tribology in industry · 2018-03-16 · satisfied modern trends in tribology. development of 3d...

97

Vol.35,No.2(2013)97‐103

TribologyinIndustry

www.tribology.fink.rs

RoughnessandTextureConceptsinTribologyN.K.Myshkin

a,A.Ya.Grigorieva

a

Metal‐PolymerResearchInstitute,Belarus.

Keywords:

SurfaceroughnessRoughnessparametersTextureWearsurfaceDebris

ABSTRACT

Current knowledge on scale and spatial organization of engineeringsurfaces is presented.Mainmethods of surface roughness analysis arediscussed. A review of theoretical and practical problems in tribologyinvolvingconceptofroughsurfaceispresented.Criticalanalysisofroughsurfacedescription in theevolution from2D to3D setofparameters iscarried out. Advantages and disadvantages of traditional andmodernapproaches of surface analysis based on concepts of roughness andtexturearediscussed.

©2013PublishedbyFacultyofEngineering

Correspondingauthor:

N.K.MyshkinMetal‐PolymerResearchInstitute,BelarusE‐mail:[email protected]

1. INTRODUCTION

Surface asperities influence practically all theaspects of solids contact. There are a lot oftheoretical and experimental data on mutualdependence of roughness and suchphenomenaas adhesion, contact stiffness, abrasion andmany others which occur during friction andwear [1‐3]. It is commonly accepted thatunderstandingof frictionphenomenaisdirectlyconnectedtoanalysisofsurfacestructureanditstransformation due towear. So, at present anyfriction and wear model is involving surfaceroughnessparameters.Traditional concept of rough surface basedmainly on profile parameters is not fullysatisfied modern trends in tribology.Development of 3D parameters did not changethe situation significantly. Only recently, a newview of surface spatial organization was

introduced.Theconceptisknownastextureandit reflects the appearance of distinctive surfacepattern.Thecurrentpaperpresentsareviewoftheproblemsofroughsurfacesanalysisintheirevolution from statistical height and stepparameters of profiles to dimensionless andscaleinvariantrepresentationofsurfacetexture.2. ROUGHNESSORIGIN

Thedeviationof a real surface from the ideallysmoothareassociatedwiththeactionofvariousfactors, which can be divided onto structural,technological, and operational ones [4‐6]. Theirfeatures define scale and textural propertieswhichformsurfacegeometric irregularitiesandasaresultdifferenttypesofroughness.Structural roughness is inherently connectedwithdiscretenatureof solids.Having smallbut

RESEARCH

N.K.MyshkinandA.Ya.Grigoriev,TribologyinIndustryVol.35,No.2(2013)97‐103

98

finite sizes, the atoms and molecules of solidsare locatedwithin a certain distance from eachother. If assume that the boundary of solidscorresponds to some constant potential ofatomic interactionwith the surrounding phase,it isevidentthat itsreliefhasaperiodicnature.SurfaceimagewithatomicresolutionpresentedonFig.1aconfirmsthat.

Fig.1. а – AFM image of pyrolytic graphite surface(amplitude of surface deviation 0.43nm);b–deformationofgoldsurfacestructureundereffectof surface forces [7] (TEM); c – surface of spiraldislocation;d–surfaceofsteelfracture.Thereliefcharacteristicsatthislevelarecloselyrelated to surface forces and here significantquantum‐mechanical effects occurs such asthermaloscillationsofatomswithamplitudesupto 10% percent of interatomic distances andspontaneouschangesintheirposition.Moreovervarious atomic stacking faults in the surfacelayerproducecompensatingdeformationsofthelayer material (Fig. 1b). However, from apracticalpointofview,thesepropertiesarenotsignificant, at least for the present level oftribologicalproblems.Strict periodicity of atomic roughness ischaracteristiconly for idealcrystals.Realsolidsare imperfect. The imperfections of atomicstructureknownasdislocations formnext levelofphysicalroughness(Fig.1c).Crystallinestructureofsolidscanformlastlevelofstructuralroughness.Unliketoprevioustypesof irregularities which characterized by sub‐nanometer heights, the size of correspondedasperities is comparable to the size of

crystallites and can reach up to hundreds ofmicrons. Usually this type of relief is observedonthefracturesurfacesofmetals(Fig.1d).Technologicalroughness is a resultofmechanical,thermaloranyothertypesofmaterialprocessing.Thesesurfacedeviationsconsistofperiodicalandrandom components. Formation of periodicalcomponentsresultsfromtheprocessesofcopyingoftoolcuttingedgesandroughnessisaffectedbytechnology(Fig.2).Theappearanceoftherandomcomponents is associated with materialdestructionduringchipformationanditsadhesionto cutting edges (build‐up), work hardening andfatiguefailureofsurfacelayers,etc.

Fig. 2. Machined surfaces (Ra 1.6): a – cylindricalgrinding;b–turning.Errors in mounting of the parts duringmachining, thepresenceofelasticdeformationsand vibration in the machine‐tool system,cutting toolwear, and so on, leads towavinessand deviations of form (hour‐glassing, faceting,barrelling,etc.)oftherealsurfaceortheprofilefromthecorrespondingparametersspecifiedonthe basis of design. These deviations can beperiodicorstochastic.Operation roughness. Main reasons of surfacedegradation of machine parts during operationarewearandcorrosion.Nowadaysitisgenerallyaccepted that mechanisms of wear andcorrosion corresponded to certainmorphological types of formed surfaces andfracturefragments(wearparticlesoroxides). Itisabasisofmodernmethodsoftribomonitoringandtriboanalysis[8,9].The theoretical background of the methods isprovided by phenomenological models offriction contact damage. While using thesemodels the actual wear mechanism isestablished basing on classification of frictionsurfacemorphology(Fig.3).


99

Fig.3.Surfacedamageatfriction:a–abrasionwear;b – plastic deformation; c – ploughing and adhesivefracture;d–fatiguewear.Thus, the delamination of thin material layersand the formation of exfoliation and spallingregions are related to fatigue wear at cyclicelastic contact. It is followedby the separationof material debris which points to plasticdeformation of the surface layer at excessiveloading and lubricant film tearing. Theappearancelinearreliefofasperitieswithsharpedgesindicatesabrasivewear.Defectsshapedasdeep tear‐outs, delaminated thin films point toadhesiveandcohesiveinteractionofthecontactsurfaces. Thus, analysis of wear debris andfriction surfaces allows for evaluation of theoperatingconditionsoftribosystem,conditionoflubricant, and provides the opportunity topredict failure of the tribosystem and takemeasurestopreventit[10].3. SCALESTRUCTUREOFROUGHLAYERAs it can be seen the heights of the surfaceasperitieslieinawiderange.Onlowersidetheyarelimitedbythedimensionsoftheatomicandsupermolecularformations,ontheupperonebymaximal heights which are proportional to thelengthoftheexaminedprofile[11].It is evident that in this case there are nolimitations on the existence of asperities invarious dimensional ranges (both spacing‐wiseand height‐wise). However, in spite of the factthat there is no universally substantiatedcriterion for distinguishing asperities on thebasisofscale,atthepresenttimethereexiststhe

concept of the surface as an ensemble ofasperities of four dimensional levels:macrodeviations, waviness, roughness, andsubroughness(Fig.4)[12].

Fig.4.Diagramoftheheightandspacingparametersofsurfaceasperities.4. SURFACEMEASUREMENT

In studying the topography the need arises forthesolutionofthreebasicproblems:descriptionof the surface, development of representativesurface evaluation systems and technicalrealization of the measurement processes. Inspite of the fact that these problems areinterdependent, the last problem is of specialimportance, since our theoretical concepts, andtherefore our understanding of how anyparticular phenomena may take place on thesurface,arebasedonthequantitativeestimates.Therefore it is evident that roughnessmeasurements are of primary importance instudying the topography. Nowadays a greatnumber of experimental methods of surfacemeasurement are used. Stylusmethods remainthemostwidespread;theyyieldresultsformingthebasisforcurrentstandards.Opticalmethodsinvolvingelectromagneticradiationsuchaslightsection, shadow projection, interferencetechniquesetc.havebecomewidelyapplied.Atomic‐force microscopy has found a widespread in surface metrology now. Figure 5represents some capabilities of differentmethods of roughness measurement and theirvertical and lateral resolution. As can be seenthere is nomethod formeasuring full range ofasperitiesdeviations.


100

Fig. 5. Resolution of various methods of roughnessmeasurement.

Fig.6.Twotypesofsurfacetextures.Table1.RoughnessparametersofsurfacesonFigure6.

Roghnessparameter

SampleonFigure5a b

Ra 3.20.5 2.30.3Rz 19.34.2 15.92.0

Rmax 15.03.7 12.21.4S 56.718.6 43.27.6Sm 165.94.4 117.78.7

The foregoing concepts form the basis of therepresentationofasurfaceinsuchdisciplinesasmechanical engineering, machine design,technology, tribology, heattransfer, and so on.On thewhole in this representation the surfaceisexaminedastherealizationofarandomfield,thecharacteristicsofwhichareevaluatedonthebasis of two‐dimensional profilogram samples[12]. In this case the system of topographyestimates is basedon analysis of thehistogramcharacteristicsoftheasperitiesinsomerangeoftheirvalues.A characteristic feature of this approach is thefact that the mutual influence andinterrelationship of the asperities are generallyignored (except for the fact that the surfacesmaybeclassifiedasisotropicoranisotropic),i.e.,the spatial organization of the asperities is nottaken into account. We can illustrate theambiguityarisinginthesurfacerepresentationsin this case. Figure 6 a, b shows photographs

(obtained on a scanning electron microscope(SEM)) of surfaces having different spatialstructure. Table 1 present the results of acomprehensivestudyoftheirmicrogeometry.It can be seen that in spite of the significantdifference between the studied objectspractically all the quantitative estimatescoincideinthelimitsofthemeasurementerrors.Itisimpossibletodeterminecriteriaonthebasisof which we can judge the difference betweenthesesurfaces.The principal reasonswhy it is not possible toevaluate the topographic properties of surfacessolelywith theaidofhistogramestimateswerediscussed in [14,15]. Specifically, it was shownthat on the basis of these characteristics wecannot construct a satisfactory prognosticmodel, since in the finalanalysis it isvalidonlyin the limits of those values thatwereused foritsconstruction.Theuseofthismodelmayleadtounexpectedresults.Forexample,accordingtothe authors of [15], by superposing theparameterswecanachieveagooddescriptionofpracticallyanyphenomenon,includingthosenotrelatingtotheexaminedobject.Consideringthatthe modern instruments yield about fiftydifferentcharacteristics(including3D)thatmaybe subsequently used, the drawbacks of thisapproachbecomestillmoreevident[6]Formorecorrectrepresentationofthesurfaceitis necessary to have the possibility ofcharacterizing it as an object, having a definitetopology. In tribology this approach isformalizedbyconceptoftexture[16,17].5. CONCEPTOFTEXTURE

Surfacetextureisratherdifficulttodefine.Mostauthors agree that this notion reflects thefeaturesofthesurfacereliefcausedbythetwo‐levelmodel of spatial relations of irregularitiesheights [18,19]. Themode of these relations atthelocallevelgovernstheshapeofirregularitiesand at the global level, the position ofirregularities relatively to each other. To someextent,theconceptoftextureunitestheideasoftreatment direction and irregularity directionaccordingtotheUSSRStandardsGOST2789–73and 9378–93. The texture is outlinedqualitativelybyseveraladjectivescharacterizing


101

the shape andmutual position of irregularities,suchasstepwiselinear,facetrandom,spherical,sphericalradial,etc.Therearenumerousapproachestothedescriptionoftexture;however,allofthemarereducedtooneof the following: comparative and parametricalapproach, usage of invariant presentations, andparameterizationofvisualcontent.Comparativemethodsarebasedonexpertvisualevaluation of the similarity of the object underinvestigation and the reference.The featuresofman’s vision allow him to notice and identifyminutedistinctions inroughness, texture,color,and shape of objects. The simplicity ofcomparative methods and the fact that theyprovidesufficientaccuracyformostapplicationsencouraged their widespread use. Thus, forqualitative evaluation of roughness, samplereference surfaces are used according to GOST9378–93(Fig.2).Comparative methods are very simple and inmost cases a set of references and an opticalmicroscope are sufficient to realize them.Theirdisadvantagesarethesubjectiveandqualitativenature of the estimates obtained. To overcomethem, the opinions and agreement of multipleexpertsareused[20].Parametric methods use different statistics ofsurfaceasperityheightsandspacing,brightness,andcolorcharacteristicsoftheirimages.In order to evaluate texture properties,roughnessparametersaremostoftenused,e.g.,according to GOST 25142–82. Since 2007, theISO 25178 standard has been used to describe3Dsurfaceproperties.However,theyaremainlysimilar to standardprofile estimates andhenceinheritalltheirshortcomings[6,15].Theadvantagesof theparametricapproacharein the simple interpretation of the respectiveestimates, while their weak descriptive abilitycanbe considered a shortcoming.Nevertheless,when a great number of similar characteristicsare used, the approach is capable to solve theproblems with accuracy sufficient for mostpracticalapplications.The essence of invariant representation is theapplicationofnormalizeddescriptionmethods,i.e.,

the representation of analysis objects in a formindependentoftheirscaleandcoordinateorigin.The simplest procedure of invariantrepresentation is based on the Fouriertransform of surface heights [21]. Withrespective normalization, the coefficients of theamplitudespectrum(Fourierdescriptors)donotdepend on the scale and position and can beconsidered as a complete (single‐valued andreversible)systemoffeatures.Morecomplicatedmethodsuse fractal compressionof imagesandwavelettransforms[22,23].Features are not defined in the given approachat all. It is believed that all elements ofnormalizedrepresentationsare features. It isofno significance that they can be numerous anddonothave visual interpretation. It is assumedthat they are analysed and classified bycomputer methods; therefore, the size of thefeaturevectorisnotimportant.Theapproachisunsuitable for research because of the absenceofanyvisualandgeometricinterpretation.Parameterization of visual content is based onthe assumption that representative descriptionof texture is the only possible by means ofestimates reflecting the visual content of theobjectsunderinvestigation.Realization of the approach is based on theintroduction of a structural element, i.e., theminimal visually perceived region of the objectunderanalysis.Itisbelievedthatthefeaturesofdislocation of structural elements relative toeach other govern local morphologicalpropertiesatsmalldistances,andglobalonesatlarge distances. Differences in the choice ofstructuralelement typeanddescriptionof theirmutual position are responsible for the varietyofthemethodsforrealizingthegivenapproach.One of the methods is based on the use of co‐occurrencematrices (COM) [16,17]. The use ofCOMismotivatedbytheknownassumptionthatthe second order probabilities of featuresderived from the images reflect their visualcontent[24].In order to describe the texture by the givenmethod, a surface region is chosen as astructural element whose position ischaracterizedbythedirectionofgradientGiand


102

distancePifromthecoordinatecenter(Fig.7a).Themutual positionof two structural elementsisspecifiedbythedistancebetweenthemρandthe difference invariant description based onCOM(Fig.7b),thenumberofpairsofstructuralelements is countedwith certain ρ andgbeingpresentatthesurfaceareaunderanalysis.Bothheight‐coded and half‐tone images can be usedtakenbyvariousmicroscopymethodsarrangedasupsidelightening.

Fig.7.Parameterizationofvisualcontentsoftexture:a–schemeofdeterminationofstructureelement;b–matrixofco‐occurrenceoftextureelements.With respective normalization, COM does notdependonscaleandobjectpositioninthefieldofview.Aswith invariantpresentations,eachof theCOM elements can be considered as a feature.However,sinceCOMelementsdefinetheareasofthesurface,andthenthepossibilityarisesofnon‐parametric comparison of objects withvisualizationoftheirsimilarityordissimilarity[25,26]. The technique involves marking the areaswhose COM elements have either close oressentially different values on the images of thecomparedobjects. In the former case, this allowsforvisualizingthesimilarityoftheobjects,andinthe latter their distinction. Figure 8 shows theresultsofthesolutionsofthisproblem.

Fig. 8. The visual matching of surface texturedifferences:a,b–surfaceof twotypesofharddrivemagneticmedia;d–differenceofthesurfaceafromb(presenceof“kidney‐like”structures).The advantage of the approach consists in itsgeneralnature,allowingustounitethefeaturesoftextureofobjects.Thismakesanalysisoftheir

morphology and classification much easier.However,itsrealizationisrathercomplicated.Prospects in development of texture analysisand applications look very promising. Theanalytical and computational tools in textureanalysis are progressing quickly [27]. Theprogressintechnologyresultsinapossibilitytouse a variety of methods for making regulartexturesandmicro‐texturese.g.bylaser[28]orpatterning with rigid asperities [29]. Thesetechnologicaladvancescanbringalotoffruitfulapplicationsinmanyareasoftribology.6. CONCLUSIONSSurface 3D organization can be described bydefinition of texture. Experience of imagerecognition theory can provide methods forrough surface texture description andvisualization of texture similarity/dissimilarity.The description of a surface texture by specialtype of COM is in a good agreement with thetexture distinctions obtained by expert visualperception. Texture analysis can be efficientlyapplied for solving practical tribologicalproblemsinmicro/nanoscale.REFERENCES[1] N.K. Myshkin, C.K. Kim, M.I Petrokovets:

IntroductiontoTribology,CheongMoonGak,1997.

[2] Micro/Nano Tribology, Ed. By B. Bhushan, CRCPress,1998.

[3] G.W. Stachowiak: Engineering Tribology.Butterworth‐Heinemann,2005.

[4] I.V. Dunin‐Barkovskii and A.N. Kartashova:MeasurementandAnalysisofSurfaceRoughness,Waviness, and Noncircularity [in Russian],Moscow,1978.

[5] T.R. Thomas: Rough Surfaces, Imperial CollegePress,London,1999.

[6] D.Whitehouse:SurfaceandTheirMeausrement,KoganPageScience,2004.

[7] M. Pruton: Surface Physics, Clarendon: Oxford,1985.

[8] B.J. Roylance, R. Dwyer‐Joyce:Wear debris andassociated wear phenomena – fundamentalresearchandpractice,Proc.Inst.Mech.Eng.PartJJEng.Tribology,No.214,pp.79‐105,2000.


103

[9] N.K. Myshkin, A.Ya. Grigoriev: Morphology:Texture,ShapeandColorofFrictionSurfacesandWearDebris, Journal of Friction andWear, Vol.29,No.3,pp.192‐199,2008.

[10] D.W. Anderson: Wear particle atlas, ReportNAEC‐92‐163.

[11] R.S.Sayles,T.R.Thomas:SurfaceTopographyasa Non‐Stationary Random Process, Nature, Vol.271,pp.431–434,1978.

[12] N.B. Demkin and E.V. Ryzhov: Surface Qualityand Contact of Machine Parts [in Russian],Moscow,1981.

[13] A.P. Khusu, Yu.R. Vitenberg, and V.A. Pal'mov:Surface Roughness (Theoretical‐ProbabilisticApproach)[inRussian],Moscow,1975.

[14] A.Ya. Grigoriev, N.K. Myshkin, O.V. Kholodilov:SurfaceMicrogeometryAnalysisMethods, SovietJournalofFrictionandWear,Vol.10,No.1,pp.138‐155,1989.

[15] M. Zecchino: Characterization surface quality:whyaverageroughness isnotenough,AP.NotesofIvecoInstrument,December,pp.24–30,2003.

[16] A.Ya.Grigoriev,S.A.Chizhik,N.K.Myshkin:Textureclassificationofengineeringsurfaceswithnanoscaleroughnes,Int.J.ofMachineToolsandManufacture,Vol.38,No.5–6,pp.719–724,1998.

[17] N. K. Myshkin, A. Ya. Grigoriev: Spatialcharacterizationof engineering surface, in Proc.of 3rd International Conference on SurfaceEngineering,Chengdu,China,pp.54–61,2002

[18] R.M. Haralick: Statistical and structuralapproachestotexture,ProceedingsIEEE,Vol.67,No.5,pp.786–804,1979.

[19] M.Sonka,V.Hlavac,R.Boyle: Imageprocessing,analysis andmachine, Boston: PWS publishing,1999.

[20] A.Ya. Grigoriev, R. Chang, E.S. Yoon, H. Kong:Classification of Wear Particles by Semantic

Features, Journal of Friction andWear, Vol. 20,No.2pp.42‐48,1999.

[21] Z.Peng,T.B.Kirk:Two‐DimensionalFastFourierTransformandPowerSpectrumforWearParticleAnalysis, Tribology Int, Vol. 30, No. 8, pp. 583‐590,1997.

[22] T.B. Kirk, G.W. Stachowiak, A.W. Batchelor:FractalParametersandComputerImageAnalysisApplied to Wear Particles Isolated byFerrography,Wear,Vol.145,pp.347‐365,1991.

[23] S.‐H. Lee, H. Zahouani, R. Caterini, T.G. Mathia:Morphological characterization of engineeredsurfacesbywavelettransform, inProc.of7thInt.Conf. on Metrology and Properties of Eng.Surfaces,Goteborg,pp.182‐190,1997.

[24] H. Tamura, S. Mori, T. Yamawaki: Texturefeaturescorrespondingtovisualperception,IEEETrans.SMC‐8,Vol.8,pp.460–473,1978.

[25] N.K. Myshkin, A.Ya. Grigoriev, S.A. Chizhik, K.Y.Choi, M.I.: Surface Roughness and TextureAnalysis inMicroscale,Wear,Vol.254,pp.1001‐1009,2003.

[26] P. Podsiadlo, G.W. Stachowiak: Development ofadvancedquantitativeanalysismethodsforwearparticlecharacterizationandclassificationtoaidtribological diagnosis, Tribology International,Vol.38,pp.887‐892,2005.

[27] P.Podsiadlo, G.W.Stachowiak: DirectionalMultiscale Analysis and Optimization for SurfaceTextures,TribologyLetters,Vol.49,pp.179‐191,2013.

[28] I. Etsion, State of the Art in Laser SurfaceTexturing, ASME J. Tribology, Vol. 127, pp. 248‐253,2005.

[29] D.T.Nguenetal.:FrictionofRubberwithSrfacesPatterned with Rigid Spherical Asperities,TribologyLetters,Vol.49,pp.135‐144,2013.

tribology in industry · 2018-03-16 · satisfied modern trends in tribology. development of 3d...

Documents