At

PP

a

ARRAA

KASSTO

1

atifcp

aa[ppF

0d

Wear 272 (2011) 122– 132

Contents lists available at ScienceDirect

Wear

j o ur nal ho me p age: www.elsev ier .com/ locate /wear

n investigation of the wear mechanism leading to self-replenishingransfer films

atrick S.M. Dougherty, Randyka Pudjoprawoto, C. Fred Higgs III ∗

article Flow and Tribology Laboratory, Carnegie Mellon University, Pittsburgh, PA, USA

r t i c l e i n f o

rticle history:eceived 26 November 2010eceived in revised form 22 July 2011ccepted 1 August 2011vailable online 5 August 2011

eywords:brasive wearolid powder lubricationelf-replenishingransfer filmptical interferometer

a b s t r a c t

Since the late 1980s, interest has risen in solid powder lubrication due to its proven ability to providelow friction and wear in interfaces unsuitable for traditional oils. This may be in the form of augment-ing oil performance as an additive, or in the form of thin, solid transfer films since it was found thatsliding materials sometimes inherently generate a film that can protect the contact interface during rel-ative motion. In particular, in situ self- replenishing solid lubrication has shown the ability to maintainlubricious conditions through the continual deposition of thin powder transfer films to “dry” surfaceasperities from a sacrificial compact. An emerging class of self-lubricating compacts is being developedby compacting lamellar powder particles into different homogeneous and heterogeneous solid lubricantstructures or “pellets.” An in situ self-replenishing solid lubricant system may be created by placing thesepellets into tribosystems in such a way that their transferred film is continuously applied to dry asperi-ties in load-bearing interfaces. Therefore, special emphasis must be given to understanding the methodof transfer film deposition and depletion, if the lubrication process is to be predicted, controlled, andoptimized. The purpose of this investigation is to examine the wear mechanisms which govern transferfilm deposition and depletion in an in situ self-replenishing system, such that a more accurate modeling

approach may be undertaken in the future. Surface analysis of contact interfaces was performed usingan optical interferometer, while friction and wear relationships were gleaned from experiments on apellet-on-disk with slider pad tribometer. Through an analysis of numerous qualitative and quantitativeparameters that describe surface topography, it was found that abrasive wear is the predominant wearmechanism governing the transfer film process. Consequently, an alternate wear description of the in situself-replenishing transfer film lubrication process is proposed.

. Introduction

While oil-based lubrication remains one of the most prolificreas of lubrication study, continued advancements in mechanicalechnologies continues to push interfacial temperatures for bear-ng related contacts beyond the capabilities of liquid lubrication. Inact, at temperatures above approximately 500 ◦C, oil based lubri-ants can actually dissociate and lose their ability to carry load orrovide ample lubrication [1].

Lamellar powders or solid lubricants have become a promisinglternative as thick film powder flows [2–4], transfer films [1,5] andlso as multifunctional additives to existing oil based lubricants6]. Despite extreme temperatures and pressures, these lamellar

owders have shown environmentally friendly potential while alsoroviding extremely low friction coefficients as low as 0.02 [7,8].urthermore, the emergence of thin transfer films applied through

∗ Corresponding author. Tel.: +1 41 2268 2486; fax: +1 41 2268 3348.E-mail address: higgs@andrew.cmu.edu (C.F. Higgs III).

043-1648/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.wear.2011.08.002

© 2011 Elsevier B.V. All rights reserved.

“pelletized” lamellar powders have proven beneficial for in situsolid lubrication [5] and special multifunctional applications suchas sliding electrical contacts [10]. However, the mechanics involvedin powder lubrication, while thought to express a mixture of bothhydrodynamic and morphological effects, are still largely misun-derstood [3,4,9,10,12].

This current work aims to elucidate the precise wear mech-anisms which produce and remove the transfer film when acompacted powder pellet under load is slid against a hard tungstencarbide disk in the presence of a loaded slider pad. In the past, it wasassumed that this was primarily governed by adhesive wear occur-ring at the pellet-on-disk and slider-on-disk interfaces. However,this paper conducts a detailed investigation of the tribosurfacesbefore and after wear testing, in which optical profilometry tech-niques are used to identify the most probable wear mechanisms.Furthermore this study also sheds light on the nature of the trans-

fer film coverage. Since the friction coefficient is assumed to varywith the fraction of the asperity domain covered by solid lubricant[5,17], there have been questions regarding whether the transferfilm coverage occurs at varying heights with uniform area, or at

P.S.M. Dougherty et al. / Wear 272 (2011) 122– 132 123

self-

vpoceitdu

2

cicdlsitAiisrmairtil

lsnmmfFg

Fig. 1. Visualization of the

arying areas with uniform heights. Results from this investigationrovide insight into these issues as well. The paper begins with anverview of the research problem with respect to observed lubri-ant deposition and depletion behaviors. It then progresses to thexperiments conducted to investigate the wear mechanism lead-ng to the pellet transfer films. Finally, the results are presented andhen discussed in order to explain why abrasive wear is likely theominant wear mode leading to the transfer films observed, andltimately the friction coefficients measured.

. Overview of the research problem

The annular configuration of the in situ self-replenishing systeman be seen in Fig. 1a. During operation, a lubricious transfer films deposited by a loaded powder pellet onto a rotating disk whicharries the film into the sliding contact of a riding slider pad. Fig. 1bisplays a schematic of the asperity regime, which helps to high-

ight the self-replenishing nature of the system by exaggerating thelider pad and pellet to be on the asperity size scale. The film itselfs partially removed by the slider pad and disk asperity interac-ion, represented in Fig. 1b by the sparsely uncovered asperities.s the disk retraces its original path, these uncovered asperities

nteract with the pellet and a new transfer film is deposited. Thust is inherently “self-replenishing” [1,5,17]. While there have beenuccesses in modeling thick-film powder lubrication as fifth-orderheological models in a continuum framework [3,4,9,11,12], theechanics of transfer film formation in situ prevents this type of

pproach for two reasons. First, these models neglect the mechan-cs of the lubricant in the asperity regime, where thin transfer filmseside. Second, thick-film volume remains constant during opera-ion whereas the volume in the interface of the thin transfer filmncreases and decreases as wear events are continuously occurringocally and globally within the film [1,5,17].

Despite modeling challenges, the transfer film form of solidubrication has great potential in the field of Tribology due to theystem’s potential to be self-contained and activated without exter-al controls [13–16]. In order to describe this type of tribosystem,odels have been suggested which account for the transfer film

echanics through the use of competing rate equations derived

rom lubricant mass balances [5,10,13,17]. In the Control Volumeractional Coverage [5] model developed by Higgs and Wornyoh, aoverning equation was developed to relate coefficient of friction

replenishing tribosystem.

(COF) to the deposition and depletion rates of the powder transferfilm. Contrary to the rheological models which focus on distributedshear stress throughout powder films, the CVFC expresses COF asa function of the lubricant film height present within the asperitydomain with uniform area coverage [5,17]. In this type of model,a low friction coefficient is achieved by ensuring a large enoughlubricant deposition rate, with respect to lubricant depletion rate,such that the lubricant coverage remains at a maximum duringoperation.

In fractional coverage models, the deposition and depletionrates of the transfer film are formulated based on Archard’s adhe-sive wear law as shown below [5,17].

V̇ = K × F × U

H(1)

where V̇ is the volumetric flow rate of worn material, K is anempirical wear coefficient representing the probability of wear, Fis the normal force between the bodies in sliding contact, and His the hardness of the body being worn. K values are found typ-ically through pin-on-disk testing in conjunction with tabulatedapproximations for two materials in adhesive wear [5,17,18].

This has three important consequences. First, these coefficientsare often situation specific to the conditions present during test-ing. Secondly, the use of tabulated values carries a large amountof inherent error due to approximation assumptions [5,17,18]. Andfinally, these K values represent the probability of a wear eventoccurring, meaning they are merely a fraction between 0 and 1meant to account for the average wear over a given time. However,due to the aforementioned transfer film models’ inherent relianceon competing wear rates to determine COF, this type of averagingfails to capture the highly transient nature of the self-replenishingsystem.

For example, Figs. 2 and 3 display the results for two dif-ferent instances where erratic wear behavior in situ leads to asubsequent inability to predict the transfer film dynamics of theself-replenishing system. In these figures, the pellet wear, displayedas the change in pellet height over time captured through a linear

differential voltage transducer (LVDT), is graphed below frictioncoefficient of the slider pad. One should note that pellet “wear”actually refers to the “deposition” of lubricant film on the rotatingdisk as shown in Fig. 1.

124 P.S.M. Dougherty et al. / Wea

Fw

pfassdciis

Ff

ig. 2. Lapses in self-replenishing: (a) slider friction coefficient vs. time and (b) pelletear length vs. time.

The beginning of Fig. 2 displays a relatively routine run-ineriod, where initial wear is high and the establishment of a trans-er film leads to a gradual decline in friction coefficient. However,s wear decreases, represented by a flattening of the LVDT wearlope, we see a drastic increase in friction coefficient. This is due totarvation of the slider-disk interface induced by ongoing lubricantepletion at the slider. A period of extremely high friction coeffi-ient continues for about 100 s from time t = 380 s to 480 s, until thenterface is replenished. Although wear (i.e. lubricant deposition)

ncreases thus returning the system to equilibrium, this highly tran-ient lapse in system lubrication is impossible to capture through

ig. 3. Indefinite starvation through unexplained transfer film dynamic: (a) sliderriction coefficient vs. time and (b) pellet wear length vs. time.

r 272 (2011) 122– 132

the existing wear treatment and could be catastrophic in practicalapplications.

While decreases in pellet wear can lead to potentially hazardousstarvation periods, there are also instances, as shown in Fig. 3,where increases in pellet wear are inexplicably insufficient to sta-bilize and lubricate the system during starvation.

As in the case of Fig. 2, low pellet wear and continued deple-tion by the slider leads to starvation present in Fig. 3 from timet = 100 s to 300 s. Contrary to predictions from our current transferfilm modeling approach, the increase in pellet wear is unable tostabilize the system, thereby leading the system to “indefinite star-vation.” This type of system breakdown implies that something elsemay be happening to hinder the re-establishment of a transfer filmand adequate lubrication conditions.

The combination of these two system failures in Figs. 1 and 2helps to highlight the limitations of current modeling frameworkfor in situ solid powder lubrication, both during the deposition ofthe transfer film as well as its operation during self-replenishment.Due to the constant nature of adhesive wear coefficients, currentmodeling techniques would predict a constant steady state COFat the slider/disk interface. However, transient film behavior leadsto erratic and unpredictable COF in both cases. This study seeksto elucidate the precise wear mechanisms at the pellet/disk andslider/disk interfaces in order to understand the evolution of a self-replenishing transfer film.

3. Experimental set-up

Transfer films were prepared and tested using a pellet-on-diskwith slider pad tribometer [5]. The pellet and slider pad COF wascaptured using calibrated load cells from Massload Co, while thepellet wear was captured using a linear differential voltage trans-ducer (LVDT). The LVDT was attached to the loading mechanismabove the pellet, such that a change in displacement, presented inFig. 3b, could be measured during pellet wear. Lubricant “pellets”were made by compacting dry molybdenum disulfide (MoS2) withan average particle diameter of 1.4 �m to a compaction pressureof 2500 psi. The slider pad and disk were constructed of sinteredtungsten carbide (WC) in order to achieve negligible wear duringexperiment [5,17].

Surface measurements were obtained using a Zygo NewView7300 white light interferometer. A 20× objective was utilized inconjunction with 620 × 480 camera mode and minimum modu-lation ratio (minmod) of 15% to ensure minimum noise into thesurface data while maintaining an XY resolution of 0.36 �m. Z res-olution was rated at 0.1 nm with repeatability up to 0.01 nm RMS.

4. Experimental procedure

In order to investigate the specific interfaces at which the wearmechanisms may affect the transfer film process, tribometer exper-imentation was coupled with surface topography measurementstaken with a scanning white light interferometer. First, standardtribometer runs were conducted in accordance with tribometerexperiments outlined by Higgs and Wornyoh [5], to analyze therelationship between friction and wear at the pellet–disk inter-face. During these tests, the tribometer was allowed to run for600 seconds with a dead load of 17.79 N (4 lbs) on the pellet. Theslider pad was employed on the disk at a weight of 10 lbs, so asto facilitate depletion of the lubricant film and further wearing ofthe pellet. The pellet COF and wear (computed from the change in

pellet length), were recorded through the load cell and LVDT.

The second branch of testing represents a combination of modi-fied tribometer runs and optical interferometer measurements. Thepurpose of these runs was to aid in the visualization of transfer film

P.S.M. Dougherty et al. / Wear 272 (2011) 122– 132 125

Fig. 4. 3D topography scan of the clean disk (pre-test).

Fig. 5. 3D topography scan of the pre-test pellet.

TR

Fig. 6. Sample of (a) adhesive wear and (b

able 1oughness parameters of the pre-test control surfaces.

Surface Ra (�m) Rms (�m)

Clean disk (Fig. 4) 0.106 0.143

Pre-test pellet (Fig. 5) 2.410 2.995

) abrasive wear from Refs. [19,20].

PV (�m) X scan length (mm) Y scan length (mm)

1.5 0.35 0.2636.633 0.71 0.53

1 / Wear 272 (2011) 122– 132

btottdsa

4

ds(

iaPae

ddwtp

4

rt(Afw

5

ttfdq

26 P.S.M. Dougherty et al.

ehavior by examining the wear surfaces involved in the deposi-ion and depletion of the transfer film itself. Preliminary surfacesf the pellet and disk are presented in Figs. 4 and 5 to act as con-rol surfaces for post-test measurements. It is important to notehat in Fig. 4, circumferential lay lines from the polishing proce-ure of the WC are aligned in the direction of disk travel at theliding interfaces. Roughness Parameters for both control surfacesre presented in Table 1.

.1. Lubricant deposition testing

The first part of testing was conducted to facilitate lubricanteposition and prepare an adequate transfer film for study. The diskurface was run-in for 300 s, using the lubricant pellet at a 17.79 N4 lb) dead weight, with no slider employed.

Both the disk and pellet were then taken to the interferometern a sealed canister for analysis. The pellet surface was inspectedt different contact points indicated by a worn, polished surface.ellet wear topographies and roughness values were mapped andnalyzed for comparison against the control surfaces and existingxamples of adhesive and abrasive wear presented in Fig. 6 [19,20].

Inspections of the established transfer films were conducted atifferent points along the disk surface to provide a picture of theisk after film run-in. Transfer film images and roughness valuesere compared against that of the clean disk presented in Fig. 4

o establish quantitative changes in surface topography due to theresence of the transfer film.

.2. Lubricant depletion testing

Once the initial measurements were taken, the disk waseturned to the tribometer for lubricant depletion testing. Theseests were conducted with the slider pad only and loaded at 40.12 N10 lbs) for 600 s to ensure a noticeable change in the transfer film.t the end of each test, the disk was taken once again to the inter-

erometer for surface measurement at different points along theear track.

. Results

In this section, COF and LVDT data are analyzed to ascertain fric-ion and wear correlations at the pellet/disk interface. Next, surface

opography data is examined to qualitatively examine each inter-ace involved in the evolution of transfer films. Lastly, roughnessata provided through surface analysis is investigated to find auantitative argument for synthesis with the friction and wear cor-

Fig. 8. 3D topography scan o

Fig. 7. The relationship between wear and friction in the self-replenishing tribosys-tem.

relations in order to propose a reprise to transfer film mechanics inlater sections.

5.1. Investigation of wear at the pellet–disk interface

Fig. 7a and b displays the results of two pellet-on-disk with sliderpad tests for pellet wear and pellet COF respectively. As explainedpreviously, areas of negligible pellet wear correspond to flatter

regions of the LVDT slope, whereas steeper regions correspond tohigh areas of wear. Using this relationship, it can be seen from thegraphs that increases in LVDT wear correspond to actual drops infriction coefficient.

f the post-test pellet.

P.S.M. Dougherty et al. / Wear 272 (2011) 122– 132 127

Table 2Roughness comparison of the post-test pellet with control surfaces.

Surface Ra (�m) Rms (�m) PV (�m) X scan length (mm) Y scan length (mm)

Clean disk (Fig. 4) 0.106 0.143 1.5 0.35 0.26Pre-test pellet (Fig. 5) 2.410 2.995 36.633 0.71 0.53Post-test pellet (Fig. 9) 0.155 0.193 1.524 0.35 0.26

n of th

ficac[

otF

Fig. 9. 3D topography sca

While adhesive wear is known to lead to increased friction coef-cient, due to the forces created between adhesive junctions on theontacting surfaces being primary components to the friction force,brasive wear is contrarily characterized by a decrease in frictionoefficient through the carving of preferential wear tracks over time22,23].

While this friction and wear relationship alludes to the presencef an alternate wear mechanism, topography measurements forhe worn MoS2 pellet, provide further evidence for abrasive wear.ig. 8 displays the surface topography for a worn test pellet after

Fig. 10. Line scans represen

e deposited transfer film.

deposition testing. When qualitatively compared with Fig. 5 of thepre-test pellet, it can be seen that significant wear of the surfacehas occurred. In the case of the fresh pellet, the surface is very ran-dom, with little orientation of the surface texture. However, afterthe establishment of the transfer film, the pellet surface in Fig. 8acquires a set of parallel trenches similar to machined lay lines or

textured surfaces similar to Fig. 4.

A drastic change in pellet topography was also observed fromthe pre-test to post-test pellet through the roughness comparisonpresented in Table 2. In particular, average roughness values have

ting disk topography.

128 P.S.M. Dougherty et al. / Wear 272 (2011) 122– 132

Fig. 11. (a) 2D line scan of collected powder on the disk surface and (b) 2D topographical map.

differ

dppcp

5

pdp

Fig. 12. Slider COF during two

ecreased from 2.410 �m to as low as 0.155 �m. In addition, it isarticularly noteworthy to compare the roughness of the post-testellet in Fig. 8 to that of the clean disk in Fig. 4. After the con-lusion of lubricant deposition testing, the pellet roughness valuesresented in Table 2 were shown to approach that of the clean disk.

.2. Investigation of the established transfer film

While images from the previous section present an idea of theellet transformation during a test, the actual transfer film, aseposited on the disk surface is displayed in Fig. 9. This 3D topogra-hy scan provides insight as to the creation of the lubricant transfer

Fig. 13. 3D topography of d

ent extended depletion tests.

film, which differs widely from the currently accepted theory ofuniform area coverage.

In order to differentiate the transfer film topography from disktopography, line scans shown in Fig. 10 were taken at differentareas present in Fig. 9. Line scans from the relatively flat areas ofFig. 10, shown in blue, displayed nearly identical average rough-ness to the virgin disk (Fig. 4), whereas the other areas, highlightedin Fig. 11, displayed large deviations from this surface rough-

ness. From this relative difference it was inferred that the disktopography is represented in blue, while the regions of differ-ing color and roughness parameters represented deposited MoS2film.

epleted transfer film.

P.S.M. Dougherty et al. / Wear 272 (2011) 122– 132 129

Fig. 14. Light intensity mapping of the clean disk (a) vs. the depleted transfer film (b).

Table 3Roughness comparison of the deposited transfer film with clean disk.

Surface Ra (�m) Rms (�m) PV (�m) X scan length (mm) Y scan length (mm)

Clean disk (Fig. 4) 0.106 0.143 1.5 0.35 0.26Blue scan average (Fig. 11) 0.103 0.134 0.865 N/A N/A

apdt

ifedddwi

5t

tonasmCt

wra

TR

Disk and film (3-D) (Fig. 10) 0.569 0.646

Disk and film (2-D) (Fig. 12) 0.675 0.770

After differentiating the topographies in Fig. 10, it can be seent first glance that the film in Fig. 9 follows a seemingly randomattern upon the surface with larger clumps spread out in the radialirection. This can be seen from the relative position of the clumpso the existing lay lines still visible on the blue disk surface.

Quantitative roughness analysis of the transfer film presentedn Table 3 reveals that the surface topography has been fully alteredrom that of the clean disk. A line profile and its roughness param-ters presented in Fig. 11a, was taken through the larger anomaliesenoted by the white line in Fig. 11b. By profiling the larger surfaceepartures, one can see from the y-axis in Fig. 11a that these pow-er clumps are as high as 3 �m, while, from the x axis, sometimes aside as 100 �m. These heights are far above that of the clean disk,

ndicating the presence of transfer film above the asperity regime.

.3. Investigation of the slider pad effect on the establishedransfer films

While these images provide an idea of transfer film conception,he results of lubricant depletion testing will help to provide an ideaf the effect of a riding slider on the fresh transfer film. It should beoted that lubricant depletion tests were conducted by imposing

state of starvation on the disk by removing the lubricant sourceuch that a riding slider could wear away the film. While currentodeling expects the removal of lubricant, and therefore a raise in

OF at the slider interface, friction data taken from many depletionests show trends to the contrary.

Fig. 12 shows the COF results of two extended depletion testshere friction behavior actually decreased despite a complete

emoval of the lubricant source. While the second case does haven aspect of initial COF increase, both tests succeed in demonstrat-

able 4oughness comparison of the deposited transfer film with clean disk.

Surface Ra (�m) Rms (�m)

Clean disk (Fig. 4) 0.106 0.143

Post depletion disk (Fig. 14) 0.103 0.134

4.081 0.35 0.262.845 N/A .26

ing that something more complex than just depletion is occurringat the slider disk interface.

By conducting surface analyses of the disk after lubricant deple-tion testing, a possible reasoning for this phenomenon is presentedin Fig. 13. Without the continued supply of lubricant, depletiontesting should result in removal of the solid lubricant from the sys-tem. From inspection of the figure however, there is a noticeabledifference between the depleted disk and the original clean disktopography in the absence of solid lubricant.

In order to verify the presence of film above the surface, the SRzparameter and light intensity maps were calculated using the opti-cal interferometer, as shown in Fig. 14. The SRz provides a uniqueunderstanding of two similar surfaces by taking an average of peakto valley height over an area. This is done to clearly differentiatebetween a surface with very uniform roughness and a surface withsimilar average roughness but deep valleys. For instance, if onesurface has a much smaller SRz value than another surface withsimilar average roughness, it implies that there may be a fillingaspect occurring for the larger valleys. From Table 4, which showsa comparison of surface roughness between the clean disk and thedisk after depletion testing, it can be seen that the SRz values forthe depleted disk were about half as much as the clean disk. Thisimplies that the valleys may have been filled with lubricant, thuslowering the SRz parameter.

A light intensity map may also display key differences betweensimilar surfaces. By showing up brightest at peaks and darkestat valleys, a distinction may be drawn between two surfaces. Forinstance if a build-up of film has occurred above the previous asper-

ities, the light intensity will be higher, and will be represented bybright regions. Likewise, if valleys have begun to fill, we wouldimagine sharp differences in bright peaks and dark valleys to bechanged to a more uniform gray.

SRz (nm) X scan length (mm) Y scan length (mm)

954.63 0.35 0.26567.38 N/A N/A

1 / Wea

bfTvs

6

raiqyt

6

rlwwwobw

sfibiwwiwawt

tctriatlptatdarat

6

tat

also conducted in order to provide greater understanding of themechanisms involved in lubricant depletion. Without the presenceof a replenishment source, an effective starvation is simulated, andthe slider should effectively “clean” the disk as the test progresses.

30 P.S.M. Dougherty et al.

While the clean disk shows a very distinct striated patternetween peaks and valleys, the depleted disk shows a more uni-orm grayish finish, with wide and very high areas of bright white.his last effect may be attributed to the build up of film deep in thealleys and smooth film in the form of “plateaus” on top of the diskurface.

. Discussion

Given the instances of unexplained wear behavior in self-eplenishing powder tribosystems explained in Figs. 2, 3, 7, 12,n investigation was undertaken to analyze interfaces involvedn the transfer film evolution. This procedure yielded a series ofualitative topography observation and quantitative surface anal-sis, which will be combined in this section to propose alternativeransfer film mechanics for an in situ self replenishing tribosystem.

.1. Transfer film inception

In order to study the creation of transfer films in a self-eplenishing system, friction and wear relationships were estab-ished at the pellet interface through tribometer testing in Fig. 7. As

ear increased at the pellet, a noticeable drop in friction over timeas observed. Although previous works assumed that adhesiveear was the primary wear mechanism involved in the creation

f MoS2 transfer films, Fig. 7 displayed an inverse relationshipetween friction and wear characteristic of pronounced abrasiveear.

While this relationship provides initial evidence against adhe-ive pellet wear, surface scans taken at the pellet interface, revealurther leveraging for identifying abrasive wear as the predom-nant mechanism of lubricant deposition. Qualitatively this cane reasoned through the comparison of pellet topography with

nstances of adhesive or abrasive wear in Fig. 6. Adhesive wearould be characterized by a random pock marking at locationshere adhesive junctions would have been formed and resulted

n material removal during sliding [19,22]. However, abrasiveear is identified as a pattern of parallel grooves where sliding

sperities have ploughed into a surface [18,20,21]. This abrasiveear landscape agrees far more closely with the post-test pellet

opography in Fig. 8.This phenomenon can be explained quantitatively as well, in

he close approach of pellet roughness parameters to that of thelean disk presented in Table 2. Despite a large difference in pre-est pellet surface roughness when compared to the clean disk, theoughness of the post-test pellet and clean disk become very sim-lar as pellet is deformed through wear. The extremely hard WCsperities can be thought to indent and then plough easily throughhe soft MoS2 surface, leaving a pattern very similar to the parallelines present on its own surface. This would result in the observedattern of parallel grooves of similar dimension to the lay lines ofhe disk surface. Given the large difference in hardness between WCnd MoS2, this type of abrasive ploughing [18] is supported by rela-ionships where interactions between materials with drasticallyifferent hardness values have been shown to exhibit primarilybrasive wear [23]. It should also be noted that the post-test pelletoughness suggests that some of the particles at the surface havelso been deformed, as evidenced by the roughness far lower thanhe characteristic diameter of the average powder particle.

.2. The pre-slider film

An early investigation of the film, before the employment ofhe riding slider, was performed through surface measurement of

transfer film after lubricant deposition testing. Recall, that theseests are conducted with only the presence of the lubricant source

r 272 (2011) 122– 132

without any type of slider. Originally, thin powder films were theo-rized to adhere uniformly to the surface in the asperity regime. Thiseffect was justified by the fact that lamellar powders, such as MoS2and graphite, possess strong bonds within lamellar sheets, yet areweak between lamellar sheets. Whole sheets of lamellar powderwere thought to be removed through shearing while evenly adher-ing to the surface. These lamellae would coat the surface, lowerfriction and prevent further wear [1,24].

Analysis of our pre-slider film provides a different picture ofthe fresh transfer film, where powder collected as random, multi-particle clumps rather than mono-layer particle sheets. 2D linescan analysis of these powder clumps shows that the film residesa full order of magnitude above the asperity domain, yet scatteredaround the surface. This is very different than despite the previ-ous modeling assumptions of transfer films residing at a variableheight and uniform area in the asperity regime. This implies twofactors: that transfer film lubrication is actually a function of areaand height based coverages, and that actual transfer film lubricationlikely begins as a combination of quasi-hydrodynamic or full filmlubrication [9,11,12], mixed lubrication, and boundary lubricatedconditions.

This random arrangement of particle clumps may also provideevidence that the pellet transfers film due to abrasive wear at theasperity scale when sliding. While adhesion would lead to moreuniform coverage of lamella layers onto the surface which wouldprevent further multi-particle build-up at a location on the disk.Abrasive wear could cause multiple clumps to fall loosely on topof each other. Disk asperities may be thought to violently ploughinto the pellet’s surface and then continue through it, causing dis-lodged clumps of particles to separate randomly from the pellet.These particles could then collect together similar to a snow storm,in the multi-particle formations seen in transfer film topographypresented in Fig. 9.

6.3. The post-slider film

Analysis of transfer films after the application of a slider were

Fig. 15. A single disk/slider pad asperity junction visualized as a rayleigh step bear-ing.

P.S.M. Dougherty et al. / Wear 272 (2011) 122– 132 131

dynam

Po

dfiSaaFt

fifitartDibpmafcms

faifisitl

7

pouioteot

Fig. 16. Schematic of alternative transfer film

ast modeling of this phenomenon has represented the slider actingnly as a mechanism of lubricant removal.

In opposition to our expectations, qualitative inspection of theisk surface topography appeared to still contain large amounts oflm. Despite similar roughness parameters to that of the clean disk,Rz analysis and light mapping were used to reveal the presence of

more uniform film residing both packed into the asperity valleysnd collected above the asperity zone. A schematic is presented inig. 15 to aid in visualizing the dynamics of the riding slider pad onhe transfer film.

Fig. 15 models the interactions between disk asperity, transferlm, and slider pad asperity as a simple Rayleigh-step bearing con-guration for visualization purposes. The contact interface betweenhe slider and disk can be imagined as a collection of contactingsperities and the voids in between. If a disk or slider asperity isepresented as the step region of the bearing, these voids betweenhe asperities can be represented then by the entrainment region.ue to the larger size of powder particles compared to disk asper-

ties, some of the powder is ejected from the interface as showny the depleted lubricant particle. However, some of the very softowder particles may be entrained into the gap leading to defor-ation and smoothening of the transfer film both over the surface

nd into the voids. Gaps that are particularly filled could actuallyorm layers of compacted powder above the disk asperities. Theseompacted regions are seen as the plateaus in the light intensityapping presented in Fig. 14b, and can act as regions of full film

olid lubrication.This type of modeling could also help to explain a phenomenon

ound during lubricant depletion testing, where friction valuesctually lower over time despite the absence of a lubricant replen-shment source (Fig. 12). The presence of a slider on the freshlm could not only act to deplete the lubricant film, but also topread the film evenly throughout the surface, This would lead toncreased lubricant coverage of the surface and therefore less fric-ion at the slider interface even in the absence of a replenishingubricant source.

. Summary and conclusions

While solid powder transfer film lubrication possesses theotential to be a viable commercial lubrication technology, muchf the predictive science for its behavior is still immature. In partic-lar, the problems of starvation and replenishment lapses impede

ts confident implementation into practical systems. It was the goalf this study to investigate the wear mechanisms which govern the

ransfer film evolution in an effort to enhance the existing knowl-dge of a self-replenishing powder transfer film tribosystem. Basedn the results from this work, Fig. 16 was developed to describehe process by which transfer films initiate and endure in a self-

ics as proposed from the results of this work.

replenishing tribosystem. In addition, a three-step hypothesis ofwhat is actually transpiring at each interface is also provided.

Pellet particles are removed primarily through abrasive wear.This is supported by both the relationship between pellet frictionduring periods of high wear and also by a surface topography evo-lution that is more characteristic of abrasive wear mechanisms.

The loose third-body particulates fall and form large, multi-particle powder clumps above the asperity regime. This can bevisualized intuitively as snow collecting on the top of a mountain.Evidence for this conclusion is provided through the investigationof disk surfaces after deposition testing. These measurements dis-play a random collection of clumps which can be as large as two orthree particle diameters, and as wide as 100 particle diameters.

As the transfer film is carried past the riding slider, a combina-tion of two effects occurs as displayed in Fig. 15. As soft powderparticles of larger diameter interact with the slider, the result isnot only a depletion or wearing away of the lubricant film, but alsoan entrainment effect where particles are drastically deformed inorder to fill the valleys and gaps between disk/slider asperity con-tacts. The slider entrainment effect can also lead to the formationof “plateaus” above disk asperities which could provide full-filmquasi-hydrodynamic lubrication. This effect may be the cause fora decrease in friction coefficient despite the imposed starvationduring lubricant depletion testing. Support for this conclusion isprovided by both qualitative inspection of the topography changein the clean disk vs. post slider disk images. Further support is alsoprovided by the changes in overall roughness parameters such asSRz, and difference in light intensity mapping between a clean diskand a post-test disk.

In closing, this work is meant to elucidate the wear mecha-nisms in a self-replenishing transfer film lubrication system wherethe sacrificial solid being worn is comprised of lubricious, lamel-lar powder. While previous modeling treatments may still provideacceptable results, inexplicable system behavior in highlightedcases stresses the need for a greater understanding of wear mecha-nisms in the transfer film process. This work aims to give evidencethat asperity-scale abrasive wear is actually dictating the tribo-logical performance of the pellet-on-disk with slider transfer filmprocess, while also helping to uncover the state of the transferfilm at different stages of the self-replenishing process. By gain-ing a proper understanding of the wear mechanisms leading tothe deposition and depletion of the transfer film in this tribosys-tem, the authors hope to be able to facilitate improved modeling ofself-replenishing transfer film experiments.

Acknowledgements

The authors would like to thank Kennametal Inc. for providingslider pad and disk samples. We would also like to thank Emmanuel

1 / Wea

WcpSt

R

[

[

[

[

[

[

[

[

[

[

[

[

[

32 P.S.M. Dougherty et al.

ornyoh for his earlier experimentation and modeling work whichontributed to our understanding of the complicated transfer filmroblem. Lastly, we would like to thank Justin Turner and Tylerteele at Zygo, for their suggestions on key optical profilometryechniques which were useful for this work.

eferences

[1] E. Wornyoh, V. Jasti, C.F. Higgs, A Review of dry particulate lubrication: powderand granular materials, Journal of Tribology (2007) 438–449.

[2] H. Heshmat, D. Brewe, Performance of powder-lubricated journal bearings withMoS2 powder: experimental study of thermal phenomena, Journal of Tribology117 (3) (1995) 506–512.

[3] H. Heshmat, M. Godet, Y. Berthier, On the role and mechanism of dry tribopar-ticulate lubrication, in: Presented at 49th STLE Annual Meeting, vol. 51, no. 7,Pittsburgh, Pennsylvania, May 1–5, 1994, STLE, Lubrication Engineering, 1995,pp. 557–564, Limiting Shear Stress.

[4] Tribology of Interface Layers, by Hooshang Heshamt, CRC Press, USA, 2010, ISBN978-0-8247-5832-5 (Hardcover: alk.paper).

[5] C.F. Higgs III, E. Wornyoh, An in situ mechanism for self-replenishing powdertransfer films: experiments and modeling, Wear 264 (1–2) (2008) 131–138.

[6] M. Lovell, C.F. Higgs, P. Deshmukh, A. Mobley, Increasing formability in sheetmetal stamping operations using environmentally friendly lubricants, Journalof Materials Processing Technology 177 (1–3) (2006) 87–90.

[7] W.O. Winer, Molybdenum-disulfide as a lubricant: a review of the fundamentalknowledge, Wear 10 (1967) 422–452.

[8] M. Lovell, P. Deshmukh, W.G. Sawyer, A. Mobley, On the friction and wear per-formance of boric acid lubricant combinations in extended duration operations,Wear 260 (11–12) (2006) 1295–1304.

[9] H. Heshmat, The rheology and hydrodynamics of dry powder lubrication, Tri-bology Transactions 34 (3) (1991) 433–439.

10] J.A. Bares, N. Argibay, P.L. Dickrell, G.R. Bourne, D.L. Burris, J.C. Ziegert, W.G.Sawyer, In situ graphite lubrication of metallic sliding electrical contacts, Wear267 (9–10) (2009) 1462–1469.

[

[

r 272 (2011) 122– 132

11] H. Heshmat, O. Pinkus, M. Godet, On a common tribological mechanismbetween interacting surfaces, STLE Transaction 32 (1) (1989) 32–41, Complex-ity of solid lubrication.

12] I. Iordanoff, Y. Berthier, S. Descartes, H. Heshmat, A review of recent approachesfor modeling solid third bodies, in: Presented at the ASME/STLE Tribology Con-ference, October 2001 San Francisco, CA, ASME Transaction Journal of Tribology124 (October) (2002) 725–735.

13] R. Kaur, H. Heshmat, 100 mm diameter self-contained solid/powder lubri-cated. Auxiliary bearing operated at 30,000 rpm, Lubrication Engineering 58(6) (2003) 13–20.

14] J.K. Lancaster, Lubrication by transferred films of solid lubricants, ASLE Trans-action 8 (1965) 146–155.

15] J.K. Lancaster, Anisotropy in the mechanical properties of lamellar solids andits effect on wear and transfer, Wear 9 (1966) 169–188.

16] A.J. Haltner, An evaluation of the role of vapor lubrication mechanisms in MoS2,Wear 7 (1964) 102–117.

17] E. Wornyoh, C.F. Higgs III, An asperity-based fractional coverage model fortransfer films on a tribological surface, Wear (2010).

18] R. Ernest, Friction and Wear of Materials, 2nd ed., Wiley-Interscience, New York,1995.

19] L. Lasa, J.M. Rodriguez-Ibabe, Effect of composition and processing route on thewear behaviour of Al–Si alloys, Scripta Materialia 46 (6) (2002) 477–481.

20] R.I. Trezona, D.N. Allsopp, Transitions between two-body and three-body abra-sive wear: influence of test conditions in the microscale abrasive wear test,Wear 225–229 (1) (1999) 205–214.

21] M.A.S. Quintanilla, D.T. Goddard, Lateral force microscopy with micrometer-sized particles: effect of wear on adhesion and friction, Wear 268 (2010)277–286.

22] J.-M. Wu, S.-J. Lin, J.-W. Yeh, S.-K. Chen, Y.-S. Huang, H.-C. Chen, Adhesive wearbehavior of AlxCoCrCuFeNi high-entropy alloys as a function of aluminum con-tent, Wear 261 (5–6) (2006) 513–519.

23] G. Dearnaley, Adhesive and abrasive wear mechanisms in ion implanted metals,Nuclear Inst. and Method in Physics Research (1985) 158–165.

24] R.F. Deacon, J.F. Goodman, Lamellar Powders in Lubrication, Proceedings ofthe Royal Society of London. Series A, Mathematical and Physical Sciences 243(February (1235)) (1958) 464–482.