III MTLTR86 6 AD-A239 958 C II!!i!1 II!I 1111 ff I JI~I' IIJ!' IJ 1i ll CERAMIC LIFE 0 PREDICTION METHODOLOGY N FINAL REPORT L.R. SWANK, J.A. MANGELS ts J.C. CAVERLY, R.K. GOVILA Research Staff Ford Motor Company P.O. Box 2053 Dearborn, Michigan 48121 May 1986 Tm Contract No. DAAG 46-77-C-0023 IPrepared for U.S. ARMY MATERIALS TECHNOLOGY LABORATORY WATERTOWN, MASSACHUSETTS 02172-0001 U.S. DEPARTMENT OF ENERGY Division of Transportation Energy Conservation 91 -09329
Transcript
III MTLTR86 6
AD-A239 958C II!!i!1 II!I 1111 ff I JI~I' IIJ!' IJ 1i ll CERAMIC LIFE
0 PREDICTION METHODOLOGYN FINAL REPORT
L.R. SWANK, J.A. MANGELSts J.C. CAVERLY, R.K. GOVILA
Research StaffFord Motor Company
P.O. Box 2053Dearborn, Michigan 48121
May 1986
Tm Contract No. DAAG 46-77-C-0023
IPrepared forU.S. ARMY MATERIALS TECHNOLOGY LABORATORYWATERTOWN, MASSACHUSETTS 02172-0001
U.S. DEPARTMENT OF ENERGY
Division of Transportation Energy Conservation
91 -09329
IIIIIU
IIIII
The findings in this report are not to be construed as an official I
Department of the Army position, unless so designated by othei
author',d documentt 3Mention of any trade names or manufacturers in this report
shall not be construed as advertising not as an off~czai
indorsement or approval of such products or companies bythe United States Government
II
DISPOSITION INSTRUCTIONS
*, 1 ,-
Do not return ir to the orginator
I
UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE (When Date Entered)
REPORT DOCUMENTATION PAGE READ INSTRUCTIONSBEFORE COMPLETING FORM
1. REPORT NUMBER 2 GOVT ACCESSION NO. 3 RECIPIENT'S CATALOG NUMBER
MTL 86 - 164. TITLE (and Subtitle) TYPE OF REPORT & PERIOD COVERED
Ceramic Life Prediction FinalMethodology - Final Report 1 Jan 83 to 31 Dec 85
6 PERFORMING ORG. REPORT NUMBER
7. AUTHOR(s) 8 CONTRACT OR GRANT NUMBER(s)
L. R. Swank, J.A Mangels, J.C. Caverly / -
and R. K. Govila DAAG-77-C-0028
9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT, PROJECT, TASK
Ford Motor Company AREA & WORK UNIT NUMBERS
Room E-3172 SRL, P.O. Box 2053Dearborn, Michigan 48121 DOE IA No:DE-AI05-840R 21411
1 . CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE
Army M4aterials Techtology Laboratory May 1986Attn: SLCMT-ISC ,3. NUMBER OF PAGES
Watertown, Massachusetts 02172 6914 MONITORING AGENCY NAME & ADDRESS(it different from Controlling Office) IS. SECURITY CLASS. (of thie report)
UnclassifiedIS. DECLASSIFICATIONODOWNGRAOING
SCHEDULE
16. DISTRIBUTION STATEMENT (of this Report)
Approved for public release; distribution unlimited.
17. DISTRIBUTION STATEMENT (of the abstract entered In Block 20, if different from Report)
18 SUPPLEMENTARY NOTES
19 KEY WORDS (Continue on reverse .itde it necessary and Identify by block number)
Ceramic materials Silicon nitrideLife expectancy Glass ceramicsDisks Gas turbineTest and evaluation Brittle materials design
20 ABSTnACT (Continue on reverse side If necesery and Identify by block number)
DD I1jAN, 1473 EDITION OF ' NOV 5 IS OBSOLfTE UNCLASSIFIEDSECURITY CLASSIFICATION C-F THIS PAGE (When Data Enrered)
UNCLASSIFIEDISCU1ITY CLASM1PICATION OF TNIS PAGEQVIoW% Data SAIeftj
Block No. 20
ABSTRACT
Fast fracture and stress rupture data were collected on twomaterials, a sintered silicon nitride and a lithium-aluminum-silicate.The fast fracture data was presented graphically in the form of Weibullplots of percent failed versus failure stress. The stress rupture resultswere presented in tabular form. Photo-micrographs were presented toillustrate the fracture surfaces of fast fracture and stress rupturefailures.
A program of specimen development was coaducted. The objective ofthe program was to develop processing techniques to make it possible tofabricate integral shaft spin disks suitable for hot -pin testing asstress rupture specimens. The hot spin disk stress rupture results wereto be used to correlate experimental time dependent failure results withanalytical time dependent failure results..r
In the specimen development program several sets of experiments wereconducted. The molding experiments determined the relationship of moldingconditions to quality after molding, and the relationship of moldingconditions to quality after binder removal. The binder removal experimentexamined the relationships between pressure and heating rate on thequality of the part. The strength experiment examined the relationshipbetween strength, microstructure, and sintering conditions,
Also as part of the specimen development program the cooling patternof the green injection molded integral shaft spin disk was studied usingfinite element techniques. This study was conducted in order to determinea cooling method that left no isolated thermal hot zone in the disk.Experience had shown that such a zone caused a void due to shrinkage. Theresults of the study were presented as temperature contour plots of thedisk and die versus time.
A program of attachment development was conducted. The integralshaft spin disk required a new attachment desipn. It is a boreless design;therefore, the tie bolt and Curvic CouplingsT used in previous hot spintesting could not be used. Attachment designs utilizing high expansionplastics to accomodate the difference in thermal expansion betweenceramics and metals were developed in bench rigs and the hot spin rig.
P' ( T E
UNCI.AFS I [E
StCURITV CL ASSI1FCATION OF THIS PAE(W 7% D* trted)
FORWARD
This report presents the work completed during the period of January1, 1.s63, through December 31, 1985, on the "Methodology for Ceramic Life
Prediction Program," initiated by Mr. Robert Schulz of the Office of
Conservation, Division of Transportation systems, Department of Energy,and monitored by the Army's Materials Technology Laboratory under ContractNumber DAAG-46-77-C-0028. Funds for this phase of the work were provided
by the Department of Energy. This work was necessary in formulating amethodology for ceramic life prediction so that ceramic materials can be
used in high temperature structural applications. The piincipalinvestigator of this program was R. R. Baker, Ceramic MaterialsDepartment, Research Staff, Ford Motor Company. The technical monitorwas Dr. E. M. Lenoe of MTL. The authors wish to thank Drs. E. M. Lr-oe,
R. N. Katz, and Mr. G. D. Quinn of MTL for suggestions in carrying out Lhe
program.
TABLE OF CONTENTS
PagzeFOREWARD ................................................. i
I Introduction ........................................... I
II. Database Collection .................................... 3
III. Specimen Development .................................. 13
IV. Attachment Development ............................... 34
V. Summary ............................................... 50
VI. Conclusions ........................................... 52
TABLE I -Flexural Stress Rupture Results forBillet A-42 .................................... 54
TABLE II -Fast Fracture Strength Data for LAS atRoom Temperature .............................. 55
TABLE III -Flexural Stress Rupture Results for LAS ..... 56
TABLE IV -LAS Thermal Stability Tests Results ......... 57
TABLE V -Experimental Design Molding Experiment ...... 61
TABLE VI -Direction of Movement of the Variables toMaximize Quality after Molding .............. 61
TABLE VII -Direction of Movement of the Variables to
Maximize Quality after Binder Removal ....... 62
TABLE VIII -Experimental Design Binder Removal Experiment..62
i The objective of this program was to establish a methodologyfor predicting the lifetime reliability of structural ceramicmaterials in high temperature applications. -. The programconsisted of two interrelated parts: one the determination ofstatistical and time-dependent strength characteristics ofselected structural ceramic materials as a basis for analyticallife prediction; two, the design and hot testing of ceramiccomponents which are exposed to high enough temperatures to havetime dependent reliability. The data gathered in part one wasused to predict the time dependent reliability of the componentsin part two. The experimental time dependent reliabilitiesdetermined by the testing required in part two were compared tothe predicted time dependent reliabilities to verify theanalytical models.
As the program evolved additional tasks were undertaken inorder to support the original program objectives. Thedesign, test, and evaluation of ceramic to metal joints wasundertaken in order to support the program requirementfor a reliable attachment method for integral shaft spindisk. Also a specimen development task was undertaken in orderto meet the requirement for an integral shaft spin disk withproperties suitable for hot testing with time-dependent failureconditions prevailing.
Early work in the program was concentrated on gathering timedependent properties of two materials; Norton's NC-132 and Ford'shot pressed silicon nitride (HPSN) containing 3.5% MgO. Theprincipal time dependent property measured was the crack velocityexponent. This was done by three methods; double torsiontesting, stress rate testing, and flexural stress rupturetesting. Also fast fracture testing with precracked specimenswas used to determine inherent flaw size. Flexural stressrupture testing was used to determine the time to failure of thematerials under constant stress and temperature conditions. Thisearly work together with a procedure for measuring fracturemechanics parameters was documented by Govila I , ,2 3 4
The next phase of the program concentrated on developing andutilizing tensile testing to gather time dependent materialparameters. Fourteen tensile stress rupture tests were conductedat 10000 C, sixteen tests at 1200'C, and eleven tests at 13000 C.In addition some precracked tensile stress rupture specimensweie tested. The crack velocity exponent was determined at1200 0C from the tensile stress rupture tests. The material
1
I
used for these tests was NC-132. In addition the crackvelocity exponent and pre-multiplier at 13000 C, 1350 0 C, and14000 C were determined for NC-132 using double torsion methods.As a follow on to previous work, additional flexural stress rateand stress rupture testing was conducted on NC-132 and Ford's3.5% MgO HPSN. This work was documented by Govila 2 ,3 ,4, 5.
Silicon carbide was added to the program early in 1980. iCarborundum's sintered alpha silicon carbide was selected forextensive testing. Fast fracture testing was conducted todetermine the materials strength versus temperature. Eight Itensile stress rupture tests were conducted at 12000 C andseven tests at 13001C. The crack velocity exoonents weredetermined from these tests. Flexural stress rupture tests
were conducted at 1300 0 C and 14000 C. The crack velocity expo inents were determined at 13000 C and 14000 C from these tests.Extensive fractography was conducted on the failed specimensto document the causes of failure, flaw size, and flawlocation. This work was published by Govila 6 ,7 , 8
As part of the program's concinuing investigation ofanalytical and experimental methods in ceramic life prediction iaimed toward utilizing structural ceramics in pra ticalapplications, a component was selected for analysis. Thecomponent selected was the hub of a hot pressed ilicon nitride Iturbine rotor. The required geometrical, material, strength, andtime dependent data was supplied by MTL. A finite elementcomputer model was prepared for the disk from chis data. The
temperature and stress distributions, the fast fracture and thetime dependent reliabilities were calculated. The results werepresented in isostress and iscthermal plots for the combined
centrifugal and thermal loadings as well as isostress plots for Ithe centrifugal loadings. The reliability versus time forthe disk to 1000 hours was calculated and presented graphically.
The work was documented in a technical report by Swank 9 .
The data generated in the early phases of the program wasused to design a NC-132 bladeless turbine disk which wP3 tested
in a hot spin test rig. The disk was designed to fail due totime dependent mechanisms. Several disks were fabricated and anexisting test rig was developed to test the disks. Ten disks
were tested at steady state under the design conditions of U2300°F rim temperature and 50,000 rpm for periods of 0.20hours to 25 hours. Six disks failed due to time dependentfailure and four tests were suspended. An experimental failure
distribution was obtained for the ten disks and presented asreliability versus time. Three different data bases (doubletorsion, stress rate, and stress rupture) were used to
calculate relidbility versus time and results were compared tothe experimental results. The best correlation with experi-mental results was with stress rupture data. The resultsof this effort were documented by Baker et allOii
I2!
In H iNI III •• •• nu annnnlm n um nnI
The results of the NC-132 bladeless disk testing suiggestedrepeating the tests with another material to confirm the con-clusion that the stress rupture data base was the best data baseto use for calculating reliability versus time. Accordingly aproposal was nade to conduct - similar series of tests utilizingan integral shaft spin disk. This disk would simplify themechanical mounting if an appropriate attachment system could bedevelo)ed. The integral shaft disk attachment would be norereliabi than the tie bolt and face spline attachment used w.ththe NC-132 disk. In addition, it offered the possibility ofbeing able to start-up and shut-down the test several time3,something not possible with a face spline attachment. This wouldmean that tests could be conducted for longer periods of timeeliminating suspensions, and giving a better experimentalfailure distribution.
This report covers the last segment of the program, wherethe effort was directed toward repeating the bladeless difktesting. Initially the program planned on using existingmaterials for disk fabrication. These materials requiredextensive flexural fast fracture and stress rupture characteri-zation to build the data base required to design the disk andselect the test conditions. As the existing materials werecharacterized it became apparent that their development was notat a state were a successful correlation program could beexecuted. At that point the program was modified to include aperiod of specimen development to bring the materials up to apoint where a successful disk program could be conducted. Alongwith this work an attachment development program was conducted inorder to develop and verify a metal to ceramic attachmentsuitable for conducting long term tests of an integral shaft spindisk.
II. DATABASE COLLECTION
A. Silicon nitride data
For tl'v characterization of a silicon nitride material Fordsupplied billet A-42 of RM-20, a sintered silicon nitride. Thismaterial contained 8 weight percent yttria as the major sinteringaddiive. This material was found to be suitable for slipcasting gas turbine engi-_ components such as rotors and adia-batic diesel engine parts. Sintering was done in a nitrogenenvironment without any over pressure. For this stuuy, thematerial was cold pressed, nitrided and sintered in the form of arectangular billet.
Flexural test specimens 1.25 inches long by 0.25 inches wideby 0.125 inches thick were machined from billet A-42. All faceswere ground lengthwise using 320 grit diamond wheels, and the
3
II
edges .iere chamfered lengthwise to prevent edge effects. :orflexural strength evaluajion, specimens were tested in four-pointbend-ng in an Instron testing machine, Model 1125, using aspecially designed self-aligning ceramic fixture made from hot-pressed SiC 2 . The inner and outer knife edges of the testing
fixture were spaced 9.5umn and 19mm apart, respectively.
The fle,,-, 'al stress rupture tests at elevated temperatures(8000 C to 12000 C) in air wcre conducted in four-point b. nding 3using the self-aligning ceramic tixture and a rapid temperatureresponse furnace. The load was applied to the test Epecimenthrough a cantilever arm, deadweight assembly. The experimental
set-up was equipped with a microswitch to cut off the power tothe furnace and the timer -t the instant failure of '-he specimenoccurred. The total time to failure was recorded. An overall-.riew of the test set-up and (o:-.iiete details regarding the design Iand operation of the st ass rupture test rig are given else-where1
At room temperature, ten specimens from billet A-42 were Itested in four point bending to determine the fast fractucestrength. The statistical variation in fracture strength at roomtemperature is showr in Fig. 1. The fracture strength variedfrom a rinimum of 785 MPa to a maximum of 988 MPa with an averagestrength of 873 MPa, standard deviation of 79 MPa and a Weibullmodulus of 13. 1
IIIII
II
I
9080-
60 - - AV 'v 873 MPa'U5 40 - -899 MPG
--C - SD -79 MPa-20 m •13 0
10 0
5
2
I I .12M 400 600 800 1000
STRESS (MPa)
Fig. 1. Weibull probability plot of billet A-42.
Examination of the fracture surfaces revealed that all
failures in specimens tested at room tempe:rature were assecLatedwith porosity in the material. Typical i ilure occurring at a
porous region is shown in Fig. 2. These porous regions -iere
approximately 50 micrometers wide and 100 micrometers deep asshown in Fig. 2. Closer examination of the flaw site revealed
that the grain morphclogy of beta silicon nitride inside the
porous region appeared to be primarily needle shape (accicular)
and less accicular away from it, Fig. 2(b).
II
~I
M2P C -785 MPa, O. c
(a) ,. (Billet A42,RM20) .--
III
204mI
Fig. 2. Typical fast fracture surface. I
Stress rupture testing was carried out in order to determine
if the RM-20 material was susceptible to instability in the
intermediate temperature range from 6000 C to 1000 0 C. A total of
six specimens were tested in stress rupture mode and the results
are summarized in Table 1. At 8001C, one specimen was tested at
an applied stress levol of 413 MPa and sustained the stress for
over 300 hours without showing any signs of bending or failure.
A second specimen tested at 482 MPa, failed in 88 hours. The
fracture surface showed a porosity associated oxidized region,
Fig. 3. Away f rom lih fracture origin, the fracture surface
showed the smooth appearance of crack propagation indicating
trans-granular fracturoe, Fig. 3(C).
IIIU
800 C-482 MPa-88 h 204(a) (RM2O0, Billet A42) - A
Fig. 3. Str,7ss rupcur f)iactuu.- surface.
At 10000C, the material showed a distinctly differentbehavior than that seen at 8C0'()C. Two specimens were tested atthe low stress level of 344 IMfa. (One cu vd306 hours withoutfailure or bending while the second speciiien failed in 35 hours.Examination of the fracture sufae evealed the presence of a
I
locally oxidized region as the failure initiation source. A Ithird specimen was tested at 413 MPa. The time to failuredecreased significantly with failure occurring in one hour.
Examination of the fracture surface revealed failure occurring ata porosity associated oxidation pit, Fig. 4. A fourth specimen,tested at 482 MPa, failed in one-half hour. In brief, this shortstudy showed that Billet A-42 material has oxidation instability Iat 10000 C. It should be pointed out that in this material, allfailures were associated with porosity and this problem can beovercome by proper sintering conditions. Current work at FordMotor Company is being directed to improve this material.
4Uniform OxidationII ,ntito
100 C-413 MPa-1 h
( a ) (M20-A42) 40
amL
'i I
Fig. 4. Stres;s rupture fracture surface, 1000C.I
I
I
B. LAS Data
For the characterization of LAS material Ford supplied LASplate serial number A18C-7. The plate was machined into.125in x .25 in x 1.25 in bars and 1.00in x 1.00in x .696inblocks. All test bars were X-rayed to insure that they werefree from internal flaws. Fast fracture testing was conductedon thirty bars to establish the baseline strength. The Instronmachine head speed was 0.5 mm/minute. The characteristic modulusof rupture was 139 MPa with a Weibull modulus of 10.0. Thedistribution mean was 133 MPa and the standard deviation was16.0 MPa. The sample range was from 99 to 160 MPa.Examination of the fracture surfaces revealed that the majorityof the failures were surface originated. The statisticalvariation in fracture strength is shown in Fig. 5. Completestrength data for the fast fracture tests is given in Table II.
99
9080 -
60
40
UNE20I -
10 - cV"133 MPOo 139 MPa
SD 16 MPO
m - 10
2
50 100 500 1000STRESS (MPa)
Fig. 5. Weibull probability plot of LAS.
Ten LAS specimens were prepared for flawed beam stressrupture testing by precracking using a Vickers diamond pyramidindenter with a 1000 gram load. This controlled precrackingmethod introduces an approximately semi-circular crack of 48 to55 micrometers deep. Table III summarizes the flawed beam
9
U
results which are discussed in detail in the following para- Ugraphs.
Two precracked specimens, numbers 31 and 32 in Table III, 5were tested at room temperature in four-point bending in order
to determine the materials strength and to show the nature of
the crack front geometry (semi-circular,ellipsoidal, or other
form). The specimens failed at 93 and 114 MPa, respectively. UTypical crack front geometry -'owing the semicircular crack
front due to precracking with -000 gram indentation load is
shown in Fig. 6, which illustrates the fracture face of Ispecimen 31.
Precrack SitewPorc '84.
1000g Vickers IdentationZ 1000g Vick
Fracture Stress -" 93 MPa at 20 C 9F 0 1in 4-Pt. Bending Fracture Stress -1I16 MPa at 20 Cin~~i 4-Pt Bening in F
(-) . |.
A" IP
Fig. 6. SEM fractographs of LAS specimens.
(a) View of fracture surface, pre-cracked Ispecimen tested at room temperature.(b) Higher magnification view, PQR is
approximately the crack front boundary.
(c) View of fracture surface, pre-cracked
stress rupture specimen. Specimen was un-
loaded and fractured at room temperature.
(d) Higher magnification view of the pre- Icracked region PQR seen in (c). Note, the
fracture surface is smooth and similar tothat seen in (b). 3
Two precracked specimens were tested in stress rupture at
871 0 C under applied stress of 40 MPa, without failure for 93
and 50 hour, respectively. After the stress rupture testing was
10
|I
completed both specimens were tested in 3-point bending using amachine head speed of 0.5 mm/minute at room temperature in orderto reveal if any subcritical crack growth occurred. Neitherspecimen failed at the precrack site, but failed away from itsuggesting crack blunting or healing. Similar behavior wasnoted for the two specimens, numbers 35 and 36 in Table III,tested at the same applied stress of 40 MPa, but at a temperatureof 927 0 C. Both sustained 50 hour without failure.
Four precracked specimens were tested at 9821C and atan applied stress of 40 MPa. All sustained 50 hour withoutfailure. These specimens were chen tested in 3-point bendingat room temperature and only one failed at the precrack site(specimen 37, Table III) and the remainder failed away fromthe precrack site suggesting crack blunting or healing. Thefracture surface for the specimen 37, which failed at theprecrack site, is shown in Fig. 6. The semi-circular crackfront region PQR is visible and shows smooth re-propagation ofthe crack. No signs of any subcritical crack growth were seen inthis specimen. This behavior is similar to that seen in a fullydense LAS in an earlier study, Govila et. al. 1 3. Therefore, itis concluded that this LAS does not undergo creep deformationat 982°C under an applied stress of 40 MPa as indicated byflexural stress rupture tests. It is quite possible that thematerial may undergo creep deformation if the applied stress isincreased.
Application of LAS as a structural ceramic materialgenerally requires long term thermal stability. To evaluatethe thermal stability of the Ford LAS, sixteen blocks wereprecision ground from the A18C-7 plate. The blocks were dividedinto four groups. Five for thermal stability testing at 1600'F,five for testing at 17000 F, five for testing at 18000 F, and oneblock to serve as a control. The blocks were measured in atemperature controlled room, with the temperature held at680 F, plus or minus 0.50 F. The pattern of measurements is shownin Fig. 7. The xyz axis was identified on each block by asmall chamfer on the 0-0-0 corner. Along the x and y axis theblocks were measured in five places, and along the z-axis theblocks were measured in the center. The reason for thelimited measurement in the z-direction was that the surfacenormal to the z-axis was used to rest the block on when it wasin the oven; therefore, it was considered inappropriate toconsider the z-direction as critical to the experimentalresults.
Electric ovens were used to conduct the thermal stabilitytests. Samples were removed from the ovens at 285, 500, and 1000hours for measurement. The results are shown in Table IV. Inaddition to dimensional measurements the blocks were weighed tofour decimal places in grams and the results are shown in thetable. The sample range was determined for each dimension and is
shown on the table.
Review of the data and comparing the results of the blocks Ithermally soaked versus that of the control block, indicate thatthe LAS was thermally stable over the temperatures and times
Itested.
I
12 I
III. SPECIMEN DEVELOPMENT
A. Introduction
The objective of the specimen development part of theprogram was to develop the methodology to fabricate the IntegralShaft Spin Disk (ISSD), Fig. 8., using an injection molded,sintered reaction bonded silicon nitride (IM-SRBSN). The ISSD'swere to be used in a series of stress rupture tests conducted inthe hot spin rig to provide additional data for the verificationof time dependent failure theories.
4.87"DIA.
- 6.82"
Fig. 8. The integral shaft spin disk (ISSD).
Ford has had considerable success with the injection moldingprocess for both thin and thick cross-section components as shownin Figures 9, 10, and 11. Relative to thick cross-sectioncomponents, Ford has demonstrated that turbocharger and AGTrotors can be molded without internal voids or externalcracks. Respectable processing yields have been obtained forturbocharger rotors. It has also been demonstrated thatthese thick cross-section components can be processed throughthe binder removal process without the creation of additionalvoids, although the processing yields are much lower.
13
fI53
TI
141
~'%rnil
Fig. 11. Ford injection molded components having thickcross-sections.
Ford has also had considerable success in the development ofsinterable silicon nitrides of which SRBSN is one type. Duringthe course of these developments, Ford has gained experience innitriding and sintering technology as well as in materialcharacterization of both fast fracture and time dependentproperties. Ford has also ecsveloped the technological expertiseto tailor a material to a particular property requirement.
The individual processing steps required for the successfulfabrication of an ISSD have been demonstrated. However, furtherdevelopment work was required to improve the process consistency,improve the overall process yields at each process step, and tofurther improve the properties of the IM-SRBSN material. Themajor development step was the scaling of the process toaccommodate the larger ISSD geometry.
15
I
The specimen development work was divided i,,to two majorareas: (1) injection molding, and (2) sintering-property dv'elop-
ment. Fig. 12. illustrates the specific processing stepsincluded in each area. The experiwental program focusedresearch on the knowai problems, which were identified prior theinitiation of the contract, and were known to affect thefabrication of a large, thick cross-section component such as the
ISSD).
INJECTION MOLDED SRBSN PROCESS FLOW SHEET
POWDER PREPARATION 3MIX POWDER WITH ADDITIVES 3
MIX POWDER WITH MELTED BINDER SYSTEM 3INJECTION M0.DI
REMOVE B',IDER FROM COMPONENT
NITRIDE TO FORM S13 N4
SINTER TO FULL DENSITY
4,ICFARACTERIZE PROPERTIES
Fig. 12. Injection molding process flow sheet. 3The following sections will describe the results of
processing experiments designed to attack specific problems,Since the program was terminated prior to the completion of the
planned program, these individual experiments stand alone. Thetasks which would have brought the experiments together,
resulting in the successful fabrication of an ISSD,
were not funded.
B. Injection molding experiment 3A set of statistically designed experiments was performed to
relate the effect of four injection molding variables tocomponent quality atter molding and after binder removal. It wasshown that the magnitude of these variables directly effect thequality of the components after binder removal. The results are
163
consistent with a general, qualitative model relating thequality of the component to the stress state developed in thecomponent during molding.
It has been demonstrated that a number of components ofdiffering geometries can be injection molded. Visual inspectionsafter molding show that all of these components can be moldedwithcut visual defects (such as cracks); however, the moldingyield may vary with the particular geometry. It has also beendemonstrated that in general these components will developcracks during the binder removal process. The severity of thesecracks appear to be a function of the component's geometry. Thecracks generally occur in locations having severe sectionchanges. These locations would be expected to be "high" stressareas.
Many discussions have centered on the question of thepresence of residual stresses within the molded article andwhether the stress state of th,_ molded article causes crackingduring 'Ander removal. It is generally agreed that changes inmolding conditions should affect the stress state of the part;however, no experimental technique has been identified which canmeasure the residual stresses in a molded ceramic article. Thepurpose of the injection molding experiment was to determine ifchanges in molding conditions result in changes in the observedcracking after binder removal.
This experiment evaluated the effect of four injectionmolding variables on part quality after molding and after binderremoval. The injection molding variables were die temperature,injection pressure, material temperature, and hold time. A24-1 fractional factorial experimental design1 4 , shown inTable V, was employed. The experiment was repeated for twocomponents, each representing some feature of the ISSD. Anumber of responses were measured and are summarized in Table VI.
Two responses are considered critical after molding:(1)density, and (2)the number of x-ray indications. The qualityparameters relating to cracks are only important after binderremoval. Table VI summarizes the direction each variablemust- be changed from the average valuc to effect an improvementin the component quality after molding.
The responses of importance after binder removal are thoserelated to cracking. Three types of cracks were common for thecomponents molded in this study. While responses of these typeof defects were analyzed afLer molding, it was determined thatthese responses after binder removal were the ones which werecritical to the production of quality components. Data wasobtained after the binder removal process. Table VII shows thedirection each variable must be changed from the average value
17
II
in order to improve component quality, that is minimize each typeof crack. 3
The nature of the experimental design resulted in thedetermination of the effect of the four variables on the variousresponses. However, because this was a fractional factorial mexperimental design, the interaction between variables could notbe determined. This design was used as a screening experiment toidentify the variables having a major effect on the responses and1n identify the direction in which to change these variables infuture experiments.
The results generally indicate that all four injectionmolding variables, die temperature, injection pressure, materialtemperature, and hold time should be reduced in order to improvethe component quality. The resulos all generally appear to beconsistent with a shriNkage/stress modul. They show that Iconditions which are thought to a yield 'w stress state in thecomponent also result in a high quality component after moldingand binder removal. Th-y also show that the molding conditionsdirectly effect the quality of tlc: component after binderremoval. 3
C. Binder removal experiment
A set of staristically designed experiments was performedto determine the effect of three binder removal processingvariables on the component quality after binder removal. Theresults indicate that complex components can be successfullyprocessed through binder removal over 10 times faster with the Iuse cr a pressurized binder removal atmosphere. The resultsalso show simple components can be successfully processed atthese high rates without the necessity of the pressurizedataosphere.
Binder removal is the most difficult processing -tep in theinjection molding process, and it has probably been the least Istudied. A large number of processing variables exist whichcould effect the quality after binder removal. This experimentstudied three of those variables.
A set of experiments were designed to determine the effectof three processing variables on the component quality afterbinder removal. The variables investigated were: (1)the pressure of the binder removal atmosphere, (2) therate of t.aperature rise, and (3) the complexity of the Igeometry of Lie component. A full factorial 23 experimentaldesign, shown in Table VfIT was employed to determine primaryeffects as well as interaction etfects. 3
18 I
Tle enalysis of the binder removal results is presented inTable IX. The results indicate that higher binder removalpercentages are obtained at high pressure than under low nressureconditions. This was found to be true for both the large andsmall component. The results also indicated that the largecomplex component exhibits higher binder removal percentages thanthe small component. This result is surprising and may be dueto thp fact that different materials were involved inthe fabrication of the large and small components.
The results were re-analyzed using only the large comp-,nent.These are summarized in Table X. These results indicatethat the maximum binder removal occurs at high pressure and lowheating rate. An interaction between these two variables ispresent, as illustrated in Fig. 13.
PERCENT BINDER REMOVAL VS PRESSURE
98 -2
94 - - -
Z
S90w
PRESSURE
Fig. 13. Percent binder removal versus pressure.
The quality of the components are determined by the nuaTherof cracks observed by visual inspection. The results Fo7: thecomplete matrix are presented in Table XI in terms of totalcracks. The results indicate t'-.at in order to minimize componentcracking the pressure shoi.ld be at the high level. These resultsalso show that small, simple components are less susceptible tocracking than large, complex components. There is astrong pressure-size interaction which is illustrated in Fig. 14.
19
1UI
NUMBER OF CRACKS VS PRESSURE
U %%U.o "- 4of5 0
Z 0 MALCOMPONENT
PRESSURE
Fig. 14. Interaction of pressure and component size on the Inumber of cracks after binder removal.
The cracking results were re-analyzed for the large 3component only. These results are summarized in Table XII.All results show that the amount of cracking in large componentscan be minimized by performing the binder removal at highpressure. Pressure-heating rate interactions are also present,as illustrated in Fig. 15.
NUMBER OF CRACKS VS PRESSURE 312
. en a
00 6
I
PRSSUR
200
PRESSUR
removal times can significantly be reduced. For largecomponents, binder removal times were reduced by a factor of sixwith no reduction in quality, as measured by percent binderremoval or number of cracks. The mechanism for the benefi-cial effect of pressure is unknown.
D. Microstructure development
The analysis of the microstructures of a number ofcompositions within the yttria/alumina system showed that acritical temperature exists above which exaggerated grain growthoccurs and below which a uniform microstructure can be obtained.Sintering time above the critical temperature was identified asthe principal parameter contributing to excessive grain growth.
Prior to the receipt of this contract, a number ofcompositions in the yttria/alumina system were dry pressed,sintered and characterized. The purpose of this study was todetermine the processing parameters responsible for thedevelopment of the unique microstructural features responsiblefor the failure origins within these materials. Attempts weremade to quantify the microstructure obtained by SEM analysis.The maximum grain size and the number of grains exceed-ing a particular size were determined.
The microstructure of the yttria/alumina SRBSN materials wasdetermined to consist of a bi-modal type distribution of needleshaped grains having length to diameter ratios of about 5/1 to10/1. A typical micrograph is shown in Fig. 16. The strengthof materials having this type of structure is in the 80-95 Ksiregion. A plot of strength versus porosity ,Fig. 17, shows thatthe strength is independent of porosity. That means another typeof defect is the strength controlling parameter. This wasdetermined to be the large, needle shaped grains.
21
I
I
IiI
28VX0e I 841
Fig. 16. Typical microstructure of the baseline SRBSN material
processed above tne critical temperature. The lowmagnification photo illustrates the exaggerated grain
growth, while the high magnification photo illustrates
the grain growth of the overall structure.
III
I
103 1 I 1 I_ "BASELINE SINTERINGco C: ONDITIONS
Zi -
U,
Ii I I I
-. 01 0 .01 .02 .03 .04VOLUME FRACTION POROSITY
Fig. 17. Strength versus porosity for the baseline SRBSN material
processed above the critical temperature.
Examination of the structure of a number of compositions
processed using a number of sintering conditions indicate that
grain growth occurs above a particular, critical temperature.
The data for maximum grain size and the number of large grains
per unit area, Figures 18, and 19 both illustrate this finding;
furthermore, these figures show that time at temperature above
the critical temperauure is the significant parameter affecting
grain growth.
23
120 0 o
100- TEMP = CRITICAL
N
z60-
: 40 0 1200
4. ITEP>TEMP 30
TIME !Fig. 18. Maximum grain size versus time. Open symbols1
indicate samples processed above the criticalItemperature (TemPc); the solid symbols indicatesamples processed below the critical temperature.
III
40
I
TEMPC 2 CRITICALTEMPERATURE
z4o
0.
(n 30 TEMP>TEMPcz7% 0
W 20 00
_J0 0U. 0 0
S/TEMP<TEMPcW 00
z TIME
Fig. 19. Number of large grains per unit area versus time.Open symbols indicate samples processed abovethe critical temperature (Tempe); the solidsymbols indicate samples processed below thecritical temperature.
These results define a set of acceptable time-temperaturesintering parameters required for obtaining a uniformmicrostructure in a sintered reaction bonded silicon nitride.These results were independent of composition and will serve toguide the processing of new compositions.
E. Strength experiment
Dry pressed SRBSN samples processed using sinteringparameters designed to produce a uniform microstructuredemonstrated strength improvements of about 35 percent over thebaseline established with maximum strengths of 133 Ksi with aWeibull modulus of 22. The strength of injection molded SRBSN,processed using these conditions was 75-88 Ksi. The bulkmicrostructure of the injection molded material was identical tothe dry pressed material, but the failure origins were different.They were identified to be metallic inclusions of silicon andiron.
25
II
The strength of SRBSN has been limited by microstructuralfeatures, especially large needle shaped grains. The workdescribed in the previous section points out processingtechniques which can minimize the exaggerated needle growth inSRBSN materials. This section describes experiments to improve ithe strength of SRBSN compositions through microstructureoptimization.
Dry pressed compositions of SRBSN were sintered using four Isets of processing conditions where the sintering temperature andtime were varied. They included a baseline where the sinter-ing temperature was above the critical temperature for grain igrowth and the time was long. The others included short timeat high temperature, long time at low temperature (below thecritical temperature) and an intermediate time at anintermediate temperature (near the critical temperature).These last three conditions were designed to generate a micro-structure free of the large needle shaped grains. Thedensity, strength and microstructure were studied for thefour processing conditions. Selected compositions of injectionmolded SRBSN were also studied using selected processingconditions. The results from the two fabrication techniques Iwere compared.
The strength results are summarized in Table XIII.The results show that the strength of the dry pressedmaterial is optimized when the sintering temperature was nearthe critical temperature. Here, the density was maxi-mized while the microstructure remained fine and uniform. IThe fracture origins of these samples could not be identified,but a plot of strength versus volume fraction porosity, Fig. 20,for a number of compositions processed using these conditionsshow that the strength controlling defect is probably porosity. ISamples processed in experiment 4 exhibited strengths of 133 Ksiand a Weibull modulus of 22. This represents a 35 percentincrease in strength over the baseline value.
lIIII
26
I
10;5
IOPTIMUM SINTERING- {/"CONDITIONS-
(BASELINE SINTERING
CONDITIONS
10 I I I I-. 01 0 .01 .02 .03 .04
VOLUME FRACTION POROSITY
Fig. 20. Strength versus volume fraction porosity fora SRBSN processed below the critical sinteringtemperature. The observed strength increase isdue to an optimized microstructure.
The injection molded SRBSN was processed using the baselineconditions and those of experiment 4. The strength of the moldedmaterial did not improve with the new sintering conditions.Analysis showed that the bulk microstructure was identical tothe corresponding dry pressed material, Fig. 21, butthe fracture origins were different. SEM analysisshowed the strength controliing defects to be metallic inclu-sions of silicon and iron, Fig. 22.
27
II
~I
' II
2 3I
using th e tiu siteig yce
II
Ii
Iusing the "optimum" sintering cycle.
28
20000
5 I
1000F
ENE-RCY(kQ )
Fig. 22. Typical fracture origins of the injection
molded SRBSN. Inclusions are compounds ofiron and silicon.
F. Integral shaft spin disk (ISSD) mold design
The most critical aspect of injection molding a thick crosssection component, like the ISSD, is the solidification behavior
of the molding material in the die cavity. If solidification is
29
II
not controlled properly, internal voids and cracks occur. 3Heretofore, the control of the solidification behavior has beensomewhat by chance. This important factor has not been con-sidered in the design of tooling for injection molded siliconnitride components.
The solidificarion behavior of the molding material iscontrolled by the temperature of the material. The molding Imaterial solidifies with decreasing temperature. The lastvolume of material to solidify tends to have a void in it due toshrinkage. This volume is the last part to cool. A properlydesigned injection molding tool controls the temperature distri-bution with time so that the last volume to cool to the solidi-fication temperature is in the sprue, and not in the part. I
The purpose of this study was: (1) to apply conventionalfinite element techniques to model the temperature distributionwith time of the injection molding material in an ISSD config- Iuration, (2) iterate the tooling design using the modeluntil the desired temperature distribution was obtained, and(3) fabricate the optimum tooling and confirm the model byproducing ISSD components.
An axisymmetric finite element heat transfer program withtransient temperature capabilities was used to study the temper- Iature distribution with time of the injection molding material in
the ISSD die cavity. The finite element model is shown in Fig.23. The model simulates three parts of the ISSD injectionmolding die, the base, the cone section, and the cap, as well asthe molding mix. The molding material includes two sections, theintegral shaft spin disk, and the sprue. After molding the sprueis machined away, but at molding time the two parts are as one. U
IBASEI
CONE IZ -AXIS
SPRUE MOLDING INTEGRAL SHAFT 3iMATERIAL SPIN DISK
Fig. 23. ISSD injection molding die finite element model. 330 U
External to the die itself is a cooling system which isused to control the molding mix temperature as it is cooled. Inthis study air impingement and water cooling systems werestudied. The hardware needed to implement the cooling systemdoes not to be modeled, since it appears in the finite elementmodel only as heat transfer coefficients.
The cooling of the molding mix and die was simulated 4ithheat transfer calculations assuming a uniform initial die temp2r-ature was 1000 F, and a uniform initial molding material was215 0 F. These are reasonable assumptions since the die and themolding material are initially heated and their temperatures aremonitored before molding a part. These are the target tempera-tures. The molding material must be heated before molding sothat it flows freely, and the die must be at the proper tempera-ture so that the incoming molding material does not stick to thedie, interfering with the flow of the molding material into thedie. The die is filled in a fraction of a second which furtherjustifies uniform initial temperatures.
The temperature distribution two minutes after the die isfilled is shown in Fig. 24. The die is being cooled with airimpinging on the acute cone, and the die is sitting on theheater at a temperature of 180 0 F. There is a hot spot of 200°Fin the middle of the disk region. If this cooling patterncontinued a shrinkage void would result.
Fig. 24. Temperature distribution two minutes after filling.
Figure 25 illustrates the temperature distribution 14minutes after filling assuming the die has been removed from theheater and sat on a large plate whose temperature is 750 F. Themiddle of the disk region is warmer than that of the rest of thepart and it will solidify last, leaving a shrinkage void.
31
IIII
___ II
Fig. 25. Temperature distribution 14 minutes after filling, Idie removed from the heater.
This is the situation to ue avoided. The die must be lefton the heater, so that heat can be delivered to the die andmolding material establishing the temperature distribution shownin Fig. 26. Here the temperatures are uniformly decreasing down
the length of the part. Heat is flowing in from the base andout the acute cone. Continued cooling with this distributionwill not leave any hot spots isolated in the interior of the
part.I
II
Fig. 26. Temperature aistribution 10 minutes after filling.
To maintain a uniformly decreasing temperature distribution 3the heater temperature must be reduced while cooling is main-tained on the cone. Fig. 27. illustrates the die at twentyminutes assuming the heater temperature was reeuced to 160°Fafter ten minutes. The temperature distribution is stilluniformly decreasing while the temperatures have dropped.
I
I
0w0
Fig 27. Temperature distribution 20 minutes after filling.
Figure 28 illustrates the temperature distribution 30minutes after filling assuming the heater temperature wasreduced at 20 minutes to 1401F. The temperatures have nowdecayed to the point where the molding material has solidifiedand no hot spots have been left behind in the molding material'sinterior. At this point in time the die could be removed fromthe heater for some further cooling and the part could t-removed from the die.
Fig. 28. Temperature distribution 30 minutes after filling.
Several temperature distributions versus tIme were calcul-ated for cooling systems where the heater temperature wasconstant with time, but none of these gave the desired result ofleaving no isolated hot spots in the molding mix interior. Itappears that heater temperature control is necessary to achievethis result.
33
I
IV. ATTACHMENT DEVELOPMENT
A. Introduction iTwo types of ceramic to metal attachments were developed to
allow hot spin testing of integral shaft spin disks. Both Iattachments used the principle of a high thermal expansionplastic sleeve trapped in a relatively constant volume createdbetween ceramic and steel shafts. One attachment uses a steellock nut to trap the plastic, and the second uses a special type Iof thread on the ceramic shaft to trap the plastic.
Both these patented attachment designs were successfully andextensively tested in bench test rigs using several plastic andceramic materials. Spin rig tests of the metal lock nutversion ran slightly more than 80 hours in the hot spin rig.
Two further versions of the screw on type attachment arepresently under construction and a new version of the lock nutattachment is presently in test.
Early life prediction tests were conducted using spin disksmachined from hot pressed billets of silicon nitride. Thesedisks were mounted on a steel shaft using curvic couplings and an Iair cooled tie boltl0 , 1 . The use of curvic couplings to attacha relative thin disk require a metal to ceramic attachment in ahot environment and the attachment may have high stresses. Forlife prediction testing this presents the potential of a failuredue to the ceramic to metal attachment instead of a failure inthe ceramic due to time dependent characteristics of the ceramic.
The casting of a spin disk, Fig. 29, with a six inchintegral shaft made possible an attachment located in the bearingcompartment. This is a relatively low temperature region and the
attachment would have low stresses. The disk would run in an Ienvironment where it would be subjected to only thermal andcentrifugal stresses. This would make a much more controlled
test for life prediction testing. I
lIII
34 I
Fig. 29. Integral shaft rotor.
After some deliberation a design was conceived which used toadvantage the relatively high thermal expansion rate of variousplastics along with their softness compared to ceramics. Thedesign concept lent itself to two configurations which weretermed the non-threaded high expansion lock attachment and thethreaded high expansion lock attachment. The non-threadedlock was the first design to be developed and will be reportedfirst. The basic principle of operation is the same for bothdesigns; however, the method of accomplishing this is different.
B. Non-threaded design high expansion lock
The most fundamental problem in joining ceramics with metalsis their greatly differing thermal expansion rates. A ratio of 5to 1 for steel and ceramic are not unusual. If joints of steeland ceramic are assembled and locked at a common temperature andthen heated, strains quickly develop in the ceramic which cancause failures at very modest temperature excursions. For thisreason attachment of metals to ceramics in areas subject totemperature variations is difficult. The idea of using ahigh expansion plastic sleeve, trapped in a volume formed byconcentric shafts was first tested in a simple bench test rigshown in Fig 30.
35
I
OUTER STEL SHAFT iINNER STEEL SHAFT
PLASTIC SLEEVEFig. 30. First design non-threaded high expansion
lock bench test rig.
This design trapped a nylon sleeve in the relatively
constant volume chamber formed by inner and outer steel shafts Iand two end caps. Nylon, with linear expansion rate of 45 to55x10"60F, would with increasing temperature, increase in volumemuch faster than the entrapping chamber. This volume
expansion would create high pressure in the nylon sleeve,locking the two steel shafts. Steel was used for both shafts inthe first design to simplify the fabrication problems. Also, tosimplify the fabrication problem, the nylon sleeve and the innersteel shaft were slip fit at room tempeiature. This provided nolocking at room temperature, a feature required for a successfulattachment. The design would demonstrate the principle and Iroom temperature locking could be incorporated into the designlater.
The two end nuts, and the two shafts were made of cold irolled steel. The sleeve was nylon and the dimensions were sizedto give the nylon sleeve a slip fit over the inner shaft and
inside the outer shaft. The end nuts were screwed down tightly Iagainst the nylon. This assembly was heated to 180°Fmetal temperature and the shaft was pushed axiallythrough the assembly. The break away load was 3200 poundsand constant motion was sustained at 2000 pounds. This test Iproved that the loads necessary to hold a rotor shaftassembly could be generated.
To accomplish room temperature locking, a new nylon sleevewas made that had a bore diameter 0.003 inches larger thanthe 0.750 inch diameter steel shaft. The outer diameter was
0.003 inch larger than the 1.000 inch inner diameter of theouter steel shaft. In order to assemble the parts, the innershaft and the sleeve were soaked in a dry ice and alcohol bath
until the sleeve had shrunk to a tight fit to the inner shaft Iand a slip fit to the outer shaft. The parts were thenassembled and allowed to return to room, temperature. Axial 3
36
U
break away load at room temperature was 800 pounds and thesliding load was 500 to 600 pounds. The parts were thenheated to 300F in order to simulate rig temperatures and re-cooled. Examination of the pafLs at 1oom temperature dis-closed that the steel outer shaft had plastically yieldpA due tothe high pressures generated. These initial bench tests provedthat the idea would work but that the nylons' expansionrate was too high, generating a very rapid internal pressurerise when heated. Also, the inner steel shaft caused theentrapped volume to be smaller for a given temperature than aninner ceramic shaft would thus making the pressure rise evengreater.
A new bench rig was designed that more closely simulated aceramic to steel attachment as shown in Fig. 31. The outersteel shaft wall was increased in thickness for additionalstrength and a silicon nitride inner shaft, of NC-132, wassubstituted for the steel inner shaft to reduce the force on thenylon sleeve's internal diameter.
LOCK NUT\ STEEL SHAFT
CERAMIC SHAFT
PLASTIC SLEEVEFig. 31. Second design non-threaded high expansion
lock bench test rig.
The test, after assembled parts came to room tempera-ture, produced no motion between the parts at 3600 pounds axialload. This high load indicated that nylon was unsatisfactory asa sleeve material, since nylons' expansion rate would causeexcessive pressures on the steel outer shaft. at anticipatedoperated temperature.
A new sleeve of Celanese Plastic's Celcon GC-25 wasfabricated. This plastics expansion rate was 22xlO- 6/oFas compared to 45 to 55xlO-6/OF for nylon. The first testusing Celcon GC-25 produced a room temperature break away loadof 3400 pounds and at 140OF the load rose to 4600 pounds.As a check against a plastic deformation of the steel, theassembly was heated to 3001F and cooled. No deformationoccurred and the assenbly remained tight at room temperature.
37
II
This successful test lead to an invention disclosure and apatent application. Patent number 4,485,545 was awarded December4, 1984. I
One further area foi improvement in the attachment designwas to increase the service temperature of the plastic sleeve.A new product of Dupont, Vespel SP-22TM , was selectedwhich has a service temperature of 500OF and an expansionrate of 15x10-6 to 20x10- 6/OF. Vespel SP-2 2TM is a 25%graphite filled polyimide resin with a bulk modulus of 475,000 Ipsi. Its higher service temperature will give more temperatureflexibility in the operation of high expansion locks.
With this last refinement to the attachment scheme it wasdecided to test a rotor in the hot spin rig and machining of ahigh expansion lock was begun on an integral shaft disk. Across-section of the assembly is shown in Fig. 32. Figure 33 isa picture of the detail parts in the assembly plus the wrenchnecessary for assembly.
CERAMICROTOR
TURBINEBEARING AREA SSTEELI
LOCK NUT SHAFT
OIL GALLEY VSEVESPEL
CERAMIC SLEEVE SPACERSHAFT
Fig. 32. Cross-sectional view of non-threaded highexpansion lock rotor assembly. I
IIII
38 3
MACHINED ROTOR
SHAFT
OCK4 --- VESPEL SLEEVE
NUT i. . SPACER
SASSEMBLY WRENCH**
Fig. 33. Non-threaded high expansion lock rotor test parts.
The assembly was balanced to 0.001 inch-oz and installedinto the hot spin rig. The initial testing consisted of nineshort runs. After each run the rotor assembly was removed andchecked balanced and visually inspected to see if any separationof the parts occurred. This series of tests proved theassembly to be stable, so an endurance test featuring increasingspeeds and temperatures was begun. This test ended after 80hours of testing when the rotor failed at 45,000 rpm and 1800°Frim temperature. The failure occurred in the ceramic partat the at the shoulder where the 0.750 diameter portion of theceramic shaft blends into the 1.310 inch bearing diameter. Theattachment section was still whole and had to be machined apartfor inspection.
Careful inspection of the failed parts suggested thatcontact with the steel locking nut in the shoulder area where thefailure occurred may have been a contributing factor. Asimplifying redesign using a VespelTM sleeve with a lockingthread ground on the sleeve was fabricated and was tested.Figure 34 shows the part which takes the place of both thesleeve and lock nut shown in Fig. 32.
39
III
.1 II
Fig. 34. All VespelTM combination lock and sleeve.
The all VespelTM combination lock and sleeve was evaluatedwith a Kyocera SN-220M integral shaft spin disk. Testing wasconducted at 1500OF rim temperature. The initial test was at2000 rpm for a period of one hour. The assembly was theninspected to make sure that no relative motion between the rotorand steel shaft had occurred. Testing was then conducted in 5hour intervals, starting at 5000 rpm and going to 25000 rpm, in5000 rpm increments. The assembly was inspected after each 5000rpm increment. After these incremental tests, twenty and one-half hours of testing were conducted at 30000 rpm at a rimtemperature of 15000 F. The combination lock and sleeve performedsuccessfully.
C. Threaded design high expansion lock IIn the non-threaded design, Fig. 32, the function of the
lock nut is to form one of the entrapping volumes' walls. The Urotor shaft is not axially locked in position by any mech-anical means but it is frictionally locked to the entrappedplastic sleeve. The spacer plug is not necessary to the designbut was used to make available hardware usable. The designthus requires two parts in addition to the two being joined.
In order to simplify the design further the function of the 3lock nut would have to be assumed by a shoulder on the ceramicshaft and the shaft would have to be locked to the steel rotorshaft. To accomplish this a threaded ceramic to steel attachmentwas designed, Fig. 35. In this design the rotor to metaljoint requires only one additional part to accomplishattachment and that is the plastic sleeve. In assembly therotor and sleeve are cooled and screwed into the rotor shaftuntil the sleeve is axially trapped. This design was granted
a patent, number 4,499,646, in February 1985.
40I
STEELVESPEL ROTOR
SLEEVESHF
CERAMIC ROTOR OIL GALLEY
Fig. 35. Threaded ceramic to steel attachment.
Parts for a hot spin rig test were fabricated. Theindividual parts and the assembly are shown in Fig. 36. Thetest was conducted at low speed, 2000 rpm and the rim tempera-ture of the rotor set at 2000OF as measured by a radiation pyro-meter. After 15 minutes of testing a failure occurred in thethread portion of the rotor shaft. Failure analysis of theparts suggested that the metal had bridged the front and rearface of two separate threads and put the thread section intension, causing a failure in the thread root. The threadsused in this design were standard 60 degree vee threads. Twopossible problems were considered as causing the failure: (1)excessive high pressures in the plastic sleeve which would causeexcessive tensile loads on the ceramic threads, and (2) adimensional mismatch between the ceramic and steel threads whichwould cause excessive stresses in the ceramic threads. A benchtest was designed to try and duplicate the hot spin rig failureand allow a convenient way to test changes in thread designs toovercome the problem. The bench rig is shown in Fig. 37.
41
III
a I,
I
J II
Fig. 36. Threaded attachment test parts and assembly.
CERAMIC VESPEL SLEEVE
• " I
Fig. 37. Bench test rig screw thread attachment.
42
In order to also generate some information on the pressurerise inside the assembly. Three strain gages were placed on theouter diameter of the steel part. The gages were temperaturecompensated and were thermally cycled three times to guaranteestable readings. The ceramic part was made from GTE's SNW-1000, this material was used because of its available. The steelpart was cold rolled steel. A thermal cycle test produced afailure on the first cycle at 2100 F. metal temperature at aninternal pressure of 1400 to 1600 psi. The failure occurred attemperature just a few degrees above the normal oil temperatureused in the hot spin rig and the test appeared to give a verygood correlation to the failure of the actual rotor test. Thefailed assembly is shown in Fig. 38.
Fig. 38. Failed bench test rig.
The pressure produced during this test was not high enoughto overload the ceramic threads in tension; hence, the problemappeared to be due to axial thread interference andsubsequent tension in the ceramic caused by expanding metal. Athread design was sought which would prevent this by assuringsufficient clearance on the unloaded side of the thread. Thiswas provided by a buttress thread whose basic form is show inFig. 39.
43
I.0075 ROOTAND TIPCLEARANCE
CONTACTSSURFACE
METAL BACK FACECLEARANCE
BOLT LOAD CERAMIC
Fig. 39. Buttress thread form. I
Included in this second screw thread rig test was a tapered
VespelTM sleeve. The outside sleeve diameter was tapered towardthe threaded end of ceramic. This change was to aid assembly.With a straight cylindrical sleeve, assembly was difficult
because the parts, as cooled, are slightly under a line to linefit. Assembly has to be accomplished quickly or the joint tends
to lock up while half assembled. A 2 degree taper on the outsideof the sleeve enables the assembly to be almost completely
secured before the mating parts touch and begin to lock up. Thismodification eased the assembly by permitting most of the
threads to engage before the VespelT M sleeve came in contact with
the steel and began to expand.
The steel shaft was strained gaged as in the previous test.An attempt was made to assemble the parts. However, the
parts began to lock up before the threads were fully engaged.Torque was applied to the ceramic head to drive the VespelTM
fully into the sleeve cavity but the SNW-1000 material failed
at the thread as shown in Fig. 40. Since this material with itsroom temperature modulus of rupture of 95000 psi, wasn't going
to stand up to the rigors of assembly, a new ceramic part was
made from NC-132, a hot pressed silicon nitride.
IiIII
I
*
Fig. 40. Buttress thread test rig, SNW-I000.
This part, along with the sleeve and strain gaged outersteel shaft, are shown in Fig. 41. They weresuccessfully assembled. The thermal cycle test began after theassembly had normalized at room temperature for 24 hours. Beforeassembly the strain gage pots had been zeroed with the steelcase unstrained. The pots were not adjusted during therest of the test. The gage readings were noted before theassembly was heated and are shown in Table XIV. The assemblywas slowly heated to 290°F and then to 341°F where it wasallowed to remain overnight. The readings were recorded, theassembly cooled to room temperature and the final readingstaken. The gages readings were averaged and converted to stressassuming the Young's modulus was 30xlO 6 psi. The stress valueswere then used to approximate the internal pressure using Eqn.1.
45
III
I
Fig. 41. NC-132 Buttress thread rig parts. IP b2__
aT -b2a2 (+ (1)
where
aT - hoop stress(psi)P - internal preszure,psib - External radius,inches (0.655)a - Internal radius,inches (0.500)R - Radius where stress is to be caiculated,inches (0.655)
Equation 1 reduces to Eqn. 2.
aT - 11.17 P (2) IOr solving for pressure.
P - 0.089 aT (3)
This formula is used to predict stress in thick walled tubes Iand the rig part doesn't strictly meet the requirements of open
46
I
soak .are very consistent. After this first test the gages wereglass beaded off the outer steel shaft and the assembly is shownin Fig. 42.
Fig. 42. NC-132 Buttress thread test rig assembly.
A second test was conducted to see how the diameter of theassembly changed with temperature. The data appears in Fig. 43.This data shows that the diameter expands at a rate consistentwith a linear thermal expansion coefficient of 8 or 9x10-6 inchesper inch per degrees Fahrenheit which is a representative valuefor steels. The plastic sleeve pressure contributes lIttleto the outer steel shaft expansion. This data is useful if ajoint were made in the area of a bearing where increases indiameter would have to be predicted so sufficient coldclearnraes could be built in.
47
II
NC 132 CERAMICVESPEL SP-22 SLEEVE
I1.3155
1.3150 FORSTEEL8
zILii 1.3145-WI-!a 1.3140
SLOPE =9x 10- IN/@F
1.3135 "
70 100 150 200 250 300TEMPERATURE (OF)
Fig. 43. Assembly diameter increase versus temperature.
After the final diameter versus temperature test, anattempt was made to unscrew the ceramic part. The parts weretightly held together and it was necessary to machine a groovearound the steel portion to allow the threads to unscrew withoutturning the ceramic relative to the sleeve. After final Idisassembly several of the ceramic threads were found to bechipped. There were some chips at assembly, but the furtherdamage was probably caused by a chip type failuze occurring atassembly with that loose chip jamming and causing much moredamage upon disassembly. The steel sleeve still surroundingthe VespelTM and the ceramic had to be pressed apart in aspecial fixture. The parts and the fixture are shown in Fig. I44.
I48!
I
Fig. 44. Disassembled NC-132 buttress thread rig.
Recent efforts have been directed at reducing the chippingtendency of the very thin edge of the 3/4-12 buttress threadsfirst used. Taps have been purchased to cut 3/4-8 buttressthreads and a 3/4-6 Acme thread (29 degrees face angle). Diamondplated grinding wheels to produce these threads on theceramic test parts are also being purchased. The present plan isto test an Acme threaded attachment in the hot spin rig withthe final design of this program as shown in Fig. 45.
29* ACME3/4-8 THREADS
CERAMIC 20 TAPEREDROTOR VESPEL \LEEVE
I RADIUSED SHOULDERS
I Fig. 45. Threaded attachment final design.
49
V. SUMMARY I
The original intent of this phase of the Life PredictionMethodolgy Program was to repeat the stress rupture tests Iconducted on NC-132 disks with a second material in order toverify the life prediction methods developed under that phase ofthe program. These tests were to be conducted with integralshaft spin disks. The integral shaft design was to make possiblea simpler and more reliable attachment than that used with theNC-132 disks. The program was to characterize the integral shaftspin disk material in order determine the test conditions for the Ispin disk. The assumption was that the material was reprodu-cible. As the characterization commenced it became apparent thatthe material was not reproducible; therefore, the program wasmodified to include a period of specimen development after whichthe characterization and testing would be resumed. Curtailmentof the funding precluded the completion of all the tasks.
Initial characterization testing was conducted on a sinteredsilicon nitride supplied by Ford. Fast fracture testing wasconducted at room temperature on ten specimens in four point ibending. The fracture strength varied from a minimum of 785 MPato a maximum of 988 MPa with an average strength of 873 MPa, anda standard deviation of 79 MPa with a Weibull modulus of 13.Stress rupture testing was conducted in the intermediate tempera-ture range of 6000 C to 10000C to determine if the material wasinstable in this range. Two specimens were tested at a stresslevel of 344 MPa. One survived 306 hours without failure of Ubending while the second specimen failed in 35 hours. Examin-ation of the fracture surface of the second specimen revealed thepresence of a locally oxidized region as the failure initiationsource. A third specimen was tested at 413 MPa. It failed inone hour. The behavior of these three specimens and otherstested indicated that the material has an instability at 10000 C.At this point the program was reviewed and modified to include aperiod of specimen development.
The objective of the specimen development part of theprogram was to develop the methodology to fabricate the integralshaft spin disk with reproducible properties, using an injectionmolded, sintered reaction bonded silicon nitride. After thespecimen development was completed the program was to resume thematerial characterization and spin testing phases of the program.The specimen development was divided into two major areas: (1)injection molding, and (2) sintering-property development. I
A set of statistically designed experiments was performed torelate the effect of four injection molding variables to com-ponent quality after injection molding and after binder removalfrom the injection molded part. The experiments evaluated theeffect of four injection molding variables on part qualityafter molding and after binder removal. The injection molding
50
|I
variables were die temperature, injection pressure, materialtemperature, and hold time. A 24-1 fractional factorial experi-mental design was employed. The experiment was repeated for twocomponents. Two responses were considered critical aftermolding: (1) density, and (2) the number of x-ray indications.The quality parameters relating to cracks are only importantafter binder removal. The results generally indicated that allfour injection molding variables, die temperature, injectionpressure, material temperature, and hold time should be reducedin order to improve the component quality. They showed thatconditions which are thought to a yield low stress state in thecomponent also result in a high quality component after moldingand binder removal. They also showed that the molding conditionsdirectly effect the quality of the component after binderremoval.
A set of statistically designed experiments was performed todetermine the effect of three binder removal processing variableson the component quality after binder removal. The variablesinvestigated were (1) the pressure of the binder removal atmo-sphere, (2) the rate of temperature rise, and (3) the complexityof the geometry of the component. A full factorial 2 3 experi-mental design was employed to determine primary effects as wellas interaction effects. The results indicated that higher binderremoval percentages are obtained at high pressure than under lowpressure conditions. This was found to be true for both largeand small components. The results also indicated that the largecomponent exhibits higher binder removal percentages than thesmall component. The results also indicated that there is aninteraction between the heating rate and the pressure of thebinder removal atmosphere.
In the sintering-property development two particular thingswere studied. These were (1) the microstructure development, and(2) strength improvements. Analysis of the microstructures of anumber of compositions in the yttria/alumina system showed that acritical temperature exists above which exaggerated grain growthoccurs and below which a uniform microstructure can be obtained.Sintering time above the critical temperature was identified asthe principal parameter contributing to excessive grain growth.Sintering experiments above and below the critical temperaturewere conducted to determine maximum grain size and number oflarge grains per unit area versus time. The results defineda set of acceptable time-temperature sintering parametersrequired for obtaining a uniform microstructure in a sinteredreaction bonded silicon nitride.
A set of experiments were conducted to improve the strengthof the sintered reaction bonded silicon nitride. Dry pressedcompositions were sintered using four sets of processing con-ditions where the sintering temperature and time were varied.They included a baseline where the sintering temperature was
51
i
above the critical temperature for grain growth and the time Uwas long. The others included a short time at high temperature,long time at low temperature and an intermediate time andtemperature. These last three sintering conditions were Idesigned to generate a microstructure free of the large needleshaped grains. The density, strength and microstructure wasstudied for the four conditions. The results showed that thestrength of the dry pressed material is optimized when thesintering temperature was near the critical temperature. Thisproduced maximum density and a fine, uniform microstructure.
In addition to the control of the microstructure andstrength of the molding material, the injection molding die usedto form the integral shaft disk requires careful design. The diemust be designed to control the solidification behavior of themolding material. The molding material solidifies and undergoesshrinkage with decreasing temperature. A properly designedinjection molding tool controls the temperature distributionwithin the die such that the last volume to cool to the solid-ification temperature is in the sprue, and not in the part.This eliminates any possibility of the formation of shrinkage ivoids within the component. A finite element heat transfer studywas conducted on a proposed integral shaft spin disk die.Temperature contours were plotted versus time for severaldifferent cooling designs. It was determined that in order toproperly control the cooling of the molding material an activetemperature control was reouired. The die must have a heater onthe sprue end of the die, and the output of the heater must be ireduced with time.
The integral shaft spin disk required the design anddevelopment of a new type attachment which uld take advantage iof the long ceramic shaft on the disk to icate the ceramic tometal attachment in a relatively cool location. This wouldincrease the reliability of the attachment. Two types of ceramicto metal attachments were developed . Both attachments used theprinciple of a high thermal expansion plastic sleeve trapped in arelatively constant volume created between ceramic and steel ishafts. One design used a steel lock nut to trap the plastic,and the second used a special type of thread on the ceramicshaft to trap the plastic. The designs were extensively testedin bench test rigs using several plastic and ceramic materials.The metal lock nut version was testeu for more than 80 hours inthe hot spin rig.
VI. CONCLUSIONS
A. The four injection molding variables, die temperature,injection pressure, material temperature, and hold timeshould be reduced in order to improve the componentquality.
52
B. Binder removal rates from injection molded parts areimproved with the use of a pressurized atmosphere.
C. A pressurized binder removal atmosphere increases compo-nent quality.
D. An active temperature control is required for injectionmolding an integral shaft disk.
E. The high expansion lock ceramic-to-metal attachment issuitable for use in a hot spin rig application.
53
I
TABLE I i
Flexural Stress Rupture Results for Billet A-42
Temperature Applied Failure Suspension Remarks
Stress Time Time°c MPa Hours Hours I800 413 - 336 Color changed to light
gray, no spot formation
800 482 88 - LOR, Color gray, no
spot formation and no Ibending
1000 344 35 LOR, Uniform oxidation,
Color white, no spot
formation and nobending
1000 344 - 306 Color white, no spot
formation and nobending
1000 413 1 - Oxidation pit, Color
whitish gray, no spot
formation and no ibending
1000 482 0.5 Porosity, Color whitish Igray, no spot formationand no bending [
LOR - Local Oxidation Region
IiIII
I
TABLE II
Fast Fracture Strength Data for LAS at Room Temperature
Specimen Fracture Specimen Fracture Specimen FractureNumber Strength Number Strength Number Strength
Publications Wholly or Partially Attributed to this Contract
1. Govila, R. K., "Methodology for Ceramic Life Prediction and
Related Proof Testing," Tech. Rept. AMMRC TR 78-29, July, 1978. I2. Govila, R. K., "Ceramic Life Prediction larameters," Tech.Rept. AMMRC TR 80-18, May, 1980. 33. Govila, R. K., "Indentation-Precracking and Double-TorsionMethods for Measuring Fracture Mechanics Parameters in Hot
Pressed Silicon Nitride," Journal of The American Ceramic ISociety, Vol. 63, No. 5-6 May-June 1980.
4. Govila, R. K., "Uniaxial Tensile and Flexural Stress RuptureStrength of Hot-Pressed Silicon Nitride," Journal of The American iCeramic Society, Vol. 65. No. 1, January, 1982.
5. Baker, R. R., Swank, L. R., and Caverly, J. C., "Ceramic Life iPrediction Methodology - Hot Spin Disc Life Program," Tech.Rept. AMMRC TR 82-26, April, 1982. 16. Swank, L. R., "Ceramic Life Prediction Methodology-AnalyticalAssessment of Selected Component Data," Tech. Rept. AMMRC TR 82-50, September 1982.
7. Govila, R. K., "High Temperature Strength Characterization of
October, 1982. I8. Govila, R. K., "Statistical Strength Evaluation of Hot-Pre-ssed Silicon Nitride," Ceramic Bulletin, 62, [11], 1983.
9. Govila, R. K., High Temperature Uniaxial Tensile StressRupture Strength of Sintered Alpha SiC," Journal of Materials
Science 18, 1967-1967, (1983).
10. Govila, R. K., "Material Parameters for Life Prediction," inCeramics for High Performance Applications III, Editors Edward M. ILenoe, R. Nathan Katz, and John J. Burke, New York, Plenum Press,
1983.
11. Govila, R. K., "Flexural Stress Rupture Strength of Sintered IAlpha Silicon Carbide," in Time-Dependent Failure Mechanisms andAssessment Methodologies, Editors J. G. Early, T. R. Shives, and
J. H. Smith, New York, Cambridge University Press, 1983. U12. Baker, R. R., Swank, L. R., and Caverly, J. C., "Ceramic LifePrediction Methodology - Hot Spin Disc Life Program," Tech. Rept.
AI4MRC TR 83-44, August, 1983.
66 i
I
13. Swank, L. R., Baker, R. R., and Lenoe, E. M., "Ceramic LifePrediction Methodology", Proceedings of the Twenty-First Automo-tive Technology Development Contractors Coordination Meeting, SAEP-138, November, 1983.
14. Covila, R. K., Phenomenology of Fracture in Sintered AlphaSilicon Carbide", Journal of Materials Science 19, 2111-2120(1984).
15. Govila, R. K., "Strength Characterization and Nature of CrackPropagation in Sintered Alpha Silicon Carbide," Proceedings:Sixth International Conference on Fracture, Editors S. R.Valluvi, P. Rama Rao, and K. N. Raju, Cambridge, England,
Pergammon Press, 1984.
TABLE XVI
Patents Wholly or Partially Attributed to this Contract
Method of Attaching a Metal Shaft to a Ceramic Shaft and ProductProduced Thereby, U. S. Patent 4,485,545.
Method of Attaching a Metal Shaft to a Ceramic Shaft and ProductProduced Thereby, U. S. Patent 4,499,646.
67
I
REFERENCES U1. Govila, R. K., "Methodology for Ceramic Life Prediction and
Related Proof Testing," Tech. Rept. AMMRC TR 78-29, July, I1978.
2. Govila, R. K., "Ceramic Life Prediction Parameters," Tech.Rept. AMMRC TR 80-18, May, 1980. I3. Govila, R. K., "Indentation-Precracking and Double-TorsionMethods for Measuring Fracture Mechanics Parameters in Hot-Pressed Silicon Nitride," Journal of The American CeramicSociety, Vol. 63, No. 5-6 May-June 1980.
4. Covila, R. K., "Uniaxial Tensile and Flexural Stress RuptureStrength of Hot-Pressed Silicon Nitride," Journal of The AmericanCeramic Society, Vol. 65. No. 1, January, 1982. 35. Govila, R. K., "Statistical Strength Evaluation of Hot-Pressed Silicon Nitride," Ceramic Bulletin, 62, (11], 1983.
6. Govila, R. K., "High Temperature Strength Characterizationof Sintered Alpha Silicon Carbide," Tech. Rept. AMMRC TR 82-51,October, 1982. 17. Covila, R. K., High Temperature Uniaxial Tensile StressRupture Strength of Sintered Alpha SiC," Journal of Materials
Science 18, 1967-1976, (1983). I8. Govila, R. K., Phenomenology of Fracture in Sintered AlphaSilicon Carbide", Journal of Materials Science 19, 2111-2120
(1984).
9. Swank, L. R., "Ceramic Life Prediction Methodology-AnalyticalAssessment of Selected Component Data," Tech. Rept. AMMRC TR 82- I50, September 1982.
10. Baker, R. R., Swank, L. R., and Caverly, J. C., "Ceramic 5Life Prediction Methodology - Hot Spin Disc Life Program," Tech.Rept. AMMRC TR 82-26, April, 1982.
11. Baker, R. R., Swank, L. R., and Caverly, J. C., "Ceramic ILife Prediction Methodology - Hot Spin Disc Life Program," Tech.Rept. AMMRC TR 83-44, August, 1983. 112. Govila, R. K., Herman J. A., and Arnon, N., "Stress RuptureTest Rig Design for Evaluating Ceramic Material Specimens," ASMEPaper #85-GT-181, March 18, 1985.
13. Govila, R. K., Kinsman, K. R., and Beardmore, P., "Fracture
Phenomenology of a Lithium-Aluminium-Silicate Glass-Ceramic,"Journal of Material Science, 13 [4] 2081-2091 (1978).
68
i
14. Lipson, C., and Sheth, N. J., Statistical Design and Analysisof Engineering Experiments, New York, McGraw-Hill, 1973.
69
DISTRIBUTION LIST
Donald F. AdamsComposite Materials Research GroupMechanical Engineering DepartmentUniversity of WyomingLaramie, WY 82071
Jane W. AdamsCorning Glass WorksSP-DV-21Corning, NY 14831
Anil K. AgarwalProduct ManagerHigh Performance CeramicsNorton CompanyOne New Bond StreetWorcester, MA 01606
Richard T. AlpaughDepartment of EnergyOffice of Transportation SystemsCE-131 FORSTL1000 Independence AvenueWashington, DC 20585
James P. ArnoldU.S. Army BelvoirR&D CenterATTN: FTRBE-EMPFort Belvoir, VA 22060
V. S. AvvaDept. of Mechanical EngineeringNorth Carolina Agricultural and
Technical State UniversityGreensboro, NC 27411
John M. BaileyResearch Consultant, Research Dept.Technical CenterCaterpillar Tractor Co.100 NE AdamsPeoria, IL 61629
Murray BaileyNASA Lewis Research Center21000 Brookpark Road, MS 77-6Cleveland, OH 44135
II
J. Gary BaldoniGTE Laboratories, Inc.40 Sylvan RoadWaltham, MA 02254
R. R. Baker 534819 Lyndon StreetLivonia, MI 48154
Ken Baumert IAir Products and Chemicals, Inc.Box 538Allentown, PA 18105
Ronald L. BeattyARCO Chemicals, Silag Operation IRoute 6, Box A
Greer, SC 29651
A. L. Bement, Jr., Vice President iTechnical ResourcesTRW, Inc.23555 Euclid AvenueCleveland, OH 44117
Clifton G. Bergeron, HeadDepartment of Ceramic Engineering I204 Ceramics BuildingUniversity of IllinoisUrbana, IL 61801
William D. BjorndahlTRW, Inc.TRW Energy Development GroupMaterials Characterization andChemical Analysis Dept.One Space ParkBuilding 01, Room 2060Redondo Beach, CA 90278
Paul N. BlumbergPresidentIntegral Technologies Inc.415 E. Plaza Drive IWestmont, IL 60559
Wolfgang D. G. Boecker iSOHIO Engineered Materials Co.Niagara Falls R&D CenterPO Box 832Niagara Falls, NY 14302
II
Seymour A. BortzManager, Nonmetallic Materials
and Composites
Materials and ManufacturingTechnology
lIT Research Institute10 West 35th StreetChicago, IL 60616
H. K. BowenDepartment of Materials Science
and Engineering, Room 12-009Massachusetts Institute ofTechnology
Cambridge, MA 02139
Richard C. Bradt
University of WashingtonRoberts Hall
FB-10
Seattle, WA 98195
Raymond J. BrattonManager, Ceramic ScienceWestinghouse Research and
Development Center1310 Beulah RoadPittsburgh, PA 15235
W. Bryzik
US Army Tank Automotive Command(TACOM)
R&D Center
Warren, MI 48090
S. T. BuljanGTE Laboratories, Inc.
40 Sylvan Road
Waltham, MA 02154
John M. Byrne, Jr.Manager, Business Development
Corporate Development DepartmentPPG Industries, Inc.
Orie PPG PlacePittsburgh, PA 15272
Donald J. Campbell
Air Force Wright Aeronautical
LaboratoryAFWAI/POXWright-Patterson AFB, OH 45433
Harry W. CarpenterRockwell InternationalRocketdyne DivisionJ39-169:HC926633 Canoga AvenueCanoga Park, CA 91304 3David CarruthersGarrett Turbine Engine Company111 South 34 StreetPO Box 5217Phoenix, AZ 85010
Se-Tak Chang IGTE Laboratories40 Sylvan RoadDept. 312 IWaltham, MA 02254
En-sheng ChenB&C Engineering Research13906 Dentwood DriveHouston, TX 77014 n
Albert A. ChesnesDirector
Heat Engine Propulsion DivisionOffice of Transportation SystemsDepartment of EnergyCE-131 FORSTL1000 Independence AvenueWashington, DC 20585
Melvin H. Chiogioji 3DirectorOffice of Transportation SystemsDepartment of EnergyCE-13 FORSTL1000 Independence Avenue, SWWashington, DC 20585
William J. ChmuraThe Torrington CompanyCorporate Research59 Field StreetTorrington, CT 06790
William L. Cleary IAssociate Division DirectorORI, Inc.1375 Piccard DriveRockville, MD 20850
II
Philip R. ComptonEnergy Sy.3tems OfficeNational keronautics andSpace AdministrationCode REC-lWashington, DC 20546
John A. CoppolaRepresentative DirectorExecutive Vice PresidentHitachi-Carborundum CompanyShinjuku-Mitsui BuildingNo. 1-1, 2-Chome, NishishinjukuSbUajuku-ku, Tokyo 160, JAPAN
C. H. CraigDepartment of Energy1000 Independence AvenueCE-131 FORSTLWashington, DC 20585
William J. CroftU. S. Army Materials Techl,,ology Laboratory (MTL)Arsenal StreetWatertown, MA 02172
Gary M. CrosbieFord Motor CompanyPO Box 2053, Room S-2079Ceramics Materials DepartmentDearborn, MI 48121
Floy-I W. Crouse, Jr.Department of EnergyMorgantown Energy Technology Lt-nterPO Box 880Morgantown, WV 26505
Raymond CutlerCeramatec, Inc.163 West 1700 SouthSalt Lake City, UT 84115
Stanley J. DapkunasOffice of Technical CoordinationFossil Energy Technical
Coordination StaffFE-14, MS B127 GTNDepartment of EnergyWashington, DC 20545
I
Robert F. Davis INo +h Carolina State UniversityMaterials Engineerin; Department
232 Riddick Laboratory IRaleigh, NC 27607
Alan L. DragooMaterials Scientist, Inorgaiic IMaterials DivisionNational Bureau of Standards
Center for Mat-r 4 1s Science UGaithersburg, '-J z0899
Keith F. Dufrane 3Battelle Columbus Laboratories
505 King Avenue
Columbus, OH 43201 1Robert J. EaganManager, Chemistry and CeramicsDepartment 1840 ISandia National Laboratories
Aibuquerque, NM 87185
Christopher A. Ebel 1Program ManagerNorton Company
High Performance CeramicsI New Bond StreetWorchester, MA 01606
J. J. Eberhardt iOffice of E1,,rgy UtilizationResearch
Department o1 EtegyCE-142 FORSTL1000 Independence AvenueWaslington, DC 20585
E. E. EcklundCfficc of Transportation Systems
Department of Energy ICE-131 FORSTL1000 Independence Avenue
Washington, DC 20585 IWiliiam A. EllingsonArg-one National Laboratory9700 South Cass AvnueArgonne, IL 60/,9
I
I
i Director, Applied Technology
LaboratoryU.S. Army Research and Technology
Laboratory (AVSCOM)ATTN: SAVDL-ATL-ATP(Graydon A. Elliott)
i• Fort Eustis, VA 23604
A. ErdelyChemical Engineer26 Avenue Gare Des Eaux-vives1208 Geneva, SWITZERLAND
3 Charles D. Estes
U.S. SenateProfessional Staff Member
Committee on AppropriationsRoom SD-152 Dirksen Senate
Office BuildingWashington, DC 20510
Anthony G. EvansUniversity of California
Santa Barbara, CA 93106
Robert C. Evans, Asst. ManagerVehicular Gas Turbine and Diesel
Project OfficeNASA-Lewis Research Center21000 Brookpark Road
Cleveland, OH 44135
John FaceyNational Aeronautics andSpace AdministrationEnergy Systems Office5 Washington, DC 20546
John W. FairbanksOffice of Advanced EnergyConversion
Department of EnergyFE-22 GTN
i Wa.,'ngton, DC 20545
rrFy P'arrellBabcock and W ilcoxPO Box 1260
Lynchburg, '.A 2/4505
III
I
Matthew K. Ferber IUniversity of Illinois-Urbana
203 Ceramic Building
105 S. Goodwin Avenue IUrbana, IL 61801
R. E. FisherPresident IAmercom, Inc.
8948 Fulibright Avenue
Chatsworth, CA 91311
H. W. FoglesongDow Corning Corporation3901 S. Saginaw RoadMidland, MI 48640
Robert C. Frank 3Manager, Non-Metallic Materials
General Electric CompanyOne Neumann Way, Mail Drop M-87PO Box 156301Cincinnati, OH 45215-6301
Frank Gac iDepartment of Materials Scienceand EngineeringUniversity of WashingtonSeattle, WA 98195
George E. GazzaMaterials Technology Laboratory ICeramics Research DivisionAMXMR - MTArsenal StreetWatertown, MA 02172
Paul Glance
Director, R&D
Concept Analysis Corporation9145 General Court
Plymouth, MI 48170
Joseph W. Glatz
Naval Air Propulsion Center iScience and Technology GroupSystems Technology Division
Bo:. 7176, PE 34
Trenton, NJ 08628
I3I
S. GoguenOffice of Transportation SystemsDepartment of EnergyCE-131 FORSTL1000 Independence AvenueWashington, DC 20585
Stephen T. GonczyAllied Signal Research CenterMaterials Science Department50 UOP PlazaDes Plaines, IL 60016-6187
Robert J. GottschallOffice of Material SciencesDepartment of EnergyER-131 GTNWashington, DC 20545
Kenneth GreenSenior Development EngineerCoors Porcelain CompanyGolden, CO 80401
Michael GreenfieldNational Aeronautics andSpace Administration
Energy Systems OfficeWashington, DC 20546
L. E. GrosecloseGeneral Motors CorporationAllison Gas Turbine DivisionP.O. Box 420Indianapolis, IN 46206-0420
T. D. Gulden, Manager
Ceramics and ChemistryGA Technologies, Inc.PO Box 81608
San Dig., CA 92138
M. D. Girne'NIPFKRPO 13o-: 2128BartleLziLle, OK 74005
H. T. HahnMechanicral EnigLneering Dep-irtment
Washui' on Uiversity
Lindri 1 ind SkinkerBox 10g3 Sr. B o 3
I ( 1
U
Nabil S. Hakim IStaff Research Engineer
Engineering R&D
General Motors CorporationDetroit Diesel Allison Division36880 Ecorse Road
Romulus, MI 48174
John M. HalsteadManager, Business Development
Structural Ceramics Division IStandard Oil Engineered Materials Company
1625 Buftalo Avenue, Bldg. 91-2
PO Fox 1054 INiagara Falls, NY 14302
R. A. Harmon25 Schalren Drive ILatham, NY 12110
Stephen D. Hartline
Norton CompanyOne New Bond StreetWorchestar, MA 01606 3Willard E. Hauth
Section Manager - CompositeDevelopment Ceramics Program IDow Corning Corporation
Midland MI 48640
Norman L. HechtUniversity of Dayton Research
Institute300 College ParkDayto-, OH 45469-0001
S. S. Hecker, Chairman 3Center for Materials ScienceLos Alaanos National Laboratory
Mail Stop K765
Los Alamos, NM 87545
Peter W. Heitman
General Motors Corporation IAllison Gas Turbine Operations
PO Box 420, W-5
indianapolis, IN 46206-0420 3IUi
H. E. HelmsGeneral Motors Corporation
Allison Gas Turbine OperationsPO Box 420Indianapolis, IN 46206-0420
Thomas L. Henson, Director ofResearch and Engineering
Chemical & Metallurgical DivisionGTE Products CorporationHawes StreetTowanda, PA 18848-0504
Thomas P. HerbellNASA Lewis Research Center21000 Brookpark RoadM/S 49-3Cleveland, OH 44135
Robert V. Hillery, ManagerCoating Materials & ProcessesGeneral Electric CompanyCincinnati, OH 45215
Jonathan W. HintonVice President and General Manager
Structural Ceramics Division
SOHIO Engineered Materials CompanyPO Box 1054Niagara Falls, NY 14302
Stephen M. Hsu
Inorganic Materials Div.Center for Materials Science
U.S. Department of CommerceNational Bureau of Standards
Richard M. SpriggsNational Materials Advisory BoardNational Research Council2101 Constitution AvenueWashington, DC 20418
M. SrinivasanStandard Oil Engineered Materials CompanyNiagara Falls R&D CenterPO Box 832)Niagara Falls, NY 14302
Gordon L. Starr, ManagerMetallic/Ceramic Materials Dept.Cummins Engine Comnny, Inc.Mail Code 50183Box 3005Columbus, 1N 47202-3005
I
Harold L. Stocker, Manager ILow Heat Rejection ProgramGeneral Motors CorporationAllison Gas Turbine Operations IPO Box 420, T-23Indianapolis, IN 46206-0420
Roger StormDirector, Niagara Falls R&D CenterStandard Oil Engineered Materials CompanyPO Box 832 UNiagara Falls, NY 14302
E. E. StrainProgram ManagerGarrett Turbine Engine CompanyIII S. 34th StreetPO Box 5217, Mail Stop 301-2N IPhoenix, AZ 85010
Thomas N. Strom iNASA Lewis Research Center21000 Brookpark Road, 77-6Cleveland, OH 44135
Karsten StyhrAiResearch Casting Co.19800 Van Ness Avenue ITorrance, CA 90509
Lewis R. SwankFord Motor CompanyPO Box 2053Building SRL, Room E3172Dearborn, MI 48121
Anthony C. TaylorStaff Director ISubcommittee on Transportation,
Aviation and MaterialsCommittee on Science andTechnology I
U.S. House of RepresentativesRoom 2321 Rayburn BuildingWashington, DC 20515 IW. H. ThielbahrChief, Energy Programs Branch 3Idaho Operations OfficeU.S. Department of Energy550 2nd StreetIdaho Falls, ID 83401
II
John K. TienDirector of Center for
Strategic Materials1137 S.W. Mudd BuildingColumbia UniversityNew York, NY 10027
T. Y. TienUniversity of MichiganDept. of Materials &Metallurgical Engineering
Dow BuildingAnn Arbor, MI 48109-2136
Nancy J. TigheNational Bureau of StandardsInorganic Materials Division 420Gaithersburg, MD 20899
Julian M. TishkoffAir Force Office of
Scientific ResearchDirectorate of Aerospace SciencesBolling AFBWashington, DC 20332
Maurice L. TortiSenior ScientistHigh Performance CeramicsNorton CompanyOne New Bond StreetWorchester, MA 01606
Louis E. TothDivision of Materials ResearchNational Science Foundation1800 G Street, N.W.Washington, DC 20550
Richard E. TresslerChairman, Ceramic Science and
Engineering DepartmentThe Pennsylvania State University201 Steidle BuildingUniversity Park, PA 16802
V. VenkateswaranStandard Oil Engineered Materials CompanyPO Box 832Niagara Falls, NY 14302
I
John B. Wachtman, Jr. IRutgers UniversityDepartment of CeramicsPO Box 901Piscataway, NJ 08854
Richard B. WallaceManager, Government Research
and Development PrugramsGeneral Motors CorporationDetroit Diesel Allison Division U36880 Ecorse RoadRomulus, MI 48174 3Harlan L. WatsonSubcommittee on Energy Researchand 2roduction
U.S. House of PepresentativesCommittee on Science andTechnology
Suite 2321, Rayburn HouseOffice Building
Washington, DC 20515
Steven G. WaxMaterials Science DivisionAdvanced Research Projects AgencyDepartment of Defense1400 Wilson BoulevardArlington, VA 22209
Albert R. C. WestwoodCorporate DirectorMartin Marietta Laboratories1450 South Rolling RoadBaltimore, MD 21227
Thomas J. WhalenPrincipal Research ScientistFord Motor CompanyScientific Lab, Room 2023Dearborn, MI 48121
Sheldon M. WiederhornInorganic Materials DivisionMechanical Properties GroupU.S. Department of CommerceNational Bureau of Standards IGaithersburg, MD 20899
III
James C. WilliamsDeanCarnegie Institute of TechnologyCarnegie-Mellon UniversitySchenley ParkPittsburgh, PA 15213
Roger R. Wills, ManagerAdvanced Ceramic ComponentsTRW Inc.Automotive Worldwide SectorValve Division1455 East 185th StreetCleveland, OH 44110
David Gordon WilsoaMassachusetts Institute of
TechnologyMechanical Engineering DepartmentRoom 3-455Cambridge, MA 02139
J. M. Wimmer, SupervisorNonmetallic Materials GroupGarrett Turbine Engine Company111 S. 34th StreetP.O. Box 5217Phoenix, AZ 85010
David Wirth, Vice PresidentTechnical Operations and Engr.Coors Porcelain Company17750 North 32 StreetGolden, CO 80401
Thomas J. WissingManager, Government
Contract AdministrationEaton CorporationEngineering & Research Center26201 Northwestern HighwayPO Box 766Southfield, MI 48037
Jamcs 3. WoodNASA Lewis Research Center21000 Brookpark RoadMail Stop 77-6Cleveland, OH 44135
i
Hun C. Yeh ICeramic SupervisorAiResearch Casting CompanyA Division of the Garrett Corp. i19800 Van Ness AvenueTorrance, CA 90509
Thomas M. Yonushonis ICummins Engine CompanyBox 3005, Mail Code 50183Columbus, IN 47202-3005 IDon Zabi-erekAir Force Wright Aeronautical
LaboratoryAFWAL/POTCWright-Patterson AFB, Ol 45433
Klaus M. ZwilskyExecutive DirectorNational Materials Advisory BoardNational Research Council2101 Constitution AvenueWashington, DC 20418 3Department of EnergyOak Ridge Operations OfficeOffice of Assistant Manager for U
Energy Research and DevelopmentPO Box EOak Ridge, TN 37831 i
Department of EnergyTechnical Information CenterOffice of Information Services IPO Box 62Oak Ridge, TN 37831
