Wear 237 2000. 9097 www.elsevier.

comrlocaterwear

Microabrasion of glass the critical role of ridge formation

P.H. Shipway ) , C.J.B. Hodge
Diision of Materials, School of Mechanical, Materials, Manufacturing Engineering and Management, Uniersity of Nottingham, Uniersity Park, Nottingham, NG7 2RD, UK Received 12 May 1999; received in revised form 3 September 1999; accepted 15 September 1999

Abstract Microabrasive wear testing of materials has been the subject of a significant amount of research as its potential for examining the wear behaviour of thin coatings and bulk materials in a sensitive manner has become apparent. The test comprises of rotation of a ball with no translation. against a testpiece with a pool of abrasive slurry surrounding the contact. While the test is elegant and simple, there are some phenomena which may occur during a test which may render the results of such an abrasive test invalid. This paper has examined the role of ridge formation in sodalime glass. Ridges are thin bands of unabraded material which form in the wear scar which, if stable, are worn down only by a sliding action against the ball. It is proposed that the ridges form in the wear scar due to inhomogeneous flow of abrasive slurry. Ridge formation is promoted by high loads and low slurry viscosities, both of which hinder entrainment of abrasive particles between the ball and specimen in the early stages of wear and also hinder entrainment of particles between the ball and any existing ridge. A regime where formation of stable ridges is suppressed has been identified in terms of applied load and slurry viscosity for a range of abrasive slurries. Stable ridge formation results in low wear rates, but when ridge formation is suppressed, wear of glass has been observed to be broadly independent of slurry viscosity and proportional to applied load. It is proposed that ridge formation is a general phenomenon in such tests, and care must be taken to ensure that it is suppressed to allow valid abrasive wear tests to be conducted. q 2000 Elsevier Science S.A. All rights reserved.
Keywords: Glass; Microabrasion; Ridges; Abrasion

1. Introduction The abrasive wear resistance of a material can be measured by a wide range of laboratory tests and the value obtained depends not only on the intrinsic properties of the material but also on the test method employed. The microabrasive wear test has been the subject of a significant amount of research in recent years as it potential for examining the wear behaviour of thin coatings and bulk materials in a sensitive manner has become apparent. A microabrasive wear test was reported by Kassman et al. w1x where a TEM dimple grinder with a crowned wheel was used to create wear scars in materials in the form of a spherical cap. Both the wheel and the sample were rotated around orthogonal axes. to produce wear; the test relied on the axis of rotation of the sample passing through the centre of the crowned wheel which was difficult to achieve with accuracy w2x. The test was developed by Rutherford

) Corresponding author. Tel.: q44-115-951-3760; fax: q44-115-9513741; e-mail: philip.shipway@nottingham.ac.uk

and Hutchings w3x; in their apparatus, a ball was rotated with no translation. while in contact with a stationary specimen with a continuously replenished pool of slurry abrasive surrounding and being drawn into the contact zone between the two. The resulting wear was again in the form of a spherical cap with radius the same as that of the ball. The test has been employed to assess the wear resistance of a wide range of materials including metals, ceramics, polymers and hard and soft coatings with both plane and non-plane geometries w113x. There are many variable parameters associated with the test such as applied load, abrasive type, abrasive concentration, ball type etc. Recent work has demonstrated that wear rate is a function of applied load, but that it does not follow classic Archard behaviour in all cases w11,13x. Furthermore, the type w11x and condition w12x of the ball have been observed to have a strong influence on wear rate and mechanism. Ridges of material in the wear scar have been observed in glass, aluminium, PMMA and steels w1113x; such ridges are thought to influence the manner in which material is removed. This paper examines both the role of these ridges in the wear of sodalime glass and

0043-1648r00r$ - see front matter q 2000 Elsevier Science S.A. All rights reserved. PII: S 0 0 4 3 - 1 6 4 8 9 9 . 0 0 3 0 9 - 9

P.H. Shipway, C.J.B. Hodge r Wear 237 (2000) 9097

91

the test parameters which promote or inhibit their formation. 2. Experimental method Microabrasion testing of glass was performed with a commercially available apparatus, the TE66 Micro-Scale Abrasion Tester Plint and Partners, Wokingham, UK.. A schematic diagram of the apparatus is shown in Fig. 1. The sample is placed in a holder block which is rotated around its pivot until the sample comes into contact with the ball. The ball is rotated about a horizontal axis parallel to the plane of the specimen surface while abrasive slurry is dripped onto the ball and entrained into the gap between the ball and specimen resulting in wear of the specimen. Specimen wear results in an indentation, which generally takes the form of a spherical cap with a geometry similar to that of the ball; a wear test can be interrupted periodically, the sample swung away from the ball and the wear crater dimensions measured with a calibrated eyepiece, from which calculations of the wear volume can be made if the spherical cap geometry of the indentation is assumed. The specimen material used in this work was sodalime glass with a hardness of 614 kg mmy2 ; the balls employed were chrome bearing steel 52100. with a hardness of 862 kg mmy2 . Microabrasion tests were conducted with slurries of SiC grade C5, F1200, approx 4 m m particle size, Washington Mills, Manchester. and Al 2 O 3 White alumina, F1200, approx 4 m m particle size, Abrasive Developments, Solihull. suspended in distilled water. Two batches of SiC, both supplied by Washington Mills were employed; both batches conformed to the companys specifications in terms of composition and particle size distribution. Experiments were conducted with SiC from Batch 1 unless otherwise stated. Various slurry concentrations were employed and are quoted either in terms of mass of

Fig. 2. Wear volume versus sliding distance for sodalime glass abraded with SiC slurry of concentration 552 grl 17.2 vol.%. at 0.5 N.

Fig. 1. Schematic diagram of the microabrasion apparatus.

abrasive per litre of slurry or in terms of a volume fraction of particles in the slurry assuming that the densities of water, Al 2 O 3 and SiC are 1000 kg my3 , 3940 kg my3 and 3217 kg my3 , respectively.. The slurry was delivered to the ball at a rate of approximately 38 m l sy1 . The dynamic viscosity of each slurry was measured with a Bohlin Rheometer fitted with a 30 mm diameter, conically ended cylinder. The nominal shear rate of the slurry during testing was 118 sy1 and the quoted value of viscosity is the average over a 2 minute test. In this configuration, the minimum viscosity that could be measured with accuracy was 10y2 Pa s which placed a limit on the range of slurries for which viscosities could be determined. The counterface balls were all 25.4 mm in diameter and were rotated so that the ball-specimen sliding speed was 0.112 m sy1 this was at the top end of the range examined by Rutherford and Hutchings w4x where they demonstrated that wear rate was independent of sliding speed. and was chosen to minimise the time required for testing. A new ball was used for each test, but the ball was run in for 200 revolutions under a 1 N load with a 24.9 vol.% SiC slurry before being employed in the test to ensure that its surface was reproducible and facilitated particle entrainment into the wear zone w12x. Various loads were applied ranging from 0.25 N to 5 N for a total sliding distance of between 16 m and 160 m depending on the observed wear rate. Some tests were stopped periodically to enable the diameter of the wear scar to be measured and assessed optically. Examination of the wear scars following testing was made by both optical Nikon Optiphot. and secondary electron scanning electron microscopy JEOL 6400.. Profiles of the wear scars were made perpendicular to the sliding direction with a Surfcom profilometer Advanced Metrology Systems, Leicester.. The subsurface regions of the wear scars were exposed for examination by fracturing the glass sample through the wear scar perpendicular to the

92

P.H. Shipway, C.J.B. Hodge r Wear 237 (2000) 9097

3. Results Fig. 2 shows an example of the linear development of wear volume with sliding distance that was generally observed in this work. The gradient of such a graph is the wear rate of the material. On examination of the wear scars, some were found to have ridges running across the scar parallel to the direction of abrasive flow; Fig. 3a

Fig. 3. a. Ridges formed in a wear scar in sodalime glass following abrasion with a SiC abrasive slurry. Abrasive movement was from right to left. b. Profile across the wear scar perpendicular to the sliding direction solid line.. The dashed line shows the predicted curve for a 12.7 mm radius ball for comparison.

direction of abrasive flow. To achieve this, a line was deeply scored with a diamond scribe on the glass sample surface on either side of the wear scar and the samples fractured along this line. To enable the volume of the wear scar to be calculated, spherical cap geometry of the wear scar was assumed as has been demonstrated to be the case w4x.. The volume, V , is related to the depth, d, of the wear scar by the following equation: d Vsp d2 ry 1. 3

where r is the radius of the cap. The depth, d, of the cap can be calculated from the optical measurements of the scar radius, a, with the following equation: d s r y 'r 2 y a2

2.

Fig. 4. a. Detail of ridge formed in wear scar in sodalime glass showing ridge terminating in the centre of the scar. b. Detail of fracture across ridge demonstrating that it is continuous with the bulk. c. Ridge close to the trailing edge of the wear scar showing a smooth transition between the ridge and the unworn area.

P.H. Shipway, C.J.B. Hodge r Wear 237 (2000) 9097

93

shows such a wear scar in the glass with ridges present. A number of short ridges can be seen at the trailing edge of the scar, with a few long ridges which stretch some way across the scar. A profile across the wear scar, perpendicular to the direction of abrasive flow, is shown in Fig. 3b along with a theoretical profile for a scar showing the form of a spherical cap with the same radius. It can be seen that the ridges sit proud of the surface by up to about 5 m m but that despite their presence the rest of the wear scar closely conforms to the spherical cap geometry expected. Fig. 4 shows details of the ridges observed on the wear scar in Fig. 3a. Each ridge stretches from the trailing edge of the wear scar to some point within the body of the wear scar. The top surfaces of the ridges exhibit a far smoother morphology than that of the surrounding material. Fig. 4a shows the area where one of the ridges stops in the middle of the wear scar, showing that there is no build up of abrasive at this point as has been reported elsewhere w12x. This particular sample has been fractured to reveal the

Fig. 6. Wear rate as a function of load for abrasion of sodalime glass with a slurry of concentration 443 g SiCrl 13.8 vol.%. indicating R. where ridges have formed.

regions below the wear scar and ridge and a detailed view of the region below the ridge is shown in Fig. 4b where it can be observed that this material is composed solely of unworn glass. Fig. 4c shows the region of the ridge close to the trailing edge of the wear scar where the smooth transition between the unworn region and the ridge can be observed. The smooth surface of the ridge indicates that no abrasive particles are entrained between the ridge and the ball and that the ball slides against the ridge to wear it down. Fig. 5 illustrates such behaviour diagrammatically. Fig. 6 shows the variation in wear rate with load for a given SiC slurry concentration of 443 g ly1 13.8 vol.%.; alongside each data point is an indication whether or not

Fig. 5. a. Schematic plan view of wear scar with stable ridges diverting the abrasive flow around them. b. Schematic cross-sectional view of wear scar showing ridge at the trailing edge supporting the ball. Abrasive does not become entrained between the ball and the ridge and is diverted out of the plane of the section see a...

Fig. 7. Wear rate versus SiC slurry concentration for abrasion of sodalime glass under 1 N applied load. R indicates where ridges have formed in the wear scars.

94

P.H. Shipway, C.J.B. Hodge r Wear 237 (2000) 9097

Fig. 8. Wear rate versus SiC slurry concentration for abrasion of sodalime glass under a range of applied loads.

Fig. 10. Wear rate of sodalime glass as a function of load for slurries of 24.9 vol.% solids. R indicates that ridges have formed in the wear scars.

ridges were observed in the wear scar. At loads of 1 N and below, the wear rate increases in proportion to the load as predicted by the Archard wear equation and no ridges are observed. However, a dramatic decrease in wear rate is observed as the load is increased above 1 N and this transition is accompanied by the presence of ridges in the wear scar. Fig. 7 shows the variation in wear rate with SiC slurry concentration for a given applied load of 1.0 N; again, alongside each data point is an indication of whether or not ridges were observed in the wear scar. At low slurry concentrations, the wear rates were low but slowly increased as the concentration increased. At these low loads, ridges were observed in the wear scars. Above a certain SiC slurry concentration in this case 291 g ly1 ., the wear

rate increased dramatically by more than 5 times. and ridges were no longer observed. Fig. 8 shows the wear rate as a function of SiC particle concentration for a range of loads. For all loads, the transition between low rate wear ridges present. and high rate wear ridges absent. can be observed. The concentration at which the transition occurs becomes higher as the applied load increases. Fig. 9 shows a map of load and concentration indicating two regimes, where ridges do and do not form in the scar. Fig. 10 shows the wear rate of glass as a function of load for slurries of 24.9 vol.% abrasive of both of the batches of SiC 801 g ly1 . and Al 2 O 3 981 g ly1 ., again with an indication of whether or not ridges formed in the wear scar. The abrasive slurries acted very differently

Fig. 9. Map of tendency to form ridges as a function of applied load and slurry concentration for SiC slurries batch 1.. Filled symbols indicate ridge formation; open symbols indicate no ridge formation.

Fig. 11. Slurry viscosity as a function of solids concentration for various abrasives.

P.H. Shipway, C.J.B. Hodge r Wear 237 (2000) 9097

95

Fig. 12. Map of tendency to form ridges as a function of applied load and slurry viscosity. Filled symbols indicate ridge formation; open symbols indicate no ridge formation. Different symbols represent the different abrasives.

which was attributed to their viscosity. A plot of slurry viscosity versus abrasive concentration for the three abrasives is shown in Fig. 11, while Fig. 12 shows a map of load and slurry viscosity indicating two regimes, where ridges do and do not form in the scar for all three abrasive types examined.

4. Discussion Fig. 2 demonstrates that wear volume increases linearly with sliding distance in this system and as such fits with the general models of abrasion. However, the rate of wear depends critically on the presence or absence of ridges which may form in the wear scar as shown in Fig. 3a. These ridges are areas of the glass which protrude above the general background of the wear scar as can be seen in the profilometry trace of Fig. 3b. Their smooth surface Fig. 4. indicates that they have not been abraded in the same manner as the surrounding material which exhibits coarse, indentation induced fracture and as such, it may be inferred that no abrasive particles have passed between the ridge surface and the ball. It is noticeable that the ridges do not alter the background profile of the wear crater in that it is still that of a spherical cap as can be seen from the close fit of the theoretical and measured shapes of the wear scar Fig. 3b.. Fig. 4 shows one of these ridges in detail. The ridge ends in the body of the wear scar as shown in Fig. 4a. The ridge material is continuous with the bulk of the glass specimen, as can be observed by its subsurface structure Fig. 4b. indicating that it has not been deposited in the wear scar but is a region of the original glass

substrate that has not been worn as much as the material which surrounds it. All ridges are linked to the trailing edge of the wear scar. In an attempt to ascertain the mechanism of formation of the ridges, further experiments were performed. Experiments where the ball and specimen are separated, observations made and the experiment then recommenced are not suitable as even slight misalignments on the replacement of the ball may affect the results in a way which would be difficult to determine. Thus an experiments were conducted with slurries of two different concentrations being used in sequence with no breaking of the specimen-ball contact or stopping of the rotation. A concentrated SiC slurry 801 g ly1 . and a dilute SiC slurry 78 g ly1 . were employed under a 1 N load. When used alone, the former never produced ridges in wear scars under these conditions whereas the latter always produced ridges. On swapping the slurries, the ball was wiped during motion with a cloth in an attempt to make the transition in concentrations in the wearing region as rapid as possible. A scar was produced with the concentrated slurry sliding distance ; 0.8 m. and then developed with the dilute slurry further sliding distances of 0.8 m and 15 m.. Following 0.8 m of sliding with the dilute slurry no ridges could be observed indicating that it was more difficult for a ridge to form in an existing ridgeless crater. This implies that ridges tend to form in the initial stages and their formation is a function of the ease with which slurry is entrained into the ballspecimen contact as argued by Trezona et al. w13x. However, following 15 m of sliding with the dilute slurry, ridges were again observed which tended to form away from the centre line of the wear scar. These ridges were forming in the region of the scar which had begun to wear within the period of sliding with the dilute slurry i.e., outside the region which was worn by the concentrated slurry. This again supports the view that ridges form as a result of poor entrainment of slurry between the unworn specimen and the ball. Further tests were performed employing the dilute slurry first, followed by the more concentrated slurry. 1.6 m of sliding with the more concentrated slurry removed all traces of the ridges which must have been present when the slurries were swapped over. This indicates that even if ridges do form in the very early stages of normal tests with the concentrated slurry, they will be quickly removed. Thus, in accordance with the findings of Allsopp et al. w12,13x, it is proposed that the ridges generally form during the early stages of development of the scar. Once a ridge has formed, it will divert the abrasive slurry flow around it see Fig. 5a. and thus the ridge will be stable. Allsopp et al. w12x suggest that in soft materials, abrasive particles may embed into the test specimen within the wear zone and thus effectively block the action of the abrasive. No particles embedded ahead of the ridge have been observed in this work Fig. 4a. and thus this mechanism is discounted.

96

P.H. Shipway, C.J.B. Hodge r Wear 237 (2000) 9097

The presence of stable ridges within the wear scar affects the subsequent wear behaviour of the specimen. It is proposed that the ridges support the load of the ball but no abrasive is entrained between the ball and the ridge Fig. 5b.; the material in the ridge thus wears by the action of the sliding of the roughened ball across it. The presence of the ridges will result in a gap between the ball surface and the glass surface wearing by abrasion and when that gap becomes larger than the abrasive particle size, wear of the glass surface will decrease dramatically until the ridge wears down and allows wear of the rest of the crater by the abrasive slurry to proceed. This is consistent with the observation that the ridges tend to be not more than 5 m m above the background wear surface similar to the abrasive particle size.. Assuming that wear of the ridges by sliding against the roughened ball is slow compared to general wear of the scar by abrasion, it is proposed that the overall wear rate will now be controlled by the wear rate of the ridges where these ridges exist. As noted by Trezona et al. w13x, the tendency for ridges to form is a function of the applied load and slurry concentration. Fig. 6 demonstrates that for a given abrasive slurry concentration, wear rate is proportional to applied load for loads up to a critical value, whereupon ridge formation becomes favourable and the wear rate decreases as previously discussed. Ridge formation tends to occur when there is difficulty in entraining slurry between the ball and unworn specimen, which is generally most severe in the initial stages of the test. Low loads and high slurry concentrations will tend to promote slurry entrainment and thus reduce the tendency for ribbon formation. If a ridge does develop, similar conditions of low loads and high slurry concentration will promote its annihilation. Higher applied loads will result in the abrasive particles failing to be entrained between the ridge and the ball; the ridge thus diverts the slurry flow around it Fig. 5a. and is stable, thus resulting in lower wear rates Fig. 6.. For a given applied load, low slurry concentrations will result in low wear rates with ridge formation Fig. 7.. Above a given concentration, the wear rate increases dramatically, accompanied by an absence of any ridges within the wear scar. At low slurry concentrations, a ridge may be stable within the wear scar; the low slurry concentration enables the abrasive particles to be easily diverted around the ridge since the large interparticle distance means that movement of particles within the slurry is less constrained. A higher slurry concentration will tend to annihilate ridges as soon as they form as the slurry is not so easily diverted around them. Ridge formation in the wear scar makes the test for abrasive wear resistance invalid since it is now the wear of the ridges by sliding against the ball that controls the overall wear behaviour. Ridge formation is promoted by high loads and low slurry concentrations. Fig. 8 shows wear rate against slurry concentration for a range of applied loads. The transition from low rate wear ridge

dominated. to high rate wear ridges absent. occurs at higher concentrations as the load is increased. Fig. 8 also demonstrates that once ridge formation has been suppressed by exceeding the critical slurry concentration for a given load, wear rate is independent of slurry concentration as is observed in the classical models of abrasive wear. To conduct a valid wear test, ridge formation in the wear scar must be suppressed. It has been observed that ridge formation is a function of load and abrasive concentration. Fig. 9 maps the tendency to form ridges as a function of these two parameters. The regime labelled "No ridge formation" covers the range of experimental parameters over which valid abrasive wear tests may be conducted for this abrasive particle type SiC, batch 1.. A second batch of SiC and alumina abrasives were employed to assess the broad applicability of the regimes identified in Fig. 9. Wear tests at a range of applied loads and slurry concentrations were performed. Fig. 10 shows the wear rates and tendency for ridge formation for slurries formed from the two batches of SiC abrasive and the alumina abrasive, all at 24.9 vol.% solids loading. The first batch of SiC suppresses ridge formation up to 4 N applied load, whereas ridge formation occurs above 2 N for the second batch of the SiC slurry and above 0.5 N for the alumina slurry. It is notable that where ridge formation has been suppressed, the wear rate of the glass at a given load is very similar for all the abrasive slurries. The particle size distributions and solids loadings of the various slurries are similar as is the angularity of the particles. Although covering a range of hardness, all the abrasive are far harder than the glass. These similarities account for the similar wear rates observed when ridge formation has been suppressed. However, it is proposed that the most significant difference between these slurries lies predominantly in their viscosities and it is this which affects the tendency for ridges to form in the wear scar. Fig. 11 shows the viscosities of the three slurries as a function of particle volume fraction. The alumina and silicon carbide slurries are expected to differ significantly in viscosity due to differences between types of bonding in the two abrasive materials, namely predominantly covalent bonding in silicon carbide with more ionic character in the alumina. Such difference in bonding affects the tendency to promote electrostatic repulsion between particles and thus inhibit flocculation within the slurry. The differences between the two batches of silicon carbide that results in the significantly different viscosities of the slurries has not yet been identified. Many of the arguments put forward to explain the load and concentration dependence of ridge formation Fig. 9. depended upon the ease by which abrasive particles in the slurry could be entrained into the wear scar andror diverted around a ridge which had formed in the wear scar. Fig. 12 shows a map of similar form to that presented in Fig. 9. However, the regimes in Fig. 12 are defined in

P.H. Shipway, C.J.B. Hodge r Wear 237 (2000) 9097

97

terms of applied load and slurry iscosity rather than slurry concentration. It can be seen that in this form, the tendency for ridge formation in sodalime glass can be generally described in terms of load and viscosity for all the three abrasives employed, despite their widely differing slurry characteristics.

loads. However, while a regime of slurry concentration and applied load has been determined to allow the abrasive wear of sodalime glass to occur without the formation of ridges, we have no information as to the direct transferability of these parameters for the tendency for ridge formation during wear of other materials.

5. Conclusions The microabrasive wear test has been applied to the wear of sodalime glass. It has been demonstrated that the wear behaviour of the glass is strongly dependent on whether or not ridges form in the wear scar. Under certain experimental conditions, these ridges tend to form in the wear scar due to abrasive particles failing to wear the glass surface evenly. Once formed, they support the load of the ball and control the overall wear rate of the material. The ridges form in the scar at high applied loads and low slurry viscosities and a regime, in terms of load and viscosity for three very different abrasive slurries., has been identified where the tendency for ridge formation is high. Ridge formation in the wear scar is associated with low wear rates. For this test to be employed as an abrasive wear test, it must be used in the regime where there is no tendency for ridges to form. Under these conditions, the wear of sodalime glass has been observed to be broadly independent of particle concentration in the slurry or viscosity. and to be proportional to load as predicted by the Archard equation. Although work presented here is limited to wear of sodalime glass, it is supposed that ridge formation is a general phenomenon since it has also been observed in ductile aluminium, steel and PMMA. To allow this test to be employed to its full potential, it must not be allowed to fall into the disrepute associated with apparently random data being generated by its use. Care must therefore be taken to ensure that all meaningful wear data are taken from scars which exhibit no ridge formation. In general terms, this calls for high slurry viscosities and low applied

References
w1x A. Kassman, S. Jacobson, L. Erickson, P. Hedenqvist, M. Olsson, A new test method for the intrinsic abrasion resistance of thin coatings, Surf. Coat. Technol. 50 1991. 7584. w2x R. Gahlin, M. Larsson, P. Hedenqvist, S. Jacobson, S. Hogmark, The crater grinder method as a means for coating wear evaluation an update, Surf. Coat. Technol. 90 1997. 107114. w3x K.L. Rutherford, I.M. Hutchings, A micro-abrasive wear test, with particular application to coated systems, Surf. Coat. Technol. 79 1996. 231239. w4x K.L. Rutherford, I.M. Hutchings, Theory and application of a micro-scale abrasive wear test, J. Test. Eval. 25 1997. 250260. w5x N. Axen, S. Jacobson, S. Hogmark, Principles for the tribological evaluation of intrinsic coating properties, Wear 203204 1997. 637641. w6x M. Choudhury, I.M. Hutchings, The effects of irradiation and ageing on the abrasive wear resistance of ultra high molecular weight polyethylene, Wear 203240 1997. 335340. w7x K.L. Rutherford, I.M. Hutchings, Micro-scale abrasive wear testing of PVD coatings on curved substrates, Tribol. Lett. 2 1996. 111. w8x N. Axen, S. Jacobson, S. Hogmark, Influence of hardness of the counterbody in three-body abrasive wear an overlooked hardness effect, Tribol. Int. 27 1994. 233241. w9x M.H. Staia, C.E. Enriquez, E.S. Puchi, D.B. Lewis, M. Jeandin, Application of ball cratering method to study abrasive wear, Surface Engineering 14 1998. 4954. w10x M.H. Staia, C. Enriquez, E.S. Puchi, Influence of heat treatment on the abrasive wear resistance of electroless NiP, Surf. Coat. Technol. 9495 1997. 543548. w11x P.H. Shipway, The role of test conditions on the micro-abrasive wear behaviour of sodalime glass, Wear 233235 1999. 191199. w12x D.N. Allsopp, R.I. Trezona, I.M. Hutchings, The effects of ball surface condition in the micro-scale abrasive wear test, Tribol. Lett. 5 1998. 259264. w13x R.I. Trezona, D.N. Allsopp, I.M. Hutchings, Transitions between two-body and three-body abrasive wear: influence of test conditions in the microscale abrasive wear test, Wear 225229 1999. 205214.