Measuring Evolution of Transfer Film–Substrate Interface Using Low Wear Alumina PTFE
Polymeric solid lubricants lay down their own wear debris onto hard metallic counterfaces to form a protective transfer film which reduces friction and wear effectively without lubrication. Adhesive shear strength at the hidden interface between the film and substrate determines the film persistence and correlates with system wear qualitatively. Previous studies showed that an ultralow wear (k ~ 10− 7 mm3/Nm) alumina-PTFE solid lubricant forms an extremely adherent and complete transfer film, and strong chemical bonds between wear debris and counterface perpetuate the film–substrate adhesion very early in the sliding. In this paper, we aimed to test the permanence of such adhesion by removing pre-developed transfer films using sliding rubber contact and measuring the topographical evolution of the interface throughout the course of a standard wear test using the well-studied alumina-PTFE system. The results unexpectedly showed continuous wear of the counterface across the wear track, and counterface wear rate decreased proportionally from 3 × 10− 7 to 3 × 10− 8 mm3/Nm with increased film area fraction and sliding distance. A proposed rule-of-mixtures wear model coincided closely with the experimental results and strongly suggested a coupled mechanism of adhesive and fatigue wear of the counterface. The upper limit of the interfacial counterface fatigue wear rate was predicted to be 3 × 10− 8 mm3/Nm.
KeywordsAlumina-PTFE Transfer film–substrate interface Counterface abrasion Wear
When slid against hard metallic counterfaces, polymeric solid lubricants sacrifice debris to the counterface in the form of a transfer film which separates the sliding bodies and reduces friction and wear. Normally, polymers wear much faster than the counterfaces and wear rates of the two are about inversely proportional to their hardness ratio as first reported by Archard . Polytetrafluorethylene (PTFE), for example, has a prohibitively high wear rate against steel and 10–40 wt% micro-sized fillers (e.g., glass fiber, metal oxides) are often used to reduce wear by up to 99.99% [2, 3]. The high content hard fillers tend to accumulate at the sliding surface and are supposed to preferentially support the normal load and reduce wear . However, they also tend to pierce the otherwise lubricious PTFE-rich transfer film and cause increased friction and counterface abrasion .
The fact that micro-fillers often reduce polymer wear at the sacrifice of increased counterface abrasion and deteriorated transfer films lead tribologists to seek nanosized fillers as alternatives since they are of the same scale of roughness asperities and much less abrasive to the counterface. Starting from 2000, Li et al. , Chen et al. , and Sawyer et al.  found that certain nanofillers could reduce polymer wear as good as micro-fillers without evident counterface abrasion. Burris and Sawyer [9, 10] and many [11, 12] later demonstrated that nanofillers like alumina could reduce PTFE wear up to 30,000 × with as little as 0.2–2% loadings. The accompanying transfer films in these systems were extremely thin, continuous, complete, and adherent [13, 14]. Krick et al.  surveyed alpha-phase alumina fillers from a range of different vendors and found that the most effective wear reducing alumina fillers in PTFE are those which form porous and friable micro-sized agglomerates that get broken down into nanosized fragments at the sliding surface. The accumulation of such nanosized filler fragments mechanically reinforces the composite’s running surface and lead to a dramatic reduction in wear.
The most crucial part of transfer film quality is the adhesion strength. Bahadur and Tabor  slid PTFE against pre-developed transfer films of low wear PTFE composites and found the transfer films immediately removed and wear of PTFE unaffected by the existing films. They suggested debris bonded weakly to counterface asperities through mechanical interlocking and transfer films were instantaneously removed and replenished during sliding. Briscoe , Bahadur and Gong , and Schwartz and Bahadur  suggested that transfer film adhesion was of chemical nature and system wear rate decreased with increased adhesion strength.
A famous ultralow wear alumina PTFE solid lubricant provides probably the best opportunity for transfer film studies to date. The system deposits nanoscale debris so strongly adherent to counterfaces that film removal is a rare event and transfer film grows thicker and more uniform continuously during sliding . Film adhesion strength is ~ 15 × greater than in unfilled PTFE  and transfer film removal rate during sliding is on the order of 10− 9 mm3/Nm, a value of ~ 1% of the composite’s wear rate and similar to wear rate of commercial wear-resistant coatings. Harris et al.  recently proved the combination of ambient air and dry sliding induces degradations of PTFE chains which chemically bond the polymer to the metallic counterface and alumina fillers to form the extremely adherent transfer film and a wear-resistant running film at the bulk’s running surface. This is a significant breakthrough in recent transfer film studies and further confirms the importance of tribochemistry in transfer film adhesion.
There is little doubt strong adhesion is crucial for high-quality transfer film formation, yet it remains unclear how long such adherence could last. It is almost intuitive to assume friction and wear only occur or initiate at the polymer-on-polymer interface between the bulk’s running surface and the transfer film top surface once a persistent film is formed and low wear reached. In other words, transfer film adhesion is permanent unless wear of film reaches the interface or delamination occurs. In this study, we aim to test this hypothesis by removing pre-developed transfer films and measuring the evolution of the film–substrate interface throughout the course of a wear test. The goal is to shed more light on the relations between transfer film adhesion, counterface abrasion, and overall wear reduction. A well-studied alumina-PTFE nanocomposite is chosen for its ultralow wear rate and extremely persistent transfer film as previously documented in detail [5, 9, 10, 13, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32]. Ultralow wear is defined in this study as having wear rates below 10− 6 mm3/Nm, a definition for wear-resistant PTFE composites suggested by Harris et al. .
2 Materials and Methods
2.1 Composite and Counterface Preparation
The polymeric solid lubricant used in this study was a 5 wt% α-Al2O3 PTFE nanocomposite. The composition and preparation method have been described by the same authors [31, 32] and are comparable to those used in many studies [9, 22, 25, 26, 27, 28, 29, 33]. The PTFE powder was a DuPont Teflon™ 7C molding resin with a reported average particle size of 40 µm. Alpha-phase alumina nanoparticles were obtained from Shanghai Maikun Chemical Co., Ltd (99.99%, product link: https://goo.gl/RR5QfV) with a reported average particle size of 30 nm measured with scanning electron microscopy (SEM) as illustrated in Fig. 1a. These nanoparticles tend to form micro-sized agglomerates likely due to their manufacturing process and high surface-to-volume ratio. Krick et al. did a thorough study on wear reducing alpha-phase alumina nanoparticles in PTFE and found a particular effective alumina filler with vendor-specified particle size of 27–43 nm (Nanostructured and Amorphous Materials, Stock #1015WW) formed similar micro-agglomerates which were porous and nanostructured under SEM inspections . Based on this finding, it is probably inaccurate to describe the alumina filler used in this study as nanosized. However, the same authors suggested that, unlike traditional micro-sized dense alumina particles, the micro-agglomerates as shown in Fig. 1a are easy to get broken up into nanosized filler fragments at the sliding surface and can reduce PTFE wear significantly through a coupled mechanical and tribochemical mechanism .
The two powders were first hand-mixed in a polyethylene container with a 5:95 wt% ratio. The mixture was added into anhydrous ethanol with a 1:2 volume ratio and dispersed using a 500 W ultrasonic horn with a 25.8-mm-diameter titanium tip; full power was pulsed on for 5 s and off for 3 s over 3 min to improve dispersion state. The mixture was drained through a lab filter paper with 8 µm pore size (grade 40, diameter 90 mm, Whatman® by Sigma–Aldrich) for 20 min to remove the bulk ethanol and vacuum dried at 90 °C for 30 min. The dry powder was cold pressed using a Φ10 × 20 mm cylindrical mold and heated to 365 at 120 °C per hour, held for 3 h, and cooled at the same rate. The sintered samples were machined into 6.4 × 6.4 × 15 mm pins prior to the wear test.
Counterface used in this study was a 120 × 30 × 3 mm grade 304 stainless steel plate pre-ground and polished to ~ 50 nm Ra using a commercial automatic fine-grinding machine (Fangda FD-24LP, China). The finished surface had no obvious directionality to the roughness. Prior to the wear test, counterfaces were ultrasonicated with acetone for 10 min, cleansed in distilled water, and air dried for 30 min.
2.2 Wear and Friction Testing
Wear test was performed in ambient air condition (20 °C, 40% RH) using a custom-built linear reciprocating tribometer as previously described in detail . The sliding conditions were selected to best match those of comparable studies [5, 13, 15, 22, 25, 26, 27, 28, 29, 33] with a normal load of 250 N, contact pressure of 6.5 MPa and sliding speed of 50 mm/s. A ‘stripe test’ was used following Harris et al. [22, 28, 30] by decreasing the reciprocating length from 90 to 20 with 10-mm intervals intermittently at predetermined sliding cycles of 1, 2, 5, 10, 20, 50, 100, and 200 k as illustrated in Fig. 1. Such test configuration preserved areas of early stage transfer films and provided the opportunities for further analyses of transfer film development. An unfilled PTFE pin was selected as the control sample and slid against the same steel counterface under identical test conditions (FN, P, V) except with a fixed stroke length of 25 mm. The control group test was stopped at 10 k sliding cycles (500 m) due to excessive wear of the pin.
2.3 Transfer Film Characterization
The difficulty of transfer film morphology characterization lies in the separation of the film from counterface. Scanning electron microscopy (SEM) was mostly used in literature as it is sensitive to surface chemistry despite its insensitivity to surface topography. Contact profilometry and white-light interferometry are good at topography characterization but poor at polymer film detection. Optical microscopy was used in this study as it provides the best contrast between transfer film and counterface as illustrated in Fig. 2. The composite used in this study has been known to produce a thin and transparent transfer film that manifests colorful interference fringes under a broadband-light-source equipped optical microscope [20, 21, 23, 25, 26]. Transfer film thickness could be determined from the fringe pattern which is basically a height contour map of the film. The film thickness (e) for a certain color fringe is a function of the interference order, k:
This is well illustrated in Fig. 2b where the maximum film thickness reaches the measuring limit and higher orders of interference fringe became indiscernible. Figure 3c shows an example of the thinnest transfer film discernable using this method (tp ~ 74 nm) and a film thickening process observed and quantified using this method.
Whereas the white regions between the island-like domains in Fig. 2 appear to be film-free, there likely exist ultra-thin transfers of aligned polymer chains in these regions beyond the detection limit of the current microscopy. Makinson and Tabor first reported that PTFE always draws fibrous transfers of about 10–40 nm thickness during sliding . Harris et al.  provided a physical framework for PTFE chain scission and transfer which suggests a single PTFE chain can get broken down into a few nanometers long filaments and transferred to the sliding counterface from pure van der Walls attractions. However, such ultra-thin films are weakly correlated with wear mode of the polymer [20, 37, 38]. A typical example is the unfilled PTFE which is a standard high wear polymer and generates a thin and continuous film underneath the usually reported thick and patchy transfer film . Based on this reasoning and the accumulation of hard fillers at the bulk’s running surface , we hypothesize that such film neither provides wear protection for the bulk composite nor the counterface. For this reason, only optically detectable and likely debris-formed regions of transfer film as shown in Fig. 2 are included in transfer film analysis in this study. A pixel intensity thresholding technique following Bhimaraj et al.  and Ye et al.  using Photoshop™ was used to convert the image into a binary map (e.g., black and white image insets in Fig. 1d) from which the transfer film area fraction, X, defined as the ratio between film-covered area and total surface area, was calculated.
2.4 Transfer Film Removal and Surface Profilometry Measurements
The primary aim of this study is to measure the evolution of transfer film–substrate interface throughout the course of a wear test. To expose the hidden interface, transfer films at different stages of development were slid against a 6-mm diameter fluoroelastomer rubber ball (E = 4.5–10.5 MPa) across the wear track width as shown in Fig. 1b. Experiments were conducted using a custom-built microtribometer equipped with in situ contact imaging capability (see Appendix for details). The sliding conditions were 3 N normal load, 2.65 MPa contact pressure, 10 mm/s sliding speed, and 10 mm reciprocating length. This method is based on a previous finding that these films adhere extremely to rubber even at very low contact pressures (e.g., 0.7 MPa) and shear stresses (e.g., 0.6 MPa), and can be easily removed by sliding against rubber, the details of which could be found in Ref.  and the Appendix. All microtribometry tests were run for 200–1000 cycles to ensure complete removal of the transfer film and an example result was shown in Fig. 1b. Exposed counterface areas were cleaned with a lint-free wipe.
3.1 Wear and Frictional Behavior of the Composite
During sliding, the composite material used in this study is known to undergo a high wear run-in period before transitioning into low wear steady-state sliding. After the first 180 m, the composite pin lost 0.37 mm3 of material, whereas it barely lost any material between 180 and 340 m as illustrated in Fig. 3a. This abrupt transition is related to the formation of a protective transfer film during the run-in and was first reported in detail by Ye et al. for this material . Previous measurements found that the exact low wear transition point occurs at about 10–20 m of sliding distance under the same sliding speed and contact pressure . For this reason, in situ wear rate at D = 180 m was computed as the slope of the wear data between 180 and 340 m using the pre-described method and had a value of 1.43 ± 6.35 × 10− 7 mm3/Nm as shown in Fig. 5a. This is the only case where the linear regression analysis window is not centered at the point of interest since it is inappropriate to use run-in data to calculate wear rate after the transition point. Wear of the composite increased gradually after the transition point and stabilized at a wear rate of 7 × 10− 7 mm3/Nm as shown in Fig. 5a, a value slightly higher than reported wear rates of the system (k ~ 1–3 × 10− 7 mm3/Nm) with fixed stroke length experiments [5, 9, 10, 13, 25, 26].
Friction coefficient remained stable throughout the test despite the changes in stroke length and reciprocating frequency. The average friction coefficient was 0.189 ± 0.011 with a 95% confidence interval and the friction coefficient decreased slightly between 1000 and 2000 sliding cycles. This is in accordance with a previous measurement reporting the material had the lowest friction right after the transition point .
3.2 Wear and Topographical Evolution of the Film–Substrate Interface
The original lapped counterface had a mean roughness of Ra = 50 nm and Rsk = 0.05. As illustrated in Fig. 4a, after the first 1000 sliding cycles, visible scratches started to scatter across the film–substrate interface resulting in an increased roughness of Ra = 150 nm and decreased skewness of Rsk = − 2.5. The average abrasion depth was δ = 0.1 µm. At 10 k cycles, the interface was ubiquitously abraded across the track width and average abrasion depth increased to δ = 0.3 µm. Roughness increased to Ra = 300 nm and skewness increased to Rsk = − 1.2. Counterface abrasion increased with increased sliding cycles, and both roughness and abrasion depth increased with increased sliding cycles as shown in Fig. 4b, c, respectively. At 200 k cycles, the original steel surface was worn down 0.8 µm on average by the sliding pin and had a roughness of Ra = 445 nm and Rsk = − 1.0.
Wear rates of the composite pin and the film–substrate interface were plotted against sliding cycles in Fig. 5a. While the pin’s wear rate increased slowly and gradually during the test, wear rate of the counterface decreased continuously from 3.0 × 10− 7 mm3/Nm at 1 k cycles to 2.6 × 10− 8 mm3/Nm at 200 k cycles. The monotonic trend of counterface wear rate inversely resembled the evolution of transfer film area fraction as shown in Fig. 5b. At 1 k cycles, the transfer film was composed of sparsely distributed 10–30 µm diameter islands of adherent transfer with a total film area fraction of 33%. At 10 k cycles, the islands grew and merged with each other resulting in an increased area fraction of 71%. Further sliding continued to increase film coverage which saturated at ~ 80% as best illustrated in Fig. 5c. Single point wear rates of an unfilled PTFE pin and the corresponding steel counterface were 4.94 ± 0.24 × 10− 4 and 7.47 ± 6.40 × 10− 9 mm3/Nm, respectively, as shown in Fig. 5a. Notice the counterface wear rate against the composite is only 3 × higher than against pure PTFE during steady-state sliding.
The slightly higher than expected wear rate of the bulk composite (7 × 10− 7 mm3/Nm) requires a few comments. The same composite pin used in this study was tested by the authors under the same sliding conditions but with a fixed stroke length (e.g., [31, 32]) and had wear rates of 1.5–3 × 10− 7 mm3/Nm, values typical to ultralow wear alumina PTFE as reported in many comparable studies [9, 22, 25, 26, 27, 28, 29, 33]. In two separate works, Harris et al. and their group reported a 0.8 × 10− 7 mm3/Nm wear rate  and a 4 × 10− 7 mm3/Nm wear rate  for the same alumina PTFE composite both run for a million cycles under identical load, speed, and roughness conditions. The first study used a fixed 25.4-mm stroke length while the second study used a similar ‘stripe’ test configuration and the stroke length decreased intermittently from 88.9 to 27.9 mm. This wear rate discrepancy was never explained but was confirmed by data in this work and two previous publications [31, 32]. We believe stroke length and reciprocating frequency have direct effects on the wear of alumina PTFE system which are beyond the scope of this work.
In the work by Harris et al. , the composite wear rate reached a lowest value of 2 × 10− 7 mm3/Nm at 100 k cycles and increased to 4 × 10− 7 mm3/Nm at 1 M cycles. In this study, the composite reached a lowest wear rate of 1.43 × 10− 7 mm3/Nm at 1 k cycles and the wear rate increased to 7 × 10− 7 mm3/Nm at 200 k cycles. Confounding factors for the wear rate discrepancy between this work and Harris et al.’s include the differences in alumina filler, dispersion and sample preparation, etc. However, transfer film produced in this study were visually indistinguishable from those reported in [20, 21, 23, 31, 32] and extremely persistent during sliding. Contact profilometry was used to acquire the topography of the transfer film at 200 k cycles which is compared with the original surface and the film-erased topography in Fig. 6. Notice the transfer film has more volume above the original counterface than below it. This has some interesting implications for transfer film thickness evaluation in tribology literature. However, this is also beyond the scope of this work and will require a standalone study.
To the authors’ best knowledge, this study is the first to measure the topographical evolution of the transfer film–substrate interface in a tribological polymer system. The primary aim of this study is to test the hypothesis that strong adhesion between transfer film and substrate protects the interface from wear and eliminates counterface abrasion locally. For this reason, a well-studied alumina PTFE system was chosen for its extremely low wear and persistent transfer film. Blanchet et al.  suggested that low wear PTFE composite transfer films developed by gradually filling in remaining areas of exposed counterface. Ye et al.  used in situ optical microscopy to show that transfer film formed by filling in such ‘free-space’ and thickening simultaneously. Optical microscopy measurements in this study confirmed a gradual filling-in process of transfer film development as illustrated in Fig. 5c. However, one evidence against the permanent adhesion theory is the 2D line scans in Fig. 4 showing abrasion of the counterface ubiquitously across the surface in spite of the high transfer film area fraction. A quick way to test the hypothesis is to compare the interface profiles between two different sliding cycles and calculate the range of incremental wear as illustrated in Fig. 7. Here, X1 and X2 are transfer film area fractions at sliding cycle N1 and N2 (N1 < N2), respectively. Mean counterface adhesive wear rate at film uncovered regions is denoted by ka and fatigue wear rate at the film–substrate interface by kf. Adhesive wear of the counterface causes increase in relative depth of film uncovered regions as illustrated in Fig. 7 (d2 > d1). The interface profile at cycle N1 is subtracted from the profile at N2 cycle, and Δ denotes regions where profile 2 is lower than profile 1. Accepting the permanent adhesion theory means kf = 0, counterface wear only occurs at film-free regions, and the subtraction data should be negative at such locations. More importantly, the area fraction of regions with negative differences over the whole track width, \(\sum \varDelta \%\), should be no greater than the film-free region percentage 1 − X1, assuming X1 ≤ X2. The result of a quick check using average 2D line scans from 10 and 100 k sliding cycles is shown in Fig. 8. Within the wear track, the area fraction of further counterface abrasion is 0.74 ± 0.25 with 90% confidence using the student t-statistics. This is much higher than the 0.29 ± 0.14 area fraction of film-free region at 10 k cycles, which suggests that film–substrate interface fatigue wear contributes significantly to the total counterface wear.
Another strong evidence suggesting the existence of counterface fatigue wear in film-covered regions is a unique pitted morphology of the film-erased counterface as best illustrated in Fig. 9. After 1 k cycles, only minor scratch marks along the sliding direction were visible at the counterface. After 50 k cycles, rows of wear pits along the sliding direction with average depth and length of 1 and 200 µm were discernable. After 200 k cycles, average pit depth remained unchanged while average pit length increased to ~ 700 µm. We believe that failure of the film–substrate interface scavenges material from the counterface bit by bit which causes the increase of pit length. The fact that average wear pit depth did not change between 50 and 200 k cycles in spite of the increased average abrasion depth from 0.48 to 0.82 µm suggests that wear occurred universally across the surface regardless of the transfer film coverage conditions.
assuming uniform pressure across the contact. Here, X is the transfer film area fraction, ka the counterface adhesive wear rate at film-free regions, and kf is the counterface fatigue wear rate at film-covered regions as illustrated in Fig. 7. From Eq. 6, the plot of ksteel versus X should be a straight line connecting (0, ka) and (1, kf). Wear rates of the film–substrate interface were plotted against transfer film area fractions in Fig. 10. The best-fit using the least square method was shown as the red solid line and the two have a strong linear correlation (R2 = 0.92). The rule-of-mixtures model was plotted as the black dashed line in which the adhesive wear rate constant, ka = 5 × 10− 7 mm3/Nm, was determined using the y-axis intercept of the red solid line. Interestingly, this value is only one order of magnitude lower than wear rates of soft radioactive metals (Cu, Zn, Ag) against PTFE (~ 3 × 10− 6 mm3/Nm) as measured by Rabinowicz and Shooter , whereas they used a much higher contact pressure (~ 100 MPa) than in this study. The fatigue wear rate constant, kf = 2.6 × 10− 8 mm3/Nm, was determined using the minimum value of counterface wear rate measured. This is our best guess of kf, since ka is 10 × greater than kf which makes it extremely unreliable to extrapolate kf from the slope constant in Eq. 6. The gray area in Fig. 10 corresponds to the mean variation between the data and the model.
In Fig. 10, there is an increased deviation of the data from the model at the bottom (X > 0.75). This, we believe, could result from two possible factors. First, as transfer film became more continuous, the average size of real microscopic contact area likely increased (Fig. 5c). From basic contact mechanics, the maximum shear stress at the film–substrate interface decreased and caused decreased fatigue wear rate at the interface. Second, for the model, we assume full contact and macroscopically uniform contact pressure at both film-covered and film-free regions such that X and 1 − X also represent the fraction of the total normal load supported on the film-covered and the bare regions, respectively. However, as the transfer film grew more complete and uniform, the residual handful film-free regions might lose contact with the bulk’s running surface during sliding due to their smaller size and thickened transfer film islands. This could also produce lower counterface wear rates than the model predicts.
There are a few important implications of the results in this study. First, contrary to the traditional hypothesis of permanent transfer film adhesion and counterface protection, wear of the counterface persists even in areas covered with a persistent transfer film, at least for the alumina PTFE system. As migration or delamination of the transfer film islands were rare events for this system , we believe fatigue failures of the film–substrate interface mostly occur within a close range (1–100 nm) to the interface underneath the apparently persistent transfer film. The mechanism of such fatigue wear might involve complex micro-flows of energy and matter near the interface and require further studies. Second, transfer film coverage correlates more strongly with wear of the counterface than bulk polymer. Previous studies have suggested that polymer wear is also strongly governed by the structure and composition of the running film at the pin’s sliding surface . Third, the counterface wear data shed some light into tribological designs where wear of the hard-metallic component than the solid lubricants is of greater concern. In such cases, interfacial fatigue wear rate sets the lower limit of the metal wear rate and could be a strong function of transfer film coverage, distribution, and adhesion strength.
A well-studied alumina-PTFE solid lubricant was known to produce extremely adherent and complete transfer films when slid against hard, metallic counterfaces. After removing the transfer film and exposing the substrate underneath, it was found that wear of the counterface occurred continuously throughout the course of the wear test even after a persistent transfer film was formed on top of the counterface. Counterface wear rate decreased from 3 × 10− 7 mm3/Nm at 1 k cycles to 3 × 10− 8 mm3/Nm at 200 k cycles and had a strong linear relation with transfer film area fraction (R2 = 0.91). A rule-of-mixtures model was proposed to predict wear rate of the counterface as a linear function of film area fraction. The model agreed with experimental results closely and strongly suggested the coexistence of adhesive and fatigue wear mechanisms of the counterface under dry sliding condition.
The authors gratefully acknowledge financial support from the National Natural Science Foundation of China (51505117 and 11472096), the Natural Science Foundation of Anhui Province (1608085QE98) and the Fundamental Research Funds for the Central Universities.
- 27.Pitenis, A.A., Ewin, J.J., Harris, K.L., Sawyer, W.G., Krick, B.A..: In vacuo tribological behavior of polytetrafluoroethylene (PTFE) and alumina nanocomposites: the importance of water for ultralow wear. Tribol. Lett. 1–9 (2014)Google Scholar
- 41.Rabinowicz, E.: Friction and wear of materials. Wiley, New York (1995)Google Scholar
- 42.Greenwood, J.A., Tabor, D. Deformation properties of friction junctions. Proc. Phys. Soc. Sect. B 68:609: (1955)Google Scholar