Understanding the Role of Directional Texture in Tribofilm Evolution

With the advances in the fields of materials and manufacturing sectors, the field of tribology has grown significantly to combat new challenges involving various components, designs, materials, and systems. The development of stronger and harder materials along with tighter tolerances in mechanical systems makes them more prone to direct contact between surfaces and material wear in several industrial applications such as gears, rails, automotive parts, and many more. Additionally, with increasing environmental concerns and the need to conserve depleting resources (by reducing energy and material waste), reducing friction via proper lubrication strategies is crucial. Using oils or greases as lubricants have been a commonly employed lubrication strategy for many years. Specifically, harsher operating conditions (high load, low-speed, and under starvation), those result in a boundary lubrication (BL) regime [1,2], pose a significant challenge as the two mating surfaces are under direct contact in this regime leading to increased frictional losses and material wear.

Typical BL strategies rely on surface protective lubricant films/tribofilms which are formed by chemical reactions between the surface and lubricant additives [2–6]. Additionally, functionally designed additives result in the formation of low shear strength tribofilms leading to effective load support and friction reduction [1]. Lamellar solid lubricant additives (micro and nanoparticles) having weak inter-layer bonding such as MoS₂, WS₂, and graphite have been widely researched as attractive BL strategies due to their excellent properties [7–16]. Apart from using functional nanoparticles as lubricant additives for effective BL, the use of textured surfaces is a widely researched lubrication strategy [17,18]. Textured features such as dimples, grooves, ellipsoids, and multi-scale textures are commonly employed and reported in the literature for lubrication applications [19–28]. However, the majority of the reported work is focused on optimizing a set of texture variables (such as size, shape, and periodicity of features) [19] for a specific tribological system [18]. The evolution of the designed textures with time in context with different lubrication regimes and mechanisms is seldom reported. In other words, texture has often been viewed as a “static” design attribute where the effect of textured surfaces on tribological response is reported under different lubricating conditions. On the other hand, the texture is dynamic due to its inherent nature in controlling various physical and chemical interactions at the mating interfaces, at micron and sub-micron scales. Texture and its evolution also contribute to the tribo-chemical reactions involving the lubricant additives and surface asperities. This dynamic nature of texture and its role in governing tribofilm dynamics is least studied in reported literature and hence is the objective of this paper.

An earlier study by the authors on untextured discs [29] indicated that the texture in the form of asperities evolves along the direction of motion while manipulating the tribofilm formation and thus controlling the friction. These experiments were performed on mirror-finished steel discs lacking any intentional and directional texture. Based on these experiments and results (as a comparison point), this paper reports the results of experiments performed on directionally textured steel surfaces where the direction of texture features aligns with the direction of movement. The evolution of texture and its effect on manipulating the tribofilm properties is discussed. Nano-engineered MoS₂-based lubricant was used as a BL additive because of its excellent tribological properties [30,31] and for consistency with the earlier experiments [29] on the untextured mirror-finished discs. The frictional response was divided into three regions of lubrication contact mechanics, and each region was experimentally studied using time intervals tests. The tribofilm (both physisorbed and chemically reacted) and texture were characterized using different characterization tools to analyze the progression of tribo-chemical and physical events at the tribological interface as discussed below.

Hardened 52100 steel discs (HRC 60) and commercially purchased 52100 steel balls (HRC 60, grade 25, S_q ∼ 0.04 µm, 6.35 mm diameter) were used for the tribological tests. Before the tribological tests, P1200 grit-sized adhesive-backed SiC abrasive paper was used to create a directional concentric texture on the steel discs. A flat chuck with the polishing paper was mounted on a vertical milling machine. The milling machine head was then lowered onto a sample disc held concentrically to fabricate directional texture on the steel disc. The resulting texture had a surface roughness of S_q ∼ 0.16 µm and a texture direction that closely aligns with the direction of circular motion in a rotational tribological test. The textured discs were then used for the experiments without any further processing. Figure 1 shows the optical micrograph (middle image of Fig. 1) of the directional concentric texture and roughness profile (right side of Fig. 1). As shown, even though the texture was directional at the sample scale (as evidenced by the presence of directional grooves), polishing with the P1200 paper resulted in locally random textures at the microscopic scale.

The tribological tests were conducted using a pin-on-disc testing apparatus in rotational mode using a 5 mm track radius. The disc and ball samples were sonicated in acetone and cleaned with isopropyl alcohol before each tribological test on a pin-on-disc tribometer (CSM instruments). A controlled amount of lubricant was applied before starting the tribological tests using disposable transfer pipettes. The tribological tests were conducted with a load of 20 N and a linear speed of 1 cm/s to represent boundary lubrication conditions. The maximum Hertzian contact pressure (ignoring the surface roughness) for the hardened 52100 discs and ball combination at 20 N load is estimated to be 1.65 GPa. MoS₂-based nano-engineered lubricant (NL) was used as a lubricant during the tribological tests to study the role of directional texture on lubricant interaction. The detailed steps for the preparation of hydrocarbon functionalized de-agglomerated MoS₂ nanoparticles (150–200 nm average size), using sequential dry and oil milling steps have been described previously [32,33]. The selection of the lubricant was made based on the earlier reports by the authors where more specific details in terms of the preparation, the dispersion stability of the lubricant, and composition can be found [32–35]. Briefly, the lubricant formulation was prepared by mixing 1% nano-engineered MoS₂ by weight as an anti-friction/anti-wear additive functionalized with 1.5% canola oil (ADM) to a non-formulated poly alfa-olefin base oil (PAO 10 with a kinematic viscosity of ∼ 10 cSt at 100 °C) along with 0.5 wt% lecithin (Alcolec-S) as an emulsifier. The mixture was sonicated before use to ensure a uniform mixture. After the tribological tests, the samples were dip cleaned in hexane solvent to remove excess oil from the wear track before further characterization. The wear track was imaged using a scanning electron microscope (Nova Nanolab) for morphological analysis. The areal texture and wear scar diameter (WSD) were measured using a laser profilometer (Keyence VK-X 250). The tribofilm chemistry and microstructure were analyzed using Raman (Renishaw InVia™ Raman microscope) and cross-sectional transmission electron microscopy (TEM) (FEI titan) techniques. The experimental characterization details are discussed as follows.

The surface texture measurements were carried out using a laser profilometer (Keyence VK-X 250) and analyzed per ISO 25178 texture parameters [36–38]. The measurements were performed with the tribofilm (physisorbed and chemically reacted) as it is part of the surface texture in a tribological contact. Two representative samples were chosen from each time interval for the measurements. On each sample, measurements (20X magnification, 2048 × 1536 pixels size) were taken at four locations on the wear track and analyzed. The areal surface texture parameters of interest were then calculated on the wear track using the MultiFileAnalyzer software. The average and standard deviation values are reported. Table 1 lists and briefly describes the areal parameters of interest that were used to characterize and compare surface texture during the experiments.

Raman spectroscopy and cross-sectional TEM measurement were performed to analyze the chemically reacted tribofilm. Wear tracks from the representative samples were first cleaned with a cotton tip and Hexane solvent to remove the physisorbed lubricant layer before Raman and TEM measurements to ensure only chemically reacted tribofilm was characterized.

Raman measurements were performed at three locations across the wear track and averaged. The background scan was taken on a raw disc with the same measurement settings and subtracted to get corrected Raman spectra. The measurements were taken at 20% laser power, 60 s acquisition time, and three accumulations using a 765 nm laser.

For TEM measurements, the area of interest was first identified using the scanning electron microscopy (SEM)/energy-dispersive X-ray spectroscopy (EDS) mapping technique before preparing the cross-sectional samples. The TEM samples were cut and prepared using the focused ion beam (FIB) technique. To understand the effect of peaks and valleys on tribofilm manipulation, the area of interest was chosen to include a valley and the adjacent peaks. The selected area was first coated with a platinum layer to protect against damage to the underlying tribofilm surface during the ion milling process.

To study the evolution of tribofilm and surface texture, a systematic set of experiments were carried out with different time durations (stop conditions). The specific values for stop conditions were identified based on the results of the control tests. The test was repeated three times (Tests 1, 2, and 3) to test the methodology and repeatability of the tribological trends in the coefficient of friction (COF). The results of the control tests are shown in Fig. 2. For control tests, the testing duration of 10,000 laps (∼8.8 h) was chosen to ensure that a stable, steady-state COF response was obtained. As, the frictional response was divided into three stages of lubrication: (1) region 1 (R1, high starting COF), (2) region 2 (R2, steadily dropping COF), and (3) region 3 (R3, stable steady-state COF). Based on earlier studies [29,34], these regions correspond to changes in the frictional response and the lubrication mechanism across the duration of the test. From the control tests, different time intervals for further detailed tribological tests were chosen to represent the three stages of lubrication. These time intervals (depicted by solid black lines in Fig. 2) are listed in Table 2. Three tests (three different discs) were carried out at each identified time interval (as experiment stop condition) in regions 1 and 2 where the deviation due to sample-to-sample variation was perceived to be more prominent while two tests were carried out at each interval in region 3 where the COF trend under steady-state remained stable. The results of experiments at these identified time intervals are discussed later.

A comparison of the COF trend observed for the control tests using a nano-engineered lubricant on textured discs with untextured discs is shown in Fig. 3. The data for untextured samples are obtained from the earlier study [29] and are used for illustrative comparison. The COF trend for both cases marked distinctive differences, specifically in terms of different lubrication regions of interest. For the untextured discs, the linear drop of coefficient of friction (R2) starts around 2000 laps as compared to around 1000 laps for the textured disc. The steady-state coefficient of friction for textured discs was achieved at a lower time duration (around 6000 laps) as compared to untextured discs (around 6500 laps). This indicated the favorable role of the directional texture in governing the tribofilm properties as studied through this research.

A comparison of selected functional areal texture parameters calculated on the raw discs (without any tribological test) is presented in Fig. 4. As shown, the reduced peak height and reduced valley depth (S_pk and S_vk) values are higher for the textured discs as compared to untextured mirror-polished discs. This indicated that the textured discs have a higher number of large peaks that could potentially get worn off during the initial contact and thus result in higher initial COF values observed in R1 as shown earlier. Similarly, the larger valley depths could allow for effective storage of the lubricant while delivering the surface protective tribofilm layers. This hypothesis was verified with the TEM characterization as discussed later in this paper.

Figure 5 shows the evolution of areal surface roughness (S_q) for directionally textured discs. In R1, the surface roughness increased at the beginning due to the initial re-adjustment of contacting surface asperities along the direction of motion, commonly understood as the break-in phenomenon. This initial re-alignment of surface features was also observed in the SEM images as discussed later. After the initial increase, the surface roughness dropped down as the core surface asperities engaged with each other. This initial rise in friction was attributed to the breaking-in of core asperities. Following R1, the surface roughness (S_q) values showed a slight increase before stabilizing to a steady-state in R3. For comparison, this trend was evaluated with the evolution of surface texture observed with untextured mirror-polished discs (with isotropic texture), which was previously reported [29]. It should be noted that the study on untextured discs was conducted with the same tribological testing conditions using the same composition of the lubricant. A comparison with untextured discs suggests that the textured discs reach a steady-state early as the S_q values are stabilized at an earlier time duration. This was attributed to the ready presence of directional peaks and valleys in the case of textured discs as opposed to isotropic asperities on the mirror-finished untextured discs. Furthermore, the error bars for the textured discs are also small in comparison with the untextured discs. This indicated that the directional texture manipulated the interaction with the lubricant particles in a favorable manner and led to a more predictable lubrication response. For the untextured discs, the synergistic evolution of surface texture and tribofilm was process-controlled and hence stochastic in nature, resulting in more variability in the calculated texture parameter values.

The arithmetic mean peak of curvature (S_pc) parameter was calculated and compared to gain further insights into the texture evolution and its correlation to the tribo-chemical reactions. The S_pc parameter characterizes the sharpness of the peaks within the measurement (see Table 1) area. Higher S_pc values mean sharper peaks while more rounded surface asperities show lower S_pc values. The evolution of mean peak curvature is shown in Fig. 6.

As shown, the peak sharpness values rise in R1 at the start of the test. This was attributed to the direct asperity-to-asperity contact under the BL mode in the running-in period. During this period, the surface asperities are broken-in, deformed, and refined along the direction of rotational motion while leading to the possible onset of tribo-chemical reactions. Similar to the trend observed for S_q, the peak sharpness dropped down in R1, stabilized in R2, and reached steady-state values in R3. This indicated that as the texture and tribofilm possibly evolved together, and the S_pc values are maintained stable due to surface protection offered by the lubricant tribofilm.

It was also noted that the variation between the calculated values as represented by the error bars is higher in R1 because of the stochastic nature of the interaction of asperity-to-asperity engagement. The variation (spread of error bars) decreased in R2 and R3 where the lubrication is likely governed by the tribofilm.

The variation in S_pc values is consistent with the observed trend in S_q (Fig. 5) and could be correlated to the functional parameters shown in Fig. 4. The higher peak heights (as characterized by S_pk values in Fig. 4) result in more friction rise in the early break-in stage which possibly aided the tribo-chemical reactions. Also, the surface asperities were already partially broken-in and aligned in the direction of motion. This behavior is different from the behavior observed on mirror-finished untextured discs [29], where the randomly oriented asperities needed a longer time to re-align and re-organize themselves in the direction of motion as compared to directionally textured asperities. Furthermore, the presence of deeper valleys (Fig. 4, S_vk) in the case of textured discs could also potentially help in the storage of lubricant particles and delivery of tribofilm on the adjacent peaks (which act as anchoring and reaction points). These observations and interpretations were studied and confirmed with the help of chemical analysis of tribofilm and TEM measurements as discussed later.

Figure 7 shows the progression of surface morphology through the three stages of lubrication identified above measured with SEM/EDS. At the beginning of R1 (at 200 laps time duration), the presence of misaligned grooves (not all grooves in the same direction) was noted. This confirms the earlier interpretation that the locally random surface asperities re-align themselves in the direction of motion during this stage (R1) and resulted in the increased COF values due to asperity-to-asperity contact. As time progressed into R2 and R3, the overall directionality of features became more consistent and uniform, as the lubrication contact mechanics is dominated by the tribofilm (both physisorbed particles and chemically reacted layers) protecting the surface asperities and preventing further drastic changes in the asperity texture.

The presence of lubricant particle clusters was observed on the wear track and was confirmed by corresponding maps (Fig. 7 bottom row). It was also noted that the EDS signal is dominated by the physisorbed lubricant clusters (which remain on the wear track even after dipping samples in hexane at the end of tribological tests), and it masks the EDS signal from the chemically reacted tribofilm. For this reason, the wear track was first cleaned with help of a cotton tip applicator to remove the physisorbed lubricant before performing chemical analysis of chemically reacted tribofilm layers using Raman and cross-sectional TEM as described below. The density of physisorbed lubricant particles on the wear track decrease with time. This suggested that more lubricant particles were possibly sheared, exfoliated, and reacted with the surface with time indicating the progression of tribo-chemical reactions.

To characterize the distribution and microstructure of lubricant tribofilm on textured surfaces, the tribofilm samples were characterized using cross-sectional TEM analysis. The area of interest was first identified using the SEM/EDS mapping technique before preparing the cross-sectional samples. The TEM samples were cut and prepared using the FIB technique. To understand the effect of peaks and valleys on tribofilm manipulation, the area of interest was chosen to include a valley and the adjacent peaks. The selected area was first coated with a platinum layer to protect against damage to the underlying tribofilm surface during the ion milling process.

The top panel of Fig. 8 shows the area of interest (which includes a valley area) that was chosen for analyzing the tribofilm microstructure on 7000 laps (R3) sample.

The storage and delivery of tribofilm layers are depicted in Fig. 8(a) for the 7000 laps (R3) sample. It can be seen from the figure that the lubricant layers are stored in the valley (Fig. 8, B) while a few monolayers of tribofilm are delivered on the adjacent peaks (Fig. 8, A, D). From the progression of images from the valley to the peak (images B-C-D), it can be seen that the lubricant layers align with the substrate and are exfoliated and reacted near the peak. The EDS line scans taken at the peak and valley regions (Fig. 8(b)) confirm the presence of MoS₂ lubricant layers. It can thus be concluded that the peaks act as anchoring points for lubricant layers to attach and react while the valleys act as storage spaces for the exfoliated and unreacted lubricant particles. Similar observations were also recorded for the tribofilm analyzed on the 3500 laps (R2) sample (not shown here due to space constraints). This observation was consistent with the previously proposed lubrication mechanisms for MoS₂ nano lubricants [30,39,40] and was tested by imaging stored lubricants in the valley region (Fig. 9) for the representative samples from all three lubrication stages.

Figure 9 shows the progression of stored lubricant layers in the valleys on the textured discs through the three stages of lubrication. As seen from the images, the thickness (denoted by white arrows) of the stored lubricant layer increases with time from R1 to R3. Additionally, at 800 laps (R1) the lamellar lubricant layers are commensurate with the substrate surface and did not show any exfoliation and intermixing. As the time increased to 3500 laps (R2) and then to 7000 laps (R3), the extent of intermixing and exfoliation of lubricant layers increased. This indicated that under the repetitive interaction between the surface asperities and the lubricant under the applied load, the lubricant particles get sheared, exfoliated, and react with the substrate peaks. As the test continues, the sheared lubricant layers get dragged, intermixed, and stored in the valleys.

Figure 10 presents the results of Raman measurements for three representative time intervals namely, 800 laps (R1), 3500 laps (R2), and 7000 laps (R3). As shown, the intensity of the Raman signal increased with time from 800 laps to 7000 laps indicating the increase in the formation and coverage of chemically reacted tribofilm. The chemical reactions between the ferrous substrate and MoS₂ nanoparticles capped with polar hydrocarbon chains result in the formation of various compounds that form the tribofilm. These compounds can be identified by comparing the Raman peak shifts with the known compounds in literature reports. Peaks of S-Mo-S bond stretch were observed at 386 cm⁻¹ (E¹_2g) and 407 cm⁻¹ (A_1g) corresponding to in-plane and out-of-plane vibrations. Additionally, peaks corresponding to MoS₂ tribofilm were also observed at 237 cm⁻¹ (E¹_2g) and 188 cm⁻¹ consistent with the literature reports [32,41]. Apart from MoS₂ peaks, several other new peaks emerged due to tribo-chemical reactions between the steel surface and lubricant. These peaks are highlighted in Fig. 10 as Fe₂O₃ at 224 cm⁻¹ (a result of the reaction between the iron and oxygen from fatty acid), FeS₂ at 361 cm⁻¹ (reaction between the iron substrate and the polar sulfur group formed as a result of exfoliation of MoS₂ particles) [42], and molybdate FeMoO₄ at 925 cm⁻¹ (a complex compound formed by the reaction between the iron substrate and functionalized, exfoliated nanoparticles) [43]. The emergence of the molybdate peak at 925 cm⁻¹ at 3500 laps confirmed the early onset and stabilization of tribo-chemical film as compared to untextured discs [29], where this peak emerged after 5000 laps of testing.

Figure 11 presents the COF plots from representative time intervals from each stage of lubrication. The overall COF trend is consistent with the trend observed in the control tests (Fig. 2). At the beginning of the test (in R1), the COF started high because of increased surface roughness on the discs (S_q and S_pc values) where initial sharp and high peaks are worn off and the frictional response is dominated by the texture. As the texture asperities interact with each other under BL to re-organize and re-align (breaking-in) in the direction of motion (SEM micrographs), the COF starts to decrease. The energy dissipation during the breaking-in process is believed to lead to the onset of tribo-chemical reactions resulting in the formation of tribofilm and lowering COF values. As time progressed (through R2), as the tribofilm formation and removal rate balanced with the rate of creation-dissolution of surface asperities, a stable steady-state frictional response was observed consistent with the trend in surface texture parameters. Furthermore, it was also noted that the sample-to-sample variation is more in R1 and R2 (800 and 3500 laps time duration) in comparison with R3. This was attributed to the locally random texture created during the polishing process, which causes sample-to-sample variation in R1, where the lubrication is governed by the direct asperity contact. As the lubrication is progressively controlled by the surface protective lubricant tribofilm in R3, the frictional response is also more controlled as seen from the 7000 laps case in Fig. 11.

The evolution of average WSD on the ball also followed a similar trend which is shown in Fig. 12. As shown, the wear scar diameter increased rapidly at the start (even at 200 laps-10 min condition) and continued to increase linearly in R1 as texture asperities are broken-in with potential realignment/reorganization. As the tribo-chemical reactions progressed and stabilized through R2 and R3, the wear scar values increase but at progressively lower linear slopes in comparison with R1, consistent with the above discussion.

Based on the aforementioned experimental results and characterization, the following lubrications progression mechanism is presented:

1. In R1, directional texture asperities created by the abrasive polishing process underwent re-alignment and break-in of sharp peaks as COF increased at the beginning of the test. After the sharp asperities characterized by reduced peak height (S_pk) parameter (Fig. 4) were broken-in, the COF starts to steadily decrease. The rise in friction and associated energy dissipation during this process along with the exfoliation of lubricant layers aligned with the substrate leads to tribo-chemical reactions and tribofilm formation on the interacting surface asperities.

2. In R2, tribofilm formation (through exfoliation and delivery of lubricant layers on peaks) led to the protection of surface asperities as COF starts decreasing and fabricated texture develops further towards stabilization of tribo-chemical reactions and physical (wear, deformation) texture. This is followed by the steady-state stage (R3), where the texture and tribofilm attained a dynamic balance as COF and surface roughness values are maintained steady owing to lubricant-coated asperities.

3. Formation and delivery of tribofilm to individual asperities is manipulated by the fabricated texture features; Peaks act as anchoring and reaction points while valleys aid storage and delivery of tribo-layers.

4. Textured discs showed an early onset and stabilization of tribofilm as confirmed by areal surface texture and Raman measurements as compared to the mirror-finished discs [29]. This is attributed to the ready presence of directional asperities as well as increased starting peak sharpness (Fig. 6, S_pc) in the case of textured discs which led to early onset and activation of tribo-chemical reactions and tribofilm formation.

To conclude, P1200 grit size SiC abrasive paper was used to realize directionally concentric texture on hardened steel discs. The directionality of features and addition of surface roughness play an important role in the early onset and stabilization of tribofilm as the asperities are already aligned in the direction favorable with the direction of motion in a pin-on-disc setup. The role of fabricated texture on the formation and progression of tribo-chemical reactions were explored using MoS_2-based nano-engineered lubricants. The time intervals of interest were identified from the control tests and divided into three distinct stages of lubrication contact mechanics and tribofilm evolution. At the beginning of the test, the initial contact between the surfaces (before tribofilm formation) led to the break-in of the asperities in R1 which triggers the tribofilm formation. The formation of tribofilm results in more rounded surface asperities as verified by the mean peak curvature (S_pc). The progression and stabilization of tribofilm concurrent with the evolving texture result in stable steady-state COF and roughness values in R3. The exfoliated and sheared lubricant gets stored in the valleys while the lamellar tribofilm layers are deposited on the adjacent peaks. The results confirm the effectiveness of adding directional texture to enhance and manipulate the properties of MoS₂ tribofilm for delivering an effective frictional response. A fundamental understanding of synergistic evolution between the tribofilm and underlying texture could be beneficial for designing textured components and lubricant chemistries for target mechanical applications.

The authors acknowledge the support and assistance from the Arkansas Bio-Nano Materials Characterization Facility and Dr. Mourad Benamara for the help in FIB and TEM measurements.

• The authors acknowledge support from the Center for Advanced Surface Engineering, under the National Science Foundation (NSF) Grant No. OIA-1457888, and the Arkansas EPSCoR Program, ASSET III. This work was performed at the University of Arkansas.

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

BL =

boundary lubrication

COF =

coefficient of friction

EDS =

energy-dispersive X-ray spectroscopy

FIB =

focused ion beam

NL =

nano-engineered lubricant

SEM =

scanning electron microscopy

TEM =

transmission electron microscopy