Tribological Coating Wear and Durability Performance Guideline for Gear Applications

For decades, the performance of tribological coatings has been studied with the goal of developing practical surface treatments to enhance the friction and wear performance of mechanical components. Known parameters that affect coating performance include contact pressure, rolling velocity, sliding velocity, lubricant film thickness, and surface roughness. From the late 1980s to the early 2000s, reports of potential performance enhancements for gears [1, 2, 3] and bearings [4, 5] began to appear in trade magazines and consortium papers. During this period, a family of thin film coatings known as diamond-like carbon (DLC) gained popularity as a tribological coating for steel substrates — especially DLC doped with transition metals. This coating family is known generically today as metal-containing diamond-like carbon (Me-DLC) or metal carbide-reinforced amorphous hydrocarbon (MC/a-C:H).

Tungsten-incorporated DLC (W-DLC or WC/a-C:H) is a coating class that has demonstrated promise for surface fatigue [3, 6, 7] and scuffing resistance [7, 8]. Like many highly engineered material types though, significant variation in coating properties is possible by tailoring the coating deposition process specifics. As a result, not all Me-DLC variants are the same. Caution is encouraged when selecting Me-DLC for use in real applications because experimental Me-DLC data published in the literature may or may not be relevant to a specific coating provided by a specific supplier. With that said, the results presented here are intended to provide a general guideline for gear train component and equipment designers as a prerequisite to prototype testing of specific application scenarios.

Performance improvements of Me-DLC-coated components are attributed to two primary factors. In the context of surface fatigue risk, the coated surface preferentially polishes the mating surface, which reduces the composite surface roughness of the contact pair and thereby improves the lubricant efficacy. In addition to mild abrasive polishing, the coating also acts as a dissimilar material between the two contacting surfaces and reduces the propensity for adhesive wear that leads to scuffing.

Coating wear was created in previous tests by pushing the limits of contact stress and slip into extreme regions [3, 6, 8]. Even though improved overall component contact fatigue and scuffing resistance were demonstrated, one outcome of the tests was significant coating wear. If the coating wears during normal operation (i.e., non-extreme situations), then it will not be available for scuffing protection during extreme events such as aircraft gearbox oil-out events or during high-torque, wide-open throttle acceleration from standstill of an automobile or heavy truck. The goal of this study is to understand coating wear risk by mapping the wear behavior of an example W-DLC coating under wide contact condition ranges and comparing this performance to the normal operating conditions of gear applications.

Background

This bench-scale study of coating wear risk is the first in a planned series of screening tests to generate data, leading to future subscale and full-scale gearbox testing. Common models for surface damage mechanisms, such as wear and smearing, typically include contact pressure and sliding distance or sliding velocity [9, 10] as factors, along with some measure of the lubrication state within the contact, such as the specific film thickness λ, which is the ratio of calculated film thickness to composite surface roughness. Rotating ball-on-disk testing was chosen as the screening test in this study because of its minimal specimen cost and wide range of possible loading conditions. Applied contact load, ball surface velocity, and disk surface velocity are among the independently controlled test factors, enabling a variety of contact pressures, sliding velocities, and lubricant-entraining velocities for the tests. Although the ball-on-disk test rig used for this study is flexible for short-duration fundamental studies, it was not designed for high-cycle fatigue testing. The number of load cycles for this study was limited to 2 x 106 cycles to prevent test rig failures. High-cycle fatigue tests on test rigs, such as rolling-sliding contact fatigue rigs or power recirculation gear test rigs, are needed to fully characterize coating wear for higher cycles.

Data generated from these tests are intended to be relevant to gear calculations that engineers use when designing gears with coatings. A brief review of various AGMA standards led to AGMA 925-A03, “Effects of Lubrication on Gear Surface Distress” [11], which includes a guideline relating the probability of wear to pitch line velocity and specific film thickness (see Figure 1). This map was developed from data analyzed by Wellauer and Holloway in 1976 [12]. It is based on laboratory tests and field applications of industrial gears fabricated from through-hardened steel ranging in size from 25 mm to 4,600 mm and lubricated with mineral-based, non-EP gear lubricants. The issue with using the pitch line velocity format for this study is the lack of any direct relationship between gear pitch line velocity and rotating ball-on-disk test operations.

Additional information is needed to relate the map in Figure 1 to the rotating ball-on-disk test controlled by fundamental load and surface velocity settings. Specific gear design information and knowledge of gear ratings, along with the location on the tooth surface where surface damage is most likely to occur, are needed. As an example, for a low-contact ratio spur gear, does wear primarily occur near the low point of single-tooth contact, which is approximately the point of maximum contact stress, but with low sliding? Does wear occur at the actual start of contact, near the root fillet where maximum sliding occurs but where there is minimal contact stress (assuming that an adequate amount of tip relief is applied to minimize tip interference)?

For this study, λ and the product of contact pressure P and sliding velocity V were proposed as an initial map of coating wear. On a tooth surface, the product PV has two local maximums: one at the location between the start of contact where maximum sliding occurs and one at the low point of single-tooth contact where the maximum contact stress occurs. Using the PV format, one expects the map of coating wear to look similar to Figure 2, with PV on the horizontal axis and λ on the vertical axis. Wear should be minimal in the upper left-hand region (indicated in blue), which corresponds to high λ, but with low pressure and low sliding. Wear should be maximum in the lower right-hand region (indicated in red) that corresponds to low λ and high PV. However, as testing progressed in this study, the data indicated that pressure was more dominant than sliding velocity, so the model was revised to consider PV^0.2 instead of PV.

Gear designs from various applications and research papers were analyzed to establish the range of contact pressure and sliding velocity that would be applicable for this study. The maximum contact pressure planned for this study was based on accelerated testing that was reported in research papers, and maximum sliding velocity was based on the highest pitch line velocities from applications. The maximum pitch line velocity analyzed was 60 m/sec., and the approximate sliding velocity at the location of maximum PV^0.2 for this application was 7 m/sec.

Procedure

W-DLC coating wear tests were performed on a rotating ball-on-disk test machine (shown in Figure 3). The ball diameter was 20.62 mm. Two coating combinations were tested: uncoated balls on coated disks and coated balls on coated disks. The test duration was 2 million cycles.

Two case-carburized materials were selected for evaluation: AMS6308 to represent aerospace gear applications and AISI 4320 to represent mobile and industrial applications. The target surface hardness was 60-63 HRC and case depth was targeted at 0.9-1.3 mm. The disk face surfaces were ground in the circumferential direction and superfinished using a chemically enhanced vibratory process. The superfinishing process did not remove all the grinding marks on the disks, so there were some visual indications of residual grind lines. The W-DLC coating was applied to the balls and disks after the superfinishing process. Coating was the final step of the process, and there were no additional finishing operations after the coating deposition. Surface roughness Sa was measured using a white light interferometric microscope. “Sa” means the surface was measured in 3D, as opposed to Ra, which is a 2D measurement with a stylus. The overall average roughness value for the disks was 0.079 μm Sa with a standard deviation of 0.024 μm; for the balls, the average was 0.142 μm Sa with a standard deviation of 0.017 μm.

Rotational directions and locations for lubrication and temperature measurements are shown in Figure 3. The disk rotated in the counterclockwise direction when viewing the disk from above. The ball rotated in the clockwise direction when looking at the end of the spindle. The surface velocity of the disk was always higher than that of the ball, imposing a slip condition with a non-zero slide-to-roll ratio. Slide-to-roll ratio is the ratio of sliding velocity to entrainment velocity. Entrainment velocity is the mean of the velocities of the contacting surfaces. Eight test runs per disk and two runs per ball were completed. Lubricant was applied directly on both the ball and disk to provide enough lubrication to assume a fully flooded contact zone. Ball and disk temperatures were measured in situ by contact thermocouples on the surfaces.

Four PAO lubricants with different viscosities were used to vary λ in the test matrix: ISO VG 10, 32, 68, and 220. The oils used did not contain any extreme pressure (EP) or antiwear (AW) additives. Contact pressures (P) ranged from 1.0 to 2.5 GPa. (As a reference, ANSI/AGMA 2101-D04 recommends a design allowable stress of 1.895 GPa for a grade 3 material.) Sliding velocities (V) ranged from 1.0 to 12.2 m/sec. and entraining (rolling) velocities ranged from 2.0 to 10.8 m/sec., which resulted in a range of slide-to-roll ratios of 0.26 to 1.85. The combinations of different lubricants, temperatures, and entrainment velocities resulted in calculated specific film thicknesses (λ) ranging from 0.2 to 3.7. The isothermal elastohydrodynamic central film thickness (h_central) [13] was used to calculate λ for each experiment.

A design-of-experiments approach varying P, V, and λ was initially used to establish the general trend line at which significant wear occurred. Once the initial trend line was established, additional tests were run in a searching method to extend the length of the trend line. Actual W-DLC coating wear is difficult to measure, since the initial coating thickness is in the range of 1–2 μm. A visual approach to estimate coating wear was used during the tests to shorten the overall test schedule. Figure 4 includes scanning electron microscope (SEM) images of an example disk with wear tracks and obvious coating wear.

As an initial assessment, the wear was designated as heavy, noticeable, or minimal. Heavy wear is shown in track 1. This run resulted in the coating being completely worn to expose the base metal of the disk. The smaller image of track 1, shown in the upper-right corner, is an SEM backscatter image in which the base material of AISI 4320 is predominantly iron and exhibits a dark appearance. Noticeable wear is shown in track 3. This amount of wear resulted in the top layer of the coating being worn and exposing the adhesion sublayer of the coating, which is made of chromium. The chromium layer is brighter in the SEM backscatter image than the W-DLC coating top layer. Minimum wear was observed in tracks 4 and 6. These runs resulted in only surface polishing with very minimal coating removal. This amount of wear was not enough to expose the chromium layer, and in the backscatter image, there is no detectable difference in brightness between the wear track and the untested coated surfaces at the edge of the tracks (as shown in the images at lower left).

Energy dispersive spectrometry (EDS) analysis was performed to verify the chemical content of three of the wear track regions. The heavy wear area shown in Figure 4 was analyzed, and EDS results are shown in Figure 5. The analysis correlated to the amount of wear. The unworn top layer (identified as “OD Coating”) has a large percentage of tungsten (87% W) but with low amounts of iron (3% Fe) and chromium (3% Cr). The area with no coating (identified as “Base Metal”) was primarily iron (89% Fe) with a low amount of chromium (1% Cr) and no tungsten. The moderately worn area (identified as “Substrate”) had a combination of tungsten (63% W) and chromium (28% Cr). It is important to emphasize that these percentage compositions are not the calibrated and quantitative elemental compositions of the coatings, but instead should be interpreted as normalized qualitative values to represent relative concentrations of the elements within the SEM-EDS sampling volume. These data indicate what portion of the coating is present in locations where visual assessments may be misleading.

Initially, an attempt was made to measure the depth of the wear tracks on the disks using a white light interferometric microscope. This method was able to identify the depth of the heavily worn tracks, but many of the more lightly worn tracks did not show significant depth. However, the 3D surface measurements did indicate a reduction in surface roughness in the areas of mild wear.

For the final evaluation of coating wear, x-ray fluorescence (XRF) measurements were made on the wear track surfaces. The percentages of W, Cr, and Fe (%W, %Cr, and %FE) in an unworn area were compared to the same measurements within each wear track. The test duration was set at 2 x 106 cycles, but some test conditions caused significant wear before completing the test. A wear factor was used to evaluate the amount of wear for each test run. The ratio of the base unworn %W₀ to wear track %W_i and the ratio of test duration to actual test cycles were used to calculate the wear factor. The wear factor was used to rank the wear results.

Where:

%W₀ is the percent tungsten of the base coating relative to XRF configuration (not absolute)

%W_i is the percent tungsten in the wear track relative to XRF configuration (not absolute)

n_i is the number of test cycles

Results

The results for the coated disk and uncoated ball did not indicate a significant difference between the AISI 4320 steel and the AMS6308 steel. The two data sets were combined, and the results are shown in Figure 6. The data seem to correlate better to a model of λ versus PV^0.2 rather than PV, so the data are plotted using PV^0.2. Note that P is in units of GPa and V in units of m/sec. (units were chosen for convenience in plotting). Linear best-fit lines are included for the three groupings of wear to illustrate the general trend. As expected, minimal wear occurs under the conditions represented in the upper left-hand region (high λ and low P and V), and heavy wear occurs in the lower right-hand region (low λ and high P and V), similar to the Figure 2 illustration.

The data used to plot Figure 6 are tabulated in rank order by wear factor in Table 1. The table is color-coded to match the color code of the plotted data. The linear best-fit lines in Figure 6 suggest that the λ value will be negative as PV^0.2 approaches zero, which is not possible. The calculated λ will approach zero but cannot be negative, so it is reasonable to assume that the transition zone between minimal wear and moderate wear can be more accurately represented by an exponential function. Therefore, an exponential curve fit (shown in grey in Figure 6) was fitted to the minimal wear data points that are located near the transition between no wear and moderate wear. The PV^0.2 numbers from runs 8, 18, 20, and 22 of the AMS6308 tests and runs 4, 8, 8a, 17, and 20 of the AISI 4320 tests represent the data points used for the exponential fit curve in Figure 6. The curve takes the form of:

Where:

y = λ

x = PV^0.2

This curve represents the transition between minimal or no coating wear and moderate coating wear. For a given P and V, this equation indicates the minimum λ to prevent wear in the rotating ball-on-disk test.

A final series of tests was completed with both the disk and the ball coated to determine if there is a significant difference in the wear characteristics when both contacting surfaces are coated. Figure 7 is a plot of the data contained in Table 2.

There are minimal data because of the limited number of coated disks available for this study. Using a similar analysis methodology to the coated-uncoated test configuration, an exponential curve was fit to the data of minimum wear. The curve takes the following form. Again, P is in GPa and V is in m/sec.

Where:

y = λ

x = PV^0.2

Both of the exponential equations used to define the transition region of coating wear, as determined from the rotating ball-on-disk tests, are plotted in Figure 8. There is a clear difference in the wear characteristics when both surfaces are coated compared to when only one surface is coated. Wear occurs at lower P and V when both contacting surfaces are coated compared to when one of the two surfaces is coated. This is because the coated-on-coated scenario is a similar interface with no one-directional preferential polishing of only one surface — that is, both coated surfaces have equally high hardness and appear to polish each other as a result, accelerating the overall coating wear rate.

Considerations

The purpose of this study was to identify steady operating condition ranges where there is a reduced risk for coating wear. Caution needs to be taken when using these data because a direct relationship has not been established between the observations from a ball-on-disk rig with point contact and full-scale gears in real applications.

Variation in results can be attributed to:

• Film thickness model: The Hamrock and Dowson model is limited to the elastohydrodynamic lubrication regime and does not account for the thermal effects associated with the sliding velocity. One way to accomplish this would be to apply a thermal correction factor as proposed by Gupta, et al. [14].
• Qualitative nature and variability of XRF measurements used to characterize wear.
• Actual lubrication temperature and condition at the point of entrainment: The lubricant was dripped on both the ball and disk surfaces. The actual lubricant temperature at the point of entrainment will not be precisely the same as the measured lubricant temperature because of heat transferred between the ball and disk surfaces and the lubricant.
• Properties of the lubricants used: Film thickness calculations for ISO VG 68 and 220 PAO oils were based on measured properties from similar ISO VG PAO oils, but not of the exact oil.

Previous scuffing resistance testing by Ribaudo et al. [8] and surface fatigue testing by Krantz et al. [6] of coated gears resulted in improved gear performance, but both reports indicated that there was coating wear at the conclusion of the testing. The approximate conditions of the gear tests are plotted with the ball-on-disk trend lines in Figure 9. These data were obtained from gear geometry and test conditions available in the reference papers and analyzed using the Ohio State Load Distribution Program (LDP) [15].

Two additional data points are included for these gears wherein the contact stress was limited to 50 percent of the grade 3 design allowable. This condition could represent a more normal operating condition for the gears. The coated gear test conditions fall within or near the region where coating wear occurred during the ball-on-disk test. Therefore, it would be reasonable to expect that the coating would wear under these conditions based on the testing and analysis presented in this work. In contrast, the 50-percent design allowable points are well within the region in which no coating wear would be expected based on ball-on-disk testing. The coating on gears designed to this level of stress under normal operating conditions are expected to have minimal wear and should be available for scuffing protection throughout the product life cycle.

Conclusions

Contact pressure is a more significant factor than sliding velocity for W-DLC coating wear, as evidenced by relative exponential weightings on the PV^0.2 factor as a best-fit parameter to the ball-on-disk experimental wear data.

The model of λ versus PV^0.2 seems to correlate well with the ball-on-disk wear results for describing W-DLC coating wear risk in the studied contact condition ranges. Higher W-DLC coating wear rates can be expected when both surfaces are coated than in the scenario where only one surface is coated. W-DLC coating wear that occurred during previous gear tests for scuffing resistance and surface fatigue seems to correlate to the risk ranges developed with ball-on-disk results in this study.

It can be expected that minimal or no coating wear will occur on gears that are operating under normal application conditions, based on expected design λ and PV^0.2 gear application design guidelines. The coating can be expected to remain intact to provide scuffing protection under momentary extreme operating conditions, such as an oil-out event. Gear testing is needed to fully validate this conclusion.

Acknowledgments

Jerry Richter for coordinating the ball-on-disk tests. Bill Hannon for providing technical input to the study. Bob Pendergrass for the SEM results. Carl Ribaudo, Matt Olinger, and Tumkur (Gopi) Prasad for coordinating the heat treatment. Rich Fowler of the University of Akron for coordinating the coating process. Curt Orkin and Sheila Cowles for providing business support for this project.

References

1. F. J. Teeter and M. Berger, “Wear Protection for Gears,” Gear Technology, pp. 27–29, March/April 1996.
2. W. R. Stott, “Myths and Miracles of Gear Coatings,” Gear Technology, pp. 35–44, July/ August 1999.
3. N. E. Anderson and L. C. Lev, “Coatings for Automotive Planetary Gearsets,” in Proceedings of DETC’03 ASME 2003 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, Chicago, 2003.
4. G. L. Doll, R. D. Evans, and C. R. Ribaudo, “Improving the Performance of Rolling Contact Bearings with Tribological Coatings,” The Minerals, Metals & Materials Society, 2005.
5. G. L. Doll, R. D. Evans, and S. P. Johnson, “Providing Oil-out Protection to Rolling Element Bearings with Coatings,” in Society of Vacuum Coaters 48th Annual Technical Conference Proceedings, 2005.
6. T. L. Krantz, C. V. Cooper, D. P. Townsend, and B. D. Hansen, “Increased Surface Fatigue Lives of Spur Gears by Application of a Coating,” in Proceedings of DETC ‘03 ASME 2003 International Design Engineering Technical Conferences and Computers and Information in Engineering Conferences, Chicago, 2003.
7. F. Joachim, N. Kurz, and B. Glatthaar, “Influence of Coatings and Surface Improvements on the Lifetime of Gears,” Gear Technology, pp. 50–56, July/August 2004.
8. C. R. Ribaudo, C. Aylott, D. Hofmann, C. V. Darragh, R. D. Evans, E. P. Cooke, and G. L. Doll, “Engineered Surfaces for Improved Gear Scuffing Resistance,” in Proceedings of DETC’03 ASME 2003 Design Engineering Technical Conferences and Computers and Information in Engineering Conference, Chicago, 2003.
9. U. Kissling and S. Hauri, “Gear Wear Calculation and Prediction,” Power Transmission Engineering, pp. 36–38, February 2015.
10. R. D. Evans, T. A. Barr, L. Houpert, and S. V. Boyd, “Prevention of Smearing damage in Cylindrical Roller Bearings,” Tribology Transactions, pp. 703–716, 2013.
11. Effect of Lubrication on Gear Surface Distress AGMA 925-A03, AGMA, 2003.
12. E. J. Wellauer and G. A. Holloway, “Application of EHD Oil Film Theory to Industrial Gear Drives,” Journal of Engineering for Industry Transactions of the ASME, vol. 98, no. series B, no. 2, pp. 626–634, May 1976.
13. B. Hamrock and D. Dowson, “Isothermal Elastohydrodynamic Lubrication of Point Contact Part III – Fully Flooded Results,” Journal of Tribology, vol. 99, pp. 264–275, 1977.
14. P. K. Gupta, H. S. Cheng, D. Zhu, N. H. Forster, and J. B. Schrand, “Viscoelastic Effects in MIL-L-7808-Type Lubricant, Part I: Analytical Formulation,” Tribology Transactions, vol. 35, no. 2, pp. 269–274, 1992.
15. Windows-LDP Load Distribution Program, Columbus, Ohio: The Ohio State University, 2014.

* Printed with permission of the copyright holder, the American Gear Manufacturers Association, 1001 N. Fairfax Street, Suite 500, Alexandria, Virginia 22314. Statements presented in this paper are those of the authors and may not represent the position or opinion of the American Gear Manufacturers Association (AGMA). This paper was presented October 2015 at the AGMA Fall Technical Meeting in Detroit, Michigan. 15FTM24.