Methods to Observe Tribological Failures in Self-Mated Steel Contacts
Authors:
Farida Ahmed Koly, Arnab Bhattacharjee, Nikhil Murthy, Benjamin Gould, Oyelayo Ajayi, Scott Walck, Cinta Lorenzo Martin, Stephen Berkebile, David L. Burris
The article 'Methods to Observe Tribological Failures in Self-Mated Steel Contacts' delves into the intricate challenges of lubrication and tribological failures in steel contacts, particularly focusing on scuffing. It discusses the environmental and practical challenges that have led to thinner lubricant films and reduced use of anti-wear additives. The authors present a novel approach using synchrotron X-rays to study scuffing in real-time, offering unprecedented insights into the mechanics of scuffing initiation and progression. The study validates the applicability of this method and provides compelling evidence of the role of debris and damage in the scuffing process, which has significant implications for the design and maintenance of mechanical components.
AI Generated
This summary of the content was generated with the help of AI.
Abstract
Scuffing, a type of wear found in highly stressed or poorly lubricated contacts, is characterized by a rapid increase in friction and severe plastic deformation of the near-surface material. Scuffing has proven difficult to study because it initiates unpredictably, progresses rapidly, and typically develops within an inaccessible contact interface. Although there have been successful in-situ studies of scuffing in real-time, the transparent counter body needed for these studies changes the interactions between the surfaces and the lubricant, which affects the scuffing process in unknown ways. This paper describes the development of X-ray-compatible tribometry to study the scuffing of self-mated steels in-situ and in real-time. The method uses a crossed cylinders configuration with a thin (500 μm thick) stationary component and a small (≈200 μm) contact width to maximize X-ray interactions with atoms within the stress field generated by the contact. The resulting instrument and method are used to benchmark the scuffing response of self-mated 52,100 steel under tribologically challenging ‘oil-off’ lubrication conditions. The results demonstrate reliable scuffing in this configuration despite the relatively small contact areas and loads used. Following scuffing, gross plastic deformation was observed on both surfaces along with significant subsurface grain refinement and flow only on the stationary surface, which experienced constant contact. Interestingly, high friction initiated at specific locations of the migratory surface, which experienced intermittent contact, and then propagated across the track over time, suggesting that local conditions of the migratory surface dominated friction leading into the failure event.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
1 Introduction
Efforts to improve environmental sustainability have led to significant lubrication challenges. Vehicle manufacturers, for example, have gradually reduced the recommended viscosity of lubricating fluids to improve fuel efficiency, but at the expense of thinner, less protective lubricating films [1]. At the same time, the anti-wear additives used to protect surfaces during periodic lubricant film collapse are being restricted or eliminated due to their potential for adverse environmental impacts [2]. Thus, engine components are being simultaneously subjected to thinner fluid films and reduced protection against film breakdown. Additionally, there are ongoing efforts to replace traditional fuels, which contain naturally protective surface-active molecules such as aromatic hydrocarbons [3], with synthetic biofuels, which do not. This presents an enormous challenge to the moving parts of the fuel pump, which rely entirely on the lubricity of the fuel for tribological function [4].
Under normal conditions, asperities make contact and wear down, but these contact periods are regularly disrupted to give time for relubrication and recovery. Often, this run-in process leads to low and stable values of friction. This normal run-in process can also include micro-scuffing (increased friction) and micro-scuff-quenching (lubrication recovery) [5]. When micro-scuffing cannot be quenched, increased friction, frictional heating, increased temperature, softening, and junction growth become self-reinforcing phenomena. This runaway positive feedback loop leads to excessive friction, gross plasticity, and in the extreme, seizure; this type of failure is known as scuffing.
Advertisement
Several approaches and test protocols have been developed to evaluate the scuffing performance of materials and lubricants. Some of the tests are based on simple contact configuration and laboratory bench top test rigs, while others are component specific and use real components for testing, e.g., Ryder gear and FZG gear testing. Different sliding modes, including unidirectional sliding, reciprocated sliding, and sliding with rolling, have been used in scuffing testing. In addition, many test variables or parameters are used to invoke scuffing failure in lubricated contacts; and the scuffing resistance is judged by the value of the parameter in which failure occurs. These include critical temperature, speed, load, and lubrication amount [6‐9]. Despite the differences in the various test methods and protocols, they all have in common progressive increase in severity of contact during test, leading to sequential failure of the lubricant fluid film, the surface tribochemical film (including oxides), and finally the near-surface material by severe plastic deformation [10].
Although there is a long history of scuffing research by practitioners and academics, the relevant mechanics remain unclear due to the difficulty in studying rapidly developing processes within an inaccessible interface. In 1937, Blok first proposed that scuffing occurs when the contact temperature, which increases with friction, pressure, and speed, reaches a critical value [11]. Dyson explained that elevated temperatures increase the risk of scuffing by impeding lubrication through a combination of boundary film desorption and the reduction of lubricant film viscosity and thickness [12]. Enthovan et al. tested this critical temperature hypothesis with direct measurements of the contact area temperature through a sapphire counterbody [13]. Contrary to their expectation that scuffing would occur at a single critical temperature, interface temperatures varied by almost 100 °C across scuffed samples. In fact, the two samples that did not scuff reached higher temperatures than any of the scuffed samples. Their results, they concluded, were inconsistent with the critical temperature hypothesis. They did, however, notice correlation between scuffing and third body mechanics, which they studied subsequently in detail using high speed optical microscopy of the contact area through the sapphire [14, 15]. First, they observed debris leaving the contact from the trailing edge and then accumulating at the leading edge (unidirectional experiments). As loads increased, debris began to enter the contact area. At a critical point, the debris became entrapped within the stationary contact area, aggregated, and formed a stable third body. Although the duration of the experiment and load at failure varied between experiments, scuffing reliably occurred shortly after third bodies stabilized within the stationary contact area. They concluded that scuffing was initiated by the debris, which impeded lubrication, rather than by excessive temperature.
Yagi et al. performed similar experiments with steel on sapphire but used a combination of visible light and synchrotron X-rays to study both debris dynamics [16, 17] and material transformations during scuffing [18, 19]. They also observed the formation of stable third bodies within the contact area and found that scuffing reliably followed. Additionally, their X-ray measurements showed that scuffing was accompanied by a transitory phase transition from martensite to austenite, the softer high temperature phase of steel. In total, the results from the literature appear to be consistent with the following positive feedback loop: 1) debris aggregate within the contact; 2) these aggregates impede lubrication; 3) impaired lubrication causes increased friction; 4) increased friction leads to increased temperature; 5) the steel transforms into its softer austenite phase at a critical temperature; 6) softening leads to junction growth and plastic deformation; 7) plastic deformation impedes lubrication.
While these in-situ studies have revealed important insights into the scuffing process, they required an optically transparent steel-sapphire pair with fundamentally different surface-to-surface and surface-to-lubricant interactions than those found in the self-mated steel contacts of primary practical interest. This paper is part of a broader effort to resolve this gap between scuffing practice and the existing scientific literature using synchrotron X-rays to study tribological failures of self-mated steel contacts in real-time. This paper develops the methods needed to achieve this outcome, validates the applicability of the approach to scuffing, and introduces evidence of a new role of debris and damage on the initiation of scuffing in self-mated steels.
Advertisement
2 Methods
2.1 Contact Design for Synchrotron X-Ray Measurements
The instrument used within the current work was designed to interface with a high energy X-ray diffraction (HEXRD) beamline, to study phase transformations, changes in the lattice strains, and grain refinement within the material as a function of depth into the contact, orientation (in the normal and sliding directions), and time (before, during, and after scuffing) at high speed and in real-time. The beamline used to conduct this experiment was beamline 1-ID at The Advanced Photon Source at Argonne National Laboratory. This beamline produces X-rays within an energy range of 42 to 136 keV. To provide transparency through a thin steel specimen with convenient X-ray diffracted beam angles for steel, the monochromator was set to E = 71.676 keV (wavelength of λ = 0.01726 nm). The unfocused beam size at this facility is 100 × 100 μm, but a more focused beam size of 50 μm wide (lateral direction) and 30 μm high (normal direction) was selected as the basis of the design.
One of the main challenges of the design was the need to restrict the beam to the contact zone. We used several strategies to help achieve this aim. First, we aimed to limit the nominal contact diameter to 100 to 200 μm, which concentrated individual asperity contacts within the beam area and reduced the potential for beam interactions with non-contacting areas. Second, we used a thin, 0.5 mm thick, stationary cylinder in a crossed-cylinder pair to limit the amount of non-contact material the beam was required to pass through before exiting the contact. Our definition of ‘stationary cylinder’ is important to note here: the contact area is stationary relative to the stationary component but moves relative to the migratory component in a tribological pair. In our case, the stationary component is also stationary relative to the instrument, which is convenient for the purpose of X-Ray analysis, while the migratory component reciprocates.
Figure 1 illustrates the situation. Figure 1A shows a side view with the X-ray beam entering one side of the stationary cylinder and the diffracted beam leaving the other. Figure 1B shows the contact area (green circle) as a significant proportion of the stationary sample thickness. The larger the contact per unit thickness, the more signal one receives from the contacting material per unit of non-contact material. However, it must be kept in mind that the nominal contact area comprises a collection of many real contacts, which tend to concentrate as the nominal area shrinks. Thus, small nominal contacts are expected to improve signal/noise. The X-ray image in Fig. 1C shows the X-ray transparency of the stationary cylinder during contact and uses a red box to highlight the approximate size and location of the focused X-ray beam used for in-situ diffraction measurements.
Fig. 1
A Side view schematic of in-situ x-ray measurements with the beam entering from one side of the instrument and exiting the other side. B Top-down view showing the approximate relationship between the beam size (red), the contact (green), and the thin stationary cylinder the beam must pass through. C X-ray tomography showing the approximate size of the 0.05 mm wide × 0.03 mm tall beam used for in-situ diffraction measurements as a red box (Color figure online)
Scuffing tends to occur at high pressure and speed. The target average contact pressure was 1 GPa, which is consistent with other studies of scuffing within the literature. To achieve these targets, we selected 6.4 mm diameter crossed cylinders under a load of 10 N, which produces a nominal contact diameter of ~ 100 μm and a mean stress of ~ 1 GPa according to Hertzian contact mechanics. The crossed-cylinder constraint also necessitated reciprocated sliding, which limits sliding speeds. Using past scuffing studies as a guide, we selected 100 mm/s as a minimum speed target for peak speed.
3 Materials
The crossed cylinders used in this study were fabricated from 6.4 mm diameter rods of 52,100 steel. The samples were prepared by first cutting stock rods into 25 mm long cylinders. These samples were rotated about their axis and wet-lapped to 100 nm Ra. Stationary samples were produced by sectioning one of these into 0.5 mm thick slices. After sectioning, stationary and migratory samples were heat-treated at 840 °C for 16 min, then removed and water quenched. This was followed by annealing for 120 min at 190 °C, resulting in a hardness of approximately 62 Rc. The edges of the stationary samples were then deburred, resulting in slight rounding of the edges. Finally, the surfaces of all cylinders were finished with a 1 μm diamond slurry, resulting in an average surface roughness of Ra = 100 nm. Figure 2 shows the optical microscopy of a representative stationary sample before and after several sliding cycles to validate contact alignment, which was affected by imperfect sample geometry and tribometer tolerancing. After method refinement, testing with N = 5 random samples showed that the center of contact was within 60 μm of the disc center with 95% confidence. However, we note that contact misalignment was an early problem for us due to the difficulty in manufacturing the thin cylinders while preserving their cylindrical geometry.
Fig. 2
Optical images of the 500 μm thick stationary cylinder following preparation A before contact and B after several sliding cycles used to mark the contact area. Following method refinement, a small preliminary study with 5 samples showed that the contact was within 60 μm of the center with 95% confidence
The custom tribometer shown in Fig. 3 is the finished product from our design process. The setup uses a vertical nanopositioning stage (PI model M605) to apply the load through a cantilever spring (k = 25 N/mm), which is used to define the relationship between stage movement and applied force. The nanopositioning stage has a range of 50 mm and a resolution of 100 nm, which yields a force resolution of 2.5 mN. Forces are measured using an ATI Nano17 six-channel force/torque transducer (50 N range ± 12.5 mN). A crank-based mechanism was used to reciprocate the migratory sample over a 10 mm long track at 4.8 Hz for a peak speed of approximately 150 mm/s. A linear variable differential transformer (LVDT) was used to measure track position and speed in real-time. A tilt stage was used to level the surface with respect to the sliding direction, which limits gradients in stationary sample elevation and normal force as a function of position on the wear track. An acoustic emission (AE) sensor was used as secondary condition monitoring.
Fig. 3
a Annotated images highlighting the features of the in-situ tribometer developed for this study. The instrument uses a computer controlled vertical nanopositioning stage to specify the applied load. The 6-channel load cell has a maximum load range of 50 N and a resolution of 12.5 mN. A crank-based mechanism creates reciprocating motion over a 10 mm long track at a frequency of 4.8 Hz and peak speed of 150 mm/s. b The X-ray beam passes beneath the vertical stage, through the contact, and the diffracted beam exits the front of the instrument without obstruction
Following sample preparation, samples were cleaned using 5 min of ultrasonication in isopropanol, dried with a lint-free laboratory wipe, and mounted into the tribometer. A cotton swab was submerged into dodecane (a low lubricity fuel) and swiped back and forth twice across the migratory cylinder to lubricate the contact but enforce starved lubrication conditions. The stationary sample was then brought into contact with a load of 10 N, which was held constant. Sliding was then initiated to target a peak speed of 150 mm/s, which corresponds to 4.8 Hz on the 10 mm long track. In-situ IR measurements have shown only a modest bulk temperature rise of < 10 degrees Celsius across the friction coefficient spectrum. Testing was performed at room temperature using N = 5 repeats to assess repeatability in the scuffing response of this model system. To mitigate the risk of system damage, the system shut itself down above a threshold friction force of 7 N.
Real-time signals from the LVDT and load cell were collected at 10 kHz over a 0.5 s acquisition window using custom LABVIEW software (~ 2.5 cycles per collection window). The data from each acquisition window were analyzed in real-time to obtain speed, friction force, normal force, and friction coefficient as a function of location on the migratory sample. To limit the analysis to roughly constant speed conditions and eliminate reversal effects, average values were obtained using only data from the central 25% of the wear track as shown in Fig. 4. In addition to average values, these ‘friction loops’ were saved and used later to gain insight into how the failure process evolved within the wear track and over time.
Fig. 4
Representative normal force, friction force, and velocity data as function of position for 1 sliding cycle with dodecane-lubricated steel-on-steel over a 10 mm track length at frequency of 4.8 Hz. Data from the central 25% (2.5 mm) of the track were used to obtain average values. In this case, the average speed was 140 ± 7 mm/s normal force was 10 ± 1 N and the average friction force was 1.5 ± 0.5 N
Ex-situ optical and scanning electron microscopy (SEM) measurements were performed prior to and following scuffing. Before imaging, samples were cleaned of lubricant and debris using isopropyl alcohol first then acetone. Optical images were collected using Nikon DS-Fi1 microscope at 10× magnification. SEM and X-ray Energy Dispersive Spectroscopy (EDS) analysis were done using a Zeiss Auriga 60 operating at 15 kV. Secondary electron (SE) and back-scattered electron (BSE) images were acquired. A ThermoFisher Scientific Helios dual beam focused ion beam (FIB) system was used to examine the subsurface microstructure. To preserve the immediate surface detail from ion beam amorphization and gallium implantation, the areas of interest were first protected with a 0.5 µm thick layer of electron beam deposited platinum, followed by a thicker layer of ion beam deposited platinum, prior to exposing the sample to an ion beam. Extended cross sections of about 50 µm in length and approximately 40 µm deep were cut using first the regular cross section tool followed by the cleaning cross section tool using a 30 keV beam with succeeding decreasing ion currents for improved polishing of the cross section. The cross-sectional areas were then ion etched to enhance the grain structure by scanning the ion beam at incident angle of 54° with an ion current of 26 pA for approximately 40 s. They were then imaged using the scanning ion beam with the same conditions and using the in-chamber electron and ion (ICE) detector.
4 Experimental Results
4.1 Scuffing Validation
The friction coefficient is shown as a function of time for three representative experiments of varying duration in Fig. 5. In the beginning, we observed effective lubrication with low stable friction as illustrated by Fig. 5A(i) (the experiment was stopped after 6 min of sliding). During this run-in phase, we consistently observed evidence of a tribofilm on the surface of the stationary sample and little evidence of wear as shown in the accompanying optical image [Fig. 5A(ii)]. This tribofilm was also consistently oxygen rich as illustrated by the corresponding XEDS measurements in Fig. 5A(iii). Thus, we have distinguished the ‘low wear regime’ as a period with friction coefficients < 0.2. Figure 5B(i) shows the transition to the ‘high wear’ regime, which we define as friction coefficients between 0.2 and 0.4. In this case, the experiment was stopped shortly after the friction coefficient increased above 0.2. Aside from higher friction coefficients, this regime is characterized by the creation and accumulation of wear debris around the contact periphery, loss of the oxygen-rich tribofilm [in the center of Fig. 5B(iii)], and a shiny metallic appearance [Fig. 5B(ii)], suggesting gradual material removal and the creation of a relatively smooth, relatively oxide-free wear surface.
Fig. 5
Friction coefficient (i) as a function of time with optical microscopy (ii) and XEDS oxygen dot mapping (iii) from the stationary component following a low wear, b high wear, and (c) scuffing. During low wear, we consistently observed oxygen-rich tribofilms on the wear surface. During high wear, we observe a shiny wear surface following cleaning. Following scuffing, we observe an extremely rough oxygen depleted surface characterized by gross plastic deformation
The complete scuffing response is shown in Fig. 5C(i). In this representative experiment, the system runs in the low friction regime for the first 5 min before experiencing a brief period of increased friction and recovery consistent with micro-scuffing and scuff-quenching as described by Ludema [5]. The system enters the high wear regime around 9 min into the experiment and friction trends toward a relatively stable value of 0.3. Shortly after the 10-min mark, friction rapidly increases to 0.7 in just a few seconds before triggering the instrument stop condition at a threshold friction force of 7 N. Following scuffing, the surface loses its smooth shiny appearance and exhibits clear visual evidence of gross plastic deformation and damage. XEDS results indicate that non-native oxides were completely removed, leaving surfaces unprotected from direct metal-to-metal interaction. This experiment meets our definition of scuffing since it involved a rapid increase in friction to beyond 0.4 with clear visual evidence of plastic deformation on the wear surface.
4.2 Repeatability of Scuffing Tests
The results of six repeat experiments are shown in Fig. 6. Friction coefficients are shown as a function of time along with optical images of the wear surface from the stationary component of the corresponding experiment. Each pair experienced scuffing as evidenced by rapidly increased friction after some time to values above 0.4 and gross plastic deformation of the wear surface. For all six experiments, the time to scuffing varied between 8 and 28 min of sliding. Four of five experiments experienced periods of low wear, high wear, and instability. Only Experiment 1 scuffed before experiencing a period of high wear sliding, and only Experiment 2 experienced micro-scuffing and scuff-quenching. Experiment 4 exhibited a very long high wear period and is the only experiment in which part of the wear surface retains the shiny appearance typically observed during the high wear regime. This suggests that only a portion of the contact scuffed before the test was stopped.
Fig. 6
Friction coefficient as a function of time for six repeat experiments with corresponding microscopy images of the stationary contact wear surfaces
Figure 7 shows friction as a function of position at various critical points in a representative experiment. During low wear sliding early in the experiment (A), we observed stable low friction across the entire wear track. At 11 min (B), we observed a frictional increase entering the reversal rather than exiting the reversal as would be expected of a static friction spike. This feature, we believe, resulted from the collection and accumulation of debris at the track ends and the bumping of the stationary contact into these accumulating debris piles. At 13.5 min (C), just prior to scuffing, we observed the same reversal feature in addition to a bump in friction at the 2.5 mm location; While the source of this feature is unknown, we can conclude that it was associated with some change on the migratory sample since the stationary sample experienced low friction everywhere but this location. In other words, this feature could not have been due to some change such as transfer or debris accumulation on the stationary component as observed in prior steel on sapphire studies of scuffing [14, 15, 17]. The persistence of this feature in subsequent cycles, illustrated by loop D, suggests that the ‘defect’ was semi-permanent such as damage or third body transfer to the migratory sample rather than loose debris. Additionally, this feature was broader in loop D suggesting progression from the initiation site. Just after a rapid increase in friction (E), the stationary sample experienced extremely high friction on one half of the track and low friction on the other, which reinforces that the migratory surface, rather than the stationary sample is dominating the initiation and progression of failure. At full scuffing (F), friction was extremely high across the entire track. Analysis of the friction loops consistently shows evidence of elevated friction leading into a reversal first, followed by high friction at one or more specific locations, followed by recovery (usually temporary) or unstable propagation across the track to failure.
Fig. 7
Friction coefficient as a function of time and corresponding friction loop (> 2 cycles per instance) for the marked times as indicated (A–F). Loop A corresponds to the early sliding situation with minimal and stable friction throughout the track. Loop B retains low friction after 10 min of sliding but shows high friction leading into the reversal due, likely, to the stationary contact encountering a debris pile. Loop C shows increased friction at a specific location within the wear track. Loop D shows that high friction is sustained at and propagated from this location. Loop E shows drastically different friction on two halves of the wear track. Loop F shows complete propagation of high friction across the wear track
It is also worth noting that each of these ‘loops’ contains 2.5 cycles of data. Friction was perfectly repeatable where one observes a ‘single’ dot at a given location. Some ‘thickening’ observed in Loop (D) indicates more rapid change at certain locations. Loop (E) was surprisingly repeatable between cycles given its substantial variation across the wear track. Loop (F) illustrates a single location of significant frictional change between cycles.
4.4 Surface and Subsurface Damage
SEM images of representative scuffed samples are shown in Fig. 8. The worn surfaces of both the stationary (A) and migratory (B) samples reveal clear evidence of severe damage within the wear zone (~ 20 μm Ra for both worn surfaces). The surface of the stationary sample exhibits the severe plastic deformation typical of scuffing. The migratory sample contains what appears to be a combination of large features (possibly from material removed) and relatively smoother grooving in the sliding direction. The subsurfaces of worn regions on stationary and migratory samples are shown in Fig. 8C and D, respectively. The stationary sample has four distinct regions. The first layer (1) is a protective Pt coating used to protect the surface from the ion beam milling process. The second layer (2) is heavily oxidized iron (XEDS not shown) and appears to be a transfer layer. The third layer exhibits significant grain refinement with plastic flow in the sliding direction. The fourth layer shows no clear evidence of flow or grain refinement and is qualitatively consistent in appearance to control sections of non-contact regions (not shown); thus, the grain refined zone, which was consistently on the order of 5–10 μm deep for scuffed stationary samples, can be attributed to shear. Interestingly, the migratory components we examined showed only minor grain refinement within the first 1–2 μm of the surface despite the significant damage observed on the surface. One possible explanation is that material removal rates were faster on the migratory sample, thereby removing the evidence of plasticity quicker. However, profilometry showed significantly less wear depth on migratory surface, which is consistent with its relatively larger wear area. Subsurface plastic flow and grain refinement are likely functions of energy absorption at a given location, which is greatest for the stationary component. Additionally, the evidence is consistent with the hypothesis that the loose debris were subjected to the greatest plastic deformation (either during or after removal) and were lost or preferentially transferred to the stationary component. The surface images are consistent with the removal of debris from the migratory (and stationary) sample, the plastic deformation of those debris by the shear process, entrainment of debris into the contact, preferential adhesion of these third bodies to the stationary component, and the abrasion of the migratory component by these oxidized, hardened, and well-adhered third bodies.
Fig. 8
A SEM image of a representative scuffed area from a stationary sample (15 kV), B SEM image of a representative scuffed area from a migratory sample (15 kV). C SEM image of a FIB cut from the scuffed area of a stationary sample (30 kV). D SEM image of a FIB cut from the scuffed area of a migratory sample (30 kV). The sliding direction is left–right as shown by the double arrow
This study aimed to bridge the gap between the practical understanding of scuffing and the existing scientific literature by enabling the use of synchrotron X-rays to study scuffing of self-mated steel contacts in real-time at high speed. This work successfully developed the methods required to achieve this outcome, validated the applicability of this approach to scuffing experimentation, and provided new insights into the role of debris and damage in the initiation of scuffing in self-mated steels (Fig. 7). The approach successfully produced scuffing in the small contacts of these experiments and demonstrated high repeatability in the results. Three regimes were consistently observed and defined as follows: low wear (COF < 0.2), high wear (0.2 < COF < 0.4), instability (rapid increase in COF to values > 0.4).
One of the early validation challenges encountered was the definition of scuffing. In a similar cross-cylinder study by Qu and Blau [20], for example, scuffing was defined as a friction coefficient > 0.2. Early in our experiments, however, we observed very smooth wear surfaces and prolonged stable sliding even at friction coefficients above 0.3 in some cases, which is inconsistent with our definition of scuffing. Our results suggest that the magnitude of friction alone is an unreliable predictor of scuffing. When a rapid change of friction (> 0.01/s over 10 s) occurred, however, it was reliably accompanied by evidence of gross plastic deformation following sliding. Thus, the rate of change was used as the primary in-situ measurement of stability/instability and predictor of gross plasticity. In these experiments, friction was always unstable above 0.4 but this threshold likely depends on the experimental conditions. Thus, the rate of frictional change is likely to be a more reliable measure of instability in studies of scuffing than the magnitude of the friction coefficient. However, the most reliable way to detect plastic deformation is with direct in-situ X-ray observation as we plan to do in the future.
A uniform oxygen-rich film was consistently observed on the stationary sample during the low friction regime. This oxide film was consistently removed when friction began to increase in the high wear regime. The oxide is clearly protective against metal–metal contact. It is unclear why the oxide formed initially or why it reliably failed in these experiments. Past studies have made similar observations of iron oxide rich films in unscuffed areas and their removal in scuffed areas [21]. It is possible that the high pressures in initially rough surface contacts promote oxide formation, but as pressures decrease after run-in, oxide formation rates may become insufficient to outpace wear. It seems likely that the tipping point involves competitive rates of formation and removal as observed in other studies [22]. Interestingly, experiments with anti-wear additives produced similar looking films comprising the primary additive constituent rather than oxygen. These additives appear to replace the oxide film, which is unstable, with a film that can replenish itself at a more sustainable rate.
While the removal of the oxide was associated with increased friction, it was not the initiator of scuffing. Early in-situ studies from Enthovan et al. and more recent studies from Yagi et al. provide compelling evidence that scuffing is debris-initiated rather than thermally initiated [14, 15, 17, 19]. According to those studies, third bodies (possibly oxides) entered the contact and began accumulating within the contact just prior to scuffing. The rationale is that the adhered third bodies disrupt lubrication, which increases friction and initiates a self-perpetuating feedback loop. The observations from the present study are mostly consistent with these findings but add several new insights. For example, the first evidence of increased friction occurred leading into the reversal. This appears to reflect the generation and accumulation of debris at the reversals. Some of these debris may be pulled into the contact area and dragged across the track by the stationary sample. However, friction typically increased and remained elevated at a discrete location on the track, which reflects a durable friction-increasing defect but only at a single spot on a much larger wear track. Had the initiator been the aggregation of debris on the stationary sample, uniformly elevated friction across the wear track would be expected. This localized region of high friction tends to propagate across the track, which eventually leads to scuffing.
This process, which appears to be a feature of self-mated steel, is precluded by steel on sapphire. Wear and other defects are unlikely or minimal when the migrating component is sapphire. Additionally, because the temperature is cooler on average and the surface energy lower, transfer of debris to the migratory alumina component is less favorable. Damage, defects, and transfer are all expected on the steel stationary component alone. In such an experiment, one would expect location insensitive friction during sliding and any relevant surface changes to occur within the contact and on the stationary component. Thus, this paper reveals what may be a key distinction between steel on sapphire and steel-on-steel experiments.
A similar unexpected disparity was observed regarding subsurface material transformations for the two surfaces. The migratory sample produced no clear evidence of change to the subsurface microstructure. The stationary sample, on the other hand, exhibited clear evidence of transfer, grain refinement, and flow within ~ 10 μm of the surface. Similar subsurface changes have been observed for the stationary component following scuffing in block-on-ring experiments [21, 23]. We are unaware of other studies showing disparities in subsurface deformation between migrating and stationary components. The observation that subsurface changes occurred first or preferentially on the stationary component is well-aligned with our strategy to use the stationary component as the X-ray transparent body. Such measurements will be well-suited for determining the timing of these subsurface changes and whether they are the cause or consequence of scuffing.
6 Conclusions
An in-situ experimental method was developed to assess evolution of damage in steel-on-steel contacts in real-time using X-ray diffraction. An instrument was designed and built for use at the Advanced Photon Source. The system required a thin stationary disk in a cross-cylinder geometry to maximize X-ray transparency and diffraction signal from the contact area. The tribometer and method reliably produced scuffing under starved lubrication with dodecane. These contacts exhibited three sliding regimes: low wear, high wear, and instability. Additionally, the results showed that high friction with a high rate of increased friction was a more reliable indicator of scuffing than high friction alone based on correlation with severe plastic deformation of the surface. In-situ friction loop analysis helped reveal new insights into its initiation and progression: debris were first entrained, followed by local damage to the migrating component, then the progression of that damage across the migrating component, which finally led to terminal frictional instability. While damage initiated on the migrating component, SEM only showed clear evidence of transfer and subsurface plasticity on the stationary component with grain refinement and flow in the sliding direction within 10 μm of the surface. These frictional and structural disparities between the stationary and migratory components provide original insights into scuffing mechanics. Critically, this observation of preferential damage to the stationary component aligns with our strategy to use this component as the target of our future X-ray diffraction measurements.
Acknowledgements
We gratefully acknowledge technical advisement from beamline scientists Dr. Peter Kenesei and Jun-Sang Park for the development of these methods. This research was sponsored by the Army Research Laboratory and was accomplished under Cooperative Agreement Number W911NF-20-2-0129. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Army Research Laboratory or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein.
Declarations
Competing Interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
