/

A*D

TECHNICAL REPORT ECOM - 01412 - F

RADIATION EFFECTS ONDIELECTRIC MATERIALS

FINAL REPORT

/. F. COL WELL, D. W De MICHELE, AND R.F. OVERMYER

DECEMBER 1966 D D C

JýNW2967T

BTHIS RESEARCH WAS SPONSORED BY THE DEFENSE ATOMIC

SUPPORT AGENCY UNDER NWER SUBTASK NO, 16,0091

ECOMUNITED STATES ARMY ELECTRONICS COMMAND . FORT MONMOUTI;, N.J.CONTRACT DA28-043 AMC-01412(E)GENERAL DYNAMICH

General Atomic DivisionSPECIAL NUCLEAR EFFIECTS LABORATORY

SAN 0100, CALIFORNIA

DISTRIBUTION OF THI1 DOCUMENT IS UNLIMITED

Technical Report ECOM--01412-F RCS OSD-'. 366QA-7474 December 1966

RADIATION EFFECTS ON DIELECTRIC MATERIALS

FINAL REPORT

1 July 1966 through 30 September 1966

Report No. 5

CONTRACT DA28-043 AMC-01412(E)

AMC Code 5900. 21. 830. 44. 00

Prepared by

J. F. Coiwell, D. W. DeMicheic, and R. F. Overmyer

General Atomic Division, General Dynamics Corporation

Spccial Nuclear Effects Laboratory

San Diego, California

For

U. S. Army Electronics Command, Fort Monmouth, N. J.

DISTRIBUTION OF THIS DOCUMENT IS UNLIMITED

This research was sponsored by the Defense AtomicSupport Agency under NWER Subtask Number 16. 0091.

ABSTRACT

A detailed description of the techniques and equipment employed inthe ion-implantation process is presented. Results of high-energyelectron irradiation of a Mylar capacitor with ion-implanted electrodesare compared to the response of a control sanrple. The ion-implantedelectrodes exhibit better carrier injection than conventional foil elec-trodes. Transient radiation effects data from irradiation of a monolithicceramic capacitor and two glass capacitors are included. One of the glasscapacitors contained a sernicrystalline dielectric with very high dielectricconstant. The transient radiation effects in Mylar verlus tempe ratur('appear to fit a band model for the conduction process in which a continuousdistribution of carrier traps is located in the forbidden zone and one signof carrier is immobilized. An examination of the energy loss processesof moderately fast electrons in insulators indicates that from 15 oV to3 eV the energy loss is primarily to excitons whereas energy loss tooptical and acoustical phonons occurs down to thermal energies.

C.

'I.-

FOREWORD

The objectives of this program were to determine the effects ofgamma and neutron irradiation on dielectric materials and electronicparts and to apply this knowledge to the accurate prediction of radiationresponse in such parts. Further, models of electronic parts were to bedeveloped which exhibit high radiation resistance.

This research was sponsored by the Defense Atomic Support Agencyunder NWer Subtask No. 16. 0091 and was authorized by the ElectronicComponents Laboratory, U. S. Army Electronics Command, FortMonmouth, N. J., under Contract DA 28-043AMC-01412(E), AMC Code5900. 21. 830. 44. 00.

iii

CONTENTS

1. INTRODUCTION ... . . . . . . .. . . . . . . 1

2. TRANSIENT RADIATION EFFECTS IN A MYLARCAPACITOR WITH ION-VIAPLANTED ELECTRODES.... 2

3. ION-IMPLANTATION TECHNIQUE ................... 7

3 . 1 I n t r o d u c t i o n t o . . . . . . . . . . . . . . . . . ..lp . . .f . . .o 7

3.2 Ion Im plantation . .................................... . 73. 3 The Vacuum System . . . ...... . . . .o . . .. . .. 83.4 Support Equipment ......................... 103. 5 Cathode and Shield Design .................... 103. 6 M etal Evaporation ........................ 153. 7 Monoenergetic Ion Source ................... 203. 8 Ion Implantation on Mylar ................... 203. 9 Conclusion ......... a . . . . .. *. . . . . . .. . #. . 24

4. TRANSIENT RADIATION EFFECTS IN MYLAR VERSUST E M P E R A T U R E . . t o . .. . . . . .. . . . . . .. . . . . . . . .a . 25

5. LINAC TESTS OF CEB AMIC AND GLASS CAPACITORS. .t. 23

6. ENERGY LOSS OF MODERATELY FAST ELECTRONSIN IN S U L A I'O R S . . . . . . . . . . . . . . . . . .o o. . . . . . . . . 30

6. 1 Introduction .... . ....... . . . .. .... . . .6. .# . 306. 2 Mathematical Details . ........ . . . ..... . .. 316. 3 Formula for Linear Range .............. 366. 4 Numerical Estimates of Range ............ .. 396. 5 Conclusions....... ...... . . . . . . . . .. . . 45

7. CONCLUSIONS ...... ........................ 46

REFERENCES ................... .. . . .... 48

Figures

2. 1--Total transient sLgnal for two Mylar capacitors 3......... 32. 2-- Prompt signal from two Mylar capacitors ............. 52. 3--Response of Mylar capacitors with + ,2. 5 V applied ...... 63. 1--Plasma implantation . . .............................. .. ....... 9

V

1. INTRODUCTION

During the final quarter of the contract, the ion-implantation processof applying electrodes to Mylar film was perfected. in this process,metallic ions are accelerated into the substrate material to form a diffuseinterface between the two materials while the surface is continually beingcleaned by sputtering with an inert gas. Contacts on metals, semicon-ductors, and insulators have demonstrated strong physical bonding, goodelectrical contact, and low impurity concentration. A detailed descriptionof the process and the equipment developed for ion implantation is includedin this report. Results of linear accelerator (Linac) irradiation of a Mylarcapacitor with ion-implanted electrodes are comp,_red with a controlsample which had foil electrodes. Although no dramatic differences wereobserved in the transient response, the signals did show dissimilarities,which are explainable on the basis of better carrier injection by the ion-implanted electrodes.

Irradiations were also performed with the Linac on a monolithic typeof ceramic capacitor and on two types of glass capacitors. One of theglass capacitors had conventional glass dielectric, whereas the uther wasconstructed with a new dielectric material, Glass-K, that has a high die-lectric constant comparable to the barium titanate ceramic formulations.The difference in properties evident in Glass-K are due to the heat treat-ment given the dielectric material to change it from an amorphous glassto one which contains microcrystals.

Theoretical examination of the energy loss of relatively low einergyelectrons in insulators is presented to strengthein the arguments rsed inthe theory of transient electrical efects in irradiated insulators.

Further consideration of the results of the transient effects meas-urerments in Mylar versus temperature that were discussed in the lastquarterly report(Z) are also included in this report.

Accelerator Pulsed Fast Assembly (APRA) irradiations were notperformed during this quarterly period as the joint construction-operating permit had not yet been issued by the Atomic Energy Commissionand the facility could not be completed.

1 , ,. .

2. TRANSIENT RADIATION EFFECTS IN A MYLAR CAPACITORWITH ION-IMPLANTED ELECTRODES

A parallel-plate capacitor with ion-implanted aJuminum electrodeson Mylar was constructed and irradiated with the Linac. The ion implanta-tion process used to apply the aluminum electrodes on the 1-mil Type CMylar is described in detail in Section 3. During each implantation, six1 in. by I in. capacitor sections were processed on one side by maskingsuch that an 0. 8 in. by 0. 8 in. electrode was implanted on each section.Twenty-four such sections were assembled with -mril ' ̀ad foil inter-leaved in alternate directions between sections to provide the electricalcontact to the implanted electrodes. Stiff fiberglass sheets held thecapacitor sections together by means of metal. clips at each end of thecapacitor. The metal clips also provided electrical contacts to the leadfoils brought out to each end. The construction was similar to that usedin commercial silvered-mica capacitors. As the experimont was per-formed in vacuum to eliminate effects due to air ionization, no pottingmaterial was used to seal the capacitor except the epoxy used to keep themetal clips in place. A similar capacitor was constructed in the samnemanner using Mylar from the same original spool of film, but without theion-implanted electrodes, for the control sample. Previous rejults oftransient effects measurements of Mylar capacitors with foil electrodeconstruction and metallized or vapor-plated Mylar had shown essentiallyno difference between the response of the two types. (3)

The two capacitors were irradlatud with 30-MeV electrons from theLinac at 2 x 109 rads (Cu)/sec with 4-pscc pulses. The test circuit,utillzinz a 10-k•l series-measuring resistor, was described in an earlierreport.(4) The results of the test were as follows:

I. The total signal (prompt plus delayed) was the same within afew percent in both types for the first pulse.

2. The total signal from the foil type for the tenth pulse was 5 to 10percent less than the signal from the ion-implanted Mylarcapacitor. The foil type achieved its final value after three tofour pulses, or about 30,000 rads (Cu), whereas the signalsfrom the ion-implanted capacitor decreased more slowly foreight to nine pulses, or about 90, 000 rads (Cu). Figure 2. 1illustrates the magnitude of the response of the first and tenthpulses for each capacitor.

2

3 .0 - _..... . .... . .. '

o MYLAR-FOIL I ST PULSE

A MYLAR-FOIL IOTH PULSE

o t A-MYLAR I ST PULSE

2.5 Al-MYLAR IOTHPULSE

Pd

0

x:1 2.0

0

0 1.50

w

LL~0z

1.0

0.5

00 5 10 15 20 25APPLIED VOLTAGE (VOLTS)

Fig. 2. 1--Total transient signal for two Mylar capacitors

3

3. The pro!.npt signal was a few percent larger in the foil-typecapacitor after correction for the different secondary emissionslgnals at OV, as indicated in Fig. Z. 2.

4. The delayed component of conductivity was about 50 percentlarger in the capacitor with the ion-implanted contacts. Thetracings of the transient signals from both capacitors inFig. 2. 3 show the difference in the delayed components afterthe initial prompt rise in the signal.

The observation that the transient-response signals decreased moreslowly with succeeding pulses in the capacitor with the Ion-implantedelectrodes indicated that polarization was still taking place and that theelectrodes were not perfectly injecting but that they were better than thefoil type. During the pulse of ionizing radiation, the Insulator-electrodeinterface is essentially bathed in a plasma such that there should havebeen little or no difference in the prompt response of the two capacitors.Slight material differences due to variatlons in the treatment of the twoMylar samnples, such as the heating during the cleaning and Implantationprocess, could account for the difference observed. The larger delayedconductivity signal in the ion-implanted type is an indication of a betterelectrode than the foil type, since the former allowed delayed charge toreach the contacts after the ionizing pulse and be observed in the extornalcircuit rather than be trapped at the insulator-eleetrodu interface.

Further invost.gatlons of the effects of varying the duw• and dowsrate are plarnned.

4

3.0

0 MYLAR-FOIL I ST PULSE

SMYLAR-FOIL I0TH PULSE

O At-MYLAR I ST PULSE

2,5 Al-MYLAR 10TH PULSE

T0x

S2.0

% ,0

z

u

I'

~ 0

0 5 10 15 20 25

APPLIED VOLTAGE (VOLTS)

Fig ., 2.--Prompt ,4ignal fro•m two Mylar 'apl•J•'.to-t4

5

" I I I i I i Ii i

VERTICAL. 2V/DIV

SWEEP; IO0.SEC/DIV

MYLAR-FOIL CAPACITOR RESPONSE,11,700 RADS tCu)

¶ i ' , I I i . .. I I I

VERTICAL' I V/DIV

SWEEP: IOuSEC/DIV

i I i - I I.A I i ]ION-IMPLANTED IAYLAR CAPACITOR

RESPONSE, 14,OO RADS (Cu)

Fig. 2,A3 .. Respons , of M lar capacitors with +22..5 volts appliod

3. ION-IMPLANTATION TECHNIQUE

3. 1. INTRODUCTION

A description of the techniques of ion implantation and the equipmentutilized is given in this section. The techniques and the necessary supportequipment have been developed in the past six months. The advantagesand limitations of ion imlantation are also pointed out.

Ion 4mplantation, or ion plating, is used to form a diffused metallicinterface between two dissimilar materials. The diffused surface layerprovides a strong physical bond, good electrical contact, and low impurityconcentrations. Insulators, semiconductors, and metals have been platedby this method, and each of these three different types of substrates re-quires a different approach to successfully implant a metal layer. Fourarticles by D. M. Mattox(5"8) were most helpful in developing the tech-niques which are described here.

3. 2. ION IMPLANTATION

Iln general, ion implantation consists in injeccing metal ions into asubstrate material. The metal ions are normally formed from a mltalvapor iv an argon plasma and the ions are accelerated toward the substrateby a large negative potential gradient. Upon arrival at the surface, thehigh-energy ions cause several interactions: they knock off physicallyadsorbed impurities; their high energy allows them to break the chemicalbonds of oxides; and,finally, their momentum carries them deep into thelattice of the substrate. Some substrate atoms are displaced and sputteredinto the vacuum, others form intorstitials in the lattice. After a period oftime, the surface of the substrate is free of chemically and physicallyadsorbed imnpurities %nd a diffused interface of the substrate and metalbegins to form. As n:,ore and more metal ions are implanted, the surfacebecomes more m.tallic.

Metal ions have been formed by two different methods. The firstmethod is to establish a plasma with a noble gas. The meta to be im-piinted is evaporated into the plasma,where it is ionized. These metalions drift through the plasma and eventually fall through the potentialgradient between the plasma and the cathode on which the substrate ismounted. The ions from this type of source are not monoenergetic. Mostof these ions are, in fact, at low energies. This type of source is easier

7

to run and the equipment requirements are not as extensive as the secondtype of source. The second method of producing metal ions does not re-quire a noble-gas partial pressure and the ions are essentially monoener-getic. The ions are formed inside a small chamber and extracted. Theseions then fall through the full cathode potential.

The plasma implantation technique is illustrated schematically inFig. 3. 1. Both methods will be discussed in more detail.

3.3. THE VACUUM SYSTEM

A high-vacuum evaporator system is used to house the implantingequipmrnt. This system has a clean base pressure of 7 X 10-7 torr. Thesystem is composed of a fore pump, a molecular- sieve cryogenic trap onthe fore line, a, 4-in. oil diffusion pump, a cryogenically cooled chevronbaffle, and an 18-in. Pyrex jar.

The stainless-steel base plate is equipped with various ports andfeodthroughs. A linear rotary-motion feedthrough is available. Paddlesattached to this feedthrough are used to baffle the pumping port when aslower pumping speed is desired and to optically shield a substrate froma metal-vapor source. An electrically insulated tube is located in thecenter of the vacuum system to leak in gas at a controlled rate. Thereare three different valves for controlling the leak rate into the system --a variable leak valve is used for a fine control of the leak rate; a manualon-off valve is provided for manual control; and a solenoid valve, which isnormally used in conjunction with the plas-na-ion source, is located inseries with the other two -valves. The solenoid valve is operated by ameter relay. The meter relay monitors the current being drawn at thecathode. If the current exceeds a predetermined value, the solenoid isclosed and the pressurc in the vacaumn system falls. With a decrease inpressure, the current also decreases. At a predetermined lower limit,the solenoid is opened and the pressure and current begin to rise slowly.Normally, the maximrui current is set at 45 mA and the minimum currentat 40 mA. The regulator can easily maintain the current within theselimits. The prescribed operation of this leak-rate control system is asfollows. The maximum and minimum values of the current is set on themeter relay; the mnanual and variable leak valves are closed. The cathodevoltage is set at the prescribed voltage; the relay control switch is turnedon. With the variable leak valve closed, the manual valve is opened;while the current meter on the power supply is monitored, the variableleak valve is slowly opened. The leak valve is adjusted until, with thesolenoid in an open position, the current rises at a very slow rate. Afterthe controls have been set in this way, the current will slowly oscillatebetween the two values set on the master relay.

8

CATHODEINSULATORFEEDTHR QUO

CATHODE ---- CATHODE BASESHIELD PLATE(FLOATING) SUBSTRATE HOLDER

______________(NEGATIVE HIGHVOLTAGE)

A )>rPLASMA SHAPE

FILAMENT SUPPORT /'2 UINU ATM

(GROND PTENIAL)@*-METAL SHIELD CUP(ELECTRICALLY

ZMOLTEN ALUMINUM

Fig. 3. 1- -- Plasma implantation

9

Thez.e are two bell jars for this system. The first--a standard18*-in. -diam jar with a dome top--was used with the first series of tests.The substrate holder used with this jar had no facilities for heating orcooling the sample. The second jar is an 18-in. -diam, 24-in. -high

cylinder, on top of which is a 19-in. -diarn stainless steel flange. Sixhigh-voltage feedthroughs are mounted on the flange. The sample holderused with this jar can be heated or cooled and the temperature of the

sample mount can be monitored during the implantation.

3.4. SUPPORT EQUIPMENT

The high-voltage cathode power supply is normally run with a nega-tive voltage output. Its output is with respect to ground. An interlock isprovided such that when the bell jar is in an open position, the high voltageis interrupted. This supply is not well regulated, and the voltage changeswith a changing current. It has a maximum voltage of 20, 000 V andmaximum current of 50 mA. The output lead of this supply is in serieswith a 50, 000-ohm resistor, whiuh acts as a partial current-limiting dc.-vice and also stabilizes the discharge. A schematic of this support sys-tem is shown in Fig. 3. 2.

3. 5. CATHODE AND SHIELD DESIGN

In order to establish a localized plasma discharge, the cathodeassembly and sample holder must be carefully designed as it is desirableto have the plasma strike only the face of the substrate holder. If theplasma strikes extraneous parts of the cathode, contaminants will besputtered into the plasma, where they will be ionized and possibly im-planted in the substrate. To localize the plasma to the area around thesubstrate holder, electrostatic shields arc used. These shield,,, alsoprevent the evaporated metal and ionized metal from coating the insulatorthat supports the cathode. As pointed out by Mattox, (5-8) these shieldsare effective only if they are placed within an electron moan free path ofthe cathode section to be shielded. A separation of 0. 25 in. is quiteeffective for pressures less than 20p. If it is desired to have the argonpressure higher than that, the separation distance must be reduced.Electrically, the shields are allowed to float. The cathode, of course, isat a high negative potential with respect to ground. Two different cathodeshield designs have worked quite well. The first cathode design, as pre-viously mentioned, has no provisions for temperature control or tempera-ture nmonitoring. A nondimensional drawing of this shield design is shownin Fig. 3. 3. The second cathode assembly, which is now being used, hasthese provisions. Schematics for the isolation transformers, heaters,and therrnocouples are given in Figs. 3. 2 and 3. 4. A scale drawing forthe second cathode design is given in Fig. 3. 5. All temperature controland monitoring can be done while the cathode assembly is at high voltage.

10

CATHODE POWERSUP PLY

CATHODE C1JPRENT-METER

M ETE R -fZ-a i HEATERRELAY IAUTO-CONTROLj TRANSFORMER

FFI 10 A C-20 KV MAX

II0OAC _____ 0L~ ... JI TRANSFORMERSOLENOID UAC 25KVVA LV E HIGH-VOLTAGE

CONTROL 110 AC

50OxKl 3AMP MAX.

CEATHOER HAE HA

R LI I

ALE1 MELAENT

I SESTEM

2-hW FILAMENT

POWER SUPPLY

FILAMENTCONTROLAUTO-TRANSFORMER

Fig. 3.2--Support equipment

INSULATED WIRE

CERAM IC-TO-METALFEE OT HROUGH HTILS TE

STAINLESS-STEEL SADF

.- INSULATOR SHIELD

PL A S M A ' --- SU B STR A T E H O LDER

3/4 IN. COPPER ROD

THERMALTANTALUM ,_ RADIATORSBOAT

CERAMIC -STAXNLESS STEELSTANDOFFSHIELD

CEIRAMIC-TO- METALFEEDTHROUGH

Fig. 3.3--First cathode assembly

12

CATHODE RESERVOIRTHERMOCOUPLE THERMOCOUPLE

F---- --

I I13-WAY SWITCH

I ° ° •I

"THERMOCOUPLE METER+40 - -200°C

"---THERMOCOUPLE METER0 0

0 - +400 C

Fig. 3. 4--Cathode-assembly termporatuire-monitoring syst em

13

PHENOLIC SUPPORT RODS

-IQUID NITROGENIQOUID NITROGEN H-FEEDRETURN

T RESERVOIRTHERMALFINS HEATER

THERMOCOUPLE

-THERMOCOUPLE.~,-AND HEATER

LEADS

RTV SEAL

CERAM IC-TO- METALNYLON SEALSCREWS 0-RING SEAL

NEIORENESEAL ,,STAINLESS CATHODESHIELD

COPPER CATHODE CATHODE HEATERS

/PERFORATED- -:L, SAPEHDRMETAL SAFETY BELL JAR -SCRPEENODE

SHIELD RESI -I SCEE

BAFESAMPLE HOLDER THERMOCOUPLE

FILAMENTANODE SHIELD

FINLAMETT

SUPPORTS

SC ALUE

ROTARY VACUUM POR~TBASE SEA

Fig. 3. 5 -- Ion- im-plantation system with cooled substrate holder

14

Both cathode assemblies must be cleaned periodically. When anabnormal discharge occurs, it usually is an indicatiu,, fhat the assemblyneeds to be cleaned. When abnormal breakdown does occur, the high-voltage cathode power supply will indicate an overcurrent condition. Thi3type of breakdown happens much too fast for the current regulator systerr,previously mentioned, to respond; repeated abnormal discharges canperforate the electrostatic shields. All metal parts are cleaned withwater and alcohol. The ceramic parts are cleaned of all metal that hasbeen deposited.

After each run, the cathode assembly is heated to expel the waterthat condenses inside the reservoir. Failure to do this will result in thecooling system becoming plugged with ice during the next cooling cycle.

Two thermocouple meters mounted inside the high-voltage sectionare used to measure the temperature from -200' C to +400° C. An insu-lated switch is provided to change scales. There are two thermocouplesmounted on the cathode assembly; one measures the temperature on theback face of the sample holder and the other monitors the temperatureinside the liquid nitrogen reservoir and provides a crude measurementof the liquid nitrogen level.

The physical shape of the sample holder determines to a large de-gree the plasma form and the implantation rate. Several different typ!,,of sample holders have bcn used. The uniformity of the implantationcan be determined by visually inspecting the plasma shape. The implai1a-tion will be thickest where the plasma intensity is greatest. Two types ofsample holder that have 'been used with success and the shape of theplasmna are shown in Fig. 3. 6. Intensifiers may be added at points wherethe plasma is weak. The intensifiers provide secondary electrons inlocalized areas and thus intensify the plasma in those areas.

3. 6. METAL EVAPORATION

After a stable, localized plasma is achieved, the metal to be im-planted must be injected into the plasma by evaporation. The evaporationof most metals is a straightforward process. Some metals, however, arenot easily evaporated. Aluminum, which is used for implantation of semi.-conductors and insulators, is particularly troublesome to evaporate.Several weeks were spent trying to develop a continuous source of alumi-num which could be used for extended periods of time. The electron bom-bardment technique would be useful in a straight evaporation process, butit is not suitable for ion implantation because of the presence of the plasmaand the high partial pressure of argon.

15

L.

CATHODE CATHODE

'PASA~)PLASMA INTENSIFIERS

CATHODE CATHODE

INTENSIFIERS

PLASMA• l , I A I J ,t,/ i

P '' lN 7V

Fig. 3. 6--Cathode ; with --nd wit:h out ititer' ifioiirH

1 6

Tungsten and tantalum boats were the first type of boats tried.These boats proved to be unsatisfactory for evaporation of alurninum.The aluminum etched the boat and large holes wvere formed during thefirst minute of evaporation. Several other materlals were tried, includ-ing boron nitride, alumina, and graphite. These, too, were found to beunsuitable. It was finally deterrined that t-ngsten filaments were thebest available method for aluminum evaporation. These filaments haveto be loaded with aluminum at the beginning of each run. The filamentsalso have to be changed after every three or four evaporation runs.Commercial filaments were found to be less versatile than those whichcould be made in the laboratory. Tungsten wire with a 0. 06 in, dian,-eter is easily formed into the desired filament shape. The filamintshown in Fig. 3. 7 has the longest filament Life. This configuration wasalso designed to approXimate a plane soirce as a plane source providesa more uniform depositiun than a point source; in addition, n'Iore uni-form heating of each strand of the filanumnt is attailied with this design.Soft aluminum is cut and formed into small hooks which are placed onthe flat section of each strand. When the aluminum melts, the surfacetunsion causes it to ball up and hang from the f1larment. The molten ballwill then spin acout the axis of thu strand and run back and Xorth alongthe flat suction of the strand. This approximAtes four line sources in aplane. Because the ball runs along the strand, the tungston Is etchedmnore uniformly, thus giving a much longer filament lifo, Aij shovvn inFig. 3. 8, the filaments are shielded to prevent coating thio vacuull) walls.A mirror placed as shownm allows one to observo the filamunt.

Nickel aud nichronie oxhibit the same characturistics as aluluminumwhen thliy arc, evaporated flrom rfrLaVtUV'y 111ULaL boats or filarI11tAs,These metals, however, can be evapourated from refractory miutal boatswith alumina and boronx niitride coatixgs. Boats of this type are comnior-cially available.

MOWetls such as copper and gold do not work well with filamentsbecause of their low surface tension in a, molten 8tiate, Tantaluin boatsare suititAlO for such mnotals. TLantaLum boats are easily formed fronm10-nill twitalum sheet.

V~ a high partial pressure of a noble gas is used for the evaporation,sorni of the equipment in the chainbor will be coated with an amorphousblack material; uoretimes this also 'occurs dol-iig implantation. Aspectral analysis hao shown that this black soot is a puro anmorphotus

form of the evaporated motal.

17

TUNGSTEN FILAMENT

TANTALUM BOAT

g3 7- - 8FVaovat io Ioato vs

18

TOP VIEW

-MIRROR

ALMNM4OKCT"..

TUNGTENSUP POR T

SiDE VIEW

Fig, 3,8--Alumrinum vapor source

19

3. 7. MONOENERGETIC ION SOURCE

To obtain deeper penetration and higher-energy metal ions, a con-ventional ion source was designed and built. The metal ions are createdinside the source and extracted. These ions then fall through the fullcathode potential. A noble gas is not required to sustain this source.

A pure aluminum plasma is established inside the source by usingan energetic electron source, a neutral aluminum vapor, and appropriateelectric and magnetic fields to enhance the ionization process. Byoptically shielding the aluminum vapor source, a low neutral efflux ismaintained. Other metals can be used for the vapor source if desirable.

Electrons emitted from a hot tungsten filament are accelerated and in-jected into a field-free region. A magnetic field of about 400 gauss isdirected normal to the direction of the electron line of flight. Aluminum

evaporated into this region is ionized by the orbiting electrons.Aluminum ions drift out of the field-free region and are extracted pastthe electron emitter into the chamber. These extracted ions then fallthrough the full cathode potential and are implanted into the substratemounted on the cathode. With the source configuration shown in Fig. 3. 9,a 30-mA beam can be maintained for approximately 2 min.

By leaking in argon, the source can be run on the residual argon.A partial pressure of 4 x 10-4 torr is required. At this pressure, themnean free path of the argon ion is well over 1 meter. This fact assuresthat the ions are monoenergetic and have not experienced any collisionsduring their flight between the source and the cathode. This monoener-getic argon ion source can be used to clean the substrate surface prior toimplantation of the metal.

This ion source has been run several times with no apparent prob-lems. This type of monoenergetic ion source has many experimentaladvantages over the plasma source. Because the metal ions are mono-energetic, the energy and depth of penetration may be controlled to prod-uce more consistent results. The absence of the plasma eliminates mostcontaminant sputtering and subsequent implantation. The possibilit ofusing this source for general production of samples is questionable.

3. 8. ION IMPLANTATION ON MYLAR

Metals, insulators, and semiconductors have all been successfullyimplanted with ionized metal. Implanted metal surfaces have been solder-ed and spot-welded with reasonable success, and ohmic contacts have beenmade on high- resistivity semiconductors.

20

r- ............O RII zI

0

I- w! 0C1

z0

+'-4

z I

4 w

U) 22

Implanting metal ions in insulators is the most difficult of the threetypes of substrates. Because of the low thermal conductivity associatedwith insulator materials, the exposed surface and implanted film may be--come very hot. The low electrical conductivity results in a large surface-charge buildup, Arcing across the surface film during the implantationwill burn holes in the film and insulators.

Glass, quartz, mica, and Mylar have been implanted with aluminumand copper. The general information for establishing the plasma and thecleaning procedures are all applicable to the process of insulator im-plantation. Most of the problems arise at the surface of the sample. Toreduce the problem of high surface temperatures, lower power levels areused in the implantation of the metal. The substrate holder is usuallycooled to a low temperature prior to starting the plasma. To reduce thesurface charge, a screen is mounted over the sample about 0. 25 in. fromthe surface. This screen is at the same potential as the cathode. Thepurpose of this screen is to establish an equipotential surface in front ofthe substrate. The screen will produce many secondary electrons du.e tothe ion bombardment. These electrons will tend to neutralize the positivecharge on the surface of the insulator. The screen also reduces the num-ber of secondary electrons which leave the surface of the insulator. Someneutral fiux and some ions will reach the insulator even if the screen isabsent. It will be observed, however, that as the conductive film beginsto form, arcing across the surface will occur. When one point of the filmtouches the mask or any part of the cathode, the surface charge will rushto the point of contact and subsequently burn the film and insulator. Ithas also been found that if both sides of a thin dielectric are to be im-planted, the front, or first, side of the film must be insulated electricallyfrom the cathode when implanting the reverse side. Mica was implantedon both sides, and it was found that the front side of the film would burnand pit when the reverse side was implanted. The reverse side was obvi-ously in a cooler environment as it was pressed against the sample holder,but yet it was destroyed each time. When Mylar was implanted, largeholes would be torn in the Mylar when implanting the second or reverseside. The substrate would remain intact until the initial metal began toimplant itself into the Mylar. It was apparent that the problem is electri-cal in nature. It is not yet clear why the phenomenon occurs. The prob-lem was eliminated by placing a 1-mil sheet of Mylar between the sub-strate holder and the dielectric sample when implanting both sides.

Using the above information, most insulators can be implanted witha tough adherent film. Mylar, because of its low melting point, requiredsome additional techniques to successfully implant it. The sample hc(lderfor Mylar is made of aluminum in the shape shown in Fig. 3. 10. TheMylar film is stretched over the curved face and spring-loaded. Loops

22

TENSION ADJUSTS

RODS"•"• A LUMINU M SAMPLE SPACER.

MYLAR_,.,,• MOUNT y f r,..

INTENSIFIER ý-SCREEN ITNIIR

MOUNTING RODSMASK

/4-

Fig. 3. lO--Mylar sample mount

23

/ °-

are made at each ena of the Mylar strip, as shown. Ordinary cellophanetape is quite suitable for this. A thin rod is placed through each loop (asshown) and hooked by the two spring -vires. The exposed surface iscleaned with ethyl alcohol. The aluminum mask is placed over the sur-face but does not touch the Mylar. The screen is placed over the maskand the intensifiers are fitted at each end to provide a uniform deposition.This unit is then screwed tightly against the cathode. Two shields areslipped down around the Mylar sample holder to limit the plasma to thesurface only. The sample holder is heated to 150°C during the pump-down. This anneals any small wrinkles and helps to clean the surface ofthe Mylar. After the pressure is below I x 10-5 torr, liquid nitrogen ispumped into the cathode reservoir. After about a 5-min delay to allowthe sample holder to reach equilibrium, the cleaning process is started.The cathode voltage is initially set at 6, 000 V. When a current density of0. 3 mA per square centimeter is maintained by adjusting the argon pres-sure, the cathode voltage is 3, ZOO V, owing to the voltage drop acrossthe stabilizing resistor and the voltage decrease in the power supply. Acleaning period of about 30 min is required at that power level to clean thesurface of the Mylar. After this cleaning period, the aluminum is intro-duced into the argon plasma. The implantation can be observed. Aninitial tan hue will first appear. This quickly changes to the metalliccolor of aluminum. After the implantation has covered the surface, thepressure can be reduced rapidly by shutting the manual control valve forthe argon leak. The plasma will reduce in intensity and stop and the sur-face will continue to be coated but with an evaporated film. The substrateholder is then heated back up to 1500 C. This anneal eliminates thewrinkles that formed due to the aluminum coating. The process is re-peated for the reverse side, but care must be taken to insulate the alumi-nized surface from the cathode.

3. 9. CONCLUSION

The results of this work have shown that ion i.mplantation can beused as a very effective technique for plating various surfaces. Althoughthe methods described here have been quite successful, simpler techniqueswill probably evolve with continued investigation. The ion-plating systemhas proved to be quite functional.

24

4. TRANSIENT RADIATION EFFECTS IN MYLAR

VERSUS TEMPERATURE

The data in Table 1 were taken during irradiation of a Mylarcapacitor from liquid nitrogen temperature (LN) up to room temperatureand were reported in the fourth quarterly report. (2) The delayed con-ductivity signals were analyzed into the individual exponential componentslisted. It was noted that the time constants of the two shortest components(-6 Lsec and - 40 Asec) were essentially independent of temperature,although the magnitudes of the components indicated fairly strong temper-ature dependence. Carrier lifetimes in the conduction band in insulatorsare usually orders of magnitude shorter than the observed decay times.Therefore, recombination processes occur after the carriers, electronsfor example, are trapped and thermally untrapped many times beforerecombining with holes at recombination centers. At room temperature,the decay of the radiation-induced conductivity was fitted to a hyperbolicform, indicative of direct recombination, only out to 100 psec. It seemsmost reasonable to attribute the observed response to a continuous distri-bution of traps in the forbidden zone in addition to the relative immobilityof one sign of carrier. Under these conditions it is possible to have anobserved time constant which is relatively insensitive to temperature anda photocurrent which increases slowly with temperature, (9) as displayedin Mylar. It is necessary, according to this model, that the rate ofexchange between conduction electrons and trapped electrons be largecolnpared to the rate of exchange with recombination centers. Therecombination centers or primary centers can be a factor of 108 largerthan the number of electrons in the conduction band and a desensitizingeffect is observed in which the photocurrent is reduced to 10-8 times thephotocurrent in the absence of trapping. Since replenishment of carriersin the conduction band from the traps takes place during the recombinationprocess, the observed time constant is considerably longer than carrierlifetime. In this model To = 10-I Tnp /nc , where T 0 is the observed delaytime, r is the average carrier lifetime, np is the number of trapped holes(-I01 5 /cm 3 ), and nc is the number of conduction electrons (107 to 1015/cjn 3 ). In the absence of replenishment, the observed time constant ismuch shorter. Depending on trap depths, both decay time constants mayoccur in the same specimen, leading to observation of long and shortcomponents in the decay curve. Thus, it is reasonable that the entiredecay curve could be analyzed into several components with time constantswhich were insensitive to temperature.

25

Nj Lfl0 0

0 C

-u2

r- v 00 - P-

Nle N 411 cc rf- ~ -0) Q

('4) 0n N- N- 0f N -

NG~~,1 _ -" 00 4

C)C

P- F- -4 -

00

U 4N N Nn

0

N (JN N- N' t -

N I I I01 0 0 0. 0 .0 .

"-4 P X X )< x <N x 1 N -RI I'D 00

0-4 r') Lnt .COLf fl

CN N, (1ON 4-NN r~1l 0

0 &I 00 006

The reduction in photocurrent at the end of the experiment comparedto the original value at room temperature is evidence of deep carriertraps which were filled during the irradiation at the low temperatures andwhich were not thermally emptied as the temperature was raised back toroom temperature. This effect has been observed in other insulators thatagain displayed the original photocurrent at room temperature after thesamples had been heated.(10

It is somewhat presumptuous to apply band models to Mylar, whichdoes not resemble the inorganic crystalline solids usually associated withsuch models; however, the behavior of the observed response may befitted to such models and therefore they are of value until additional ex-periments lead to more definitive models.

27

5. LINAC 'TESTS OF CERAMIC AND GLASS CAPACITORS

Linac irradiations of two new types of capacitors wore performed tosee if a significant improvement in transient radiation response was evi-dent. Only one sample of each type was tested as this was only a surveyto determine whether further testing would be desirable. Irradiationexposure rates of 107 to 109 rads (Cu)/soc were employed.

The ceramic capacitor has a capacity of 0. 1 /r at a working voltageof 100 V. Thin films of ceramnic and electrode matorial are fused Into amonolithic solid that is 0. 3 in. by 0. 3 in. by 0. 1 in. in sizo. The largetemperature coefficient of these high-dielectric-constant cor anmi forunu-lations, which have a ± 15% chanige in capacity over the temperatureinterval -55 'C to 125 *C, limits their use to noncritical circuit applica-tions. No delayed conductivity was observed with 4-psuc Linau pulses.The prompt coefficient, Kp, was calculated from the expression I K) VC

where V is voltage applied in volts, C is capacity in farads, 'Y is theexposure rate in R/sec, A is a constant between 0. 5 anid 1. 0, and 6 1 is thoprompt current in amperes. The resulting value, Kp) = 1. 1 X10, is com-parable with previously ineasurd monolithic capacitors fron-mi -othuwmanufacturer. ( 1)

Two glass capacitors from Corning Glass Works wore irradiated,one of which--a conventional type with a 0. 01-pd" capacity--has beenavailable for several years. It has a temperature coefficiunt of 140 ppn•over the temperature range -55 'C to 125 'C, resulting in loss than ±LI2%capacity change over this temperature interval. It is a stable capat:itor,having a dielectric constant of about 8. 4. The second glass capacitor hasa dielectric called Glass.-K. Three formulations of this material areavailable with dielectric constants up to two orders of magnitude largerthan that of ordinary glass. This high dielectric constant is achievedthrough heat treatment of the glass, which forms mnicrocrystals thruugh-out the dielectric. As the dielectric constant is increased, the tempera-ture stability decreases, as with ceramic dielectrics. Table 2 lists themaximum capacity available with each temperature stability cinaracturisticover the ter-,.,erature range -55 'C to +125 'C. All the listed capacitor'shave the same physical size--0. 25 in. long by 0. 14 in. dianmetur.

28

'Table 2

N'axinium Temper aturcCapac itance Stability Coofficiont,

(JAI"' Characte ristic N%

0. 0 T1 +2, - 10

0. 039 U +2, -15

0. 100 V +10, -40

Sarnplhs having U and V characturistics ware ordered, but the 0. 1-j,, V-typa capacitors arrived too late fur tuetbig. Under thu manic testconditiols Ui8 thotse used for the ceramic capacitor, the pr'ompt coefficientfor thu ordinary glait4 capacitor was Kp u 4.5 x 10-7 and for thu Glass-Kcapacitur with 0. 039 'F capacity and utability characteristic U it was

p r- 2, 3 x 10-7. No delayod conductivity was obsurvud with 4-Asoc Linacpuluuu. It is possiblu that the dieluctric Lorniulatiun with characteristicV would have wi even urnallur Kpp howuver, itU wide turnpuraturcu ttabilitytolurw'co would prevent itU uue ib critical circuits in the •uau way thatcu ramic c apac ito rs aru limited,

- - -. ~r-~ , - .

6. ENERGY LQ50S OF MODERATELY FAST

ELECTRONS IN INSULATORS

6.1 INTRODTJCTION

Electron energy loss in materials is a process that has been studied

extensively; however, little attention has been given to the later stages of

the process.(12) In order to obtain better numerical estimates for the

theory of transient radiation respoinse discussed in a previous report,(l)

the slowing down of electrons whose energy is 100 eV or less is examined

in detail. It is assumed that the Bethe theory, based on electronic excita-

tions, is sufficient to giv, he range down to this energy. Thus, the mech-

anisms whereby an elec' 'on transfers energy to those excitations in a

solid which have low energy ai'c treated here,

The Harmiltonian for the undisturbed solid contains ternms which give

rise to all of the excitations that can exist in the solid. It is not necessary

to write down this I-Tamiltonian but merely to list the types of excitations

that are known to exist and their approximate or ,rgy: acoustic phonions

(0 to 0.03 eV), optical phonons (0.03 eV), dipolar relaxation (characterized

by a time rd' with 0:5 1±/r d 0.03 eV), excitons (2 to 10 eV), band-to-band

electronic transitions (2 eV and greater), and plasmons (15 eV). There

are also interactions between these excitations.

The term solid in the above context is usually taken to mean a per-

fect crystal, It is fair to say that this is because the mathematics are

thereby simplified rather than because the phenomena are different. For

example, there is probably not much difference in the range of a fast

electron in a solid metal or a liquid metal or, more appropriate for the

iroblem at hand, in quartz or glass. Thus, the term solid is extended to

mean any dense material, which, in this context, includes organic

30

polymers and liquids. Of course, the initial testing of the theory will best

be done in a well-understood material, i.e., in a perfect crystal.

A crude estimate is made of the contribution of various loss mech-

anisms for an energy low enough that band-to-band and plasmon excita-

Lions are not important, It is foutnd that the losses to excitons and to

acoutstic phonons dominate.

6.2 MATHEIMATICAL DETAILS

The method used here to treat the problem is to relate the energy

loss of the electroun to the imaginary part of the dielectric constant (which

in turn gives the energy loss of a photon propagating through the crystal).

This method has been fruitful in studying the energy lost to collective

1,lasrna oscillations. It is fitting to note, however, that this method(14)

was first used in a problem akin to the one studied here.

If an extra electron is added to the solid, there will be a coupling

between the excitations listed above and the charge, For those excitations

that involve only electrons, the coupling can be written most easily in

terms of the interaction energy of two charge densities, The charge den.

sity of the electrons in the solid, in terms of its Fourier cotefficients in

space, is -upI, The charge due to the extra electron, written in terms ofits Fourier coefficients in space and time, is -cu0ke , (C4,tvntionally,

this interaction is considered to be turned on slowly from t = -•0 by soei,I6t

factor e, , with 6'-)0. Au this is considered to be a device emplh0yd to

ensure that the correct contours will be taken when doing the Fourier'

integrals, this factor is, for Convenience, omitted. The subsc ripts oln.r

are also omitted.) The form (1 ) of the electron interaction is

4+ i i wt)

kk

Now consider1 the energy, of in~teraction with the optical j)hl•inol,

The optical phonons will, in general, be accompanled by a polarization

31

of the molecules. The strength of the mode of wave vector k is labeled

Q.(k), where j stands for both the vector component of the displacement3 .thand the type of mode, A phenomenological relation between the I

component of the macroscopic electric polarization M and the opticali

mode Q. is assumed; this relation has the form

M. = eM..Q. (2)1 13 J

The convention that repeated indices are to be summed will be followed.

It should be noted here that M has been used to denote polarization; the

symbol P, which is commonly used for polarization, will be employed for

a different quantity. The extra energy that exists when a field E ex(r)

generated by the extra electron is applied and the material has a polari-

zation M is

(r)ph I M * (r) d 3 r ( -(k) M(k) , (3)

k

The field generated by the extra charge ar must satisfy Poisson' s equation

ik' .Zex (k) = -4'r a , (4)

Further, if the retardation effects are ignored, the field is longitdcti•kal

(alonlg k), tio the solution to Eq. (4) is

ItE =i ;t14ecr a N•

The use of Eq. (5) fo Ee allowu Eq. (3) to be wil'itton in much tlluex

same foi'm as lfjq. (1). Thus, the total Interaction found so far can be

deiscribed in turmns of a more elaborate charge opri'tLtor:

1 ) - l-Mt, (6),

and thws

III -t 1 4- 1 (4 7ro 2 / I 2) 1_ 4 iwtint t k occ (7)

k

3 2

If moleculcs in the solid have a permanent dipole moment, it can be in-

cluded formally as a zero-frequency phonon mode. If this dipole moment

has two (or more) stable positions, two or more zero-frequency modes

can be used. The transitions between these modes give rise to the dipole

l e (16,17)

There is one other loss mechanism that occurs - the coupling to

acoustic phonons. It may be that this mechanism appears such a low

energy that it is not of interest--i.e., by the time an electron' s energy

drops to kT, it may be meaningless to inquire into its range frorn that

point on; even so, it should be mentioned for completeness, The deforma-

Lion potential model., which is valid for semiconductors, will be used.(18'19)

Here, the interaction is assumed to be proportional to the strain, which in

turn is the derivative of the displacement, The displacement of the

acoustic modes, which carry no polarization, is labeled-q. The interaction

energy will be

I =+ i47rZ /k2q iWt,.*0q ,r (8)ac ij Ij

k

whereuj ij is the deformation potential tension that is set up when all

electron is present. 4ij has been normalized somewhat differently than

usual so that it will frt the standard format used,

'J,'lhu coupling between the electron and the various excitation modes

of the solid have now been obtained in Eqs. (7) and (8). 'TIherre is now the

i)roblem of cal culating the energy transferred (or lost) from the electron

to the solid. The easiest way to calculate this energy loss is to calclllate

lie probability that the solid has beun excited fromn its initial state, 0, to

a state ii, Conservation of unorgy then ensures that the electron has lost

eCner'gy E - 0 tiw . That the expression for e1nergy loss is idontical,

to a certain expression involving the dielectric constant remains to b.)e

8lhowii,

3 3

The energy loss, W(kw), will be found here without formal proof.

This will be done because the answer can be essentially copied out of

Pines. (1 The difference is that where Pines treats a problem with the

single operator p k' the sum

+* +Pk + iki13jqj' (9)

is -ased here. Thus, Pine' s "golden rule" result and the generalized

interaction derived here is written as

27r /4ire 2 ) (+2W(k w) 2 2 k I(P i + ik q .Jq)n0 6(w - W1nO). (10)

t k nnn

Actually, Eq. (10) probably requires considerably more justification

than given here. In the first place, it is based on the assumption that the

electron is sufficiently localized in space that it has a unifuor, momenturn

density. In other words, we have set lol2 = 1. It is conceivable that at

the lower energies, the wave packet describing the electron will be sp)read

out so much that Ial2 must be a not-so-slowly-varying function of k. A

second point that must be checked is whether the interactions are indeed

weak enough that only the first term in the perturbation soeries can be

retained, Arisumlng this will be done in the future, Eq. (10) will bo

accepted as valid for tlhe moment,

For' convenience later, Eq, (10) is written in a slightly different,

form. Q, q, and /) aire all, boson operators, This nmuans, fi,) r ({ample,

that (q) 0 is zero uhlss the only difference between state n and Ot.te 0

is that n contains onei more acoustic phonon of type J. It 1s astuitned that

the wave functions can be written as a pr-oduct of states containing

acoustic 1)honoas (labeled II) and states containing optical pllozlnH and

electronic excitations (labeled (y) Thus, Fdq, (10) (. an bo wt\'ittlut iii two

parts:

34

W(kw) =37 (47re 22 1P kJ2 6(U) 0 0-w

+ (k q (W A (11)

wii > f30 0 - I >[3

Now, the calculation of the dielectric constant, c, can be considered.

If a polarization such as M is present, E must satisfy Poisson' s equation,

in which the source of E is all of the free charge minus the divergence

of , (20)

ik Ek -4rc [e(j) +pxt) + ik M (12)

in Eq. (12), p ext is thu external charge used to establish a field to probe

the solid, The definition of C(ko) is that 1E obeys a Poisson equiation in

which only the external charge appears; thus,

Cik I E -i 4'ru t (13)

E,.~tU.Ioos (12) and (1:3) may ue combined to yield

C" 4 v 'r 1) x t

() 1I, in to rnmlt] of thu op)urato 1 P dufined in Eq. (6),

1 (I i)L (It -- ' ) 4" 1) U xt,-; -

Thc vuluu iluauuit'ud for 1/(c wtill d( po •tt l. (1'P) hv 1) iu Hilimo•-mi'oo d

tatutL Of tho solid in• thu pvrUuU11cv of 1)~t

Again, thiLA it; exactly tliho saire u qation au 1•i'n u ubtainu fulo 1/(

'I'llti it , 1W()i])l fo) r (1 ,) 1ay 1w u pl)(ut , d to f il4,ow I ( l j'(oH • j- , I j) ' iHW M W h,1IIO~ ' IJYfI)hi

k(16)

i(,,( /• • I q "-2- I()• ),0 W W

.s2

6(

The last term in Eq. (17) represents a gain of energy by the field, i.e.,

stimulated emission. If the solid is at zero temperature, such processes

would be impossible (i.e., con0 > 0 for all n). Even at elevated tempera-

tures, the number of such processes should be small enough that they

can be ignored.

A comparison of Eq. (11) for the energy loss with Eq. (17) for

Ims(/e) allows the expression for energy loss to be written as follows:

87re 2 1

+ 27r+ (i k..\* q+) 2

/3

X MW( go W) " o(W0 + W)], (18)

According to Eq. (18), there is a term in the energy-loss expression that

is not given by the dielectric constant, This is becvuse, as stated pre-

viously, the acoustic modes are not: accompanied by a polarizatiou of any

sort, so the externally generated field is not coupled to thefso modes, If'

this is correct, it is logical to ask why the electron is coulpled to them,

The answer is that the allowed energy levels depend on the position of tho

atoms, Any change in the putitito, be it a uniform cumpto-nssion or an

acoustic wave, indteh(1s a chango in the elorgy of the cl.ctrvons, The

electrons can follow the lattice adiabatically so an electric field im never

guneratud during thie process, The proof (or disproof, as the case may, be)

of the above statements is an interesting physical problem, As mentioned

before, It iN not :clear whlth, r this proceoms affects the total range, of the

electron but it is most important at vory low energies,

6A3 FORMULA FOR LINEAR RANGE

If Eq. (18) is 1-, ctptecd for the energy loss, a large part of tie p,'ob.

1011 Is solve(d concUptually. Much detailed analytic work remdins,

36

however, before the range can be found. As a first step, let us find the

linear range. This is given, in the continuous method, by the expression(21)

E0R = I9a)"J -dE/dx(Ef

In Eq. (19a), E 0 is the initial energy and Ef is the final energy. The

energy loss per unit range, dE/dx, is found from the energy loss per unit

time, dE/dt, by

dE =1 dE (I9b)dx v dt

where v is the velocity, Now W is the probability of losing energy t(,') and

mnomentum Ilk per unit time per unit volume in k space; thus the expecta.

tion (mean value) of the energy loss per unit time is

"" (2r) j d3 k 'h W(kI ), (20)

The perturbation theory by which W(kw) was derived must con.

serve energy and momentum (even though this was not explicitly

indicatd), Thus, if Pil P r are the initial and final electron momentum

fol. a, given collision,

1'3 lf + Ihk (ZI la)

1M 2M 1 f21 p. P1 +•'w (•.1 b)

U I--W--" (t'1~ (Zic)t~t- P~

We as sumu hk X Pi (which may not be true when the electron onorgy is

lo st than t ijVlb t 11 UV), so t:010 last turin' in Eq. (21 c) cuan be droppedd, Thiii,

th, hsum ovor cw van bu rup)laced by iai integral timnes a delta ft'•ution:

dit 1] P 0k kd Jr It -( - k -v) Im 'C- (22)

We iiow i so cy lindrical cooirdinat-as for k, with the axim along v, and Into -

Sr"Itv ov .Ir cIkil to rem( ove tle del-ta ftictioun, We also awistme c to be

37

independent of k. All of these steps reduce Eq. (22) to

dE 2e 2 1- kL dkj.= - dw Irnr (2:3)

vdt 7r--" k2 + (w/v)2

The conservation of momentum has not been completely specified

by just ensuring that Eq. (21c) is satisfied, If we look at the following

scatte'i'ig diagram

MAXIMUM VALUE OF411I

MAXIMUM VALUE OF AA,

%vw tcie that requiiring hkil to b• smnall altou fui'com k. to be uiall. ',hu si

w% take

Whe•ou Vy i s•me 111.1uber of uordo r unity, Tlihe intgiz'a1l of P"q. (23)) ovoi It

will thus give a numik r' ic al Lactor of 1/ 2 logM (Y4 +I 1), wvhich io I loucoft) I-th

druppod,

F'iaally, wo i,'u miak that the e0lectron can U Cai'v(ly lo9 muv• 0'eo gy

il 11 collision thall it IwoL initially; i'v, therO P 1 all Ul)jW bomild )11 (0,

1,q*tiutiozi (23) hal now beou red' ced to the oimple forim

IE 2ttI_ -1 (/ -

0

38,

6.4 NUMERICAL ESTIMATES OF RANGE

The job now (apart from justifying a great number of assumptions I

is to evaluate the integral in Eq, (25) for materials of interest, which has

not actually been done yet. It has, however, proven to be utseful in evalluat-

ing the order of magnitude of dE/dt as it; is determined by the various

excitations that occur, As an example, suppose the dielectric constant is

composed of the sum of three terms

C + C d + Cph 4-, Cex (Z6)

whie re

* l T dipole tfrm, (27a)

ph .. 2 optical pholnoii, (271))

w /w/ p) iw 1)

(w x 0 x

1% I' VO flhl)l•llte ~n ~ A bilnidto-biaind turi n aid a pl&llli tpa•io n mihould also

,IM iliCluldhd for the tente rgy louiesi below 5 to 2,5 oV( but move tlnti'

would be rqulyiied to inlcludv them,

Th' oi'h i veo iicilitriulintlu llto C ph iiud C 0 htve1t it r'kpid varialtion

only in tho i'coioui uf thoir i'-oiounat freqtutncy, We tian the rel-i'o mauke

the apipl-~ii, a ion

C1 (1 f lW07' d)

S . ... h-• W (2 d )

I(w A0 1) +1orO1 - (1 /l4. p) 1 pi t w (28)4

39

c 3 1 - / (o/A) ex)2 + iwrex)[I - , 2 + [Wr , 1]2 Wex (28c)

In Eqs, (28a) and (28b),

id = I + c2 + c3 (29a)

p Ph = 3 (29b)

At this stage we are interested in a rough answer (the input c is

only a guess based on a plauMible model), so we shial estiniato the

integrals in a crude mannor, For o n 1oph,

-I I W (I + Wrd) -rwlmrnc ct 1•-...... -- I for Wo • 7d 3

1 + (rT d d (

We assume rd 1X W ph, vo the large constant region of the intorMlLud willd

dominate the integral,

The intagrand in tho region of a resonanuce itco m -1 _ . .___.__,_._ 31

(I - (w/o 0 ) 1 + (wr

This is a peak with n haight of I/r and a width of about co0z , ThMS, thUUU

int•grali are indupendont of tho damping and doplnd only on thu umuergy of

the romonakive

We can now collect thu above estimate to find the onergy low0,1 pur

unit di •itca& , as given by Eq, (2,S), to be

40

d) 2 r d2

1 ~ -( tE) ( Ph 2(~uh 11W, E -ii~w ex (32)

2 2 2xe

In2 E q, (3 2),D 0A( in thu 13oliv radiusj,

The vil' iouli (Ouilgch 11go 1-v fni, unuiogh rumovu d that the highu mt to rm

appenvi'ng in E~q. (32) will dominiate, Thim boing tho case, we call 11111kc thu-

s implh' phy 8ictil inte rp aO tatioll 0111t, thO I-1-ang(I fol 010h VaI'IHMIA 1'OCOiN HO ii

will jo nt acdd, Thi't its, thu P0 will bo Ix ('01-taml val11gO1 Itu , while the

ouc(Ltronl ii losin~g energ d''cown Lu U,~ 11110010hu' PL111Ig WhilO it 108ON0$ uno gy

down to E P Oel 1hu

d tph OX

1 (it71 (L c ) W 4 2 11 ) (34u)

itp a0 Ir L2) T)Iph I(i/10 P"p) 1) (3 4b)

Rd ~d (I pht C, ( Td1 0/1)J 3c

Fo, a'exatimplu, taippume un ilImulatuto hu a n excituni biwcli at 3 CV,

optl 'iC1 n aMIL lo i t .3 ý 10' UV, luld dipuluv v i'uixatlon tiio Um oi ~iow i0t I IV,'

(b' / r (I 1k I) o~ V'), mid wo inqiav i ~nto tho rruige jof0 3m it I5vV olot Pu)).

From Fq(3411), thii cle ctrotiwll \61 ft'avl abotit 25~ whilco lo m1 nM onerg

downi to .3 vV'V ;It. will 4then tr avel of thu ortde r of 10O A (ama feti11i ng T phthtl a'inaitu wvil Io 10 if F a ph ill am~ ii-g fAm 3) while mloilWIU clo)wn to

x 10 ujV, iandt, 1,i1111ly, it will t.raivel about i x' 10' as nit slow MIO owln

-3 -ato, say, 3 X 10 eV. (This last range assumes c S 10 ; an it should be

if the material is a good dielectric,)

As a final estimate, the energy loss to acoustic phonons is examined,

which, again, is done with the aid of a simple model rather than relying

too strongly on first principles. The last term in Eq. (18), which gives

the energy lost to acoustic phonons, involves exactly the same matrix

elements that are needed to find the relaxation time that determines the

mobility, (Already another simplification has been made. It is assumed

the electrons do not couple to the optical modes so well that they are

w-companied by a virtual polarization cloud- i.e,, the material could be

silicon ,, probably Mylar, but not magnesium oxide.)

These matrix elements do not depend on the energy of the electron.(2)

The energy loss per unit time given by Eq. (20) is thuis appr'oxinjatly

equal to

dE -3

whore A( w) is the volume of k spa.co allowad in the scatttrHng event that

creates a phinon of ene rgy hw and W(k,w) has boen vot; equal to thi con.

stant W,

A liagrim of the initial and fitAal momentum is slhwn in Fig, t, 1.

The energy and momuntum conuetrvation rules thlt mriust accompaily

Fig, 6,1 are, for omall WP,

M

and

1p (k + g) ,I(~1

whore g in a r scip.ýocal lattice vectov, This vector must be added to k to

include Umklapp processes which can occur in solids. Since onergie" am

low as 5 eV are of concern here, the occurrence of ouch proceumve io

important since only if Urnklapp scatterings occur can the entire ci 'cl.

shown in Fig. 6.1 bu reached by P

42

P1 gPo

Pig g,0iv mýlwlu \( ctuigovnmta edt

ti(ýRttt1'1 *Di11w 1Wir mhuiliM t\V( wi&tIA)lt U111 1daEIL pontivL u.i tAP, kriiual viiht o rau oll th I~S u Ik-ai nI 1 g1 , i 1w I tt Is aU fe por.l

43

The actual value of A(k) will be rather complicated, but the order

of magnitude in clearly iseen to be

A(k) ~44irP(AP -P 1v ~4 rM Mtow(8

Thus~,

dE (Zr2-Zwmý t)

~(2ir ) WP M (kO0) (39)

Lin E Cj. (39), tile N ur )Ve P, tho, pholun Vile 1-gie hia, 11 been replaced by tho

ave rago phionon vite rgy Thi tiquantity, Ill. Wella, ham bee cii rplaced by thlt

Dabyv to lp 1, Atli PC 0.

Frini Eq. (39 ), the onte riiy Io#N ])VP wilt di NWhWCO IN A (1oiii1itait;I1

indlepenI-dvit ot the, electr1onl oilcr~y. 'hig caun bo lnureapi'utod III te -I'III of

a Vory vnI)II1pQ Iiiodu1l ItInel0y, two,4 thlt oati-toii hitu a con Nttwit iiiotian rvee

pat, ~,With

And that tit Oealh Nutitte d1 iM It; 11) Nr4 Al an a~kiloit. (A' OciteM 10,

Theli likuloiini til Vdalue o, x cnn be uNtfi~lated Iroin the mo(bilitY p.

J'"o P a Wood ci ikNittat ( P, a Ilo hi lit y of aboult 30 ei' a N CC V INi N Niat)1 'Phd i"I

CuilUNP P OIldw to u P Lilaxat bit tiat iv of

-'s uV 2.7 100 NL)U

Thu ye lovity, of tin u le tnnio with 0cite rgy W1 PiN a 1it tAW Illo Puti 0111 (0oil

VIII NeC a 0O

h 10 cm 7d ~

Tihu lilut IIICLLI P '4.1im

44

if 114 0 : 3 L-V and E 0.03 eV, then

it IV 2 1 (44)

which is About thr~ sarine as that found Lfor optical phonIons,

T~he linear ranlge, of course, greatly overeotinuratow thu tnt-):A projected

PRIAI0 hI, travervsed by the electron, Fortuniately, thin p rojec ted I'ange

CAnl he utttily e gtirii-t0ed for energ~y loss to phIonunfs According to the

niodel tAlready nieitioiied, the electron loses the moat-no tamount of energy,

k 0, which is typically 0.03 tuV, At oach collision, and thuso it Must wuffer

N collisions, where

N _tE0/kO0 (45)

is tkbout 100 for all E% 0f 3 uV, 0vroii ottndard rtundorn-wulk theoory, onec

knows that the pi-oJected viunge is sirnply

1hi~c th(I loss to op~tical pholtfoliS is about the mameiv hut perha)lpN

somelwhat largerP, the pi-uJated range duec to both procosmmci wilt ba about

one -half of the value given in I~14j (46).

6Ab C~ON( Al 8 IONS

It mhoutld bo runiamnburud that only a rouigh wthiamto of tho range hlits

boun niade hvire I hvon if the appeuxiimaticons madea inl tho derivations of

tho finlal fo r illlas al.e valid, thu0 ma1tevial plir ailute v umo d to ffind nm

Vill valutis dre wily a guessi bused Onl typical valuesl 0ecounite rod for. well.-

itudliud mate vialm. Nttiurmb V pecific to a p~ractical insmulatol m uch as

Ivylu. Iavo RIot Iuliwaii

It' thiese v)uflbe I' du not pirove to be too widu~ of the niark, h()wIIVor,

then it. is foutud thut the clectr-on will mlow downi very rapidly (a linelar

rvalnfc of 41 A) fromr 14 vv to 3 UY am it loma No I enry to ectioltC)II and the ii

it will lose une vgy about uquamlly to uptical mind acoustical phuonws, Thu

proje~( ted vange is cutiniated to bo about 10 3 k Its thle clectronl dlwos downi

fromn 3 vV to the rynai eneorgies,

45

7, CONCLUSIONS

The ton implantation 1.rcep has been successfully t:itt 'dv( to appl~ymetalick cantactm to igamiconductors and insulators. The trant~iontresponge of a Mylar capacitor with ion-implanted czontacts dimplayedreduced polarization offaets but nio mignificant dflcroUan in rc~pofnt1 Mhitch

would make Itm ume destrablco in radiation onvironinentm, Theu ton-implanted olactroden elxhibited better carrier lnjc'ctlon than foil oloctro(IUNand allowed ohmici contactm to be applied to Pjamiconductox-N with a w~dorange of renintivity I consequently, the procesm han advitntagom in rometti'ch.Work where n implo r niothodN fail to pro vide good contactN, The produc-tion uae of the procami for caIpaciturs tit doubtful owing to th. lengthycloadnlbg procodura awnd the difficulty Ui' achieving a Untifornit plaiayla Withthe asmuciated equiplilant nu1co~tit~ry fur imlplaintationl onl a 0oi~ftbfluouk fiilm,

The traulimont rmdiationl romponoo exhibittud by a Mylar capacitorvurioui tonmpu riture is cofnsikntft with the res)oflNe prodictud by a bandmo1dol of the conductionl procepi ill which traps are unlifOrml-1y diN tribUtUdin the Lo rbiddtin vmnu eAnd uoiN ii~g of carrie r itN imnu-nbilized. Undo r Nuchconditiofnsp the photuctirrunt iw extreumoly Maldli Compared to that dimplay -

ed w iUlout t rapphig and it Wo &k (topondoico uto the lphotoPur ront onl tonipo Puattiv'o I ti prod Ic i .ict , TliiU obH 1' 1'V Illd o (1C y 1i~ ,11110 ('1 tat ct Ctnl 1)o littlly Or dtlvf4of illaugitude loiigur tb an tho vo ritge car rio lifetinio and I imNU hiulitivo toIto iIpPI' at0 1 ,

Now mu t~hodi of fabric ating caurainly and glaN N cLpac ito ro wttit hi~ghdlioloutric cunnftwtH ldN ve y~oldlud likrg( cupacity-to- volwum ratitom and

c urit uquily th prl-Ol)ipt tdIkidto~l l. N wp(lNe U4 WINalll. Tho uN u of th~oN 0

U aa~ toI'N i t ill lulltejd, IIOWeVer,' N inc-o the tompo rathure CUUMlCAMANt

of1 capacL(ity tue latruti land thiN preo vntw tho Li~ ut ofmuch cttmc itori itIcr1'itic al circuit ajpjliicat buNl who no the autcIlVity IIIutit: unoialn c olnw(ti~i A fr

p)ropor j. iric tt Ame lion N ovo it l'argea' tompo~ratu ni ranig u.

I''xa1tIl hul. i.on of thu 0110 rgy lorid i of IlUlodorato ly flwti . olloL' . I(lw HLit

in11 Ulato rti, which iti nuc-clNu~ary for thu theoury of tit~iNlouat radiutionl offel IN

tit IliNuidtorN * () hicdicatos t~hat ono rgy liUNN to excitont im the pl'illal'ylpro1otiti that occurm tin tho onorgy range from F) oX" to 3 UV. 4linorgy lolitito opticadl a~nd acous tica] 131,111011 t~hui occurm (town to thormal ano rgiu il.

PrOViOeIN work undoer th lu contract haw mhown that anom-aloum.ly largo'tVWimml~oit ufftoctN mPeiNul'oC tit 01a-111,1VdURod mica crpiwp~ito VN could bo

46

as4c ribed to leakage currantit thrtough air trapped iflm4idt! the ocapac)i(ito r(11, (, tiu. II~ j,,t-oj(4) Coinpartsoon of the tl'aflB lant 1ue1)ofln of a comrncilrc iali-iica capacitor with that of a Nfinillar capacitor that had ai coiidtctivocorAinig Oil the 01nCIIp)I4latioii do moiitrated that cluct~ric chiarg~e frotiloniizod air cani Ib trappo ~o4i the itioulator surface, leadiuig to incvroasediniitial traimisoit siiam Irradiationi of movoral capacitors u1nitgdiffaronit dielectric materi~als Iii a TRIGA reac tor indicated that noautronlresponse wtan least saisitive bi caranic. capacitors followed by ilica,tanltalum oxide, antid Mylar dielectric i. (2 3 ) A me~thod for varlficuatioun ofa model fof i~outro11 irradiation of dilelctrics by proton irradiationi wat4do nertbad, t'A capacitor oxc~hanigt program ill which the mama capacitornweare raoaurad by GOnolral Atomnic and bito riational 3unfionaiw Machinoptfiditcatud that measuremontit performed with anl 1G eircuit in which chargetrwisifol wan measured yielded comparable roalt) to I'l allu rei alit ofradiation" induced current with current tprobon.

47

I. DaMichlal, D. W, , D. K. Nichols, &nid V. A. 5. van Lint, RadlinitonIrfet nDeecrcMtran Quarterly Report No, 3, UI. S. AvmyElectronics Command, Toehnicad Report IflOM -01412-3, July 19606

Quarterlly Raeport No. 4, Ui. S. Armiy IEluctronics Commandl~,Techlnical Report; EClOM-Ol 41I.bA4 October )966,

3. van Lint, V. A. 5., 13. E,. Kott, and D, E'. W~illis. 'l'-iiitLlýlttiýiEffecto on-- .~lacti'Punic Partn' Report. No, 8, EighthII Qua rh.' ly F4rogMr1'sm Report, Contract DA3 6-039SC-891 96, Omiaral Aotomic Dlvi sioh-,Clne val Dynaiic m Corpjoration, Report GIA. UH11 Smic 1 964.

4. Ovat-nyer, It. F'.0 D. K(. Nicholm, and1 V. A.. J, van Lint., kiLct1itLiol1Xffect8 oti Dieivec tvic NMv Ha rls, (Ittarte lily Re po rvt No. 1, U, 8, Army1Eluetionicio Comm'an~d, Tachnical Report EfCOM -01412-1,December 9b

41. attox, D. M., 'l11irlm Dupowitioll UtiffiR Ac colurutud Ions,'' 8tidhiCo vImpv i'ti ,on , Ropr id HC W-44~-6-4. 1,44, Novninbe r 1 964,

6. ..... I____ 'Motalliving (Zurkalilc Uingu a Gam Divuhadfgo' ,' 8111(111Co rp1)0 a.tion Rup i'lid S1-Ru 64- 13 30, 8upteitnbor 19A.4

7. ý A0 1inte t'facu Forimation and the Adhoio §ofl Doilmep 4i~tuThill Fi'lmsn Samidit Cm-o Pv0atlon, Ivofloj4 P aph H (- 8~jnumavy 196½..

8. 'LDom1gn consvidev~ationsm fur, ton1 Plaxting",' 81undiltCo ~pu0 atiun, Repr'int SC-H-6-99?, January 1966.

9.- Rtumv, A., RGA IRoviv , Vol. 12, Nu. 3, 1951, 1p. 362.

10.. van Lint, V. A. J., E. , CLWilmer, R. Don~on, 11, lioriyu, 1-1 K. Liuntz,D. X(. Nichols, ~tud It, A. Pull, Tjvunviunt Radiatdion 1,ffoct, F~inalReport-, C'ontract DA-49 -186-AMC-65(X), Goncral Atornic Divibion,Gencr~al Dynamrics Curporatiun, Report CIA-6248, March 22, 196ý'..

48

ii. 1, van Lint, V'. A. J. R. F. Ovv rrnyo r, tald 1). K'. Nich ols * 1''ransiolntRadCiatlion E~ffects onl Liluctronlic Parti~ , F'tnal Ratport, Con1t.ract. )A3 6-039SC -89196, Gcoiiwral Atomic DiviM Ion, Gunn ral flyncuic r4 C01111 I'llt~ioi, Re'porti GA-05~34, July 10, 19 65.

1 ~. Sturnbo,1111t 1.' , 11. M. I In MathodNm of Expc rivimotql I-mxyn4ic.1, Vol. 5i,

Part. A., Ch uIcn-Shbi[ng Wui Mnd E. C. L. W nan, udiN. , Acadounik Promm ,Nvw York, 1961, pp. 1-76,

13. 1'rublllchi, 1-. , and Hf. Pu zo r, Proc. Ph yt. Soc. (London), V ol, A.68,

14, ti_______ ,ad H(. L, Platzman, "E1nargy Losm of Moving Elvlotronnto Dipolar IAulaxt)ttion, 1' Phyr1 f. R'ov, I Vol. 92~,1 PM3, p. IM".

15., Pine m , D., Elaion1Cft~ary Exc itaticoni In So011dm W, A. hi~on mitn, IJIC,

Now York, 19 3, lip.W ff.

16, Dvb~y u, P. ,Pol~r 'M oloc ultim, Dovo r Public ationM * 19 9(), sac, ,9

17, K~it~tocl, C. Int roduc tion to Solid 8t-Ato Physiem, ad vd, I John wii iy

ev soolm hw In, Now York,1 195K~3

1900 Slocm (lQ , 6, ý1 9),

191. PI~ij!(1 of ho 'Phtoory or'Sl~N Cumbridgv Univor1'

fjkty~~ ~ ~ 6-raV t 94 o.(. 13,

4) 01 FO yiwi1an, A~. P. 1 etd, 4, 1'.I'll e Vlynm4l"n 1htoturto s oilP 1-yN ic , Vol. 11,

Addlioun- Wwo~ly i-re.4w, 1964, Soc. 10, 4.

z I1I Nicholti , 1).) K. , un V. A, J'. vim Lint# 80lid 8tot~ti t1I Yý CN , Vol. Ili,

2 3. Ovorrnyor', 11, F. , 1). K. N ichol~ , nn.I V. A. J. van Lint, 1RndiatimiE ffoctti onl Diluct~ri~c Matortial, Quarto rly lRoport. Nu. 2, U,. S. A rmnyEloct ronlcia Cormmand, Tochulicw~. Ru port EUCOM -014 12-21 Ma rcb 1966.

4 C)

tavaurhil' ,e ru iw, ea mirua, Seat or me e acr rasetsel an Indo~mmifl d"018110" #"Us# he bnlastei wftn the overel1 to#~I mit OIs 01841118)IO- ORMINA TIN 0 AC~tIVITY (Coitiotaioauthor) I&. ftfONT? 41%BRIuTtY tfl ýAIGIICgtgcAON

0'ncr'ril Atoniic Divin ion Unc 1&1mifit't1Genera ~l Dynaieimlc Corpora~tion III GROUP

RADIATION WF1IC T:J ON DIELECTR IC MATME LALS

* P liA I Ike)()V.t(I1 Ju-l _ 1966 _thrqou h 30 Septembrc 1- 1966)A. AUrHONO~) (Lost name. fiftl inametnis.119)

3. F1, Colwell, D, W, De Michele, R, F, vomyi

6, 111POW'r-r-1p. TOTA6 No NOF PACSK 76, NO, OF 04110

4o ONfRAd* Wst INANT NO. I&. ORISINATOR1S RKPORT HUMBIR(S)

DAZ8 -O43 1%MC -0 141 Z(Ir) OA-7474'A C odo

4ý9 0, 2 , 830, 400 6. ATm~m rPom we( ) (Atilo othati~umbsts fhatmiay be aseil"Od

10. A VA I~t ANIL ITY 'LIMITA ?ION NOTICIA

Dist ribution of thIN documaci t ill untinlititd

11B111101611MINTARN6711 7 * hliruIoL1liti h W4~j 1 1,I 1110NICAING MILITARY ACT IvIyY

mpoaisortid by tho Dtifi Lu14 At rn ic tJ,Si Armi~y Eictrut nIle'm Colnit-ml~nd

Supp1ort Agii IW y tinido r NWER~1 Subtdoik 16 , 009 1 Feo rt Mon.mouuth, N. J. 07703

II A111TRACT/

Aý ciptai4lIii(I (t c ('I' ti on of thu t tichnitim~ is wa t111iqip4 Jt-nen lt milulpl oyod ill ( illlull wim)laikilttitin pl'ocom MN lI PrONtlitod. 1HuNVIt of higih uiizW rgy 010111-01n1 il-sldUA-itLou of' it My4%I, r vsiitol.o With ionuiulmplauitt'd (Ii hluthruu Ill-() conliparmnl to thilI'VMIIOli1hi O)f A~ CUuIt 1-01 NtfllPl The Lol-imlahInlileid oouctriuctinti xhibit litttor al rir Is

uu I tLuni th i~l coiw ulit uiui Il fol i li (1wi-rOtlO N, T' I-I Nu iii V~dd t~il O0 i Ifu I d itsi f i in11

i' r lidtt 111lo i of it Iloiii010it-hit! vl-tinii' ~ iVc apJa citr UI'ii d (tW.,) A iN NA u apacit.l- a1rNtl i

lIncludod. Onuol Ofu 01 It-INl c t-plACUI icr O~lItti iud t-i 111011miC a'y~tt-li itt (1iiiC-1 h~ uwithlvtl vy 11 ighd ti ( I lilctp IC (' v0 list alit tThu tra' i iulditin rai'tid-t onl u froAtN ill My hiar ci' N UN~

tIi 1ljW'tl0ýjJ) VIU W fi ~~~ 41 ~ to it eLbuil d modu f~i or tilt! ilc LW~dltOu 4OU(111 ill W1N L 10 It h Ut 1tilluouli diNt -1ibi't lull1 of caiip1 1,i r I ntN lN luclititu ill tho fUu'biddc ii 40110 411d o110 NMipti carm ioi 4 '& 1IIrinimoi 11izod, Ail examina~ llt~ion of div (1110 P y losINNP g)( I-ov N tim (if

uIiodct' i'atciy fam u lt. icti-i ilOl i jInN ulsutUi' N Iili(1211 thN1141 fi-0uii 15~ 0VY t0 3 (IV 010cllittrgy IONN1 iN JiaHIl41v1-l~y Io OXsc toiiii whil i'iltN u inii ry IuNN to optica il 1 IcitmlMI'itJIl1-41011011 01 0CMU1fl Clown~l to 1111-114 0id 0lhuMiOiN,

DD I JAN 4 1473_ _ _

Socurity Cluuuification

- zwuurily 11101111 lulli____ KY JneMLINK LINK 11 LINK C

Radfiltm lol flth''t.M clidlio 1cwty1r

Siirldiit. iou IY0H' It cl p A Pc itoiNs

toll 11- l a it 'It lol

Fi liAct. 1)31l tvlw('g y lorm

INTRNUCTIONN1, 01t11ItlNA'TINO AC'TIVIrllt Millet the name and address imp slid by security viaasefielsialn, usingl standard statementsof the olonitavltr, sub'onitraeltr, gratee80 Department of Do. flOO lot5tenlse avetivity or olothermeatlinaln Numerioal* author) issuing (1) "Qualified re14U011t111`1 lay Obtain POPics111 Of thiNthe reports report tiom DWI.2a. HItACONT NKCIiUJITY CLANNIVICATIONt Knite the over. (2 "'Forelo anuneement and dietise innat on of thipall fifeurity (aeles inof the reports Indloste whethersptyDO INntohriil"114sotrieled Dasta" Iis Ineluded, Markliv Is to be In aeornreotbyDCinnt uhrieLant',i Wilith agatpruptate feourity fegtdetioejns (3) "OU, 1. Clovernment agltenlo"lemay obltain Corteso of

211,OROUI Aunitolledownradng i siolfid inDo)Di-thitl report dljetetly (ram DDC, Other (4takl lod DDCtIh tive SOU . oi Au iti vil F ovlo downrding ri al M anuel , Kinlh erO users #hall requ elt through othe Nfoup n umber, A Iwo, wiwn spl~ithmleh, shlow that optional - AAImarokingsk hafve teen1 %sloo for (Irecup 3 and (Iroup 4 a@ author- (4) flu, 0 military ageno0ili may obtainl epoiue Of thislteedl, report diroutly fronm DDC1 Other qualifliedl users3, I4KION'T TIL'i'L~ Killer ithe uomplete report title in all 411ll1 request. throughvapitaol letiters Titles In All ititoes should lie unuiessilied,It a ceceactillN61tt title v'aoluot be seleuted without (1ls4si1ie6a.Iloino lihow title voIsas ieliost ion in ll vepit lail In parnthOai a (4) ' 'All disitibution tof this report Iis tontrolloitL Quid.it11101oooditl y foll owing till title, tied DDC users *altal request through4. I) KNt('IOT#IVle!, NOT0'l'1 If apsproprtiate, efnite the tyef (if -.11or1t, i 01 1o erim,1111 Iprogres, su~lmmery, annual, icr final, If the reolio has beenlvl shodliif tW the Oflitip of To'l'hnitIa0hYo t hto f il ita14y iv a eI0 when a cipouif (Ii replortilogt periud Is gerviesa, Department (of Commercle, ((r vasle to the# publelt, mndii

l'ovotodOat. this faot and 01116f the lrinu, if known,No AUtIION(H)I 1iliotr thw name(s) of author(s) as ahown onl 1i, 6UI'PLEMKNTANY NQTKRi Use for additional omploina'(it to t he' report, 1(otp loo st ooame, Nor t names mid~diale inital tory n0otesIf 111lit ar y, ithow rank ontd bronlih of( seviu, Ill ite nme ofthe4 prIcet1114 "Ii ll ~tor lte an absolulte mInhimum requirements 1 2, VIPONNOHINfl MILITAHY ACTIIVITi YItaiot e the namel ofit, 14KI'0RTl DAIA'l' Kniotr the date olt the rPolort as dlay, the dearlirlental prooi:l offittue or laborslot,~ aplloorlog (pay.

y"Ai lt i (111, por 1 moe ten"li dop ppoa inj o) h resvearth arddevelopments onrlde address,monceth, y'rool, ors flat# yefr ItIII mIre ltha loll 1a3aers AIIjUTHACTI Kilter an oabsrailt tviviiii a tritef andi fautual

summary of the dounmenti indiunt lye of thefi report, i van thouugh'1 a 'l1Vl' Al. NUMII K,14 010 l'AOR~ 1I'l Th total lease Count It may Olso appear elsew110ret In timhe bdy cflice fle edcnius I ro.shouli ~dFollow 11oumal4 Ila Inst io prolloimdcc, en, iCe, enter the 1 ertl.f additionsl Ipu lot0 i rotitcilroil, as vont incailon shoa t sheoll'cececcdce'r (if palges uonotin og intormat lonuI) 46100l.e70j. NtJMIW KN 010 K KI" KNICNC~ K niKute t14e total numtcer tit It ts l1t1g1ly de t1rebletst the0 il atrai shlofl ulasce head replortsflu br umni a 110t ll w the orii, lie unolasml l ted, Rouhl jnagra ph(i oll th aboleat ray lsall 0114d withNos, CONTR14ACT ON O A NT NUM1111l4i ifotiapropriatil, ente oi andinduct on ot( the mil itary useoucity v Isasifica tion uflhi't In.010e spgi ivi abe numlbeir (if ho1 ic otraulit $?kiral undeor wiltith formation In this pars graplh, repro wanted AN ('1N). (N). (0'), fit (1))Ill vlsie Weu ONtwa wrttton TI lhorill Is no limita1tioin L11c 11e0 longil ut otm the nh otr Ilow.8t), *,', & Iell t'NOJRCT NUMflKu Kieter thle appropriolof ever, this sulggested lsength li frOm 1I0 to) 2211 woridsceellitar departilent idenoitiu ation, suob as project nlumber, 4KYWO)H KywrdaetehosIlymaigutrm

sul~o~e't umbr, ystm nmbes, asknumerdoor nicort phra seal that cohavloterive a repoirt ando nesy he usedl ofsVa. ORICHGNAT0I4't REP~lORT NUMBERKI4( Kilter the )((I- litle% entnive' (or ustaolgsinlg thu rep)ort Keoy words mtost I)evicat ioptrt nunehar by wlcivi thle diouuoni will be idenet ifde d opoi'tud so that not oomirity vuoln l (hat ion Iis required,. lidenti.sodll rid I led Iby thel original hag acitivit y, Tl'Ii number mlust (ier. Nueth as0 ejuipment 11odel0 desI oat tion, trw de fowene, military

)0unique to ilisthitieprt. pirtjuiotuods name, eotgraphi tivbatir on, may ho usved ofa key01. OTI 9W N'PIORT NU M11I KWR) If the repoirt haes tbeeo words but will be, to llwow by ant Indiumtilon of tavihnie Iol von.

a aw Mold ccyot hr rpor oualces (~thr h~ ite'at, Tlis somIignomeot of litkil, rules, and willtght al is )pt Ionalof Ia)) (if" ýqitfloaiwr), Asol) cnot r tiis number(s),to. AVAIL, AIIIIJl Y/I.IMI'lATION NOTICIM~ Enter any lim.ltttiaouns on fuccthcor diiiweicninitticon of thei report, other than those

(11i0 111.101 Bil

Security Clcesslllication