8Wear Debris

Classification

8.1 Introduction 8.2 How Wear Debris Is Generated8.3 Collection of Wear Debris8.4 Diagnostics with Wear Debris

Particle Counting • Biological Debris Sampling • Reworked Wear Debris

8.5 Conclusions

8.1 Introduction

Wear generates debris. The debris comes in a wide variety of sizes and shapes. Wear debris turns motoroil black. You can see it on your hands if you shake hands with your garage mechanic. The black in usedoil is like a pigment — it is nanometer-size colloidal metal particles (and carbon) suspended in the oil.If you take an oil filter apart, you will find other types of wear debris. Some are shiny metal particlesvisible to the naked eye and in the millimeter size category. If you are in the surface mining business,you will find even larger metal particles in the gravel produced by rock crushers. This is metal that isgouged from the metal jaws in the crusher. In a factory grinding room, dust swept up from the floorand examined in a scanning electron microscope will show perfect micron-size metal spheres from rapidlysolidifying drops of molten metal generated by the grinding wheel. Examples of these unique debrisparticles will be found later in this chapter.

Wear debris represents loss of geometric accuracy of moving contacting parts. It can also foul orificesand close spaced parts. Although the total material lost as wear debris in a machine is minute comparedto the volume and weight of the moving parts, it can signal failure of gears or bearings, and expensiverepairs or warranty payments.

Because wear debris can be carried in a circulating oil lubrication system, the condition of an engineor a critical machine can be diagnosed by analyzing a sample of the oil. Debris can be removed from theoil by centrifuging or by a magnetic process. The separated debris can then be examined in a microscopeto determine morphology, or it can be chemically analyzed. A standard test, the spectrographic oil analysisprogram (SOAP), has been used for condition monitoring by railroads, truck fleets, and the military fortanks and aircraft.

Until the Industrial Revolution and the development of steam and internal combustion engines, wearwas taken for granted. When wear of machinery and tools became a problem, the development of wear-resistant materials and the study of wear itself increased rapidly. Now we have a large body of informationdeveloped from experience and scientific investigation that can be used in the design and maintenanceof more reliable and economical machinery.

William A. GlaeserBattelle

8.2 How Wear Debris Is Generated

Early wear theory, as related by Bowden and Tabor (1964) and Holm (1946) indicated that real surfacescontact at high spots in the surface microtopography (asperities) and wear was caused by the shearingoff of these contacts. The theory was not clear, however, on how the wear particles were separated fromcontacting asperities. One obvious mechanism is cutting, as in machining. Under abrasive wear condi-tions, it is presumed that hard particles embedded in one surface will plow through a softer counterfaceand produce microchips. The debris thus generated looks like the example shown in Figure 8.1. Thecurled form that the debris particles assume reminds one of chips from machining. In fact this signalsa severe type of abrasive wear. The attack angle, hardness, and depth of penetration of abrading particlesdetermine whether cutting-type abrasion will occur as shown by Mulhearn and Samuels (1962). Afrequency distribution of attack angles for abrasive particles on SiC bonded abrasive is shown inFigure 8.2. A number of metals are listed in Table 8.1 together with their critical attack angles — or the

FIGURE 8.1 SEM micrograph of cutting-type debris from abrasive wear, M2 tool steel. (From Rigney, D.A. (1992),The role of characterization in understanding debris generation, Tribology Series, 21, Wear Particles: From the Cradleto the Grave, Elsevier. With permission.)

FIGURE 8.2 Diagram showing the effect of cutting angle on hard particle cutting wear.

100

80

60

40

20

00 20 40 60 80 100 120 140 160 180

Attack angle - α

Noncutting points Cutting points

Fre

qu

ency

Criticalattackangle

α

angle above which cutting will occur. Note that the lower the critical angle, the greater the percentage ofparticles that will cut. Note also that hardened steel is quite susceptible to cutting abrasion. In general,the higher the hardness of the metal, the more susceptible to cutting wear it is. In order for cuttingabrasion to take place, the hardness of the abrasive medium must be over 1.2 times that of the materialbeing abraded.

Contact load between sliding surfaces does not have to be large for cutting wear to take place. In fact,Zhao and Bhushan (1998) have demonstrated that cutting abrasion can occur with normal loads of 40 to80 µN using a diamond pyramidal point on single crystal silicon. The debris produced was similar inmorphology to metal-cutting debris (curly and stringy). The size of the chips was about 50 nm wide by200 nm long.

Wear debris can also be generated by plastic deformation. Hard asperities, plowing over a softer surfacewithout cutting will produce ridges. Ridges can then be flattened by further contact. Extrusions or lipsare formed. These lips are broken off and become flat wear flakes. The process is illustrated in Figure 8.3.Extrusions can also occur at the exit border of a wear scar formed by a curved surface rotating againsta flat surface. An example is shown in Figure 8.4. The example in Figure 8.4 is from a block-on-ring testin which a ring rotates against a block.

Wear debris can also be generated by material transfer from one surface to another. An asperity plowingthrough a surface will cold weld to the mating surface and the weld will shear off, leaving some of theasperity stuck to the mating surface. An in situ scanning electron microscope (SEM) wear experiment byGlaeser (1981) demonstrated this process. A SEM photomicrograph shows this in Figure 8.5. You can seethe track the asperity made as it penetrated the other surface. Then, suddenly, adhesion occurs and a pieceof the asperity breaks off and sticks to the surface as a lump. Figure 8.6 shows the lump after the rubbingsurface has passed over it many times. The lump has been squashed and is breaking up into smaller particles.

A unique type of wear debris is found in grinding swarf and from electric spark discharge damage tometal surfaces. It is shown in Figure 8.7. Each particle is spherical. A dendritic microstructure is oftenvisible on the sphere surface. The particles are formed from molten droplets caused by frictional heatingof the grinding process. They can also be produced by the hot arc developed in electric discharge damageto metal surfaces or from electric arc welding. Spherical debris can be found in oil samples from machineryoperating in the vicinity of grinding operations.

Rubber produces wear particles by a different process than metals. During abrasion, the rubber isheated by frictional energy, softens, and sticks to the abrading surface. The rubber then stretches andforms abrasion ridges oriented normal to the direction of sliding. An example is shown in Figure 8.8.Debris comes from these ridges in the form of rolled up particles. A familiar example is the debris thata rubber eraser produces.

Most of the wear debris formed by the above mechanisms is visible using light microscopy. Submicrondebris can be found in SEM analysis of material filtered from used oil. In a series of experiments designedto observe the generation of wear debris under lubricated conditions, a block-on-ring wear tester wasused by Glaeser (1983). The block was Cu-3.2 wt% Al alloy and the ring was carburized steel. A drop of

TABLE 8.1

HVMaterial kg/mm2 Critical Attack Angle

Lead 5 50° to 60°α-brass 180 50° to 60°Aluminum 35 80° to 90°Aluminum alloy 145 ~35°Copper 120 40° to 50°Nickel 350 60° to 70°1040 steel, annealed 180–260 40° to 90°1082 steel, quenched and tempered 860 ~30°Tool steel, quenched and tempered 760 ~30°

silicone oil was applied to the ring so that it clung to the bottom of the ring. The configuration is shownin Figure 8.9. The drop was illuminated with a light source and was observed through a microscope. Asthe ring rotated, oil from the drop circulated through the block contact and back to the drop, carryingany debris generated into the drop. At the beginning of rotation, the oil drop began to fill with metalflakes that sparkled as they tumbled in the drop. Later in the test, the drop became cloudy. At this point,the experiment was terminated and the drop removed for analysis. The large wear debris was removedby centrifuging. The still cloudy oil was then filtered through a Nucleopore polycarbonate filter.

FIGURE 8.3 Ways in which wear debris is formed by plowing action.

FIGURE 8.4 SEM micrograph of stainless steel extrusions.

PLOWING MICRO CUTTING EXTRUSION

EXTRUSION WEDGING FORWARD EXTRUSION

GOUGING

The filter was mounted on a carbon grid for an electron microscope and the filter dissolved with asolvent. The remaining material was then examined in a transmission electron microscope and was foundto consist of equiaxed nanoparticles embedded in a gel. An example of the debris is shown in an STEMmicrograph, Figure 8.10. The size of the round solid particle indicated in Figure 8.10 is about 130 nm.EDX analysis of the particle showed it to be 80% Cu, 1.6% Fe, and 2% Al. The balance was silicon. The

FIGURE 8.5 SEM micrograph of in-SEM wear experiment, showing asperity plowing, adhesion, and shear off.Transfer of iron to copper–nickel alloy.

FIGURE 8.6 SEM micrograph of transfer lump after being run over many times by the mating surface.

silicone oil had formed a gel with the submicron metal particles. The large bronze flakes that appearedinitially were from the wear-in process as the roughness in the bronze surface was removed. Heavy surfaceflow produced smeared layers. As these layers were subjected to increasing numbers of rubbing cycles,particles broke off and were trapped in the surface. These particles were reduced to very small particlesby comminution, as happens in the ball milling process. SEM analysis revealed particle reduction inprogress on the aluminum bronze block as shown in Figure 8.11. The submicron particles producedreacted with the oxidizing lubricant to form a gel which clouded the oil drop. This type of wear is oftenfound in boundary-lubricated or very thin film elastohydrodynamic-lubricated contacting surfaces.

8.3 Collection of Wear Debris

In circulating oil systems, the SOAP (spectrographic oil analysis program) process of debris monitoringhas proved valuable in certain applications. Instead of extracting the debris from the oil, the whole oil

FIGURE 8.7 Spherical wear debris from grinding swarf.

FIGURE 8.8 Rubber abrasion pattern.

sample is subjected to spectrographic analysis. A summary of this type of oil analysis was reported byLukas and Anderson (1998). Metal content is measured in ppm. The following potential failures can bedetected in this way:

• Broken piston rings, worn or scuffed cylinder walls, and worn valve guides

• Worn ball or roller bearings and/or retainers

• Worn journal bearings

• Worn spline couplings

• Worn cams and followers

• Scored pistons in hydraulic pumps

FIGURE 8.9 Experimental setup for viewing wear particles in an oil drop in a block-on-ring wear tester.

FIGURE 8.10 TEM micrograph of a submicron wear particle embedded in a gel.

This method works well for colloidal suspensions of metals in oil. However, it is not practical foranalysis of larger pieces. By examining the shape, surface texture, and chemistry of larger particles, onecan obtain supplemental information on failure of mechanical components.

Wear debris can be collected by allowing it to fall on a glass slide. The debris can then be fixed to theslide and examined under a microscope. In a circulating oil system, a filter in the line can be used to collectdebris. Debris can be washed out of the filter with solvent, or the filter element, if of an organic material,can be dissolved by an appropriate solvent and the remaining metallic debris collected. Ferrography, devel-oped in the 1970s and described by Seifert, Westcott, and Bowen (1972), uses a magnetic field to collectferromagnetic particles from a fluid. The fluid with suspended metal particles is allowed to flow over a glassslide inclined at a low angle. The glass slide is held to a strong permanent magnet in such a way that amagnetic gradient is developed along the length of the slide. In this way, the particles are separated by size.Ferrous particles separate in strings, with the largest particles at one end. Weak magnetic particles depositrandomly on the slide. A complete issue of the journal Wear (1983) was devoted to ferrography.

Centrifuging suspensions of wear debris in oil-solvent solutions will separate metal particles accordingto mass. The particles will be distributed in layers. Once the centrifuging is complete, the fluid is decantedfrom the vessel and the wear particles removed by washing out with solvent. The solvent volume can bereduced by evaporation, the wear particles suspended by ultrasound and the remaining fluid poured outon a glass slide. The wear particles will be well dispersed on the slide when the solvent evaporates. If onewishes to separate the particles by mass, the layers can be removed by pipette and deposited on separateslides.

Wear tests will often leave wear debris on one of the wear specimens. For instance in a fretting test inwhich a ball is oscillated against a flat, debris will appear at the edges of the wear scar. The debris canbe picked up with a cellulose acetate film dampened with solvent. The film can then be dissolved, leavingthe debris free for examination.

8.4 Diagnostics with Wear Debris

As has been noted, wear debris comes in all sizes and shapes. The way in which the debris has beenformed can be deduced from size, shape, and surface texture. For instance, cutting wear debris is uniquein having a curly and stringy shape. The presence of cutting wear debris in an oil sample signals a severewear problem needing immediate attention.

FIGURE 8.11 Wear scar of aluminum bronze block showing the process of comminution of wear debris.

Use of wear debris as a diagnostic tool for the health of operating machinery is difficult. Recognitionof significant morphological characteristics, color, and chemical make up of the variety of particles andelements that turn up in a sample requires experience. The SOAP program has proven effective for truck,tank, and railroad diesels, hydraulic systems, gear boxes, and compressors. The diagnosis is based on thehistory of concentration of specific elements found in oil samples. The particles involved are under 10 µmin size and more likely to be of submicron size. They are detected by spectrographic methods which givequantitative results, usually in ppm. The quantitative results are plotted as a function of running hours.An example of such a plot is shown in Figure 8.12. The trends for several metals are compared with theoriginal oil sample taken at hour one. By experience, these trends (for instance, sudden continuingincrease in iron content) can be related to pending wear failures of certain components.

8.4.1 Particle Counting

Particles from a given sample of oil separated by ferrograph or other methods can be classified as to sizeand a size-frequency distribution plot made. Assuming that the larger particles are related to severe wear,significant increases in their content compared with the running-in plot should signal shut down andmaintenance. A sample plot from a review by Lockwood and Dally (1992) is shown in Figure 8.13. Itshows the change in size distribution as running hours accumulate.

The shape, thickness, and chemistry of a wear particle can tell much about how it was formed.Individual particles can be examined by SEM. An atlas of particle types has been assembled by Anderson(1982) and contains a wide variety of wear particles collect by ferrography. Micrographs, some in color,are presented together with the conditions that produced the wear particles. A page from the atlas isshown in Figure 8.14. The figure shows the way in which the particles line up in strings in the ferrographs.Individual particles can be seen in SEM micrographs in Figures 8.15 and 8.16. Some of the more easilyrecognized characteristics of wear particles include:

• Wear-in conditions. Flat, fairly smooth particles about 10 µm or smaller formed from broken offlips and extrusions produced by machining.

• Cutting. Curled strings and crumpled flakes resulting from abrasive wear process.

• Deformation. Flat flakes with parallel striations from breakout of deformed surface material causedby heavy sliding contact.

FIGURE 8.12 Plot of iron and chromium content as a function of hours of engine operation. (From spectrographicanalyses of oil samples.)

120

110

100

90

80

70

60

50

40

30

20

10

0

10 20 30 40 50 60 70 80 90 100 110 120

pp

mIron

Chromium

Hours

• Contact fatigue. Chunk-like particles 50 µm in size resulting from contact fatigue spalls. Someinvestigators, including Douglas Scott (1983) found spherical debris 1 to 5 µm in size having thesame composition as the steel of ball bearings in engine oil and attributed them to contact fatigue.One must be careful in drawing conclusions from spherical debris. It has been found in hydraulicsystems and engines having no rolling contact bearing in the oil stream. J.E. Davies (1986) foundspherical particles in diesel engine oil. A dendritic pattern was shown on the particle surfaces asshown in Figure 8.7. These were solidified liquid drops. The spherisurface features and theircomposition should be determined to diagnose their source.

• Babbitt fatigue. Silver-colored flakes of babbitt analyzing for tin or lead (by EDS).

• Fretting. For iron base materials, orange powder with particle size on the order of 100 nm andcontaining alpha Fe2O3 (hematite).

Under heavy load, surface features can be produced that are peculiar to the surface morphology. Forinstance, Sarkar (1983) noted that the surface of an Al–Si alloy pin which had been run against a hardeneddisk had developed “inclined shear plates” on the surface. A particle, extracted from the wear debrisshowed the same morphology. Therefore, it was recognized as a piece of the pin surface broken outduring heavy sliding contact. The pin and particle are shown in Figure 8.15.

Ferrographs will also collect nonmagnetic and nonmetallic particles. They will tend to precipitate outof the fluid as it moves over the glass slide. The same analysis can be performed on wear particles separatedfrom oil by other means (filters, centrifuging). Organic materials can be detected by viewing the slidewith transmitted light. Polarized light will show up crystalline minerals.

8.4.2 Biological Debris Sampling

Human joints are similar to lubricated mechanical components. The hip joint is a ball-in-socket config-uration with cartilage sliding against cartilage, lubricated with a slippery lubricant called synovial fluid.As these joints age, the cartilage is worn away until exposed bone slides on bone. This causes irritationof the joint and inflammation and pain. In the last 25 to 30 years, worn joints have been replaced withartificial components. One part of the joint prosthesis is made of a tough plastic compatible with bodytissue (ultra-high-molecular weight polyethylene). The other component is a metal alloy (titanium alloyor cobalt base alloy). The combination operates with relatively low friction but is subject to wear. Boththe original human joint or the prosthesis can be tested for wear by using the ferrographic method asshown by Mills and Hunter (1983). An example of extracted cartilage particles is shown in Figure 8.16.

FIGURE 8.13 Plot of wear particle size distribution from oil analysis.

0.1 1 10 100 103

Time

Wear particle size, µm

Catastrophic failureAdvanced failure modeOnset of severe wear modeBenign wear

Wea

r p

arti

cle

con

cen

trat

ion

.

par

ticl

es/1

0 m

L

In biological joints, samples of synovial fluid were extracted and treated with a magnetizing solution byEvans and Bowen (1980). The particles recovered by ferrography can be identified by stains or X-rayanalysis.

Examination of tissues around failed total hip replacements for UHMWPE wear debris requires aspecial process described by Campbell, Ma, Yeom et al. (1995). Tissue samples were processed and thepolymer wear debris extracted and deposited on Nucleopore polycarbonate filters. The filter plates

FIGURE 8.14 Page from Wear Particle Atlas.

containing the wear debris were sputter coated with gold palladium and examined in an SEM. A typicalSEM micrograph of the debris is shown in Figure 8.17.

8.4.3 Reworked Wear Debris

It must be assumed that debris collected from a wear process is as it was when separated from a surfaceif one is to deduce how it was generated. This often is not the case. Wear debris can go through a contactzone many times and in the process be altered in size and shape. This is true in roller bearings, recipro-cating flat-on-flat sliders, splines, gear couplings, and wet clutches. Ductile metals can be mashed outinto extremely thin flakes — thin enough to transmit electrons. When these submicron particles areexamined in a transmitting electron microscope (TEM), they appear transparent. An example is shownin Figure 8.18. In experiments to produce this type of particle, iron powders were ball milled underlubricated conditions by Glaeser (1991) and similar submicron flakes produced as shown in Figure 8.19.

8.5 Conclusions

Wear debris comes in many sizes and shapes. Study of wear debris morphology and chemistry can provideclues to the condition of various lubricated parts in a machine and the wear mechanisms extant — all

FIGURE 8.15 SEM micrographs of a wear scar and a wear debris particle showing the same surface roughnessfeatures.

FIGURE 8.16 SEM micrograph of cartilage particles extracted from a human joint and separated by ferrography.

this without tearing an engine down. By monitoring the amounts of given metals in a circulating oilsystem as a function of running time, one can detect the onset of a mechanical part (bearing, gear, pistonring, or seal) failure. This program (SOAP) has been successfully used for a number of years for conditionmonitoring of engines.

Individual wear particles can provide clues to the wear mechanism that produced them. For instance,a sudden increase in bronze particles from an engine with bronze bearings may be the result of ingestionof abrasive contaminants. Just monitoring the increase in copper in the oil would not indicate the changein wear mode. Bronze particle size and shape would suggest this. Failure analyses can be enhanced bythe examination of wear particle size and shape.

FIGURE 8.17 SEM micrograph of UHMWPE wear debris from a human joint prosthesis.

FIGURE 8.18 TEM micrograph of a very thin submicron wear flake.

Researchers can use wear particle analysis as a tool to follow the wear process as it changes with changesin operating conditions (bearing load, change in environment, increased sliding velocity). Classificationof wear debris according to size, thickness, shape, color, and chemistry and relating debris types to wearmechanisms or failure of mechanical components has been the goal of debris atlases. However, care mustbe exercised in using this information since the character of wear debris is often influenced by factorsother than wear mode or parts failure. Diagnostics with wear debris requires experience in interpretingand recognizing significant characteristics.

References

Anderson, D.P. (1982), Wear Particle Atlas, NAEC Report 92-163, Lakehurst, NJ.Bowden, F.P. and Tabor, D. (1964), The friction and lubrication of solids, Part 1 (1950) and Part II,

Oxford Press, London.Bowen, E.R., Scott, D., Seifert, W.W., and Westcott, V.C. (1964), Ferrography, Tribology International,

9(3), 109-115.Campbell, P, Ma, S., Yeom, B., McKellop, H., Schmalzried, T.P., and Amstuz, H.C. (1995), Isolation of

predominantly submicron sized UHMWPE particles from periprosthetic tissue, J. Biomed MaterialRes., 29, 127-131.

Davies, J.E. (1986), Spherical debris in engine oil, Wear, 107-189.Evans, C.H., Bowen, E.R. et al. (1980), Synovial fluid analysis by ferrography, J. Biochem. Biophys. Meth.,

2, 11-18.Glaeser, W.A. (1981), Wear experiments in the scanning electron microscope, Wear, 73, 371-386.Glaeser, W.A. (1983), The nature of wear debris generated during lubricated wear, ASLE Transactions,

26, 517-522.Glaeser, W.A. (1991), Ball mill simulation of wear debris attrition, Proceedings, 18th Leeds–Lyon Sym-

posium on Tribology, Tribology Series 21, Elsevier, 515-521.Holm, R. (1946), Electric Contacts, Almquist and Wicksells, Uppsala, Sweden.International Conference on Advances in Ferrography (1983), Wear, 90, 1-181.Lockwood, F.F. and Dally (1992), Lubricant Analysis, 18th edition, Metals Handbook, Tribology, ASM, 19,

300-312.Lukas, M. and Anderson, D.P. (1998), Laboratory used oil analysis methods, Lubrication Engineering,

STLE, 31-35.

FIGURE 8.19 TEM micrograph of an iron particle subjected to ball milling to simulate wear debris subject torepeated passage through a rolling contract.

Mills, G.H. and Hunter, J.A. (1983), A preliminary use of ferrography in the study of arthritic diseases,Wear, 90, 107-111.

Mulhearn, T.O. and Samuels, L.E. (1962), Abrasion of metals, Wear, 5, 478-498.Rigney, D.A. (1992), The role of characterization in understanding debris generation, Tribology Series

21, Wear Particles: From the Cradle to the Grave, Elsevier, 405-412.Sarkar, A., D. (1983), The role of wear debris in the study of wear, Wear, 90, 39-47.Scott, D. (1983), The application of ferrography to the condition monitoring of gas turbines, Wear, 90,

21-29.Seifert, W.W. and Westcott, V.C. (1972), A method for the study of wear particles in lubricating oil, Wear,

21, 22-42.Zhao, X. and Bhushan, B. (1998), Material removal mechanisms of single-crystal silicon on nanoscale

and at ultralow loads, Wear, 223, 66-78.