Paper VII (v) Chemical and Microstructural Aspects of Debris Formation in Mild Sliding Wear

Paper VII (v) Chemical and Microstructural Aspects of Debris Formation in Mild Sliding Wear

Wear Particles - D. Dowson et al. (Editors) 0 1992 Elsevier Science Publishers B.V. All rights reserved. 293 Paper VII (v) Chemical and Microstruct...

1MB Sizes 50 Downloads 48 Views

Wear Particles - D. Dowson et al. (Editors) 0 1992 Elsevier Science Publishers B.V. All rights reserved.

293

Paper VII (v)

Chemical and Microstructural Aspects of Debris Formation in Mild Sliding Wear S. Fayeulle and I.L. Singer

ABSTRACT Wear tests were perforrnedon TiC hardcoating in air, at roornternperature in pin-on-disktesters. Wear scars and surfacefilrns were analyzed by scanning Auger rnicroscopy (SAM), secondary electron rnicroscopy (SEM), X-ray photoelectronspectroscopy (XPS) and transrnission electron rnicroscopy (TEM) equippedwith energy dispersive X-ray analysis (EDX). Cornposition of the debris and the debrishider interfacewere studied by SAM and SEM and the phases and rnorphology of the debris by TEM. SAM of transfer filrns showed titaniurn Oxide, not TiC transferredfrorn TiC to steel and sapphire riders. Cornparisonbetween analyses of transfer layers and rernaining coatings suggestedthat shear took place at the oxidekarbide interface. TEM diffraction Patterns indicatedthat the debris stripped frorn steel riders contained binary and/or ternary Oxide cornpounds Fe203and/or FeTi0,and that the ones stripped frorn sapphire balls were pure Ti%. These phases were accounted for by equilibriurn ternary and quaternary phases diagrarns which were calculated frorn therrnochernicaldata. This thermochernicalmodel is used to discuss the friction and wear results obtained for other treatrnents: MoS2, TiN and Ti and Ti+C irnplantation and to explain the various phases found in the debris particles.Chernica1 and rnechanicaldifferences achieved when sliders are changed frorn steel to sapphire will be ernphasized. Relations between friction, chemistry and rnorphology of debris particles will be discussed.

1. INTRODUCTION

that chemistry controls the wear process.

Wear of solids in sliding contact is usually treated as a rnechanicalprocess, and rnany well-developed models of mechanicalwear, based on deformation and fracture, have been published (1,2). However, chemistry can play an irnportant role in debris forrnation, especially during mild wear, when the average wear rate is less than a monolayer per cycle of sliding. In rnost wear processes, sliding rarely occurs along the original substratehider surface, but at sorne weaker interface. The Separation process allows surface layers frorn the Substrate to transfer to the rider and leaves "freshly" separated, highly reactive surfaces exposed to surrounding gases. Ac the transfer layers becorne thinner, the chernistry becomes increasingly m r e irnportant. Stated another way, the lower the wear rate, the rnore likely

Therrnodynarnicshas been used by tribologists to explain rnany aspects of wear, frorn chernical reactivity of solid lubricants (3,4,5) to adhesion and shearing of well-characterizedinterfaces (6). Therrnochernicalcalculations have been used extensively to account for tribochernical (rnechanochernical) reactions occurring in ball rnillc (7). While the irnportance of chernistry in tribology is well docurnented (8,9), and even glamorized by the discipline of "Tribochernistry," there have been few atternpts to develop wear rnodels based on chernistry. Quinn (10) and later Fischer (11) developed models to account for the wear loss of Oxides on steels, but the rnodels were rnainly rnechanical. Two wear rnodels based on therrnochernical calculations have been proposed

294

in the last decade. Krarner and Suh (12) modelled the high temperature wear of cutting tools in terms of their dissolution rate against various rnaterials. Gerkema (13) provided a modelfor the failure of lubricatingrnetal films based on the interfacial energies of the sliding contacts. This paper presentsthe resutts of a chernical and rnicrostructuralanalysis of the wear behavior of TiC hard coatings. A thermochernical model developed for analyzing b w wear rnaterials (14)l is extended to analyse the behavior of TiC coating. The therrnochernicalcalculations and equilibriurn diagrams needed to account for both solid and gaseous products of the sliding contact are presented. Surface analytical studies that reveal the chernical aspects of wear and the usefulness of the thermochemicalapproach are then sumrnarizedfor other surface treatrnents such as solids lubricating filrns of MoS2, TiN hardcoatingand Ti and Ti+C implantedsteels . Finally, lirnitations and recornrnended extensions of the present model as well as opportunitiesfor applying the thermochernical methods in future friction and wear behavior are discussed.

sputter depth profiling using a 3 keV Ar+ ion beam. Depth Profiles were quantified using the sensitivity factors determinedfrorn reference materials or with values suppliedwith the PHI 660 SAM Software.

3. RESULTS TiC-mated M2 Substrates were polished before friction test. Unidirectionalsliding experirnents with steel balls against these polished surfaces gave friction coefficient around 0.2 on the 15th pass. Very little debns was Seen in the wear track. Sorne wear particles were attachedto the ball and a few big scratches were Seen (Fig. 1). Experiments with sapphire balls gave friction coefficient of 0.18 after 15 passes. Nearly IX)debris was detected on the ball as shown on Fig. 2.

2. EXPERIMENTAL TiC coatings 0.2 to 5 prn thick, were deposited by reactive rnagnetron sputtering onto M2 tool steel. Frictionand wear rneasurementson TiC coatings were performed at rmrn ternperature under unlubncated sliding conditions in ambient air (20% < RH < 55%). Tests were performed in two sliding configurations: unidirectional and reciprocating sliding and continuous ball-on-dick geornetries. The balls were uncoated 52100 steel or sapphire, with diarnetersfrorn 3.2 to 12.7 rnm. Loads frorn 2 to 10 N were appliedto the balls, giving initial rnean Hertzian contact stresses, ignoring the TiC contribution, of 0.6 GPa to 0.9 GPa. Wear scars and transfer layerswere examined optically by interference (Nornarskyand Michelson) rnicroscopy. Transfer layers and loose debris were removed from uncoated balls using a cellulose acetate Stripping technique. Microstnictures of the stripped layers were exarnined in a 200CX JEOL scanning transmission electron rnicroscope (TEM), operated at 200 keV, equipped with energy dispersive X-ray analysis (EDX). Phases were identified by selected area electron diff raction (SAD) and EDX, and morphologies were characterizedby bright fiekl (BF) rnicroscopy. Layer thickness was assessed qualitatively by brghtness contrast. Films which were uniforrnly transparent across the viewing field were judged to be ?hin flakes.” Transfer layers on ball riders were exarnined by scanning Auger rnicroscopy (SAM) and secondary electron microscopy (SEM) in a PHI 660 Analyzer. Composition of debris and of the interfacial film left frorn the Stripping process was deterrnined by Auger

Fig.1: Wear scar on steel ball after 15 passes against TiC coated M2 steel

Fg.2: Wear scar on sapphire ball after 15 passes against TiC coated M2 steel Longer duration tests with sapphire balls were perforrned on ball-on-disk machine. Friction coefficient varied frorn 0.18 to 0.2. Michelson interferometry Observations revealed that depth of the wear track was about 150 nm after the 2000th

295

Fg.4: Schematic of the wear process for bw pass and about 300 nrn after the 7000th pass. The wear rate was thus lirnitedto less than a monolayer per cycle of sliing. Auger and XPS analyses of polished TiC revealed a thin titanium Oxide surfacefilm, cornparableto the one observed on TiN coating (15). SAM perforrned on 10th pass debris generated during sliding against steel ball indicatedthat the thin particles in the contact area were at this Stage, mainly formed of titaniurn Oxide with iron. For bnger duration tests, debris were stripped from both steel and sapphire balis and examined by TEM and SAD. SAD patterns obtained from debris collected on steel balls indexedwell to a rhombohedral structure similar to the one observed when friction is achieved against TiN coating (15). EDX analysis always showed Ti and Fe Peaks. As in the case of TIN, it has been concluded here that debris generated by steel sliding against TiC in air is the rhombohedralternary Oxide of the solid Solution series joining FeTi03 to Fe203. In addition, a few other Spots of the diffraction patterns were attributedto Ti02 and to metallic iron. SAD investigationof debris stripped from sapphire balls indicatedonly one phase, Ti02 rutile, and EDX analysis showed only titaniurn. Bright field microscopyof debris in both cases (steel and sapphire sliders) showed the classical thin and uniformflake-motphology (Fig. 3) which has been fully describedfor friction against TiN (15) and which has also been encountered during friction experirnents involving MoS2 coatings (16,17) or Tiimplanted steel surfaces (18).

Fg.3: Bright field micrograph of thin flake debris

friction surfaces 4. WEAR PROCESSES AND THERMOCHEMICAL

ANALYSIS

The wear of low friction, b w wear TiC surfaces can be describedschematically as follow (Fg. 4): initially (ieft), the titanium Oxide surface film is removed from the TiC Substrate and transfers to the rider (steel or sapphire). Next (Center), the transfer layer reacts with the surroundinggases (e.g. 02) and, ii allowed chemically, the rider material, M. Finally (right),as more layers accumulate and the transfer layers thicken, debris particles fall off onto the wear track. The question of whether or not a reaction proceeds can be answered by thermochemical calculations: if A and B are the reactantsand C and D are albwed products, then the more stable cornbination is the one with the lower Gibbs free energy, G. Thus, for the reaction aA+ bB = CC +dD one evaluates the Gibbs free energy for each side - ~SO298 (in practice, one computes GT = 6 ~ 0 2 9 8 using room temperature vaiues of 6 ~ ~ 2 and 98 SO298 (19)). The side with the iowest sum is the more stable. For Systems of three or more elernents, we have performed these calculations with a cornputer prograrnthat balances equations by the method of determinantsthen evaluates G with data readfrorn a data base. Results of these thermochernicalcalculations are most easily surnmarized graphcally as equilibriurn phase diagrams. The procedures for constructing ternary (19,20,21,22) and quaternary (15,16,23,24) phasediagrams have been discussed in detail by several authors. Fig. 5 shows an isothermal section of theTi-C-0 phase diagram. Incases where empirically determined tie-lines have not been established (25), they were calculated frorn standard enthalpy and entropy data of the elernents and cornpounds depicted (19). The reaction line frorn oxygen to TiC is represented by the dashed line. It is shown that the ternary diagram constrains TiC to oxidiie to Ti203. In Order to analyze the reaction of TiC in contact with steel in oxygen, one must examine all Fe-TiC reactions and all Oxidation products of those reactions, i.e. Oxidation reactions of Fe-C and Fe-Ti

296

cornpounds as weil as the Oxidation reactions of Ti-C cornpou nds. 0

/ I

TiC

c.

Fig.5: Calculated isotherrnal section of the Ti-C-0 phase diagrarn One graphical representationof these reactions is illustrated in Fig. 6. The three Oxidation ternaries are attached to the three binary sides of the Fe-Ti-C ternary diagrarn and a pyramidalrepresentationof this diagrarn is rnade by "folding" each of the outer three triangles upwards along the sides of the Fe-TiC triangle. The four exterior surfaces of this isotherrnalsection of a quaternary phase diagrarn are ternary phase diagrams with tie-lines. In addition, its interior is subdivided into four-phase equilibrium regions by intenor tie-lines and tie-planes. Ac was done with ternary diagrarns, reaction lines can be drawn through the interior of quaternary diagrarns. In general, any point (composition) on the reaction line is actually a rnixture of four phases, deterrnined by the vertices of the four tie-planes surrounding it.

Quaternary tie-lines and tie-planes have been calculated using a procedure developed by Bhansali et al(24). Phases resultingfrom the contact of steel with TiC in air are determined as follow: reaction lines are drawn between the possible reactants: the products are identified as the vertices of the tie-lines and the tieplanes through which the reaction line passes. The reaction line between Oxidation products of Fe (Fe203 and Fe304) and TiC passes through the tie plane connecting Fe3C to FeTiO3 and Ti02. Therefore, the predicied stable Oxidation products are Fe203 and the Oxides found at the vertices FeTi03 and Ti02 (and not T203 even if this one is the Oxidation product of TiC in the ternary diagrarn). Reaciionsfor sapphire in contaci with TiC in oxygen are interpretedby the calculated AI-Ti-C-0 quaternary phase diagram shown in Fig. 7. In this case, Al203 has tie-lines to TiC and to A12Ti05, which joins to the Oxides of TiC, narnely Ti02 and Ti203. These connections lirnit the possible Oxides to A1203, A12Ti05, and Ti02 and Ti203. The experimentally detected phase, Ti02 (rutile), is consistent with the calculated phase diagrarn; rutile is the most stable of the three natural Ti02 polytypes (the others being anatase and brookite) at roorn temperature. It is not the direct Oxidation product of TiC (Ti203). However, therrnochemical calculations Show that Ti203 is less stable than Ti02 at T=298K by 160 kcal/mole 02, indicating that the transfer film oxidized to cornpletion. The other phase that might have been expected, A12Ti05, was not Seen in these low Speed tests, but it rnight be generated during higher Speed or longer duration tests. 0

A U

a:FeC03 b:FeTi03

Fe

Feli

Fig. 6: Calculated isotherrnalquaternary phase diagrarn for Fe- Ti-C-0.

Fig.7: Calculated isothermal quatemary phase diagramfor AI-Ti-C-0.

291

5.GENERALIZATION TO OTHER SURFACE TREATMENTS 5.1 MoS2 coatings (16,17,26) Surface analyticalstudies and friction and wear tests against MoS2-coatedsteels are reviewed briefly. Sputtered MoS2 coatings have surface Oxide filrns and contain significant concentrations of carbon and oxygen (up to 30%). This section presents evidence for the three stages of wear depicted in Fig. 4 and describes how phase diagrarns can be used to account for the wear products. In four-ball tests, durability of the 1 p rn thick MoS2 film was quite high: roughly 100,000 contacts against WC:Co in dry Ar; that translates to an average wear rate of about 100 contact per nrn bst, or a wear rate of 1 nm/lOO contacts. In dry air the durabilities were a factor of 4 less: about 25,000 contacts against steel and WC:Co. Debris stripped frorn the contact Zone of the two different riiers, after 1 and 10 rnin tests, were exarnined by TEM. MoS2 was the rnain phase detected in dry Ar tests. In air tests, in addition to MoS2, Mo03 and several ternary Oxides of the rider material (iron frorn steel and cobalt frorn WC:Co) were found. Moreover, two of the phases that had hexagonal structures, MoS2 and COMOO3, showed basal planes parallel to the sliding direction -- a texture conducive to easy shear. The quaternary phase diagrarn for the Mo-S-Fe-0 systern, shown in Fig. 8, gives a good account of the phases detected. The iwo iron- molybdenurn Oxides, FeMo04 and Fe2Mo04, are stable reaction products of Fe-Mo cornpounds and Mo03 is the most stable Oxide phase of Mo. We cannot yet explain why the other two iron- molybdenurn Oxides were not detected. 0

Fe,MOZ

Fig.8: Sirnplified phase diagrarn of Mo-S-Fe-0.

Mo

Auger sputter depth Profiles of (stripped) contact spots and debris flakes on steel riders provide additional evidence for the reactions depicted by the quaternary phase diagrarn. In general all of the Profiles gave "cornpositions" that could be located as elernental rnixtures within the volurne bounded by the shaded area and the Fe-MoS2 tie-line. The intetfacial"bonding" film was an oxidized MoS2 film, sornewhat suifur rich at the surface, mergingwith a thicker oxidized steel surface. The depth Profile of a debris particle attachedto the contact Spot appears to have been a layered rnixtureof Fe Oxide and (Fe,Mo) Oxide containing S. It is also possible that the ternary Fe cornpounds could have been forrned by the rider breaching the MoS2 layer and rubbing underlying Fe. However, the presence of COMOO3 phase in debris of MoS2-coatedsteel vs WC:Co is evidence that the rider reacted with the MoS2. Finally, two distinct morphologies were observed in the TEM. MoS2 and the COMOO3, stripped from the Center of the contact Zone were uniforrnly thin over distances several rnicrornetersacross; we call this morphology "thin flakes." (Both also showed a preferredorientation of basal planes parallel to the sliding surface.) In contrast, most Oxide debris, and occasionally loose powdery MoS2 debris, were irregularly shaped and nonuniformly thick, a rnorphologywe call "spherical clusters." 5.2 TiN coatings (15) Sliding tests with steel balls against polished TiN coatings gave friction coefficients increasingfrorn 0.2 to 0.6 in proportion to the amount of debris transferred back to the coating. Friction coefficients with sapphire balls started sornewhat lower and usually remained bw, about 0.1. Auger and XPS analysis of polishedTiN revealed a thin surface Oxide sirnilar to the one detected on TiC surfaces. This surface Oxide is consistent with the XPS results of Emsberger et al(27) who showed that TiN develops a thin, therrnally stable Oxide layer about 1 nrn thick. Auger sputter depth Profiles of tranfer film and debris particle achieved after friction tests indicated that the thicker areas were Oxide films containingTi and Fe in roughly 3:l proportions and the thinner parts were thin titaniurn Oxide film on metallic iron. XPS analysis confirrnedthe transfer film contained no detectible TiN . TEM investigationsof debris flakes frorn both riders also find identical phases to those of TiC and sirnilar morphologies. These phases are consistent with the phases shown in the quaternary phase diagrarns representing steel (Fe) and sapphire (A1203) against TiN in air. Reactionsfor steel in contact with TiN in oxygen are interpreted on the calculated FeTi-N-0 quaternary phase diagram shown in Fig. 9. If the Oxidation products of Fe (Fe304 and Fe203) and TiN (Ti02) were connected by a reaction line (not shown), it would pass through the tie-plane

298

(Ti++C+) or Ti+ into steel reduces the friction coeff icient from 0.6 to 0.2-0.3, about that of a TiC coating. TEM revealed that both steel and sapphire riders generated debris with the FeTi03/ -Fe203 phase. This phase is consistent with that Seen in the quaternary phase diagram for Fe-Ti-C-0, shown in Fig. 6. Williamson et al(31) have recently identified the surface film on oxidized Ti-implanted steel to be ilmenite (FeTi03). By contrast, the debris generated against steel were predominantly spherical Clusters of Fe304.

connecting Fe2N to FeTi03 to Ti02. Therefore, the predicted stable Oxidation products are Fe203 and the Oxides found at the vertices, FeTi03 and Ti02, the phases found in the wear debris. 0

6. DlSCUSSlON

Fe

FfJli

Ti

Fig.9: Calculated isothermal quaternary phase diagram for Fe-Ti-N-0. Reactionsfor sapphire in contact with TiN in oxygen are interpretedby the calculated AI-Ti-N-0 quaternary phase diagram which has the Same Al-Ti0 ternary and internaltie-line from Al203 to TiN as the AI-Ti-C-0diagram. This case is even simpler since Ti02 is the direct Oxidation product of TiN. Al203 has tie-lines to TiN and to A12Ti05, which joins to the Oxides of TiN. These connections limit the possible Oxides as previously to A1203, A12Ti05, and Ti02.

5.3Ti+ - and (Ti+ + C+)-implantedsteel(28,99) Sliding tests with steel balls against polished (Ti++C+)- implanted steel gave friction coeff icients virtually identicalto those of polished TiC coatings ( 0.2, depending on debris transfer). Test with sapphire balls gave friction coefficients about 20% lower, but Same debris behavior. Friction coeff icients with Ti+-implanted steel showed somewhat higher friction coefficients (about = 0.3) and somewhat more debris formation. Steel vs steel gave =0.6 at all times. High-fluenceTi+ implantationdramatically alters the surface chemistry and structure of steel (28). The surface Oxide consists of an outer iron Oxide on top of an inner titanium Oxide. The Oxide grows on a metallic surface consistingof an amorphous irontitanium with carbon which diffuses into the Substrate during implantation. lmplanting C+ on top of Ti+ increasesthe subsurface carbon concentration and hardens the surface by precipitatingTiC particles(29,30). lmplanting

6.1 Chemical aspects of friction and wear. Study of the tribological behavior of TiC coatings Shows that the wear process develops in several Stages: - a thin Oxide surface film is removedfrom the coating - the transferfilm thickens and chemical reactions may occur - debris particlesgrow and detach from the rider. This three step-modelis common to all low friction, low wear surfaces, i.e. TiC, MoS2, TiN, Ti implanted surfaces. The therrnochemical reactions responsible for the compositionof debris particles and of interfacial bonding layers are described using ternary and quaternary diagrams.These reactionscan provide insight into the mechanismsof wear. If wear occurs by the removal of a thin layer of material, then the interface between the layer and the Substrate may be an easily sheared plane. MoS2 is known to have easy shear planes located along the S-S bonds oriented parallel to the basal planes. Less well understood is the mechanism of easy shear of thin Oxide films on TiN/TiC coatingsor (Ti++C+)-implanted steels. One possibility is that the Oxide film growth is diffusion limited, and the film that forms-albeit thin-attains one stable phase. If this were so, then the phase (e.g. Ti02 on TiN) would likely have a tie-line with the Substrate, indicating minimal driving force for adhesion. In (Ti++C+)-implantedsteel, the TiC-rich interface may act as a diifusion barrier to hinder interdiifusion of Fe and 0 between the surface film and the Substrate, thereby allowing the surface film to oxidize rapidly to completion (as it may on TiN and TiC). Another possibility is that the phases themselves act as low frictionfilms. Ti02, forexample, is softer that TiN or TiC, and a soft, sacrificial layer between hard counterfaceswould produce lower friction. Lally et al(31) suggested that the interface separating hematite particles that precipitated out of an ilmenite matrix wouM have a very low energy, making them more easily sheared. By contrast, Fe does not produce a thin, stable Oxide layer, and its Oxide is as hard if not harder than

299

Fig. 10: Schernatic of the wear process the Fe itself. Chernicalreactions can also account for the low wear rates involved. Thin, sacrificialOxide layers could reducethe wear rate of materials to less than a monolayer per cycle. If a monolayer of Oxide film (roughly 0.3 nrn thick) were lost on each cycle, then a TiN Substrate would loose less than 0.1 nrn of material, since the atornic volurne of the Ti02 molecule is 1.6 times that of TiN and oxygen rnakes up 2/3 of Ti02. Chernicalreactions also assist in the bonding of transfer layers to riders . If a transier layer has the right chernistry and structure, as alluded above, it can provide a more slippery interface than the original sliding surfaces. Chernical reactions can also be used to distinguish low friction wear processes frorn high friction ones. We postulatedthat low friction is associated with easy shear of thin (norninallya monolayer thick) transfer layers along a welldefined interface. Wear processes that separate thin Oxide layers along a well-defined shear plane should produce stoichiometric Oxides and flat, uniformly (thin) flakes. This morphology is characteristicof debris frorn low friction sliding in all cases given here. By contrast, wear processesthat separate layers along a less well-defined interface(e.g. not an easy shear plane) should give higher friction coefficients and produce debris having very different material characteristics. If, for exarnple, shear ocwrs below the metalioxide interface, the metallic portion of the debris would not likely oxidize to completion, resulting in sub-Oxides. However, Oxidation of the metallic portion would Set up Stresses that would cause the debris to curl. The latter cornposition and morphology are characteristic of Fe3Mebris frorn high friction ( = 0.6) sliding of steel against steel. The morphologies and cornpositions of transfer filrns removed frorn the low and high friction surfaces are illustrated schernatically in Fig. 9.

of low and high friction surfaces 6.2 Thermodynarnicalaspects of friction and wear. Equilibriurn phase diagrarns have been used successfully to account for the cornpositions of debris generated during the wear of low wear, low friction surfaces. Nonetheless, a large body of literature demonstrates that equilibriurn therrnodynarnicsdoes not apply to rnany tribochernical(i.e. rnechanically-induced) reactions(7). Furthermore, recent investigations into rnechanical mixingof metals suggest that metastable phases (e.g. amorphous intermetallics)are energetically preferred over highly defective crystalline phases, at lower ternperatures where kinetics of rnixing rnay be favorable and defect annihilation is slow (33,34). Why, then, should equilibriurn thermodynarnics hold for debris generated during sliding contacts? Here we address one aspect of this issue: the kinetics responsible for phase forrnation at low sliding Speeds and at low ternperatures. Diffusionof both rnetal and oxygen atoms in transfer layers will be enhanced by the high defect concentrations produced by deforrnation aWor fracture accornpanyingthe removal of a surface film. Moreover, since the layers are so thin (often 10 nrn or less), they can react to cornpletion (e.g. Fe and Ti interdiffusion and Oxidation) in very short times, even at low temperatures. Thus, we can rationalize equilibrationat roorn ternperature in the wear processes discussed here by defect-enhanced kinetics and short diffusion distances, without the need to invoke "hot spot" rnechanisrns. Fischer [9] elaborateson sirnilar rnechanisrns for tribochernical reactions in greater detail. Finally, the calculatedternary and quaternary diagrarns presented here have several lirnitations (14). But thermodynarnic rnodeling provides a scientific basis for designing future lubrication Systems. If reaction products can be predicted, alloys and surface treatrnents can be tailored to produce beneficial transfer filrns (e.g. phases that

300

are softer than the Substrate), and benign wear debris (e.9. phases that are seif lubricatinglike basaloriented hexagonal platelets). 7. CONCLUSIONS Thermochernicalanalysis has been shown to be a powerful rnethodfor describing chernical reactions that accornpany the wear of low wear, low frictiin suriaces. Ternary and quaternary diagrarns, in most instancescakulated frorn Standard thermochernical data, have accounted for the phases of debris and agree qualitativelywith the Oxidation and alloying processes leading to debris formation. Finally, the themochernical model provides a mrnrnon understanding of the wear of rnechanically different rnaterials. 8. ACKNOWLEDGMENTS The authors thank P. Ehni, for his suriace analytical contributions, M.C. Einloth for programrning the ternary diagrarn calculations, A. Bhansalifor his quaternary tie-line prograrn, and L. Seitzrnan for valuable cornrnents. This work was rnade possible by funds frorn DARPA, SDlO and ONR.

REFERENCES 1. Ling F.F. Surface Mechanics (Wiley Interscience, New York, 1973). 2. Suh N.P. Tribophysics (Prentice-Hall,New Jersey, 1986). 3 Orcutt F.K., Krause H.H., and Allen C.M. Wear,S (1962) 345. 4. Sliney H.E., Strom T.N., and Allen G.P., ASLE Trans., & (1965) 307. 5. Johnson R.L. and Sliney H.E. Proc. AFML-MRI Conf. on Solid Lubricants (1970) p. 40. 6. Buckley D.H. Surface effects in Adhesion, Friction, Wear and Lubrication(Elsevier,Arnsterdarn, 1981) and references therein. 7. HeinickeG. Tribochernistry, (Carl Hanser Verlag, Munchen, 1984). 8. Bentley R.M. and Duquette D.J. in Fundarnentals of Frictionand Wear of Materials, edited by D.A. Rigney (ASM, Metals Park, OH, 1981) p. 291. 9. Fischer T.E. Ann. Rev. Mater. Sci., 111 (1988) 303. 10. Quinn T.F.J. Tribol. Int.,X(1983) 257 and 305. 11. Sexton M.D. and Fischer T.E. Wear, 96(1984) 17.

12. Krarner B.H. and Suh N.P. ASME Trans. J. Eng. Ind., m ( 1 9 7 9 ) 303. 13. GerkernaJ. Wearm(1985) 241. 14. Singer I.L. in Proc. Int. Conf. on Metallurgical Coatings, San Diego,(l991) to be published 15. Singer I.L. , Fayeulle S. and Ehni P.D. , in Wear of Materials 1991 ed.K. Luderna (ASME, NY, 1991), to be published. 16. Fayeulle S. ,Singer I.L. , and Ehni P.D. , in Mechanics of Coatings,Leeds-Lyon 16, Tribology Series, 17. edited by D. Dowson, C.M. Taylor, M. Godet (Elsevier, GB, 1990) p. 129. 17. Fayeulle S. , Ehni P.D. and Singer I.L., Surf. Coat. Technol.,fi(1990) 93. 18. Fayeulle S. and Singer I.L., Materials Sci. and Engineering A l 15 (1989) 285. 19. Kubaschewski 0. and Alm& C.B., MetallurgicalThermochernistry, 5th edition (Pergamon Press, Oxford, 1979) Table A. 20. Thurmond C.D. ,Schwartz G.P. ,Karnrnlott G.W. 1980) and Schwartz B. , J. Electrochem. SOC.,=( 1366. 21, Beyers R., J. Applied Physics (1984) 147; R. Beyers, R. Sinclair and M.E. Thornas, J. Vac. Sci. Technol. 82 (1984) 781. 22. Singer I.L. and Wandass J.H. , in StructureProperty Relationships in Surface-Modified Cerarnics C.J. McHargue (ed.), (Kluwer Acadernic Publishers, 1989) pp. 199-208. 23. Lin J.-C. , Hsieh K.-C.,Schulz K.J. and Chang Y.A. J.Mater. Res.,9(1988) 148. 24. BhansaliA.S. , Sinclair R. and Morgan A.E., J. Appl. Physics, 68 (1990)1043. 25. For exarnple, as found in Phase diagrarns for ceramists Vols. 1-5, (Am. Cerarnics SOC.,Colurnbus, Ohio, 1964 - 1981). 26. Ehni P.D. and Singer I.L., in New Materials Approaches to Tribology: Theory and Applications, edited by L. Pope, L. Fehrenbacher and W. Winer, MRS Symposium, 140 (MRS, Pgh. PA, 1989). 27. Emsberger C. , Nickerson J. , Srnith T.,Miller A.E. and Banks D., J.Vac. Sci. TechnolA (1986) 2784. 28. Singer I.L., Appl. Sud. Science 18 (1984) 28. 29. Singer I.L. and Jeffries R.A., Appl. Phys. Leii. G(1983) 925. 30. Oliver W.C., Hutchings R., Pethica J.B.,Singer I.L. and Hubler G.K., in Ion Implantationand Ion Beam Processingof Materials, edited by G.K.Hubler, C.W. White, O.W. Holland, C.R. Clayton, Elsevier, NY (1984) pp. 603-608. 31. Williarnson D., private cornmunciation (1990);to be cornpleted later. 32. Lally J.S. , Heuer A.H. and Nord G.L., Jr. in "Electronic Microscopy in Mineralogy (SpringerVerlag, Berlin, 1976) pp. 214-219. 33. Koch C.C. , Cavin O.B., McKarney C.G. and Scarbrough J.O., Appl. Phys. Lett., G(1983) 1017. 34. Schwarz R.B., Petrich R.R. and Saw C.K., J. Non-Cryst. Solids, =(1985) 281.