IC Copy No. - Defense Technical Information Center

1 MANUAL OFMETAL-TO- CERAMICSEALINO TECHNIQUESELECTRONIC TUBE DIVISIONSPERRY GYROSCOPE COUPANVIN DIVISION OF SPERRY RAND CORPORATIONOREAT NECK, NEW YORKP repared byENGINEERING JohnsonPUBLICATIONS under the sponsorship ofROME AIR DEVELOPMENT CENTERAIR FORCE SYISTEIS COMMANDUNITED STATES AIR PORCEGRIFFISS AIR FORCE BASE, NEW YORKC ontract No. AF301602)-2371 Project No. 5573 Task No. 557303LI Copy Pub. No. NA-27-90\ ~ ' May 196* NPATENT NOMICEWhen Government drawings, specifications, or other dataare used for any purpose other than in connectiontwith adefinitely related Government procurement operation, theUnited States Government thereby incurs no responsibilitynor any obligation whatsoever and the fact that the Gov-ernment may have formulated, furnished, or in any waysupplied the said drawings, specifications or other datais not to be regarded by implication or otherwise as inany manner licensing the holder or any other personorcorporation, or conveying any rights or permission tomanufacture, use, or sell any patented invention that mayin any way be related NOTICEQ ualified requestors mayobtain copies of this documentfrom the Defense Documentation Center forScientific andTechnical Information , ArlingtonHall Station, Arlington12, Virginia.

2 DDCservicebfor the Department of Defensecontractors are available through the "Field of InterestRegister"ona "need-to-know" certified by the cognizantmilitary agency of their project or Section PageI INTRODUCTION 1II SELECTION OF RAWMATERIALS 52-1. Ceramics 52-2. Metals 6 III seal design 93-1. Basic Thin-WallOD CompressionSeals 93-2. ID or Pin Seals 11"3-3. Butt Seals 133-4. NonstandardSeals 17IV PARTS PREPARATION"AND METALLIZING 194-1. Cleaning ofCeramics 194-2. Cleaning andPlating of Metals 214-3. Metallizing 21"4-4. Evaluation ofMetallizing 284-5. Electroplating 30V FIXTURES ANDASSEMBLY 33jVI YPRAZING 37iiiCONTENTS (Cont)Section PageVII CONTROL TESTING 39 VIII FAILURE ANALYSIS 4*AppendixA RECOMMENDEDTOLERANCES ANDMETALLIZINGALLOWANCES 43B METALLIZING MIXTURES 45C COMBINATIONS OFCERAMIC BODY,METALLIZING MIXTURE,AND SINTERING CYCLE 47D BRAZING MATERIALS 49E TYPICAL PLATINGBATHS ANDPROCEDURES 51ivII ILLUSTRATIONSF igure Page1 Expansion Coefficientsof Typical CeramicI Bodies and Metals 72 Thin-Wall ODI Compression seal 103 Transitions to Thick-Wall Metal Members 12I4 Self-Jigging Seals inWhich Metal MemberI Performs Jigging 145 Hollow-Pin SealsStrengthened withSolid Pin Insert 146 Typical Butt-SealI Assemblies 157 Back-up of DuctileMetal seal withI Second or BlankCeramic 16I 8 Thick-Wall seal 189 Rectangular Seals 18I 10 Hand-Coating Equipment 2311 Roller-Coating Equipment 2412 Spray-Coating Equipment 2413 Dewpoint Cup 27 IVIISECTION II NITRODUCT ION"Metal-to-ceramic seals have been inexistence for thousands of years, since earlyman first made decorative enamel-copper trin-kets.

3 Little progress was made until the nine-teenth century with the advent of the internalcombustion engine and the need for spark then, there have been considerable tech-nological advances, particularly with regard tothe electronic tube industry. To further knowl-edge in this area, the Electronic Tube Divisionof Sperry Gyroscope Company conducted a sealtechnology study for the Rome Air DevelopmentCenter, Air Force Systems Command, underContract No. AF30(602)-Z371. This manual,discussing the procedures to fabricate metal-to-ceramic seals, is based on the study the study program, the effectsknown variables had upon the ultimate strengthof metal-to-ceramic seals were statistically designed experiments wereperformed to examine the significance of thevariables and their interactions on over-allseal strength. The results of the program weremasked to a great extent by a high residualerror due to uncontrolled and/or unknown vari-ables; only the most marked effects could beobserved and assigned statistical of the high coefficients of varia-tion (or dispersion) of the experiments, manyof the recomn ended processes are of an advi-sory nature.

4 Ather than of specific main contribution of thismanual is in cataloging the variety of processesrequired to fabricate metal-to-ceramic seals,with recommendations as to the relative degreesof importance in controlling various phases ofthe operations. Some of the techniques em-ployed in fabricating seals and controlling theknown variables are described; the ceramicengineer can select those suitable for his par-ticular needs .It must be remembered, how--vever, that fabrication of seals depends to alarge extent upon the skills of the operators;and reproducibility and reliability can begained only through proper training and The organization of this manual ispatterned after the normal work flow, from rawmaterials to finished product:"* Selection of raw materials"* seal design "* Parts preparation and metallizing"* Fixtures and assembly"* Brazing"* Control testing"* Failure must be emphasized that this manualprovides recommendations and not absoluteinstructions.

5 With the large variety of ceramicsand metallizing mixtures available, an almostinfinite number of combinations is possible,each possessing individual characteristics. Itis true that many combinations with similarcharacteristics can be grouped together, butproficiency in the use of a particular ceramic-metallizing combination is obtained through* careful observation of all phases of the techniques which can be employedin the processing steps have not been mentionedfor the sake of brevity. Substitutions can bemade because of available equipment and facili-ties. For instance, if electroplating facilitiesare not available, the application of nickelous"oxide or cuprous oxide, with subsequent reduc-ing atmosphere sintering, can replace the elec-troplated layers. Similarly, some electroless"platings can be used if phosphorous contamina-tion is permissible.

6 The ceramic engineermust select several combinations of ceramic,metallizing, metal members, and brazingmaterials and then work with these to establishbasic parameters of operation. A skilledengineer soon develops a basic understandingof his selected combinations, and is better ableto employ these materials for his work still remains todetermine the basic mechanisms involved inthese processes. Several investigators havehypothesized the mechanisms, but none hassucceeded in accurately describing the observed3phenomena. Manufacturing of metal -to -ceramicseals currently straddles the line betweenscience and skilled craft; a few years ago itwas completely craft. The immediate futureshould benefit as mo re scientific data arecollected, as the mechanisms are more com-pletely I ISELECTION OF RAW MATERIALSI2-1. CERAHICSj The choice of the ceramic body is basicto the process of sealing.

7 Selection is primarilydetermined by the application of the device inwhich the seal is incorporated, because of themechanical and electrical properties of differentceramics. Some of the properties to considerare thermal shock resistance, r-f losses,softening temperature, thermal expansion,porosity or vacuum-tightness, degassing, andphysical percentage of alumina in the body isproportional to the ease of sealing. Manyceramics from a variety of suppliers are avail-able covering a range from 85- to 100-percentalumina. Dense vacuum-tight aluminas aremade by mixing a small percentage of thevitreous or glassy material referred to as aflux with the refractory polycrystalline material,and heating the composition beyond the meltingpoint of the flux. Two alumina bodies with thesame percent alumina content will not neces-sarily have the same physical and electricalproperties, because the composition of the fluxcan vary.

8 In general, ceramics increase inprice as the purity of alumina increases. The5 Ilowest-purity alumina which will be satisfactoryprovides the greatest cost savings in rawmaterial and yield in 2. METALSThe selection of the metal member ofthe seal is somewhat governed by the physicaland electrical requirements, but it is moredependent upon the ceramic body such as magnetic properties,r-f and d-c conductivity, and structuralstrength determine the selection of a group ofmetals. A particular metal member is thenselected on the basis of its coefficient ofexpansion. Ideally, the ceramic and the metalshould be identically matched, so that theyexpand and contract at the same rate throughoutthe brazing and operating cycles; stressescaused by dissimilar expansion characteristicsare thereby eliminated. It should be noted thatthe stringent thermal cycling tests performedto evaluate the reliability of finished seals arebased on the stresses developed between thedissimilarly expanded ceramic and coefficients of some typicalceramic bodies and some of the metals morecommonly employed in sealing to the ceramicsare shown in figure 1.

9 No ideal combinationsare presently available, but the metals andceramics currently used allow construction ofhighly reliable seals. Some of the metals6!100i I IGLASS-SEALING ALLOY(GRAY) KOVAR A AND FERNICO,-. 90 --70 -7"C ) -- Z" 0 -ALLOY T2 Cu( )c ALUMINA50O -L --JJ C SAl -00)g ,(WESGO AL-300O/-0 (WESGOAL-IO09)40 ---CARIBON STEEL0W ( )d" No )10 d00 l00 200 300 400 500 600 700 600 900 TEMPERATURE (*C)i FIGURE I. EXPANSION COEFFICIENTS OF TYPICALCERAMIC BODIES AND METALS77 Inormally encountered in fabricating thevariety of seals described in Section III areKovar, No. 42 alloy, Sylvania No. 4, nickel,molybdenum, and IIISEAL DESIGN3-1. BASIC THIN-WALL O COMPRESSION ' seal 'SThe thin-wall OD compression seal (figure 2) employs a thin surrounding metalmember which relies on its compliance to com-pensate for the mechanical stresses caused bythe thermal expansion mismatch.

10 Wall thick-ness usually ranges from 0. 005 inch to 0. 025inch. The metal member should be as thin aspossible while still offering the mechanicalstrength required for handling and designing this type of seal , theengineer must rely heavily on judgment and pre-vious experience. Small seals-those about0. 25 inch and less in diameter--are more diffi-cult to seal successfully than are seals having,for example, a diameter of 1 or 2 possible, small seals should usethinner metals. seal thickness or length is animportant factor. Extremely thin ceramicmembers, which provide short leak paths, areto be avoided wherever possible. Seals havingthicknesses or leak distances of 0. 125 inch orgreater offer fewer problems than those in the0. 050- to 0. 030-inch range. In designingseals, it is necessary to allow proper clear-ance between the ceramic and metal member9 IROOMTEMPERATUREIi IIMISMATCHSTRESSHEATEDMETAL CERAMICFIGURE 2.