Embed Size (px)
Citation preview
- I STIATIO OF THE UT OF ,E,.,-•1., PATERIA
IN AICRAFT ENI1NE BEAVINS
0 <
John W. LucekNorton CompanyI Now Bon7StreetWorcester, MA 01606 14
Paui i. CowleyFederal Mogul Corporation3980 Research Park DriveAnn Arbor, MI 48104
1 June 1979
Final Report for Period 14 June 1976 19 December 1978
Contract: NASC N00019-76-C-02S1
Appz.ovvA. for public rele,,a, distribution urslimited
Prepamrd forLa
L DEPARTMENT OF TYE NAVYN#al Air Systems CommandCode AIR-5203ZAWashington, DCL 20360 h
NORTOK COM!PANYIVJYJSThIAL CERAMICS DIVISION
WONCESTER, MISSACt.jI1ETTS 01606
L ,!-Sij
ERRATA Hp.31 ADDENDUM
Data in Table 9 tabulates maximum tensile stresses at failure Iload resulting from the Hertzian contact. The average compressivestresses at failure (tensile failure), failure load - contact area,are! NC-132 - 10.3 GN/m•; Si 3 N4 /ZrO 2 - 10.9 GN/m 2 ; and Si 3N4/Y 2 O3 -
13.4 GN/m 2 .
*itI
LI
CVMTVim CLSIIAOM W OJM4A PPAUM
'ITERIAL IN~ AIR~CRAFT ENLUNE ERtINGS._ 14ZW6 - 1ob EC79W "a *Ono. or"'.AlU
'ayFederal Mogu lorp N 19-76-C- 2Sl,~Worcester, M-A 061606 Ann Arbor,, MI 48104_________
Norton CompanyIndustriai ceramics DivisionWorcester,_MA__01606 ________
I f. CON Tm PL U &NA OFP PsC N AM E AN D DR S ---- - -Depavtment of the Navy 09,Naval Air Systems CommaridWashington, D0 20360 _____________
IN ITOW 49 N A 0 A ADORIESS.91 dift""wt (MM C00100111100 0911iff) IS. SCUNI CLASS. (of this ,gpef)
DCASR Unclassified666 Summer Street Iýa.DECT~kASWI t A TIOWONDOWN OR AOINBoston, MA 02210 L
116. DISTRIMgiW STATEMENT (oehi lAD R0..)
Approved for public release; distribution unlimited.
Ceramic Finishing* Hertzian Contact
SO. LUST*ACT (Cotm ee. an ronmro side Of necessar and fridaltj r bybloc numbat)* . program to initiate life factor correlation between hot-pressed
zilicon nitride rolling elements and M-5O steel hybrid rollingelements has been completed. All materials for component fabrica-tion have been procured and qualified. Retainers and test toolingare complete. Complications in rolling contact fatigue qualifi-cation Of roller stock, traced to specimen preparation techniquesdid not allcw completion of bearing fabrication. A considerable£forjdwa ddcated, to sjudX~ of the RCF Qualif ication --or ceramic00ON"?'a 1473. EDITION OF NOV 6ItSOBSOLETE UNCLASSIFIED
82CURITY CLAMSFICATION OF THIS PAGE (Mone DoMIfi
K ic
- ~ ''rT~ "'4- -, ft"
UNiCLASSIFIEDVCUlVY ¢CLAWFICATION OF THIS PA6I6= j. Xe.O
0. ABSTRACT (continued)
aterials resulting in development of machining techniques toproduce bearing quality silicon nitride surfaces without 4super-finishingf. Static Hertzian indentation of silicon nitridesurfaces, fracture energy measurements, and radioactive gaspenetrant tracers were found to be potentially valuable evalua-tion techniques for ceramics in the rolling contact environment.
DD."
By -
t 71- I A'. a i a ,, oj.
DI• st pccIla
S~NCWASIFIEDD-
Ot
SUMMI4ARY
To date, the evaluation of hybrid rolling contact bearingshas been carried out under "accelerated" test conditions or ona marginally significant statistical sample. The objective ofthis program was to initiate testing on a significant populationto determine life-factors in a loading environment more closelysimulating an actual application. Subtasks under this objectivewere to:
1. Design and build 60 mm bore, 20 roller complementhybrid SiN%/M-SO and all steel M-SO bearings.
2. Initiate "slightly accelerated" testing.
3. Experiment with static Hertzian incentation as aqualification tool.
4. Tvaluate SisN4-Y,0O hot pressed ceramics.
Because of complications encountered in the rolling contactfatigue qualification of the silicon nitride roller stock, con-siderable effort was expended to study the techniques used fromthe standpoints of specimen preparation, evaluation, and testing.This delay resulted n not completing bearing fabrication; norinitiating testing. A follow-on program is currently underwayto complete these objectives.
A design study aimed at reducing the higher relative contactstresses in the hybrid bearing was carried out. The resultinganalysis required geometries impractical to manufacture and werenot pursued further. At the close of this contract all materialsfor bearing fabrication have been procured, qualified, and accep-ted. The retainers for the 20 roller complement bearings arecomplete. All tooling for testing the hybrid and all-steelbearings is on hand.
A major effort was made to clarify the variables associatedwith the RCF qualification of hot-pressed silicon nitride mater-ials. The material properties, specimen fabrication techniques,and test methods all contribute equally to the data generated inthe RCF test. In this program, the recurring anomalous data ob-tained was directly related to the techniques used to preparespecimens. The honing and lapping techniques previously usedfor final finishing of RCF rods imparted a degree of variabilityto the surface properties of the material. Once this had beendetermined alternate finishing techniques were sought thatwould yiela uniform surface properties. Systematic studies toprepare bearing quality finishes by grinding resulted in uniformlyfine, damage-free surfaces that yielded acceptable RCF behavior
S~1
allowing material qualification. It is felt that the machiningsequence developed can yield an optimized surface compared toearlier work and is applicable to bearing production as well astest specimen preparation.
Experimental study of static Hertzian indentation )f siliconnitride surfaces has demonstrated the technique is sensitive toboth machining variables and material properties; ft is of usefor ranking materials with regard to tracture energy, surface flawdistributions, and stress corrosion rates. Evaluation of bothSi*N%/ZrOt and Si3N%/Y&Os hot pressed ceramics have shown themboth superior to NC-132 (SiIN,1/'gO) in terms of resistance tostatic Hertzian stresses. The rolling contact fatiguo behaviorremains to be determined. Fracture energy measurements did notdifferentiate between NC-132 Sit,:4 and SiiNa/YaOs materials.Efforts to characterize machining damaged strfaces in the RC'studies included evaluation by radioactive gas penetrarcrs.
The KET* technique resolved surface structures not otherwisedetectable, but, as yet, no clear correlation between these struc-tu.res and RCF performance exists.
The evaluation of RCF specimen surfaces delayed bearing fab-rication but also resulted in closer control oi this qualificationtechnique and the optimization of machining methods for finishingSi.N4 ceramics in bearing applications. Test tooling -nd somecomponent fabrication are complete. No new impediments to thedevelopment of ceramic rolling contact bearings were discovered.
!I
*QUAL-X, Inc. Hilliard, Ohio
2
FOREWORD
This final report describes efforts at the Norton Companyunder contract to Naval Air Systems Command for the period 14June 1976 through 19 December 1978 on Contract N00019-76-C-0251.The objective of this program was to investigate the behaviorof ceramic materials in rolling contact environments.
This program was administered by bot) Mr, C. Bersch and Mr.P. Weinberg, NAVAIR, Washington, DC. The program was initiatedby Dr. H. R. Baumgartnor, Norton Company, who was Principal in-vestigator through 20 December 1976. For the remaining portionof the program, Mr. J. W. Lucek served as principal investigator;technical management was provided by Dr. M. L. Torti. Bearingdesign, material qualification, and component manufacturo effortswere carried out by the Bearing Research Group of Federal-1ogulCorporation, Ann Arbor, Michigan 48104; Mr. P. Cowley was prin-cipal investigator at that facility.
The authors wish to thank the research support groups atboth the Norton Company and Federal-Mogul Corporation. They wishto acknowled e the consultation time provided by Dr. Baumgartnerthroughout the program. The authors also wish to thank the NAVAIRadministrators for their cooperation.
3
TABLE OF CONTENTS
k!
Summary 1
Foreword 3
I. IN'TRODUCTION 7
11. BiARING DESIGN
Ill. MATERIAL QUALIFICATION: 10
IV. RCF TESTING OF NC-132 SILICON NITRIDE 10A. IntroductionB. Diamond Honed SpecimensC. Lapped RodsD. Techni~que InvestigationsE. Conclusions
V. RCF SPECIMEN PRODUCTION DEVELOPMENT 20
VI. COMPONENT MANUFACTURE 24
VII. EXPERIMENTAL MATERIALS TESTING 24A. Fracture EnergyB. Hertzian Contct Characterization
REFERENCES 32
APPENDIX A
APPENDIX 8
Distribution List
777--e L i
LIST OF FIGURES
Figure
Number Pg
I Rod 13 failure spall, SOX SIM is
2 Wear tracks, Rods 13 and 14 16
3 RCF rod, Federal-Mogul 23
4 Hertzian indenter 26
S Equipment layout 26
6A NC-132 silicon nitride - Hertzian contact 27behavior - 4.75 am radius WC indenter
6D Alternate composition hot-pressed siliconnitride - Hertzian contact behavior - 284.75 mm radius VC indenter
7 dertzian cone crack, SOX SEX 29
8 Hertzian contact areas of NC-132 30
LIST OF TABLES
TableNjubor
1 Summary of R"F Tests on Silicon Nitride and
M-SO CVM Steel as a Function of Loading
2 •a-~da• •iN., Mterial Qualifications 11
3 RCF Data Summary - 810,000 psi Stress 14
4 RCF Life Versus Finish for MC-i326 18
S RCF Lives of D~.amord Ground NC-132 at 1
800,000 psi Contact Stress
6 RCF Rod Diamond Grinding Parameters 22
7 Grinding Development RCF Specimens 22
8 Fracture Energy Measurements 25
9 Hortzian Indentation Testing, Hot-Pressed 31Silicon Nitride, 0.1875" WC Indenter
6 a
I. INTRODUCTION
Hot-pressed silicon nitride has been shown tn be a viablematerial for use In rolling contact bearings. Programs under thesponsorship of the Naval Air Systems Command have shown thatroperly machined silicon nitride surfaces are superior to AISI
CM -SO stael In terms of rolling contact fatigue life.$ Hybrid,as well as all silicon nitride test bearings have been manufac-tured and successfully telted under accelerated life conditions.,'High speed rig tests (2,5 million Di) of hybrid silicop nitrideball bearings hay* demonstrated that such bearinga generate lessheat than an equivalent steel bearing. Hybrid roller bearingsperformed acceptably in a similar environment.' A contract to theArmy Materials and Mechanics Research Center vo develop finalfinishing techniques for silicon nitride componentse has shownthat diasona honing as well 83 silicon carbide and diamond grind-Ing can produce long contact fatigue life surfaces. A recent allsilicon nitride bearing test carried out under a joint Navy/Armycontract has demonstrated the feasibility of operating ceramicroller bearings both with minimal lubrication and without lubri-cation for sustained periods oa time in a J402 gas turbine engine.?
Silicon nitride rolling contact bearings are of interest inhostile chemical and thermal environments for several reason3s:
1. The low density of silicon nitride rolling elements,approximately one-thlrd that of tool steel, allows for higherresidual bearing capacity through the reduction of centrifugalbody forces. This lower density also reduces element skiddingduring rotational velocity transients.
2. The low coefficient of thermal expansion of thisceramic material, under conditions of a radial thermal differ-ential through A bearing, reduces skidding tendencis due to lossof internal clearance.
3. Silicon nitride's low cnefficient of friction andhigh temperature hardness potentially allow operation withoutexternal lubrication. The response of hot pressed NC-137 sili-con nitride *, stress fields is essentially elastic to by uvnd10000C. This may allow relaxation of current cooling and lubri-cation requirements in some applications.
4. The silicon nitride ceramic is essentially inert inthe bearing environment. Oxidation does not begin until 7000Cand reaction with fused salts i3 negligible until 700"C also.In light of the above considerations, it is unlikely that anylong term reactions with lubricents should occur.
Previous work has shown that the formation of Hertzian crackswhich lead to bearing component failure is highly dependent on
7 _
the component composAtion, finish and may be influenced by itsenvironment through stress corrosion mechanisms. Yttria containintsilicon nitride (YIO./SiIN,) hot ressed materials have shown me-chanical properties superior to tose of current NSO/SiqN4 hotpressed material. Initial evaluation of this material indicatesthat it possesses a highor modulus of rupture, lover strengthvariations, and a higher fracture energy than regular hot-pressedsilicon nitrides WC-132.'
Barlier work with silicon nitride rolling contact elementtesting has indicated there may be a "transition stress level"below which silicon nitride bearing life may be considerablygiuater than calculated by conventil load/life formula. AtHertz stress levels of 000 000 psi, rolling contact fatiguetesting (RCP) of silicon nitride has demonstrated fati uue livesn imes greater than those of M-S0 tool steel.' See Table 1.
The actual advantage is not known as an insufficient number ofelement failures occurred; most tests were suspensions and, there-fore, the data is of limited use fer life calculations. This"transition stress level" is felt to be well above bearing deiAgnlimits, but is below the stresses encountered in conventional"accelerated-life" testing. Testing of bearings under load con-ditions more representative of an actual application may demon-strate the existence of such a transition level.
I1, BEARING DES16H
The bearing design intended for testing in this program wasa slightly modified Dower Aircraft design utilized in previoussilicon nitride bearing development work .t, It is comprised ofa 60 ma bore inner race assembly with a separable outer race, 20roller complement, and a one piece silver-plated machined steelretainer.
Two design modifications were considered for this program:
A. Increasing the silicon nitride roller crown radiusto reduce the stresses in the contact zone. Silicon nitride'shigh Young's modulus results in higher stresses than in an all-steel bearing under the sane conditions. This modification wasnot recommended becauso the larger radius would be in racticalto manufacture and might compromise the crown's Capacity to ac-comodate 3isalignaent.
B. Crowning of the outer raceway to bias the contactstress to the outer race; aore closely simulating a high speedapplication. This change was also not pursued because of uncer-tain mating crown alignment effects.
8
TABLE 1
Sumary of RCF Tests on Silicon Nitrideand N-S0 CVN Steel as a Punction of Loading
Hertz Stress Silicon Nitride 4-SO
6000000 16 :uspensions in range of LI: a 2.38 million
30.92 - 93.5S million cyclescycles LSO i 3.70 million
cycles
700,000 6 failures in range of LIO- 1.60 million3.85 - 141.S4 million cyclescycles LSO a 2.58 million
cycles
10 suspensions in range of32.57 - 113.44 millioncycles
750,000 4 failures in range of Nc tests conducted19.72 - 100.61 mIllioncycles
2 sispensions at 97.64 and125.74 million cycles
800,000 11 failures in range of Testing caused ex-1.35 - 47.36 million cessive deformationcycles of the steel bars.
9I
III, MATERIAL QUALIFICATIONS
A. Silicon Nitride
Four billets of NC-132 s'licon nitride have been producedfor use as roller stock in this program. Standard mate-ie1 luali-fications have indicated that all of the billets provided meet thepublished specifications for this material. A summary of thesequalifications appears in Table 2. Four billets provided for rollerstock materiel all exceeded the strength density and chemical spe-cifications of NC-132 silicon nitride. Additional testing todetermine fracture energy, also shows that the material exhibitsbehavior typical of material used in past bearing programs. Basedon the poor rolling contact fatigue performance (below) of billet300502 it has been rejected as roller stock material. This maybe due to the low concentration of the magnesium oxide/silicateintergranular phase. The last billet, 428841, was not providedas roller stock material; but was supplied to Federal-Mogul formaterial to be used in grinding studies. The roller stock materialhas been qualified in rolling contact fatigue at program completion.
B. Metallic Components
The first lot of 4340 steel for bearing retainers was judgedacceptable by the fourth month of the program and retainer fabri-zation was initiated. The first lot of M-S0 race material wasjudged unacceptable by reason of Class 8 carbide segregation andwas rejected by Federal-Mogul for this program and a reorder placed.The second lot of race material was accepted and reserved for fab-rication after roller stock qualification. M-SO roller stock wasaccepted by the eighth month of the program.
IV. RCF TESTING OF NC-132 SILICON NITRIDE
The billets of NC-1on2 siliron nitoide supplied as rollerstock for this program initially exhibited highly variable rollingcontact fatigue performance. Previously qualified specimens ofsilicon nitride also exhibited a larger proportion of early fa-tigue failures than they had when originally qualified. Becauseof the inability to q?aqlify otherwise acceptable ceramic rollerstock, or requalify reviously accepted specimens, an effort wasmounted to clarify tae RCF variables pertinent to ceramic materials.
The study revealed the primary reason for fatigue life scatterwas the technique used for machining specimens. Lapping and honingdid not provide silicon nitride surfaces with uniform properties.Some scatter was attributed to the test equipment. Both of thesecomplications were resolved in this program. A method to fabricateRCF spe-imens by diamond grinding was developed that resulted inextremely uniform finisbes and acceptable geometries. Testing ofof these specimens on repaired test rigs resulted in acceptance ofthe silicon nitride billet stock.
"1-
00~r4-
o 000A l m"
"0 r- e4 N0 n00t m0Lo 05 t4 * 4 Ok
td 4 NCD 0 P "4
0 0 0 00t; 0 0 00= 40U 9-4 M
0~ ~ s-4 P-40 " O ~ *U) ~ ~ ~ ~ ~ ~ . *-4~. 0 * ~ 9f
"4J
"CD 0 LZ 0""4 V-4 m~* N 0 w0e
C r-4 "4 LM %C M"4P r*1Ul V4
""4 C;4 ; 41f-H
cuu
0 bb '-S"P4
41 bo (* h b- * 9 l c 0 :34) 4) -r4 F4 +-J.. ~ ) 60'
r4 ro U) 0
0k u
"41-2 'e4 m
"Q 0 oo41L
0 0 0 Al In 0 L
FU A- A0'"WJ 4i
A. Introduction
The rolling contact fatigue test is a rapid meany of eval-
ring bearJng materials. The tVst equipment applies a high wagni-tude (700-800 KSI), cyclic contact Vtr0ss to the tlst sample, acylindrical rod, a proAimately I cm OD x 7.5 cm long. This rodis rotated arosind Its cyiindrical axis at 10,000 rpm between twocrowned steel discs which apply a contact stress to a 0.1 cm widecircumferential band on the specimen. The test band undergoes1.2 million stress cycles per hour and at the stress balneg used,several times the design stress for bearings, M-S0 steels failreproducibly in 4-S hours. Use of this test to q'alify ceramicmaterials for rolling contact 1earitngi poses unique complicationsin that ceramics do not po3sess microplcsticity at ambient testtemperatures. Ceramics fail when the local stress exceeds theinherent strength of the material. Several aspects of RCF ma-terial evaluation can result in magnification of the nominallyapplied stress.
1. Material Flaws - inhomogeneities at or near the highlystressed contact surface can act as stress concentrators;i.e., pores, or large inclusions.
2. Machining Flaws - on the surface, scratches or sMall pull-outs have sharp boundaries and reatly multiply nominalstresses. A uniformly fine finish, with no large "occa-sional" flaws, is a necessity.
3. Subsurface Damage - even though the contact stress field'smagnitude decreases rapidly with penetration; it can inter-
sect subsurface grinding damage left by improper machining.Resulting failure is net typical v1' the material, but of thefinishing technique.
4. Tost Rig Alignment - causes additional strain in thesample. Ceramic's high Young's modulus makos these strainsinduce higher magnitude stresses.
S. Loading Disc Condition - the roughness of the loading disccrown can play a major role in the RCP performance of bothceramic and metallic specimens. Line asperities from crowngrinding act as point indenters on the test surface, gener-ating very hiah magnitude, localized stresses. Metallicmaterials yield microstructurally to reduce these stresses;ceramics, generally, do not. This variable is closely con-trolled as a result.
Thus the RCF test results reflect several, interdependantvariables in addition to basic material integrity. In all cases,metallic materials will be more forgiving of anomalies than
ceramics; reproducible test results on ceramics are a direct fun-tion of the uniformity that can be maintained in each variable.
12
The following sections discuss the RCF performance of the supplied
roller stock in light of th- above considerations.
B. Diamond Honed Specimens
The first set of RCF specimens provided for qualificationof NC-132 billets, 300S02, 300503, and 300304 exhibited only oneof eight lives in excess of three times the steel Q-bar life; thequalification standard. These specimens, of marginal geometry,and exhibiting varying degrees of longitudinal and spiral scratcheshad originally been judgid suitable for material qualification butafter early results a second lot was prepared. This second lot ofPCF rods, also prepared 'y 320 grit diamond cylindrical grindingfollowed by a 400 grit diamond hone, Pxbibited considerably bettergeometry but surface finishes were poor; in the range of 2-1/2 to6 micro inches AA. This average roughness, in excess of that con-sidered acceptable, was further cornplicated by dense distributionsof fine scratches. These fine scratches are felt to contributeheavily to poor rolling contact fatigue performance because theyact as preferential initiation sites for Hartzian cone cracks dueto their extremely high stress concentration factor. The surfacesof th's second lot of specimens were further prepared by handfinishing (diamond stropping) to remove visual evidence of scratches.RCF test results were still unacceptable, probably due to the addi-tional finishing operation's not removing subsurface damage. Table3 suamarizes these results.
C. Lapped Rods
The third set of RCF specimens for billets 300502 thrcugh300504 was prepared by a lapping technique. This series of testspecimens exhibited better RCF performance than thos previous toit, but still possesseZ an unacceptably high percentage of verylow fatigue lives at the 800,000 psi stress level. In addition,the character of both the failure spall and the waar track re-sulting from the RCF evaluation differed from that observed inprevious programs. Figure 1 shows a failure spall on Rod 13,billet 300502, in whi-h not only a spall but an area where sur-face material has been removed is apparent. figures 2A and B,
illuminated by polarized light to highlight cracks by internalreflection, show cracks both within and at the edges of the weartrack. The cracks within the wear track may be indicative ofeither si~idding or a weakened surface layer of material. A de-tailed review of both the RCF testing techniquen as well as thespecimen preparation operations was initiated.
D., Technique Investigations
Ceramic materials are brittle and highly susceptible topoint loading. An increase in the surface roughness of the steelloading discs used in the RCF test could adversely affect the con-tact fatigue life of the silicon nitride ceramic. Review of thetechniques used in finishing the loading discs showed that the
13
TABLE 3RCF Data Sumary - 800,000 psi Stress
AxialSurface .Roughness RCF Lives
Rod Billet C o(x I0 - Comments1 300502 1.9 0.8 Narginal geometry,2 300502 2.4 17.76, 0.2, 1.6 originally aocepted for
mate•i'el qualification.Cylindrical grind/4003 300503 3.0 4.5 grit diamond hone
4 300503 2.6 (probably contaminated)43ngitudinal scratches, A5 300504 3.2 spi-al hone marks, 5026 300504 3.8 7.8S shore chatter.
7 300502 5.7 4.1S, 5.4, 0.4 Geometry acceptable,honed, stropped withdiamond to remove9 300503 2.2 scratches, ?rcbably
10 300503 3.6 didn't remove sub-11 300504 2.8 surface damage.12 300504 4.9 16.2 i
A SitWa-YaO$ 2.5B SiINa-Ya0g 3.9
13 300502 U1.0 0.2, 0.8 Cylindrical ground 22014 S00S02 u1. 0 0.2 diamond 0.001"/passinfoad, cast iron ring15 300503 P1.0 4.3, 9.4, So, S* lapped, light random16 300503 %.1.0 0.1 scratoes *13 not ca-
pletely -loan, geometry17 300504 ul1.0 12.05, 6.2, 2.6, accepthble. Definite1S.3S 0.2, 8.03 subsurface damage.
18 300504 "U1.0 12.0, 13.5, 6.60.02, 8.6, 9.8
19 <1.0 0.7, 29S, 2, S* , $2.252.5St
20 <1.0 2.3, 0.4, 35.•S 320 Cia. cylindrical111.2S grinei. O.0002"/pasn in-
*Suspension > L115 million at ?4APC, Trenton feed. Cast iron ringt700,OOO psi lap. Geametry/surface 4
finish acceptableS *Suspension t 600 KSI
14
I'
i i
t:I,[U
15
2A
Rod 13 150X,crossed nichols
2B
Rod 14 150X,crossed nichols
FIGURES 2A • B - Wear tracks Rods 13 and 14
16 _
technique itself had not changed; but that it is a hand operationand subject to operator variance. The roughness of the loadingdiscs used may have increased marginally during this program but,in the final review, it was felt that this wa's not a major factorin the change in RCF performance. Early in the testing of thethird series of RCF specimens, the controlled -lot of steel"-bars" began to show a large variability in their contact fatiguelives. This behavior was traced to a worn set of contacts in a
Fbalancing circuit for the load cell on the RCF tester. These worncontacts made exact determination of the load applied to a speci-men during testing uncertain. It is not known whether this condi-tion existed, to a lesser degree, during the testing of the firsttwo lots of RCF rods. After repair of the bridge circuit testingAresumed.
Review of the machining parameters for the third lot of RrFspecimens revealed that improper final finishing operations wereused. Specimens were cylindrically ground with a 320 grit diamondwheel to 0.005" oversize radius. An irifoed of approximately 0.0002'"per pass was used in this machining operation; samples were thendelivered for lapping. Prior to this operation, however, a 220grit diamond wheel at 0.001"1 per pass was used to reduce the radiusto 0.001"1 oversize. The specimens were then cast iron ring lappedwith 6 and 3 micron diamond paste to provide a matte finish; thisI was followed by a SnO2 slurry polish to remove any very finescratches. The resulting geometry and surface finish were excel-lent; however, there was undoubtedly considerable subs;urface damagein these specimens. Investigation of these rods with the KET sur-face characterization technique at QUAL-X Industries, Inc. did notgreatly clarify the surface characteristics of these specimens; butdid point out that the technique has potential use in ceramic sur-
face characterization. See Appendix A.
A replacement billet, 323010, was pressed after billet number300502 was judged unsuitable. All RCF test results for 300502 hadbeen extreme2y poor and it was felt that a second lot of materialmight shed further light on the RCF testing situation by samplingsilicon nitride materials from completely different lots than thoseoriginally provided. RCF specimens from this billet were ground
with a 320 grit diamond wheel at 0.0002"1 per pass radial infeed to0.001"1 oversizc ý!Xiameter, These specimens were then cast-iron ring
of this billet, still yielded fatigue lives ranging from less thana million stress cycles to over a hundred millhcn stress cycles.Additional testing on billet 300503 at 700,000 psi Hertz stressrather than the customary 800,000 psi Hertz stress yielded two sus-pensions at 43 and 58 million stress cycles. Testing of billet300503 and 323010 at the Naval Air Propulsion Center in Trenton,New Jersey, yielded four suspensions all greater than 115 mill.ionstress cycles. These results cannot be compared to the Federal-Mogul results as the test condl.tions differ greatly between thetwo facilities. It is also disturbing that no differentiation
17
between Rod 15 (300503) and Rod 19 (323010) was possible at NAPCeven though Rod IS was felt to have suffered serious subsurfacemachining damage.
Final finishing of this third lot of RCF specimens oy ringlapping was chosen as the most expedient method of obtaining anacceptable surface finish and geometry. Final fiiishing of RCFspecimens by honing did not provide acceptable specimens. Tosome degree, the variability noted in the testing of the lappedspecimens can he Justified on a historic basis. A previrus pro-gram, dedicated to investigating finishing techniques for siliconnitride, noted that honinZ as a final finishing operation providedspecimens which gave re r.iducibly long contact fatigue lives.5RCF specimens final finished by lapping techniques, from the samematerial and the same program, vxh2.bited considerably lower livesand much more scatter. See Table 4. For example, a specimen ofacceptable zilicon nitride material finished by lapping plus aSnO slurry polishl gave rolling contact fatigue lives ranging from0.S to 88 million stress cycles at the Federal-MogUl test facility.Rod 20, from this program and finished by the same teclmiqu•,yielded lives ranging from C.4 million to 111 million stresscycles; also at the Federal-Mogul facility. Accordingly, someof the variability noted in this program may be traced to thefinishing technique used.
TABLB 4
RCF Life Versus Finish for NC-1324
Finish RCF Life (!:ycles 106)
320 Diamond Cylindrical Grind 16!, 24, 5
320 Diamond Cylindrical Grind 77, 71, 34, 62, 12, -40, 11,plus 400 grit hone 11, 11, 12, 11, 10, 10, 11
320 Diamond Cylindrical Grind 121, 52, 51, 11, 11, 11, 9,plus 600 grit hone 11, 10, 12, 11
320 Diamond Cylindr-cal Grind 5, 3, 11, 2, 8plus 6V diamond lap
320 Diamond Cylindrical Grind 30, 89, 0.6, 39, 12, 11, 11plus 6V diamond lapplus SnOa slurry polish
220 Diamond Cylindrical Grind SO, 40, 3plus 600 grit hone
18
An additional component of variability may he attributed tothe RCF testing equipment itself. Coincident wich the testing ofthe lapped silicon nitride rods from this program, several testswere run on RCF specimens from other bearing programs that had beenused to qualify silicon nitride stock based on uniformly acceptablefatigue performance. These specimens now yielded variable fatiguelives ranging between 0.7 and S5 million stress cycles; similar tothe resulti currently being obtained on roller stock supplied.
E. Conclusions
A review of these results indicated that RCF rods fvcmLots A and B, both honed, gave highly variable and unacceptablecontact fatigue performance primarily because of surface scr&tchescaused by a contaminated diamond hone. The variability exhibitedby Lot C is attributed to several sources; machining damage is acertainty, load cell variability noted previously, and that leppedspecimens have historically exhibited wider scatter than properlyhcned specimens. Given that some scatter was expected from thelappe' rods and that the RCF te3ters were yieldin3 more variableresults than usual it was decided that either or both factors wereaffecting the RCI lives.
Within vprograms constraints, an effort was mounted at Federal-Mogul to eliminpte the effect of machiaing variables in RCF per-formance by systtmatic grinding studies, These efforts are dis-cussed ia Smc*Jor V of this report. The reral, of this study wastiat excellent surface properties were obtained without resortingto either lapping or honing as final finishing steps. RCF testingdata on ttase rods are presented in Table S.
TABLE Sr
RCF Lt.es of Diamond Ground NC-132at 800,000 psi Contact Stress
Surface Roundness RCF Lives*Billet/Rod Finish p" AA El" AA x 10" cycles
300S03/lT 1.5 is 6.3, 6.4, 6.6
300503/lh 1.5 20-30 15.9, 16.3, 6.2
300504/21 1.s 25 6.4, 6.8, 7.8
300504/2H 1.s 15 6.8, 6.8, 6.7
323010/3T 1.5 20 6.7, 8.7, 7.2
323010/3H 1.S 30-SO 7.3, 7.3, 6.7
*All tests were suspended after exceeding Q-bar lives rangingfrom 4.5-6.7 x 10' cycles 9 700 KSI with an average life of5.6 x 10' cycles
19
RCF testing of specimens fabricated by the doveloped grindingtechniques was carried out only to detect the early failures thathad been experienced previously. It was decided that if no failuresoccurred before the M-SO qualification bar life limit (S-6 x 10"ttress cycles) was reached, the material would qualify as rollerstock. This premise Is based on the fact that occasional longlives were noted in the damated bars. If the low lives were eli-minated by a machining tec~uique* sufficient data existed on thelonger lives normally expected.
Table S shows that all testin3 carried out on the ground
specimens was suspended at a level higher than the M-SO Q-bar life.There were no early failures. This data, combined with the occa-sional long lives exhibited by machining dazagt4 specimens, quali-fied billets 323010 '00303, and 300S04 as roller stock. The dataalso stiongly highlighted that uniformity of finish is a primeconsideration in these specimens. Lapped rods from billet 323010were, by averaging profilometry, smoother than the Federal-Mojulgrcund specimen.:, but at 800 KSI conti&ý.t ress levels yielded4 of 6 lives lower than the M-50 Q-bar. 1:e role of sub-surfacedamage in RCF life variability is also clear from comparison oflapped and ground data.
V. RCF SPECIMEN PRODUCTION DEVELOPMENT
The variability noted in RCF testing of current NC-132 billetstock material necessitated the development oi a machining tech-nique providing suitable specimens. Efforts carried out byFederal-Mogul were aimed at eliminating the final honing orlapping operation and reproducibly producing specimens by diamondgrinding only. A three phase effort was coriducted as outlinedbelow:
Establish conditions and parameters required to producethe desired uniform finish and geometry.
•roduce six RCF specimens from previously qualifiedsilicon nitride stock and conduct RCF tests todemonstrate the finis;iing technique's suitability.
Produce six RCF specimens from current billets (2 fromeach) and conduct sufficient tests to qualify material.
The finished RCF specimen is a right cylinder, approximately3 inches long and 0.3750 inches (*0 -0.0002) in diameter. Theroundness is controlled to less than 50 microinch and the surfacefinish is normally 4 microinch AA or better. The material blankform is typically 6 inches iong with a 0.4 inch square cross sec-tion and is sliced from the material to be qualified.
The grinding parameters established in the first phase thatwere employed in specimen preparation for subsequent work ore
20
summarized in Table 6. Characteristics of specimen% prepared inthe first and second phases is presented in Tablo 7. Prior tocylindrical grinding, the first phase bars wora surface groitnd toan octagonal cross section to reduce cylindrical grindinS time.This step was eliminated in phases two ani three; specimens wererough ground directly from square stock. Steel female centerswere cemented on the bar ends as shown in Figure 3A.
Figure 3B depicts the surface obtained by final grinding witha 600 grit diamond whiel on a cylindrical grinder equipped with afive micron particle filter in the grinding coolant line. Thissurface exhibits no large randoa flaws (common without the filter)and very little smearing. The surface finish is 1-2 microinchAA.The mottled structure evident 1.n Figure 38 is due to a high magne-sium silicate concentration in the test material; that the struc-ture is clearly evident attests to the minimal surface damageimparted in this final grinding procedure. While the surfacefinish of these first phase rods is quite acceptable, the geometrywas not satisfactory. These rods were 30-60 .'croinch out ofround and the 6 inch rods exhibit a 0.0005 inch barrel shape dueto deflection under the high finish grinding pressures,
The geometries of the second and third hase rods were im-proved by slicing the 6 inch rods into 3 in lengths betweenrough and intermediate grinds. The roundness was improved byimproving the steel female centers used to hold the rod.
The 150 and 320 grit diamond wheels were trued with a Nortonbrake controlled truing device used with a 60 grit silicon carbidewheel. They were dressed with fixtured Norton SiC dressing sticks.The 600 grit wheel was trued with the fixttired dressing sticks andcleaned with chlorethane prior to grinding.
The 320 grit diamond wheel produced a 3-4 microinch surfacefinish when the wheel was optimally conditioned. Freshly dressed,tha resulting finish was likely to be much rougher, perhaps 8-14microinch. The finish with a 320 grit wheel is, thus, difficultto control due to the critical nature of the wheel conditioning.The 600 grit wheel, consistently produced a 1-2 microin,h finishwhen the grinding stock was 0.0002"/side or less. Finish andgeometry generally deteriorated with greater stock removal.
The procedures outlined in Table 6 allow the production ofsound sil con nitride surfaces; possessing a uniform and repro-ducible machining flaw size distribution. Prior to the study,this information did not exist. Careful mounting of minimumlen th specimens during final grinding operations will reducedeflection and insure roundness is obtained.
21
TABLE 6
RCF Rod Diamond Grinding Parameters
!Mh Intetmediate Sent-Finish Finish
Wheel Grit* ISO 320 320 600
Wheel Speed (SFPlN) 6400 6400 6400 6400
Voik Speed (RPMl) Soo Soo Soo Soo
Table Speed (Il4) 30 30 30 is
I- eed (In./Pass) 0.0005** 0.0002S 0.0001 0.0001
After Grind:Rod Size (dia.) 0.396 0.385 0.3754 0.3750
Stock (In/side) 0.01 0.00S 0.0002 -U-
*Diamond Wheel Specifications10S SD150R100B69:?I • SD320R1003696CJ a D6OON7SB3S
"**0.00025 after rol is round
SITABLE 7
Grinding Development RCF Specimens
Surface Finish Roundness RCF LivesRod lljt iv" AA pi .. x 1 0S cycles
1-6 Scrap 1.5-2.0 30-60 ND
IA 428841 1.5-2.0 2S 17.4S, 2.9, 17.5SO
1I 428841 1.5-2.0 30 12.6S, 12.9S
ZA 428841 1.0-1.S 20 19.7S, 16.652B 4288,91 2.0-2.S 45 16.85
3A 428841 1.5-2.0 20 14.OS, 12.6
S - Suspension*Previous lives
22
, • - - .
3A
3 B
FIGURES 3A Vi B RCF rod, Fe e-ok lc-!ij- oguij
23 1
II
S ..... _ • ...... • .. .... ... .: .. .... ... ... ... . ... ... ... .... ...... .. ... .... .. ....i " :, ;: 2 -..7 -• •I
VI, COMPONENT MANUFACTUREOther than the material qualification difficulties discussed
in the previous section, metallic bearing component fabricationproceeded smoothly. At the close of the current contract, thefollowing work had been completed:
1. Retainers for both the metallic and hybrid bearingsare complete.
2. Endurance test tooling is complete.
3. M-SO steel roller material has been accepted.Final machining of steel rollers will be done atthe same time as the NC-132 rollers in order toobtain maximum continuity.
4. M-SO steel race materials for both hybrid andmetallic beatings has been accepted and isready for machining.
5. Silicon nitride roller material has now beenqualified and accepted for fabrication.
VII. EXPERIMENTAL MATERIALS TESTING
A. Fracture Energy
Fracture energies for the three original NC-132 billetsand the Si9N4-Y 2O3 (NCX-34) were determined by the double canti-lever beam technique at the Naval Research Laboratories. Thistesting technique is believed to accurately characterize hot-pressed silicon nitrides in the rolling contact environment;* butit yields data with a high standard deviation. The results in-dicate that there is some experimental difference betwe 'en 300502Aand 300504 but this is due to a single, unexplained, very high
value for 300504. Disregarding that one data point, there is nostatistical difference between the billets. See Table 8.
Testing of the SisNt-Y2O, (NCX-34) hot pressed material in-dicates its fracture energy falls in the same range as the NC-132silicon nitride and say be slightly (S percent) higher. This isslightly at variance with expected results based on the MOR dataon these two materials, which indicates that NCX-34 should possessa measureably h•..r fracture energy.$
B. Hertzian Contact Characterization
The rolling contact bearing environment is characterizedby essentially static loading to very high magnitudes. SeeAppendix B. Static indentation of various silicon nitride
24
TABLE 8
Fracture Energy Measurements
Individual Values Sample Mean StandardMaterial T____ .....__/tM Deviation
NC-132 27.8. 33.3, 23.1 6.2(300502) 18.0, 19.6, 30.7NC-132 A08 2 3 . .
(300S03) 32.1, 40.8, 24.6 32.5 8.1
NC-132 30.7, 29.9, 3S.6 39.6 15.6(300504) 34.5, 67.2
NCX-34(Si,$N- 18-62 34 14S~81 Y103) :
materials, using MgO, YwOs, and ZrO2 as densification aids, with
various surfacb finishes and in two environments using tungsteaicarbide indenters was studied as a possible materials characteri-zation tool. The Hertzian stress fields resulting whon semi-S~infinite plates are indented with spherical tools have been de-scribed in the literature.'," The use of acoustic emissionequipment to detect the formation of the "cone" crack has alsobeen described for glass specimens."1
Hertzian indentation testing of silicon nittide surfaces withtungsten carbide balls at constant indentation rates* was carriedout to determine the technique's sensitivity to several materialvariables. Failure was detected by acoustic emission equipment**and verified by metallographic investigation.
Plates of silicon nitride materials nominally 1" x 3" x 1/8"were indented with 0.1875" diameter WC spheres*** at a crossheadspeed of 0.02"/minute. A piezo-electric transducer was attachedto the specimen whose signal after amplification was integratedelectronically for display. Recording of both load and summedacoustic emission was done on & dual strip chart recorder. Figures4 and S show the equipment layout. Figures 6A and B show typical
load/emission plots. The large emission of sound was demonstratedto correspond to the formation of the cone crack, the physicalnature of which is depicted in Figure 7.
*Instron Mechanical Tester, Canton, MA**Duneagan Acoustic Count Totalizer NS-lA***Kennemetal Inc. Latrobe, PA
25 L
FIGURE 4 - Hertzian indenter 1
OfI
FIGURE 5 Equipment layout .
26 .
. .
JIMI4)
U3
4)0
S M C4
27
31
0 L
6) u
0
eq 41
228
ii•
li
FIGURE 7 H ertzian cone crack SOX SEM
The cone cracks are difficult to detect in an opaque material.Dye penetration characterization of the cone cracks is possible,but of low sensitivity. Under c"-Dssed nichols in reflected light,internal reflection of light in tha crack allows the cracks to bedetected and accurateiy measured. Figures 8A and B show both thecnntact ;:one where the carbide ball disturbed the machined surface(shear stress) and an arc of a cone crack under crossed nichols.
The results of the testing on standardized surfaces are pre-seidted in Table 9. The testing shows that the technique is sen-sitVve to changes in material (fracttre energy, most likely); butdoes not accurately reflect, measured differences in MOR or smallmeasured changes in fracture energy in a single material. Testingin a 100 percent humidity environment was carried out. NC-132samples were loaded to 7S-85 percent of their mean measured failureload, and held for up to fifteen minutes in the saturated environ-ment. Loading then proceeded to failure; no statistical differencein Pc, the critical load, was detected. Thus, the test has notI . demonstrated that stress corrosion mechanisms operate in NC-132.
Testing of abraded surfaces did not greatly reduce the load re-quired for fracture; however, heavy oxidation coatings did impairperformance. The heavily oxidized surfaces do not have the inte-grity of NC-132, most likely due to high contact stresses being
29
rii
A
Contact area9Ox
Crack front240X
FIGIIRE 8 I-ertzian contact areas of NC-132.
3 0
--------------------------------------------------- -.
TABLE 9
Hertslan Indentation TestingHot-Pressed Silicon Nitride,
0.187s" WC Indenter
Maximum StandardModulus of Fracture Contact Deviation
Rupture Uerfy Stress and SapleMaterial •,,)/t _Site..
MC-131(30 751 0. 2610 103/36(300302)
NC-132NC-13) 917 33 25S0 69/33r•0QS03)
(0C5132 910 40 2S70 62/S4(300S04)
',N%+S% ZrOt 896 87 2700 00/11
AN*+$% Y&O& 96S 34 2820 160/11
set up in the SiOa layer rather than in a "crack-blunted" siliconnitride surface.,
The Hertz indentation test is of sufficient sensitivity todetect large changes in material behavior. It is interesting tonote that static loading failures occurred in the region of 400KSI Hertz stress. RCF tests are run at 800 KSI; indicating thelarge effect of the lubricant film on stress reduction. Currentlythe test cannot be recommended as a qualification tool for bearingmaterials.
31
+. t
r
REFERENCES
1. W. M. Wheildon, H. R. Baumgartner, D. V. Sundberg, M. L. rortiFinal Report on Contract N00019-72-C-0299, Feb. 1973.
2. H. R. Baumgartner, D. V. Sundberg and W. M. Wheildon, finalReport on Contract N00019-73-C-0193, Oct. 1973.
3. H. R. Baumgartner and P. E. Cowley, Final Report on ContractN00019-74-C-0157, August 1975.
4. H. M. Dalal• et.al., Final Report on Contract N00019-74-C-0168.
S. H. R. Baumgartner, G. S. Calvert, P. E. Cowley, Final Reporton Contract N00019-7S-C-0197.
6. H. R. Baumgartner, P. E. Cowley, Final Report on ContractDAAG46-74-C-005S.
7. Hamburg, et.al., Final Report on Contract N00140-76-C-1104.
8. G. O. Weaver and J. W. Lucek - "Optimization of SiyN•-YO 1Hot Pressed Materials", Bull. Am. Ceram. Soc. 57-12(1978) 1131.
9. T. R. Wilshaw, J. Physics D. Appl. Phys. 4 (1971) 1567.
10. F. B. Langitan and B. R. Lawn, J. Appl. Phys. 41 (1970) 3357.11. W. B. Swindlehurst and I. R. Wilshaw, J. Mat. Science, 11
(1963) 16S3.
32
APPENDIX A
Evaluation of KET*Surface Characterization Technique
for Si3N4 Surfaces
prepared forNaval Air Systems Command
April 1978
J. W. LucekResearch Engineer
Industrial Ceramics Division
*Registered Trademark Qual-X. Inc.
A-i
A!bau ~mguA k~im
Evaluation of KETSurface Characterization Techinque
For Si3N4 Surfaces
SUMMARY
KIT surface characterization of NC-132* silicon nitride
specimens with bearing quality surface finishes was carried out.
Characterization by several other techniques; fluorescent dye
penetrant, SEM, Dispersive Wavelength X-ray analysis (BIAX). and
optical microscopy failed to substantiate the surface structure
revealed in the KIT signatures. Correlation of the [UT signs-
tures to RCF data for the surfaces proved difficult.
INTRODUCTION
The KHT surface characterization technique utilizes a surface
adsorbed radioactive tracer as a 'penetrant" to detect microscopt-
cally fine surface flaws. Krypton gas enriched with 3 6 Kr85 is
adsorbed on surfaces and then allowed to diffuse off while exposing
photographic emulsion. Cracks and pores in the surface retain moretaI krypton; due to higher surface area and more tortuous diffusion
i ~ paths; than a flat, smooth surface and darken the emulsion to a
greater degree.
Several RCF test specimens of silicon nitride with various
surface preparation techniques and documented contact fatigue
behavior were surface characterized by lET. Efforts were made to
relate the KET signature to either finish technique or RQF life.
Pores or machining flaws in HP Si3N4 are not easily detectable
by traditional penetration techniques.
*Norton Company
A-2
1.t. ,• .......•_. .' frauu,, A•ihh, /:w * h--m a
NMOR N COMPANY
IBT C ,WMCTIRIZATIGo
Four runs were made on each sample to verify that the
signatures obtained were due to the sample surface structure
and not process variables. The typical pore size of NC-132
silicon nitride 4s less than Su and the machining flaws ox-
pected wsr* evea finer.
Process variables with high speed emulsion could conceivably
play an important part in the results obtained. KET signatures
similar to each other were obtained each time. It is felt thatthe vignatures obtained are characteristic of the surface studied
and not due to a controlled process variable.
Three characteristic surface signatures were obtained from
the RCF rods. The first signature is fairly uniform in density.
This type of signature would be expected from a uniform distribu-tion of fine pores. This particular rod was ground with 150 and
320 grit diamond wheels sequentially and finished with 1000 grit
SiC. This rod had highly variable (poor) RCF characteristics. A
The second signature is one that shows varying degress ofresponse to KET. This density variation should represent a con-
siderable diiference in porosity of the sample. This rod was
150 and 320 grit ground and then 400 grit diamond honed. its
RCF performance was excellent. Another rod with a similar sig-
nature, 100 grit and 320 grit diamond ground with no iinal step,
also gave excellent life in RCF testing. Traditional liquid dye
penetrant, optical, and SEM characterization do not reveal any
of this structure. See Figure 1.
A-3
AikWIAr WnLi.-fA m i&A" idA ~
NORTON COMPANY
The third signature shows an "island" structure where the
indications are betwvoa SO0 and 2000p in diameter. This parti-
cular sigtature was found an throe rods. Sample 73-2S was 220
grit gSotad and then lapped with 6 and 3jI diamond paste. Rods
1S and 17 were 220 grit diamond ground, lapped with successively
finer diamond to 3P diamond, and polished with CeO2 . All threeare felt to have suffered varying degrees of subsurface damage
in their processing. RCF performance on all three was fair to
poor. See Figures 2 and 3.
This is the most interesting structure, particularly in the
light that all these rods are felt to have some sub-surface damage.
One of these samples was given a thorough optical, SEN. and EDAX
characterization in efforts to relate the "island" structure to
discernablo surface details. A reconstruction of the island
structure signature at 100X appears in Figure 4. Indications of
such a structure were looked for in various SEN modes. See Figures
5-8. No concentration gradients of the elements detected by dis-
persive x-ray clearly fall in a similar pattern. Figure 8, SEM
in backscatter mode 1000X, clearly shows the pores in the surface;
but scanning the surface shows that they are uniformly dispersed.tA
The source of the potassium in the EDAX characterization is un-
known. The potassium concentration in NC-132 silicon nitride is
considerably below the threshold of sensitivity for dispersive 'I
x-ray analysis. Conference with both Qual-X and Kodak have not
revealed any likely source. Optical microscopy of the same area
A-4
L W IiWiih'// S'awak, ~diei'. Ih Am A A ammmAk
NDRTONCMNAfAY
with crossed nichols did not detect fine surface cracks due to
internal reflection; nor any other indications of flaws. Fluorescent
dye penetration revealed none of the island structure.
The [UT penetration technique routinely detected the spalls
and Hertzian cracks due to RCF testing. It is interesting to note
that in most cases the test bands are lighter than the surrounding
material in the signature, perhaps indicating the surface struc-
ture is filled with wear debris from th. testing.
CONCLUSIONS
The XiT signature for a very uniformly finished, high contact
fatigue life sample indicates considerable variations in surface
porosity. Such a surface would not be expected to give good per-
formance. The KIT indication of uniform porosity, a desireable
characteristic, occurred on a rod with highly variable RCF per-
formanze. Rods felt to have been damaged in their near surface
layers with poor RCF performance gave an island structure in the
lET signature. This indication is significant; but based on the
previous two signature types this result must not be accepted as
completely reliable. If the third signature is an indication of
subsurface danage in HP Si3N4, KET is the only NDE technique to
reveal it to date. That single characteristic makes this technique
of interest for ceramic bearing surface characterization.
RECOMBNDAT IONS
Because of the technological promise of the KET characteriza-
tion technique further work on subsurface damaged materials is in
order. A-S
W I~at~w g,.~. &.~,/e.A-5
- w• • • ••. .....~r . •- '•• .*•• • . :.-•. wvl• • ,•-• • .q•r ,'" Ifl • .. Y•'• <.•7-'7•'": : .- ,•
1. Simpler specimen geometries - test bars.
2. Various finishing techniques. ,..
3. Relate to simple strength characteristics - flexure.
4. ConStant base material.
5. Documented control of KET technique
A. film-lot
B. emulsion contact with surface
C. sample cleaning procedures
D. uniform processing techniques. -.
Once such an investigation has been made, the known strengthand finishing relationships will be deftiAtively related to charac-
teristic XET usefulness of the technique can be "deiermined.
A S
h" W mo*9&#*#1&f~vh Air vff 07"~r.offf
A-7q
AWbIAP A NOwO COMPANY---
NORTON COMPANY
FIGURE 3 - Rod 72-25 KET Signature 2.5X
A-8
S~~Industriol Ceromics aivision /Reseorch • Deva/opmunt •
NORTN 'COPANVY
Ati
IA
to's
0 64
df, -9
F ~ IOURS 4 ' lau iaul OC~rc"I
AL
• ~NORTONI "COMMNtY
' ~Ii
,, I
.I
-. A
SEM - Secondary Emission - bOOX 1]
SA-I lna#tralCermic Ovi~on/R~tc•l avB¢I
--
" - -
NORTON COMPANY
.I ...
I , ,
'I W c ' .p1.
"I , A•
A-11
VI IV '$DAX Fe concentration map IOOX
11
! -
EDAX W concentration map bOOX
,I FIGURE 6 A-ll
NORTrON COMPANY
EDAX Element Plot
AC
EDAX K Concentration Map 10OX
FIGURE7
NORTON lCOMPANY
12
FIGURE 8 ~MBackscattered Modeg0 spot 1000x
A-13
4 W /n u~fr,/ ~ara~'c: /,~W n/6'segrh d ev1/prwq
!I
APPENDIX B
Semi-Static Loading on RC Environment
Response of'ceramic materials to stress fields must be treated
7. dynamically if the stress fronts advance at significant portion of
the velocity of sound on the material. Thus, the rate of speed
with which the-Hertzian contact zone advances must be considerably
less than 101-104 m/s. In the RCF test, the test specimen rotates
at 1050 radians/second. The calculations below justify the con-
sideration of this test as essentially static, and thus, similur
to indentationHertzian testing as far as loading rate is considered.
The contact zone in the 800 KSI RCF test is an ellipse due to
the fact that the steel rollers are crowned. The major axis of the
contact area, parallel to the RCF rod axis, is the direction in
which the major stresses are applied and approximately 0.015 inches
long. Assuming the contact area to be planar, a reasonable assump-
tion, the minor axis is 0.13" long.
At a rotational velocity of 1050 radians/second the sample's
major axis will 4ncrease from 0 4 0.015" in 0.005 sec. Thus the
contact najor axis increases at an average rate of 0"/second or
0.07 m/sec; considerably lower than the acoustic velocity.
The initial increase ii, "diameter" of the contact area occurs
at a rate approaching infinity in the unlubricated state. The
actual rate is much lower and the sizes are only approximations
because of the lubrication film present during testing.
B-1
DISTRIBUTION LIST
NO, Of Copies
Naval Air Systems Ceumand • 20WeabIngton, D.C. 20361Attention: AIR 950D 11
310C 1330 1536 1520321 6
Office of Naval Research 1Washington, D.C. 20360Attention: Code 471
White Oak Laboratory 1Naval Surface Weapons CenterWhite Oak, MarylarA 20910 ,Attention: Code i301
Naval Research Laboratory 1Washington, D.C. 20390Attention: Code 6360
David W. Taylor Naval Ship Research & Development Center 1Annapolis, Maryland 21402Attention: W. Smith, Code 2832
Naval Air Prcpulsion Center 1Trenton, Now Jersey 08628Attention: R. Valori, Code PE 72
Naval Undersea CenterSan Diego, Califorri±a 92132Attention: Dr. J. Stachiw
Naval Air Development CenterMaterials Application Branch, Code 6061Warminster, Pennsylvania 18974
Air Force Materials Laboratory 4Wright-Patterson Air Force BaseDayton, Ohio -45433Attention: D. H..Grahams, LLM 1
Dr. R. R•b, LLK 1Dr. P. Land, LPJ 1Lt. B. Tagge, LTM 1
Air Force Aero Propulsion Laboratory 1Wright-Patterson Air Force Base, Ohio 45433Attention: Ron Dayton (SFL)
SI i
'I'
Aepoteo molov Laboratory
U.S. Almy aeoaTeohly Laboratvris'
FotRustles, Vrgsin ',23604Attentions DAYDL-ATU4TP (Mr. Pause)
U.8,, A t Rasearch Offlee IBox CH& Duke StationDurum, North Carolina 27706
Attentiant CIDARD
U.S. Ar• 1MRDCrt Blvoirs Virginia 22060
Attention& W. Mc•overn (S1MM-3P)
Arq? Materials &W Mechanics Research CenterWatertowmn Massachusetts 02172Attentiow: Dr. R. N. WAcs
NASA ,.adquertersWashington D.C. 20546Attention: J. J. Gengler, RIM
HASA Lews Research Center 221000 Brookpark RoadCleveland$ chio 44135
Attention: Dr. 1. laretiky 1W. A. Sanders (49-1) 1
and Dr. T. aergell
Defense Advanced Research Project Office 21.400 Wallacen Boulevard
Arlington, Virginia 22209~tcentiont Dr. Van Reuth 1
Mr . Buckley 1
Inorganic Materials DivisionInstitute for Materials PtesearchNational Bureau of StandardsWashington. D.C. 20234
University of Palifornia 1Lamrence Berkeley LaboratoryHeart Minina Build•tgBerkeley, California 94720Attention: Dr. L. 'roschauer
; • 2
ILI
Deparcment of tlgtoerins -.University of Calionla1-s Aoseles., California 9004Attentmimo n rs. J. V. Knapp mad 0. Sinee
Department of Metallurgy 1Case-Vastern Reserve UniversityCleveland, Ohio 44106Attentions Dr. A. Heuer
Engineering Experiment Station 1Georgia Institute of TechnologyAtlanta, Georgia 30332Attention: J. D. Walton
Department of Engineering Research 1North Carolina State UniversityRaleigh, Nori.h Carolina 27607Attention: Dr. H. PaluourMaterials Research Laboratory 1
Pennsylvania State UniversityI' University Park, Pennsylvania 16802
Attentions Prof. Rustum Roy
Rensselaer Polytechnic Institute 1"110 lighth StreetTroy, New York 12181Attention: R. J. Diefendcrf
School of Ceramics 1Rutgers, The State UniversityNew Brunswick, New Jersey 08903
Virginia Polytechnic Institute 1Minerals EngineeringBlacksburg, Virginia 24060
Attention: Dr. D. P. H. Hasselman
Aerospace Corporation 1Materials LaboratoryP.O. Box 95085Los Angeles, California 90045
Supervisor, Materials Ingineering 1Departlmnt 93-39(AiRoaserch Manufacturing Company of Arisona402 South 36th StreetPhoenix, Ariaona 85034
Li
AV6 Waptom latIM
At~tca: Tern Yeah
stgatloa, Commctlest 097
Attutli: Mr. 3.. Na~ale.
Ut&U fttfpe I~~a*t* S
Deel wraltme 6 MU 43M t
ma*~ and Covde Inoato Qut
MItll ginulval mattt505 E~s a~mo
P.O. pe.. ut
Attentlem: MR. battlesht
Cale %*usa rctome"
eCoAlosalm Cintor
Root Peorias.no Ulime411Attentlas: A. It. Comady
km 4"state Callego e, msyl',amia 3E~1
lsta Lam. Clalfif gxaa 70
4
go. of Copies 'C06CS Niwcalaln CAmPaaWGo math Streetaelden , Colmad. NOAMLI &uAtention: Rees"Tab Dspetin
Co1wbus India"a 47201Anetmsil Mr. L. EAM, Director ot Imssafth
cotieo-Vtiot c~am~y
eiimh "Mrmastical DivisionI am Passaic StreetIs.4-Ilidgo New Jersey 07075
DlylSIM Textron Corporation27 Booth StreetNew ahitafs coamecticit 06030
?.AG. er1mi"~ Corporatto.
Samford, Cmmicb15 08104Attentions Josephyar
3M00 Remenadhar DriveDAm rbors, l~chigm 28121Attenttisa Dk. G.lov ser1
Prodcrt, isgiopa. Group
Genral lectori Company
CeavcmtMic hioa 45215
GgmarL Electrig 10 Laboratories4P.O. Box Ischenectady, mew York 12301 lSpace fciences LaboratoryGoeneral Ilectric CompuiyP.O. Box 8555phi~ladelphia, rnuaay.vmla 110
S
No. of Copies
Detroit Diesel Allison Division 1General Motors Corp-orationP.O. Box U94Indianpolis, Indiana 46206Attention: Dr. It. Berman
OD Division 1
General Motors Corporationmayes StrsetSandusky, OhioAttention: H. Woerhle
Rughes Aircraft Company 1Aerospace GroupR&D DivisionCulver City, California 90130
UT Research Institute10 Waest 35th StreetChicago, Illinois 60616Attention: Ceramics Division
Industrial Tectonics, Inc.18301 Santa ?a AvenueCompton, California 90224Atteation: Hmans R. Signer
Keveki-Berylco IndustryBox 1462Reading, Pennsylvania 19603Attentlin: Mr. R. J. Longnebker
Rosearch and Developmnat Division 1Arthur D. Little Company
Acorn ParkCambridge, Massachusetts 02140
Mechanical Technology, Inc. 1968 Albany-Shaker RoadLatham, New York 12110Attention: Dr. B. F. Finkin
North American Rockvwll "cl$euce Center 1P.O. box 1085Thousand Oaks, California 91360
lollway lBering Company IDivL3ion Lips Corporation7600 Morgan RoadLivarpool, New York 1306&Attention: B. Dalton
6
. .tt L o..hC .. ,: .d ~ r-.. t, .j'x ..;.Ž. ... ~ .m ... d........
Ceramic Division .Saudi. CorpoationAlbquarqua• 1•w lMmico 57101
tIbWQievtil and Vesearch CWnter 1SKY Industries, Inc.1100 First AvenueKing of rusmia, IPennsylvania 19406Attentions L. Sibley
Solar DIvisto•laternatiaonal Harvester Company
2200 Pacific HiglwaySan Diep, Califorota 92112Attention: Dr. A. 0. Metcalfe
Southaest Research Instltute 1P.O. Drover 28510San Antonio, Texas 76228
Materials Science- G 041zxeering Laboratory 1Stanford ReseaTch InetrQkteMenmlo Park, California 14025 IAttention: Dr. Cublclotit
Teldyne CAR 11330 Laskay lodToledo, Ohio 43601Atteatiou IR. BeckThe Tinkon CompanyCanton, Chic 44704
Attentton: R. Couish
Marlin Mockwell, Divisio of TRW 1J~wtmn, Nom York 14701Mtantiou: John C. Lwtruce and A. S. lowin
Union Carbide C*r cration 1Pame Techtical Centert.0. lox 6116C1lvGI*ml Vk0o 44101
Mat~erils S~ciences taborst~orUnit~ed Airtcraft Corporationj
last iartford, Connecticut 06101Attentions Dr, J. 1. Slrewan 7
Pratt G witney Aircraft Division ItnLted ALrcraft G sotratiloamst Eartford, Connecticut 06106Mtentioan Paul Draw W2-.
Pratt A Witney Aircraft Division 1Ulaited Aircraft Corporationmiddletom, Conmecticu 06108Attention: L. 1. Vrledrich, MMPratt & Vbtvtmny JAlrcraf Divi~sionUited Aircraft Corporation
FPloridaI "D Center •
Vest Pla leach, Florda .;Attentions Mrt. J. Hisser
Aistromuclear LsboratorY1
Vestinghouse Research Cororrton
aurch£l]l Boronh&Pittsburgh, Pe ylvenla 15235Attentions Dr. t. Bretton
Vnisi Research Corporation1Velled Lke, 1Ichig:1 48086
I+K,•
a -
- i•• "'
