Introduction

Previous research has shown that for metals in a tribological contact, a feedback loop exists between the friction coefficient, the stress state and the microstructure [1]. During a tribological experiment, the friction and the normal force determine the stress state that leads to a microstructural evolution in the subsurface area [2, 3]. In turn, the resulting deformation layer leads to a change in the friction coefficient. So far, various microstructural characteristics have been observed in coarse grained, face-centred cubic (fcc) metallic materials under tribological load, such as a dislocation trace line [4, 5], subgrain formation [2, 6], deformation twinning [7, 8], band-like patterns [9,10,11] and nanocrystalline grains [7, 12].

In essence, a tribological system features two forces controlling the microstructural evolution: first, the normal force, which can be easily adjusted and second, the friction force, which is a function of the tribological system itself. Further parameters determining the deformation layer are cycle number [4, 13, 14], sliding velocity [15, 16], temperature [16] and ambient environment [17].

The quinary single phase high-entropy alloy (HEA) CoCrFeMnNi [18], exhibiting a medium stacking fault energy, was chosen as the material of interest because it covers dislocation-mediated microstructures at low strains and twin-induced plasticity at higher strains, although this knowledge stems from uniaxial load experiments [19, 20]. To the authors’ best knowledge, only three other research groups have investigated the tribological properties of single phase CoCrFeMnNi so far. Ayyagari et al. analysed the wear behaviour in dry and marine environments [21]. Joseph et al. [22] investigated the friction and wear behaviour at temperatures ranging from room temperature up to 900 °C. Jones et al. examined the interplay between grain refinement and the friction and wear behaviour after several thousand sliding cycles [12]. The two normal loads they applied resulted in different grain sizes in the subsurface area [12]. In our earlier work, we analysed the microstructural evolution with increasing cycle number with a focus on the onset of microstructural changes at high friction coefficients [7]. In contrast, the influence of the friction and normal force is investigated after a single-trace experiment in this manuscript. Besides that, the pronounced adhesive tendency of CoCrFeMnNi enables us to cover friction coefficients between 0.1 and 1 by using different counter body materials and environments. Furthermore, the influence of an increased normal load is investigated.

The polycrystallinity of CoCrFeMnNi makes it necessary to take the crystal orientation into account. It is known from uniaxial tests that the initial grain orientation is decisive for the occurrence of deformation twinning [23, 24]. In contrast to high SFE materials [25,26,27], there is only little data available on medium SFE materials [7, 13, Variation of friction forces

The tribologically imposed stress states of the materials constituting the contact is a function of the friction coefficient [1, 32]. To investigate the effect of the stress state on the microstructural evolution, the tribological systems were varied to achieve varying friction coefficients, see Fig. 1. Steady-state friction coefficients are already published in literature for several tribological systems containing CoCrFeMnNi [12, 21, 22], which are similar to the friction coefficients in the present study. The obtained variation of the friction coefficients stems from various contributions as briefly discussed below.

The work of adhesion is considered in the first place, which comprises the surface free energies of the two materials and their interface [33]. Neither the surface free energy of CoCrFeMnNi nor the ones of the interfaces are published in literature to the best of the authors’ knowledge. The values of the surface free energy of the counter body materials [34] do not directly correlate with the measured friction coefficients. No final conclusion can therefore be drawn on the influence of the work of adhesion on the friction coefficient in the chosen tribological systems. However, wear tracks with a low friction coefficient have a scratched appearance, and flake formation was observed in wear tracks with a high friction coefficient (Fig. 2). Furthermore, the contact areas of the counter bodies with a high friction coefficient exhibit material transfer. Flake formation and material transfer are clear indicators of adhesive wear [33]. These factors therefore provide a qualitative indication of different adhesive forces within the chosen tribological systems, but the adhesive forces will not be further analysed in the current manuscript.

The second parameter is the surface roughness of the different counter body materials. It was proposed in literature [35,36,37] that with an increasing surface roughness the contact area decreases, which can result in a lower friction coefficient. This tendency was confirmed in our experiments.

The environmental changes from air with 50%RH to dry N2 resulted in a lower friction coefficient. Surface softening was reported for sapphire with increasing humidity [38][38] which can facilitate shearing of the surface and lead to an increased contact area [40]. These two effects might therefore cancel each other out [40].

Three aspects, namely adhesive force, surface roughness and surface softening, were considered to explain the varying friction coefficients. So far, it is unclear to which extent they contribute to the friction forces.

Influence of initial grain orientation

The initial grain orientation can strongly influence the deformation mechanism, especially in the case of deformation twinning under uniaxial load [23, 24]. The influence of the initial grain orientation under tribological load was investigated by comparing two subsurface deformation layers for the same tribological system and, therefore, also the same applied surface stress state.

The tribological system in focus was chosen to be SiC/2N/air, as it does not exhibit material transfer and the friction coefficient is nearly constant over the entire sliding distance (see Fig. 1). The two grains in question were presented in Fig. 3a + b and Fig. 4, demonstrating their different microstructural evolution. The thickness of the tribologically induced subsurface layer is small for both grains, so the orientation underneath this deformed layer was interpreted as the initial grain orientation (see TKD results in Fig. 4).

The SAGB identified in Fig. 4a has been described as a dislocation trace line (DTL) in earlier publications [4, 5, 13]. The DTL in CoCrFeMnNi will be further discussed in the subsection ‘Dislocation trace line and crystal rotation’. The line-type features in Fig. 3b were clearly identified as twins using FFT in Fig. 5. Therefore, there is a need for understanding why plastic deformation is mediated by dislocation slip in one grain and twinning-induced plasticity in the other; under otherwise identical tribological conditions. The initial grain orientation as well as the type of applied load is decisive for deformation twinning [24]. The loading was the same for both grains. The initial crystal directions of the two grains are quite close to each other parallel to SD and TD (Fig. 4g + i). Thus, it is evident that the initial crystal direction parallel to ND under the given loading parameters is decisive for the dominating deformation mechanism. This is unexpected, as it was demonstrated in a previous study that the initial grain direction parallel to SD is decisive for the occurrence of deformation twinning [7]. The differences between the published results in [7] and the results in Fig. 4 are the selected counter body materials, resulting in a significantly reduced friction coefficient and no material transfer here and in a higher friction coefficient and material transfer in [7]. The lack of material transfer results in lower adhesive forces, which in turn decrease the tensile stresses parallel to SD at the trailing edge of the counter body. Further details about the twinning mechanism under tribological load are discussed in the section ‘Twinning’.

The increase in misorientation in Fig. 4j is caused by dislocation motion. In both grains this increase starts at 0.5 µm, which shows that the critical resolved shear stress (CRSS) for dislocation glide is reached at a similar depth. A difference between the two grain orientations can be observed in the maximum misorientation from the bulk to the surface which, is approximately 25° higher in the grain with the DTL than in the grain with twins. The misorientation profiles (Fig. 4j), however, only include the misorientation changes caused by dislocation motion. This indicates a difference in the dislocation activity. The discontinuity in misorientation corresponds to the SAGB in the grain shown in Fig. 4a–c. The higher the misorientation, the more dislocations are stored in the lattice. The formation of a SAGB leads to a reduction in the stored lattice energy, therefore the discontinuity could be favoured [41].

The two investigated grains under the same tribological loading reveal that not only the stress state, but also the initial grain orientation can be decisive for the dominating deformation mechanism under tribological load.

Influence of a higher normal load

For medium stacking fault energy materials like CoCrFeMnNi, increasing the stress under uniaxial loading increases the probability for the material to twin. We therefore expected to observe more and larger twins at a normal load of 5 N than at a normal load of 2 N. However, the microstructures in Fig. 3g + h and Fig. 6d–i for the SiC/5N/air experiments do not show any evidence of twinning. In contrast, we found DTL formation [4, 5, 13]. The differences between the subsurface microstructures in these two grains are the DTL depths and the grain boundary type of the DTL, which is a SAGB (Fig. 6d) in one grain and a HAGB in the other (Fig. 6g). Additionally, the area between the DTL and the surface is dominated by subgrains in Fig. 6g and does not exhibit a colour-gradient as observed in Fig. 6d. The microstructural features are formed in both grains by dislocation glide and self-organisation. The same was observed and already discussed for the grain in Fig. 3a. Both experiments showed a similar friction coefficient, which also means that the friction force is higher with normal load. This is why the influence of the increased friction force on the DTL depth cannot be separated from the increased normal load. The higher normal load results in a larger misorientation difference at the discontinuities (Figs. 4j, 6j). This can be interpreted as stronger dislocation activity with higher normal forces. Contrary to expectations, an increase in normal load did not increase the twinning probability under tribological load, but rather the amount of dislocation activation and motion.

Dislocation trace line and crystal rotation

Five grains shown in Fig. 3 exhibit a DTL [4, 5, 42] even though their grain orientations and tribological loading differ. Dislocation self-organisation is presumed to be the responsible mechanism for DTL formation [4, 42]. CoCrFeMnNi has a medium SFE [43] resulting in mainly planar slip. It was reported in previous work that these planar slip materials exhibit band-like patterns in the subsurface microstructure after tribological loading [7, 9, 11, 13]. However, exceeding a critical dislocation density might have led here to dislocation self-organisation forming the DTL.

The DTL is only observed in systems with a low friction coefficient and without flake formation in the wear track. The stress state beneath the moving sphere changes with varying friction coefficient [32]. At the same time, the friction force can be interpreted as the sum of forces arising by ploughing and by adhesion [45], with the two being difficult to separate. Therefore, the occurrence of material transfer can further alter the stress state, and to our knowledge, no adequate stress field model exists to cover this behaviour.

The depth of the DTL cannot be correlated with the applied load because the DTL for the experiment SiC/2N/N2 with the lowest friction coefficient and a normal load of 2N is located at the largest distance to the surface (Fig. 3c). This seems to be contrary to results presented in reference [44], but if the lower and not the upper DTL is considered in [44], the results agree well.

The crystal rotation during tribological loading is an important parameter and can influence the dominant deformation mechanism. The smallest rotation of a crystal direction parallel to a principal sample axis is observed for the crystal direction parallel to TD in Figs. 4 and 6. This means that up to a certain stress level, TD is favoured as rotation axis. The crystal rotation around TD was also observed in other studies [4, 10, 44, 46].

The crystal direction parallel to SD in the SiC/2N/air experiment (Fig. 3g), coloured in dark green, first moves along the [001]–[111] bisector towards the [001] pole with decreasing distance to the surface until [113], where the crystal rotates towards the [001]–[011] bisector. This means that the given rotation changes its direction. This might be explained by a change in glide behaviour as the direction parallel to SD is close to [113] at the turning point. In uniaxial tension or compression experiments, the stacking fault width is independent of the stress state along the line between [012] and [113]. If this line is crossed under uniaxial loading, the stacking fault width either decreases or increases, depending on the applied stress [24]. Further tests are required to identify whether this also holds true for tribological experiments.

Summarising this section, dislocation mediated microstructures of CoCrFeMnNi are caused by low friction coefficients, and they favour the formation of a DTL. The crystal rotation is mainly about TD for low friction coefficients.

Twinning

Besides dislocation activity, also twins were observed. The occurrence of twins can be predicted for uniaxial loading when the crystal orientation is known. This is not the case for a tribological load because of the associated complex stress field.

Three different stages of twinning are observed: The twin in the SiC/2N/air experiment (Fig. 3b) is small, and only one twin system was active. The twin interacts with the surface. The deformation layer in the experiment sapphire/2N/N2 (Fig. 3d) also shows only one active twin system. However, the twin itself is large and wide, and it does not have any connection to the nanocrystalline layer above. The grain in the Si3N4/2N/air experiment (Fig. 3e) exhibits two activated twin systems under similar tribological loading. The twins interact with the SAGB in the near-surface region.

As mentioned in the previous section, the force necessary to activate twinning in the SiC/2N/air experiment (Fig. 3b) is most likely a compression stress parallel to ND. The initial grain direction parallel to ND of the grain featuring twins is not within the [001]-[012]-[113] sector, which precludes twinning under uniaxial compression load. Assuming that the force parallel to ND leads to twinning, the observed crystal rotation towards the [001] pole is necessary for twin formation, as the rotation increases the probability of twinning. Since the twins are small and close to the surface, the twins were most probably formed by slip of Shockley partial dislocations on adjacent {111} planes starting from the surface. This mechanism was observed in MD simulations for nanocrystalline grains [47] and thin films [48]. Further analyses of the MD simulation in [49] given in Figure S1 also show that twins form at the surface under tribological load. The HRTEM image in Fig. 5 suggests the same, as the twin is thicker at the surface. Shockley partial dislocations are nucleated at the surface, and some of them may glide over longer distances than others owing to local differences in the elemental distribution [50]. Steps at the twin boundary are clearly visible and support this hypothesis. In Ref. [16], twinning is observed in Cu under tribological load at high sliding speeds and at cryogenic temperatures, and it is supposed to start at the surface roughness sites as well.

The twins generated in the experiment with sapphire/2N/N2 and detected using TKD measurements presented in Fig. 7a–c are of lenticular shape, which is typical of deformation twins in coarse-grained materials. This is an indicator for the Venables mechanism [23] being active.

In the systems discussed so far, only one active twin system was detected, while for Si3N4/2N/air (Fig. 3e), two twin systems are present. In the near-surface region in Fig. 7d–f some SAGBs are visible. The ends of the twins closer to the surface interact with the SAGB. It may be possible that Shockley partial dislocations nucleate at the SAGB and glide on adjacent {111} planes, forming the twins. This seems likely as the twins have their largest width at the intersection with the SAGB. Based on MD simulations, easier activation of twinning processes was observed at rough grain boundaries [47], as is the case for the SAGB in Fig. 7d–f. Nevertheless, the Venables twinning mechanism cannot be ruled out completely. It is possible that the twins were nucleated in the bulk region and then interact with the SAGB. Additionally, as mentioned earlier, part of the twin boundary exhibits HAGB characteristics, likely a result from the compounded effect of significant interaction between twin boundary and dislocations. This implies that in the Si3N4/2N/air case not just twin activity, but also twin-boundary–dislocation interaction is high. In addition, the tribological system Si3N4/2N/air with a different initial grain orientation results in the microstructure presented in Fig. 8. Here, neither a DTL nor twins are detected. Instead, a colour-gradient in the TKD measurement is observed and interpreted as crystal rotation [51] caused by dislocation motion. This demonstrates the tremendous influence of the initial grain orientation on the dominant deformation mechanisms.

Based on this discussion, two different types of twins were identified: The first one seems to be caused by compression stresses along ND when Shockley partial dislocations nucleate at the surface and glide on adjacent {111} planes. These twins form under lower friction forces and in the absence of material transfer. They are also small in size and width, effectively revealed only via HRTEM and not even by TKD. The second type of twins might arise from the Venables mechanism and be caused by a tensile stress parallel to SD.

Conclusions

We investigated the influence of the friction and the normal force on the dominating deformation mechanisms in CoCrFeMnNi for tribological single-trace experiments. A range of friction forces resulted from selecting varying counter body materials and atmospheres. The use of a polycrystalline material allowed us to give first insights into the role of the initial grain orientation. The ensuing subsurface microstructures were investigated by applying STEM, TKD and HRTEM. The following conclusions can be drawn:

  • With the selected tribological systems, dislocation-mediated as well as twinning-induced plasticity microstructures were obtained.

  • The dominating deformation mechanism under tribological load is not only dependent on the friction and normal force, but also on the initial grain orientation and the direction of the crystal rotation.

  • The responsible force direction and mechanism for deformation twins depends on the tribological loading. With low friction coefficients, the crystal direction parallel to ND seems to be decisive for twin formation. This results in compression twins. These twins are most likely formed by glide of partial dislocations from the surface on adjacent {111} planes. With higher friction coefficients, tensile stresses parallel to SD may lead to twinning. These stresses were caused by the adhesive forces and have a tensile character at the trailing edge of the sphere. This means that the crystal orientation has to favour twins under uniaxial tensile load parallel to SD. As these twins are lenticular shaped, they are most likely formed by the Venables mechanism.

  • In materials with a medium stacking fault energy such as CoCrFeMnNi, a dislocation trace line (DTL) can be formed, which is unexpected based on the planar slip behaviour. The loading condition most likely influences the degree of the misorientation discontinuity along the DTL.

  • Twins are favoured for higher friction forces, and a DTL is preferential at higher normal loads.

  • The crystal rotation occurs nearly perfectly around TD at low friction forces.