ical and chemicalproperties of the switch gas'can result fromThe switch performance can be optimized by heating of the gas, or as a cuimulative result of
choosing a gas or gas combination that exhibits a the length of the conduction period or dutymobility and attachment rate which strongly vary cycle. It may be possible to avoid some of thesewith thi jt ei ependent electric field across the potetialproblems with the proper gasswitch. ' ' An example of such optimization mixture. Any reduction of E/N, where N is theis illustrated in Fig. S where P for a 10% 02 - gas density, either local or global at any time90% N2 and 10% 02 - 90% Ar mixturi is plotted for can trigger avalanche 1pn|gation. For exampl. anI atm as a functlon of initial applied voltage attachment instability 0' " has been suggested asacross the discharge, V (0) . This is also the a cause of local increase tn o which whenvoltage across the discarge when p is evaluated accompanied by local heating ang ; consequent(Eq. (3)). In an actual applicatioR we desire decrease in M during the conduction phase of an e-p to be low during conduction, i.e., when Vd is beam substained discharge can result in alw and increase during opening, when Vd becomes breakdown. Decreases in N resulting from heating
D larger. We see that the Ar-02 mixture shows at inhomogenities in the discharge or atmarkedly different behavior from the N2- 0 mixture electrodes so been observed to causeat different applied voltages. At low voTtages breakdown. 1 ,1 9 a(Vd(O)>lkV, for Vd(O) ( 1 kV sheath effects
dominate) the resistivity is 2 to 3 times lower Under certain conditions in our experiment afor Ar-O2 than "2-02. As the voltage increases, breakdown is observed sometime after the e-beam
-98-
controlled discharge plasma current has stopped physics. In what follows we assume the load pulseflowing. This effect is illustrated in Fig. 6, characteristics (current, voltage, FOIq) arewhere the time to breakdown, T. . (measured from ecified a priori.the Initiation of I ), Is plotted as function ofthe peak discharge plasma current, 10 . This data We begin by defining the energ transfer 0was taken with a 101-90%,N 2 mixture It 1 atm. In efficiency, 11, asorder to observe this effect, the applied voltagemust be 80-901 of the self break voltage. As seen ,in the figure. Tincreases rapidly as it (and L L (10)the energy dissipated in the gas) is decriased. ( ( Z)LI z + nb b'CThe measured times are of the order of acoustictransit times. This effect Is not observed if the Here n is the number of pulses, t and V are theapplied voltage is 4 751 of the breakdown voltage. (single pulse) peak current and vbitage &elivered 0and is beng studied further at present. to the load, Ib and b are the (single pulse) e-
beam current agd voltage, L is the storageinductance, I Is thepeak switch current, t isthe characteritM c load pulse width, and Tr Is'the
I0% 2 -90%N 2 time interval during which the switch is CP-Iatm conducting.V6. " WkV To elucidate better the energy loss term,. we
rewte Eq. (10) as n a c/Ct +1), wher C. thegain factor, is the ratio of the energ deliveredto the load to the total enerwy used In mking theswitch conduct. Thus
T0L 01 SRSTC*!bVbTCwhere Rf Is the average switch resistance during
uconductron, ISM is the average switch currentIduring conduction, and T is the time Interval
during which the switch janges from conducting tononconducting. The factor k is a dimensionless
• constant that when multiplie by the peak powerdelivered to the load gives the average power
10- dissipated in the switch during the opening 0phase. For resipive or Inductiveloads ka - 0.2S.
For reasonable energ efficiency (it ) 0.5) werequire C • 1. Therefore, each term in thedenominator of Eq. (11) must be sufficienty lessthan the numerator. Thus we set
J 0 O'90' LT0 000 2 M00 3000 4000 1 R a (12 )
41IA) ISW RSj"C , 9C I VOL TL*(~Fig. 6. Plot of time to breakdown, measured (100
from the initition of *-bem 'a .1controlled discharge plasma current, as b9
a function of peak o-beam cntrolleddischarge plasma current, wp. here g, 9C6# 1 such that
Switch Design Formalism 0 o * ( 1 (13)
The design of any EBCS oust address two major The gas breakdown problem Is avoidedconcerns: (1) a required overall system efficiency Initially by satisfying the constraintmust be satisfied; and (2) the self-sparkingthreshold must never be exceeded, so the switch E8gas can recover to its original dielectric V. " sS o (PI ) " (14)strength. In this section we review a self- L Tconsistent procedure that can be used to estimatethe switch gas composition, pressure, area, Here Vo is the maximum expected voltage acrosslength, and e-beam generator characteristics for a the swtc (i.e., the peak load voltage), E isgiven application. For details of this work, the the static breakdown electric field at atmophericreader should consult Ref. B. This design pressure P , s ( 1 is a dimensionless safetyformalism emphasizes the overall energ transfer factor, ans P Ts the switch pressure.efficiency of the switch-inductive store system, Additionally, cumulative heating of the gas mustthe circuit requirements, and the switch be sufficiently constrained so that any reduction
-99-
. . . . .. .. . . . . . . . . . . . . . .. . . . . . . .. . . . . . . . . . .
In switch gas density does not significantly lower peakoload current of 10 u 30 kA, peak load voltagethe self-sparking threshold. Energy is deposited of VL - 200 kV, and 1o;d pulse width ofin the switch primarily by resistive heating u 100 ns. This choice of parameters isduring the conduction phase, by resistive heating rllevant to several pulse power applications.3,8during the opening phase, and by direct deposition For technological reasons dealing with the beamby the -beam. Thus, a second constraint to avoid decay time, we take T a 100 ns. We chose thebreakdown is a storage inductance chlrging time to
be " 10 t or I us. Thus the quarter period2 of the chsrg ing circut (the switch conduction
1 L L SW RS 1C + kblbVbC" time) is I us, giving for a (typical)W I ,F capacitor, a lystem inductance
of L u 0.65 uH. I1W , the peak value of thesH---) A(Pt) (15) switch current, isZ 10. This gives a charging
voltage for the capacior of V - 25 kV. Thevoltage produced is then VL = V:I.o) 9Here k is the fraction of the beam energy L a) 200 kY.deposited In the switch gas, s H is a dimensionless L "safety factor, and MB is the deposited energy perunit volum, required for breakdown at atmospheric Tpressure. I - I
We my now proceed to compute the switchpressure. Using Eqs. (12ab.bc) and (14), the C[CS IIIdefinition of c, scaling relation Eq. (9), and RLrecalling that Ec- ISWSW/1 we have
p- 1 gOgCgbsBfo E02 T L_p -b -
The factor k I Ta/T Is the number of T periodsnecessary foP the sw?tch to Interrupt thl currentwhen the beam current is zero. Using Fig. 7. Schematic of single pulse, InductiveEq. (S) with the ionization term set to zero, one storage generator using an EBCS.can she that k - 5 for an attachment dominatedswitch.7 , 8 The ?raction of attaching gas can be We choose an energy transfer efficiencyestimated from the calculated pressure, required of n a 0.5, assume the switch will be attachmentefficiency, and known attachment rates. The "g" dominated giving kp - 6, take k a 0.2S for afactors in Eq. (16) are chosen so that P can be 0small (for ease of switch construction) consistent resistive load, sq a 0.75, H -0.5, E8 - 20with the constraint of Eq. (13). An optimum kV/cm, and WS 0.1S J/c 3
* Then the results ofchoice for go, 9C. and gb can be obtained by usingthe method of Lagrange undetermined multipliers, the previous section can be used to obtain theThis removes the Implied arbitrary nature of Eq. following parameters: P a 12 atm JEq. (16)),(13). The result is go,?C-g . For example * u I cm (Eq. (14)), and A a 350 cm ,radiusif n is chosen to be 0.75, then C-3 and g-0.1. - 10 cm (Eq. (15)). The current gain, from Eq.
(12C) with 9b - 0.3, is c a 30. This gives an e-Once-P.Is known, Eq. (14) gives the switch beam current of lb a 1 kA. The beam decay time
length, t, and Eq. (15) gives the switch area, should be << -r"A. The switch dimensions thus computed insuresthat the switch will be large enough that thebreakdown threshold will not be reached. Conclusion
The e-beam generator requirements are The results of this work indicate that thedetermined from Vh, [b- and A. Ib is computed E8CS is a viable opening switch concept.from Eq. (12c). The e-beam generator must Experiments with e-beam controlled dischargesactually provide a somhat higher current than lb verify the conceptual understanding of the physicsto account for current lost to the structure which govern the switch behavior. Dischargesupporting the interface between vacuum and high parameters may be optimized for switch applicationpressure. The @-beam generator must also supply a through proper choice of gas mixture and operatingbeam of area A. Note that in order for the beam regime. Backed by this understanding an EBCSto traverse the switch length Vhwfll depend on design procedure was outlined and a single pulsethe product P1 , as does VL. Therefore, one can system Ias designed to demonstrate a high powershow that typically Vb I RV with kb a 0.3. (- 6x0 W) inductive store - opening switch
system.
Single Pulse System This work would not have been possibleIn this section we outline the design of a without the expert technical assistance of J.N.
single pulse, high power (- 6x10'0 W), inductive Cameron and H. Hall.storage system presently under construction. Thecircuit requirements are presented first. Thedesign procedure outlined in the previous section Nork supported by NSWC, Dahlgren, VA. andis then used to determine the switch parameters. ONR, Arlington, VA.
ajaycor, Inc., Alexandria, VA 22304.Figure 7 Is a circuit diagram of the proposed
system. The desired load pulse parameters art-100-
e
References
1. R.D. Ford, 0. Jenkins, W.H. Lupton. and 1.M.Vitkovitsky. Rev. SO. Inst. 52 694 (1981).
2. S.A. Nesar and H.H. Woodson, roceedings ofthe Sixth Symqositu on Engineering Problemsof Fusion Research, IEEE Pub. ft. 75CHN097-5-NPS (1975), p. 316.
3. W.A. Barletta, Lawrence Livemore LaboratoryRePort UCRL-67288 (1981).
4. R.J. Comosso, R.F. Fernsler, V.E. Scherrer,and 1-4. Vltkovitsky. IEEE Trans. Plasma Si.PS-1O, 241 (1982) and references therein.
S. .T.-kline, Ibid. p 224.6. N.t. Hallada, P. Bletzlnger, and W.F. Bailey,
Ibid, p. 218.7. R.F. Fernsler, D. Conte, and I.4.
Vitkovitsky, IEEE Trans. Plasma SO. PS-S176, (1980).
8. R.J. Commisso, R.F. Fernsler, V.E. Scherrer,and L.M. Vitkovltsky, Naval ResearchLaboratory Memo Report 4975 (1982). To besubmitted to Rev. Sdt. Inst.
9. B. Fell, ft .. Coxisso, V.E. Scherrer. andI.. Vtkolvtsky, J. Appl. Phys. S3 2818(1962).
10. S.C. Orem., Basic Data of Plasm Physics (MITPress, cambridge, 1961). pp. 87-94. 0
11. ft.F. Fernsler, A.M. Ali, J.R. Grieg, and I.M.Vithovitsky. Meval Research Laboratory MemoReport 4110 (1979).
12. L.A. Christophorou, S.A. IHunter, J.G. Carterand L.A. Mthis, Appl. Phys. Lett. 41 147(19 2).
13. J.V. Ozimlanski and L.E. Kline, FinalTechnical Report No. 80-0120, Aero Propulsion •Laboratory. Wirg.ht-Patterson AFS (1979).
14. J.8. Hasted, Phsics of Atomic Collisions(Amrican Elsevier, New Tork, 91z), pp. I06-310.
1. L. Christophorou, paper in this conference.16. U.H. Douglas-Hamilton and Siva A. Mni, Appl.
Phys. Latter. 23, 503 (1973).17. O.K. Douglas-I01lton and S.A. Mni, J..ppl.
Phys. 45, 4406 (1974).18. M.M. J~reeva, 1.6. Piftiantsev, V.0.
PIs'mennyi, V.M. Polushkin, A.T. Rakhtmov,N.A. Timofeev, and E.0. Treneva, Soy. J.Plasm Phys. 3 (6). 770 (1977).
19. S.A. Genkin, 1 6a. D. Korolev, V.G. Rabotkin,and AP. Khuzeev, Soy. J. Plasma Phys. 7 (3),327 (1981).
•
-l~l-
STUDY OF GAS MIXTURES FOR E-BEAM CONTROLLED SWITCHES
V.E. Scherrer, R.J. Commisso, R.F. Fernsler,and I.M. Vitkovitsky
Naval Research LaboratoryWashington, D.C. 20375
ABSTRACT
An electron-beam sustained diffuse discharge has been suggested as a possiblefast, high power, repetitive opening switch for compact pulsed power systemsemploying inductive storage. The dependence of gas resistivity at the time ofmaximum discharge current for an N2-02 mixture, an Ar-02 mixture, pure CH4 ,
and CH4 with C2F6 has been measured as a function of reduced electric field,
E/N, at various pressures in such a discharge. Effects associated with E/Ndependent mobility and attachment rate that are beneficial to switchapplications have been observed.
KEYWORDS
Diffuse discharge, opening switch, electron-beam controlled discharge,inductive storage.
I NTRODUCTI ON
There is presently much interest in the use of an electron-beam fe-beam)controlled diffuse discharge as a repetitive opening switch in an inductivestorage pulsed power generator (Fernsler; 1980, Commisso, 1982; Hallada, 1982;Kline, 1982; Bletzinger, 1983; Commisso, 1983; Lowry, 1983; Commisso,1984). The conductivity of this discharge is controlled by the competitionbetween e-beam ionization and the various attachment and recombinationprocesses characteristic of the gas or gas mixture. When the e-beam pulse isremoved the gas may rapidly return to the normal, highly resistive state.This transition may be particularly rapid if attachment is the dominant lossprocess for discharge electrons (Fernsler, 1980; Commisso, 1982).
-102-
An important physics aspect of such an opening switch is the optimization ofthe gas mixture. Assuming that the electron mobility and attachment rates are *independent of E/N, where E is the electric field across the switch and N isthe gas density, one can show that the product of the opening time and theminimum discharge resistivity is a constant (Scherrer, 1982). The openingtime is the time interval during which the discharge undergoes a change fromlow to high resistivity. This suggests that it may be difficult to realize aswitch with both low resistivity during conduction and a short opening time. *However, by choosing gas mixtures whose associated mobilities and attachmentrates are strong functions of E/N it may be possible to satisfy thesecontradictory requirements, as has been suggested (Fernsler, 1980;Christophorou, 1982, 1983; Commisso, 1982,1983,1984). In this paper wecompare the measured discharge resistivity during the conduction phase ofseveral gas mixtures. The qualitative comparison of these data and of the 0discharge current waveforms demonstrates that indeed the gas mixture can beoptimized to give a low resistivity at low E/N (conduction phase) with theresistivity increasing rapidly as E/N increases (opening phase).
DESCRIPTION OF EXPERIMENT 0
The apparatus has been described previously (Commisso, 1983). A schematic ofthe system is shown in Fig. 1. The system comprises three principal parts: ahigh voltage pulser (Fell, 1982), which uses fuses to produce a high voltagepulse; an e-beam diode, which provides the ionizing e-beam; and a gas celltest chamber, typically pressurized to 1-5 atm. At the typical chargingvoltage on Co (240 F) of = 9 kV and with Lo - 7 H, e-beam pulsesof = 180 kY with 200 ns FWHM are produced. The anode plate is drilled with0.5-cm dia. holes over an 8-cm dia. with a geometrical ratio of open area tototal area of 0.68.
I S
ANODE
DISCHARGEELECTRODESI.----" 0
CATHODE
S2 O a I L CEA VO
' MYLARWINDOW
-DIODE- GAS DISCHARGE-t-iO5 Tar" REGION
(-0I Ton) to
Fig. 1. Schematic of experimental apparatus.
-103-
The current source driving the discharge was a-capacitor, C = 1 UF, charge&-at-0-30 kV. The discharge system inductance, L, was 430 nH. Unless otherwisEstated, the discharge electrode separation was Z - I cm. A 50-um mylar window
maintained the pressure differential between the discharge vessel and the
evacuated (-1O" 4 torr) e-beam diode region while allowing the high energy
portion of the e-beam to enter the discharge region.
EXPERIMENTAL RESULTS
Figure 2 is a plot of the discharge resistivity, po, at peak discharge
current as a function of E/No , where No is the ambient gas density and E s
Vd(0)/Z is the average electric field across the switch of electrodeseparation Z at the time of peak discharge current. The voltage across theswitch at this time is simply the charge voltage of the ancillary capacitor,
Vd(0 ) because the inductive correction to the discharge voltage may be
neglected (Scherrer, 1982). For values of Vd(O) < 1 kV, E may be over
estimated, as discussed later in the paper. The data shown are for 1:4mixtures of 02-N2 and 02-Ar at ambient pressures of 1 and 5 atm. Note that
for the 02-Ar case there Is a broad minimum at 6-8 Td for 1 atm and a strong
minimum at 2-4 Td for 6 atm. The 02 -N2 data show no such minimum.
S . . p. -u I
C. 120 027 80%A J' 2
%O-O%Au
Fig. 2. Discharge resistivity at peak .diachar~e current)r O
as a funct:ion of reduced field strength, E/N , for
00
2 -N2 and .O2 -Ar mixtures at 1 and 5 at.
In Fig. 3 we present data taken with 1 atm of CH4 both with and without 1%
the attaching gas C2F6 (Christophorou, 1983). A broad minimum is observed at
18-21 Td. The addition of 1% C2 F6 increases the minimum resistivity 6
-104-
j
SWS
U .Jb LSA/cU.
p- I
I
i%
Fig. 3. Coparison of p0a as a fzmction of E/No for C%4with an d wiihout 1% V2 6 at I. atm.
• I I
400-
100%C cm
Fig. 4. C=p on Po asa fuiction of E/N for C:with and without 1% C2 F a 5 ti.
-105-
by - 10%. However, as E/No is increased the resistivity for the 1% C2F6 case
increases by more than a factor of 2.
This high E/N. behavior is also observed for the same data taken at 5 atm,as shown in Fig. 4. The resistivity minimum has become more pronounced anmoved to lower E/No - 4-5 Td.
Figure 5 is a representative time history of the discharge current compari.;pure CH4 and CH4 with 1% C2F6, both at 1 atm. The decay time (90% to 10%) of
the discharge current for the 1% C2F6 case is nearly an order of magnitude
smaller than the 100% CH4 case.
Fig. 5. Tine history of discharge plasma current I Pfor
CK4 - with and without 1% CAF at 1 atm.initial (E/N) o-M 20 Td.
DISCUSSION
Several aspects associated with the inte-rpretation of these data beardiscussion before proceeding. The magnitude of the resistivity quoted is_obtained at the time of peak discharge current and is thus a minimum value for
our particular choice of parameters. By varying E/Nn we attempt to stmulat-the transient variation of E/N in an actual switch as it evolves from the
conduction to opening phase. However, in a real switch the e-beam would beabsent during the opening phase making the value of the resistivitysignificantly higher than what is quoted here. Thus, the general trend of hedata, rather than their actual magnitude, will be interpreted.
Another concern is the effect of the potential associated with the cathodesheath. At low values of E/N , the sheath potential (Vs - ) is a
significant fraction of the applied voltage Vd(O ). The effective field in the
body of the discharge Is therefore given by E - (d(O)-Vs)/Z rather than o
-106-
E Vd(O)/Z. This discrepancy is minimized over the same range of E/No by
i operating at higher pressure and/or larger electrode separation Z. In these
cases Vd(O) must be increased to maintain the same E/No making the cathode
sheath potential a smaller fraction of Vd(O).
Using the appropriate Ohm's law and assuming an attachment dominated discharge*
in equilibrium (dnp/dt 0, where np is the discharge plasma electron density)
the discharge resistivity is (Commisso, 1984)
N a CLPo (CbO _0 (1)
Here u is the electron mobility, Jb is the e-beam current density, Na is the
density of the attaching gas, c is the effective cross section for beam
ionization, and a is the attachment rate coefficient (cm3/sec). For a given'
gas and e-beam, po depends on E/No through the parameters a and U
For the 02-N2 mixture, a and P depend only weakly on E/No (Dutton, 1975;
Brown, 1967) as is observed in the data of Fig. 2. The apparent rapidincrease in P0 at E/No < 8 Td is most likely a result of overestimating E, as
discussed earlier. For example at E/No = 8 Td for 1 atm, E = 2 kV/cm. If the
sheath potential is f 1 kV, the effective field is E - 1 kV/cm
(for Z = 1 cm) and a more realistic value for E/No is * 4 Td.The situation in Fig. 2 is very different for 02-Ar. Here p0 varies with E/Notin such a.way as to strongly enhance switching operation. That is,po 0
increases with E/No. The minimum of P0 moves from - 6-8 Td for 1 atm
to - 4 Td for 6 atm. This is again mainly a cathode sheath effect, asoutlined above. Estimates at 1 atm for the three body attachment rate, whichis weakly E/No dependent, and the two body dissociative attachment rate, which
is strongly E/N0 dependent, indicate they are comparable for a discharge
plasma electron temperature of Te - 1 eV . At 6 atm three body attachment is
dominant. Thus, from Eq. (1) the mobility must be E/N0 dependent as isexpected because Ar exhibits a strong Ramsaur-Townsend behavior. Theinterpretation of the 02-Ar data at 1 atm would be'more straightforward if the
.electron energy as a function of E/No were known.
The CH4 data In Figs. 3 and 4 clearly show the effects of adding the attaching
gas C2F6 . The attachment rate for this gas Is known to be E/No dependent(Christophorou, 1983). Although the minimum resistivity is higher 0
by - 10-20% ,the resistivity at higher E/No is greater by a factor - 2 when 1%of C2F6 Is used. The same sheath effect is seen here as observed with 02-Ar.
The resistivity minimum at 5 atm has become very sharp and moved to 4 Td fromthe broad minimum at 18-21 Td with 1 atm. The quantitative agreement betweenthe location of the minima in this case (again assuming a cathode sheath S
-107-
potential of - 1 kV) is not as good as the 02-Ar case. Note that, as with Ar,
CH4 exhibits a strong Ramsauer-Townsend effect. -
The enhanced resistivity at higher E/No manifests Itself as a shorter openin-
time, i.e., faster decay of the discharge plasma current Ip. This Is
demonstrated In Fig. 5 where Ip for 100% CH4 and 99% CH4-1% C2F6 at I atm is
plotted as a function of time. The time from the 90% to 10% pointis - 100 ns with C2F2, compared with - 700 ns without C2F6 . The e-beam deca''time is - 100 ns.
CONCLUSIONS
The data presented here demonstrate that an e-beam controlled discharge canoptimized for switching applications by choosing a gas mixture whose mobilit.and attachment rate are strong functions of E/No. The optimization allows alow resistivity during the conduction with a short opening time.
ACKNOWLEDGEMENTS
The authors wish to thank H. Hall and J.M. Cameron for their expert technicaassistance and Dr. L. G. Christophorou for suggesting the use of CH4 withC2F6 . This work was supported by the office of Naval Research and the Naval
Surface Weapons Center, Oahlgren, VA.
REFERENCES
Brown, S.C. (1967). Basic Data of Plasma Physics, MIT Press,Cambridge, 84.
Bletzinger, P. (1983). Fourth IEEE Pulsed Power Conference,Albuquerque, NM, paper 3.3.
Christophorou, L.G., S.R. Hunter, J.G. Carter, and R.A. Mathis,(1982). Appl. Phys. Lett. 41, 147-149.
Chrtstophorou, L.G., S.R. Hunter, J.G. Carter, S.M. Spyrou, andV.K. Lakdawala, (1983). Fourth IEEE Pulsed Power Conference, Albuquerque,NM, paper 32.1.
Commisso, R.J., R.F. Fernsler, V.E. Scherrer, and I.M.Vitkovitsky, (1982). IEEE Trans. Plasma Sci. PS-10, 241-245.
Commisso, R.J., R.F. Fernsler, V.E. Scherrer, and I.M.Vitkovitsky (1983). Fourth IEEE Pulsed Power Conference, Albuquerque, NM,paper 7.1.
Commisso, R.J., R.F. Fernsler, V.E. Scherrer, and I.M.Vltkovitsky (1984). Rev. Sol. Instrum., to be published.
Dutton, J. (1975). J. Phys. Chem. Ref. Data 4, 631-684.Fell, B., R.J. Commisso, V.E. Scherrer, and I.M. Vitkovitsky
(1982). J. Appl. Phys. 53, 2818-2824.Fernsler, R.F., 0. Conte, and I.M. Vltkovitsky (1980).
IEEE Trans. Plasma Sc. PS-8, 176-180.Hallada, M.R., P. Bletzinger, and W.F. Bailey (1982).
-108-
IEEE Trans. Plasma Sci. PS-1O, 218-223.Kline, L.E., (1982). IEEE Trans. Plasma Set. PS-I0, 224-233.Lowry, J.F., L.E. Kline, and J.V.R. Heberlein (1983).
Fourth IEEE Pulsed Power Conference, Albuquerque, NM, paper 7.2.Scherrer, V.E., R.J. Commisso, R.F. Fernsler, and I.M.
Vltkovitsky (1982). 1982 Fifteenth Power Modulator Symposium, Baltimore,MO. IEEE Cat. No. 82 CH1785-5, 146-152.
0
-109-
-19
1412 111.1 1111 LUAI t(INAL EUNI I NI U13 (1UN Ii M SLII EIdRlAY U-19. 1997
$oes Sqie .arhmts; hame ulm.2t,..Iad
that lmi.a an elJetalwi Ia . Of AlfqlaitA WWVY aMWmumemet dew.&1 Is Pazsed through, a 9" loated Pivd irer o 'I 1rst ail
btauwen twe oz.,saiua. the vas can 1w made, to vandmaoL. finjl Call for Pdvzrs'O.0i Wwe the asItaga aMe. the V34-,Otz0d IS well saimmiacaaikNA far instruCLia.s~bill.' the seaU-imosL-doom valu. 1
- CIComm a Is In proepay our e abstract.admae"e "hem, lmilaft Procaeses aS0Uu With uhe
beas= a rapid thee ree.klatmms &Wd electronalamang. t e laoS lb.. tes rn W1rmm daud and &Woiuject .4tevi amma:apgiudt ad the reistivity of the visa m he rom
trailad bp l teolectron be&%. SbAmewes. It the been_________Profits Is uniom a vouetzia discharge is praioed
&*a. ho electrodeas toamosy doe Im the SuabjeC c11ug0?7 nmber:*leotm bass musta~aed lame devices. DY ea"Ing afoam zdatore Of .l4mretesm ative ad as.)eatle._____________meWatiwe gees the diohedm.r say be Os to "tuen Ogrga..4 maia) very rapidly sinS. ies a- ) Prefer oral sessioaGm'davu W.AM5 the ..Wmlead ilarie Slams. iseapee;%rsaa mlc owns, a disadmr" ivroually ametal as Prefer postsr assioaa mwtkin plum Me (Opinge ori elus) for pass"dr-mwr war&ir flbprefrence
we r"ala onriwma to i.vetivas Limsma&,rale of oleitarn-ben merallad &Ulm*n ( )Iges requests for
dkavye. an £adheevely drIviem slob plaemn ofthis abstractarmdioe.t450o W. -1.5 km) tbai As OR11mbla 09
gr.4mq asihmr m or ttai boom wamema ampastad by_____________31W ;:I Is weed to stmaiy the eombctsm, aid romouy
fpaugoti.10 at a varlay of gapea a"i gas sitarms. 0_________
Slu bea A Injected Ibramgb a 0.0020 mylaw vacomlAartsin late the diaetmarva regioe (XI Abe Wharm,____________mmLvrne Is aidwed betwes twosaleow ae drivenby a I ve apsafttr. Disaga Isla- W..v etsaigeveeta amitom. eletron bas doaisafty. and ~ ~ la brysakA.9era~*y. To dute. ela.@wenaarlw goee In.cluium ft. ". C04 a" I&Am gao" In mister..________With an-atthai..g 0010808ab as ft mod AV. tpdimitmamita prsdietim jar, the tins history' ag &hedi4Ma y ardat aug mativity for the nmae at____________
aD ws-n'e WA~f"Will ha smpaced with aq"aheesl 1 mM tjrpsurLtCmjwwi~e. 11% 1 Sal results tow the disehaad
owwm ad ,aiy Ado ar to aqwee with om-6mwata disaharmt ehavior. Cormt multirlirnUms Tf MrM-S)
ti. he ratio Of5She diaee a" n mn twooumearmm esmains o up to b -to he. hmex.____________
observe.mtdb U iaie.6
.tS te.msami.. VA, 23Nd -0 a M3A Of Ii.. CmsstttnMoalvtm beiap Cetatg wim. as. 3036 an Plasma scwg~ ad
Ai~ & aor. rs. sm nextS jet- Pulsed to~ a 3cout.. LuldAt. = £293%). viacol-4. y f
SJ.P. ftrnli. . IM VIraL let. passedMom" CadE.. AMOAMIC. 12 11971. llICS4I-6
S63 AMldmek and Z.2. Iluaraes. Sow. Task.-MoW. lost. 3.. 9117.1.
4E., aml. .1. ju. *IUr Txdaa. PLAsmai Sol..
% ir.romir. 1. cnga, amd z.1. vicknveeby.u=2 Tvam. rfm sck. rv-S. 276 (19502 -
"a. -u. 3.J. casae. v.z. Sur.rand
lb%$ eor&. Or a re4-.ac~iblQ fiClliuilt, plau Luc Xerox copies guost te received 1101 LAIt'Uan rtklULSt. 1912 at IhPi following address: A..). A1cock. Chailrma. 1912International Conloreaco w. Pldsalw SciCwva CIO Nfational Researc~h Council of Casmia.Ottawa. awitarie. CUsad CIA ON6G.
Abstract SubmittedFor the Twenty-fourth Annual Meeting
Division of Plasma PhysicsNovember 1 to 5, 1982
*CategorY Number and Subject 1.122 Blectron Deam/Plasma Xntezactions4.17 OerApplications
O0neory MJExperiment* Parameter Scaling in cstron-eam Controlled
Dici ~ASEE.+ V.1. SCherrer, AU. C=Misso*, R.v.Fenlr, and IJI. Vitkcovitsk7, -Naval RteseaurchLaborata.1 The. usie of an electron beam to controlthe conductivity of a gas between two electrodes haspotentia ap cation a-&a high power repetitiveOpening orclosing switch. 1 ,2 an inductively drivene-bea, diede -is used-to generate a 200-keY, 200-na
* 1MM electrou bemn. which is Injected into a V2-02* ges mixemke between *two electrodes. The discharge
resistivity at peak discharge current, the currentgain (discharge plasma current/injected beam current)..and Opening time are determineid as a function ofpercent attaching gas (02) , injected beamt current
U (1L-5 A/CK 2) and ambient pressure (1-5 atm). Thieresults axe c~rdto computer simulations.
41SqpPorted by NU1C Dahlgren, VA and OMR, AringtonVa*JAY=, Inc., A lazandria, VA 223043 1. Ferasler, D. Coaits, and. X. Vitkovitsky, I=
* ~Trans. Plasm Sci. MA-, 176 (180).... ....2R,; Comiso, 3.?. Jerusler, V.1. Scherrer, and
I.3. Vitkavitsky, -to be published, Iz" Trans.Planes Sci. (1982).
M* Prefer Poster Session Submitted by:
o Prefer Oral Session eA*h (- 1-
o No Prefenc (sinatutd of APS member)
o Special Requests for placement r J.Ca ssof this *abstc Atm nam 4770t
ol Special Facilities Requested Wsiton, D.C. 20375(e.g., movie projector) (address)
This form, or a reasonable facsimile, plus Two Xerox Copies must be receivedNO LATER THAN NOON, July 30, 1982, at the following address:
Ms. Barbara SafartyPrinceton Plasma Physics LaboratoryP.O. Box 451Princeton, New Jersey 08544
wV
. ~LLt
z ww 'ala-
4I. 9 m m1 0 - (.**r% w wil)I
lii *~ 4 i~ii
trIL
Oldl
LI 6
cno ~cn=- - -J
4 b.LL
t~i ww -i
0 w in-01Q-
00
-j C)
0
w
1. 0 zc
a iar
cr.
cni
m F-
CLl
Li z
0 ww m-
000Z,
-(14
- . I . I
Ius
0 us
US
co 0 .(
- * - U
co IL L IICi
IL . ca.0 0 4
co soa L * On z*o a i
0 Id 01. c0 = IL- II3c CL.> 0
Go 0 3 CD(0 U
> w~ a.us =C 4c a a
0 U. 0 U
* Ir
V~L
I r Ql S-
IZ*a
T S °
20% 0 2 -80% N2v V =8kV 600E (VOLTS -500
2..300 1
" c * ,7 , , ' , , , u , , , , , S I , , , , , , , , i , ,3 t0
20 Plh 2000
15 i /. 15004 "./ \Vd °oo
> 100- 200
0 0
150 4 300
~1000 Iwo
J -- - - -- -
0 00
0 1 1 i t (ns)1W -37-0
\lb
'pO
I I I -
I * 0
1j '-4 0I AII i
* . S0
S
U i0
-~ ~ -~ -~. 0
8N S
'Ut
------ S.-
S
zb. -U '. .44
- .~ 0- E S..h
o
.10.-
- inD~NfO 'S.
* U U I*-~-4 154 S0. fl054 \. \~~IH
I I Qp.
S'-p. 0 0 0§ 0 0N N
0
'S
-119-
S
hi
00
6 10
CQ CD
-00
UI) z
0wI
0 ' F 0 0 0 .I
lo I Nr K ~SI fl * I I.
o
-04-Eaa
I, c)
a. 4o
_0
Ca , a 0N _'f 0 A 0
-122-
.2000, E/P2 10.5 V/cm - torr1002 90% NZ
- 0 1 aotm 4
114 3at
1500 * '.i-
ci.
50549
II
0 0.5 .0 1.
Jb (A/CM)
- -223-
90-E/P= !0.5V/cm-torr10% 02 - 90/ N?
80- - !tm
-- a* 2otm
70-
60z
050-zw
'40 - )
30-
20-
I0
0 ! - I t t t I I l , I ! | I .L ! I
0 0.5 1.0 1.5 2.0Jb(A/CM2)
-124-
' . -, .
0*0
CA w0
CC 0M O Id
(AN 0A
V- CD
CS 0 0
cr..
0.'
UU
IL *s si g t I r;I 1
(Ew*) IOA
-126-
Studies of Electron-Beam Controlled Diffuse MY 23-25, 1983Dischar es , V.. Scherra, RLi. Commisso*, .F.
-esler*, and .LM. Vltkovitsky, Haval ResearchLaborator7, Washington, DC 20375 - When an electronbea. of sufficient energy passes'through a gas column athigh pressure. (- 760 Torr) and 'Voltage is applied acrossthe column, a diffuse discharge can be formed. TheI condu-ct:vLty of this discharge is dominated by electron-neutral collisions and is therefore controlled by the SU3JECr CATEGORY:fractional ionization of the gas. The fractional
ionization of the colun is in turn determined by a GcompetitiLon between electron-beam (e-beam) ionization .-ARC TEH- & GSEUSEand recombination and (more importantly) attachment NI4BER: .14-processes in the gas. When the electron beam ceases,the conductivity of the gas column can return quickly tothat of the neutral gas. Investigations into these e-beam controlled diffuse discharges indicate potentialapplication for repetitive (- 10 kRz, 14 -"burs-mode), high power (- 1010 W) switching. 1 " Theseapplications :nclude both opening and closing switches. 4 X PREFER POSTER SE
Usin N2-02 combinations at pressures rangingbetween 760 and 2,280 Torr we have studied the .....dependance of such discharge parameters as resistivity,current gain,. and opening time on percentage ofattaching gas, e-beam current density, and applieddischarge voltage. These results are coppared to zero-dimensional numerical simulations in which the gaschemistry Is coupled to the discharge circuit.
Uder, certain conditions the gas is observed tobreakdown after the e-beam ceases. The time interval . . SU4ITED BY: ,)?etween the. e-beaum turing off-. and the breakdown.- -- -~-
occurring can be varied. Experiments aimed atunderstanding this phenomenon will be discussed. Robert . Commisso
Code 4770Other experilents, vhich use gases that exhibit thelamsauer-Townsend collision cross section behavior, Naval Research Labor!have been performed. ashington, DC 20375
t Work supported by MvnaL Surface Weapons Center, 202-767-2468Dahlgren, VA
* JAYCOM, Inc., Alexandria, VA 22304
R.J. Commisso, R.F. Fernsler, V.E. Scherrer, and IJZ.Vitkovitsky, IEEE Trans. Plasma Sci. PS-10, 2412(1982).-2L. %line, IEEE Trans. Plasma Sci. PS-10, 224 (1982).
MLR. Ballada, P. Bletzinger, and W.F. -aily, IEEETrans. Plasma Sci. PS-10, 218 (1982).R.J. Commisso, R.F. Fersler, V.E. Scherrer, and 1.111.Vitkovitsky, NRL Memorandum Report 4975 (1982).J.3. Basted, "Physics and Atomic Cblisions," AericanElsevier Publishing Co., Ny (1972), p. 306.
-227-
Lai i1- U 0
X4 zco
z tiCJ
U)U
04
mu-
0 I-c
w co
-12 8-
- 004Jw
0~ _JC
wifro
Uv v
IL
I 0
,z o o W
LLJ W
ii I w
0 CO
IrF-
-W30
EIP=i0.5 V/cm-torr--20%0 2 - 80% N2 50
- - 400U 0 300-
>100 '200
.0 02000-
ipI
1500-
I ~ 1000
a 500-
300 -i
200
* 100-
0 100 200 300 400 500 600TIME (ns)
________________ o
I I I t~Ci
IIIII 0*
I N
I'IN
I 0
III S
U
I
I N
I,,. c'E1
I 6
I1*I
0- - 4
I
qlog
I I I 0('J 0
d 0 cS(Ut>
(Z) III-132-
I
w
THEORETICAL MODEI
DISCHARGE PLASMA ELECTRONS:
(, --).- BODY DISSOCIATE ATTACHMENT : e 4 Ax. --)A
O N -' 3 BODY ATTACHMENT :e + A * "-t "I B-3'2.- BODY DISSOCIATIVE RECOMBINATION: e 4 A. -- AtA
OTHER SPECIES:
CHARGE CONSERVATION:.t,- '- 2: Nt
A A
DISCHARGE PLASMA CURREfiT:
CIRCUIT EQUATION:
DISC ARGE RESISTIVITY:
IN EQUILIBRIUM:
- d- P (o (P, GIAS 91=P-6~T.
-133-
. . . . ... . ....... . .. . . . .. .. . . . . . . . . . .I . . ... . . . .I| I
E/ Pa 10.5 V/cm- ton'1.5 Pulatin
lf,$SIMULATION ~ 0-.80% N2
i01EXPERIMENT'JbA/m
1.0 T50 ns
CL -
* i~S1MLATI0NE/P=10.5 V/cm-ton S
i.,EXPER1MONT Pu3atminlEXPERIMENT20%oz-. 80% N
2.0-' Jbx GoA/cm 2
I I (&.3-.If
I Pc562.Q- cmTO r20 ns
1.0l .0o .41 11 l S1t l i l f
-50~~ S 0 50 10 102020 0 5 0 5
TIM (ns
-14
JI
-135
APPROXIMATE
FRAMING CAMERA IiAGPHOTOGRAPHY I OIS~HRDES
JIFi
TIM t05g/d-136-
1 100.0-10% Op- 90% N?Pulatm.
ftvS 8- 15.5kV.
Vd (0) =12.5kV
*Vd (0)=14kW
1.0-
010 00 2000 3000 4000
io (A)-137-
SUMM1ARY
1. DISCHARGE PARAMETERS SCALE AS PREDICTED BY THEORY/
SIMULATION (CAN DESIGN SWITCHES WITH SOME CONFIDENCE).
2. DISCHARGE CHARACTERISTICS DESIRED FOR SWITCH
APPLICATIONS MAY BE OBTAINED THROUGH OPTIMUM CHOICE
OF GAS.
3. OBSERVED LATE TIME-BREAKDOWN IS MOST LIKELY A RESULT
OF LOCAL HEATING OR SURFACE EFFECTS.
-138-
.Abstxace Sub=LttedSar the Twenty-fifth Annual Ifettlg
Miviiau of fl~a Physics,Xaysome 7 to 11, 1983
Categozy Ihbar*ad Sujc= 4-14 OTHER rCTXS
QO Mo03y ff]EXPe-
Xn~sd~q~tins nto* Xlect~on-Rea= ControlledDiffuse jggqs+ V.r. SER.M1 R..J.. COMSO,
K.!. YDNSU',and'-1 LB.. VIvI!, Naval eeacT~~ a a
* controlledCflfuse discharges to fast, higI. power,* epetitiva opening sitche represents a possible
sol03ution to soo af. the switching problem aocate&.wift iniductive storAgi.. 1 , 2 We repoxt on experizentsIn vhich the physics of these discharges relevant toswi~tching plcain is Investigated-. VariousMixtures of ~rOand IT~2 axe compared for switchng
* ~io.~pe an of emrW deposition -on
4Var#c -uprt by ONR, Arlington, VA arA HSWC,flahigran, VA.* 3AYCOR Xnc., Alexanaria, VA 22304.1 3.Y. Ylerzsler, D. Conte, and 1JCM. ViLtkovitsky, T-T'rans. PI= Sci. PS-S, 176 (1960).
- .2 p~j. Concesso, "i. pezler, V.E. Sehez= re& -
I.E. IVSitlovitskcy, TrM-Tans. Plasma SCi.. P-241. W192), and references therein-.
CONCEPT
-140-
cc0
(0 X0 Cc
w CL
I t
I.z
-141-
go P 0I.z
APPARATUS -
-142-
F.o
wwI
-i'
Cw w~iX0 ~
M _n
ww
I-r
w ul
LI
LeL
C5 T)-143-
- -~ - ' '~- -
I
I
TYPICAL EXAMPLE
I
i
I
I
t-144-
I 6
lIE 300700
E/P=10.5 V/cm-tory 0
2020% Q2 - 80% Np 5005
>1400
2100
0 0200
1500
-~1000
-500-
0
300 i
200-
.0
4 0 100 200 300 400 500 600TIME Cns)
-.14 5-
uu -- - -------
*mo
I0
Ii
IU >IZ U,
146
4
, *1ac
*
I
* THEORY *
I
-.
* 4
p
*U
_
I
I
b
I-147-
- i/
THEORETICAL MODF.
DISCHARGE PLASMA ELECTRONS:
i CC --.2 - BODY DISSOCIATE ATTACHMENT : e -A1 --. ecK2.,v -*3 - BODY ATTACHMENT -A .1--- A-t,
..t "-92 - BODY DISSOCIATIVE RECOMBINATION: e AL -A &
OTHER SPECIES:
CHARGE CONSERVATION:
CIRCUIT EQUATION: L.
DISCHARGE PLASMA CURRENT:
..o
DISCHARGE RESISTIVITY:.-p-
IN EQUILIBRIUM.:
at
~~C l - (p (o fPcoNSTA..) Ag.
-1T4 8~-148-
THEORETICAL MODEL
ATTA C 4,MENT 1: COMfR I A/AT IOAf
U~~ -V P___
ot: A
0(I I~.,,'/43
-149-
E/ Pu[0.5V/cmn-torrL.5 -s Puatm
in,SIMULATON sv.- 20%02 -80% N2
in,EXPER1MENT ~ /m
P 317 12- cm1.0- F50ns
Ua5
30
In, IMULTIONE/P=10.5V/cm-torrt n IMULTIONPs 3atm
- in, EXPERIMENT 20%02-80% N2
2.0-II Jbs 6A/cm2
4c J Pe562Sl- cm
1.0 I
o / IIIA
-50 0 .50 100 150 200 250 300 350 400 450TIME C ns)
-150-
4
.1* -.. - .---- ahi
luc
* '* DISCHARGE PARAMETERSCALING
*
- ... - ..--.-. ,. - -- -.-. - .. -.
SI
S
S
U
S£
S
SI
-151-
SI
70
601 Xp= j tm10% 02- 90% NZ,
x x (E/N) 0 3ZTd
40 *.
30-
20-
OL
c0
~80-P=2atm
70- 10% 02- 90% N2
400
20c
( to- K
0 I .0 0.2 CA 0.6 0.8 1.0 1.2 1.4 16 1.8 2.0 12
Ji, (A/cm2)-52
20001 (E/N) 0 =32 Td
10%02- 90% N2
1500
ci'
0 9'0 90. 10 .
Jb A/
ic153
(E/N) 0= 32 Td
20 b 5A/cm2-
20 1 atmSS A 3citm
CD
z
a:10-
0 10 2030 4050 60 7080 90 100% OZIN NZ
3 0 0 0 f i l l [g s oa u a u m g I a i u a g a i a u a a ~ s a .
2500o CE/N)o 32TdJb 5A/cm2
I ata
3atm
2000. -ellt
I ~c 15 0 0 -
1000/
40 100000
50000,
0 1 1 L 11 I L1 fle lf ill 111 1ti
0 10 20 30 40 50 60 70 80 90 100%02 IN N2
L -155-
I -
* DISCHARGE STABIUTY
-.-. -.. *.- * .--- -- - --
I
I
I
I
I
-A
-156-
I
-- , .\ . . - -I!" --" - '" -- ... -. - . .. ' " . . . I . . ..
.\ (E/N)o -49Td
0-(E/N) o =55Td10% .?_.- 90% NZ
EsB- 15.5 kV/cm
U) 10. :
. ,
cIO
AMBIENTENERGY/PARTICLE o-
0 1.0 2.0 3.0 ;S a o * l Il. . e - 2 1 r" *' m r% - - , - , " - - 1 5 7 -
U APPROXIMATE
FIELD OF VIEW
FRAMING CAMERA* PHTOGRPHYELECTRODES
40f1-
YOn XPSRTIME 0.5/A/dia
in lOOns EXPOSURE
TIME C.Spgs/dlv)
o~p~ Vd(kV)
0-
0-0C
>>
0- 0
N L 44
W (J/CM 3)-159-
* GAS OPTIMIZATION.
-160-
Jb.=: A/cmZP=6atm20%02~ 80%o Ar
Ic~ P=I atm
IQ -20% 02- 80% N52
P=latm 4
20% Op 80%O/ Ar
0 4 8' 12 16 20 24 28L E/ No(Td)
c INTERPRETATION OF
fvs. u
NORMAL GAS (02-N2): CONSTANT p.
RISE AT LOW E/N DUE TO CATHODE SHEATH (Vc"500V)
ANOMOLOUS GAS (02 -AR): VARIABLE j~
RISE OF P. WITH E/N DESIRABLE FOR SWITCHES
• PROBABLE CAUSES:
RAMSAUER EFFECT ON MOBILITY/CEN)
AND/OR
i DISSOCIATIVE ATTACHMENT RATE oa(EIN)
PRESSURE SCALING OF fo SUGGESTS BOTH PROCESSES PLAY
A ROLE:
LOW P, e+o2-?- 04-0- WITH Ok(EIN)
HIGH P, ea+ o 2 +AR. +AR WITH/(E/N)
-162-
1m
--
p
U . S
SUMMARYI
. - - -~ - p
*
. I
p
*41-163-
I
- -.. - -.,..b . -
06. 1. WE HAVE PRESENTED A CONCEPT FOR AN E-BEAM CONTROLLED
OPENING SWITCH, DESCRIBED EXPERIMENTAL AND THEORETICAL
METHODS. FOR INVESTIGATING IT, AND USED THESE METHODS TO
OBTAIN EXPERIMENTAL AND THEORETICAL RESULTS.
2. EXPERIMENTAL RESULTS FOR 20%02-80N 2 AGREE WITH THOSE
OBTAINED BY TIME DEPENDENT COMPUTER CODE SIMULATION.
3. OUR DISCHARGE APPEARS TO BE RECOMBINATION DOMINATED.
THE IMPORTANT DISCHARGE PARAMETERS FOR SWITCH APPLICATIONS,
fo AND 6, SCALE AS THEORY PREDICTS.
4. DISCHARGE INSTABILITY LEADING TO DELAYED BREAKDOWN CAN
BE AVOIDED BY CHOICE OF OPERATING CONDITIONS.
5. PROPER CHOICE OF GAS CAN OPTIMIZE DISCHARGE FOR SWITCHAPPLI CATIONS.
I
lI
-164-1 "
APPENDIX II
*PAPER PRESENTED ON PDS
REFERENCE 12
a o
-165-
1982 IEEE INTERNATIONAL COiFERENJCE UN PLASIA SCIEUCC
"AY 1-19, 1982I j9xperiments with Plasma Switchin.* I.J. Commisso,
and I. :. vitkovitsky, naval Maserch LaboratoryApplicatLon of inductive storage to present requIrement:for Lpulaed, high peowr generators demnds a switch witht*.., ..-,lo ir:! i, n cXl"-ctorintics: it cond'ict1 :.A of cuxrent at 101 rcsitancg (<<l,2) for a tide on
the order of IS, open predictably (i.e., becra resistive Please refer to "First and
with respect to same load) in a tim -<lOOns, and remain Final Call for Parers"open against voltage of IV. In practice, combining afnounr.cr;;,nL for instruction!
12 in preparing your abstract.all of these requirements in a single device has provent, b e- trem.ly difficult. However, some success has
been achieved by using neveral typoes of switches insuccession to obtain the required power amplification.
1 Su;)JuLL c1Levory i,.e:
One such component switch is the exploding wire fuse.
It generally conducts for tens of microsecoids, opensin -150ns and can have recovery strengths of -20kV/cm. Subject category fwmzer:The main disadvantages of the wire fuse
are the joule
heating losses and the necessity for physical]y re-placing it after each use. In addition, the fuse
opening tise is about a factor of ten too long for. same ) Prefer oral sessionhigh rower applications.
one possibility as an improvement over the wirefuse is the use of a plasma as a switch medium. Plasmas ( ) Prefer poster session
have been observed to conduct hundreds of kilcamperes ( ) Ho prefernceof current for 100ns and to open in tens of nano-seconds against several hundred kilovolts.
2 A plasma ( ) Special requests for4. switch is also, at least conceptually, resettable; so placement of this abstraC8 that not only is replacement after each use notc necessary but repetitively pulsed operation may beL feasible.
in this paper we report an investigation to< detersine the feasibility of one scheme for a plasma-rtch. The scheme involves use of a deflagrationplasma gun
3 as a source of helium plasma with high
dir.," ted one.gy k107 eu/s). The plasma so produced.riz.s through a vacuum region and then enters the Subiltted by:gap bet -een two electrodes that have a potential
Sapplied across them. The conduction and opening '- 1. properties of the helium plasma in the electrode
region are then studied. The diagnostics presently . -.einclude voltage and current monitors, streak/framingcamer , faraday cup pres, and spectroscopy. Atpresent, 'the following operating parmters are used (sam rialae typewritte")
' for the gun- VG 12 kV, IG G 2 0
0 kA, CG -55 UF: andfor the switch: - 10 v, I =OkA, C. . 1.85 Uf.s , (ful I addre.si)
As observed rith the streak/framing camera, theplas=a gun ejects three, not totally distinctpla .oids; each correlated in time with an extemum
'- of the gun current with each subsequent one mcvingmore slowly than its predecessor. The first of
, these plasmoids has two cponents, each with a I em a n 7ber of the Coaafttc" different speed. ased on the time of current n a ience and
initiation at the switch, a third component of the on Plasma Science a4J
first planid moving almost twice as fast as the Application -fastest visible front is conjectured. No openingaction has been observed as yet, probably as a result ( ) yes ( ) noof excess plasma and poor plasma spatial distribution.The effects of limiting apperatures and of a crowbaron the pum are being investigated.
*l'ork supported by Office of Naval Research.
tJAYCOR, Inc., Alexandria, VA 22304
1R.D. Ford, D. Jenkins, W.H. Lupton, and r.M.
Vitiovitsky, Third ZEEE Int. Pulaed Power Conf.,Alb-jqirque, IMd (1981) paper 7.2.
2c.u. %ndel and S.A. Goldstein, J. Appl. Phys. 48,1004 (1977).
3 D*. Cheng, Mic. Fue. 10, 305 (1970).
L This form, or a reasonahle fdcSimile, pls two XWrox copies mttst ba rcceeivo: NOT LArL
110MI FEBRUARY ]ST. 1992 at the followinj .,dress: A.J. Alcock, Chfirin,International Conference on Plasma Scien:', c/o Nlation. Research Council of Canada,Ottawa, Ontario, Canada KIA 06.
-166-
_..I
5
S
SaS -
ja.
2
-J0.. 7
z -
* I.7*-. 0 *1
I*4 .. ,~ IS
I ISV.
4.I -
I
-167-
/ ,--.
S
/ .-- -
p w1
C.)U
0. 4J-C Z
J- 0
CAMERA FIELD OF VIEW
-Z GUN
VIEW PORT VGX12KV
f/s f/16
t a 4.4 is t'.4*&s
tu 3.4pus t:7.4ps
*ta2.4 1&s ts 64ps
tul1.4ps ta54~
t aO.4ps t ,.4L
-= 0- GUN CURRENT INITIATION (lOips half period)
-1269-
I It_
aa
*j 2L4
0
o >
aL Ic
IL
-170-
0 cr0*,.. w
144mz-
I3 :1%flI 7
00
lie]
w Ex > >
2U le
0~ CI:
0
OmI
2w I1- .1'
Io .
z La4.
-J
0 a*1*
A
-172-
/ ..
- ~ END .~-m.~.--- .&
0 00
P"MW.
go
* 2z; I- S~~L&J
U~aO 49
T Ti
0- %
jU~LOIi
0 s
* S'X>/
ow
-173
-7
j 38 cm
FARADAY 5m-oCUPS c
DISCHARGEVESSEL -~APERTURES t-PHOTODIODE
~ FRAMINGCAMERA VGU14 KV
FULL DISCHARGE FUSE(f/16)
a Cu wirest 8.0.43 0I1cm x.032 cm dia.)
GUN IIIIII.. fll
CURRENTC (130 KA/div) 0
PHOTODIODESIGNAL
(ARBITRARY) ____________
C 00 51015 20 2530 3540 0 5 10 15 2025 30 3540
TIME (jus) TIME (ps) A
-174-
Iy M --. 0 -. cof4 vf
I. Fos+esi- piOasi'fo. Cowmpanev%+ CvnQ+ seev% Op4;lco&llj)9
12.~
I f-%r53 CvIii 4YNO >,-. *67r w
L3
3. Slower Covw~.~tv ^rS o.~-e yrtot i-C3
iS -y'o LVr*.vA j" -P4 nc.~o
iC~')~ ~~v~OS. (~hybe ao4t'esf cI iwmo. 'Jl
-175-
~0
uu . .. ..
• -17s--.
.I _
cooM, 0
zz
UL-(1
iL~i
-I Z
U) 44
td0 0 %F
-177-
U-
o a.CD ~ w0
E Cl- - U.
U F-)
F-
C4O
a. i w
--
in 0 i
LOz
-178-
KNOW , x
CAVEl 0 0*.i lIU
E~~LI IEZEI IKRLUU nuL
-i -
LU _U U
CDo
LC)
W to
o- U, a0 tonlU) in
*. 4 1
Xa