UNCLASSIFIED

AD 248 535

ARMED SERVICES TECHNICAL INR)MATION WYARLINGTON HALL STATIONARLINGTON 12, VIRGINIA

0Zmoro04 0 06UNCLASSIFIED

Best Available Copy

NOTICX: When govermnt or other davings, ELpci-ficatioas or other data are used tor any purposeother than in connection vitt a lefinitely relatedgoverment procurement opcration, the U. S.Oomant thereby incurs no responsibility, nor anyob2lIBtion whatsoever; and the fact that the (overn-ment &Ay have formalated, furnished, or in any wysup.lied the maid d awings, specifications, or otherdata Is not to be regarded by implication or other-visa as in any anner licensing the bolder or anyother peromo or corporation, or ctnveying any rightsor permission to mnufacture, use or sell anypatented invention that my in any wy be relatedthereto.

C ,

9.' ' '%.

DC-T--60236OCTOSER 1960

METAL-TO-CERAMIC SEALTECHNOLOGY STUDY

FINAL TECHNICAL REPORT

Covering the Period

22 Jun* 1959 to 22 September 1960

ELECTRONIC TUBE DIVISION

SPERRY GYROSCOPE COMPANYDIVISION OF SPERRY RAND CORPORATION

GREAT NECK, NEW YORK

SPERRY REPORT NO. NA-8240-4216

CONTRACT NO. AF 30(60212047

PRPARED FOR

ROME AIR DEVELOPMENT CENTERAIR RESEARCH AND DEVELOPMENT COMMAND

UNITED STATES AIR FORCE

GRIFFISS AIR FORCE BASENEW YORk

PATENT NOTICE: When Government drawings, specifications, orother data are used for any purpose other then in connectionwAjh a definitely related Government procurement operationthe United States Government thereby incurs no responsibilitiesnor any obligation whatsoever and the fact that the Governmentmay have formulated, furnished, or in any way supplied the saiddrawings, specifications or other date is not to he regarded byimplication or otherwi3e as in any manner licensing the holderor any other person or corporation, or conveying any rights orpermission to manufacture, use, or sell any patented inventionthat may in any way be related thereto.

Qualified requestors may obtain copies of this report from the ASTIADocument Service Center, Arlington Hall Station, Arlington 12. Virginia.ASTIA Services for the Department of Defense :.,ntractors are availableth-ough the "Field of Interest Register" on a need-to-know" certifiedby the cosnizant militazy agency of their projc . or contraLt.

RtADC-Tt-.60-236 COPY No. 4 LOCTooIW 190o

METAL-TO-CERAMIC SEALTECHNOLOGY STUDY

FINAL TECI INICAL REPORT

22 Jun. 1959 to 22 Soeomba 1960

ELECTONIC TUBE DIVISIONSPERRY GYROSCOPE COMPANYKYWON OF SPUE RAWCWAIO

GRtEAT NECK, NEW YORK

590MY AFFMR NO. NA-4240-4216 S. S. Ce". k.

COWTACf NO. AF 301621204 J. .IMPs~

PSOJUC 140. 450W TASK NO. 45152 K. ft. Ieyaw, hf.

L Chop*"e

PmuAWu roROME AIR DEVELOPMENT CENTER

AIR RESEARCH AND DEVELOPMENT COMMANDUNITED STATES ANR FORCE.GRIFFISS AIR FOPCE BASE

NEW YORK

1a

TABLE OF CONTENT.

Paragraph Page

SECTION A - INTRODUCTION

1 Purpose of Program 12 Phases of ProGram 2

a. Phase I - Literature Survey andAnalysis 2

b. Phase II - Metallizing Investigation 3c. Phase III- Brazing Investigation 3

* d. Phase IV - Tasting Investigation 3e. Phase V - Temperature Investigation 3

* f. Phase VI - Stress Investigation 4

SECTION B - DISCUSSION

3 Phase I - Literature Survoy and Analysisa. Introduction 5b. Alumina Reactio,. .heory of Adherence 6

c. Glass Phase Migration Theory of Adherence 9d. Other Factors Affecting Ad' ence 10e. Related Systems and Porcelain-

Enamel Reactions 114 Phase II - Metallizing Inveztigat - 12

a. Investigating Design 12

(1) Alumina Reaction Theory 13'2) Molybdenum Oxide Mechanism 15(3) Glass'Migration Theory 15(4) Molybdenum Sintering Mechanism 16

(5) Glass AddItive Mechanism 17

.(6) Other Mechanisms 17

i

TABLE OF CONTENTS (Cont)

Paragraph Page

4 b. Evaluation of Metallizing

(Cont) Compositions 17

(1) Materials and Methods 17

(2) Adherence Test 19

(3) Torque P.e Test 20

(4) Torqae Peel Test Results 21

(5) Compression Test 24

(6) Compression Test Results 25

(7) Tensile Test 28(8) Tensile Test Results 29

(9) Final Observations 315 Phase - - Brazing Investigation 32

6 Phase IV - Testing Investigation 32

a. Comparison of Test Methods 32

b. Tensile Test Specimen Design 35

C. Photoelastic Study 35

d. One-Piece Test Specimens 37e. Tensile Specimen Gasket Materials 38

7 Phrase V- Temperature Investigation 38

8 Phasa VI - Stress Investigation 3da. Introduction 38b. Theory 39

c. Experimental Procedures 42

(1) Stress-Strain Measurements 43

(2) Seal Residual StressMeasurements 44

(3) Calculations 45d. Experimental Results 46e. Summary of Strqzs Study 47

ii

TABLE OF CONTENTS (Cont)

Paragraph Page

SECTION C - CONCLUSIONS

9 Conclusions 48

SECTION D - RECOMMENDATIONS

10 Recowmendations for iuture Work 50

BIBLIOGRAPHY 52

APPENDIX - TABLES

Table Page

1 Summary of Thermcdynamic Calculetions 55

2 Resulto vf Adherence Tests and Torque PeelTests Performed on Specimens 1 etallizedWith Compositions Based on the AluminaReaction Theory 56

Results of Adherence Tests and TorquePeel Tests Performed on SpecimensMetallized With Compositions Based onthe Molytdenum Oxide Theory 59

4 Materials Which Lower Glass Viscosity 59

5 Results of Adherence Tests and TorquePeel Tests Performed on SpecimensMetallized With Compositions Ba'sed cnthe Glass Migration Theory 60

6 Results of Adherence Tests and TorquePeel Tests Performed on SpecimensMetallized With Compositions Based onthe Molybdenum Jintering Theory 62

iii

TABLE OF CONTENTS (Cont)

Table Page

7 Rosults of Adherence Tasts and TorquePeel Tests Performed on SpecimensMetallized With Compositions Based onthe Glass Additive Theory 63

8 Results of Adherence Tests and TorquePeel Tests Performed on SpccimensMetallized With Other Compositions 64

9 Results of Compression Tests FromMetallized Discs Sintered at 13000CUsing Experimental Metallizing Mixtures 67

10 Results of Compression Tests FromMetallized Discs Sintered at 14000CUsing Experimental Metallizing Mixtures 68

1 Results oP Comyression Tests FromMetallizoo )iscs Sintered at 15000CUsing Experimental Metallizing Mixtures 70

12 Rsults of Compression Tests FromMetallized D.scs Sintered at 16000CUsing Experimental Metallizing Mixtures 73

1 Results of Compression Tests FromMetallized Discs Sintered at 17000CUsing Experimental Metallizing Mixtures 77

24 Summary of Compression Test Results 79

15 Results of Tensile Tests Using Sperry'sDesign No. 1 and ExperimentalMetallizing Mixtures Sintered at 13000C 80

16 Results of Tensile Tests Using Sperry'sDesign No. 1 and ExperimentalMetallizing Mixtires Sintered at 14000C 80

17 Results of Tensile Tests Usirg SperrylsDesign No. 1 and ExperimcntalMetallizing Mixtures Siz.'tced at 15000C 81

iv

TABLE OF CONTENTS (Cont)

Table Page

18 Results of Tensile Tests Using Sperry'sDesign No. 1 and ExperimentalMetallizing Mixtures Sintered at 16000C 82

19 Results of Tensile Tests Using Sperry'sDesign No. 1 and ExlerimentalMetallizing Mixtu-es Sintered at 1700C 83

20 Summary of Tensile Test Results 8421 Recommended Metallizing Mixtures 8522 Analysis of Comparison Test Data 85

23 Stress Values 86

V

LIST OF ILL'"RATIONS

Figure Title Page

1 Metallizing Compositions as Applied toTest Discs 87

2 Furnace Temperature Profile 87

3 Torque Peal Testing Fixture 88

4 Compression Test: Specimen 89

5 Sections of Two Compression SpecimensStill Vacuum-Tight After Distortingfrom Over 5000-Pound Load 39

6 Tensile Test Specimens 90

7 Tensile Specl- -i After Testing 9.

8 Comparison Test Specimens 91

9 Schematic Diagram of Drnm Peel Apparatua 92

10 Sections of Test Specimens With StressPatterns Shown by Photoelastic Technique 93

11 Redesigned Seal TensileLTest Specimens 94

12 Seal Elements Undergoing Stress 95

13 Stress-Strain Graphs 96

14 Stress-Strain Graphs 97

15 Stress-Strain Graphs 98

16 Transverse Tensile Stress-Strain Specimenin Modified High-Temperature Extensometer 99

17 Csramic-to-Metal Seal Test Specimen forStress Investigation I0C

vi

LIS OF ILLUSTRATION.S CCon 2

18 In: rumente Stress Specimen P019 Stre~s Specimen Etching Apparatus

10220 Stress Versus Stran for "A" Nickel. 10221 St"ess Versus 3traln for 303 Stainless

Stel10422 Linear Expansion Versus Temperature 104

vil

FOREWORD

This study was conducted by theTechniques Sectionof the Electronic Tube Division, Sperry Gyroscope Company

Division of Sperry Rand Corporation, Great Neck, New York.

The Sperry Materials Laboratory also -articipated in the

program; this Zroap was responsible for the stress analysis

portion of the study and for other metallurgical and

iLeasurement aspects. Sincere appreciation is expressed

to Lt. Charles Martin of RADC, the Contract Officer;

Mr. James McLinden of Sperry, Techniques Section Head;

Mr. Frank Vanderveer of the Sperry Materials Laboratory;

and to the many other individuals whose contributions

made the work possible.

viii

ABSTRACT

To achieve ceramic-to-metal seals demonstrating

strengths as high as the ceramic member itself required a

thorough testing program for their measurement and evalua-

tion. A study was also conducted on the analytical and

experimental nature of seal stresses.

A literature survey on ceramic-to-metal sealingtechniques, adherence theory, and allied systems disclosed

limited published work and no procedures for achieving

ultra-high-strength seals or seals to pure high alumina.

Reported work on adherence mechanisms is limited to

chemical reacticn and molybdenum oxide reaction theories.

Two additional theories were formulated for this

study--one proposing the migration of the glass in the

ceramic into thc metallizing mixture, and the otherrecognizing the need for promoting metallizi ..intering.

These theories, together with thermodynamic and equilibrium

calculations, allowed 200 metallizing compositions to be

formi lated.

Three sintering temperatures were chosendepending on composltinn, for each of the 200 metallizing

mixtures. Each mixture was applied to specimens of 94-,

96-, and 99.6-percent alumina. Testing involved a screening

technique whereby the most promising composltions wera

carried through to increasingly refined test techniques(scratch and peel, circumferential seal, and finally tensiletests). Approximately 3200 specimens were prepared and

tested.

ix

The toile test specimen was redesigned to eliminate

snoulder breaks when evaluating ultra-high-strntn seals.

A photoelastic study was made to learn more about atress

distribution in this specimen.

Extremely strong seals were developed for all the

ceramic bodies considered. A -1.1e variety of sealing com-

positions was disclosed which produced seals stronger than

those previously reported. Cpreful analysis of the data

indicated that the Glass Migration Theory should receivi

careful consideration and that simple chemical reactions were

not enough to explain seal adherence.

A stuuy was made of the origin and nature of seal

stresses resulting from the dissimilar physical properties

of metals and ceramics. A method to calculate stresses in

ceramic-to-metal seals is theorized. Measurements of the

properties of the metal and of residual stresses in seals

were made, showing excellent agreement with calculated

stresses.

x

SECTION A

INTRUDUCTI ON

1. PURPOSE OF PROGRA1

This final report, discussing a study of metal-to-

ceramic seal technology, h9s been prepared for the U.S. Air

Force Air Research and Development Command by the Sperry

Gyroscope Company Division of Sperry Rand Corporation$ Great

Neck, New York. The work described herein was performed

under Contract No. AF30(602)2047 during the period 22 June

1959 to 22 September 1960.

The objective of this study was tc advance ceramic-to-netal seal technology to the point where stronger seals--

seals with bond strengths approaching tensile strengths of

25,000 psi, the strength of the nnramic nember--could be de-

veloped and produced. It was anticipated that through

theoretical and experimental investigations, seals would be

developed which not only were stronger as far as bond strength,

was concerned, but also would be more satisfactory in meeting

the 4- ands for increased electrical, mechanicall and thermal

requirements which are being imposed by present and futurehigh-performance electronic-tube applications.

Two major directions were explored to fulfillthe

goal of the program. The first was a comp1lation of possible

metallizing mixtures through theoretical considerations such

as high-temperature phase equilibria, thermodynamics, and

equilibria calculations, with subsequent experimental fabria-tion and testing of 200 of the most promising mixtures. The

second direction was an analytic and .xperimental investigation

1I

of Lhe stresses developed at the ceramic-metal interface due

to differences in the coefficients of tr,0rmaJ axpansion of the

metals and the ceramics. High-temperature prope:ties of

several metals and ceramics were measured, and stresses in the

ceramic-metal interface were predicted on the basis of these

data. A method of calculating stresses knowing the physical

properties of the materials involved was determined. These

calculated stresses were then compared with stresses actuallr

measured in faLricated cerazic-to-metal seal assemblies.

Two technical papers resulted from this contract

and a third is anticipated. These were "Theory of Adherence

in Ceramic-to-Metal Seals" by S.S. Cole, Jr., and H.W.

Larisch, presented at the Fifth Tube Techniques Conference;

"The Glass Migration Mechanism of Ceramic-to-Metal Seal

Adherence" by S.S. Cole, Jr., and G. Sommer, present6d at the

Electronic Division of the American Ceramic Society; and

"The Calculation of Stress In Ciramc-to-Netal Seals" by

S.S. Cole, Jr., and S. Inge, which is proposed for presenta-

tion at the Annual Meeting of the American Ceramic Society.

2. PHASES OF THE PROGRAM

The goal of this prog.am was achieved through six

major phases of itidy, the first five of which were intimately

interdependent. These were as follows:

a. Phase I - Literature Survey and Analysis - A generalsurvey was made of all literature related to cerumlc-to-metal

seal technology. This survey included a study of ceramic-

metal reactions which were not necessarily related to common

seal technologj, but were useful In gaining more fundamental

knowledge concerning bond mechanisms.

2

b. Pnase II - Metallizing Investigation- A comprehensiveinvestigation of metallizing materials and their fabrication,

application, and processing a conducted. Pha .e equilibria,

thermodyraamics, and previous work en this field were considered.

Two hundr6d experimental metallizing compositions were prepared

and tested by a series of successfully refined tests. Efforts

were made to relate data to the nature of the bond mechanism

or mechanisms between various metallizing material, -ndceramics.

c. Phase III - Brazing Investigation - An investigation ofthe effect of solders, their composition, thickness, time in

the liquid state, the effect of weights, and the role of the

metal members on brazing was planned. However, this phase was

nct necessary to fulfillthe pu- i of the program. ConvGn-

tional brazing techniques were I capable of producing the

desired strengths provided a me64Alizing mixture was properly

sintered onto any particular ceramic in question.

d. Phase IV - Testing Investigation - A broad and definitivetesting program was undertaken to evaluate testing techniques

and variables. Testing methods included the torque peel,drur peel, tensile, and compression tests, and also the leak

checking of compression-test assemblies. A comparison oftesting methods was conducted on a standardized metallizing

mixture, which was applied, sinteredp and brazed under

standardized conditions. The results of this comparison were

statistically analyzed.

e. Pha-e V'- Temperature Investigation - This phase was a

study of the effect of temperature cycling on seal igt1'and seal vacuum tightness from sub-zero to elevatedtemperatures.

3

f. Phase VI - Stress Investigation - A study of the stresses

involved in simple ceramic-to-imetal seal tructures due todifferences between the thermal expansio;. rates of metals andof ceramics was conducted. To gain basic knowledge in theseareas) comparisons were made between calculated and measuredstresses in ceramic-to-retal seal assemblies.

4

SECTION B

DISCUSSION

3. PHASE I- LITERATURE SURVEY AN D ANALYSIS

a. Introduction

Phase I consisted of a review of published literature

concerning ceramic-to-metal seals, with particular emphasisdirect I toward the refractory-metal process because of its

wide use in the electronic-tube industry. To metallize

metal-ceramic sealp by the refractory-metal process, a thincoating of finely ground metal particles is placed on theceramic surface and heated to teweratures in the range of

12000C to 13000 C. During the heating process, the metalparticles sinter and adhere to the ceramic and to each other.The result is a hard, rough coating to which other metals canbe bonded. The metallizing is electrically conductive; butbecause of its extreme thinness. it does not lend itself tothe common metal-working processes such as machining or drilling.

A wide variety Gf metals were found satisfactoryfor the metallizing process. Historically, the first refrac-tory-metal seqls were composed of molybderum or tungstenmetal and iron. 2'-' Vaious additions were made to themolybdenum or tungsten, including manganese, titanium,

nickel, iron oxide, manganese oxidej and glasses. In additionto being carried outinavacuum, the sintering process wasconducted in atmospheres of hydrogen, argon% dissociated

ammonia, producer gas, and natural Ras.

*The references cited are located in t..e Bibliography fcllowingthe teat.

5

The investigation of adherence mechanisms revealed astate of considerable complexity and one not easily satisfied

by any single theory. This condition was curthcr complicated

by the fact that witbin the refractcry-metal or Telefunkensealing group there are several basic types of metallizingmixtures which are placed on various high-alumina ceramics

(ranging from 85-percent to almost 100-percent alumina). Itis likely that different adherence mechanisms are operativeas the metallizing types and alimina content of a body are

changed.

b. Alumina Reaction Theory of Adherence

The earliest references of adhering molybdenum andtungsten seals to ceramics are by H. Pulfrich 4 -1 0 and

H. Vatterlll 6 . Working basically with steatite rather thanhigh-alumina bodies, Pulfrich was nevertheless aware of therole of chemical reactions and liquid phases. He recognizedthe need to heat the metallizing to temperatures which ap-

proach softening or eutectic ooints in the ceramic. Pulfrichstated that the furnace atmosphere should contain sufficienthydrogen to maintain most of the molybdenum as a metal, butalso sufficient oxygen (about 0.25 percent) to form a trace

of molybdenum oxide. This oxide was said to melt and flow tothe ceramic surface tnd there promote bonding. The possi-

bility that adherence iay be due to the glassy phase of theceramic was also recognized.

In 1953, Pincus, after considering some basicchemical reactions and after several microscopic observations,drew some basic conclusions1 7 . He postulated that manganesein a wet hydrogen atmosphere will oxidize to manganous oxiae,a reaction completed at 10000C. As the temperature incrcases,

6

a solid-state reaction begins to form the compound manganesealuminate, MnO.A 203, also called manganese spinel. A further

increase in temperature produces a molten -r zi4g condition ofthis compound at the ieramic-metal interface. At 14000C, an

appreciable sintering of molybdenum particles to ee.ch other

has taken place, and the spinel has begun to lock tnishardened layer to the ceramic. Increasing the temperature

furthqr causes the mass of manganese spinel to begin to

crystallize, thereby forming ga xyrte, a second crystalline

form of spinel. Finally, prec.ipitation of corundum (A1203)

crystals will occur. This precipitation, Pincus advccateso

heralds the general weakening of the seal.

The Alumina Reaction Theory predicts that seals

made to a 100-percent alumina body should be as strong, or

stronger, than those made to a 90-percent alumina body.Experimentally it has been universally observed that as thealumina content increases seals become more difficult to make.

In a later paper on adherence mechanisms, Pincus

postulated that bonds between pure molybdenum and high-aluminaceramics, though weak, were chemical in nature and depended

upon a reaction between molybdanum oxide and aluminum oxide18.

A nuvaber of reasons exist ahich cause certain doubt.

Molybdenum oxide, either MoO2 or MoO3, has ne'.:r been reportedas occurring after firing molybdenum metal in a hydrogen

atmosphere. An argument has been put forth that hydrogenatmospheres heavily laden with water vipor will supply the

neiessary oxygen to allow the reaction

Mo + 2H20 - 2H2 + MoO2

to take place. Although this reaction is thermodynami.allypredictable at rather low temperatures 19 "2 , repeated attempts

7

to achieve it at Sperry and at other organizations have failed.Even if the oxide is formed, for example, by heating in air orby prolonged low-temperature heating in ,tt Jbd..rogen, it is sovolatile that it immediately vaporizes at temperatures higher

than 6000C to 7000 C.

An immediate question concerns the purpose of watervapor in hydrogen gas. To form spinal, it is necessary to

satisfy the reaction

Mn + H 20 - MnO + H2

A second important function is in the promotion o± sintering,as water vapors are known to aid sintering to a very marked

degree.

It is questionable whether the molybdenum-iron systemwill behave the same. To date, no work has been reported,probably because of the comparatively small use of this mixture.A suspicion that there is considerable complicity in thissysLem is warranted because the reaction

s'e + H20 FeO + H2

can be expected to be highly temperature and dew-point sen-

sitf.re, as predicted by thermodynamic calculations.

Denton and Rawon investigated two refractory-metaltechniques, namely ths molybdenum-manganese and the molybdictrioxide techniques 4 . The importance of the minor constit-

uents of the ceramic in the sealing mechanism Is pointed outin their paper. They conclude that in the molybdenum-manganese

technique, acidic oxides, such as Si02 , are likely to have animportant effect on the metallziag behavior on the ceramic.In tie molybdic trioxide techniquep however, bazic oxides,

such as MgO and CaO, may be more important. Denton and Rawsonals.. conclude that the texture of the ceramic is of importancein controlling the lijteraction between thp ceramic and themetallizing layer, fine grain ceramics being, in general,easier to metallize. The Alumina Reaction Theory is supportedby these researchers.

c. Glass Phase Migration Theory of Adherence

In contrast to the tyce of mixture in which 15 to 30percent of the -ix is manganese, iron, or some other metal,

a second basic metallizing type can be considered. This is

the group in which the metallizing mixture i- " rgely molybd-rnumand a small addition of an active matsrial, u- ly titanium.Working with a mix of 94-percent molybdenum an. enttitanium, Cole and Hynes investigated the effec imiccomposition on seal strength 2 5 . This work suggidependence of seal strength on both glass content and glass

composition within the high-alumina body. No attempt was made

to theorize a sealing mechanism, although subsequent studiespointed to a very probable mechanism in this system. Titanium,like manganese, will readily form an oxide in a wet hydrogenatmosphere according to the reaction

T. + 2H,0 TiO2 +o2H 2

It is very probable that the titanium dioxide entersthe glassy phase of the ceramic and causes a reduction inviscosity. This, in turn, enables the glass to flow slightlyand enter the interstices of the molybdenim coatng. Micro-scopic examinations supported this theory; in addition thedecrease in seal strength with increasing alumina content is

preeictable.

9

d. Other Factors Affecting Adherence

Although adherence of the molybdenum to the ceramic

is the most elusive aspect of ceramic s6zs, tere are otherfacts to consider. The plating of the molybdenum coating

adheres to the metallized coating because of mechanical mAans.

This plating is not well bcnded and may be easily peeled off

if it is allowed to become too hilck. If, however, the platedcoating is fired, a solid-state sintering reaction occursbetween the molybdenum particls idtering to the ceramic andthe plated metal. The rough and somewhat porous molybdenum

boundary layer, into which the plated metal has diffused, canbe seen under a microscope. It was observed that this diffusionis substantially increased by firing the coating.

Whether or not the plate is fired prior to brazingis a disputable question, because it undergoes a heating opera-tion during brazing. What actually happens during brazing

depends mainly on what solders and plates have been chosen.Tf opper plate and copper solder are used, both will melt andenter the porous molybdenum coating heavily. If a highermelting plate is used (for example, copper-plated metal memberand silver-copper eutectic solder, or nickel-plated metalmem'er and copper solder), the occurrences in brazing arecomplicated. Phase diagrams found in the Metals Handbook26

are helpful in this respect. In general, the solder will reactwith the plate and the final alloy can be roughly estimated byuse of the phase diagrams. The degree of reaction will be

determined by the plating thickness and by the a3ount of timethe solder is allowed to remain liquid. Usually a microscopicexamination of the seal will not show any trace of the plate.

However, this is not always true; occaiionally the resultirggraded alloy will be seen, or the plate can be observed to be

10

nearly unaffected by the solder. Concern about the intricacies

of the brazirg operation becomes a secondary problem because

seals have been found to fail repeatedly at tre ceraic-to-metal

interface.

e. Related Systems and Porcelain-Enamel Reactions

Some qualitative information from related systems,

such as the fabrication of cermets, and also from porcelain-

enamal reactions aided in tne uaidea'standing of seal mechani.ms.

The reactions which take place during the formation of cermet

bodies occur, in the majority of cases, between ceramic-typematerials such as titanium or silicon carbide, and metals

such as iron, nickel, and chromium and/or alloys such as

Haynes Alloy No. 1, Haynes Alloy No. 25, and Nichrome. Thesereactions involve the dispersing of the ceramic ccnstituentin the form of grains within a continuous metallic phase.The dispersing takes place during a liquid-or iolid-sta;esinterine operations Such as (1) hot pressing (simultaneousheating and pressing in an induction furnace in the presence

of a protective atmosphere), (2) cold pressing and subsequentsintering in a protective atmosphere furnace, or (3) vacuuminfiltration (diffusion of metal or alloy into a ceramic-

type pozous skeleton m" ial in a vacuum furnace).

Porcelain-enamel reactions occurring in the fusingof fritted glasses to hot-rolled enameling iron involve the

interaction of oxides, carbonates, nitrates, and fluorides

(after their smelting, in which case the lecs scable carbonates

and nitrates are converted to stable oxides) with iron and its

various oxides in the presence of heat. The adherence

phenomena ples.?nt after these reactions hava been com;letedare complex and subject to continual review and debate.

Neither of the two forementioned ,,. ,ects seem to involve re-actions of alumina and metal to cny ex" nt.

11

4. PHASE II - METALJJ/71G INVESTIGATION

a. Investigatio,. Lveign

The metallJzin? investigat~on, one of uine majorefforts in this study, was developed around five bisic sealingmechanisms. These ares

* The Alumina Reaction Theory, which depends on a

chemical reactic- of the metallizing composition

and the ceramic.

o The Molybdenum Oxide Theory, which depends on the

reaction of molybdenum oxide with ceramic.

e The Glass Migration Theory, which depends on glassmigration from the ceramic into the motallizingcoating.

* The Molybdenum Sintering Mechanism, which recognizes

the need for adequate molybdenum sintering.

* The Glass Additive Mechanism, which suggests that

-nals can be accomplished by adding glass to the

metallizing composition.

In addition, a category of compositions which does

not appiar to conform to any theory, but which has been reported

to be of high seal zlrength was investigated.

The above categories, used with thermodynamic and

equilibrit m-diagram data where possible, alloved the formulation

of th6 200 metallizing compositions listed in t e various table,in the Appendix. These compositions were determined in the

following manner.

12

(1) Alumina Reaction Theory

The Alumina Reaction Theory, a. proposed by A. Pincus,

predicts a compound formation between the aluminh and one of the

metals used in the metallizing mixtures. By measuring the freeenergies and heats of formation in related chemical reactions,

it is possible to predict whether other reactions can be ex-

pected to occur. As an example of such a thermodynamic pre-

diction, water has a free energy of formation of -52,360 calories

per mole* at a tAmperature of 5000K. This can be written as

H2 +1 0 2-----4W H2 O;F = -52,360 cal

At the same temperature, the reduction of titanium dioxide can be

written as

T1O2----- Ti + 02;&F = ..203,450 cal

To predict the reaction of titanium metal with watervapor or hydrogen, such ab is present in sintering furnaces,the above reactions are added:

2H2 + 02 - 2H20; LF = -104,720 cal

TiO2 Ti + 02; AF = +203,450 cal

2H2 + TiO2 - - Ti + 2H2 0;AF = +98,730 cal

Because this reaction has a positive free energy, it w.6l proceedto the left, forming TiO2 at 5000 K, providing there is no very

large excess of hydrogen present. However, in a wet hydrogenatmosphere, a large excess of hydrogen is present. If thewater content becomes low enoug the reaction will tend to driveto the right despite the large positive&F. This possibility

*The minus sign indicates the tendency for the reaction to pro-ceed to the right; a positiveAF indicates a tendency to proceedto the left.

13

may be predicted by a consideration of the partial pressures ofthe two gases, hydrogen and water vapor, which are present.These factors are related by the followtii fc:.oula:

ALF = -RT 1n PH (at equilibrium)

Ri PH2

which is rewritten as

PH20 _AFIn .

PH RT2

where

PH20 = partial pressure water

PH2 = partial pressure hydrogen

F = free energy of reactionR = molar gas constantT = absolute temperature

If ln(PH2O/P12 ) is numerically greater tnan AFiRTI,the equation will not be satisfied and the titanium will tendto oxidize. However, if lr,(PH20/PH2 ) is less than -AF/RT,reduction can be predicted. At a dew point of OC,

PH20in 2- - =-3.4

2

and at 5000K,

-98

Oxidation can be expected because ln(PH 2C/PH2 ) is numericallygreater than -AF/RT. Previously, this has been proven validby experimental means. It can further be shown that a dewpoint of about -104OC is required at 5000K to cause T10 2 to

14

rAduce. As in all cases, it must be pointed out that thermo-

dynamic predictions do not consider reantion rate or any of

several other factors which may slow or stop a :eaction from

proceeding; these are, nevertheless, very accurate predictions.

Having predicted whether an oxide or a metal will be

present, it is possible to predict from phase d4agrams the

reactivity of aluminum oxide with other oxides. The diagrams

also give a good indication of the temperatures at which the

reactions can be expected tc occur. Table 1* lists the metals

whose oxides will react with aluminum oxide, their tendency

to oxidize, the melting point of the metal and its oxide,and the lowest melting eutectic between AI203 and the metallicoxide. From this table, compositions have been formulated

which should behave according to the alumina reaction mechan-ism. These are shown in table 2.

(2) Molybdenum Oxide Theory

The second category (suggested by Pincus) is basedon the chemical reaction between the primary material andthe ceramic. This would involve, for example, the formationof a small amount of molybdenum oxide, which would then reactin zhe same rashion with the ceramic. Table 3 lists thecompositions containing additions of MoO3 to apply this theory

of adherence.

(3) Ilass Migration Theory

The Glass Migrationi Theory, which recognizes the

importance of the glassy phase in the ceramic, was developed

from the work conducted by Cole and Hynes2 . It is proposed

*Tables are grouped in numerical ordor at the rear of the report.

15

that certain metals, such as titanium, after having oxidized

in a wet hydrogen atmosphere enter the glassy phase of the

ceramic and lower the viscosity of that .! ass. The glass is

then free to flow out slightly and lock the ceii ic to the

somewhat porous molybuenum coating which remains.

An entirely different direction is used in the

approach to the Glass Migration Theory. Over the period of a

great many years, the glass, enamel, and glaze manufacturers

have been able to determine the materials which are known to

lower the viscosity of glassis; a substantial list of materials

has evolved which is suspected to affect greatly glass vis-

cosity. These are shown in table 4 along with their principal

sources. From this table, compositions were made which should

behave according to the theory (see table 5).

(4) Molybdenum Sintering Mechanism

The sintering mechanism recognizes the need for ac-

complishing thorough sintering of the molybdenum particles and

cells. For 3intering to occur, the sintering particles must

be in intimate contact with each other so that bonding can take

place at the point of contact. Theoretically, anything which

wouLd increase this contact area hucld enhance subsequent

sinterlng by supplying more bonding points. Any increase in

temperature not abo*e the melting point will enhance the

sintering rate because cf increased diffusion rate and plastic

deformation.

From the standpoints of mutual solubility, crystal

structure, and atomic size, the following elements were pre-

dicted for addition: titanium, vanadium, chromium, iron, cobalt,

nickel, zirconium, niobium, and tantalum. Of these, oLLy iron,

16

nickel, and cobalt will not oxidize in a wet hydrogen atmos-phere according to thermodynamic calculations. Compositionsformulated on the basis of these considertlonz ire shown intable 6.

(5) Glass Additive Mechanism

This mechanism po stulates that glassy or glass-forming materials can be added to a metallizing mixture com-posed basically of a refractory aetal such as molybdenum. The

glass thus added is then able to fuse to both the ceramic and

the metal particles.

A series of compositions based on the Glass Additive

Mechanism were made (table 7). They were largely determined

by an extensive literature search and a study of previous work

by other researcners in this field. Certain problems arise

in a study of this mechanism. The glass must have a high

softening point, must be capable of wetting both the high

alu t !uid the molybdenum grain, and must have fair mechanicalstrength. However, the glass must not be reduced in a wethydrogen furnace.

'6) Other Mechanisms

In addition to the above, the literature search

yielded a series of compositions which were considered worthy

of trial at various temperaturs and on the various high-

alumina bodies. These compositions are shown in table 8.

b. Evaluation of Metallizing Compositions

(1) Materials and Methods

Based on the above information, the ceramics listed

on the following page werechosen to be metallized and evaluated.

These are typical high-alumina ceramis which are of interest

to the electronic-tube industry.

17

* AD-94 - A dense 94-percent alumina body manufacturedby the Coors Porcelain Co.

* AD-96 - A dense 96-percent alunmira body manufacturedby the Coors Porcelain Co.

* AL-23 - A dense 99.6-percent alumina body suppliedby Materials For Electronics, Inc.

The following milling procedures were used:

* Inside dimensions of the steel mills selected because

of their capability for more efficient grinding thanporcelain, are 5.5 x 6 inches.

* 0.5-inch-diameter steel balls were placed 2 inchesdeep in the bottom of the mill. About six one-inchsteel balls were also added to each mill.

* Materials shown in tables 2 through 8 were weighed

and placed in the mill.

e A binder of 60 ml acetone, 60 ml amyl acetate, and25 ml 8-percent nitrocellulose lacquer was added.

* If, after 2 hours of milling, viscosity was not lessthan 50 cpa, binder additions were made in quantities

of 70 ml until the viscosity wis less than 50 cps.

* Milling was conducted for 24 hours at a mill speed

of 60 rpm.

* Mills were emptied and cleaned with acetone prior torecharging.

* Resulting mixtures were stored in glass .ars with

polyethylene-lined caps.

The compositions were then sintered to various ceramicsat various temperatures, and tested by a successive seriCs of

tests designed to be more exacting as the less promising com-

positions were eliminated.

18

Ao loop'20P-°

hp

Bo

(2) Adherence Test

Three different metallizing compositions were painted

on 1.5-inch discs of each of the three cerami,;q studied (see

figure 1). These were than sintered in molybdenum boats,

each holding 27 discs. The furnace atmosphere was wet

dissociated ammonia, the dewpoint of which was held between

+600F and +850F by passing the furnace gasses through a con-

trolled-temperature water tank. Figure 2 illustrates the

furnace-temperature profile and the sintering cycle of the

metallized ceramics for the l00 0 C sintering temperature.

The same sintering cycle in a furnace-temperature profile

similar to that shown in figure 2 "' used in the firings at

12500C, 13500C, 1400OC, and 1500°C. Because of a furnace

limitation, a different furnace was required for the 16000C

and 17000 C sinterings; the ;ame atmospnere, dewpoint, and

30-minute time period in , - hot zone of the furnace wereused.

After sintering, the coatings were adhfterence tested

using a scalpel and 30-power binocular microscopd. A system

of rough grading the coatings was devised as follows:

" Poor no adherence, coatings curl, no effort

expended in removal , cracks or holes

visible.

" Fair moderate adherence, moderate coating hard-

ness and cohesion.

" Good motallizing coating absolutely not removable,

ceramic removedp hard dense metal films.

Results of the adherence tests conducted on all discs

sintered at all proposed temperatures are shown in tables 2through 8. The adherence testing program was originally

19

designed to eliminate only those experimental metallizing

mixtures which exhibited grossly poor adherence, at all

siintering temperatures, to all three ceramic bodies considered

in this program. It soon became apparent tiat few metallizing

mixtures could be eliminated by this technique. N#. mixtures

yielded poor ratings to all ceramizs at all temperattres,

and few were found that could not be rated at least fair or

fair-to-good on some ceramics at some sintering temperatures.

Peel tests were next performed on all stripes of experimental

metallizing mixtures resulting frcm the adherence test pro-

gram. Because both tests were conducted on all combinations

of mix, body, and sintering temperature, results of both

tests are discussed concurrently in the following paragraphs.

(3) Torque Peel Test

The torque peel test was chosen as a quick, in-

expensive test of ceramic-to-metal seal strength from the

results of Phase IV, Testing Investigation. After completing

the adherence test, each stripe of experimental metallizing

on alldiscs of each of the three bodies was plated with

0.0005-inch hard nickel. A 1.25- x 0.375- x 0.010-inchKovar strip was then brazed to each stripe using a 0.002-inch

shim of OFHC copper. These Kovar strips were then peeled

from the ceramic by the torque wrench and torque peel fixture

illustrated in figure 3.

Results of the torque peel tests are also included

in tables 2 through 8. The massive amount of data contained

in these tables was difficult to analyze and was therefore

handled in the following manner. All metallizing compositions

were separated into groups according to the type of secondarymetal, oxide, or mineral present in the metallizing mixture.

A tabulation was then made of the number of times each

20

additive to a refractory-metal base produced superior peelte-t values. Each additive was further weighted according to

torque peel strength and whether it was used singly or in

combination with another material. This technite produced

numbers which could then be ued to correlate the number oftimes strong seals were produced against the number of times

each type of metallizing additin was tried. It was thus

possible to determine the general effect of each additive tothe refractory metal.

In audition, the total number of high peel test

values was correlated against sintering temperatures in

general, and against the alumina content of the body to which

it was applied. Each metallizing composition and compositionaltype which produced a high seal strength was further examined

to determine its optimum sintering temperature, the com-

position of the alumina body to which it best adheres, and

the effect of concentration of the additive on seal strength.

(4) Torque Peel Test Results

The rbsults of this analysis indicated that cerium

oxide and thorium oyide additions to molybdenum produced the

highest frequency of strong seals when used on Body AD-94.

Theae .ere closely followed by additions of titanium andtungsten. In decreesing order of improved torque peel strength

on AD-94 were additicns to molybdenum of talc, manganese,

sodium carbonate, titanium carbide, kaolin, and feldspar.*

Pure molybdenum trioxide produced promising seals if con-

sidered only singly, especially at the lower sintering tempera-

tures.

*The composition of talc is approximately MgO'SiO2, that offeldspar is K20.A1203.6SiO2, and ka0'lin is A1203.2SiO 2.2H20.

21

Seal strengths were generally lower on Body AD-96

than on AD-94, with cerium oxide and thcrium oxide producing

good results. Feldspar, talc, titanium, zironi.'r, titanium

carbide, kaolin, and silicon dioxide, .n that approximate order,

all produced promising seals. Pure molybdenum trioxide again

rated fairly well if considered singly.

Seal strengths were much lower on Body AL-23 than

on AD-94 or AD-961 feldspar, silica, and talc additions,

however, appear tn produce the be.t results. Barium oxide,

calcium oxide, zirconium dioxide, and manganese also seemed

to help. It is significant that silica and silica-bearing

minerals appoared to produce the most satisfactory seals to

AL-23, indicating that the formation of a silica glass or analumina silicate compound in the metallizing layer or at the

metallizing-ceramic interface may encourage high-strength

seals to this high-alumina ceramic.

It is also interesting to note tha feldspar andsilica additions had the same approximate efect on sealstrengths regardless of the choice of ceramic. This may

occur because additional amoiuts of glassy phase over some

optimum amount do not aid sealing, or because a chemicalcompound may be forming with the alumina. Talc producedstronger seals to the lower aluminas, possibly because of

the silica contained in this mineral.

These ratings are mentioned in a general ratherthan a specific manner because of the many variables involved.

Also, a composition will often yield a high'torque peel valueon a particular body at two widely separated sintering tem-peratures, but a low value when sinterod to a temperature

b"tween them. In these cases, it is assumed that processing

variables in plating, brazing, or testing affected the reault;

22

this fact is considered in the evaluations. An example is

Composition 24 in which the 15000C torque peel results on

AD-94 were considerably loijc- than the 14O 0 C a:id 1600Cvalues. Manganese dioxide, hc..efora, was considared slightll

better for use in molybdenum on AD-94 than a r,,)-.''tc:i cP1-

culation indicated. This type of rating was ce.:1- _ed -_tis-

factory becauic its only functicn was to choos. IVs, miALures

to be considered for further testin&.

Many trends were apparent when a careu1i analysis ofadherence test and torque peel test data was made. For

example, the silica and silicate additions, including feldsparand talc, ylelded highest torque peel values when sintered atthe higher temperatures. This suggested that migration ofsilicate glasses, the viscosity of which lowers with increasingtemperature, may be responsible for stronger seals. A chewtcal

compourd might also be formed, and thus would be more complireat higher temperatures. Alternatively, the manganese com-

pounds (for example, manganese dioxide in Compositions 23 and

24, and lithium manganate in Composition 88) yielded highest

values when sintered in the lower range of temperatures in-vestigated. The metallic additions to molybdenum, such as

nickl, iron, cobalt, and tungsten, yielded highest strengths

when sintered at the higher temperatures. With the exception

of irnn, which may oxidize under some sintering ^or ltions,these additions, being soluble in molybdenum, may promote

sintering and, thus, high seal strengths.

It can be seen that there was little difference

whether the element titanium was added as metallic titanium

or as the oxide (Compositions 32 and 33, respectively).

This supported previous data that titanium oxidizes tutitanium dioxide in wet cracked ammonia at elevated tempera-tures. Two additional observations w.-e (1) small additions

23

of titanium or the oxide to molybdenum produced as good or

better results as large additions and (2) these mixtures adhered

better to tha lower-alumina (94-percent A1203) than to the

higher-alumina (99.6-percent A1203) ceramic. These observationssuggested that the Glass Migration Theory rather than the

Alum±na Reaction Theory was operative. The same general ob-servations are true for both manganese and manganese dioxide

(Compositions 49 and 50).

The amount of titanium dioxide necessary to produce

maximum strength appeared slightly higher than titanium metal;this was expected because the metal oxidized during sintering.

Titanium also improved strength over a wide range, but smaller

amount- were effective; an optimum addition appeared likely at

about 3 percent by weight.

Additions of talc and feldspar also seemed to showan optimum strength at about 3 to 5 percent, though smaller

amounts yielded higher strengths than pure molybdenum.Cerium oxide, though helpful over a wide range, apparently

was best in small amounts, 0.5 percent yielding noticeably

improved strengths.

Many materials appeared to improve adherence whenadded in very small amounts, but degraded seal strength whenadded in larger amounts. Barium oxide, lithium manganate,

lithium titanate, and lithium carbonate showed this effect.Additions of metallic nickel and cobalt were also helpfulwhen added in small amounts, less than 1 percent, but theydegraded adherence when the concentration increased.

(5) Compression Test

The compression test was applied to the most promisingcompositions found by peel testing. Figure 4 illustrates thetest specimens and fixture used. A 0.125-inch ceramic disc

24

was metallized on its outside diameter and then brazed intc a

tight-fitting Kovar .Aeeve to form a vacuum-tight assembly.

This assembly was tested by compressing two 0.3.25-inch tight-

fitting rubber discs in the fixture. The loading rate used

on the Baldwin Universal tester was 6000 pounds per minute to

a load of 2000 pounds. The assembly was then checked again fo,

vacuum tightness and loaded at a rate of 3000 pounds per minute

in 100 pound increments, leak checking after each successive

increment of load until a leak was detected. Of interest was

an audible crack at failure, always detected while the speci-

men was being loaded, which allowed the operator to determineexactly when the leak occurred.

Results of the compression tests are shown in tables9 through 13, which present data from metallized discs sinteredat 1300OC, 1400OC, 1500C, 1600C, and 1700C, respectively.

The maximum load values recorded are slightly over 5000 pounds,indicated by the symbol > 5000, because at that developedpressure the rubber discs flowed past the expanded Kovar sleeve

in an almost liquid condition. Later tests were limited to

4000 pounds since the rubber distorted and shredded. Figure

5 shows a section of two assemblies which would not fail even

after a loading of>5000 pounds. These assemblies, though

distorted, are still vacuum tight.

(6) Compression Tesc Results

It can be seen from table 8 that only two mixtures,

Composition 50 and Composition 199, showeu promise of high-strength seals when sintered to 1300C. Composition 50 is amolybdenum-mangarese mixture with an addition of silica, andComposition 199 is a 100-percent molybdic trioxide mixture.

25

Many more met3llizing compositions yielded promising

ceramic-to-metal seal strengths when sintered at 14000C, as

indicated by the compression values shown !.n table 9. For

example, Composition 43 wich an addition of feldsptr and

Composition 45 with an addition of talc produced fairly high-

compression values on high-alumina AL-23. Glassy phase

resulting from diffusion of these silicate materials may be'nstrumental in the sealing of the materials. Composition 47

with silica and MnO incorporated in the mixture and Composition50 with silica and manganese zhowed promising results at thissintering temperature.

A general summary of the compression test data forapproximately 1500 test specimens is shown in table 14. Thetype of mixture which produced the highest seal strengths,along with symbols which represent metallizing systems, arepresented. Where two or more compositions were formulatedwithin a given system, some using the metal and others usingthe oxide, only Che metallic symbol describes the system.Where only the oxide was used, it is so written. The figureof merit represents the number of times any metallizing systemproduced strong seals as opposed to the number of mixtureswhict were formulated and tried on the system. For example,if five cerium oxide and molybdenum mixtures were tested andfour were associated with strong seals, the figure of meritwould be 4/5. In general, strong seals are defined as thosewhich are equal or superior to the strength which would begenerated using a 20-percent manganese, 80-percent molybdenummetallizing mixture. Many attempted metallizing systemsfailed to produce strong seals, resulting in a figure of meritof 0/4, for example; these are not iicluded in the table.

26

Several observations can be made by studying table 14.One of the most important is that a very I.arga number of meta.s

or oxides can be added to molybdenur to produce satisfactoryceramic-to-metal seals. No fewer than 16 mtals or their oxides

were found to produce seals of the same or higher quality as

manganese.

An immediate observation can also be made regardingthe sharp decrease in metallizing systems which were satis-

factory cn the higher-alumina ceramics t as compared with the94-percent alumina body. The increased difficulty in sealingto high alumina is thus apparent in most metallizing systems,

and not only in thc molybdenum-manganese systems as has beengenerally conceded in the industry.

To accomplish seals to the 99.6-percent aluminaceramic, AL-23, silica additives, at 14000C and 150oC, weresuitable in a large percentage of cases. Examples of thisare Compositions R39 and R50, which showed higher strengthsto the AL-23 body than to the lower-alumina Coors ceramics.At 16000C and 17000C, titanium additives and molybdenum andglass additions were most satisfactory. Very few promisingmet.lli zing mixtures were generated &t 13000C, but the numberincreased steadily to 1600 0C and then decreased at 17000C.

As the sintjring temperature was increased, a steadydecline was noted in the number of metallizing mixtures con-taining manganese. At 1400C, on Coors AD-94 and Af-96, allbut one composition were manganese bearing, at 1700C, nonecontained manganese. This decrease also occurred between15000C and 16000C.

Titanium was a material very frequently associatedwith high seal strength. At 14000C, slightly less than one

27

third of the high-strength compositions contained titanium ortitanium dioxide; at l5000C, more than half i.d The percentageindicated a marked decrda.ef however, at 16000C and 17000C.

Materials which were added solely to promotemolybdenum sintering were found frequently at 1600C and 1700C.

These are metals such as iron, nicKel, cobalt, and tungsten

which will remain in the metall.c state in a wet crackedammonia atmosphere. As show, by the figures of merit, nearly

40 percent of the compositions at 16000C contained sintering

promoters and nearly 60 percent did so at 17000C. The balanceof suitable mixtures at these temperatures was rather random,

with silica additions being the most common.

Pure molybdenum and molybdenum oxide made manysatisfactory seals. The oxide was promising at 13000C and

14000C, whereas pure molybdenum worked well at 16000C and17000C. Oxides other than molybdenum oxide did not offer any

measurable advantage over pure metallic additions, undoubtedlybecause the metals quickly developed their stable oxidation

levels anyway with cracked ammonia atmospheres.

%7) Tensile Test

Those combinations of experimental metallizingmixtures, ceramic bodies, and sintering temperatures that

yielded the highest values in the compression test werechosen for tensile testing. Sperry Tensile Design No. 1(figure 6a) was used for the tensile tests. Because of thevariation between the three specimens of each combinationand because of the general decrease in strength noted with

the higher alumina bodies, tensile tests were conducted oncombinations yielding the following o: higher compression

test valuess

28

AD-94 AD-96 AL-23

Compositions havingindividual loadvalues of 5000 1b 350C lb 3000 lb

Compositions havingaverage load values of 3000 lb 2500 lb 2000 lb

After cleaning and air firing the tensile specimens,two coats of a metallizing mixture were hand painted on theceramic surface, each coat beig sin(.ered at the appropriatotemperature. Each metallized surface was then hard nickel

plated. Two half specimens were then brazed together usingOFHC copper shim stock in a suitable brazing Jig. Tensile

testing was done on a Baldwin Universal Tester (60,OO0-pound

capacity), using a load rate of 3000 pounds per minate.

Rubber gaskets between the surface of the ceramic and the

steel pulling fixtures were used to equalize the stresses atthe shoulder of the tensile specimen. Tensile testing data

are shown in tables 15 through 19, for specimens sintered at

130000, 14000C, 1500-C, 1600OC, and 17000C, respectively.

(8) Tensile Test Results

It is apparent from the tables that some combina-tions yielded tensil values in excess of the 25,OOO-psi goal

of this study. All tensile test values were corrected forthe bending moment induced by the nonlinear loading, as dis-

cussad in paragraph 6 of this report. Composition'65 on Body

AD-94 at a 1500C sintering temperature, for eynmple, yielded

one tensile test value of 28,400 psi, with an average tensile

value for three duplicate spiciments of 21,400 psi. As shown

in figure 7 it is the ceramic which fractures in the vicinityof the seal area, but not through it, for these higher-

strength ceramic-to-metal seals. Left to right, figure 7

29

illustrates a specimen of AD-94 metallized with Composition 65,yialding a tensile strength of 28,400 psi; a specimen of AD-96metallized with Composition 72, yielding a tensile strength of22,000 psi; and a specimen of AL-23 metallized with Composition50, yielding a tensile strength of 16,100 psi. in each casethe sintering temperature was 15000C.

It should be noted that the appearance of the fractuiis not the most reliable means for determining if the seal isstronger than the ceramic. Th AD-94 specimen which failed ?.t28,400 psi shows a failure .argely in the seal area. However,the AL-23 specimen definitely has a seal stronger than theceramic, although it failed at 16,000 psi. The contributionof thermal-mismatch stress of the solder itself will undoubtedlyaffect the physical nature of the failure. This is an area inwhich more knowledge is needed. It is of interest that in over400 tensile pulls only one case of shoulder fracture is en-countered.

The analysis of the tensile test data was conductedby using the same techniques applied to the compression testdata, as described in paragraph 2b(6). Table 20 shows themetallizing systems which produced the strongest seals. Aswar the case for the compression test results, one noticesimmediately the wide variety of compositional systems whichproduced strong seals, as well as the sharp decrease of com-pojitions which were suitable for the 96-percent and the 99.6-par-ent aluminas.

Compositions containing manganese predominated at13000C and 1400C. At 15000C, manganese-bearing compositionsdecreased sharply, with two of six superior compositions con-taining ma~iganese. At 160OOC and 1700C, only one c-mpoitioncontained manganese.

30

All satisfactory seals to AL-23 contained silica and,

in iost cases, manganese. These seals were sintered at 1300OC,

14000C and 1500C. No strong seals were tade at 16000C and17000C to AL-23, althougi several ccmpositions were attempted.

Titanium did not produce superior seals as oftenwhen using the tenstile test as It did with the compression test.These compositions also predominated at lower temperatures.

At 16000C and 17000C, there was only one superior titanium-

bearing composition.

The wide variety of compositional systems at 16000C

and 1700oC is rather difficult to understand. At these tem-

peratures, 8 out of 26 compositions contain a sintering pro-moter as the only mlybdenum addition. Silica was present inanother five compositit..s. The remaining compositions consistedof additions of thoria, zinc, titanium, ceria, zirconia, andmanganese, as well as pure molybdenum. No significance in thislist could be found, except to note the variety involved. Acheck on each composition involved, however, showed that none

of these metallic additions was in excess of 15 percent byweight.

(9) Final Observations

Final observations in this phase o.' the study programshould be made by referring to table 21. The seven most

promising compositions and their sintering temperatures are

shown for the three bodies studied. The bost values for AD-94

were obtained using Composition 65, a 2.5-.?ercent titaniumaddition to pure molybdenum.

The tensile test average iA 21,400 psi. Both Com-

positions 91 and 141 produced averages ovw:r 15,000 psi. One

basic formulation, molybdenum-silica-manginiese, was excellent

31

on AL-23 and AD-96. Molybdenum-ceria alro prcdtced high-

strength seals to AD-96. All sever, compositions s hown in

table 21 are extremely promising and an investigation of their

reliability is highly desirable.

5. PHASE III - BRAZING INVESTIGATION

The brazing of ceramic-to-metal seals is recognizedas paramount in achieving reliable and strong seals. Factorssuch as Joint clearance, solder type, and soak time and rate,to name a few, have been established as important variables.Although a study of this phase was planned, it was found that

seals can be made as strong as the ceramic by using standardSperry brazing techniques, provided a superior metallizing

mixture was properly sintered onto the particular ceramic

considered.

6. PHASE IV - TESTING INVESTIGATION

a. Comparison of Test Methods

An accurate and thorough testing endeavor wasnecessaxy for the proper evaluation of this program, becausethe value of the data obtained was to be determined largely

by the test pieces and test methods selected. Such a programwith emphasis on reproducibility was therefore conducted byevaluating the tensile, the compression, the torqle peel, and

the drum peel tests. Sixty-two ASTM tensile test specimenswere prepared. These, in addition to the two half specimensbonded together, had two Kova7 strips bonded to the flatsurface of one of the halves to measure torque peel, and twopreformed Kovar strips sealed to the cntside diameter of eachof the two shoulders to measure drum pe'el. All brazes were

32

madi simultaneously, producing the specimens shown in figure 8.The ceramic used was Coors AD-949 metal2zed Ith. an 80-percent

molybdenum, 20-percent wanganese mi-cure$ sintered at 15000C,plated with hard nickel, and brazed using OFHC copper shim

stock.

Torque peel data were evaluated by inserting thespecimen into the fixture shown in figure 3. Flat Kovar stripswere crimped and engaged into E :lot built in the cylinder ofthe test fixture as shown. T'he load in inch-pounds was readon the -torque wrench by the operator, who recorded the maximumvalue indicated.

The drum peel test was chusen to eliminate theoperator variables in rate of application of the load and inthe reading of the data. Figure 9 illustratms the fixtureused. The load was applied at a constant, reproducible rateand recorded as pounds pulled versus time on strip chart papor.Figure 6b is an example of the tensile specimens of the designof which was originally proposed by ASTM. The method used wasthat under discussion by Group V-D, Subcommittee F-I of thatorganization.

To evaluate seal strength by compression testing, aspecimen and a fixtur3 as shown in figure 4 are used. Thistest has a distinct Pdvantage in that the con-figuration ofthe test specimen closely resembles the geometry often usedin tubes. Compressive forces loaded at a prescribed rateto the ends of the fixture produce hydrostatic forces in therubber discs$ these forces tear the ceramic from the Kovarsleeve. Failure is defined as the load at which the assem-bly is no longer vacuum tight.

Twenty-five comparison specimens for compressiontesting were prepared from Coors AD-94 ceramic, Kovar metal,

33

and the 80-20 molybdenum-manganese metallizing mixture. These

were sintered at 15000C, nickel plated, and OFHC copper brazed.Results of the test comparison series arG listed in table 22;

which shows an average strength value, the stancard deviations

the coefficient of variation, and the number of trials for

each compression, tensile, drum peel, and torque peel test.

It can be seen in table 22a that the compressiontest, with a coefficient of variation of only 10.5 percent,

had the greatest reproducibility and least variation. The

tensile test with a coefficient of variation of 27.8 percentproved to be less reproducible than the compression test.

The torque peal is significantly more reproducible than the

drum peel test and therefore was chosen for the initial

evaluation of the 200 experimental metallizing mixtures pre-pared later.

Table 22b presents a similar comparison in which n,the number of trialss was in all cases equal to 25. Themost significant improvement was in the torque peel test,indicating it was the most sensitive to changes in brazingconditions and also suggesting it would be the most sensitive

to variations in seal strength.

The averages in table 22a tend to indicate that the

following measures of seal strength are equivalent for thepaiticul-r ceramic, metallizing mixture, and conditions studied:

* Tensile Test 2011-lb load, 11,061 psi

* Compression Test 2428-lb load

* Drur "eel test 8.08-lb load

* Tc _ 1. Test 2.85.-in.-lb torque

34

On the basis of the above single-point curves, a

linear extrapolation suggests that the following values are

equivalent to a 25,000-psi tensile strength:

a Compression Test 5200 pou-nds

* Drum Peel Load 17.25 pounds

* Torque Peel 6.13 inch-pounds

As previously mer.timn;r, a limitation was found in

the compression test in tha. in pressures above a 5000-pound

load the rubber discs flowed in an almost liquid condition pastthe expanded Kovar sleeves.

b. Tensile Test Specimen Design

During the course 'of this investigation it was foundthat the tensile test specimen now under consioeration by ASTH

(figure 6b) had a serious shortcoming for the measurement of

very strong ceramic-to-metal seals. This fault is that failuresoccurred at the shoulder rather than in the seal area. ASTMI

in efforts Lo eliminate this fault, has considered an increase

in fracture pathp as shown in figure 6c. The calculations

presented in the Third Technical Noteindicated that an increase

in 4h& radius of curvature at the shoulder should decrease

stresses in that arqa2 7 . Because it was necessary to makecertain assumptions between the case cited by Timashenko andthe one in question, it was decided to measure the exact

stress picture using the photoeleastic technique.

c. Photoeleastic Study

Full-scale models of the three designs in questionwere made of a standard photoelastic material, as describedin the Fourth Technical Note 1 . By stressing models at elevatedtemperatures and chilling them while . till stressed, the

35

stresses produced in a three-dimensionE.l object can be viewed

by carefully sectioning and polishing the object after it has

cooled to room temperature. A photograph 'f a flat slice of

each of the three specimers with tho Ctress pattern "locked in"is shown in figure 10. The specimens are, left to right,Sperry Tensile Design No. 1 (designed under this contract),ASTM Tensile Design No. 2, and the design under present usety ASTM.

The fringes visible in the ASTM sample show thatthere is a higher level of stress in the shoulder than in theseal area by a ratio of approximately 2.5 to 2.0, thus pre-dicting failure in the shoulder area. Design No. 2 shows

approximately equal stress in both the shoulder area and the

seal area, so a failure at either point is equally probable.Sperry Design No. 1, however, shows a greater tensile stress

at the seal area than at any other point and thus should failat the seal.

The photoelastic study revealed another character-

istic of thesw Scecimens which was not suspected previously.

The bending moment induced by loading which was not directlyabove the fracture area gave rise to nonuniform tensile load-

ing in Ihat area. This was a significant discovery because

it meant that all tensile values measured to date were actually

higher than a straight-forward calculation would indicate.

Additional tensile stresses on the outside surface of the

specimen and compressing stresses on the inside were caused

by the bending moment. This changed the stress ditribution

from that of pure tensile stress to that of a combined tensile

and bending stre-s.

Careful analysis of the stress pattern allowed the

calculation of this maximum, combined %tress, presented in the

36

Foi.pth Technical Note. Based on the assumption that the bending

stress varied uniformly across the section, a maximum combinedstress in Sperry Tensile Design No. 1 was calculated to be L33times the direct tensile stress. Similar values for the ASTM

standard specimen and ASTM's Design No. 2 were 1.09 and 1.18,

respectively.

It can be concluded that the ASTM specimens giveunreliable results when shoulder breaks occur because of the

stress concentrations in the shoulders. Design No. 1 is the

preferred specimen from this point of view. All three speci-mens showed undesirable bending effects which cause testedtensile specimens to fail at lower than true tensile valuesby the amounts calculated. It may be possible to eliminatethis effect by redesigning the specimen so that the centerof the seal area and the point of loading are colinear. Forthe present program, more realistic tensile strength valueswere obtained by multiplying the tested value by the calcul-ated factors.

d. One-Piece Test Specimens

For further evidence that the foregoing calculationsand ob.ervations were corrects the two new tensile specimens(figure 6a and 6c) were manufactured into one-piece testspecimens; tensile tecting would thus reveal the location andmagnitude of the fracture. As illustrated in figure 31,Sperry Tensile Design No. 1 fractured at the desired locationwhereas Design No. 2 did not. An average or four replicaspecimens of each body produced the average values shown on the

following page. It should be noted that these figures includethe correction factor due to tbe bending moment inducted bynonlinear loading.

37

Tensile Specimen Average PSI

Sperry Design No. 1 25,300

ASTM Design No. 2 16,000

e. Tensile Specimen Gasket Materials

A series of tensile tests was conducted to compare

soft lead and rubber gaskets for use between the shoulders of

tensile specimens and the steel pulling grips. No differences

in tensile valves were found; thereeore, rubber gaskets were

used because of the greater simplicity of assembly.

7. PHASE V - TEMPERATURE INVESTIGATION

Because of the vast quantity of work necessary to

complete Phases II and IV, ti.. metallizing and testing in-

vestigations, time and funds were not availabla to conduct a

ma.ior effort in this phase. This is a recommended area forfuture study.

8. PHASE VI - STRESS INVESTIGATION

a. Introduction

The importance of stresses in ceramic-to-m'tal sealsis generally acknowJedged by all who have become involved in

seal design, manufacture, or application. Distinctly different

characteristics prevail for electrical ceramici and the -tde

variety of metals to which they may be sealed. Because of

these differences, in particular their thermal expansion and

ductility properties, substantial stress is known to exist

when a brazed seal is cooled to room temperature. Althoughdesign techniques are known to circumvent these residua2

stresses, virtually nothing of a quantitative nature is known.

38

Considerable work has been done on stresses present in glass-to-metal seals, enamel coatings, etc. However, no publishedpapers could be found regarding stresses 4a ceramic-to-metalseals.

It was the goal of this phase to discover more aboutthe nature of seal stresses. In particular, it was hoped tkatseal stresses could be calculated, knowing enough about thematerials involved. This goal has been realized. More

basically, it appears that streis onstants can be calcu-lated, not only for ceramic-to-metal seals, but also for any

two, brazed materials having different properties.

b. Theory

Consider an infinitely small seal element as shown

in figure 12a prior to brazing. Imagine the top cube to bethe higher expanding member and 3 to be the total expansion

mismatch in inches per inch. Assume, for the initial argu-ments, that tha bottom memb6r is totally unyielding and doesnot expand. When the system is heated, the configurationshown in figure 12b is realized. The seal is thus achievedand the system is allowed to cool, resulting in the con-fig, ration shown in figure 12c. Considering a single planein the element, illustrated by figure 12d, the distance 8is also equal to the total strain developed in the coolingprocess; this is a key point in the argument.

The acceptance of the preceding argument, equatingthermal expansion mismatch to total strain, provides the first

means for the calculation of seal stresses.

Imagine that the thermal expansion mismatch betweenthe top and, for the moment, the nonexpanding bottom member,as shown in figure 13a, is the distance 8 in inches per inch

39

at the temperature at which the solder melts. Because S isalso equal to total strain, it can be transposed to a stress-strain curve as shown in figure 13b. The stre:s level can thenbe read directly from th3 ordinate. If the stress generatedexceeds the yield point of the metalg the procedure does notchange, but will result in a transposition such as figure 13cillustrates.

A correct stress determination at the seal interfacewould result from the foregoing procedures if (l) the bottommember were nonexpanding and nonyieding and (2) a singlestress-strain curve were operative at all temperatures.Neither is the case.

The acttal, final configuration of the element inquestion would be as shown in figure 14a. The top memberwould be pulled into tension and the bottom into compression.Defining tension as positive strain and compression as nega-tive strain, both can be plotted using the same ordinatel asshown in figure 14b. Thermal expansion mismatch can then bemeasured as shown in figure 14c and transposed to produce asomewhat lower stress, for a fixcd Sy than would be the caseneglecting the lower expanding member.

The problem of changes in the btress-strain curvewith changing temperature presents a somewhat more difficultsituation. A different curve would be operative at any tem-perature, and a solid surface could actually be generated forthe three-component system: stress-strain-temperature. Ifthe equation of such a surface coul. -a Aeterminedt a mathe-matical solution could be achieved ..... .L the stress-ratecurve. Such a procedure would be extremely difficult, prrb-ably resulting in a machine calculation. A simpler graphicalapproximation is preferred, as described in the followingparagraphs.

40

Assume that Lhe ceramic member expands, but that it

is totally nonyielding. Such an assumption could be circum-vented, as previously described, by plot; .'ng nega';Ive strain.

Even in actual measurements, however, the yield contributionof the ceramic can be shown to be so small that its neglect

produces virtually no error.

Consider a family of stress-strain curves, as shownin figure 15a, in which virtually no stress could be developedregardless of strain at temperatures down to 5000 and then

suddenly the room-temperature curve became operative. Theresult would be that only the strain or thermal mismatch

developing from 5000C to room temperature would be of im-portance, and the mismatch in this temperature range could betransposed, as previously discussed.

Next, consider a family of stress-strain curves in

which a measurable curve is operative from braze temperature to5000C, and then suddenly a second curve becomes operative forall temperatures between 5000C and room temperature. This sys-tem is shown in figure 1b. The transposition can then be

carried out by using the 500C to braze temperature thermal-expansion mismatch. If no further temperature drop were ex-

perienced, the stress level would be determined. At 4990C,however, th- second curve would become operative. Stress

level would not have changed substantially from 5000C to 499°C,only the curve which is operative; one would begin to accumu-

late stress on a new curve beginning from point A. From pointA, the thermal mismatch from 5000C to roo. tempexature isagain transposed to determine the final stress level. Thisprocedure can be applied to approximate the final stress level,providing the necessary curves are available. The accuracy

increases with the number of curves; in this study four curves

were found to be quito satisfactory.

41

One basic observation should be made at this point.A seal stress thus determined and existing at the seal inter-face is a constant value for any two matei-lals for a givenbraze temperature. It is not geometrically sensitive at theinterface, and operates equally in any direction in the sealplane. However, at any point above the interface, the stresswill be influenced by the geometry of the seal; factors suchas bending in butt seals may interrupt its normal development.In the experimental study of LhLsb stressesp samples wereused which developed stress uniformly about an axis. Theexperimental evidence in support of these theories demon-strates excellent agreement.

c. Experimental Procedure

To investigate the preceding theory required ex-perimental testing in two distinct areas. One was thedevelopment of stress-strain curves at elevated temperaturesfor the same metals, and at annealing temperatures used formxking seals. The second was the fabrication of seals inwhich the stress could be measured by strain gages. For thesecond area, the technique was to apply the gages to the metalmembir, remove the ceramic, and measure the metal relaxationor strain which took place. This provided sufficient data

to aaable a comparison of calculated stress with measured

The metals chosen for the study were 303 stainlesssteel and "A" grade nickel. The stainless steel %as selectedbecause of its high-temperature strength and nickel becauseof the reverse tendency. Copper solder was chosen because ofits ease in acid removal and generally common usage.

42

(1) Stress-Strain Measurements

Considerable data can be found in the literature re-garding the high-temperature characterist.cs oj. nickel andstairless steel. ThA data presented some difficulty in that itwas not available for the same annealing times and temperatureswhich would be experienced in brazing. In addition, heatingrates were different. For these reasons, precise measurementswere made after the metal was exposed to the brazing cycleused in making the seals.

The initial plan was to machine stress-strain speci-mens from the same tubular stock from which the seal metals

were to be fabricated. It became apparent, howeverl thatseveral problems were involved with this procedure. Fabrica-tion of the specimens provided the first difficulty. Thestock had to be cut and rolled flat, requiring high pressures.Because of spring-back, the pieces had a low residualcurvature. A similar shaped mold had to be formed to over-bend the pieces. Secondly, considerable cold working wasintroduced which could be removed only by undesirable anneal-ing. Finally, the samples thus produced showed poor stress-strain repeatability.

Because of these problems it was decided to machinespecimens from sheet stock, which, upon checking, was foundto have virtually the same hardness as the metal stock to beused for seals. Exact discussions of the hardness of themetal, the rolling and machining technique, and the specimendesign can be found in the Second and Third Technical Notes1 .

The specimens, after being passed through the brazingfurnace, were placed in a hiah-temperature extensometerl a-shown in figure 16. Preliminary test runs were made to testthe equipment, and the technique was refined.

43

(2) Seal Residual Stress Measurements

Because of ease of calculations and to achieve uni-form stress distribution, a test specimen with a design as

shown in figure 17 was chosen. The sxal.l shoulder on theinside diameter of the metal was provided for placenent ofsolder. A wall thickness of 0.100 inch was used for the

stainless seal and a wall thickne3s of 0.080 inch for ther ickel seal. The differences developed because of stockavailability. These can be easi.y taken care of mathematicallyas shown in paragraph 8z(2). Jamples were brazed using OFECcopper solder.

Five resistance-wire strain gages were bonded tothe outer circumference of the metal, three to measuretangential strain and two to measure axial strain, as shown

in figure 17. The sample was fitted with a collar equippedwith terminals. Lead wires from the gages were attached tothe terrinzals, which in turn were equipped with lead wiresto the measuring bridge cii cuit. An instrumented sample

is shown in fIgure 18.

Solder removal vas accomplished by leaching out thesolder with a 25-percent ammonium persulfate solution at 1200F.Using a jeweler's saw to remove the solder and direct axiaJ

loading to remove the ceramics were considered. Thesetechniques were abandoned, however, when early trials showedthe leaching technique to be the most promising.

To facilitatc exposure of the solder to the ammoniupersulfate without attacking the gages, a closed circulatorysystem was developed. Samples were provided with teflon-

gasketed aluminum caps through which the leaching compoundcould be circulated. A bank of four specimens was then set

44

up and ammonium persulfate was drawn through the system inse-ies(figure 19). It was found highly desirable to generatea mild negative pressure within the cirn'alatry system thusgreatly reducivg leaks. This was accomplished ty pullingrather than pushing the ammonium persulfate through the system.Gage readings were taken continuously during the leachingprocess. The solder was completely removed in about 10 hoursfor the nickel and in about 24 nours for the stainless steel.

(3) Calculations

In the previously discussed theory, a method forcalculating stresses at the seal interface was described.Experimental procedures allowed measuring the actual tangentialstress on the outer circumferential face oi the metal. Tocalculate the stress from the inner face to the outer face,formulas which are available in the literature can bA used

2

St r -r 2 r02

where

St = tangential unit stress at radius r

r = any radius

Pi= interna2 radial pressure

=o outside radius

ri = inside radius

For the special case where

S = tangential stress on outside surfacet0

St = tangential stress on inside surface

45

it can be shown that

. 2 S ri2St 00t r 2 + r i2

This expression can then be used to calculate theoutside tangential stress using the value predicted for insidetangential stress as discussed previously. A comparisonbetween measured and calculated stresses can then be made.

One final considerabion remains. Because the elastic

limit of the metal will have been exceeded when brazing,residual stresses can be expected in the metal ring despite

*the fact that the ceramic has been removed. This conditiondoes not invalidate the use of the above formula, because, insolder removal, unloading will then take place according toHcokels Law29 . Simply, more stress relief can be accomplishedby machining the inside diameter of the metal ring.

d. Experimental Results

The r'esults of the stress-strain measurements areshown in figures 20 and 21 for the nickel and 303 stainlessstee., respectively. Thermal expansion curves which were used

are illustrated in figure 22. Table 23 shows the results ofmeasured as well as calculated stress. Outside tangentialstress is designatec by the symbol S.o, and outside axial

stress by Sao. It should be noted that no means could befound to calculate Sao from the inside face to enable a com-parison with measured values. The inside tangential stressor brazing constant K, calculated as previously described,represents a stress in any direction within the seal plane.Although approximately 15 samples were made in each metal,

46

time was available to measure only 4 of each; these were chosen

at random. Sample dimensicns and physical constants used in

the calculations are as follows:

Young's modulus for nickel = 30 x 106 psi/in./in.

Young's modulus for 303 stainless steel = 29 x 106 psi/in./i

Ceramic 0.D. = 1.577 inches

Nickel O.D. before braze = 1.743 inches

Nickel O.D. after braze 1.765 inches

Stainless steel O.D. before braze = 1.783 inches

Stainless steel O.D. after bra-a = 1.803 inches

e. Summary of Stress Study

A study of table 22 shows excellent theoretical andexperimental agreement within the limits of the testing tech-niques. It appears that a stress constant does exist in brazingceramic-to-metal seals and that it can be used in calculatingseal stresses. The implications of these findings are note-worthy. Tables of stress constants could be generated forcombinations of materials at various temperatures. The firstmeans would thus be provided for the calculation of previouslyundeterninable stresses. Thusga method would be available toenable the choice o' low stress combinations as well as theab4lity of making use of prestressed geometries or structures.

47

SECTION C

CONCLUSIONS

9. CONCLUSIONS

The following can be drawn from the data presented:

a. Ceramic-to-metal seal strengths as high as 28, )0 psi intension were found for the 94-,jercent alumina body. The maxi-mum strengths decreased - as the alumina content of the ceramicmember increased - to 229000 psi in tension for the 96-percentalumina body and to 16,100 psi for the 99.6-percent body.Sealing techniques were developed which produce seals as

strong as the ceramic.

b. The optimum sintering temperatures vary with theparticular metallizing mixture under consideration; however,in most cases, these were in the 15000C to 16000C sintering-temperature range.

c. At tht lowest sintering temperatures (13)oeC and 14000C),the strongest seals resulted from straight molybdic trioxide

metallizing mixtures or molybdenun-mangangese base mixtureswiti additions of silica or titanlum helpful. Molybdenum-

manganese mixtures were less 1requently associated with high-strength beals at 15000C. Additions of titanium and ceria

to pure molybdenum were stronger after sintering at this

temperature. At highest sintering temperatures (16000C and

17000C), those additives which promoted molybdenum sinteringgenerally yielded the highest seal strengths.

d. Metallizing compositions for AL-23, a 99.6-percent

alumina ceramic, almost invariably required silica or silicate-

bearing minerals such as feldspar or talc to yield strong

ceramic-to-metal seals.

48

e. A photoelastic study of the tensile test specimens dis-

closed and allowed the calculation of bending moment forces

in the seal area due to nonlinear loadirn. In addition, the

stress concentrations which can cause shoulder breaks are now

well understood.

f. A tensile test design was developed in which shoulderbreaks, even in ultra-high-strength ceramic-to-metal seals,are no longer a problem. Only one shoulder break was en-

countered in over 400 fabricated and tested asemblies.

g. A basic investigation of ceramic-to-metal seal stress

developed a method for solder removal, thereby enabling

accurate stress measurements. Theoretical work indicatesthat a brazing stress constant exists which can be calculated

and used for the correct determination of seal stresses, not

only for ceramic-to-metal seals, but also for any two dis-

similar materials brazed together. Experimental data in

support of the brazing-constant theory demonstrated excellent

agreement.

h. The results of this investigation support, in general,

the Glass Migration Theory of adherence.

49

SECTION D

RECOMMENDATIONS

10. RECOMMENDATIONS FOR FUTURE WORK

Seal reliability usirg the improved sealing com-

positions du'e1.cyed under this contract, should be studied.This should include the importance of minor and major changesin processing variables including a wide range of factors t

such as furnace atmosphere, metallizing thickness, sinteringtime and temperature, plating thickness, and brazing pro-cedures. Criteria should be established to bracket the maxi-mum variation which can be tolerated in these areas. Workis also recommended to determine the compatibility of themetallizing procedures for tube usage. Factors such as out-gassing, cathode poisoning, life, thermal cyclingg and r-fcharacteristics require study.

Of equal importance is the study of leak path.The mechanism of leaks, their origin, and elimination require

immediate study and understanding.

Long-range studies are required in the areas ofceramic-to-metal seal adherence, particularly where 100-percent aluminas ar6 involved. When the sealing of materials

othar than alumina is considered, large savings in engineering,manpower, and money can be realized if complete understandingof adkerence mechanism is available.

Tho investigation of stresses should be continued.

With the mechanics - stress calculation now available, work

should be initiatea to determine the brazing constant for largenumbers of materials.

50

Additional work on the ceramic itself should be con-sider-d in long-range thinkirg. This should include the use

of beryllia ceramics. The creation of higher-strength, lower-

loss ceramics should be encouraged. Basic understanding of

electrical-loss characteristics in the various solid and

liquid phases in the ceramic is important.

In general, efforts should be made to keep the

development of ceramic and ceramic-sealing technology at the

same level as tube technology.

51

BIBLIOGRAPHY

1. Metal-to-Ceramic Seal Technology Study, Fii. through

Fourth Technical Notes, RADC-TN-59-370, RADC-IN-60-53,

RADC-TN-60-108, RADC-TN-60-192, furnished by Sperry

Gyroscope Company to U.S. Air Force under Contract No.

AF 30(602)2047.

2. Kohl, W., "Flectron Tubes Fkr Critical Environments",

WADC-TR 57-434, March 1958.

3. Van Houten, G., "A Survey of Cp"amic-to-Metal Bonding",American Ceramic Society Bulletin Vol. 38, June 1959,pp. 305-307.

4. Pulfrich, H., U.S. Patent 2,163,407 (1939).

5. Pulfrich, H., U.S. Patent 2,163,408 (1939).

6. Pulfrich; H-, U.S. Patent 2,163,409 (1939).

7. Pulfrich, H., U.S. Patent 2,163,410 (1939).

8. Jenkins, D., "Ceramic-to-Metal Sealing; Its Development

and Use in the American Radio Valve Industry", ElectronicEngineering, .ondon, Vol. 27, July 1944, pp. 390-394.

9. Pulfrich, H., U.S. Patent 2,174,390 (1940).

10. Pulfrich, H., German Patent 659,427 (1938).

11. Vatter, H., "On the History of Ceramic-to-Metal SealingTechniques", Vol. 4, February 1956.

52

12. Vatter, H., German Patent 645,871 (1935).

13. Vatter, H., German Patent 682,962 (1939).

14. Vatter, H., German Patent 689,504 (1938).

15. Vatter, H., German Patent 706,045 (1938).

16. Vatter, H., German Patent 720,064 (1936).

17. Pincus, A., "Metallographic Examination of Ceramic-to-Metal

Scals", J. Am. Ceram. So-., Vol. 36, May 1953, pp. 152-.158.

18. Pincus, A., "Mechanism of Ceramic-to-Metal Adherence",Ceramic Age, Marcri 1954, pp. 16-20, 30-32.

19. Armour Research Foundation, Illinois Institute of Technology,"Literature Review and Induc al Survey of Brazing", FinalReport, Project No. 90-1060B for Frankford Arsenal.

20. Gulbransen and Wysong, "Thin Oxide Films on Molybdenum",A.I.M.M.E. (Metals Tech) 14 T.P. No. 2226, 1-17, September

1947.

21. Dushman, S., Scientific Foundations of Vacuum Technig'ne,Now York, John Wiley & Sons, 1949.

22. Kubaschewski and Evans, Metallurgical Thermochemistry,Pergamon Press, 1953, Table E.

23. Coughlin, J.9 Contributions to the Data on TheoreticalMetallurgy (XII Heats and Free Energics of formation of

Inorganic Oxides), Bulletin 542, Bureau of Mines, pp. 23, 31.

24. Denton, E.P. and Rawson, H., "The Metallizing of HighAlumina Ceramics", Brit. Ceram. Soc. Trans., 1960, pp. 25-

37.

53

25. Cole, S.S. and Hynes, F.J., "Some Parameters AffectingCeramic-to-Metal Seal Strength of a High Alumina Body",

Bul. Am. Ceram. Soc., Vol. 37, No. 3. 1958, pp. 135-138.

26. American jociety for Mstals, Metals Handbookt ASM, 1948,

pp. 1146-1240.

27. Timoshenko, and MacCullough, Elements of Strength ofMaterials, New York, D. Van Nostrand Co., Third Edition,

June 1949, pp. 25-26.

28. Timoshenko, Strength of Materials, Part II, "AdvancedTheory and Problems", New York, D. Veui Nostrand, p. 211.

29. Ibid, p. 389.

54

£AFJQIZ

TAMU IAlMUA 0F ThlW41cca CALCULAI0

lL4 N&Itiz .-lwmtlo Tapwture tktJ . amd

P.".. Ift 49t N2 of metal (0c, of~4 (C) 0 AI20 (-Q

b ,4AIu. 650 1920 3Mw

c. Oidise fl 2570 1IO

cc amwiep um5 I AO 100

V4 Qxdlse 650 2800 1900

ft Msakow 327 no go a.teom

at Oxdils.- 1 171Z: Im

.4 OxIiJus 1.w 2000

11 OxiJ~sot 19 1300 1700

zr 0U.4. 1700 2=0 1700

Xn i i"S 126 178 is

1,e Dow point 1300 1 ,O13

TABLE 2krcUTI OF ADHMW E d TRC.4 A"!) rL$'IJE PEEL TOTJ PF"R'MI' ON JPE1N43Z NMIAL1Z A d,.-oaQ;MO&. RLM ON M; hUI4NA RUACTION THM-

A er nt Torqua. Poo b*.1 r4 * Pool8 + : i Vialue-

210 3o 1500 P F 0 0 3/4 235M, 130 p p p /i. 3/4011600 F W? 1/2 1/3 1/2 16 1600 7lv? 1 1!4 3/4 1/A10. 1700 Pr if P 1/2 1/2 532.0 1700 0 1W F 3/ /2 W2210M No u P P 1/2 3/4 0 25 N 1500 a Fp 1/2 3/402 I=0 7 1 F G 0 0 ' 1600 a 0a 51/2.31/ 1/-4260 1500 G G G 22 11 1/2 51 V42 1700 a, 312212t 2210 o 500 p p 1/0 3/ 10 11 10/L 3/

3, 1 0 p T 1 3/ 100 0 6 1 1/ 212 212

3 10 Pp P 2 ." 1/4 18 160,0 0 To0 . 4 1 1/2 1,11 6 0m 7 0 1 2 1 1 / 2 1 9 1 7 0 0 G 1 1 0 W 2 1 2 /27G 1500 l JI 0 " 1-500 2, lb 3/4 3/4 3/4150M i 1700 OP F 2 0 6 2n p70 /. 11/2 : 1/221 ,o 1400 P 7 3 25* /I/ 1500 G0 ? 3/ 3/41

61500 p o 0 ) 11 20 100 t 0 G G 3 1 3,2 1/419 3 Z102 1L60 P P P0 1 /41/4 60 L; 1700 4 0 17 3A/43 0255M 140 400 if 7 0 3 0 3/4. 25'j)b 1400 0 107 1/ 7/81 3/461' 00 P pG 3/43 y1221 1'O ~ 3/4 1/219.212 1600 T p 1 2 3,.1 J. 4.i 10 0 2 if 3 5/6, 5/621

' No 1500 P P 1 4/20 210 , 10 G 00 F 3/4 3/ 1/21600 P P 3/;- 1/2 3/4 22 1500 010 1 1 3/4 3/41.062,0 1700 F Pf P goI . 160c, C G 0 1 3/41 0210 No 1500 G P 11 1 0 210 1400 o o T 21/21 3A4

160 a F P y 3 / -iiF e 7 2 1/-5 5 1700 27 W I 4 3 1.'2 23 1500 0 1 2 1O102 1 1 1 142, 2 1600 a' a p 1 3/4 1S Mb 1500 1 3/4 1 20 255 1No 40 10 7 0? 7 21/21 11600 0 3/2, 1/2 24 1500 0 P7 1/ 3/41902, 10 0 1 3 12 1 /,0 1600 43 7/ 3/8

4 13 2 3/4IC 1600 !j if" 11/21 0 25 ill" a a 0/ 4 2 023.0 1700 10W 3 1/22 0 5/ I0 1100 0W 2 3/4 5/d20 No 00 70 ; 1 1/210 21 |M. 1400 a 07 -3/40 01600 a a 1, / 150c 0 c1 3/4 3/4 0La.10 0 G 1 1 2 1/2 314 1/4

2..5W 150 1 p 1 1/2L 210 X. 107 3/4 1/2

1 / ' /4

12 1600 3p 1 /4 27 1300 001 0 3/421/.1 3/4300o 1700 ; Priw 2212 1/2 14 907 133 1010 2 33/

255-, 1300 P 3/4 1/2 3/4 - -I13. 630.0 Coo W7 0 3/4 255 12b 0 W r 1/2 0 0-- 1500 G 707 F 1/4 1 21/2 90 1300 FQ700? 1/2 1/21 1/4

45?.e 1350 WW Jr P 3/h4 11, 3/425,1 b 1500 P P P 0 0 0160 P? " P7 1 3/0A41 Y34 ItO pP 2/ 1/20 210~ No 3 m p TP 0 0 0

29 1300 p Rp p 0 1/ 01255M Va 00 070? 1/2 3/40 129 N203 1350 PpF P 1/4 0 015 Ne IDU o; P 1 1/2 3/4 -

*P..per, 7-fair, 0..g

56

7Am... 2 (C nt.)RL3UL73 OF ADNME.E Ta~t AANao E ?I. jC &)-. atnWM L~xihLZ22J~CNP.~uLM~da~ ld ?8i AIIMINA Rk&ACT 0, THAL..!

6i0o 1.0 W 9 I . J 0 2 10 N 13 00 0or 2100 1 p pr 1 2 1 11 0 43 1/ 2 /31

2i . 0 P P 1/2 / 4 130 ) *Id

3/ 4 3 /b31 21 0. 3 L1 0 0 2 0F z1 I~ sw 0 U m p a 32 / 2 /48 2 1/50 FI16G p 0 C 1 /2 1 1 2~55 Pt 110 w m 3/ 1/2 3/b125 No 1300 a G " 4. 43. 34d 50 0 78 / /

321!0 a F sa 100 g g /b45 T.3?a 0 U /I 3'. 2~ N1 /b 1100 G p p 21i~ 21210 100 P pr :3b 3/ /NO 22?.

100 63;210 ) 1500 IF F oF ac 1 / ! /33IL40 G C / 3/4 1 / /A 0 /2.3 1/210 T0 10 G j? .. 1 2 1/ , 255 14M 1 3/4 3/41,00 U C / 71

30 a 1; / 3 /- 03-l'00 P7 C / 1 2

4 211/21/ 3/k 112 20, 100 0 0 7 /21

0* 13e Fd 1FF 1.1/ , 1M, 1.~ 3 1/2.'.21 21020 10 )L 0 N 3b 3'0 SIN2 1 . r0 7 i 3/I. /2 .34 371,, 1(". 0 212 11/.95 10 0G 0 3/1, 1 ~ 15 / /

3/4oa 2 .f.211)0-g30g

255 Y G yr 1/2 1 .57l

TAULE 2 (Cont.)RU LTS OF AEI{M E fl -TS AND '.WE r= , ?T. MP" W pII ON SPE'lqN YEJALLIZE dllT

CON-I'ON-Z USD ON THE ,LJKIW REAC'ON lJWW

1 1 0 0 7 4 112 3/4 15 ft V 0 To 10 314 3/4 3/4

Aa ) ] ro e o-. . 2 3 ; 3k zoK Tor0o Pee Adee. isiu Pll /

260No 1400 PC F F 17 /2 7/6 3/4 10. 150 0 8 E /2 1/405/

10 51 & 150 0 F 13/2 3/4 3/4 1 1 0 K1 100 P 7 1 1/ 1/2 1/12 1600 007 o 1/2 3/ 1 100 0 3 1/1/

3// 3/5240 lb 1. 0 Y 70 1 1/2 W / 4 - -0 0 1/3-/24M 3 2.,4100 0G 1/23/A 3/4 1500 Mi 7 /l2 7/8 3/4

S0 500 1 0 1/4 1 3/4 -- /-160 0 0?F 2 3 1 240o .300 o Ii 5/8 3/4 1/2

1501. . 400 0 11/4 3/411/4 45 F 1 1/0 1 22 3/4A 1 5 I . 3/4 1/2 1/2 1500 17 1/2 5/80 A O0 MN IQ 11/2 1 /2 -

- 240 140 0 M o 1/2 1/21 160M 1300 M i 3/4 2 1500 0 0 1 1/2 3/8 1/2

1 7 6 0 o. 3 / 1 6 0 7 1 1 1 3 / 4 / 4 1601. 21s 00 3/00 N/ 3/4----?

1500 17?? 1/2 5/8 1 1/2 340 130 11 1/4 1/4 1/4

240NO 1300 N 1 3/4 1/2 30. e 1400 "FT1 1/4 1/2 014 30 NN 1500 7177 1/2 1/2 1/2

30?To 1400 0o0 2 1/2 1 1/2 -1500 11717 /V8 1/4 1/2 240 W 125C P1 0 1/2 1/2

1- - -- 6 1300 1t7171 1/2 0 02401No 1.300 10 71 2 3/4 3/4 6o01e 1350 IF7IF11 3/4*0 1/4

149 45 N'. 1500 1' P 1/2 1/2 1/2151ft 140 a0MM 3/4 3/4 1/2 - -~-

1500 a0 IT 7 3/4 1/4 1/2 2401.o 1400200 1500a M F

60o 160

-FPlw, F-fair l -go

"Torive in losl-poeds

58

M=T Or AM==1 T~rj AND T~t PL.Z TE1YM F1MM4W 00^~WALL21ZD UM1 CQU00flI1S &%:;= On TIM MW3XOWJE 01 ?1

Am Tghg'. Pwlme Tr, ~

50 ob 1 500 P P p 31 0 / L50 V 1500 a r F P1551604 0 NWPC / 3/4 b. 1600 P 3/ 1/ 1/91~0 woo 0 Pr 3/. 0 91b0 1701) 6 1 1 2 3/4 1/2

100 loc. 1112 ZO

150b 10 PP P 3/4- ki J/A 130 No 1900 a a P 1 1/2. 3/1b

156 1- :t 0 160 a/ 1 M,' L 1//2110 _____

116.03 . 00 NN N0 11/2 11/2 3/i. f91bl 1 G P IF 3/4. 3. /

13b 1500 06 7/6 a 130 110 1 PPh I 1/' 12115 160 a NP 2 7/8 0/ .3 11010 100 7 O 1 1/1 1/ 3/41b 3 160 11 NCO 162 G0a 1A 12 31N

130MGO 164W I ry 1/2 j/6 2 66 o; 16M 10 H? w I/t I /-wV0 15V No 12" r Pr W 1/2 0 /I

4lM1 1500 PP 0 1/6k p23P1 A 30W W 12 12k3 1~ toP 0 1/4 /11/4 1500 a IF F / ld 3/1-

5 11506 No P5 P P P 9/ /010M 160 U PS 1/21 2

41 bo I r " 6 1W1 1/ I/ 3/2

3w M I - 1300 PP P 2 1/2. 1 1.1150 M 1500 1 NP p 3/0 1W) 16.00 P P p /6 3A6 3/1.

i~o a tt, 5 0 6 1. 1/ 3 A170D 0 Nr p / !/" 1300 p 1 1/2 3141 P 1/ /.171100 W 'a 1 1 / 1

TW3ZIS~LA LM 1U

kft'Lrial sa~w" motsiall arm"

TLo L3 U 2 0" UaI3[ o C.O02 U22102 U.2T103

Utku Cwrbomto

*Pri.w -ohitttvy bl

59

YAN1 5

RESULT Or AMOEC TEST AMl T022M lPn. 1531 PU122I0M ON3MWALLMD Wrfli CWW00TIW SAME ON 17ME GZASS MMIE.RA2 THM

I1I I

hernw Torq.. r.. Mh zOSlc Torq. fte5

US .a - J"W.55 b 2500 a01 I 212 / 285 N 500 T o e 3/2 3/4 7/8

65 2600 001a 2* 4 1 79 b 0 10r 1 1/ 1/2 1/4*7.5 T1 117M G a YO I 1 26. 0. Al.. M~o 4 1 07.511~~~~ ~ ~~ 130 "0 0 ___I2.6C 3 1700 02W 4 2

2d5b No 500 0 G0f1 2 1 I./As 3/2*' 2v %, 500 0 01F 121/2 3/*3/4.'66 1 0 G O F 1/2 2 1 v 7 P 3/ 1/20

2.5 T 1700 a001 9 3/4j 53.2 K203 700 01071 21/2 3/4 o

M ob 250010 a1 2 3/* 3/2* 299S5 b !SO00 001p 3P,11,/2 3/267 1600 000 3 2 3/0 0 01 3 3/2 1/*

30 1700 0 R 4 5 1/ .4 B 1 00MT 2 1 11,

292.5 ) .1500 P ae 2 1/2 1/2 2d5 M. 5 P 3/* 1/2 3/2a 1600 0 2 1/2 6 3/ *62 160 N N 2 1/21/4

9.2 C40217 Io a0 7 4 1/, 1.640 70 0 1 1.1/S 1 1/

2M b 2.0 0 . 3/ 3/2* 270N. 1.500 111 5/ 1/2j11/269 2 a00 001 2 2* 1 . 63 o0 1 /4. 1/2, 1/2

19.-C.0 2 1700 0 G /2 5 33.650 700 P 10 3/J.

27o No 1 50 0 1 11/2 3/* 3/* 240 No 1500 N1T 1/2 1/2 3/TO 1600 P a 1 2 21/1 * 1600 01 7 3/ ,/2 1/4

36.6012 17D 00 1 67.2 NO G a F 3/41 1/2

24O b 1500 0 1 11 2 1/2 1 1/2 29.5 No 1500 1 0W 1/2 1/2 1/271 1600 00 a 0 ~ 2 1 85 2±00 P31W p 1 5 1/2

60T1 100 0 T 7 1 112 3/4 ,*333 3 1700 0 r 1 . 1

240 OD 1500 001 2 2 3/4* 2961VA 150 1 1 1/2 1/2 3/1672 1600 a a 2 2 1/2"e 60 a111 31/2 3/* 1/

73.6 602 1700 01 r y 4 3 /1 1/4 21 N333 1700 0101P 3 3A, 1/.

292.5b N 1500 PC 1 1/2 3/* 1/2 5me 1500 1 1 0 1/2 1/2 1/273 2,um0 a P• 2 3/* 518 67 2600 I"O 1 1/ 1/2*

17.366 02001700 0 1 5 3/4 3/4 86 1203 7W0 P y 2 3/1 /2296 b 150 11r1 3/1 3/4 1/2 2h6mN 130D0 Y 3 3/ 2

74 = 3 1 1 2 11/2 1/4 3.75 lthlua8.6 %W03 1700 0 PC011 . 1/4I 88 mNe..

0 - - s 11*00 111 1 3/* 1/2 0a05 W. 1500 10 • P 1 1/2 7/6 2500 0 7/8 T/ 5/8

75 600 0 10" 3 1 1/4* -."F . . . .32*6 X200317700 011 .- 1.1/2 3/410 205b 10 1 /*

- . -- - T.5 Llthlm71" 1500 0 0 1 3/* 1 5/6 891 N-

7 160D 0 01 0 3/1 5/8 1/2 s" Iw e WWW 0 1/4 1/269.2 M 30170 a 7011 1 1 3/4. 150 a a0 3/h 5O 1/2

292.51b 1500 G 1 3/4 3/1*0 285 lb 130C 01 1 1/2 3 1/477 1600 0 10 3i4 1/2 2/ 15 Ltbim

13.3 K03 170 0 0) 2 3/4 1/490 WbmeF /1 58 /296m 1500 0 1 1 3/A 5/8 1500 0 1 a7 1. 3/ 3/4

6600 0 1 0 3/1b 11/* 1/1l6.7 Im a a0 .3 710 0

O-gm6

60

TABUZ 3 (Coat.)YZMMC AMMZC 1291 AM TCRWE PZP2 ?3MS P'DlRMVJ) 00ZEIMMUJZU2 WM~ COOIN' DA=~ OW TIM raAS3 NIGATI1M r.5U

-dsa Tor__ -w A___e _ Tq-P

*m W. 130D 7 F7 3V8 1/2 1./2 240 No 1300 r F 7 3/4 I2? 1/230 LiU.1u (10 tl

VALL. 1%00 10 F I 1 1/4 3/1. 311- mt- 1400 WI s PC 1 3/4, 3/41500 M In Pr 2 5/6 5/d .11500 a07 2 1/2 3/41

9 2 6 0 L i t i n a 9 3 .7 5 & 1 t h i s 7 0

nam* 1400 TO ? N 3/4 5/ /h nat. 140 r7 1 0 0

29 D 1300 p0011 7/d j 3/- 1450 o ~ 0 077 32/h 0/' 0

3.75 1th1;& 99~ 7.5 Liud 140 pr r 3 1/4 1/4

mat 140 PCr F 56- t 1500 7 7 7 A/ /. 112

1500 a PC 1 3/4 3/'. 2U5 - 1330 T771PI 3/41 1/20

2 2.5 ND 1300 7 7 F 1 1/4 1 Li4 100 Catb..*)Llthlmi m5t. 1Am0 31 7 1 1120 0

94 Tlts- 1500 r r vp 3/4 V/4 3/4bat. 11.00 707T7 3 1 !/4 3/4 -

500 0 1777P 3/41 1/2 13/4 270 w 1300 F7 if 1/2 1/20

26 N 1300 F 7 F 2 3/-. 1/4 101 Catt.--15 Llthi, os. 1l.O r r vp 3/ 0 095 !it - 151 a 7 if 77 3p'. 3/4 13/A

saL. 11.00 1077 1 1/2 1 3/4 -1500 7 77 1,2 112 7/:) 240 14 1300 F Y Vf2.2 3/1-0

- 'A___ L1 - i-~2m ob 1300 7 p r /I 1 3/4 102 Cart-30 14101,. rat, 1400 if prf 1/4 0 0

9b Tt- 1500 if 17 7 3/4 314 3/4-Date 11.0 N No P0 3 3 3/4

1500 0 YO1 311b 1/2 3/4-

61

TMU. 6RM O A3M~ TUSS AND TOR=0 PM. T="S r--VM ONW

=LuAU. w173 cw08ITC7 v36E 03 rm3 h12K-Nm snnUDM 7310

- bmhm~n. Tosia. N. dh-tl" Iatlgo V~

3w 50 PP Pr P 10 jO 0 7m o~ 1500 P P P 1/2 1/2 I'll1300 MM??r p 3/4 1/2 0 115 1300 F 14, M 1/1. 1/2 1/,107 !350 PIP???F 3/4 1/4.1/6 303 170 Pl p? 5/y p I

150010 aP 7/06 5/80 3w No 1500 0 FP 3/I. 3/4 1 41600 !'aW 6, 11/2 1/6,116 JA e 0G Pr5 1/12110.) co 10 GOP 3 3F5

3)N 1500 0? 7 P 3/4 '0 - - --108 1600 a10 YO6F1/21 2/ 97 X 15 0 OP P 1/2 1/2 1/2

O.9 f 17W0 GO GM 5 3 1A117 160 0 yupr 3 1/2-2WND 1500 0 1PP 7/d 7/8 0 cc-M r mP ,- -/ 1

109 160 M10 641/222 1/ 291 1500 PP P 1/2 1/2 7/83P ye o 200 G a /. 4 1 / 1ie 16. 0 pr p 5 1/2 1/4

- - 9o 1700 PC 7N 11/A 3/.1.291 O 15 0 OP P 5/8 1/2 1/2 - -

110 16'..) OP PP 6 1 1/2 1/4 204 150 0? /2 / 3/Pm 1.000 322 1/4119 N 1600 0 0M 1/2 5/8 3/4

- -e -W -YO 30' 3 /I 1600 G WPr / 1/ 2270 W0 15 0 OP P 4 1/2 1/......0.... I - I 11 t~

11,1 al OP Pt 7 3/1, 1/6 30041 No 3 0 is? 3/6 5/8 3/630ft 1700 G P 1 1/4 1 1/4 10 11600 0M 6y" .1/2 11/2 1/6

-- 0.9 w 1700 G PPM A 1 1/30 N 1500 aP P 7V8 J/'. 1/-

112 1bo 0 10 PP 3 1/2 3 1/2 1/4 297 N 1500 mP P 1 3/6 1/0.931I 170 010 PC .0 1 0 121 1600 0 YOPI 5 1/2641/4 /

M ft Iwo a " 11 1/2 1/2 0 3 I' rI o 1 0113 1o 0M 1Fp 1/. 1/6 29 N 100 mvp pP 3/4 5/8 3/A

31 NI P0 PIP 1 0 0 1.22 1600 a0m0W 8 2 1/2

29106 150 a P P 1/2 1/2 - - - -1700 3V/41lk 1600 1m Py 1 1/4 1/" 270 M 1500 101??F 3/6, 5/8 3/A

9321 10 OWf 3/1 1/20 123 1600 o0 6 i 2 1/2 1/230 w 1700 0 PPM 7 3/4 3/4

62

7202SO W AUE T= AMD TQR 7. MA M OPZ2 MSCUMIDAULIN5 C O7TIONS 008 TI2 GLAU ADMITSME

1 ein. 1ous aume. Value-

270 No 1.4O F 11 Pr 1/P Ti 17 2;1 No 14A0 0 W i P 1/2 0 1/2105 LW0 a a 0 3/. 3' 5/8 1 )AMuun

30 oftfpr 160 0 0 F A. 2 J 155 Bill .. * 1I / ,

285 w 14 v rc Pr Prw 7/8 - / 5W. 70W 5/8 1/2 1/2106 15w a r 7? J/4 5/8 12-

15 Fldspr low 070 3o r 5/d 2W3 230 WWO ifPrP 1/2 1/A. "'/224 o ~ 1300 m0 70 r 3/A. 3/. . 156 172321.3 0ft. class 7- 0 1 I 1/A. 14 V

1.1 0 ro 1500 7 a 7 V/' 1/21 1/212 $102

Coo LOW0 or? r 1 7/6 VS8128, Ib 1300 F Pr P 3/6.0 1/2

200 No 13.0 m7r7r 3/6. 3/. A 15V0 - -?ho3 .f 05 No 1300 7 Wif 1/4.0 0

125 101 n125 1,? 1400 7 Wi? 1 0 0

22 1150 m 010F 3/A. 1/2 1/I.1500 a0r 3/. 3V. 1/2 2853 13o 7ic r Wm 3/A. 1/2 C

- - -1 2 15 A*203 4 lIAC 1W 5/d 02g13 153W0 00a T/8 3/A. 5/3 1500 WW? 5 /8 1/I0

11.1 & k, IA .4 r 31/2 1/2 -9 Tale 1737W 3 5 V 281 No 1400 PWiNP 13/. 1/2' 1/2

-.-- - 197 1500 r ? ",/F 1/2 1/22-.0 Ob 13X2 WWI 0 a2 0 15810, 16W 00? 6 5

30 A (N

20 Ta4il. 1.-00 a. 1 1 1/2 1/. JAl1!,. rF r F / 1/2

63

?AU I128i0MCs AM T AM TMM TS18 PV40

uCin DUALLW 17 ONPIIW0M

Ad4 - o Torq IJ eue PealAevo w at -

It1 V.1IF

2m0 v 3001 if P? P 1/'. o 04 N'oU 1400 0 10 r 3 3/. 1*5t P p P P 0 0 1/2j V 150 . a " 1 1/. 5/ 1/2

W2 F6a 1700 F M P 3/4, 0 o 9 Tale 1600 3 3 F 3 /1I l/l

'TT 310 ,M. 14.2 15 F. 150 ro F PIP lt 3/ll 1/2

31, Co 1400 FG ? P i ,, V 1/$. 9 Tole 1600 0 0 N/? 1 2 2 2/1- Fit 1/.o 7/0 1 o f 3/ / 1/2!

13' .I9 me i- " Ci - 1/2

.40 HD 15 00 1P 3 112 5/9 J1 / 2 1.0 /I1 2No 1400 0 7 1/ 1 1 / P m ff 5 sI0 2

-~o a____ 7 -y - 2 . Te. 1M N 5 o-11

2W0No 140 ao ? 1 ,/9 m/ nim/ 9 a300 r 2 3/ 3/ 4 /2

9 Tt 1600 0 aF 3 1 1 / .1 iiop-'. IC 2 3 /

atiC 1700 0 F 1 //. 5//1 1/- -1.30 ?- 225 10 0 21/2 1/2 5/8

131 0 1W 0 7~ fi 3/'. I/ 1/a, a0g 150 0 1/2 5. /8 I/A

21 T1 1500 i if P V 5/8 1/ . 0 iN 10 C 0 i 311/ 1/ 3/

35.58.0 1i I FV I if /. 3/.. I 1 / - ............. ).J -tn o e -l l

L- - -a -AO 0~0 123 FW if 1120 1/' N31.I 1.0 6. mento-!i tl

2.0 No 1500 0 1 11/21 7/ 8 15 7 100 . 1300 a0?? 3 /'l 1/'1 6. TIC 1500 ? 3/8 0/8 5/8 1 1V0 No 1330 P 0W 3/'. 1/2 1/'

20 PA 1(0 0 ?F *1 J/42 L12 5 u- L00 F P 02 0

2'.O i '. 0 701070 2/22 3/'. I00 1230 if if? 3/60l ,/81135 30 N 1500 000 ?1 r8 0/0.0 1 AO 1 30 0 Wo 1/212 1/2 1/.3075Or 14.1n 0010P 3/23 s /, I5 7 1 /. 36

- o _____o o o ] 1:) ,l ft20 lo l ifo , a~ a/ ./

137 301 150 0 ' 3 5/ 3 5 l /81 100 0e 1300 1 1 7 3/ 1/2 1/3102 1600 00 1/2421/21/'. 3 1500 P 3 ' 0 /

2- 0 No 1 O F0 F0 2 112 2 31 o IM PN 0

13.. 30 If 1500 00a0 51' /8 7 /8 la650Coo, 1600 001 N/2 1/2 1/)-

30 T1 1W0 0 0 o ' .1/ 2 3/ 1 5 6 2 0. 3 150 0 i /8 12 1/62-O -1 -4 a 1 00 11 /1 3/.20" 13 1 /8 3/'.o - - r - -

1 3 30?1 l 1500 .60 Cm7! 1 1600 0 0 10 / 8 1 11/2

41' 01... 1500 a0? a .l221/ 2 1/ 1/8 50 ft2Q 135 r Vf PC 1/2 1/ 1/215000 - _ - -0 0

19.2 CoM M6OO 10 70 0 M 1/2 1/2 L/ '.3, 10 0 0 a /24 4 m 102 116.00 0 0 Pr 5/ 6 4 1/10

.86 i1'. 3 100 0 0 0 a/8 1/2 5 1/2Soollad~~!Io 15 M F 31b 3 34WW o to 7 lil

1'-p .Y-ai4 -wpa

6~4

FSKMOr AUMSW TSMAMA ( Mcentnw

SoI)M HULUZO WMs 0OrA CG0SITIM

240 V0s 16.0 POO a V4 3/4 1/3 135 ft i400 a a P 1 l,' it/

34011 1".X iff i 3/4 1/2 '65Zircon 1300 000 1/4, 1/4, 1/

167 *1.11o 66'l.1t. 10 iff py 3/1 3/4 3/14 12 Ee.1Sn4 i~liil IW0O if M P 3/. 3/1. 3/4 1791 isf21 0'al~

^0 W03 1400 o pa r 3/4 3/4 w '4la1300 0 0 a 1/2 1 1/2 3/4, oust.

60304, 100 a00 1 3/4 1/2 05000 1WO0a800 3/4 3/41 1/4in L~ 10 - / 1w0 P PPF 0 - 0 10

LO9 L20aS. 340 Carsiang 1300 0 0 1 /b 1/.I 1/1,603 1Lbw 0 NCO 1 1/2 3/41La 7052

- - 1500 0; 0 a 3/4 5/8 3/4 -~ 201- - 153. 10 0 / t!4 I/;.

60 ~ 1300 1 F Wf I/'. 3/4 V.1500 F F p 0 0 0

.7Q 3 CA 240 Cy' 1400 N /a I/A 1/15 No 1400 if if if 3/4 3/4 L/2 60 0'N130

1500 if Pr 1/2 112, 3/4 11 S..1Z.4 1501M f N I/3 / /

60 3t 1300 ifilf / /. 010 16W.Pr PrF IaIF

171 3 Cu .24 M1 1250 0 0 0 0 0 1/2I Co 1140 Py iff 1/4 3/8 1/4 60 O'somel130U 0 1 /.40 1/11500 pP p v L/2 5/0 5/816 I 56.11.4 115 0 81a 0 0

-~ W3. 1 f0 p a 5/8 1!!! 1/260V 1300 Mfi ifP 1/4 3/h L/2 50

172 3 CA L20~is 0: )O 6 0 0 1/b.

1500 p p p 0 1/12 0 6 sin~- - 13 01...

60o 130yn pr rV 1/b -A 3/1, #M003ff?. w ?. 1400 NO N 3/a 5/8 5/8

173 3 O 1500 a 0, 0 6 /8 112 5/815s 1400 vb O P 0 1/" 3/4 1/ -..---------..-.--

1500 PPF P 1/2 1/2 1/3 6r, AL~o 10 PPN /0 L/240I.130 QP /

60 me 10 Pr if if 1/ 1/4 0

174 3. CA 0O 1.0 P N Nl 1/1. 5/6 5/8140 00?, 1/& 112 1/4 10 PP F 3/b 3/4 11=00 r-V? 3A1 1/21 1/- 12-A

29 L- I-----. 13 ,owl 13W 080 0 0 u

175 3Cm 4 (I0 0 PU 1 1/4 1 /.- 7/ 8 Lo15 L1O 1600 0 0 If - '31/2 1/4 h 1.

1I"3 no 0 IWO a Pa 01I1/4 3/1 1 /3 6Dc 1400 080 0 0 0179 ;1 - 1500 P'aUp 11 5/.8 ./ 150C000 3/4 5/ 5/8

" L1,00 1600 0 a V 2

*FP..w, -fair, 0goq..

65

TAILI 8 (Coot.)

M~ O AnN ?T= M TONQJE&. F .tl I ER1MW ONSPKD NTALLIZE WM~ MMhE CIX'02.1T0IM

A hohe. T'.rqw P0.1 Adha Torqo. Nfj

9_ I _ _A1203 130 0 LA/I 1/'. 1/4. 1 l.par 13030 0 G G / 3/1, 3/.go O'suI 3m 130 '90 m o 14W0 yo00 c 1/ I' 3/81IN Se.ling 1500 0 0 0 / 112 0

01..s 150 000 15/8 15/8 3/'.

iz) KO IM0 0 C, 0 2 3 1 1/. o1# 100 P 7 Pf 1/2 0 0111 i 1'.00 rPP r?

30 Al~' 1300 F? F P 3/- 112 3/'. 150C 0003 1/'. 1/. 1/4.181 seli.n {~ l,08 1.2.' P P 0 0 0 1/12

01*.. 1500 a0100 3/ 8 J 192 1300 OP pp 0 1 1/4.#" I..: 1350 P P p 1/'. 112 1 i1

2.0 No 130W 0 P01W 3A I 1500 00a0 1/4 1 1/'. 1/'.

180 F.14cpu, 1330 YOPCa 003/ 3/6 -.0 M 1300 a a a 3/1- 3A) 3/.1500 00a0 1/20o 1/'. 193 150 No 1.00 7010a0 3/. 3/. 3/'.

18 TIAW 30 G G i2 /8 3/ 1- 1500 006G . 3/. 5/8 1/2

1891207.e 1-00 70 P 1/? 1/2 1/'. 10 No 1250 PP P 1 0 01500 0 00 0 0 1/. 14 125 Y. 1300 WWW1 1 /2 0

15 C' 1350 pr PP J7 0/.j i/21500 pP P 3/'. 1A'.0

-ftm to too&.p

66

TAME~ 9

R4OSILTOf 01COK~tL-.10 ?T2TS 7lJJk dALLLZM 01202

SLUT=2~ AT I )0.3C WIM3 L4PEMkMlT.L HNrA2LIZING NwIXyRL

Cmp i~ti. Cmpraa~ion L ad (lb) B. cmw.iation Cum*taaJ. ad (lb)

deight (1p) AD-94 AD-6 23 1 d d.idh (P.) AD-%4 9," L23

V7 2L0 Ho 1501) 95 285 4o 9009) t 1.100 15 Litimt 6.21

033)'

48 21 o230C 2h.A N 096.5 .1.02 10 j ISp.

49 165 700 2w lw 1700

26 " 15 va1733-

5C .255 16 .1) ZW 150 192, Mo 22D48 ;; ) 105 Mn 30

22 Y- .~ 15 re3035' 3-Q.2

68 ,46 )k. 199 300 MWa0 *30 28003.75 Lith!- 5.20 130

H1.Ui2t& I&;,=T0!.

No~m

67

TAULE 10AxsULs W cm1Usa0W ThFT2 Vm mzrALwtX DIMS

£LNTMW AT 1AO0 USING WftOTA. IgRTALLIAZING NIMMIS

-igb (p) AD% A-96 L.-' 4pght(RD AD.ft AD-%6 IA-23

2,1 M5, i"4, 4 N 3100"n.!uo 48 22

tt 30000

23 2103k No*1) 1700 50 55N m I2U 2y"0142 mNn. ZY00 lax0 45 S102 2120 23.X

24. 25' No *Z. 1 2O 4360 i24. 150 pk 3C50 2!'00a1 WAO- 2400 NX 9. -1 220 1200

I~~ 126C0

25 25 NO IA 161V0 7A 2 l.b 37058 22W 1"0 3.73 LitImI. 2500

-* 'e. Xahpnto .

35 Z4u W 500 9 1292.5K 2455c 3.0 300 7.5 1"~ma 2096.5 3LC2 Uwiamato ju

433 210C

36 25K 1500 500 93 2% No 170025 3.0 W, 19c1 .7 Lumim am0

38 Z15 lk 1 13-ii, 1600 94 -92.5 P. 230D 00031 C&0 15W3 1.W 00 C 7.5 bLihm 2=.0 500

1500- 1233- 1 e3

43 21.3V. 2&X0 2W, I'= 95 285 M. 2100r, P.1 ...* 2300 370u 35X2 1) LIt~lmi 2320

44 Z011M 32012 96 V7) P 10 045 We..~~ 4X40 r, Llt~mm 2100 1.20

T ltamtw

45 21.1 P.Sc. 99 .)2 5 w 15009T4.! 260 7.5 LitkJfli 1500

Zarbomte

47 '~ -1'0:, 1-, 2851 11SW158 M__i5 .. I_

'A 'I4rl

68

SINMR= AT IWO-C 05IW, L:PEPrmITM. N.TAU.IZ:JG MtfTUKZ

Ii Com".1Uon C~sWab~i L..S (lb) " iua Cr w%..fr LoW (1b;

.07 IX N 60 140 LW N 2;.w

1.20 240b No0" MV A51 mn a F. 20

L29 240 No 26W 141 ZNV,51 ft 5.0t 12M9 TI Koi

11) 20 ILM, l1lL IM 7M WlaiO

12 ZrO 0 2 2 man.tot

13x aW0 o 3400 AM2 US 4 ZOb 242b w 400 22DO 51 N Z9 TI Am 7 1 6 kbnt.

13 40 NoI 1200 US 24C N 120030 M 1600 30~ 110

135 Z40 No 000 I=2 LIS) Z5 N 250030 Nn 4W0 240 60 f 4900

Ral '333-L36 240 . 2200 164 24D WO Ad

)0? 3300 60 M%02 AwA

'33P.1.)7 210 N. am5 2601 178 ISO0? 21W2

30 Ph 3eZ 2500 190 3)2Mo

A : - __ *,I ft-fl

69

ThA U

RIULTS 0F CUWS.;J1 T371 MM~ MZALLI13 DVIC8

sIXTEW AT WIXU.V WI! nPiALMMAL meAULJZZ00 MIrjJ2~

Coapotio Cmp.sion LoWA (lb) oo.oause W On Lo#A (lb)

ga1gt (p) AD-9i. Af-6 A23 60*t (a) F D' -- %9 *-3

2 10 f 24W0 2100 2100 25 255Mft 2500 270 SODLa coo 2500 2100 2100 56 VA 5=0 llow 700

Z"~ 2100 am 2300 1500 1(02 2100D 210080

ii No10 9004 210 H. WOl 90 -.1 I=0

1%, lNgO 400 35002200 1000

3. 251 NO 3WO6 255N I.X 451 3Ti 3100

9..5 1.10 0 330011.00

0 10KO ph 12. 00 500 1600-)2.5 ?Io, -.,N0 9W. 310.. 600 1=0

;G 'Wo 14M0

h$3 210K ND00O 130011 210 &a 290' 90 ?.14aP, 1600 1i.00

121 Zr0. 1oO 42 15000.120W3 -9 20

44T 102 Irm.~. 255K 1"~ 600

13 2!5 No 1751 45 I=~ipm63 C.0 1dw K20 1600

1700

A.25K >50002! 2-0M. J 210=2 3500

270C

Ts.eotad ror beaU. t. itt~w

70

TAKE 11 (Cont.)

MRi=TS 0F CO4U100 TESTS 1?M KMILL1ZE MXW-SITSM AT~ 13LXA- UICt UiflWIAL XXTAU2.Z)W MIXTIMM.

Lht.) -~i 096 4.O La., (eP) ADl.94 4fl-96 AL-23

O 25M 1W 2_00~ 68 02. No0 7004SSO2ioo 230 9200

20 O300T 70 MC . 3000

51 ZaO No 25I 368300, )O96.5 sikO 1000936 Z.70I .04

56 150 NO 24110 .2300 '@ 6 T

91 2300 2300 1700

58 15O No 2.00 >> .T 47=91 "40 IWO m

193 sl 5 212.5 ft 1100

62 M1no16091 Mo ~ 2500

1.2. ZrO2 A1 M4 lb >5"0

)o UP1ISA > 500065 292.514 He 5G 2700

> 60 U2.103 i6b 225Me > 5Q2 2100 3133?

15 r.. > 5co zO 3 240 NO 40

>rwo 12Z'2 00

30 50 135 2.0 b >5=0 1600

137 -0G MD O oo3c m 3000 220

71

TABU: 11 (rant.)r1LTs or l~Pf~ow IsTs PIM MALUJZF DI9cs

3Muin AT 1 100 c mwI tmmmU "?WU13l WMII

CpO.itioa C-upiw~.L~n 1-4 (.b)P A.Lbt (4p) AD) AD-9b AL-2j

1,38 2i.UJ No 21

30 TiC I N-V

45 240 No 41C0

to N~onIc .

15 Al (Olt),

I A to IAlt>

&-in lt,-~~050W 0.~

IA. d.5 No

M~.

f,r ~Lwil@ twtir -

72

TABLE U.UMIL?1; OF C pMLZ~t0I rk;;-.; 7M 14 WALLIZED DIS

SLiNTOW AT 1A(,.C UJI;:G kEkCflQUiTAL 1KLTALLIZU)G 14fl ME

CM Wa mpooio IW lb CpItio CAI L. Lw. (lb)

weihttp A-%AD* AL23.6sh (P) A0-4 A6hL2

.5 21014 No10 32 2551No 17090M V4 9-uA I 20

6 255 No. 21 LL~1w 33 2aw No. 2D LOW96.5 S1OQ2 Q~ 400 1900 M5( T103 19W wm

2w 190C. aim. 1011 21 N 1W 34 25514. I

122.5 rAO 1700 7 TI2 um

9 21014. 30 15OU 35 21014 M-0904 zo 1000o 505.0 wIO00

96. 5 S1O -MJ

17 25514 No 20 1420 43 2103me. WC "MW 1W

31W. I,'00 td?, 1IUW1

is 255 No 30 1720 4 25, No 1W 700 IMO05 Zn 2)00 %Q ~ 45 ?.Ziyar 23W 00S 900

19 M Yk( 2120 4.0 44. 25514 No 50 60054 zoO '4w;; 23)o 43 7*0? IWO 00 900

Z, 255 M. 4=7 9w-. 53 25534 2.A IJD 90060 Zr02 low 4A0 =2 90 w 10

21 255 Ka 2w0 5 255K No __45 Kn 16-1; A84103 1900

te. 21 ..

2/. 251- V, WOO 57 IN5014. 071 *42 ,4"0 91 3 SOU

'9 215K me 0 am 65 292.5 No 3100 MOO5s o I r. 7. 5 T 2)00 ism

2100 j 23i33131 2 3 No 2120 1;2W 1410 66 2851no 2m0 13W0

73

TrA.I 12 (cant.)

IMJJTS 0? , ltE SMI0 TI FRO XOTALLIZDU DISS.AITUM AT 16000C UIW LV~1MFNTAL XMOALITMID KIfl 3lr5

CmpoWttom Compmalon .ood (lb) Coupwiton C; ws.tor. Wd (lb)

.4.4bt (@B) A. 19 A..-Q AL-SI 0 j~0t (gin) A01 L9 A-.3

6 270 No 300 1J00 1" 205 34o 1600 4; am030 Ti 31--, 5W0 15 alc 1000 800 700

3001000* 173

(A 292.5 No >4 400 105 ?0 go l 0 o0 009.2 Cs.0 900 500 30 lapa 3400 300 1.100

69 205 40 2500 700 136 2651No 330 600

11.6 Co02 2900 500 15 FWapm 'M= 900

70 2703Me 2300 300 207 300 No 3100 600

36.6 C002 2 00 9 50000 O1500

' 1133

72 9240 lo0 400 26 300N0 .m00073.6 o02 220 1.200 0.9 re 2400

71. 2963b 2000 0 1O 3 297 4o 'm 9008.6 6o2OD3 220 500 3 fe I=~ '

,2400 9330 533-

75 285340 230 LID 291 No 23W000m3.6 U 3 S00 9?.* 2200 1300

2333 0 .

81 292.S5ND 1300 A001 270 m40064 ,, 190 700 o TotAM

17330 5330 )353Y65 292.5 No 900 2.12 3m050 3000 900

1000 OO 0.96 Ki 4000 MOD

066296 )4 260 116 300)Me 30 230021660 XO 0 .9 0C' 3900 900

103 2 . 1"3 2"l Wo 2.17 29714A .4000 2D03 .o 1200 30 1000 3 c 4000 600

0.D0 1.* 1031-It 7

*AV-wegT Solo~ted for Tenil. Toting

71+.

TABU12 (C-'t.)AW2 ZS OF COURMSION mTST FRU IWALLIzZO DISCS

JMN?'n 4T 16IwJ, UsdNQ EXPEMIWIA METALI=ND ,LXWM

!I Co..1tio Coopm. im Lod (lb) C-j A.I I*" o (lb)

113 2916 No 135 li 2400 No

13D X.13P No 00 AM 13o6 249 o6" 26W20.94 d N00 1400 330W 1900 130

4L 2*0,

121 2#7 0 41000 11.0 137 2106 M. aiw 1600 u3d '400 130 )U 6b 200 70 S

30 TI41 " ftff 0

1222 2916 VA p x 700 138 30P me I0 6009 d '00000 )o PA6 1300 ID0

_ J0OTIC

123 270 P. 41000 1100 140 260 me moo ow I304 '40M 900 Um 2600 1low 900

5; -W 9 h1.L

zio No0P 2700 w4. 291 P. 370 17005160 2O 9 TaLe4 3600 20

1in 2AD me 1700 14 276PM 3600 7W51 ft 2500 15 n 3000 709 T1 IN A f

13. 241) m 20 1700 U43 27960 3w0 90051 W, 1W a 1S f. 3600 MOO

4Ta. 91" 3~)L

U3 00 go 2 0 1"m I" 7m60 3900 9001.26N0 .0 0 156 No 2w su

1Z U102 140J smAli -;

322/0 No 2900 1600 U3 24060 1700 I=60 TLC 3300 g00 .1 No 15010 um~

131 206ft 3600 1 . 153 22356. 10 1.200306 No 1' 4m; 1-60 N4 00 1400

zw t. d.d for Tomuhi TseU~og

TABLN 12 (C nt.,

RMTS? OF CC~lKM=SION TROTS Z~ rNTALLIZED DI~S

SINT=E AT 1600-C USING EIXflRflAL METALIZUNG KMWE

cooition Coqwemoion Load (lb) Compit~ti. Cm.aiw Load (lb)

ADAL-23 webirh'() h- "~ L2o igbt (on) AD-% ~ htAD-96%rf4AL2

142 20 16C197 261 M. '400 27C0 70r,60 M0 2 1603 1300 15 3L02 4000 26Co 000

Z167- 1700- 23W 2Y'900

115 255 N~o 2Z.00 2300 :2.99 300 N,03 260 900jo ma 1000 230C 2300 ODD15 L12002 = y"

5 m . 'A g ~ 22000

90 DMCO3 3000 170060 WO02 7 F3

T Walod tlr Taoil. Ta tig

76

?AR.E 13iWLTS 0F MOURMSSON Thar. MPJ 4=WLIZX D!XC3

jT13M A7 I~tU0C UZIIG LEW~TAL NETALLZM K2UT MI;St

.1 COWDiontl. j{prwuo Ll(1)cl , a t loAw (lb)'hi6t tp) AD-% hAD-96 A-23 hadlht AA- 1)- AL.-23

S 2m0 X 1Y6X 1300 65 292.5 No ]202

7.5 -T 11750-20

1) 2WJ4 W, C (.v 6 235M No L 21002w ~15 11 4WL

113 U13 w 1900 67 270 N 22W.) 1000112 ZV 2001 3u TI 1600 aw

23j 5, am 500 68 292.3 m 160075 NCO 1.300 700 9.2 C0

17 255 No 900 W~ 235M No ~ 1100=2 SOO ~ 830 .. C402 2300 1100"

IN 255X YW 3600 8.) ) 2704 I'43 Zft 3000 22.3 36.0 C42

I t 71 2h03No 3000

19 255 So 3500 2DO C )o

1 3514 '1230 272 )bM 2300

33 235 go 50073.6 002 20

601 'h00900W-1

7 7. "29640031 2150 W.10 '4100

906 120 goo

51 8 4NO2 1600

AS82102 2x9)22 1.

T J.aoteM to TMoa TottiM

77

amTs rau 13 (coat.)

SINTUS A? ::"ZO Dr3.

T-'" 'U- COMMOSOO Lld (lb) I

am c"PositiftM94*t (P) A,% Ag"O ALQj aw ica LOW (lb",

86 2% Y. AP-N AD-* A4432L J " 3100

;am IM 300 W. -VOO

IN. 0.9 w

IM 300 K. '40M I"M 17W 121 m pb 37W

IS T 3 -0 30M

Ics 3m X.4.9 ro ow I./ou L23 27r, Oft

30 a

3 to aw 132 240 bS g. 60 Tw

291 V. '4= um9 ft 26M im 44 1

low2m Im

122 3w K. 13W1=

Ift 90 AfI201Amy 21739; 90 0 ;WW Awzg.w ou. 3300

IJA 3W 6000.9 C* I"

194 all F. asm ilmU7 Wn )b 31M 15W 70D

3 Co ft

'A=

for T-us TootLg

78

fABLh 1.4

...MWX VA CJftL11trN Tk2,1 VW2LT3

ADALh A>-*eAL2

M-Zn 2/4 NJV11/

M. Tl 2/i. *2 Y 1/3 1.!*

1/2't, 1/1

N-d 1/3

Ho-n 31 '1i,1/ 1-

N -dIao.,2 1/2N -31D2 1/1

X-Tt 1/7

H 4-M-1SiA2 1/2

Mo1?J. '1 H-CO2 1/1

N-F*AL-03-jC,21/1

mHo-1 2 034a AO 1 t& 2 3 s 2 22M4iO-12

PM /2 MoN-M2 1/1 Ih1-fl I1, -?1LiW 1/5 No03-4 1/1 AMD80 /

L/i KC0 /

j M H-Nn-5iC 2 3/o9-Ag-L20- 12 s-i* /

M _- __ N- -zo / -- 1233I

79

RESULTS Of TISIL9 TESTS USING S~kZR'1 DWqW ND- IVT~ F"11Ah5TAL W~~ALLZ1EG MIXTURS SUNRAL AT 1JOW

jcWapoitioo TovilLod p~i)j cmps-ition lowc (psi)

50 255 N~o 160 1570S IO 11 0 5

ASSJ2 .0 12 0 ?W ) 300 1~0 1.50221 MR =w__

43~ ~131 21K.i: -J

AM11 OF TENSILE sm.S MING 815313 DUXMh AU.

AN UUNA MWUCT ALLIZIM U" 3DIZT ATJRUP

e oe an Tvni)A Load (psi) Omwtm Tomils LW (psi)wa0o (w Z *-96 AL-23 de, h (4 ±1AD%4 A AL-)

A, 3 o2290 6250 129 2I ' 0090Yla"9000 74 51n 90

U 255 Y. 4mO 1()6 139 240 N s43 ?.lI~por 390 750 360 1 .

0500043 210 Y. 570 700 19.2 Co

90 Ta" ni 7 5 7A 4.8

74 4 135 240 r.0 mm030K 12300)O

47 96.5 SL2 304 ?1f

210KM. .I37 2.10 Mao Ilam "m30 Mn 9150 5110

;.9 255 102 m00 ISSO N.~

26OG153 225 No 50

50 2557Y.60 142D0 15 AL(ON)322K Me. 57 I"U A3 "0k

S$246 W 15150 6 lf 00 Ms3.75 LIihm 564

Noar-t 175 150.. Nu66I 90K0

I_- -~ 60 -"-2 5,.

80

U=WTS OF TINS1IL ?T0 USING SPMVal; DESIGN No. IAND CLIMYETA.L XL7&hWZLiG MI1TUm .. , 2Q2 At 1500-C

cositim U"I1I. L A (pal) 0,O.IUO.. Lo.:. .w, (psi)

IIh (m) -94 O9 AL-2) Io A sm AD-% 1- L2

1.1 210 H 17W 87 M Mc, 13810m 2 930 3021i 43$00

25 2AL N 65WX 74 201 535046 150 36.8 0.02 "M0

32 255 No 5"13 72 2.1.C, ML4 2o45 ?1 9150 73.6 CaG2 6.36 1&))u

56 15U o 1,e~ 91 2731No 1555091 Ih03 8500 y 111mW 1570r,

126 cao I awa __t

66 292.5 No 28"0 600 L35 241~ ft007..1T 1935 10000) 30 No 15150

66 20l M. 734. 153 211 ft

15 TI 12152 W P 11

49 255 No L130048 .31C: 163

1300

50 5 NO 1A05.X.

48 .110 16100

81

onD L*JL~kVTAMC NALIZII M1 .Thkfl C'UE A1

tfl ('1 A 1, 1 'mght (0) AAA)AC-9 2

9 2iJ 11, to-..l

12 J55K No "&,l W II NoK 1J5SLU5) Aw lw5 j. 7. m

Cc 70

Z') 2)51 me 110

02'mL 121 H" ww

-71 *02X 1; WI

ff. 1 1d

U1 M1K 1w. a41 291 179m

24 C4402 1

9W8 CW3 90%,1 "N

Ofi 240 Va

70 80 Y.15 Li.o

159 ftAp 95

170017 282 U

tAMLk 191.* .'O fr=. TW ? SIW Zpaff. AP214~C. 1

AND -:O!AIXtc2AL N.TA J471I;G MHIIuk. 4I-l.=h AT 17DC

15 255M L4 19 29'M, LO4i :A 11950Jl s-

IL 291X. I.

19 255 1413GX56 A) l7Wu

vLot 111 3-.. 7,Xa 5Mr

31 2!Z N. IlL il..90 7 L. .. 9 C. L

L. - 1050

4"La 9 Cia- L-15" C* 2*

66 285No05W a&Ah

9.2 C. 1.72% 117 291 X W

ft 3 3.;

7v 2m VA36.8 2~ 9550

73 292.5 IL17.3 U42CO3 5150

73 296 r !x6.7 11;03 114

*Ivwvvg

83

_________________ ~IJ6ARY OF Th2XSlE MI R2JSLT:;

)4etaflisiog Systm Merit metaijio 3tm Iit ,t&U ±zio Spm

Ma-Th I'llX-Z 2/2

Mo-Ti 12

M*-C@O2 1/1N o-Ni 3/1Mo-Co 112

mo-4 .1

M-Lt203-M"02 1/iMo.Mg-SL02 11/1

M-U.203so,-lOi/ M-Mo-LiCC, L/i

Mo-,d ,/

Mo-zr 1/i~M-ut 2/3

Mo-Th '

:o-F-4I2o3.sAo2Mgo1 1/i

Mo-C, 24 1/1 1

M*"I / tG0 v4o-.u 2 2/2x M-? 3/5NO-L12Mo 1/1

No-LAAM03 1/i M-,A20r=M2 1/1M No-c -A-30 2 i/1 N0460-102 2/2M 24h*-Ti 1/2

Mo-o-O 2 I/4 Mo-Mo-S 2 i I MN-W 1/1

?A"L 21

RkIOWWD MECALLILIAWl HIMXIRS

Copostion Cmoaltion AVA S1Jnteri~g N"j Test v .s~ im Too±A Test

65 292. 5 No 1500 2... 2=l0'4000 19350

7.5 ft. 16=0

91 270 No 13M 00o.4000 5700

)0 Ll 30 14W

UL1 291 No 1600 300 123009 Talc 3"0 17900

12 24 o1500 2 3SW V&V

73.6 4wO 4000

5 55 No 1200 2 2M 15704 SL2 101300

222 Mo 227002

50 255 No 1500 2 2O14=C

22 AM 1 15=~

1.9 255 Mo 1500 2.75 AQ20 11300

26 _ 2M2OU

TAUJ 22AMLISIJ OF C4WAM25 ?W? DATA

at. All Test Daa

Data* Tenosl Compresuons Oren NFj Tosss PuaL

o 52011.5 lb, Mid6 psi 24.0 M LaW 254C,27.8 10.5 41 64 I 47.57

561 15 122

b. 3sleat. '"LI Date. N - 2S

*AvyAP (;), BAda~rd d~wiatloo (s), GOOMOOIc.t of mtraioo (OWS), &W Nmb.. t trio". (Nip f. tamelA G=WveaLm,drum peels aM toue peal tests. A *twarad 80-poo t Me amS 20-pee. Mo ai21.a1W mixture. ad taftimado1Atering and braving coondition* were s-sO.

85

AaLi; 23

.31 .,elimw (mma'urd, Pd.i) (--urn-, Psi) (cajcu~atag, pqi (calc~a ted. poi)

N-6 450C 65cfaa- N9_w67W 660

t- ce .11c -1300 10550N-11 - 300 L=ZX

-aio -5 -16500 2037COwt2-6i- * -11600 201700 ~ m210

86

'4 -ro

D'sc A Disc 9 IS" Cfr-l'oop5 "OCIt cc I AD-I tCO*$ PORCELAIN CC I09601451? AL-2S

IUA?(RA&S foa I4ECTrollif. b

F IGURE I

MEITALLIZING COMPOSITIOSS APPLIED TC TEST DISCS

0-00 TEMPfRATJRES DETERMINED AIYH C"IOML-ALUM(L 1t10110COUPI.E-- - -- TEMPERATURES DETERIMNE WiTU OP?,CA. PYROVMETIP

%500 0 POSIIONS OP 90AT imIUERALS INDICATE DWELL TIME IIIN INUTESI

LPOOFILE0 20 40 so so 100 l17 lAO IGO 19S, too

DISTA06CE IN1CHES

FUNACE TEMPE!RATURE PROF ILE

87

40AZED STRIPSOf KOVARCIAMIC

FIGUREi 3

TORQUIE PEEL TESTIN~G FIXYURFt

88

FIGUIE it

ccIPeO.siN TEST SP(CIMEN

MTA L POTTING COMPOUND

CtRA~MC

St' •ONS OF TWO COMPESSION SPECIENS STItL VACUUM-TI. T AFTER OISTORTING F!1 OVER 500-POUVO LOA')

89

S .o0' 10.015'

0.425, 10'0

-- 0.402" 10.006'

0.2S2, l0.003'

II

0. B~RAZED SURFACESat

S0. 5t *

-SK[R*Y TENSILE DESIGN 40.1 b1**,' b. ASTM

r~'1

IIN

M42t0 5 .I Is -4"100w

o..26* 1o.oOOC I0.402' *.00*--0 .50, to 005-

WCTALSIZCO FL.AT TO .0002S*

su atPAIRAILC. TO .~~$ A 4 1-W(T...LIZto AND

710" 0.001' *AA 0 WUVACII

Is I .. * a- **'1 FLATo

V i T a I

aTM~~~~~~~ ~TISL TTESTL TETSPCME IT OOrEOTP44t Wct

2 1u~

T(jL goT~i(II5

* COORS Ao-94 *IT"w I COORS IA6,-4 WIT C O IJSSiT AL-23 *J00.PC ohobTv *5 AT i50O-C COMPQ T ON 12 AT 15006C COMPOSITION 50 AT 15006C

TENSILE VALUE I 26,40OPS TEN4SILE VALUE-22OOQ PSi TENSILE VALLE-4,I1i PSI

I GURE?7

TENSILE SPECIMENS AFTER TESTINIG

TEST STRIPS

CNRIC.TEST SPiCIPfS

91

UWVI*SA J*

C RAC 01 A4O

IT PI £~A

@AL SAL OO4'm~A*92*

e. SEAL ILE10MM 101t T 10141606E b. SEAL ELEMENT AT SIING TauPN tATU

*SEAL ELEMENT AT 40O0 TEMPERATURE 4. PLAwl VIE1W OF SEAL ELEMENT

FIGURE 12

SEAL ELEMENTS UNOEREORS STRESS

95

I

TEMPERATURC (OC)

0.

IL

STRAIN (IcMES/IW1CH)

STRAIfSS(INRAI! GRAPHS

96

~OU~S~lE ST*I~ TIENSILE STRAIN- (liNCNES/INCM) +

NiSftd EXPANOINO (NEAL.)

LOW IM0ANONS(ERSUICI

TEUPWQATURE 40C

C

STRESS-STRAto GAAfqIS

97

s:m*I IPmCMINt#T

PROM CoC @*z TO*ya

/00O 510C

&-INA MISMA C T Ml

1 TIMPERAY URS

THERMAL MISMATCH FROM SO00CTOgAETtUPgRAUSE

1.. -A

STRESS-STRAIN GRAPHS

98

TRANSVtRSE TIENSILIE STMESS-SlM11,1 SPECIM 0 IN

t40IfEO Ni4-TEMD4PfTUPE EXTESOWfT(

99

TAML~~TIAXIA STRAAiN GAGEOU£

"Ua

me TM.

FIGURE! 17

IRMIC-TO-4,1ETAL SEAL TEST SPICIMIN FOR STRIES

INKSTIGAT ON

100

IftRUI4ENTEO STRESS $PtCiMf%

101

AO"4L%*E: PEECMt N %,NtJ RMP NqTJ ' ftd'A

12,000 .... ...."77 777

. ......... ....... ........ .........

7::*:.:: -- 77 --: ..... ....

... ........

... ....... 77 -. = .-T7-77-7- t::::'7 ................. .... 2 NIKKO T901INI]SATUIlt.. .... .. .... ... .... ..... ......

am .... ......

... ...... . .... .... .. .....

woo'.. ................. .... ..

7 7 ..... ....

4000 ... ...... ... . .......- . ... . ......

......... ... ..... ... .-7-.,77

2000 ......... ..... . ..... ......... ...

11-4 a: . .. ............ ...... . . ......... ............ 77........ ...

0 ...... T ::

0 0001 *on 0004 owa 0006 0006 ocor Gam

F I oAt 20

STNJU VIM$ $TRAIN FOR 'A' NICKEL

103

24.000aw p.~SI3T~tIW

.. .. .... .. .. ....

20.000~o3 ~ O 00 ~ 0 46 0

.... .... .. ... ...

.977 ~ Flim '7 .7 .. =

..,U .....S ....6 T~v~5

.. ~ ~ ~ ~ ~ 1 ............

24 T

20

... .. .. .

... ... .. . .....

0~~C 400 m w oo a ~ o im gao=

1'.. ......... ...

.... ~ ~ ~ ~ ~ ~ S ...... . .. ....

..EA ........ iSS IIPIAIJ

...0. .. .

DISTRIBUTION LIST

No. of Copies

2 RADC (RCLTP)Griffiss AFB, New York

I RADC (RCAT)Griffiss AF'B, New York

1 RADC (FrOIL-2'Griffiss AFB, New York

10 Armed Serriecs Tc:hnical Information AgencyArlington 1'a~l StatiunArlington 3, 1- Virginia

1 AFCCDD (CRQSL-l)L. 3. Hanscom Field, Massachusetts

2 AUL (AUL-7736)Maxwell AF2, Alabama

1 WADD (WCOSI-3)Wright-Patterson AFB, Ohio

1 ChiefNaval Research LaboratoryATTN: Code 2021Washington 25, D.C.

1 Air Force Field RepresentativeNaval Research LaboratoryATTN: Code 1010Washington 25, D.C.

1 Commanding OfficerSignal Corps Engineering LaboratoriesATTN: Technical Reports LibraryFort Monmouth, New Jersey

1 1RADC (RCOL, Capt. Norton)Criffiss AFB, New Yoek

I RADC (RCOIA, Mr. Malloy)Griffiss AFB, New York

1 Physical Electronics Labs2493 Pulgas AvenueE. Palo Alto, California

1 Western Cold & Platinum Co.525 Harbor BoulevardBelmont, California

DISTRIBUTION LIST (Cont)

No. of Copies

1 Raytheon CompanyWaltham, MassachusettsATTN: F. J. Fallom - Dept. 30,99

CO, TSAS R&D LabsFort Monmouth, New JerseyATTN: SIGRA/SL-PRT

ARDC, E" -pean OfficeShell Building, 60 Rue CantersteenBrussels, Belgium

SecretariatAdvisory Group on Electron Tubes346 BroadwiayNew York 13, New York

1 Polytechnic Institute of BrooklynMicrowave Research Institute55 Johnson StreetBrooklyn 1, New York

1 California Institute of TechnologyDepartment of Electrical EngineeringPasadena, CaliforniaATTN: Prof. L. M. Field

1 Harvard UniversityTechnical Reports CollectionRoom 303A, Pierce HallCambridge 38, MassachusettsATTN: LibrarianUniversity of IllinoisElectrical Engineering DepartmentUrbana, IllinoisATTN: Electron Tube SectionMassachusetts Institute of TechnologyResearch Laboratory of ElectronicsCambridge 39, MassachusettsATTN: Documents LibraryUniversity of MichiganEngineering Research institute351 E. Engineering BuildingAnn Arbor, MichiganATTN: J. E. Howe, Research Associates

DISTRIbUTION LIST (Cont)

No. of Copies

Stanford Research InstituteMenlo Park, CaliforniATTN: Document- Center

University of CaliforniaElectrical Engineering DepartmentBerkeley, CaliforniaATTN: Prof. J. Whinnery

University of ColoradoDepartment of Electrical EngineeringBoulder, CcioradoATTN: Prof. W. Worcester

Johns Hopkins UniversityFadiation Laboratory1315 St. Paul StreetBaltimore 2, MarylandATTN: Librarian

Linfield Research InstituteMcMirnville, OregonATTN: Dr. W. P. Dyke, Director

University of WashingtonDepartment of Electrical EngineeringSeattle 5, WashingtonATTN: A. E. Harrison

Cornell UniversityDepartment of Electrical EngineeringIthaca, New YorkATTN: C. Dalman

Bell Telephone LaboratoriesMurray Hill LaboratoriesMurray Hill, New JerseyATTN: Electronics Research DepartmentColumbia UniversityColumbia Radiation Laboratory538 W. 120th StreetNew York 27, New York

Ritl-McCullough, Inc.San Brunc, CaliforniaATTN: D. Priest

DISTRIBUTION LIST (Cont)

No. of Copies

Federal Telecommunications Laboratories500 Washington AvenueNutley, New JerseyATTN: LibrarianFrenchtown Porcelain CompanyFrenchtoun, New JerseyATTN: F. J. HymesGeneral Elect:c CompanyElectron Tube Division of Research LaboratoryThe KnollsSchenectady, Jew YorkATTh: E. L. McArthur

General Electric Microwave Laboratory601 California AvenuePalo Alto, CaliforniaATTN: Dr. L. Coughanour

Hughes Aircraft CompanyResearch & Development LibraryCulver City. CaliforniaATTN: Engineering LibrarianHuggins Laboratories711 Hamilton AvenueMenlo Park, CaliforniaATTN: D. A. Roberts

Litton Industries960 Industrial RoadSan Carlos, CaliforniaATTN: N. Moore

Philips Research LaboratoriesIrvington-on-Hudson, New YorkATWNs Dr. B. Arfin

Polarad Electronics Corporation43-20 34th StreetLong Island City 1, New YorkATTN: Vacuum Tube Engineerirn Department

Radio Corporation of AmericaElectron Tube, Chemical & Physical LaboratoryLancaster, PennsylvaniaATTN: R. .1. Zachariason

DISTRIBUrION LIST (Cont)

No. of Copies

1 RCA LaboratoriesElectronics Research LircoratcryPrinceton, New JerseyATTN: E. W. Herold, DirectorRIaytheon Company, Spancer LaboratoryMicrowave & Power Tiibe DivisionBurlington, Missachusetts

1 ATTN: L. Clampitt1 W. C. Brown

Sperry Rand C2rporationSperry Electron Tube DivisionGalasville, FloridaATTN: P. GergmanSylvania Electronic Products, Inc.Physics LaboratoryBayside, L. I., New YorkATTN: Dr. R. HlitterSylvania Electric Products, Inc.Microwave Tube Laboratory500 , yn AvenueMt. V. , CaliforniaATTN: Technical Library

1 Varian Associates611 Hansen WayPalo Alto, CaliforniaATTNt Technical LibraryStanford UniversityStanford Electronics LaboratoriesElectron Devices LaboratoryStanford, CaliforniaAITN: D. A. WatkinsNationul Bureau of StandardsElectricity and Electronics DivisionWashington 25, D.C.ATTN: Dr. C. P. Marsden

DISTRIBUTION L13T (Cont)

No. of Copies

1 Ceramics for Industry Corp.Cotta n, PlaceMineola, L. I., New YorkATTNs T. S. StoinislawCoors Porcelain CompanyGolden, ColoradcoATTN: Lawrence Farv'iara