Coupling of surface and volume behaviors to understand the generation of TTS in pure magnesium proceeded by HPT

This study analyzes the formation of tribologically transformed structures (TTS) in pure magnesium (Mg) using high-pressure torsion (HPT) processing. Generally, studies conducted in such conditions do not focus on surface behaviors. The correlation between the friction and wear phenomena at the surface and the microstructural changes was investigated to supplement the knowledge on TTS formation during the first stage of rotation. An RHEOS apparatus was used to test the samples with an average grain size of 70 μm under a mean pressure of 1 GPa and a rotation speed of 0.5 rpm. The samples were conducted in an unconstrained setup at room temperature. Surface and microstructure changes were examined using optical microscopy, scanning electron microscopy (SEM), and the focused ion beam (FIB). Observation of surfaces shows that friction between the anvils and the surfaces of the sample was set to satisfy the sticking condition. Three different zones in surface contact are identified: the centre zone, the adhesion/sliding zones, and the edge zone, which generate consequently different behaviors. It was found that 0.5 turns of HPT produced a significant refinement in the grain size of the processed Mg. The TTS were considered a zone with a fine microstructure, where the initial grain size was reduced to the range of 300 nm to 1000 nm. The results show that TTS produced in these conditions are not homogenous. The deformation occurs differently, so the TTS were less or more refined. According to the different observations, a scenario of surface degradation was established. The accommodation mechanisms considered are the rupture and shear modes, which occurred, respectively, in the first material and the third body.


Introduction
In the industry, there are a large number of applications in which the parts of the machine into contact are in relative motion.This contact is subject to various alterations, leading to the wear and detachment of particles that form the third body.The detachment of particles is generally preceded by the formation of what is named Tribologically Transformed Structures (TTS) [1][2][3][4][5][6].This phenomenon is related to the formation of a nanocrystalline structure near the bulk surface [1,2,7].Commonly, the term TTS is used to describe all processes that occurred in the skin of bodies in contact and altered their essential properties such as geometrical properties, mechanical properties and thermal properties, but not chemical ones [8].Other terms are also used to define these transformations: Tribological Transformations of the Surface [9,10] and Superficial Tribological Transformation (STT) [11,12] which refer to the location where this layer is formed in the bulk; the White Layer (WL) or White Etching Layer (WEL) [13][14][15], which reflect the appearance of the metal after etching with Nital.
According to the behavior of the material in contact which depends on the material properties (chemical composition, crystallographic structure, grain size, etc) and contact conditions (solicitation mode, speed deformation, temperature, etc), TTS involve several mechanisms (plastic deformation, phase transformations, cracking, etc) that which evolve over time.The expanded definition of TTS is to consider it as a state of the structure of a material obtained under the applied stress to describe a nanoscale grain refinement, whatever any intermediate process.
The TTS were obtained on various materials and in a multitude of situations, mainly in fretting, rolling contacts, and conditions of impact, where are commonly characterized by a marked resistance to chemical etching and high hardness.The values of hardness are significantly greater than those in the bulk material [13].
The first definition of TTS was given by Maurice Godet, considering TTS as the natural response that material uses to protect itself against wear [16], which has helped to develop a new idea based on the control of particle detachment by controlling TTS formation.Thus, researchers have made considerable efforts to clarify the mechanisms responsible for TTS formation, as well as the mechanisms leading to their destruction.Various explanations have been proposed.TTS were generally considered to be formed by temperature and the effects of plastic deformation.However, numerous researches have demonstrated that TTS could result at low temperatures under hydrostatic pressure and shearing gradients [17][18][19].Similarly, the result obtained by Eleöd in [20] demonstrated the preponderance of the hydrostatic pressure compared to the temperature effect.
Thus, severe plastic deformation (SPD) technologies have attracted more attention to studying the wear phenomenon and tribological mechanisms.Many of these technologies have been used, including equalchannel angular pressing (ECAP) [21], friction stir processing (FSP) [22], high-pressure torsion (HPT) [23,24], and other techniques.The authors suggest the possibility of exploring these techniques to improve the properties of materials and wear resistance.
In this study, the experiments were conducted under HPT conditions.This method, introduced by Bridgman in 1935 [25], involves the application of simultaneous compression and shear strains to achieve significant grain refinement.In fact, the HPT method allows precise control of process parameters like shear strain, hydrostatic pressure, and temperature that permit reproduction of contact conditions, such as in real contact.For that reason, recent experiments have investigated these advantages to study the conditions of TTS formation that occur principally under high pressure and shear strain [4,23,26].
The effect of deformation gradients was investigated through modelling in [3,4,24], where the authors confirmed the presence of a deformation gradient in the radial and in depth, for which the TTS formation is considered related to high hydrostatic stress components, high local plastic strains and high strain gradients.Generally, in HPT studies, the behavior of rubbed surfaces is not taken into account and is not exposed.Initially, the process was used to develop nanostructured materials where the primary interest concerning surface behavior was to avoid slippage at the contact interface.Hence, most of the research studies deal with the problem of slippage.Very few investigations have been conducted to date to discuss the evolution of the rubbed surfaces, principally in pure Mg.Accordingly, the present work was initiated to provide the first information about surface behavior and obtain a better understanding of the TTS formation in the investigated condition range.The aim of this work is to establish a scenario of surface degradation.

HPT processing
The HPT method has been applied to a wide range of materials: pure metals, alloys, composites, and ceramics.Generally, the HPT can be applied according to three different HPT facilities depending on the lateral flow in the sample: constrained, quasi-constrained, and unconstrained flow.In this work, experiments were conducted in an unconstrained condition.In such a case, the material is free to flow outward when the high pressure is applied, thereby making the sample much thinner compared to other facilities.

Apparatus
The tests were performed using the RHEOS apparatus at room temperature.The RHEOS is an apparatus available at the LaMCoS (INSA-Lyon).This apparatus was composed of two opposing anvils arranged vertically.Conical anvils were made of tungsten carbide.The sample placed on the lower anvil was compressed between the two anvils.For the first time, hydrostatic compression was imposed on the sample.Once the pressure stabilizes, the lower anvil is rotated at a constant speed.The method of HPT is limited to small samples in the form of a disk with a diameter of 6 mm and a thickness of 0.3 mm.The samples were cut from an extruded bar having a diameter of 6 mm.To success the cut and avoid the deformation of samples, the bar is inserted into a tube of internal diameter equal to 6 mm and of outside diameter equal to 8 mm.Samples were cut by a precision chainsaw, grinded, and polished mechanically.In each experiment, particular attention was given to the anvil misalignment and the problems of slippage at the interface of contact.To ensure good contact between the sample and anvils, anvil misalignment was regularly controlled.For more caution, the upper anvil was used with a larger diameter than the diameter of the lower anvil (for example, a combination of anvils with a diameter of 6 mm and another with a diameter of 10 mm).To avoid the slippage problem at the interface anvils/samples, surfaces were roughened to ensure better adhesion, (figure 1).Before each test, the anvils were polished to a mirror finish, and then roughened mechanically to obtain a roughness of about 0.1 μm.After that, the anvils and the samples were cleaned in ultrasonic alcohol baths and finally, etched with nitric acid for a few seconds.
This apparatus allows for carrying out a great number of experiments in view of the fact that it presents a broad range of varying parameters.In the present work, the load applied on the anvils was 27 kN, which led to a nominal pressure of 1 GPa on the sample's surfaces.The sample was rotated for 0.5 turn at a speed of 0.5 rpm.
Due to the lack of detail of surface behavior in most HPT studies and, principally, the lack of studies concerning the generation of TTS in the Mg under HPT conditions, the input parameters were selected from previous studies [3,4,11,27].During each test, normal loads and, consequently, the pressure, sample thickness, and torque were continuously measured.

Material
The present investigation was conducted using a high purity magnesium (99,97%).The chemical composition and the principal mechanical proprieties of the initial material are show in table 1.The material has an initial average grain size of 70 μm as shown in figure 2. The selection of pure material is justified by the fact that it presents a plastic deformation behavior that is less complicated without any effects of the additional elements, the impurity atoms, the phase transitions, etc In what follows, the TTS are considered the Tribologically Transformed Structures.

Characterization
In this study, the samples were subjected to the characterization of the surface and the microstructure evolution across the thickness using an optical microscope (OM), a Scanning Electron Microscope (SEM), and a Focused Ion Beam (FIB).Observations were made on both faces of the sample in different areas of contact: adhesion and sliding zones.

Results and discussion
3.1.Torque and thickness measurements Figure 3 presents the evolution of the torque (C) and the sample thickness as a function of the angular position (q).The momentary angular position is deduced from the time since the motor started rotation at the preselected speed.Under compression, the deformation of the material rigorously follows the variation in the applied load, wherever the increase in this load introduces further material hardening.During the test, the material was strengthened via work hardening.At the end of the compression, the thickness is reduced to 207 μm, where the variation was approximately 31% of the initial thickness.
When torsion is imposed, the sample continues to be deformed, and the material outflows from the contact.In the first moment of the rotation (angle of rotation approximately 2°), the torsion torque increases quickly up to 3 N.m.This is due to the evolution of the friction at the interface, which consequently induces a twisting in the volume of the sample.The thickness of the sample decreased as the torque increased.The decrease in thickness during this stage was approximately 80%.The torque stabilized after 60°of rotation and reached a maximum value of 3.7 N.m.The shape of the curve-related torque and rotation angle -is strongly dependent on the friction coefficient between the sample and anvil.The stabilization is due to the condition of contact at the sample/anvil interface, which remains unchanged.The correlation between the change in sample thickness and the torque is due to extra work, which is involved in the stage of rotation, associated with the extrusion of the material.After the HPT, the thickness was reduced to approximately one eighth of the initial thickness.In the non-constrained condition, the thickness reduction is significant and more important than in the other HPT facility conditions because there is a continuous decrease in the sample thickness.The torque measurements permit the calculation of the stress strain imposed on the disk by the equation [28,29]: The equivalent strain is given by the relationship: where N is the number of revolutions, r is the radius of the disk and h is the thickness of the sample.To consider the change in the thickness, h was calculated as the mean of the thickness at the begining of the rotation and the thickness at the end of the test.From the equation (1), the shear strain is equal to zero at the centre of the sample and increases linearly with the radius.This relationship implies that deformation is heterogeneous in the radial direction.The stress-strain curve that corresponds to these calculations is illustrated in figure 4. At the beginning, the Stress-Strain curve shows a work hardening, and it appears with a steady state of strain attained at large strains.

Rubbed surfaces morphology
The shape of the Mg sample after the HPT processing is shown in figure 5.It can be observed that the form of the deformed sample is different from the initial state.The details of the rubbed surfaces are shown in figure 6.
The thickness of the sample varied along the radius.Kim [30] explained that the change in the thickness of the sample with distance from the centre results from the higher compressive plastic stress in the centre compared to the outer part during the loading state and the elastic recovery during the unloading, which puts the peripheral region under tensile stress and the central region under compressive stress.
The outflow of the material along the radial direction was greater than the result in [4] tested under the same conditions, which produced a small flow.It is clear that adherence reduces the lateral flow of the sample; however, it can not be eliminated.The formation of the ribbon at the edge was discussed in FEM investigations [31][32][33].Pereira [31] explained that this region is subjected to tensile stresses for the reason that the surface is not in contact with the anvils of the ribbon; where it is free to grow.Furthermore, because the ribbon does not touch  the anvils, it is expected that the temperature of the sample will be slightly lower compared to the other HPT facilities.The search in [33] stated that the extent of the material outflow depends, in addition to the friction conditions at the interface anvil/disk, on the relative size of the sample, where its length may be diminished by increasing the ratio of r/h.
In this work, the analysis of the friction interface allows the division of the contact area into three zones, as shown in figure 6(c).There are some differences between the appearances of the lower and upper surfaces, especially when the tests were conducted with a combination of two anvils with different diameters.Thus, the material is more confined in the lower contact, for the reason that the material flows outwards on the edge and comes bent on the anvil.In the case of the lower contact, the central zone is characterized by the presence of the scratches of the anvils.There has been a plastic flow of Mg surface roughness.This zone is surrounded by a corona of some detached particles, accompanied by some marks of microsliding.These microslidings are not oriented, as they should appear, at every point on the rubbed surface perpendicular to the radial direction.As shown in (figure 7(a)), the friction trace is oriented in the same direction, and for that reason, these microslidings are considered a consequence of the unloading stage.In general, this zone indicates the occurrence of good adhesion at the anvilsample interfaces, therefore, it is called the adhesion zone.The intermediate zone is also marked by the scratches of the anvils and the presence of some detached particles.The morphology of the sample surfaces highlighted the occurrence of adhesion.The zone at the edge is characterized by the plastic flow towards the outside of the contact and the presence of the marks of the anvils.It is further characterized by the ejection of detached particles as well as a few marks of microsliding.
The surface obtained in this study is less degraded than that obtained in the case of iron or steel [3].Consequently, this part of the contact area is known as the sliding zone.A closer analysis of the edge highlights the formation of a layer at the edge of the sample that flows out of the contact surface.This prolongation of the metal is called a tongue.Indeed, by looking at the edge exactly through the holes formed by the detachment of the particles, the sliding marks are visible and are in the prolongation of the marks outside the edge (figure 7).In addition, the view in z-contrast mode confirms the occurrence of good adhesion at the edge, where the anvil's marks are obvious, and also around the zones of detachment of particles.
Figure 8(b) shows the surface at the centre of the contact, which appears slightly cracked.This suggests that the layer formed at the centre of the contact bursts at the time of the contact opening.In this zone, the adhesion is high, and the layer formed, with a fragile structure, can not accommodate the mechanical stress imposed at the end of the test, which is the vertical unloading force, and it cracks.The transfer of Mg on the antagonistic anvil in the form of islands of the third body is important.The morphology of the surface of the adjacent anvil is shown in figure 9.The figure highlights adhesion by the presence of the detachment of material from the sample surface.The details in figure 9 show important islets of the third body that spread and adhere to the anvil.In the central and inter zone zones, the detached material was trapped in the cavity of the anvil and between its asperities.
At the edge, the particles constantly detached from the surface, mainly at the upper contact, agglomerated, and formed a barrier to the ejection of other particles from the contact.The third zone forms the source of the detached particles.In addition, it can be called a debris retention zone.
Once the test ended, the majority of the samples remained adhered to one of the anvils.These observations translate the range of adhesion reached mainly in the centre of the surface compared to the remainder of the surface.The trace in the centre has never been evident in the case of the iron, while in the current work, it has been observed in 30 tests carried out on pure Mg.For that, the trace can be considered the signature of the pure Mg conducted under unconstrained HPT conditions.The central zone is usually subjected to higher pressures where the pressure contact exceeds the nominal pressure, especially at the end of the compression.
The nominal pressure was calculated using classical elasticity theory: P = F/A, where F is the force applied to an area A, assuming a uniform pressure distribution.In practice, the real pressure is higher than the nominal pressure.Bridgman suggested that the pressure should be higher at the centre than at the periphery [24,25].In the central region, the friction stress is high enough to achieve a sticking condition at the interface sample/anvil [30].Several studies have analyzed the evolution of accumulative shear strain and the hardening behavior during the HPT process in unconstrained contact.The simulation results in [3] show that higher compressive stresses are observed in the disk centre and lower stresses at the edge.This can be explained by the effect of the friction present in the centre, where the material meets a higher resistance to flow outward than it does at the edge of the disk.
This denotes that the normal force in this zone is large enough to provide sticking to the anvils, whereas at the edge, slippage takes place.Similarly, the results obtained in [34] illustrated that the microhardness and equivalent plastic strain in the middle regions of the disks were higher than those in other regions.In the case of iron [27], the pressure in the centre is higher than 1GPa, while it is three times higher in deformed grade steel.Busquet [3] suggested that the maximal pressure decreases when the anvil rotation increases, and the distribution tends to become approximately uniform for the highest angles.The author shows that the maximum pressure occurs in an annular region (between a radius of 0.8 mm and 1.5 mm).Searchers in [27] checked that the adhesion radius rA is practically equal to the radial location of the maximal equivalent strain.The observation cross-section was made in the region around the edge (figure 10).These observations give a better image of the presence of the thin layer.

Grain refinement
The resulting microstructure is illustrated in figures 11, 12, and 13. Figure 11 shows the SEM observation realized in the cross-sectional plane of the sample.Overall, there is a significant refinement of the microstructure compared with the initial grain microstructure.Figure 12 exhibits the FIB images made in different regions of the rubbed surface: in the centre of the disk, in the corona trace, and in the outer trace.Comparing the three zones, the refinement occurred preferentially in the corona trace, thereby in the adhesion-sliding zone.In addition, some grains are slightly elongated.In the central zone, where the grain is most constrained, grains were also refined; nevertheless, some coarse grains are detected, especially near the surface.The localization of ultrafine grains is observed below 4 μm the surface.Figueiredo [35] shows that fine grains are formed at the grain boundaries of coarse grains and gradually consume the whole structure.There is an obvious structural variation in the direction of the thickness as well as in the radial direction, which indicates the heterogeneity of the plastic deformation.Overall, the grain size is reduced to the range of 300 nm to 1000 nm. Figure 12 also depicts the presence of some microcracks and voids below and parallel to the rubbed surface, which eventually promote material detachment.
There is some difference in grain size obtained in the different zones of the sample surface.The mean size of the grains in the centre and in the corona zone is close, but the difference is notable compared to other parts where the grain is less refined.It is also apparent that there is some difference in the grain morphology.In general, the grains tend to be equiaxed except in the corona trace, where some grains are elongated.Elongated grains were observed in [36][37][38][39].The researchers suggest the possibility of fragmentation of the elongated grains into equiaxed fine grains.Compared to the results of the pure Mg deformation under different conditions (applied pressures, velocities, number of revolutions with different initial grain sizes, etc), which are summarized in table 2, the present work successfully achieved significant grain refinement.
Figure 13 displays the microstructure developed in the subsurface at a distance of 10 μm below the surface.The grain is less refined compared to the grain at the surface, and the structure produced is also heterogeneous.It is clear that the microstructure observed in the corona trace is less heterogeneous than in the other regions.The findings in previous investigations also showed the formation of a strong bimodal microstructure composed of large recrystallized grains beyond the micron range coexisting with grains in the submicron range [48].Coarser recrystallized grains have also been reported in [49].
In general, the multi-modal grain size distribution is observed in the early stage of deformation and gradually evolves to a homogeneous distribution of ultrafine grains [35].Searchers in [50] claim that the inhomogeneous grain structure is clearly a transitional structure during grain refinement.The inhomogeneous transformation is mainly explained by the fact that the sample was not deformed in an identical manner.On the other hand, elongated grains reveal the insufficient conditions of the experiments to achieve a homogeneous microstructure.Effectively, the deformation is not homogeneous and is mainly related to the distribution of stresses and strains in the disk and also to the friction variation at the interface sample/anvil.
When the adhesion condition dominates, little flow is observed, and the shearing takes place within the sample.
However, with lower friction, a greater flow of material is observed at the edge of the disk.For that, it is necessary to improve adhesion by giving the surface a roughness favourable to mechanical hanging to maintain it in integral contact with the anvils.In recent research, adhesion is not considered the dominant factor in achieving a better result for refining structure and thereby the TTS.It must be associated with the effect of pressure on the plastic deformation [26].That is, the formation of such a structure requires the stabilization of some compromise between a significant increase in strength and good ductility.
During the friction, more than one accommodation mechanism of the plastic deformation can coexist in the contact, and these mechanisms can change, therefore different mechanical response can occur.As highlighted in the above results, an inhomogeneous behavior was recorded on the sample surfaces, in the bulk, and therefore in the TTS production.
Preliminary research [4,26,27] has demonstrated that the transformed layers in the finest microstructure were produced in a specific volume linked with sliding/adhesion contact conditions [3].In this work, the microstructure transformed in this zone also shows a finer grain size.The analysis of its surface shows that the detachment of particles occurs at the opening of the contact when the two surfaces are separated.The islands of detached particles remained stuck on the anvil.These observations reveal that the TTS formed in this zone is more brittle than the initial material.It cracks at the opening of the contact and will fragment, leading to the formation of the third body S3.
Previous results in [51] showed that cracks could initiate and propagate at the interfaces of WEL and the sublayer, leading to the RCF of the bearings.In this work, it was clear that there was a separation between the formed layer and the bulk.The deformation accommodation mechanisms are therefore identified as a combination of the rupture mode and the shear mode (the S1M3 and the S1M2).In the border zone, particle detachment occurs during rotation and upon contact opening.The islands of detached particles are compacted and remain stuck on the anvil surfaces.The accommodation mechanisms considered are the rupture mode and the shear mode, which occur first in the first material (S1M2 and S1M3) and later in the third body, wherever the accommodation mechanisms become (S3M3).In fact, during the unloading, some particles rub on the sample surface when the two surfaces are separated gradually, which is why a few micro-slidings appear on the sample surface.

Conclusion
The generation of the TTS in the case of pure Mg was investigated in this study.In this experiment, axial torque is combined with high axial compression.In such loading conditions, the sample was exposed to shear force, axial compression, and hydrostatic pressure.The sample was successfully processed with unconstrained HPT under 1 GPa, 0.5 turns, and 0.5 rpm.The examination of the rubbed surfaces led to the identification of three zones: the centre zone, the adhesion/sliding zones, and the edge zone, which generate consequently different behaviors.Various responses were observed: particle detachment, accumulation of the third bodies, and transfer of matter, which are governed mainly by plastic deformation, adhesion phenomena, cracking, and abrasion.
The corona trace was formed by an important particle detachment, which highlights the level of stress strain in this area.This finding was coherent with the results obtained in the previous studies.It was evident that the condition of adherence was important to allow significant grain refinement.The initial grain size was reduced approximately by a factor of 230.The structure refinement was also observed in depth; however, the structure obtained was not homogeneous, neither along the radius nor across the thickness.It is clear that TTS formation involves competition between surface and bulk behaviors, in addition to debris generation.

Figure 3 .
Figure 3. Measured thickness and torque as function of the rotation angle.

Figure 6 .
Figure 6.Deformed sample of Mg, (a) lower surfaces, (b) upper surface, (c) schematics different zones on the lower surface: 1-the central zone, 2-the intermediate zone, 3-the zone at the edge.

Figure 7 .
Figure 7. Details of the rubbed surfaces -lower contact: (a), (b) shows the enlarged images of the regions labelled 1, (c) and (d) shows the enlarged images of the regions labelled 2 and 3 in figure 6.

Figure 9 .
Figure 9. Details of the lower anvil surface.

Figure 10 .
Figure 10.View of the tongue cross the section.

Figure 12 .
Figure 12.Local views of the microstructure in the different zones of the top of Mg rubbed surface slight.

Table 1 .
Main characteristics of Mg.

Table 2 .
Summary of final grain size of pure Mg processed by HPT at room temperature for different conditions.C: Constrained contact, QC: Quasi-Constrained contact, UC: Unconstrained contact) *