Surface and Coatings Technology 155(2002) 169–175

0257-8972/02/$ - see front matter� 2002 Elsevier Science B.V. All rights reserved.PII: S0257-8972Ž02.00045-2

High temperature fretting behaviour of plasma vapour deposition TiNcoatings

A. Ramalho *, J.-P. Celisa, b

Dep. Eng. Mecanica, Polo II, Universidade de Coimbra – FCTUC, Pinhal de Marrocos, P-3030 Coimbra, Portugala ˆ ´Dept. MTM, Katholieke Universiteit Leuven, Kasteelpark Arenberg 44, B-3001 Leuven, Belgiumb

Received 7 August 2001; accepted in revised form 1 February 2002

Abstract

Fretting tests(mode I) were performed on TiN coatings at test temperatures of 23–5008C. The evolution of the coefficient offriction with the number of fretting cycles for tests performed in that temperature range was recorded. Differences in that evolutionwere analysed based on the frictional energy dissipated in the sliding contact during the fretting tests. That analysis demonstratedthe role of dissipated frictional energy and thermal energy in initiating a structural modification of the debris. The structuralmodification of the debris coincides with the transition from high friction conditions taking place in the presence of amorphousdebris to low friction conditions in the presence of nano-crystalline debris.� 2002 Elsevier Science B.V. All rights reserved.

Keywords: Fretting; Coatings; High-temperature wear; Friction; Crystallization

1. Introduction

Fretting is a wear process induced by the reciprocatingmovement at small displacement amplitudes of a count-er-body, and takes place in many technical systems. Thecontact region remains to a large extent isolated fromthe surrounding atmosphere. The debris formed and thesurface reactions induced during fretting, play a majorrole in the fretting wear process.Nowadays it is generally accepted that tribological

contacts need to be analysed based on a tribo-systemapproach. The energy dissipated by friction is the maininput of energy in a tribo-system. Therefore all theenergy-consuming processes in sliding contacts, inclu-sive wear, are directly or indirectly dependent on thedissipated energy. This is the basis for expressing thewear rate as a volumetric wear per unit of frictionalenergy dissipated during wear tests. In fretting tests, acumulative dissipated energy is obtained by summingup the energy dissipated during the successive frettingcycles. The advantage of plotting the volumetric wearvs. the cumulative dissipated energy is that both are

*Corresponding author. Tel.:q351-239-790700; fax:q351-239-790701.

E-mail address: amilcar.ramalho@dem.uc.pt(A. Ramalho).

cumulative phenomena. As a result, fretting tests per-formed under different sets of test conditions, includingthe duration of the test, can be compared as long as themain wear mechanism remains unchangedw1x.That approach has been successfully used for the

laboratory investigation of wear-resistant TiN coatingswidely used on cutting and forming toolsw2–7x. Theeffect of relative humidity, normal load, and counter-face material, has been extensively studied by Mohr-bacher et al.w4,5x. These studies demonstrated thatrelative humidity plays an important role on the oxida-tional wear of TiN in non-lubricated fretting tests.Recently, de Wit et al.w7x studied the mechanismassociated with the changes in friction observed duringnon-lubricated fretting tests performed with TiN-coatedsteel sliding against corundum balls. An investigationof the material transfer to the corundum counter-bodyrevealed that for fretting tests performed at room tem-perature, both amorphous and nano-crystalline debrisare detected. The appearance of amorphous andyornano-crystalline debris depends on the contact load andthe environmental conditions, and also on the testingtime at which these debris are formed. The structure ofthe debris determines also whether a high or a lowcoefficient of friction is noticedw7x. The structure of the

170 A. Ramalho, J.-P.-P. Celis / Surface and Coatings Technology 155 (2002) 169–175

Fig. 1. Schematic diagram of the high temperature fretting test appa-ratus. 1, Normal-load actuator; 2, load-cell; 3, pivot; 4, quartz force-transducers; 5, upper specimen holder; 6, lower specimen holder; 7,oven; 8, motorised table.

debris evolves in tests performed at room temperaturewith the dissipated frictional energy, and this dissipatedfrictional energy is affected by environmental conditions,mainly the relative humidityw7x.The dissipated frictional energy is thus a key param-

eter determining the fretting behaviour of coatings likeplasma vapour deposition(PVD) TiN. As a conse-quence, the test temperature ought to be an importantparameter in fretting. Nevertheless, to our best knowl-edge, such an effect of temperature on the frettingbehaviour of TiN PVD coatings has not yet beenreported. In this paper, the fretting behaviour of TiNhigh-speed steel coatings sliding against corundum attemperatures up to 5008C is presented and discussed.

2. Experimental

High-speed steel ASP23 flat specimens quenched andtempered at a hardness of 63 HRC were polished to aroughnessR 0.02 mm and subsequently PVD-coateda

with TiN. The 4-mm-thick coatings were deposited usinga Balzers triode ion-plating equipment(WTCM, Die-penbeek, Belgium). The roughness of the coating was0.06mm (R ). Polished corundum balls with a diametera

of 10 mm, a hardness of 2000 HVN, and a surfaceroughnessR 0.2 mm, were used as a counter-body.a

Fretting tests mode I were carried out in a high temper-ature fretting test rig as outlined in Fig. 1 and describedin detail in ref. w8x. The normal load, applied by anelectro-magnetic linear actuator(1), was 10 N andremained constant during the tests. A triangular shapewave with a peak-to-peak displacement amplitude of

100 mm, and a frequency of 2 Hz was applied to thelower flat specimens by a linear motorised table guidedby air bearings(8). A linear encoder with a resolutionof 0.1 mm was used to assure feedback to the linearACyDC motor. Quartz force transducers(4) were linkeddirectly to the upper stationary ball specimen in orderto measure the tangential force. All the tests were donein a background laboratory ambient air of 238C and50% relative humidity. A specially designed oven(7)using MoSi heating elements surrounded the fretting2

contact area. Ceramic specimen holders(5, 6) havebeen previewed to withstand temperatures up to 12008C. In this work, the temperature was varied between60 8C and 5008C, controlled by a thermocouple mount-ed on the TiN-coated specimen. This temperature isreferred hereafter as the ‘test temperature’, and is notthe real contact temperature at the sliding fretting contactarea.The tangential force and the displacement amplitude

were measured on-line during the fretting tests at anacquisition rate of minimum 500 pointsycycle. Thenumber of cycles was selected in such a way that wear-through of the coating did not occur during the frettingtests performed in the temperature range mentionedabove.Prior to the fretting tests, the specimen and the

counter-body were cleaned with acetone and alcohol.After the fretting tests, the morphology of the weartracks was observed by scanning electron microscopy(SEM). Wear scars were investigated by laser profilo-metry (Rodenstock RM600), and the wear volume wasobtained by integrating a cross-sectional profile areas asdescribed earlierw9x.The tangential force was measured on-line during the

fretting tests. The energy dissipated during the succes-sive fretting loops was calculated based on the instan-taneous values of the tangential force and displacementaccording to a procedure published earlierw5x. Thiscumulative dissipated energy was calculated for thedifferent fretting tests performed.

3. Results

The evolution of the tangential force during thesuccessive fretting cycles is shown as a fretting log inFig. 2 for the case of a fretting test done at 1008C.This fretting log represents hysteresis loops obtained onplotting the tangential force vs. displacement at discretenumbers of fretting cycles. The shape of the hysteresisloops does not change during the fretting tests and theloops are almost parallelograms, characteristic for fret-ting tests performed under gross slip. Starting as smallloops, the width of these loops increases progressivelyup to a maximum value, and then decreases suddenly atapproximately 1000 cycles for the case shown in Fig.2. Hysteresis loops corresponding to cycle 1000 are

171A. Ramalho, J.-P.-P. Celis / Surface and Coatings Technology 155 (2002) 169–175

Fig. 2. Fretting log for the case of a fretting test done in ambient airat a temperature of 1008C. The normal load was 10 N and the dis-placement amplitude was 100mm.

Fig. 3. Hysteresis loops corresponding to cycle 1000 for fretting testsperformed at test temperatures varying between 608C and 5008C(normal load 10 N, displacement amplitude 100mm).

Fig. 4. Evolution of the mean coefficient of friction during the testsperformed at temperatures in the range of 608C to 5008C (normalload 10 N, displacement amplitude 100mm).

shown in Fig. 3 for fretting tests performed at testtemperatures between 608C and 5008C. The coefficientof friction was derived from these hysteresis loops asthe ratio between the mean value of the tangential forceand the normal load applied. Since the variation of thetangential force is small in the tests performed duringthis study, the mean value of the coefficient of frictionis used. That mean value was calculated from thehysteresis loop area since all tests were performed undergross slip conditions, and since the displacement con-troller is very reliable. The evolution of the meancoefficient of friction during the fretting tests is shownin Fig. 4 for tests performed at temperatures between60 8C and 5008C. The use of a logarithmic scale forthe number of cycles allows a better observation of thevariation of the coefficient of friction during the running-in phase. The evolution of the coefficient of frictionwith the number of cycles agrees well with previouswork w4x, and is characterised by five periods. The firstone is a running-in period with a low and constantcoefficient of friction, followed by a second periodwhere a rise of the coefficient of friction takes place.During the third period, the coefficient of friction reach-es high values and remains more or less constant.Finally, a rapid drop of the coefficient of friction takesplace and a steady state period is established till wear-through occurs. Besides this general evolution appearingin all tests performed, some major differences betweenthe tests were noticed. So, e.g. the coefficient of frictionincreases in the running-in period with increasing testtemperature. The maximum value of the coefficient offriction in the third period is 0.7 for fretting tests doneat 60 8C, while a value near 0.5 is noticed at all the

other test temperatures. Finally, the number of frettingcycles at which the fourth period starts, decreases withincreasing test temperature. Only the fretting test doneat 5008C is an exception to this trend.The fretting tests result in elliptical wear scars on TiN

coated HSS steel samples. Fig. 5a,b shows such wearscars measured on TiN by contactless laser profilometryafter fretting tests performed under gross slip conditionsagainst a corundum ball for 5000 cycles at a testtemperature of 608C. In the rather exceptional case thatwear-through did occur within the number of frettingcycles selected(Fig. 5b), a large increase of the weardepth is noticed as pointed by an arrow in Fig. 5b. Thewear loss was calculated from topographical analysessuch as the ones shown in Fig. 5. The wear loss on TiN

172 A. Ramalho, J.-P.-P. Celis / Surface and Coatings Technology 155 (2002) 169–175

Fig. 5. Three-dimensional topography profile representation of the TiNfretting scar. Fretting test done at 1008C for 5000 cycles at a normalload of 10 N and a lateral displacement of 100mm. (a) Profilometryin the case wear-through did not take place.(b) Profilometery in theexceptional case of wear-through.

Fig. 7. Morphology of TiN fretting scar observed in SEM. Frettingtest done at 1008C for 5000 cycles at a normal load of 10 N and adisplacement amplitude of 100mm.

Fig. 6. Wear volume of TiN coatings as a function of the cumulativedissipated energy. Dashed line represents the results obtained at 238Cby Mohrbacher et al.w6x.

Fig. 8. Wear scars of test temperature of 1008C. (a) Three-dimen-sional profile of a wear scar on the flat TiN surface,(b) SEM of thecorresponding wear track on the corundum ball.

coatings tested at different test temperatures is plottedin Fig. 6 against the cumulative dissipated energy. InFig. 6, data published by Mohrbacher et al.w5x on asimilar material combination but tested at 238C, are

represented as a dotted line. Mohrbacher found a linearrelationship between the volumetric wear loss and thecumulative dissipated energy for fretting tests performedunder gross slip conditions in ambient air of 238C, andat different relative humidity, normal loads, and frequen-cies. The test done presently at 608C fits well with thedata of Mohrbacher, notwithstanding the fact that acompletely different fretting test equipment was usedwith a different rigidity and frame structure. Even more

173A. Ramalho, J.-P.-P. Celis / Surface and Coatings Technology 155 (2002) 169–175

Fig. 9. Outline of the friction evolution to identify the points used inthe energetic analysis. The three curves shown are taken from Fig. 4and correspond to fretting tests done at 60, 200 and 4008C,respectively.

interesting is to note that for the experimental dataobtained at test temperatures between 100 and 5008Cin ambient air, a linear dependence is also noticedbetween dissipated energy and volumetric wear loss.The best linear fit for these high temperature frettingtests, has a slope comparable to the one reported byMohrbacher, but the line is shifted so that a slightlyhigher wear loss is taking place at a given amount ofdissipated energy.Typical fretting wear scars morphologies are shown

in Figs. 7 and 8. Fig. 7 shows the wear scar on the TiNcoated surface after a fretting test done at 1008C for5000 cycles. Fig. 8a shows the profilometry of that wearscar on the TiN-coated surface, and Fig. 8b the mor-phology of the corresponding wear scar on the corundumcounter-body. The morphology of the wear scar on theTiN coated surface is characterised by a central contactarea exhibiting fine grooves and showing traces of somepolishing action by the debris, as well as accumulatedamounts of debris located at the periphery of the contactzone. These debris are quite small. On the corundumcounter-body, a transfer layer adheres well in the contactarea. EDX-analyses revealed the presence of Ti and Oin this transferred material. On comparing Fig. 8a,b, itappears that the transfer layer sticking to defined areasof the ball leads to a reduction of the wear of thecorresponding TiN counterparts. In fact, the central partof the ball contact area, referred as 1 in Fig. 8b, has acorresponding area on the counter-face(Fig. 8a) exhib-iting a low wear compared with the surrounding regions2 and 3 without any evidence of transferred material onthe ball scar.

4. Discussion

The shift in the fretting wear data obtained in attemperatures of 1008C up to 400 8C in comparisonwith the fretting data obtained at 238C by Mohrbacheret al. (see Fig. 6), indicates that the wear degradationprocess is influenced by the test temperature in thatrange. The linear relationship between the volumetricwear and dissipated energy for fretting tests performedeither at low(range of 238C up to 608C) or at hightemperature(range of 1008C up to 4008C), indicatesthat the test temperature does not basically affect thewear mechanism. That lateral displacement of the curveindicates that the achievement of a given volumetricwear loss requires lesser dissipated frictional energy infretting tests done at high temperature than in frettingtests done at low temperature. Having a closer look onthe evolution of the coefficient of friction during frettingtests, we notice that the coefficient of friction remainslow at the start of the fretting tests performed either atlow or high temperature. The general evolution of the

coefficient of friction with the number of fretting cyclesis similar for all fretting tests done. The coefficient offriction progressively increases until a maximum valueis reached, followed by a more steep decrease. However,two major differences are noticed between tests per-formed at different test temperatures(see Fig. 9). Thefirst difference is that the coefficient of friction at thestart of the fretting tests increases with increasing testtemperature. Secondly, the number of fretting cyclesnecessary to achieve the drop in the coefficient offriction once the maximum value is reached(see pointsB), decreases with increasing test temperature.To get a better insight into these phenomena, two

typical amounts of frictional energy dissipated in thefretting contacts were calculated. The first one is usedto explain the increase of the coefficient of friction atthe start of the fretting tests with test temperature. Heretothe fretting test performed at 608C was taken as areference. The frictional energy dissipated in that frettingtest was calculated up to the time that a coefficient offriction is reached corresponding to the coefficient offriction at the start of a fretting test performed at ahigher temperature. It was obtained by integrating thefretting hysteresis loops in the reference test up to thecycle corresponding to point A in Fig. 9. That energycan be considered as the equivalent energy in thereference test required to create in the contact area, amaterial surface condition similar to the one on the TiNsample at the start of the fretting test at the correspond-ing higher test temperature. That equivalent energy isplotted in Fig. 10 as a function of the temperaturedifference with the reference fretting test at 608C. Alinear relationship is found for fretting tests performedin the range of 200–5008C. This indicates that the

174 A. Ramalho, J.-P.-P. Celis / Surface and Coatings Technology 155 (2002) 169–175

Fig. 10. Effect of the test temperature on the equivalent energy nec-essary to reach the starting value of coefficient of friction at 608C(see point A in Fig. 9) for fretting tests performed at 10 N and 100mm displacement amplitude.

Fig. 11. Cumulative dissipated energy up to the start of the drop ofthe coefficient of friction,(see points B in Fig. 9) for fretting testsperformed at 10 N and 100mm displacement amplitude.

surface modification induced on TiN during a frettingtest, is equivalent to a thermal oxidation requiring athreshold contact temperature of approximately 1608C(DT of 100 8C above the temperature of the referencetest).The second one is the frictional energy dissipated in

the fretting contact up to the time that the drop in thecoefficient of friction takes place(see points B in Fig.9). This dissipated frictional energy was calculated forfretting tests performed at test temperatures between 60and 4008C (Fig. 11). That dissipated frictional energyinduces the transformation of amorphous debris formedin the sliding contact during the first three periods ofthe fretting tests(see Fig. 4), into nano-crystalline ones.Such a transformation has also be reported by de Wit etal. w7x. That dissipated frictional energy decreases line-arly with increasing temperature(Fig. 11). The input ofthermal energy takes thus up a part of the energyrequired for initiating the crystallisation of the amor-phous titania debris. It has to be noticed that the resultsobtained at a test temperature of 5008C, do not fit withthe data obtained at lower test temperatures. That canbe explained by looking on previous work on TiNw7,10xthat showed that the evolution of the coefficient offriction depends on the formation of a transfer layer onthe surface of the counter-body. The formation of sucha transfer layer consisting of titanium and oxygen, wasdescribed by Singerw11x as originating from debrisresulting from the degradation of the TiN coating. Theachievement of a constant coefficient of friction dependsthen on the evolution in the structure of that transferlayer. Both de Wit et al.w7x and Hsu et al.w10x foundout that the heating up of TiN in an oven at 5008C,induces the crystallisation of TiN into Ti-O. In view of

these facts, the results of the fretting test shown in Figs.4 and 11 can be explained as follows:

– during fretting tests done at 608C, the crystallisationof the debris is induced by the frictional energydissipated in the sliding contacts. Initiation of crys-tallisation requires a large number of fretting cycles,and the coefficient of friction remains high during alarge number of fretting cycles;

– during fretting tests performed between 1008C and400 8C, the dissipated frictional energy required toinduce crystallisation decreases since thermal oxida-tion becomes more active, and the coefficient offriction lowers after a smaller number of frettingcycles;

– for fretting tests performed at 5008C and higher, thefrictional behaviour is different right from the start ofthe fretting tests. This is linked to a spontaneousthermal oxidation of TiN on exposure to ambient airat these temperatures as reported previouslyw7,10x,so that the crystallisation of the debris does notrequire any additional frictional energy.

From these experimental facts, it may be concludedthat the test temperature during fretting tests acts as anadditional input of energy contributing to the crystallis-ation of the debris. At increasing test temperature, thefrictional energy required to induce crystallisation low-ers. This conclusion agrees well with the oxidationmechanism put forward by Mohrbacher et al.w5x toexplain the fretting wear of TiN coatings. The testtemperature thus appears as an additional input of energyfor crystallisation of the debris, proportional to the testtemperature increase.

175A. Ramalho, J.-P.-P. Celis / Surface and Coatings Technology 155 (2002) 169–175

5. Conclusions

The effect of temperature on the fretting behaviour ofPVD TiN coatings sliding against corundum was inves-tigated up to 5008C. The experimental results could beexplained based on a structural transformation fromamorphous to nano-crystalline debris originating fromthe TiN coating that takes place at a certain momentduring fretting tests. This transformation of debris gov-erns the evolution from a high to a low coefficient offriction, and determines the extent of the fretting wearon TiN. The energy needed to activate that structuraltransformation of the debris, is shown to originate notonly from the frictional energy dissipated in the slidingcontacts during the fretting tests, but also from theexternal heat input when fretting tests are performed attemperatures above room temperature. At increasing testtemperature in the range of 608C up to 4008C, theexternal heat reduces accordingly the frictional energyneeded to get low friction nano-crystalline debris onTiN. For fretting tests performed at temperatures above500 8C, the oxidation of TiN on exposure to the ambientair is sufficient to create an oxide layer on top of TiN.

Acknowledgments

A.R. thanks the Portuguese Foundation for Scienceand Technology for granting a sabbatical bursary. Partof this work was done within a project funded by theBelgian Government(contract IUAP P4y33) and aproject funded by the CEC(contract BRITEyEURAM3EUPIII BRPR-CT97-0380). The technical assistance ofing. Marc Peeters in setting up the high temperaturefretting equipment and the fretting test series is mostappreciated. The authors acknowledge the help of ir. M.

Van Stappen(WTCM-Diepenbeek, Belgium) in provid-ing TiN-coated samples.

References

w1x J.P. Celis, L. Stals, E. Vancoille, H. Mohrbacher, Wear testingof hard coatings: more than wear rate?, Surface Eng. 14(1998)232–239.

w2x M. Shima, J. Okado, I.R. McColl, R.B. Waterhouse, T. Hase-gawa, M. Kasaya, Influence of substrate material and hardnesson the fretting behavior of TiN, Wear 225–229(1999) 38–45.

w3x G. Xu, Z. Zhou, J. Liu, X. Ma, An investigation of frettingbehavior of ion-plated TiN, magnetron-sputtered MoS and2

their composite coatings, Wear 225–229(1999) 46–52.w4x H. Mohrbacher, B. Blanpain, J.-P. Celis, J.R. Roos, The

influence of humidity on the fretting behavior of PVD TiNcoatings, Wear 180(1995) 43–52.

w5x H. Mohrbacher, B. Blanpain, J.-P. Celis, J.R. Roos, L. Stals,M. Van Stappen, Oxidational wear of TiN coatings on toolsteel and nitrided tool steel in unlubricated fretting, Wear 188(1995) 130–137.

w6x P.-Q. Wu, H. Chen, M. Van Stappen, L. Stals, J.-P. Celis,Comparison of fretting wear of uncoated and PVD TiN coatedhigh-speed steel under different testing conditions, Surf. Coat.Technol. 127(2000) 114–119.

w7x E. de Wit, L. Froyen, J.-P. Celis, The crystallization ofamorphous debris on titanium nitride coatings influenced bysliding wear conditions, Wear 221(1998) 124–133.

w8x A. Ramalho, J.-P. Celis, Fretting laboratory tests: the role ofthe mechanical response of the test rigs, Tribol. Lett.,submitted.

w9x B. Blanpain, J.-P. Celis, J.R. Roos, J. Ebberink, J. Smeets, Acomparative study of the fretting wear of hard carbon coatings,Thin Solid Films 223(1993) 65–71.

w10x L.S. Hsu, R. Rujkorakarn, J.R. Sites, C.Y. She, Thermallyinduced crystallization of amorphous titania films, J. Appl.Phys. 59(10) (1986) 3475–3480.

w11x I.L. Singer, A thermochemical model for analysing low wear-rate materials, Surf. Coat. Technol. 49(1991) 474–481.