Effect of lubricants on bearing damage in rolling-sliding conditions: Evolution of white etching cracks

Wear 398–399 (2018) 165–177

Contents lists available at ScienceDirect

Wear journal homepage: www.elsevier.com/locate/wear

Effect of lubricants on bearing damage in rolling-sliding conditions: Evolution of white etching cracks

T

⁎

Mohanchand Paladugu , Douglas R. Lucas, R. Scott Hyde The Timken Company World Headquarters (WHQ), North Canton, OH 44720, USA

A R T I C L E I N F O

A B S T R A C T

Keywords: Steel Bearings Lubrication oil Additives Rolling-sliding Sliding friction Cracking White etching cracking Axial cracks

To understand the role of lubricants in generating white etching cracks (WECs), cylindrical roller thrust bearings were life-tested in different oils in rolling-sliding conditions. The bearings were damaged prematurely when the tests were performed in an oil with additives (so-called “WEC critical oil”). Bearing life in the WEC critical oil was less than 5% of that of the mineral oil without additives. In the bearings tested in the WEC critical oil, posttest investigations revealed a non-uniform ~ 100 nm thick tribo-film and micro-scale line cracks parallel to the contact line in the slide zones of the tribo-surface. In contrast, typical “point surface origin” (PSO)-type damages were observed when the bearings were tested in the mineral oil. Electron microscopy investigations showed that the line cracks in the case of the WEC critical oil tests propagated deeper into the subsurface. Only in the negative slide zone, the cracks then became WECs. The PSO damages generated in the mineral oil testing did not show subsurface cracking or white etching features. The results support the conclusion that, in the case of the WEC critical oil, a tribo-film formed and caused an increase in surface shear forces, which consequently resulted in the line cracks on tribo-surface that subsequently propagated deep into the subsurface.

1. Introduction Key mechanical engineering components are tested rigorously in application-relevant tests to ensure high reliability of the components in the actual application and to avoid damage in severe conditions. Roller bearings are key components in rotating machinery, and avoiding premature bearing damage is essential for uninterrupted operation of the machinery and for minimal maintenance costs. Due to the existing dynamic loading and severe operating conditions in wind turbines, wind turbine gear box bearings have been reported to become prematurely damaged [1]. But replacing them with new bearings at top of the wind tower adds to the maintenance cost immensely. Investigations of prematurely damaged wind turbine bearings have shown cracks in the bearing steel. So, in addition to choosing the right bearing design, a suitable steel grade, an appropriate heat-treating condition and high steel cleanness must be present to minimize the cracks generation in the bearings. For example, in wind turbine gear boxes, bearings face nonRCF (rolling contact fatigue) loads from the shaft; these loads include installation stresses (hoop stresses) and bending stresses. Under these stresses, it is reported that through-hardened bearings develop axial cracks and damage prematurely [2]. Owing to their tough core, casehardened bearings can accommodate these loads and hence minimize

⁎

the chances of axial cracking. In addition, the microstructure should be designed to resist fatigue damage from dynamic loading in wind turbines. In fact, investigations of the field-returned bearings suggest that case-carburized bearings with more than 20% retained austenite in the case-microstructure can achieve very long lives in the wind turbine environment [3]. However, evidencing these insights in a laboratory test is a challenge because of the unavailability of a laboratory-scale test method that can represent the environmental and mechanical loading conditions of wind turbines. Developing such a test method is still a challenge to date. Deeper optical metallographic study of the prematurely failed turbine bearings has revealed white etching matter (WEM) along the cracks in the steel, as well [3]. The white appearance comes from the nanograined ferrite material present along the crack faces. This nanograined material doesn’t get etched or etched uniformly during the etching step of the metallographic sample preparation. Therefore, it doesn’t produce a local contrast by diffracting the optical light, and consequently appears as white in optical metallography observations. Since WEM was found along the cracks, it was initially assumed that WEM is a cause of bearing damage [4,5]. When WEM is found along a crack, the crack is often called a “white etching crack” (WEC). Several investigators have reported different

Corresponding author. E-mail address: [email protected] (M. Paladugu).

https://doi.org/10.1016/j.wear.2017.12.001 Received 25 October 2017; Received in revised form 30 November 2017; Accepted 1 December 2017 Available online 08 December 2017 0043-1648/ © 2017 Elsevier B.V. All rights reserved.

Wear 398–399 (2018) 165–177

M. Paladugu et al.

roller assembly is shown in Fig. S3 of the Supplementary material). During bearing operation, load was applied normally on the bearing while the rollers and one of the raceways (washers) are under rotation. Because of the difference in local circumference and the resulting rotation speed differences along the “line of contact” between the rollers and the raceways, about 11% sliding is generated at the ends of the contact line, as shown in the bottom sketch of Fig. 1a [13,20]. Negative sliding generates toward the inner radius of the raceway because the velocity of the rollers’ surface is higher than that of the raceway's surface. Toward the outer radius, positive slide generates on the raceway because the velocity of the rollers’ surface is lower than that of the raceway's surface. Two CRTBs were assembled into the test rig as shown in the schematic in Fig. 1b (cross section of the test rig). The four washers of CRTBs are depicted in blue and the roller assemblies are shown in red. During the bearings operation, the washers placed at extreme ends are stationary and the other two inner washers are under rotation. The sliding applies to the stationary washers as well because of the difference in the local circumference along the contact line. The rollers also experience the sliding, except the sign of sliding is opposite to that of the washers. Thrust load was applied on the bearings by compressing the springs to the required level. In this study, a load of 60 kN was applied (~ 1.9 GPa of contact pressure), the drive shaft was rotated at a speed of 750 rpm, and the temperature of the bearings was set at 100 °C. These test conditions are proven to produce WECs repeatedly. As shown in the schematic (Fig. 1b), oil was supplied to each bearing; the used oil flow to each bearing was 0.2 L/min. An oil tank was connected to the test rig from which the oil was circulated to each bearing and collected back into the tank. During the bearing test, damage to the bearings was detected by attaching a vibration sensor (accelerometer) to the bearing assembly. As can be seen in Fig. 1c, the vibration sensor is attached to the top of the bearing assembly. During each test, as the bearing damage begins to occur, the vibration signal increases. The tests can be setup to terminate at certain level of vibration increase. In these life-test experiments, to investigate initial stages of the damage to the bearings, tests were terminated at early stages of vibration signal increase by adjusting the respective sensor levels. The stop criterion of the test was set at 100% vibration increase compared to vibration at the start of test.

ways of producing WECs in laboratory-scale bearing and RCF tests and have discussed underlying mechanisms. The following test parameters are proven to reproduce WECs: i) using low-reference lubricant oil in rolling-sliding conditions [6–11]; and ii) using test components with material and surface finish defects (e.g., inclusions, cracks, intergranular embrittlement and rough surfaces) [12–14]. Recent reports published on the topic suggest that WECs are not a root cause of bearing damage after all; instead, WEM along a crack is a consequence of the crack faces rubbing [2,12,13]. In other words, the cracks form first, and when the crack faces rub each other, WEM forms along the cracks. However, in contrast to these insights, there are other published reports that suggest WECs are a type of failure mode and cause premature bearing failures [15–17]. More study is required to elucidate the mechanisms associated with WEC formation and to evaluate the relationship between WEC formation and bearing life. In fact, the method of testing the bearings in low reference oil under rolling-sliding conditions is used to compare lives of Cylindrical Roller Thrust Bearings (CRTBs) made with different materials and heat treatment conditions [16]; the used test apparatus is called FE-8 test rig. This test rig is a commercially available and is recommended for lubricant evaluation by the DIN 51819 standard [18]. Using this test rig and a low reference oil, several published reports suggest that WECs form consistently in the bearings and cause premature bearing damage [8–11,15,16,19,20]. More detailed studies are needed relating to influence of lubricants on bearings’ damage and respective damage mechanisms. To address this, we performed life tests on CRTBs in the FE-8 test rig in different lubricant oils. The tested bearings were investigated in detail by optical and electron microscopy, and accounted the associated damage mechanisms with respect to the used lubricant oils. 2. Materials and methods 2.1. Bearing test rig and test rig parameters An FE-8 test rig was used to test the bearings in different oils. The test rig is made by Schaeffler Technologies, Schweinfurt, Germany. Cylindrical roller thrust bearings (CRTBs) were used in this study. Fig. 1a shows an example of the CRTB; the rollers between the two raceways (washers) are held in place by a brass cage (a photograph of a

Fig. 1. (a) A photograph showing an example of a cylindrical roller thrust bearing (CRTB). The schematic diagram in the bottom shows sliding along the line of contact. (b) A schematic cross section of the bearing assembly in the test rig. (c) Photograph of the test rig with vibration sensor on top of the test head. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

166

Wear 398–399 (2018) 165–177

M. Paladugu et al.

2.2. Bearing materials The CRBTs (rollers and raceways) were made of 52100 grade steel. The steel's composition in wt% was C: 1.05%; Mn: 0.29%; Si: 0.28%; Cr: 1.49%; Al: 0.015%; S and P less than 0.001%. The bearing components were in austempered heat treated condition. To avoid embrittlement of the steel, it was austenitized in a lower temperature range than that reported in our previous investigations [12,13]. The bearing steel contained a lower bainite microstructure with 3% retained austenite. The size of the grains was between 20 and 30 µm. The average surface roughness (Ra) of the rollers and raceways was between 70 and 90 nm. 2.3. Lubricant oils In this study, two different lubricants were used: one a life testing ISO VG68 mineral oil [21] and the other a commercially available SAE 75W-80 oil [9]. The SAE 75W-80 oil is the “WEC critical oil” specified in this report. Both the oils have a viscosity of about 65 mm2/s at 40 °C. In their fresh condition (before the life tests), both the oils were analyzed for the presence of metallic elements, contaminants and sulfur. The metallic elements were measured by rotating disc electrode atomic emission spectrometry (as per ASTM D6595) [22] and the sulfur was measured by wavelength dispersive X-ray fluorescence spectrometry (as per ASTM D2622) [23]. The water content of the oils was measured by a titration-based technique, the Karl Fischer method [24]. The compositions of the oils as measured are shown in Table 1. As can be seen in Table 1, compared to the mineral oil, the WEC critical oil has large amounts of metallic elements and sulfur. These elements are found in various oil additives. In addition, the water content of the WEC critical oil (4174 ppm) is considerably higher than that of the mineral oil (44 ppm). To accelerate the bearing damage, in some of the bearing life tests 10% water was added to the mineral oil, stirred thoroughly and used as the test lubricant (“mineral oil and water lubricant”). It has been reported that the addition of water to lubricant oils causes hydrogen intrusion into the steel surfaces of a tribo-contact [25,26]. To minimize water evaporation in the life tests we conducted with mineral oil and water lubricant, the test temperature was set to 50 °C.

Fig. 2. A plot showing the life times of CRTBs in different lubricant oils. Tests were conducted in a FE-8 test rig made by Schaeffler Technologies. Test conditions were 60 kN load (1.8 GPa contact pressure), 750 rpm shaft rotation speed. The test rig temperature was set at 100 °C.

technique to reveal the WEM along the cracks.

3. Results 3.1. Overview of damages produced in different oils Fig. 2 shows a plot of bearing life in the different oils. The CRTBs tested in the mineral oil showed damage (vibration increase) after about 980 h, whereas the bearings tested in the WEC critical oil showed damage at about 40 h. When the bearings were tested in the mixture of mineral oil with 10% water, the bearings became damaged at about 450 h of testing. It is known from the literature that addition of water to the lubricant decreases bearing life [27]. These reported lifetimes were found to be repeatable. In the case of WEC critical oil, a weibull plot of life times from six different tests is shown in Fig. S1 of Supplementary material; high weibull slope (9.35) resemble the consistency of early bearing damages in the WEC critical oil. In the case of mineral oil and “mineral oil and water” lubricants, because of long lifetimes, only 2–3 tests were conducted in respective lubricants, and the life times are ± 50 h to the reported values in Fig. 2 To understand bearing life differences with respect to the lubricant oils, we examined the respective damages in the tested bearings. In the current bearing tests, since damages were found mainly on raceways, the raceways of the tested bearings were analyzed by optical and electron microscopy. To identify the early stages of the damage, we investigated the micro-scale damages present on the post-test raceways. All the four post-test raceways of each bearing were examined for the micro-scale damages. Fig. 3a shows a typical photograph of a raceway after the test in the WEC critical oil; the “high slide regions” near the ends of the contact line appear as bright rings (clearly shown in the inset). Careful optical and electron microscopy observations of those bright rings showed micro-scale line cracks on the steel's surface along the line of contact; an optical image of a typical crack from the negative slide zone is shown in Fig. 3b. The corresponding direction of the rollers’ surface during bearing operation is shown in the image. These kinds of line cracks were seen both in the negative and positive slide areas (see Fig. 3a inset). In fact, these line cracks were observed to be present only on the races tested in the WEC critical oil. In addition to the line cracks present on some of the raceways, there is at least one large, visible spall (to naked eye) present on at least one of the raceways that has caused the vibration signal increase from the test rig that consequently stopped the bearings’ life tests. These kinds of macro spalls (produced by extending the test time even after 100% vibration increase) have been shown in various references [8,9,19] and also shown in the images in Fig. S2 of the Supplementary material. For understanding, an image of typical post-test roller assembly is shown in

2.4. Materials characterization Post-test bearing components were cleaned in ultrasonic bath of hexane followed by Isopropanol. The cleaned bearing components were characterized by optical microscopy and electron microscopy. Metallography samples were etched using a 2% Nital solution. A scanning electron microscope (SEM) was used to characterize the materials. We used a FEI Versa 3D Dual Beam SEM equipped with scanning ion microscopy (SIM) and with focused ion beam (FIB) milling and scanning transmission electron microscopy (STEM) capabilities. As shown in our past investigations [12,13], SIM imaging is able to reveal the features of the local microstructure due to the different beam penetration depths of the ion beam based on the crystallographic orientations of the microstructure (channeling effect). Since WEM is composed of nanograin structures, we could thus use the SIM imaging Table 1 Various additive elements and water present in the WEC critical oil and the mineral oil.

B (ppm) Na (ppm) Ca (ppm) P (ppm) Zn (ppm) S (ppm) Water (ppm)

SAE 75W-80 (WEC critical oil)

ISO VG68 mineral oil

581 1442 4479 1496 2201 7946 4174

0 0 6 0 68 400 44

167

Wear 398–399 (2018) 165–177

M. Paladugu et al.

Fig. 3. (a) Photograph of a washer (raceway) tested in the WEC critical oil; inset is the magnified image showing the slide zones in the raceway. (b) Optical microscope image of a typical line crack in the slide zone shown in (a). (c) Photograph of post-test raceway from mineral oil; slide zones are shown by arrows. (d) Optical microscope image of typical PSO damage in the slide zone shown in (c). (e) Optical microscope image of a typical micro-pit on the post-test raceway from the mineral oil and water lubricant.

Fig. S3. Fig. 3c shows a post-test raceway from a life test in the mineral oil. In comparison to Fig. 3a, there are no distinct bright rings present in the high slide regions of Fig. 3c. In contrast to the line-like cracks observed in the WEC critical oil tests, the damage in the mineral oil tests is much broader and resembles point surface origin (PSO)-like damage (as shown in Fig. 3d). PSO is a well-known damage mode in rolling contacting components where the damage starts at a point on the raceway and broadens further along the roller's direction. As a consequence, PSO damage is shaped like an arrowhead and is shallow at the arrowhead's point, which is the site of damage initiation [28]. In the mineral oil and water lubricant tests, the damage characteristics on the raceways were observed to be micro-pits, as shown in Fig. 3e. To understand the differences in post-test raceway wear, surface profiles were measured using a contact profilometer. Fig. 4 shows the typical surface profiles of the tested raceways in the three different lubricants. As can be seen in Fig. 4, the raceways tested in the WEC critical oil show remarkable wear in the slide zone (at the end of the contact line) compared to the raceways tested in the mineral oil and mineral oil and water lubricants.

Fig. 4. Surface profiles of raceways tested in different oils. A stylus-based profilometer (TSK Zeiss gage) was used to measure the profiles. The inset is a magnified photograph of the raceway tested in WEC critical shown in Fig. 3a.

to the different oils, the raceways in the slide zones (shown by arrows in the insets of Fig. 3a and c) were further characterized and compared to each other. Fig. 5a–c, respectively, show the tribo-surfaces of the raceways from the bearings tested in mineral oil, WEC critical oil and

3.2. Characterization of tribo-surfaces produced in different oils To further understand the tribo-surface characteristics with respect 168

Wear 398–399 (2018) 165–177

M. Paladugu et al.

Fig. 5. Optical images of tribo-surfaces (in the slide zones) of post-test raceways from different oils: (a) mineral oil; (b) WEC critical oil; (c) mineral oil and water.

described in Table 1; the source of the oxygen may have been the water content present in the WEC critical oil. Fig. 7c shows an EDS spectrum taken on the bright areas of Fig. 6b. Fig. 7c shows that no significant intensities from the additive elements of the WEC critical oil are present. This result suggests that a non-uniform tribo-layer was formed from the additives of the WEC critical oil. In addition to the raceways, tribofilm was also observed on the surface of post-test rollers from the WEC critical oil. The morphology and composition characteristics of the tribofilm are shown in Fig. S3 of the Supplementary material; these characteristics are not exactly same as that of the raceways (Figs. 5b, 6b, 7b). To further understand the characteristics of the WEC critical oil tribo-layer on the raceways, STEM (scanning transmission electron microscopy) was conducted in the chosen locations. Fig. 8a shows a SEM image of the tribo-layer. It is of interest to note from the SEM image that the tribo-film has wavy and rough topography in the sliding direction; these topological features resemble shear marks on the tribofilm. A STEM sample was prepared from the transition region of bright and dark areas indicated by the dotted line in Fig. 8a; a platinum layer was deposited on the transition region and the sample was prepared using FIB milling. Fig. 8b shows a STEM image from the transition region. In support of the SEM and EDS observations, in Fig. 8b one can see a tribo-layer on the steel surface with varying thickness. The STEM revealed varying thicknesses from about 100 nm to < 5 nm. It can be recalled from Figs. 6b and 7b that a higher-intensity EDS spectrum from the additive elements was observed in the darker regions, suggesting that the darker areas in the SEM images (Figs. 6b and 8a) have a thicker tribo-layer and the brighter areas have a thinner tribo-layer. Similar STEM investigations were conducted on the FeO tribo-layer formed in the case of mineral oil and water lubricant. The FeO layer also has a thickness in the range of 100–150 nm with a non-uniform thickness distribution.

mineral oil and water. The tribo-surface from the bearings tested in mineral oil appears relatively smooth with a Ra of 30–40 nm (Fig. 5a), but in the case of WEC critical oil (Fig. 5b), the surface appears to be uneven with patch-like features (Ra of ~ 80 to 100 nm). The tribosurface from the bearings tested in mineral oil and water (Fig. 5c) appears to be less smooth (Ra of ~ 60 to 80 nm) compared to the bearings tested in mineral oil (Fig. 5a). It is of interest to note that these morphologies were observed in areas of high sliding only. No significant difference between the surface morphology was observed in no- or low-sliding areas (i.e., near the middle of the contact line. We therefore conducted electron microscopy investigations on the tribo-surfaces to further understand their morphology and chemical composition. Fig. 6a–c, respectively, show SEM images of tribo-surfaces from the bearings tested in mineral oil, WEC critical oil and mineral oil and water. In support of the optical image in Fig. 5b, the tribo-surface produced by the WEC critical oil (Fig. 6b) appears to be uneven with bright and dark areas. In addition, the surface appears to have some deposit. In contrast, the tribo-surfaces from the mineral oil (Fig. 6a) and mineral oil and water (Fig. 6c) tests do not show any patch-like morphologies. To understand the chemical composition details of the tribo-surfaces, EDS (energy dispersive spectroscopy) analyses were conducted. Fig. 7a and d, respectively, show typical EDS spectrums taken on the tribo-surfaces from the mineral oil and mineral oil and water lubricants (shown in Fig. 6a–c). A comparison of Fig. 7a and d suggests that an iron oxide layer is generated on the tribo-surface of the bearing tested in mineral oil and water, whereas in the case of the mineral oil, the tribo-surface has no layers and the EDS spectrum shows the steel's composition. Quantification of the EDS spectrum in Fig. 7d (mineral oil and water) shows nearly equal atomic percentages of iron and oxygen, suggesting that the formed oxide is in FeO form. Fig. 7b and c are the EDS spectrums taken at different locations of the tribo-surface from the WEC critical oil shown in Fig. 6b. As can be seen in Fig. 7b (taken on the dark area of Fig. 6b), the tribo-surface has a layer comprising Zn, P, S, Ca, Na and O. In fact all of these elements actually existed in the original WEC critical oil, as

3.3. Electron microscopy characterization of the tribo-surface damage To more thoroughly examine the damage shown in Fig. 3b and d,

Fig. 6. SEM images of tribo-surfaces (in the slide zones) of post-test raceways from different oils: (a) mineral oil; (b) WEC critical oil; (c) mineral oil and water.

169

Wear 398–399 (2018) 165–177

M. Paladugu et al.

Fig. 7. EDS measurements taken on tribo-surfaces (in the slide zones) of post-test raceways from different oils as shown in Fig. 6: (a) mineral oil; (b) thicker tribo-film from WEC critical oil, dark areas in Fig. 6b; (c) thinner tribo-film from WEC critical oil, bright areas in Fig. 6b; (d) mineral oil and water.

SEM and SIM studies were conducted on the respective areas. Fig. 9a and b, respectively, show SEM images of a typical line crack (produced in the WEC critical oil) and typical PSO damage (produced in the mineral oil). The directions of the roller's surface during the bearing operation are shown by arrows in the respective images. The damages shown in both Fig. 9a and b were found in the negative slide areas. To further investigate the damage characteristics, FIB milling was performed on the damaged areas shown by the dotted lines in Fig. 9a and b; respective SEM images of FIB trenches are shown in Fig. 9c and d. In the WEC critical oil tests (Fig. 9c), the line crack on the surface extended downward into the material and further propagated into the material at an inclined angle (< 45°) in the direction opposite to the roller direction. In addition, in Fig. 9c, there is a narrower crack branching from or connecting to the inclined part of the crack. In contrast, in the mineral oil tests (PSO damage), the crack from the PSO damage formed in the roller direction and did not propagate downward. To understand the microstructural characteristics around the cracked areas, SIM imaging was conducted. Fig. 9e and f are the SIM images taken from Fig. 9c and d, respectively. In the case of WEC critical oil (Fig. 9e), nano grained structure (WEM) can be seen along the faces of the inclined part of the crack, and no WEM appears on the vertical part of the crack. Conversely, in the case of the mineral oil (Fig. 9f), no contrast due to nanograins (WEM) is present. In the mineral oil and water lubricant tests, the damage was in the form of micro-pits. Similar to the mineral oil results, no microstructural changes were found to be associated with the damage from the tests with mineral oil and water lubricant. Our detailed metallography and electron microscopy investigations showed no formation of WEM in the cases of mineral oil and mineral oil and water lubricants.

Fig. 8. (a) SEM image of tribo-film produced in the WEC critical oil tests. The dotted line indicates the STEM sample location. (b) A STEM image showing the tribo-layer on the steel from the location shown in (a) by the dotted line.

170

Wear 398–399 (2018) 165–177

M. Paladugu et al.

Fig. 9. (a) and (c) are SEM images of a line crack before and after FIB milling, respectively; the line crack was generated in the negative slide zone of the raceway tested in WEC critical oil. (e) SIM image of the location shown in (c). (b) and (d) are SEM images of PSO damage before and after FIB milling, respectively; the PSO damage was generated on the tribosurface tested in mineral oil. (f) SIM image of the location shown in (d).

Fig. 10. (a) SEM image of a line crack generated in the positive slide zone of the raceway that was tested in the WEC critical oil. (b) SIM image of an FIB milled trench along the dotted line shown in (a).

As shown in the SEM images in Fig. 11a and b, the surface cracks formed in the WEC critical oil have a wide bright area adjacent to each crack. This can be clearly seen in optical images (for example, Fig. 3b). In fact, all the observed line cracks have this kind of morphology adjacent to each crack. As discussed in Fig. 8, this bright appearance implies a thinner tribo-film. Based on this evidence, we expect that this bright region adjacent to each crack forms as the tribo-film wears out during the crack formation stage. It is worthwhile to mention that, instead of originating at the surface and propagating downward, it may be possible for the cracks to form in the subsurface and propagate toward the surface. If the cracks form in the subsurface first and move toward the surface, one would expect to see irregular crack networks on the tribo-surface. The fact that the cracks have line-like morphologies and are formed along the contact line suggests that they formed on the surface first and propagated downward into the subsurface. In order to further verify this

3.3.1. Electron microscopy characterization of the tribo-surface damage from the WEC critical oil We studied the formation characteristics of WEM in the WEC critical oil by conducting detailed SEM and SIM investigations on the line cracks produced in the positive slide zone (refer to Fig. 3a). Fig. 10a shows an SEM image of a line crack produced in the positive slide zone. As with the negative slide zone (Fig. 9a, c and e), FIB milling and subsequent SIM imaging were conducted on the line cracks (as shown by the dotted line). Fig. 10b shows an SIM image of an FIB milled surface crack in which the crack extended downward into the material and further propagated on an angle into the material in the direction of the roller. It is of interest to note that – in contrast to the WEM seen along the crack faces in the negative slide zone (Fig. 9e) – in the positive slide zone (Fig. 10b), no WEM is present along the crack. We have repeatedly observed this difference between the positive and negative slide regions in our electron microscopy observations. 171

Wear 398–399 (2018) 165–177

M. Paladugu et al.

Fig. 11. (a)–(c) are SEM images of line cracks generated in the negative slide zone of raceway that was tested in the WEC critical oil. (d) SEM image of an FIB milled trench along the dotted line shown in (c).

Fig. 12. (a) A schematic diagram showing a line crack on the tribo-surface and its propagation downward vertically and then on an incline. Metallographic observation (parallel to the contact line) away from the line crack shows a subsurface crack with no connection to the surface. (b) A metallograph showing a white etching crack (WEC) with no connection to the surface; the sample is prepared from the raceway tested in the WEC critical oil. WECs were seen in the negative slide region.

3.3.2. STEM characterization of microstructure deformation on the tribosurfaces Next, we conducted FIB milling and SIM and STEM imaging at the locations near cracks (Fig. 11d) to examine the microstructure details. Fig. 13a shows an SIM image of the FIB milled area shown in Fig. 11d. It is of interest to note from Fig. 13a that the microstructure near the surface appears to be fine or fragmented compared to the microstructure in the lower regions. In fact, this relatively fine and fragmented microstructure suggests deformation damage to the material [13], and the difference can be clearly observed in the STEM images. Fig. 13b shows a low-magnification STEM image of the region. The microstructure in the upper region (4–5 µm deep, shown in Fig. 13c) appears to be much finer than the microstructure underneath (Fig. 13d). This further confirms the significant deformation damage in the near-surface region of the raceway. To evaluate the extent of microstructure damage in the mineral oil tests, a location was chosen near the PSO damage, as shown in Fig. 13e, and respective STEM investigations were conducted. As it can be seen in Fig. 13f (a low-magnification STEM image), the top ~ 2 µm of the microstructure is fragmented due to the deformation damage. This difference can be clearly seen by comparing Fig. 13g–h. As shown in Fig. 13b and f, the surface layers from the WEC critical oil tests (40 h of life) appeared to have deeper damage (finer microstructure) than the mineral oil tested layers (980 h of life). To further verify this deformation damage, untested

hypothesis, we chose several locations that appeared ready to crack based on the bright appearance shown in Fig. 11a and b, and analyzed them using FIB milling. Fig. 11c shows an SEM image of one of those locations; line cracks can be observed in the brighter areas at the bottom and top (shown by the rectangles and respective inset images). No cracks can be seen in the brighter area in the middle, so FIB milling was carried out in this region (as shown by the dotted line) to see whether any cracks had formed in the subsurface region. Fig. 11d shows an SEM image of the FIB milled subsurface, and one can see that no cracks were found in the subsurface area. To verify this observation, additional FIB milling and SEM investigations were conducted on several locations similar to Fig. 11c. The subsurface cracks were observed to be present only when there were line cracks on the surface. These results suggest that the line cracks form on the tribo-surface and propagate downward on an incline (Fig. 9c and e), deep into the subsurface region. Our detailed microscopy investigations suggested that, because of this inclined propagation of the cracks (as shown in schematic Fig. 12a), when the washers (raceways) were cut along the contact line for metallographic observations, WECs were observed in the subsurface region without a connection to the surface in the plane of view (Fig. 12b).

172

Wear 398–399 (2018) 165–177

M. Paladugu et al.

Fig. 13. (a) SIM image of the FIB milled location shown in Fig. 11(d). (b) STEM image showing the microstructural details near the surface of the material shown in Fig. 11(d). (c) and (d) are STEM images from the top and bottom regions, respectively, as indicated in (b). (e) SEM image of a tribo-surface tested in mineral oil. Image shows PSO damage initiated from a grind line. The location of the STEM sample is shown by the dotted line. (f) STEM image showing microstructural details in the near-surface region from the location shown in (e). (g) and (h) are STEM images from the top and bottom regions, respectively, as indicated in (f).

formation of tribo-film are the reason for the manifestation of the bright rings in the slide zones (Figs. 3a and 3c). In the mineral oil and water lubricant, the water reacted with the steel surface and created an iron oxide layer as the tribo-film. For the experiment, we set the temperature of the test at 50 °C to minimize water evaporation from the lubricant. However, because of the heat from the rotation of the loaded bearings, the temperature of the lubricant rose to around 95 °C. To determine the amount of water dissolved in the oil, the lubricant was collected after the test and analyzed by the Karl Fischer method. The measured amount of water was about 3%. In addition, a quantity of water was observed to have segregated to the bottom of the oil tank due to immiscibility and density differences. Since the lubricant temperature was about 95 °C, the oil was expected to be further enriched by the vapor from the underlying segregated water. Therefore, during the bearing test, the actual amount of water in the oil would have been higher than the measured 3%. Also, since the lubricant temperature was raised from 50 °C to 95 °C by the heat delivered from the bearing rotation, the actual temperature at the line of contact would have been much higher than the lubricant temperature. The fact that the oxide layer formed only in the slide zone confirms that the water in the mineral oil reacted with the tribo-surfaces and formed an iron oxide layer. Therefore, it is logical to conclude that, in the presence of water molecules, the high sliding forces at the edges of the contact line and the associated shearing of the local surface asperities

raceways were characterized in similar way and observed undeformed microstructures at the surface of untested raceways (similar to the bulk microstructure, as the one shown in Fig. 13h. 4. Discussion 4.1. Formation of tribo-layers Within the test parameters used in the FE-8 test rig, the calculated oil film thickness was about 0.13 µm at the rolling contact [29], and the average surface roughness (Ra) of the rollers and raceways was between 0.07 and 0.09 µm. These parameters generated a mixed lubrication condition (lambda ratio of 1.15) at the contact. As a consequence, the high sliding conditions generated toward the ends of the contact line (Fig. 1a schematic) caused shearing and possible wear of the asperities (or solid-solid contacts) on the contacting surfaces. The considerable wear was seen in the sliding zones in the case of the WEC critical oil (Fig. 4) suggests that the shearing forces are considerably strong in the WEC critical oil. In addition, the fact that the tribo-layers (discussed in Figs. 5–7) are formed only in the high sliding areas actually suggests that the physical contact of the asperities and associated surface shearing forces produces sliding-induced tribo-chemical reactions between the lubricant and the contacting surfaces. This preferential shearing of asperities (or solid-solid contacts) and simultaneous 173

Wear 398–399 (2018) 165–177

M. Paladugu et al.

rollers (seen in Figs. 3d and 9b). In fact, PSO damage is the most common type of damage mode in bearings. In this study, mixed lubrication conditions combined with sliding occurred along the contact line. Mixed lubrication conditions combined with sliding promote asperity contacts on the contacting surfaces. Therefore, it is reasonable to expect the generation of PSO damage after several hundred hours of loaded bearing operation. In agreement with the typical characteristics of PSO damage (arrowhead shape), as shown in Figs. 9b and 9d, the damage was shallow at its origin point, and then broadened and propagated in the rolling direction [28]. The water in the mineral oil and water lubricant promoted the oxidation and corrosion of the raceway. Therefore, it is reasonable to assume that the corrosion caused the micro-pits formation. In addition, when water is mixed with mineral oil, the resultant lubricant tends to have a lower viscosity and hence lower film thickness. Therefore, it is logical to expect damage to the raceway surface much earlier than in the case of pure mineral oil. In contrast to the PSO damage incurred with the mineral oil, the raceways tested in the WEC critical oil have line cracks along the contact line (Figs. 3b, 9a, 10a, 11 a and c). From the surface morphology (top-view) of the line cracks, these cracks appear to be similar to the reported axial cracks in wind turbine bearings [33–35]. As demonstrated in Figs. 9–12, these line cracks originated at the surface and propagated downward vertically first and then further on an incline. This cracking behavior was not found when the other lubricants were used. As can be seen in Figs. 9a and 9c (negative slide zone), the crack propagated on an incline in the opposite direction of the roller direction. In contrast, in the positive slide zone (Figs. 10a and 10b), the crack propagated in the direction of the roller. These kinds of crack propagation are schematically shown in Fig. 14a and b. In the negative slide zone, the velocity of the roller's surface is higher than the velocity of the raceway's surface. As a consequence, the traction force is applied on the raceway in the same direction as it's moving direction (velocity direction) [36]. In Fig. 14a, this is schematically shown by an arrow, Ftraction. In the positive slide zone, Fig. 14b, the velocity of the roller's surface is lower than the velocity of the raceway's surface; this result in traction force applied on the raceway in opposite direction to the raceway moving direction [36]. This traction force is schematically depicted in Fig. 14b. This analysis indicates that, in both the positive and negative slide zones, cracks are incline-oriented in the respective directions of the traction forces applied on the raceway. This result further suggests that the traction forces applied on raceways (from the roller's surface) play a decisive role in influencing the direction of crack orientation. Because of domination of surface forces, in their initial stages of generation, cracks tend to propagate vertically from the surface, and as the cracks reach the subsurface region, the cracks tend to incline and also branch in other directions under the influence of subsurface shear stresses (Figs. 9c and 9d). In the current bearing tests, maximum orthogonal and maximum unidirectional shear stresses occur in the depths of about 90 µm and 140 µm respectively beneath the surface of the raceway. As illustrated in Fig. 13, significant microstructure deformation occurred in the near-surface region of the sliding zones, suggesting that significant surface shear stresses were exerted by the sliding conditions. The deeper surface damage and wear in the slide zones (Fig. 4) indicates that the stresses were higher in the tests conducted with WEC critical oil. Because cracks are initiated along the contact line and are vertically propagated downward into the material (10–20 µm from the surface) only in the case of WEC critical oil, we posit that these cracks are initiated by the significantly higher shearing forces applied on the tribo-surface in the sliding zones. Since the tribo-film is the major difference between the WEC critical oil and the mineral oil, we conclude that the high shearing forces are most likely caused by the tribo-film formed from the additives in the WEC critical oil [31]. An iron oxide non-uniform tribo-film was observed in the mineral oil and water lubricant tested surfaces; however, the line cracks and premature bearing

Fig. 14. (a) and (b) are schematic diagrams showing the negative and positive sliding regions, respectively; in each case, orientation of crack with respect to the traction force acting on the raceway can be seen. The WEM matter formed along the crack is shown in the negative slide region.

(or solid-solid contacts) promoted the oxidation reaction in those regions. Based on the stoichiometry of the formed tribolayer, the possible oxidation reaction can be written as H2O + Fe = FeO + H2. The tribo-layer composition analysis of the WEC critical oil (Figs. 6b, 7b and 7c) revealed the presence of elements from the oil additives (Table 1). Table 1 shows that, compared to the mineral oil, the WEC critical oil has significantly higher amounts of Zn, Ca, S, P, Na and B. Since the WEC critical oil is commercially available oil, we do not know exactly which additive chemicals it contained. However, as is known from the literature [30], these elements come from anti-wear (AW) and extreme pressure (EP) additives; detergent additives; oxidation, corrosion and rust inhibitors; etc. Formation of a tribo-layer only in the slide zones containing the additive elements of the WEC critical oil further confirms that the surface shearing forces in the slide zones caused the tribo-layer. In addition, as shown in Fig. 8, the tribo-layer is not uniform and has significant thickness variations along the sliding direction. This kind of zinc dialkyl-dithiophosphate (ZDDP)-based tribo-film has been reported to increase surface traction forces in rolling-sliding test conditions [31]. The formation of tribofilm in the slide zones is consistent with previous reports [32,33] in literature, where Ca-P tribofilm and Zn-S tribofilm are shown. In this report, the tribo films from the WEC critical oil have most of the additive elements in the oil, including Zn, P, S, Ca, Na and O as shown in the EDS spectrums. Figs. 5a, 6a and 7a show that in the mineral oil tests, no tribo-layer can be found in the SEM and optical microcopy characterizations. But since the same mixed lubrication conditions existed along the contact line, one would expect an iron oxide tribo-layer to be formed in these experiments as well, even though the mineral oil had no significant additives and water contamination. We believe that this oxide tribolayer might have formed but is too thin to be detected by the measuring sensitivity of the EDS and SEM techniques. 4.2. Formation of PSO damage and line cracks As discussed in Figs. 3c and 3d, PSO damage occurred on the bearing races that were tested in the mineral oil. PSO damage initiates on a tribo-surface when the stress concentration applied on an asperity contact is high enough to cause plastic deformation of the asperity and physical removal of material at that contact. Once the material removal is initiated, and as the stress cycles progress, the damage broadens and the material is pushed by the rolling elements in the direction of the 174

Wear 398–399 (2018) 165–177

M. Paladugu et al.

corrosive environments – Salt Spray Corrosion Test as per ASTM B117 standard for 48 h and 100 h [48]. With both the test durations, WEC critical oil applied bearings were found to be significantly more corrosion resistant than the mineral oil applied bearings. Based on this result, corrosion induced electrochemical reactions in the WEC critical oil are not as aggressive as in the case of mineral oil. Since atomic hydrogen evolution is result of electrochemical reactions, this result indirectly suggest that, if at all any atomic hydrogen evolution occurs from the lubricant oils during the bearing tests, hydrogen generation from the WEC critical oil will not be considerably high compared to the mineral oil. These analyses strongly suggest that the hydrogen intrusion is not a cause for line-cracking (early damage from WEC critical oil), and it further reinforces the hypothesis that the line cracking is result of high surface shear stresses; in this report, these high surface shear stresses are caused by the tribo-film produced from the WEC critical oil. Based on explained analyses, it is logical to believe that the location of the line crack and time to generate a line crack are determined by the local intensity of asperity contacts at that location and the respective magnitude of the surface shear forces. The line cracks generated earlier have more chances of propagating into a macro-spall (shown in Fig. S2 of Supplementary material) and cause the vibration increase from the test rig; this vibration increase ultimately stop the bearing's life test. In addition, based on the discussion in the manuscript, the line-cracks generated on the positive slide zone (crack faces tends to move away from each other) have more chances of propagating into the macrospall. In agreement to this analysis, we observed more spalls on the positive slide region, and some examples are shown in the images in Fig. S2 of the Supplementary material. Based on these analyses, the life test in the FE-8 rig with the WEC critical oil is actually an aggressive, highly accelerated RCF test with high surface shear forces.

damage were not seen in this case – implying that an FeO layer by itself cannot cause the surface cracking and consequent early failures. This result further suggests that the composition of tribolayer play significant role in resulting surface shear forces [31]. Some studies in the literature used similar WEC critical oils and hypothesized that the hydrogen liberated at the tribo-contact caused the cracks [11,20,37]. The moisture content of the WEC critical oil was put forth as the source of the hydrogen [37]. As we know, atomic hydrogen liberation from water containing lubricant oils is indirectly measured using in-situ current density measurements [25,26]. The water content in the mineral oil and water lubricant (3 wt%) is much higher than the water content in the WEC critical oil (4174 ppm). Therefore, if hydrogen is the reason for the cracking, one would expect similar cracking and early failures in the case of mineral oil and water. The influence of hydrogen diffusion into steel and the consequent embrittlement phenomenon have been well documented for more than half a century [38]. In this phenomenon, atomic hydrogen diffuses into the steel and segregates in defect sites and interfaces and cause embrittlement when the atomic hydrogen tends to combine into molecular hydrogen. In the manufacturing industry, the problem of hydrogen intrusion into steel is often encountered in the electroplating step when the process parameters are not controlled as required. In such cases, after electroplating, heating the electroplated components to about 150 °C liberates the hydrogen from the steel [39,40]. This is a wellknown and proven practice in the manufacturing industry and is known to work consistently. In the experiments described here, the bearing test temperature was set at 100 °C, which means the entire test rig temperature was maintained at 100 °C through external heating (the heating coils shown in Fig. 1c). In addition, in the case of the WEC critical oil, heat from the bearing rotation added to this temperature and caused the bearing temperatures to increase to about 110-120 °C (as per the thermocouples attached to the stationary washers’ outer diameters). As evidenced by this temperature raise, the actual temperature at the rolling contact line would have been much higher than 120 °C. This analysis suggests that even if any hydrogen intruded into the steel, the chances of retention in steel (by segregation at interfaces in the microstructure of steel) would be minimal due to the high temperature at the tribo-contact and the steel. In a recent report [37] in literature, thermal desorption spectroscopy (TDS) results of WEC critical oil tested post-test rollers from FE-8 rig was shown. Temperature of the rollers was raised at a rate of 10 °C/ min and hydrogen desorption was measured simultaneously. The hydrogen started to evolve sharply around 300 °C and no hydrogen evolution was noticed below 200–250 °C. TDS studies of hydrogen liberation from steels were very well reported in literature [41–45]. In all these reports, the hydrogen tends to liberate from steel in the temperature range of 100–150 °C. In fact, this temperature range is consistent with the hydrogen liberation treatment used in the manufacturing industries mentioned above. In contrast, the hydrogen liberation from the FE-8 tested rollers occurs in the range of 300 °C [37]; it may be possible that the hydrogen liberation from the WEC critical oil tested FE-8 CRTB components was associated with the decomposition (or modification) of tribolayers and its associated organic constituents present on the surfaces [11,37]. It is worthwhile to mention that, in literature [5], the calcium sulfonates (additive) in the WEC critical oil is assumed to act as a poisoner for recombination of atomic hydrogen and thus promote atomic hydrogen generation. In fact, in the studies relating to sulfide stress cracking and stress corrosion cracking, sulfur was thought as a poisoner for hydrogen recombination, mainly in the presence of corrosive environments (in specific, associated electrochemical reactions) and H2S [46,47]. However, how calcium sulfonates promote the atomic hydrogen generation and evidence for associated electrochemical reactions are not demonstrated in the literature. In this context, to verify the corrosivity of the WEC critical oil in comparison to the mineral oil, respective oils were applied on fresh bearings and subjected them to

4.2.1. Verification of hypothesis relating to the WEC critical oil To verify the role of the tribo-film, additional experiments were conducted. In these experiments, the raceways were coated with a wear-resistant and self-lubricated diamond-like carbon (DLC)-based coating [49] and the raceways were tested in the WEC critical oil with the same test parameters used in the original experiment. The early failures of raceways were not observed, and analysis of the post-test raceways showed that the DLC layers were intact on the surfaces of the raceways, and no tribo-film was found on the raceways. More detailed testing and investigations are in progress to address the benefit of the DLC coating in this context. In contrast to the DLC-coated raceways, black oxide-coated raceways [50] did not show any improvement in or benefit to bearing life. Investigations on post-test bearings showed that the black oxide layers were worn off in the sliding zone, and the tribofilm and the line cracks were observed as discussed in this report. In another set of experiments, WEC critical oil was used in bearing tests where no sliding was involved [21]. In the absence of sliding, no significant tribo-film was observed; hence, premature bearing failures were not observed. In fact, in the absence of sliding conditions, bearing life in the WEC critical oil was observed to be longer than the bearing life in the mineral oil. 4.3. Formation of white etching matter along cracks in the case of WEC critical oil As illustrated in Figs. 9a, 9c, 9e, white etching matter is formed along the crack in the negative slide zone of the raceways tested in the WEC critical oil. It is of interest to note that the WEM formed only in the inclined portion of the crack, and not in the vertical portion of the crack (top 10–20 µm). In contrast, in the positive slide zone (Fig. 10b), no WEM is formed along the crack. This difference can be explained as follows: Fig. 14 shows that cracks tend to orient in the direction of the traction forces applied on the raceways. In the negative slide zone, since the direction of traction 175

Wear 398–399 (2018) 165–177

M. Paladugu et al.

5. Conclusions

force is in the same direction as the raceway movement (Fig. 14a), upon each load cycle, the crack faces move away (driven by traction force) and come closer (influence from raceway movement). This rubbing of the crack faces generates a nanograined structure along the crack faces, and this nanograined structure is the WEM. Some reports in literature propose the rubbing as a possible mechanism for WEM formation [51–53]; the results shown in this report provide an evidence to this mechanism [12,13]. As can be seen in Fig. 9e, no WEM is formed along the vertical portion of the crack, and WEM is observed only along the inclined portion of the crack. This result further supports the hypothesis that the crack faces are rubbing, since the vertical portion does not have sufficient rubbing action under the influence of surface and subsurface shear forces. In contrast, in the case of positive sliding, since the direction of traction force is in opposite direction to the raceway movement (Fig. 14b), the crack faces tends to move away from each other and the chances of crack faces to come close and rub are less. As a consequence, no WEM is present along the crack faces in the positive slide zone. The differences in the characteristics of the line cracks generated in positive and negative slide zones were confirmed by characterizing at least five cracks in each slide zone in the similar manner shown in this manuscript. The morphological similarity of line cracks (shown in this report) with the axial cracks in field returned wind turbine bearings [33-35] suggests that the formation of axial cracks is associated with high surface shear forces and sliding generated in the bearings of wind turbine gear boxes during their service. It may be worthwhile to discuss the following. In literature [5], it was postulated that i) the crack faces in negative slide region would rub each other and generate nascent steel surfaces; whereas, in the positive slide region, the crack faces would not have nascent surfaces (lack of rubbing); and ii) the nascent steel surfaces promote the generation of atomic hydrogen in negative slide region and cause WEM. In fact, there is no evidence available in literature relating to “how a nascent steel surface alone catalyze the atomic hydrogen generation”. If a nascent steel surface is the one that promote the atomic hydrogen generation, the hydrogen generation is not possible at the tribo-contact as there was significant tribofilm present at the tribo-contact. If sulfur is the element that promotes the hydrogen generation, one would expect to see sulfur contamination along the crack faces in negative slide zone. Our detailed electron microscopy (FIB and SEM) investigations did not evidence any sulfur along the crack faces. In addition, if lubricant oil is the source of hydrogen in the presence of nascent steel surface, oil needs to go into the nanometer scale crack opening (Fig. 9e) and wet the crack faces (if surface tension forces of oil and crack faces allow it). We have not seen any remnants of oil or its additive elements in these cracks in our detailed FIB and electron microscopy investigations. Therefore, there is no evidence to suggest that the hydrogen is generated in the cracks through oil and crack faces interaction (or) hydrogen is the cause for cracking. Our detailed FIB milling and SIM images showed that in some instances, WEM appeared to exist without the associated crack in the plane of view. This may lead to misinterpretation of the result that WEM generates first and the crack comes next. However, further FIB milling for a few microns in the viewing direction revealed the cracks associated with the WEM. If WEM causes the cracking, one should expect to see WEM in the positive slide zone as well. The fact that no WEM is associated with the cracking in the positive slide zone suggests that WEM is a consequence of the cracking and associated rubbing phenomena, rather than the cause. In addition, because of this rubbing phenomenon, WEM was seen on both the crack faces as well as on one of the two crack faces. The related example images are shown in Fig. S4 of the Supplementary material. Based on the above explained analyses, the location of WEM on the crack faces is related to which of the crack faces undergone the severe plastic deformation during the crack faces movement towards each other and consequent rubbing.

In this report, CRTBs were tested in different lubricant oils in rolling-sliding conditions. The oils used were: 1) commercially available SAE 75W-80 gear oil – called “WEC critical oil”, which contains significant amounts of oil additives; 2) mineral oil without additives; and 3) mineral oil without additives plus water. We determined the lifetimes of bearings in different oils and accounted for the associated damage and its respective mechanisms. A detailed examination of the post-test bearings revealed the following conclusions. 1) Bearing life in the WEC critical oil is less than 5% of life in the mineral oil. When water is mixed with the mineral oil, bearing life is half that of the mineral oil. 2) Different damage modes were observed in the sliding zones with respect to the oils used. WEC critical oil produced micro-scale line cracks on the respective tribo-surfaces along the contact line. In contrast, use of mineral oil produced typical arrowhead-shaped PSO damages. Micro-pits were observed with the use of the mineral oil and water lubricant. 3) The WEC critical oil produced a tribo-film in the sliding zones composed primarily of the additive elements of the oil. In the case of “mineral oil and water” lubricant, water oxidized the tribo-surface and caused an iron oxide tribo-film. 4) The WEC critical oil generated micro sized line cracks (along rolling contact line in the slide zones) that propagated downward vertically and then further on an incline. Only the cracks generated in the negative slide region had white etching matter along the crack faces (WECs). No line cracking or associated WEM was observed in the cases of the mineral oil and mineral oil and water lubricants. 5) While the line cracking was not observed in the tests with mineral oil, typical damage mode was point surface origin damages. In the case of mineral oil and water lubricant, the primary damage mode was oxidation induced micro-pits. 6) Based on the crack morphologies and the crack orientations with respect to the roller movement and applied traction forces, in the case of the WEC critical oil, the line cracks generation is attributed to the high surface shear forces applied on the contacting surfaces. The high surface shear forces are caused by the tribo-film formed in the sliding zones from the oil additives of the WEC critical oil. The morphological similarity between the line cracks (shown in this report) and the axial cracks in bearings of wind turbine gear boxes suggests that the formation of axial cracks is associated with high surface shear forces and sliding. 7) In the negative slide region, since the traction force is applied in the direction of raceway movement, upon each load cycle, the crack faces rub each other and caused WEM formation along the cracks. 8) In the case of positive slide, the applied traction force has opposite direction to the raceway movement and hence the crack faces tend to move away rather than rubbing; as a consequence, no WEM was formed in the positive slide region. Acknowledgments We gratefully acknowledge Stephen P. Johnson, Director of The Timken Company R&D department, for allowing us to publish this paper and commenting on the manuscript. In addition, we acknowledge Jerry P. Rhodes, General Manager of Engineering Fundamentals & Physical Testing department of Timken Company for making useful comments on the manuscript. In addition, the following associates of TIMKEN-WHQ are gratefully acknowledged: i) Dr. Carl H. Hager, Mr. James R. Gnagy, and Dr. William Hannon for their discussions relating to tribology fundamentals; ii) Mr. Gerald A. Richter and Mr. Jeremy S. Kimble for assisting with bearing tests on the FE-8 test rig; iii) Mr. Matthew R. Boyle and Mr. Robert A. Pendergrass for assisting with optical metallography and electron microscopy experiments 176

Wear 398–399 (2018) 165–177

M. Paladugu et al.

respectively; iv) Mr. Mai Lee for assisting with the profilometer measurements.

[25] B. Han, J. Binns, I. Nedelcu, In situ detection of hydrogen uptake from lubricated rubbing contacts, Tribol. Online 11 (2016) 450–454. [26] E.A. Esfahani, A. Morina, B. Han, I. Nedelcu, M.C.P. van Eijk, A. Neville, Development of a novel in-situ technique for hydrogen uptake evaluation from a lubricated tribocontact, Tribol. Int. 113 (2017) 433–442. [27] R.E. Cantley, The effect of water in lubricating oil on bearing fatigue life, ASLE Trans. 20 (1977) 244–248. [28] M. Kotzalas, Rolling contact fatigue wear, in: Handbook of Lubrication and Tribology, Volume II, CRC Press, 2012, pp. 1–12. [29] A. Ruellan, X. Kleber, F. Ville, J. Cavoret, B. Liatard, Understanding white etching cracks in rolling element bearings: formation mechanisms and influent tribochemical drivers, Part J: Proc. Inst. Mech. Eng. J. Eng. Tribol. 229 (2015) 886–901. [30] L. Deters, Tribology, in: K.-H. Grote, E.K. Antonsson (Eds.), Springer Handbook of Mechanical Engineering, Springer Berlin Heidelberg, Berlin, Heidelberg, 2009, pp. 295–326. [31] L. Taylor, A. Dratva, H. Spikes, Friction and wear behavior of zinc dialkyldithiophosphate additive, Tribol. Trans. 43 (2000) 469–479. [32] J. Franke, W. Holweger, H. Surborg, T. Blass, J. Fahl, T. Elfrath, T. Abe, D. Merk, Influence of tribolayer on rolling bearing fatigue performed on a FE8 test rig, in: TAE 19th international colloquium tribology, Ostfildern, Germany, 2014, 1–13. [33] Ruellan, Tribological Analysis of White Etching Crack (WEC) Failures in Rolling Element Bearings (Ph.D. thesis), INSA de Lyon, Lyon, France, 2015. [34] a) J. Gegner, Tribological aspects of rolling bearing failures, in: C.-H. Kuo (Ed.), ‘Tribology-lubricants and Lubrication, InTech, Rijeka, 2011, pp. 33–94; b) J. Gegner, W. Nierlich, Service loading analysis of wind turbine gearbox rolling bearings based on X-ray diffraction residual stress measurements, Mater. Sci. Forum 768–769 (2014) 723–732. [35] H. Uyama: The mechanism of white structure flaking in rolling bearings, in: NREL Wind Turbine Tribology Seminar, Renaissance Boulder, Flatiron Hotel, Broomfield, CO, USA, November 15–17, National Renewable Energy Laboratory (NREL), 2011. [36] N. Soda, T. Yamamoto, Effect of tangential traction and roughness on crack initiation/propagation during rolling contact, ASLE Trans. 25 (1982) 198–206. [37] J. Franke, J.T. Carey, S. Korres, T. Haque, P.W. Jacobs, J. Loos, W. Kruhoeffer, White etching cracking — simulation in bearing rig and bench tests, Tribol. Trans. (2017) 1–11. [38] M. Devanathan, Z. Stachurski, W. Beck, A technique for the evaluation of hydrogen embrittlement characteristics of electroplating baths, J. Electrochem. Soc. 110 (1963) 886–890. [39] M.F. Stevenson Sr., Cadmium Plating, Surface Engineering, Vol 5, ASM Handbook, ASM International, 1994, pp. 215–226. [40] G. Krauss, Steels: Processing, Structure, and Performance, ASM International, Materials Park, OH, USA, 2005, pp. 469–470. [41] R. Silverstein, D. Eliezer, Mechanisms of hydrogen trapping in austenitic, duplex, and super martensitic stainless steels, J. Alloy. Compd. 720 (2017) 451–459. [42] D. Pérez Escobar, T. Depover, E. Wallaert, L. Duprez, M. Verhaege, K. Verbeken, Thermal desorption spectroscopy study of the interaction between hydrogen and different microstructural constituents in lab cast Fe–C alloys, Corros. Sci. 65 (2012) 199–208. [43] D. Pérez Escobar, K. Verbeken, L. Duprez, M. Verhaege, Evaluation of hydrogen trapping in high strength steels by thermal desorption spectroscopy, Mater. Sci. Eng.: A 551 (2012) 50–58. [44] S. Frappart, A. Oudriss, X. Feaugas, J. Creus, J. Bouhattate, F. Thébault, L. Delattre, H. Marchebois, Hydrogen trapping in martensitic steel investigated using electrochemical permeation and thermal desorption spectroscopy, Scr. Mater. 65 (2011) 859–862. [45] D. Pérez Escobar, T. Depover, L. Duprez, K. Verbeken, M. Verhaege, Combined thermal desorption spectroscopy, differential scanning calorimetry, scanning electron microscopy and X-ray diffraction study of hydrogen trapping in cold deformed TRIP steel, Acta Mater. 60 (2012) 2593–2605. [46] B. Berkowitz, H. Horowitz, The role of H2S in the corrosion and hydrogen embrittlement of steel, J. Electrochem. Soc. 129 (1982) 468–474. [47] A.W. Thompson, I. Bernstein, The role of metallurgical variables in hydrogen-assisted environmental fracture, in: Advances in Corrosion Science and Technology, Springer, 1980, pp. 53–175. [48] ASTM, B117-16 Standard Practice for Operating Salt Spray (Fog) Apparatus, ASTM International, West Conshohocken, PA, 2016. [49] T. Scharf, M. Romanes, K. Mahdak, J. Hwang, R. Banerjee, R. Evans, G. Doll, Atomic-scale structure and composition of tungsten carbide reinforced diamondlike carbon films, Appl. Phys. Lett. 93 (2008) 151909. [50] R.D. Evans, C.H. Hager Jr, Y.S. Kang, G.L. Doll, Comparison of black oxide and tungsten carbide-reinforced diamond-like carbon (WC/aC: H) surface treatments for rolling element bearings, Tribol. Trans. 58 (2015) 444–453. [51] B. Gould, A. Greco, K. Stadler, E. Vegter, X. Xiao, Using advanced tomography techniques to investigate the development of White Etching Cracks in a prematurely failed field bearing, Tribol. Int. 116 (2017) 362–370. [52] H.K.D.H. Bhadeshia, W. Solano-Alvarez, Critical Assessment 13: elimination of white etching matter in bearing steels, Mater. Sci. Technol. 31 (2015) 1011–1015. [53] Y. Kadin, M.Y. Sherif, Energy dissipation at rubbing crack faces in rolling contact fatigue as the mechanism of white etching area formation, Int. J. Fatigue 96 (2017) 114–126.

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at doi:10.1016/j.wear.2017.12.001. References [1] M.N. Kotzalas, G.L. Doll, Tribological advancements for reliable wind turbine performance, Philos. Trans. R. Soc. Lond. A: Math. Phys. Eng. Sci. 368 (2010) 4829–4850. [2] J. Lai, K. Stadler, Investigation on the mechanisms of white etching crack (WEC) formation in rolling contact fatigue and identification of a root cause for bearing premature failure, Wear 364 (2016) 244–256. [3] R. Errichello, R. Budny, R. Eckert, Investigations of bearing failures associated with white etching areas (WEAs) in wind turbine gearboxes, Tribol. Trans. 56 (2013) 1069–1076. [4] M. Evans, White structure flaking (WSF) in wind turbine gearbox bearings: effects of ‘butterflies' and white etching cracks (WECs), Mater. Sci. Technol. 28 (2012) 3–22. [5] M.H. Evans, An updated review: white etching cracks (WECs) and axial cracks in wind turbine gearbox bearings, Mater. Sci. Technol. 32 (2016) 1133–1169. [6] B. Gould, A. Greco, Investigating the process of white etching crack initiation in bearing steel, Tribol. Lett. 62 (2016) 26. [7] B. Gould, A. Greco, The influence of sliding and contact severity on the generation of white etching cracks, Tribol. Lett. 60 (2015) 29. [8] W. Holweger, M. Wolf, D. Merk, T. Blass, M. Goss, J. Loos, S. Barteldes, A. Jakovics, White etching crack root cause investigations, Tribol. Trans. 58 (2015) 59–69. [9] W. Kruhöffer, J. Loos, WEC formation in rolling bearings under mixed friction: influences and “Friction Energy Accumulation” as indicator, Tribol. Trans. 60 (2017) 516–529. [10] M.-H. Evans, A. Richardson, L. Wang, R. Wood, W. Anderson, Confirming subsurface initiation at non-metallic inclusions as one mechanism for white etching crack (WEC) formation, Tribol. Int. 75 (2014) 87–97. [11] a) A. Richardson, M.-H. Evans, L. Wang, R. Wood, M. Ingram, Thermal desorption analysis of hydrogen in non-hydrogen-charged rolling contact fatigue-tested 100Cr6 steel, Tribol. Lett. 66 (2018) 4; b) A.D. Richardson, M.-H. Evans, L. Wang, R.J.K. Wood, M. Ingram, B. Meuth, The evolution of white etching cracks (WECs) in rolling contact fatigue-tested 100Cr6 steel, Tribol. Lett. 66 (2017) 6. [12] M. Paladugu, R.S. Hyde, White etching matter promoted by intergranular embrittlement, Scr. Mater. 130 (2017) 219–222. [13] M. Paladugu, R.S. Hyde, Microstructure deformation and white etching matter formation along cracks, Wear 390–391 (2017) 367–375. [14] W. Solano-Alvarez, H. Bhadeshia, White-etching matter in bearing steel. Part II: distinguishing cause and effect in bearing steel failure, Metall. Mater. Trans. A 45 (2014) 4916–4931. [15] A.M. Diederichs, S. Barteldes, A. Schwedt, J. Mayer, W. Holweger, Study of subsurface initiation mechanism for white etching crack formation, Mater. Sci. Technol. 32 (2016) 1170–1178. [16] T. Blass, M. Dinkel, W. Trojahn, Bearing performance as a function of structure and heat treatment, Mater. Sci. Technol. 32 (2016) 1079–1085. [17] V. Smelova, A. Schwedt, L. Wang, W. Holweger, J. Mayer, Microstructural changes in White Etching Cracks (WECs) and their relationship with those in dark region (DER) and White Etching Bands (WEBs) due to rolling contact Fatigue (RCF), Int. J. Fatigue 100 (2017) 148–158. [18] DIN 51 819, Testing of Lubricants – Mechanical-Dynamic Testing in the Roller Bearing Test Apparatus FE8–Part-1: General Working Principles, BeuthVerlag GmbH, Berlin, 1999. [19] H.K. Danielsen, F.G. Guzmán, K.V. Dahl, Y. Li, J. Wu, G. Jacobs, G. Burghardt, S. Faester, H. Alimadadi, S. Goto, Multiscale characterization of White Etching Cracks (WEC) in a 100Cr6 bearing from a thrust bearing test rig, Wear 370 (2017) 73–82. [20] F.G. Guzmán, M. Oezel, G. Jacobs, G. Burghardt, C. Broeckmann, T. Janitzky, Reproduction of white etching cracks under rolling contact loading on thrust bearing and two-disc test rigs, Wear 390 (2017) 23–32. [21] J. Gnagy, L. Houpert, F. Chevalier, Rolling bearing stress based life — Part II: experimental calibration and validation, J. Tribol. 134 (2) (2012) 021104. [22] ASTM D6595 – 17, Standard Test Method for Determination of Wear Metals and Contaminants in Used Lubricating Oils or Used Hydraulic Fluids by Rotating Disc Electrode Atomic Emission Spectrometry. [23] ASTM D2622 – 16, Standard Test Method for Sulfur in Petroleum Products by Wavelength Dispersive X-ray Fluorescence Spectrometry. [24] ASTM D6304 – 16e1, Standard Test Method for Determination of Water in Petroleum Products, Lubricating Oils, and Additives by Coulometric Karl Fischer Titration.

177