The Role of Strain Rate Response on Tribological Behavior of Metals

The fundamental mechanisms that govern friction and wear have been the subject of considerable study for centuries [1–3]. These studies, ranging from macro [1,2] to micro [3] to nano [4–6] scales, have often generated fundamental questions in addition to the answers they have provided. In the opinion of the authors, one factor that has been studied but requires further investigation is the potential role of strain rate response on the friction and wear behavior of contacting materials.

When two surfaces slide against each other at a particular strain rate and temperature, local contact at the asperities creates large plastic strains and associated changes in the near surface microstructure. The strain rate behavior of a metal is associated with its microstructural response to imposed conditions of strain, strain rate, and temperature. Previous research has demonstrated that different materials exhibit a large spectrum of microstructural responses [7]. The various microstructural responses of materials include dynamic recrystallization (DRX), dynamic recovery (DRY), adiabatic shear banding (ASB), flow banding (FB), wedge cracking (WC), void formation (VF), super-plastic deformation (SPD), inter-crystalline cracking (ICC), prior particle boundary (PPB) cracking [7]. Some of these responses (DRX, DRY and SPD) lead to microstructural evolutions that produce desirable properties while others (ASB, FB, WC, VF, PPB and ICC) can lead to undesirable microstructures that destroy the overall integrity of the material. For a given metal or alloy, the specific microstructural evolution (strain rate response) is a direct function of the strain rate and temperature that it experiences. This premise can also be extended to contact conditions that produce significant wear because during the wear process a large gradient of strain exists in the subsurface regions [8]. During aggressive sliding contact; therefore, a particular combination of strain rate and temperature exists at the contact interface and along the sub-surface regions where plastic deformation is occurring. These regions may produce a deleterious strain rate response that allows for the formation, nucleation, and propagation of cracks that generating material wear debris. Kailas and Biswas [9,10] proposed the strain rate response approach to explain the wear behavior of metals. It is reported [9,10] that titanium, when compared to other metals, is more prone at room temperature to adiabatic shear banding at high strain rates. Adiabatic shear banding is a microstructural evolution that causes flow localization and regions of crack nucleation. From the microstructures evolved in uni-axial compression tests [11], it was stated that Ti undergoes intense adiabatic shear banding when compressed at high strain rates (>10 s⁻¹) and at lower temperatures (<575 K). It was also observed experimentally that wear rate reduces with an increase in sliding speed. This was postulated to be due to the reduction in the intensity of ASB (microstructural instability) in near surface regions of the titanium pin [11]. Strain rate response has also been correlated with the wear rate and microstructural evolution in the near surface region [9–11].

In the present investigation, the role of strain rate response during abrasive contact was investigated by sliding commercially pure titanium and a-iron pins against textured H-11 die steel flats at room temperature. Previous work has shown that the strain rate response of a material during sliding plays an important role in determining its wear rate and wear mechanism [12]. Efforts were made earlier [13–15] to study the effect of surface texture on the friction and transfer layer formation for various materials using an inclined scratch test device. In these studies, it was determined that coefficient of friction and formation of a transfer layer depended on material flow and the nature of surface texture imposed during sliding. To build on these prior investigations, the relationship between strain rate and friction and wear were studied for a-iron and titanium by sliding them against a harder material with different surface textures. Titanium was specifically chosen because it is prone to ASB at higher strain rates of deformation. The surface texture of the hard material, which changes the constraints to flow, can be used to vary the strain rates imposed on the softer material during sliding contact. Comparing the friction and wear results for iron and titanium also allow the delineation of the role of relative hardness of the softer material during abrasive contact.

In this work, the tribological response of two pin materials—commercially pure titanium (99.9 wt. %) and pure a-iron (99.9 wt. %)—were investigated using a standard pin-on-plate testing apparatus [16]. The counterpart flat plate surface was made of H-11 die steel. The pins were 10 mm long, 3 mm in diameter and had a tip radius of 1.5 mm. The dimensions of the steel flats were 28 mm × 20 mm × 10 mm (thickness). The pins were electro-polished to remove any mechanically work hardened layer that would have formed during the machining process. Hardness measurements of the pins and steel flat were made at room temperature using a Vickers micro hardness tester with 100 gm load and ten second dwell time. Average hardness values of the pins (titanium and iron) and flat (H-11 die steel), obtained from five indentations, were presented in Table 1.

Four different kinds of surfaces of the H-11 die steel were chosen for this study, which included the “Unidirectional,” the “8-Ground” and the “Random” surfaces. The Unidirectional and 8-Ground surfaces with varying roughness were created by dry grinding the flats with Silicon Carbide (SiC) emery papers of 220, 400, 600, 800 or 1000 grit sizes. For the Unidirectional case, care was taken so that the grinding marks were unidirectional in nature. The 8-Ground surface was generated by moving the steel plate against dry emery papers along a path with the shape of an “8” for 500 cycles. The Random surface with varying roughness was generated under wet grinding conditions using a polishing wheel and one of three abrasive media—SiC powder (220, 600 and 1000 grit), Al₂O₃ powder (0.017 μm), and diamond paste (1–3 μm). Before each experiment, the pins and steel flats were thoroughly cleaned with an aqueous soap solution; this was followed by further cleaning the pins and flats with acetone in an ultrasonic cleaner. The surface roughness of the flats was recorded using an optical profilometer. The 3-D surface profiles of these surfaces are shown in Fig. 1. It should be noted that the three surfaces have the same R_a, but are morphologically very different.

Titanium (Ti) and a-iron (Fe) pins were slid perpendicular to the unidirectional texture (U-PD), parallel to the unidirectional texture (U-PL) and on the “8-Ground” and “Random” surfaces (the direction of sliding did not matter). The tests were carried out on an inclined pin-on-plate sliding tester designed and developed for carrying out such tests [16]. A schematic diagram of the inclined pin-on-plate sliding tester is shown in Fig. 2. By keeping the flats inclined, it is possible to conduct sliding tests where the normal load and tangential load increases monotonically. Thus, the coefficient of friction at various loads can be got from a single test. The sliding speed was kept at 2 mm s⁻¹ in an ambient environment.

Five parallel sliding tests of 10 mm length were done separately under dry and lubricated conditions on the same steel flat. Each scratch was produced by a single sliding event. The lubricant used was an engine lubricant containing Zinc Dialkyl Dithio Phosphate (ZDDP) as an additive. A few drops of the lubricant were smeared on the surface before the lubricated test. A fresh pin was used for the lubricated tests. It should be noted that the sliding tests essentially became a flat on flat contact after around 2–3 mm of sliding. The results presented are for the fourth scratch by which time the radius of the contact area of the flat pin was around 0.25 mm. It was observed that the coefficient of friction did not vary significantly for all these five wear tracks. After the tests, the pins and steel flats were observed using a scanning electron microscope (SEM) to study their surface morphology.

To study the microstructure at different strain rates, uni-axial compression tests [7] were done on cylindrical pure Fe at constant true strain rates of 0.01 and 100 s⁻¹ at room temperature. The tests were conducted using a computer controlled servo hydraulic test machine (DARTEC, UK), which had the advantage to change the speed of the ram exponentially so that constant true strain rates were maintained. The ratio of height to diameter of the sample was maintained at 1.5. After the experiment, the deformed specimens were sectioned along the compression axis and the microstructure evolved was studied using conventional metallographic methods. Experiments were conducted under similar testing conditions for the case of pure Ti and the results of which are reported elsewhere [17].

A typical graph of the variation of the normal force and tangential force with sliding distance when Ti pins slid perpendicular to the unidirectional grinding marks on the inclined steel flat under dry conditions is shown in Fig. 3(a). The variation of the coefficient of friction (calculated by resolving the normal and tangential force along the incline of the H-11 steel flat) with sliding distance is also shown in the same figure. It is interesting to note that coefficient of friction essentially remained constant over the entire sliding distance (up to a normal force of around 120 N). Figure 3(b) shows the variation of coefficient of friction with sliding distance when Ti pins are slid over surfaces with different roughness under both dry and lubricated conditions. The results are for the experiments conducted when the Ti pins are slid perpendicular to the grinding marks on the uni-directionally ground surfaces (U-PD). It can be seen that the roughness of the surface, within the present range of roughness tested, does not have any discernible influence on the coefficient of friction. The lubricated experiments had a lower coefficient of friction. Similar results were obtained for the experiments when Fe pins were slid on the H-11 steel flats. Figures 3(c) and 3(d) summarize the results obtained for the various surfaces when sliding the Ti and Fe pins, respectively. The error bars in the figures indicate the maximum and minimum values of the friction for the five surface roughness values of each surface texture tested. The connecting line in each figure represents the average coefficient of friction for the corresponding texture. It can be seen that the coefficient of friction does not vary much for the Ti when moving from the U-PD to 8-ground and then U-PL surfaces under dry conditions. However, a decreasing tendency in friction values can be seen when moving from the U-PD to 8-ground and then to the U-PL surfaces. Further, a clear decreasing tendency can be observed when moving from U-PL to the Random surfaces (Fig. 3(c)). There is no marked variation in friction under the lubricated conditions except for the Random surface. Interestingly, the lubricant does not seem to have any effect for the Random surface (Fig. 3(c)). The nature of the surface does not have an effect when Fe is slid on the H-11 steel flat, except that a drop is seen for the Random surface (Fig. 3(d)). All the Fe experiments when the surfaces are lubricated have a lower friction.

Figures 4(a) and 4(b) show the SEM micrographs and morphology of the H-11 surface for Ti and Fe, respectively, under both the dry and lubricated conditions. Some interesting features are noticeable in these two figures. Careful scrutiny of the results of Ti for a U-PD surface (Fig. 4(a)) reveals a large amount of transfer layer of Ti under both dry and lubricated conditions. The amount of transfer layer is lower for the lubricated sliding, but still significant. The amount of transfer layer is comparatively lower for the 8-Ground and U-PL surfaces than the U-PD surfaces and it was higher for the dry conditions when compared to the lubricated experiments. Here too some amount of transfer layer is seen under lubricated conditions. For the Random surface no transfer layer is seen under both the dry and lubricated conditions (Fig. 4(a)). Considering the case of Fe pins sliding on the H-11 steel surface (Fig. 4(b)), the surface morphology changed completely under dry conditions for all the surfaces except Random. Under lubricated conditions the surface morphology of the initial surface was preserved for all the surfaces and no significant transfer layer developed. This is in direct contrast to the Ti experimental results. Another noteworthy finding is that there is significant transfer of Ti on the U-PD surface while no such transfer layer of Fe on the H-11 steel flat occurs. In fact, the softer Fe abrades the H-11 steel flat changing the morphology of the initial surface whereas the harder Ti is not able to abrade the surface for the U-PD condition. It should be noted that the H-11 steel flat is significantly harder than both the Ti and Fe used for the present study.

Previous work by the authors [9,12] has shown that the strain rate response of a material can play an important role in controlling the wear rate and wear behavior of metals. The threshold strain rate was found to be ≈10 s⁻¹ for Ti at room temperature [18]. At strain rates greater than 10 s⁻¹, the material exhibits adiabatic shear banding. At strain rates below the threshold (less than 10 s⁻¹), the material deforms in a homogeneous fashion. During sliding, the interaction taking place at the asperity level imposes a strain rate of the order of 100 s⁻¹ [9]. The microstructure of the Ti deformed at room temperature at 100 s⁻¹ and 0.1 s⁻¹ are shown in Figs. 5(a) and 5(b), respectively [17]. The microstructural response of Ti when deformed at 100 s⁻¹ and room temperature shows adiabatic shear banding (ASB), while no ASB or other deleterious microstructure can be seen when deformed at 0.1 s⁻¹ and room temperature. Adiabatic shear banding leads to paths of lower energy crack propagation. In the experiments, Fe did not show similar unstable behavior. The microstructures of the Fe deformed at room temperature at 100 s⁻¹ and 0.01 s⁻¹ are shown in Figs. 6(a) and 6(b), respectively. It is important to note that the figures demonstrate that no ASB or other deleterious microstructure can be found.

When Ti is slid on H-11 steel flat it is expected to undergo ASB at the asperity level leading to fracture and transfer onto the surface. However, such a transfer layer is only seen for the U-PD surface and not the Random surface. This is because the U-PD surface imposes a higher constraint to flow at the asperity level than the Random surface. This constraint to flow will leads to a greater degree of plane strain conditions and thus a higher strain rate. One can expect more intense ASB when sliding over the U-PD surface. If it was indeed ASB that caused the large amount of transfer layer then such a transfer layer should also exist for the lubricated experiment as the lubricant will not alter the strain rates near the surface considerably. This is exactly what was observed for the U-PD lubricated experiments (Fig. 4(a)). Likewise, on a Random surface, the strain rates are lower so the material does not undergo ASB and thus the Ti pins abrade the surface for both the dry and lubricated experiments (Fig. 4(a)). The fact that the Ti pins transfers on an U-PD surface but abrades the Random surface can also be seen from the profiles of the tracks taken at a normal load of 60 N (Figs. 7(a) and 7(b)). In the case of Fe pins there is no ASB formation at high strain rates and thus it is less likely to fracture and form a transfer layer. Thus the Fe pin would be more likely to abrade the steel surface for both the U-PD and Random surface. This is confirmed from Fig. 4(b), where the SEM micrographs of the surfaces of H-11 steel flat for the U-PD surface and the Random surface are shown. The abrasion is also confirmed by the profiles of the tracks taken for Fe (Figs. 7(c) and 7(d)) at a load of 60 N.

It was stated earlier that the constraint to flow for the softer material increases the strain rates when moving from the random surfaces to U-PL, 8-ground and U-PD surfaces. The strain rates generated by the random surfaces were the lowest and this amount of strain rates were not enough for formation of adiabatic shear bands in the microstructure of Ti during sliding. Consequently, the Ti pin behaves like a harder material (similar to Fe) and abrade the steel surface. Hence, no transfer layer was formed on the random surface. In this case the hardness difference between the die steel and Ti surfaces is less and thus lower friction coefficient. When the Ti pins slid on U-PL surfaces, the constraint induced by the U-PL surfaces is comparatively more than that of random surfaces and thus amount of strain rates generated in the Ti pin would be more. This amount of strain rates are enough for the formation of adiabatic shear banding in the Ti microstructure. Consequently, the Ti pins behave like a softer material during sliding. Hence, transfer layer was formed on the U-PL surfaces. In this case the hardness difference between the die steel and Ti surfaces is more than that with random surfaces and thus the friction coefficient increases. Once the threshold strain rate is reached for the formation of adiabatic shear banding in the Ti material, further increases in strain rates does not drastically vary the tribological properties. This means that there is a significant variation in the tribological properties when moving from the absence of adiabatic shear banding to the presence of adiabatic shear banding in the microstructure of Ti. This can be clearly seen when comparing the friction plots and SEM micrographs of U-PL and random surfaces. The constraint induced by the 8-ground surface is more than the U-PL surface; thus the amount of strain rates and the intensity of adiabatic shear banding are comparatively more. However, the coefficient of friction and amount of transfer layer formed on the steel surfaces did not vary significantly when moving from U-PL to the 8-ground surfaces. The same phenomena can be expected for the U-PD surfaces that induce highest constraint to flow (among the surfaces studied), highest strain rates and thus the adiabatic shear banding. Hence, the coefficient of friction and transfer layer did not significantly vary when moving from U-PL to 8-ground and U-PD surfaces.

In the literature, efforts have been made to study the strain rate response of materials by sliding them against wedges of various angles [17]. Higher coefficient of friction and higher strain rate values in the deforming material were observed when sliding tests were conducted at higher wedge (attack) angles. Other [19,20] efforts were made to study the role of surface texture on the coefficient of friction and transfer layer formation. Four kinds of surface texture, as in the present case, were attained on the steel plate surfaces. The surface textures were characterized in terms of roughness parameters. In the experiments it was found that the friction coefficient depends on the asperity slope of the harder surfaces. More specifically, higher coefficient of friction and transfer layer were observed when the asperity slope of the harder mating surface was increased. Higher asperity slope values were found for the U-PD surfaces and decreased for the 8-ground, U-PL, and was the lowest for the Random surfaces. It was concluded that the coefficient of friction and transfer layer formation for the U-PD case was higher due to higher asperity slope (similar to higher wedge angle) that induces a high degree of plane strain condition at the asperity level. For the random surfaces, the extent of plane strain condition is lower due to lower asperity slope; this resulted in lower coefficient of friction and transfer layer formation [19,20]. Additional efforts were made to study the strain rate in the softer material by conducting subsurface deformation studies of the softer pin materials that were slide against the four surface textures (U-PD, 8-ground, U-PL and random [8]). The strain was found to be highest for the U-PD case, followed by 8-ground, U-PL, and was the lowest for the random surfaces. Based on these analyses, it can be inferred that U-PD surfaces induce higher amounts of strain rate and random surfaces induces lower amounts of strain rate in the deforming material during sliding. The estimate of the strain rate for the U-PD surface is ≈100 s⁻¹ [9], which is well above the threshold (10 s⁻¹) value.

In the literature, efforts have also been made to study the relation between interfacial friction and strain rate of various materials [21–23]. As seen from the upper bound calculations for a wedge sliding over a surface [21], the effect of the interfacial friction on the strain rate, when the wedge attack angle is small is low. In the present surface, the attack angle for both the unidirectional surface and random surface as given by the average slope is less than 3 deg. [16,19,20] and thus the interfacial friction should not influence the strain rate to a large extent. It is thus clear that it is the nature of the surfaces that is influencing the strain rate and transfer behavior observed. Further, it should also be mentioned that the strain rate response of a material generally changes gradually with strain rate with a complete change taking place when the strain rates changes by more than an order of magnitude.

It is important to note that the lubricant does not have any influence on coefficient of friction for the random surfaces. Efforts were earlier [19,20] made to characterize the lubricated surfaces in terms of roughness parameters. It was found that the average asperity slope was found to be highest for the U-PD surfaces followed by the 8-ground, U-PL and was the lowest for the randomly polished surfaces. Hence, the lubricant does not get trapped between the asperities to build up pressures for random surfaces during sliding. Thus, differences in friction values between dry and lubricated test is the lowest for random surfaces. Several other researchers also showed that build up pressure of lubricants depends on surface texture [24–26].

It is interesting to note that the unidirectional and 8-ground surfaces were generated using the same grinding media (i.e., SiC emery papers of different grit sizes). The random surfaces, however, were generated using different grinding media - SiC powder (220 grit, 600 grit, and 1000 grit), Al₂O₃ powder and diamond paste. This was performed in order to vary the surface roughness of the flats as outlined in the experimental section. As different grinding media are used for the random surfaces, different physical and chemical properties could exist (e.g., in the case of diamond paste polished surface) and this would influence frictional behavior. It is important to note in the present investigation that the random surfaces were also created by using SiC powders and water as the lubricant. This surface and the surface generated by the emery papers would have the same chemistry as the SiC polishing for the unidirectional surface was also the same. If the chemistry plays an important role in the friction then there would be a difference in the coefficient of friction between the surfaces created by the random surface using various media. The fact that this is not found indicates that the chemistry of the surface does not play as significant a role as surface texture in the friction variation.

Based on the experiments conducted in the present work, it is clear that for transfer layer to form and for a material to abrade the surface the strain rate response of the material plays a crucial role. We have also seen that the range in which the strain rate response would play a role would depend on the hardness ratio of the mating surfaces. The degree of constraint to flow imposed by a surface would dictate the strain rate levels on the plastically deforming material at the asperity level. Thus, four factors, the hardness, the strain rate response, the constraint to flow at the asperity level and the lubricant would dictate the “abrasivity” of a material in sliding. When one moves from micro-level interaction, as in the present set of experiments, to a nano-level interaction the one thing that changes is the strain rate, as this depends on the height of asperities that interact. Therefore, extrapolating these results to nano-level interaction could be erroneous if the strain rate response of the material at higher strain rates is not considered. This work also gives a possible direction in which one should address the design of lubricants. The lubricant should make it possible to reduce the interaction and strain rate significantly (which does not necessarily happen under boundary lubrication conditions) or change the chemistry of the surface in such a way that it makes the material near the surface undergo a strain rate response that does not lead to shear and fracture.

In the present investigation, the role of strain rate response on wear behavior of materials was studied by sliding pure titanium and a-iron pins at room temperature against an H-11 die steel flat of varying textures. It was observed that titanium, a harder material when compared to iron, does not abrade the surface of H-11 steel under specific surface textures. However, iron abrades the surface of H-11 steel of all surface textures. It is demonstrated that the surface textures can impose variable strain rates during sliding. Specifically, it was shown that sliding perpendicular to the unidirectional grinding marks imposes higher strain rates, while sliding on the Random surface imposes lower strain rates at the asperity level. Titanium is prone to adiabatic shear banding at higher strain rates and when titanium pins slide perpendicular to the unidirectional grinding marks, a transfer layer of titanium forms on the steel flat owing to adiabatic shear banding. However, titanium pins abrade the random surface during sliding as the nature of constraint does not impose strain rates of the order which can lead to adiabatic shear banding. Iron does not undergo adiabatic shear banding and hence, it abrades all surfaces during sliding.