Influence of microstructure on tribologically mixed layers

Influence of microstructure on tribologically mixed layers

Wear 271 (2011) 792–801 Contents lists available at ScienceDirect Wear journal homepage: www.elsevier.com/locate/wear Influence of microstructure on...

3MB Sizes 1 Downloads 8 Views

Wear 271 (2011) 792–801

Contents lists available at ScienceDirect

Wear journal homepage: www.elsevier.com/locate/wear

Influence of microstructure on tribologically mixed layers W. Tuckart a,∗ , L. Iurman a , E. Forlerer b a b

Departamento de Ingeniería - Universidad Nacional del Sur Avd. Alem 1253, 8000 Bahía Blanca, Argentina Departamento de Materiales, CNEA-CAC, Av. Gral. Paz 1499 (B1650KNA), San Martín, Pcia. Bs. As., Argentina

a r t i c l e

i n f o

Article history: Received 13 July 2010 Received in revised form 1 March 2011 Accepted 25 March 2011 Available online 16 April 2011 Keywords: MML Accumulated damage Wear reduction

a b s t r a c t The purpose of this study is to determine the influence of microstructure on the formation and mechanical stability of tribologically mixed layers acting as a wear reduction factor. To this purpose, AISI 420 steel specimens of different initial microstructure were subjected to annealing, quenching and tempering to 673 K and to quenching and tempering to 943 K under pure sliding conditions. The wear level for each case was evaluated on the basis of the applied load levels and in relation to the morphological features of the underlying surface affected by wear. Also locally stressed field gradients were identified by means of microhardness profiles, which were then correlated with measurements of the resulting plastic subsurface deformations on the basis of the distance to the tribolayer. On the basis of the results, the conditions of formation and mechanical stability, the relations between the mechanical properties of the initial microstructure and the needed level of stress due to the resulting plastic deformation for the tribological layer to act as factor of wear rate reduction were assessed and are herein discussed. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Wear rate controls the transition from severe wear to mild wear, especially for non-lubricated systems and it is strongly related to the magnitude and distribution of local strains and with stress conditions variation within a small volume of the material in the subsurface area. In the past, several researchers analyzed (particularly in Al alloys against ferrous alloys) the influence of mechanically mixed layers (or MML) produced during sliding process as the ones that result from plastic flow, debris oxidation, fragmentation, transfer, compactation, and mechanical mixing. Such layers can conduce to protective effect of the rubbing wear affected region, increasing the tribosystem resistance wear [1–5]. On the basis of the results of wear tests made on Al–SiC composite materials, Venkataraman and Sundararajan [6] concluded that for the formation of a tribo-layer it is essential to exceed a threshold of subsurface strain resulting from shear stresses whose magnitude depends upon the existing SiC particles concentration values. The same authors in another research paper [7] with Al alloys and the Al-MMC concluded that bulk strength does not correlate

∗ Corresponding author at: Engineering Department - Universidad Nacional del Sur Avd. Alem 1253, 1◦ , 8000 Bahía Blanca – R, Argentina. Tel.: +54 291 4595156; fax: +54 291 4595157. E-mail address: [email protected] (W. Tuckart). 0043-1648/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2011.03.012

with the wear and friction behaviour. Additionally, there exists a strong correlation between transition behaviour from mild to severe wear, and hardness, thickness and composition of MML. On the other hand, the removal of the MML or its non-formation is responsible for the severe wear regime. From a low alloy steel study, Tarassov and Kolubaev [8] reported complex dependence between load and sliding speed with the coefficient of friction due to structural changes occurring in the specimens subsurface layers under the cooperative action of both temperature and deformation. However, the effect of microstructure characteristics in iron alloy on the mechanical stability of tribologically mixed layers enabling the latter to act as a means of wear rate control has not been fully analyzed so far. The purpose of this paper is to identify the influence of initial microstructure on the formation and stability of a tribologically mixed layer that may restrain the wear rate of the tribosystem. The accumulated gradients of damage resulting from subsurface plastic deformation are herein estimated by determining the stress fields accumulated by deformation. The attained results were correlated with those obtained from bulk wear and mechanical properties in order to discuss the effects of subsurface hardening on friction and wear tribolayer behaviour. In this study, it has been experimentally demonstrated that for the tribological layer to act as wear-protective not only are necessary the conditions to enable its formation, but also the attainment of its mechanical stability.

W. Tuckart et al. / Wear 271 (2011) 792–801

793

Table 1 Maximum contact stress in the initial time as a function of the normal load applied. Applied load [N] Maximum contact stress [MPa]

200 1638.4

275 1821.9

2. Materials and methods 2.1. Material and equipment For this purpose AISI 420 steel (wt%: 0.42 C%, 13.89 Cr%, 0.75 Mn%, 0.33 Si%, 0.0027 P%, 0.0098 S% and Fe balance) 40 mm diameter and 5 mm thick annular test samples were prepared. Some of the samples were tested in such as-received microstructure conditions (named “Rec”), while others were austenized to 1303 K during 30 min followed by oil quenching. A number of specimens were tempered to 673 K for 45 min (T673). Still other samples were tempered to 943 K for 45 min (T943). For each case, counting of carbide size distribution in the initial microstructures was made using a Nikon NIS Elements D3.1 software. Wear tests under pure sliding conditions (one sample mobile on a fixed one) with an Amsler wear machine using normal constant loads of 200, 275, 350, 425 and 500 N were carried out, at a temperature of 293 K and 40–50% ambient relative humidity. The initial contact stresses produced were calculated using the Hertz theory [9] and shown in Table 1 ( = 0.3 and E = 200 GPa) [10]. The relative speed between the probes with sliding contact was of 0.80 m/s, for a total of 15,000 rotations (equivalent to 1728 m of distance). In this experimental work only the fixed specimen was analyzed as it was the one subjected to pure sliding conditions. Quantitative wear by the weight difference technique with a scale of 10−4 grf sensibility was used and the debris produced during the tests were collected every 216 m of sliding distance (5 min) and weighed after the test. Although the debris came from both specimens, for the most part they correspond to fixed specimen, mainly during the running-in stage due to change in the geometry contact – noconforming to conforming. The triboxides phases were identified by X-rays diffractometry (XRD) by means of a RIGAKU Denkid, model max IIIC equipment, between 2 = 20–90 with a Cu K␣ = 1, 5405 A˚ with a conventional /2 Bragg–Brentano symmetric geometry. Rietveld method by means of MDI Jade software was used to analyze the X-rays diffraction patterns. During the tests, friction torque and temperature in the specimen near the contact zone were registered and fed into computer via interface as functions of time (60 sample by min); specifically, temperature was measured by a chromel–alumel 1 mm diameter thermocouple inserted into the fixed probe ∼1 mm below the sliding surface as shown in Fig. 1. After the test, the width of the wear

Fig. 1. Reference pattern of the wear scar and thermocouple disposition in fixed probe during the testing.

350 1974.5

425 2107.2

500 2225.4

scar was measured to determine the final contact stresses as indicated in Fig. 1. The tribosurface and subsurface were analyzed by optical microscopy (OM) and the bulk by scanning electronic microscope (SEM) on Philips XL30, with secondary electrons detector. The mechanical characterization of the bulk was made with Vickers hardness and the microstructure was studied by optical microscopy and 0.020 and 0.040 kgf Vickers microhardness tests. The metallographic sections were begun by cutting crosssectional to the tribosurface and the lateral side ground manually using 2000 grit abrasive paper and finally polished with 0.5 ␮m Al2 O3 water-based slurries. For optical microscopy ‘Marble’ reagent (100 ml H2 O, 100 ml hydrochloric acid and 20 g copper sulphate) was used. 2.2. Estimate of subsurface stress fields By applying the model proposed by Nobre et al. [11] and using Vickers microhardness profiles, local stress fields were estimated. This method linearly relates the increase in hardness with the field of stresses that may be estimated for every level of plastic deformation in terms of hardness using the following equation:  = EH

 1 + H  H0

where  is the local stress,  EH is the substrate elastic limit,  is a dimensionless constant, H is the relative hardness variation and H0 the substrate hardness. According to Nobre [11], the value of  is 2.8 when microhardness values are used for the calculations. Because this model ignores possible microstructurals changes like phase transformations caused by heating friction, it was applied only on samples in which the metallographic subsurface studies did not show evidence of transformation in the tribolayer/bulk interface region. 3. Results 3.1. Microstructural characterization Fig. 2 shows the samples microstructures with one aspect in common: in all them a fine distribution of particles (FeCrC carbides precipitation) presenting different shapes can be observed; namely, rounded, globular whereas others are elongated and bigger, localised either in grain boundaries for Rec samples or in grain boundaries of the previous austenite for the T673 and T943 microstructure. The Rec sample with ferritic matrix posses a 245 ± 3 HV10 microstructure hardness, while both T673 (attained 542 HV10 ) and T943 (attained 322 HV10 hardness level) have martensitic matrix due to quenching and tempering heat treatment. This is in coincidence with a part of the PhD dissertation realized by Tuckart [12], carried out in AISI 420 with the same austenised and quenched conditions and analyzed using carbon thin foils in transmission electron microscopy (TEM), from which it was observed that the precipitate carbides present from martensite microstructures (T673 and T943) are bigger than those of the annealing condition. The sizes distribution is dependent on austenizing temperature [13].

794

W. Tuckart et al. / Wear 271 (2011) 792–801

Fig. 3. Distribution of amount of particles as a function of their size.

3.2. Sliding wear tests The results of wear tests in terms of weight loss according to the applied test load are presented in Fig. 4. The figure shows that all microstructures exhibit a maximum weight loss which is determined for the T943 and Rec samples at the test load of 350 N, and for T673 sample in 275 N. Loss weight depletion was observed both after and before these “threshold load” values. For Rec specimens under an intermediate load, wear was increased four times as compared to that recorded at lower load; on the other hand, a double weight loss reduction was determined regarding the wear recorded at 500 N. This behaviour is due to the formation of a tribological layer and shall be later analyzed in higher detail. On T673 specimens, at the applied load of 275 N weight loss rises three times higher than at 200 N; however, when a bigger load was applied, a low wear reduction was observed due to the influence of white etching formation. This shall also be analyzed later on. In a similar way, in T943 specimen wear increase was observed when the applied load grew from 200 to 350 N; i.e., 6 times higher. And from that point to 500 N, wear reduction of about 1.3 times was observed. Additionally, the contact stresses measured after the test from wear scar caused by wear processes are shown in Table 2. From this, a rise of load bearing capacity when the normal load applied was bigger than the threshold value can be inferred. Fig. 2. Image of SEM of microstructure at 10 k×: (a) Rec, (b) T673 and (c) T943.

Usually, such material shows fracture toughness values of the order of 49.7 and 83.1 MPa (m)1/2 for tempering temperatures of 723 and 923 K, respectively; i.e., martensitic microstructures like those studied in the present work [14]. In order to estimate the characteristics of particle size distribution in the different microstructures, the sizes of the carbides inside an area of 300 ␮m2 were determined. The size measurement was established from the longitudinal axis and the particles in grain boundaries were excluded. The results are shown in Fig. 3 and from this it follows that most of the particles are mainly 250–500 nm long. The T943 sample presents quantitatively a lot more particles than the other microstructures and on the contrary, T673 sample shows fewer particles and a higher presence of coarse precipitates (see the range at 1000–1750 nm in Fig. 3).

Fig. 4. The variation of the weight loss as a function of test applied load.

W. Tuckart et al. / Wear 271 (2011) 792–801

795

Table 2 Contact stresses measured after the tests from the produced wear scar (see Fig. 1). Sample

Load [N] 200

275

350

425

500

3.939 5.762 5.554

5.316 7.957 6.997

7.426 9.776 8.371

Contact stress [MPa] Rec T673 T943

3.764 4.484 5.706

4.298 3.679 5.715

On the basis of the remarkable difference in wear behaviour with the applied load, it was decided to focus on the study of samples with smaller, higher and threshold load values with the aim of analyzing the process that influences the wear.

Fig. 6. OM micrograph in side view at 2000× from Rec-500 N specimen. Microindentations with the same load on MML and PDR can be observed.

3.3. Features of the subsurface beneath the tribosurface

Fig. 5. Image of SEM in side view of tribological layers. Rec samples tested at 200 N at 1500× (a), 350 N at 750× (b) and 500 N 750× (c).

The metallographic analysis of the subsurface region, just beneath the worn surface in Rec samples shows that, during a tribological process occurred under all the applied test loads, a stratified surface layer is formed in which no carbides as those in the surface layer (see Fig. 3) were found. This layer has been called “mechanically mixer layer” (MML) [1,2]. Whereas MML of samples tested at 200 and 350 N exhibited a highly fragmented tribolayer; i.e., low continuity, MML of samples tested at 500 N presented a compact and highly stratified one (Fig. 5). In the same figure, a region beneath the MML with material flow lines that progressively bend towards the sliding direction and that according to the applied load have different thickness (as measured from the MML to the bulk) can be observed. This is named “plastically deformed region” (PDR). Besides, on MML/PDR interface no significative microstructural transformation could be evidenced, as seen in Fig. 6. Furthermore, in samples Rec-500 N evidence of turbulent plastic flow was observed. According to Venkataraman and Sundararajan [6], such condition is the result of large subsurface shear instability that allows mechanical mixing. On T673 samples, subsurface metallographic studies in lateral view carried out after the test revealed that only in the specimens tested at the loads of 200 N and 275 N, a tribological layer could be observed. At a minor load, one third of the contact area was naked, specifically, on starting region of contact. Meanwhile, the rest of the surface was covered by a MML with both thickness varying from a few ␮m to 200 ␮m and ∼822 HV0.02 – presenting damage due to contact fatigue cracks (see Fig. 7). Besides, cracks similar to those developed by thermal fatigue were indentified. This damage could be produced by residuals stresses during the cooling after the test. Beneath the MML, a well compact and defined layer was identified as non-stratified, without material flow lines or grain boundaries – with ∼5 to 75 ␮m thickness and 595 HV0.02 . Below this layer, a soft zone with 381.9 ± 21 HV0.02 and 90 ␮m of thickness was found underneath which, bulk hardness grows up at 529 HV0.02 . These layers could be caused by effect of friction heating and it was the only one subsurface with this feature in the present work. When the wear test load of 275 N was applied, a thinner MML formation regarding the previously mentioned one could be observed. However, higher damage by contact fatigue cracks was detected although the MML presented similarities such as covering level on tribosurface (see Fig. 8).

796

W. Tuckart et al. / Wear 271 (2011) 792–801

Fig. 7. OM micrograph in side view at 200× from T673-200 N specimen. MML and thermal affected layer may be observed.

Fig. 9. OM micrograph in side view at 500× from T673-500 N specimen. “White etching” layer can be observed.

On the other hand, all T943 samples showed the continuous formation with 1100–900 HV0.02 40 ␮m thickness MML in probes that were tested at the test load of 200 N (see Fig. 10a). Likewise, it was observed an 800 HV0.02 30–50 ␮m thickness layer in samples tested at 350 N and an 80–100 ␮m thick layer in samples tested at 500 N (see Fig. 10b). An overview about the results obtained from the microstructural characterization of samples that shows MML and PDR is presented in Table 3. 3.4. Flow stress fields in the plastically deformed region

Fig. 8. OM micrograph in side view at 1000× from T673-275 N specimen. Arrows show cracks on MML.

At the wear test load of 350 N, a well-known “white etching layer” (WEL) formation – named due to its feature appearance under optical microscopes after etching – from 6 ␮m of thickness and 911 ± 20 HV0.02 was observed. The thickness of this layer grows with the increase in load up to 18–25 ␮m when 500 N was applied. Below the WEL formation was observed a region with ∼500 ␮m of thickness and approximately 649 HV0.02 . Because subsurface studies did not show material flow lines, this harder zone could be promoted by the friction heating, considering that the AISI 420 shows a secondary hardening at tempering temperatures of 753–923 K [10]. The OM micrograph in Fig. 9 illustrates an image of WEL of a side view of the sliding subsurface a T673-500 N sample.

Fig. 11 shows the results of local stress fields of T943 specimens tested at 350 and 500 N presented together with those obtained from Rec samples to enable the comparison. It can be observed that T943-500 N specimens reached a maximum stress of 1823 MPa on the surface which progressively decreases to the bulk value at 80 ␮m of the tribolayer, while for T943-350 N specimens, surface stress values are in the order of 779 MPa and progressively decrease to the normal value at the depth of 10 ␮m of the tribolayer. These results show that under 350 N, the values of the stress level acting on the subsurface of Rec and T943 samples are similar. However, at the test load of 500 N, T943 samples have on their surface a stress field that is 280 MPa higher than for Rec samples, though in the latter the field reaches a depth of 230 ␮m. 3.5. Wear mechanism All the Rec samples show the same wear behaviour – with different wear rate as shown in Fig. 12. At the beginning of the run, due to the initial high contact stress is supported by a small bulk volume (see Table 1), intense adhesive wear occurs as the principal wear mechanism (see Fig. 13), thus high wear rate was developed

Table 3 Overview of subsurface microstructure results obtained from probes with MML. Sample Rec Applied load [N] MML thickness [␮m] MML hardness [HV0.02 ] MML continuity level PDR thickness [␮m] PDR maximum hardness [HV0.02 ] Substrate hardness [HV0.02 ]

200 ∼20 790 ± 15 Very low ∼60 373 ± 12 263 ± 16

T673 350 ∼50 800 ± 30 Low ∼60 406 ± 22 263 ± 16

500 ∼50–100 810 ± 27 High ∼230 537 ± 25 263 ± 16

200 ∼200 822 ± 24 High – – 546 ± 20

T943 275 ∼60 888 ± 13 High ∼5 594 ± 15 546 ± 20

200 ∼25–40 922 ± 27 High ∼15 464 ± 2 360 ± 2

350 ∼30–50 802 ± 30 High ∼5 396 ± 17 360 ± 2

500 ∼80–100 807 ± 25 High ∼15 546 ± 15 360 ± 7

W. Tuckart et al. / Wear 271 (2011) 792–801

797

Fig. 12. Debris as a function of sliding distance in Rec samples.

Fig. 10. OM micrograph in side view at from T943 specimens: (a) tested at 200 N at 1000× and (b) tested at 350 N at 750×.

which promoted the modification of the contact geometry – from non-conforming to conforming – thereby, contact stresses are reduced and the stationary specimen made a “shaving effect” on rotating counterface. That condition causes the change of the wear mechanism. Debris distribution as function of sliding distance (see Fig. 11) lets infer a wear rate down after 648 m of slide because oxidative wear was developed. In this process, the stationary disc worn surface is under continuous contact, oxygen is ineffective in diffusing towards this surface, so the formation of these oxides mainly

Fig. 11. The variation of the local fields of stress caused by work-hardening as a function of the depth from MML on Rec and T943 samples tested at 350 and 500 N.

Fig. 13. SEM micrographs at 200× in upper view of tribosurface Rec-500 N after 216 m of slides.

takes place on the tribosurface of the rotating sample during the out-of-contact time [15]. In Fig. 14, the triboxides collected in Rec-500 N after 648 m of slide are shown and their X-ray diffraction pattern is presented in Fig. 15. It is worth noticing a prevalence of sharp peaks of phase hematite iron oxide ␣-Fe2 O3 , and others of phase ␥-Fe2 O3 and also characteristic diffraction peaks of ␣-Fe [16].

Fig. 14. SEM micrograph at 1750× of debris collected from Rec-500 N after 648 m of slide.

798

W. Tuckart et al. / Wear 271 (2011) 792–801

Fig. 15. X-ray diffraction pattern between 29◦ and 70◦ of triboxides collected in oxidative wear stage of test with 500 N.

Fig. 18. Debris as function of sliding distance in T943 samples.

In particles and tribosurfaces from 648 m of slide to finishing the test, the mechanism was oxidative wear with triboxides similar to those present in Rec samples. On the other hand, in T673-275 N initially contact fatigue wear was found and after 1080 m of slide until the end of test, oxidative wear mechanism was detected. With 500 N of load test, a similar mechanism to that developed in Rec samples was observed. Regarding T943 samples, the examination on tribosurfaces and the ejecting particles from the contact zone suggest the presence of wear adhesive and oxidative mechanism, such is the case in Rec samples (see Fig. 18). 3.6. Coefficient of friction and evolution of contact temperature

Fig. 16. Debris as function of sliding distance in T673 samples.

The analyses of worn surfaces and the debris collected during the wear tests in T673 samples under 200 N of load applied show a debris loss rise from 648 m of slide (see Fig. 16). This condition suggests that the development of contact fatigue wear mechanism increases ejected debris weight without detecting oxidative wear. Fig. 17 shows an image of collected laminar particles from 432 to 648 m of slide.

Fig. 19 shows the coefficient of friction (COF) evolution during some tests and its load dependence. From this, it can be inferred that at upper loads, Rec samples show higher value of COF – approx. 0.76, typically of seizure process – with respect to T673 sample. Although the T943 sample registers evidence a smaller COF value (0.65), after the running-in (after of 216 m of slide) all specimens show similar friction behaviour. At the test load of 200 N, similar behaviour (from ∼0.4 to ∼0.3 COF value) was registered in all samples. On the other hand, the evolution of the temperature during the test and its dependence with the load applied is illustrated in Fig. 20. 4. Discussion 4.1. The mechanically mixed layer (MML)

Fig. 17. SEM micrograph at 1500× of debris collected from T673-200 N after 432–648 m of slide.

The analysis of the results obtained from metallographic studies and tests carried out to determine wear from weight loss made on Rec and T943 samples, allowed us to conclude that wear rate is controlled by the formation of a tribological layer (MML). It is also worth noting that for steel wear a critical load value is not inferred as has been observed for metal matrix composite materials [17]. The interpretation between temperature level vs. weight loss on Rec and T943 probes (see Figs. 4 and 20), shows that when the test load rises from 350 N to 500 N, wear rate tends to decrease as temperature grows – also with work-hardened depth (Fig. 11). Those observations are similar to the study on relations between temperature, friction and wear made by Khanafi-Benghalem et al. [18] and reveal the complex relations involved during the sliding wear process. The total COF value during sliding process is the result of different micro-mechanisms that influence it. The greater COF values

W. Tuckart et al. / Wear 271 (2011) 792–801

Fig. 19. COF on Rec and T943 samples tested with 200, 350 and 500 N and on T673 tested under loads of 200, 275 and 500 N.

obtained of Rec samples with regard to T943 could be caused by higher subsurface plastic deformation level developed mainly during the running-in stage (see Fig. 19). In the same way, the mechanical energy of frictional contact is transformed and can be stored in the tribosystem or dissipated in a number of different ways just as heat, sound, incandescent light, or the creation of new surfaces by fracture [19]. As the energy process is shearing in different manners, it is possible that Rec samples tested at 350 and 500 N have similar friction behaviour during the sliding process, but different wear rate (see Figs. 4 and 19). On the other hand, the best wear behaviour in the present study was found with T943 probes when 200 N was applied. That condition and the better results on Rec and T943 samples tested at 425 and 500 N regarding samples evaluated with other loads values can be attributable to the influence of MML that turns out to be “functional”, i.e., it is continuous and capable of decreasing the wear rate of the tribological system (see Fig. 4). However, once the layer is formed, although not “functional”, it increases the wear rate of the system due to the fact that while

799

Fig. 20. Temperature near the contact zone registered in Rec, T673 and T943 samples.

stability conditions are not attained, the layer is ejected from the worn surface, e.g., Rec 350 N samples. As postulated by Fillot et al. [20], MML formation involves two competing processes: one that entails agglomeration, mechanical mixing, compactation, sinterization of particles and extrusion of MML (observed like lip formation to the periphery) and another that implies the partial transfer, fracturing and ejection of the tribolayer. For the tribolayer to be formed, the first of the above mentioned processes should prevail over the second. From metallographic of Rec 350 N and 500 N probes in particular, one can observe the formation of MML (Fig. 5); however, they present remarkably different wear behaviour between them, despite initially having the same ductile microstructure. The extensive subsurface feature studies of these samples do not reveal evidence of significant phase transformation in the MML/PDR interface region (see Fig. 6), but different subsurface plastic deformation levels can be observed in Fig. 5. According to Alpas et al. [21], during sliding contact, a great hydrostatic pressure may be developed in the area that lies beneath the contact

800

W. Tuckart et al. / Wear 271 (2011) 792–801

Fig. 21. COF during the first distance of slide of Rec, T673 and T943 samples tested at 200 N load.

surface. This improves ductility and delays the onset and progression of damages. On the contrary, an extended plastic deformation in the deformed area has the opposite effect of promoting failure. When hydrostatic pressure is large, a high level of local stresses is required to nucleate voids. In this sense, the results of local stress fields shown in Fig. 11 reveal that in Rec 500 N PDR probes a high level of hydrostatic pressure is attained ( = 1544 MPa), considering that the maximum Ultimate Tensile Strength value (in as-annealed conditions  max = 690 MPa [10]) has been widely exceeded. Under high applied load values, MML attains “mechanical stability”; i.e., conditions are given for both its formation and structural integrity – enough to be functional. Such conditions are promoted by the non-homogeneous deformation which, in turn, brings about hardening due to the plastic deformation in the PDR zone. It is inferred that in this situation, the local hydrostatic pressure level could prevail in the subsurface region of the contact surface and this in turn, facilitates ductility regarding the damage promoted by the same plastic deformation. In Rec-350 N samples, the value of the maximum local stress is lower ( = 1014 MPa) than in the 500 N and therefore, so should its spherical state of stress be. Under these conditions for attaining hardening levels similar to those produced on T943-350 N (see Fig. 11), a longer sliding distance is required. Such condition is corroborated by bigger amount of collected debris in Rec-350 N respect to Rec-500 N during the running-in (see Fig. 12). At this stage, the occurrence of damage and coalescence of voids prevails over hardening due to plastic deformation and therefore, although the MML is formed, it does not become functional as it does not attain the needed mechanical stability conditions and so it is ejected from the surface. That condition is suggested by the hardness measured in the PDR, in particular under great loads (see Table 3). The best wear behaviour observed in T943 sample, in comparison with Rec samples, is also associated with the fact that an “intermediate” microstructure is present which, though having less ductility than the annealed one, maintains some degree of plastic deformation hardening capability. By comparing the thickness values presented in Table 3 and the results of the stress fields from Fig. 11, it may be inferred that the levels of spherical state stress effects developed on T943 samples enable the prevalence of the PDR plastic deformation hardening over the damage gradient caused by plastic deformation. That is evident from the COF registers during the first metres of slide in lower load test (see Fig. 21), where the initial COF value of Rec samples is bigger (0.54). This increment in size could corre-

spond to subsurface plastic deformation with intense accumulated damage causing higher initial wear rate during the running-in stage (Figs. 12 and 19). On the other hand, T673 probe only shows a curve – with a maximum 0.40 COF value – without initial sharp peak due to the lack of plastic deformation; i.e., it is consistent with the development of damage by contact fatigue wear. Finally, T943 sample shows that the COF rises to intermediate value, which promotes better conditions for the development of a functional MML, i.e., it is continuous and able to protect the tribosurface with regard to other microstructures still with initial higher hardness. As above mentioned, MML formation requires mechanical mixing; i.e., the existence of a turbulent plastic flow in the near surface layer such as that observed on ductile microstructure Rec and T943 samples [22]. However, harder reinforcing particles than those of the mating matrix materials are also required for its mechanical stability. This was identified by Venkataraman and Sundararajan [7] from a study of sliding wear of metal matrix composite (MMC) type Al–SiC between 10% and 40% SiC in the MMC. In this sense, Kato [23] supplied different sizes of FeO to sliding Fe–Fe and found that only with particles <500 nm is a severe–mild wear transition promoted. Both the low fracture toughness and the great number of particles with 1000–1750 nm size in the initial microstructure – which should reinforce even bigger oxide particles during the wear process – could explain the development of nonfunctional MML and the predominance of contact fatigue wear as the main wear mechanism in T673 specimens tested under low load. 4.2. The white etching layer (WEL) Although layers affected or produced by the friction heating were found, such as those shown in Figs. 7 and 9, the records obtained do not reveal extreme temperature levels. It is clear that intense heating was produced within a few microns at surface level attributable to plowing mechanism caused by interaction of high deformed protuberance transfer due to intense adhesion. In fact, incandescent debris was observed several times during T673-500 N test. In such a way, when high hardness microstructure was evaluated under bigger contact stress, wear rate decreased due to the influence of a WEL formation with 911 ± 20 HV0.02 . This seems in agreement with the observations obtained by Sun et al. [24] from tests made under similar conditions on AISI 440C martensitic stainless steel. They observed a wear reduced level due to the effect of WEL with 20 ␮m of thickness and 1300 HV0.1 . We speculate that the bigger hardness level regarding the white etching layer observed in the present work is due to greater carbon concentration of AISI 440C steel (1.25%C). WELs are intensely studied in order to evaluate their influence on surface rails performance. However, their composition and structure as well as their origin are still under discussion [25,26]. Besides, Tarassov and Kolubaev [8] found a WEL formation during the friction sliding of martensitic steel and they conclude that under high pressure and sliding conditions, the induced intense local temperature close to austenite phase transformation and selfquenching by the fast cooling sublayers results in the formation of martensite microstructure with extremely small grain size and morphology and hence, with high hardness. Those features are consistent with the results obtained from higher hardness microstructure that were evaluated in our test conditions, but just when initially ∼2000 MPa or larger contact pressure was applied (see Table 1). Under lower pressure levels, the damage by contact fatigue increases the contact area (see Table 2) and reduces local stress, not providing conditions for the WEL development.

W. Tuckart et al. / Wear 271 (2011) 792–801

5. Conclusions On the basis of the results obtained from the study of subsurface microstructure before and after the wear caused by dry sliding of martensitic AISI 420 stainless steel, the following conclusions may be drawn: The wear rate of a material with intermediate ductility is controlled by the formation of a tribologically mixed layer, which once mechanical stability conditions have been attained, becomes functional as a factor of wear reduction of the tribosystem. This fact explains the minor wear rate under a minor pressure in materials with higher hardness level. The condition of a functional tribologically mixed layer as a factor of wear reduction has also been observed in probes with greater initial ductility, but just when the friction process promoted a high strain at a subsurface level. However, the tribolayer may be formed, but does not become functional, on the contrary, it is ejected and wear level is increased. In the material with initial higher hardness level and low deformation hardening capacity used for this study, no tribolayer functional formation was observed. However, at highest loads the wear was reduced due to the influence of the white etching layer formation. Acknowledgements The authors wish to express herein their appreciation for the support given to the Engineering Departament of Universidad Nacional del Sur and especially to PhD Irwin L. Singer from NRL for his valuable contributions to the discussion of this paper. References [1] D.A. Rigney, Transfer, mixing and associated chemical and processes during the sliding of ductile materials, Wear 245 (2000) 1–9. [2] J.L. Young Jr., D Kuhlmann Winsdorf, R. Hull, The generation of mechanically mixed layers (MMLs) during sliding contact and the effects of lubricant thereon, Wear 246 (2000) 74–90. [3] M.J. Ghazali, W.M. Rainforth, M.Z. Omar, A comparative study of mechanically mixed layers (MMLs) characteristics of commercial aluminium alloys sliding against alumina and steel sliders, J. Mater. Proc. Technol. 201 (2008) 662–668. [4] R.N. Rao, S. Das, D.P. Mondal, G. Dixit, Dry sliding wear behaviour of cast high strength aluminium alloy (Al–Zn–Mg) and hard particle composites, Wear 267 (2009) 1688–1695.

801

[5] R.N. Rao, S. Das, D.P. Mondal, G. Dixit, Effect of heat treatment on the sliding wear behaviour of aluminium alloy (Al–Zn–Mg) hard particle composite, Tribol. Int. 43 (2010) 330–339. [6] B. Venkataraman, G. Sundararajan, The sliding wear behaviour of Al–SiC particle composites. II. The characterization of subsurface deformation and correlation with wear behavior, Acta Mater. 44 (1996) 461–473. [7] B. Venkataraman, G. Sundararajan, Correlation between the characteristics of the mechanically mixed layer and wear behaviour of aluminium, Al-7075 alloy and Al-MMCs, Wear 245 (2000) 22–38. [8] S. Tarassov, A. Kolubaev, Effect of friction on subsurface layer microstructure in austenitic and martensitic steels, Wear 231 (1999) 228–234. [9] K.L. Johnson, Contact Mechanics, University Press, Cambridge, 1999. [10] Stainless steel wrought in properties and selections irons Steel and HighPerformance Alloy, ASM Metals Handbook, vol. 1, ASM International, OH, 1994, pp. 841–907. [11] J.P. Nobre, A.M. Dias, M. Kornmeier, An empirical methodology to estimate a local yield stress in work-hardened surface layers, Exp. Mech. 44 (2004) 76–84. [12] W. Tuckart, PhD Thesis, Universidad Nacional del Sur, Argentina, 2008. [13] A. Nasery Isfahanya, H. Saghafiana, G. Borhani, The effect of heat treatment on mechanical properties and corrosion behavior of AISI420 martensitic stainless steel, J. Alloys Compd. 509 (2011) 3931–3936. [14] Fatigue and Fracture Properties of Stainless Steels, ASM Metals Handbook, vol. 19, ASM International, Ohio, 1996, pp. 1826–1827. [15] G. Strafellini, A. Molinari, Dry sliding wear Ti–6Al–4V alloy as influenced by the counterface and sliding conditions, Wear 236 (1999) 328–338. [16] T. Quinn, J. Sullivan, D. Rowson, Origins and development of oxidational wear at low ambient temperatures, Wear 94 (1984) 175–191. [17] S. Yu, H. Ishii, K. Tohgo, Y. Cho, D. Diao, Temperature dependence of sliding wear behavior in SiC whisker or SiC particulate reinforced 6061, aluminium alloy composite, Wear 213 (1997) 21–28. [18] N Khanafi-Benghalem, E. Felder, K. Loucif, P. Montmitonnet, Plastic deformation of 25CrMo4 steel during wear: effect of the temperature, the normal force, the sliding velocity and the structural state, Wear 268 (2010) 23– 40. [19] P.J. Blau, The significance and use of the friction coefficient, Tribol. Int. 34 (2001) 585–591. [20] N. Fillot, I. Iordanoff, Y. Berthier, Wear modeling and the third body concept, Wear 262 (2007) 949–957. [21] A. Alpas, H. Hu, J. Zhang, Plastic deformation and damage accumulation below the worn surface, Wear 162–164 (1993) 188–195. [22] A. Rosenfield, A shear instability model of sliding wear, Wear 116 (1987) 319–328. [23] H. Kato, Severe–mild wear transition by supply of oxide particles on sliding surface, Wear 255 (2003) 426–429. [24] Y. Sun, T. Bell, G. Wood, Wear behaviour of plasma-nitrided martensitic stainless steel, Wear 178 (1994) 131–138. [25] A. Pyzalla, L. Wang, E. Wild, T. Wroblewski, Changes in microstructure, texture and residual stresses on the surface of a rail resulting from friction and wear, Wear 251 (2001) 901–907. [26] W Osterle, H. Rooch, A. Pyzalla, L. Wang, Investigation of white etching layers on rails by optical microscopy, electron microscopy, X-ray and synchrotron X-ray diffraction, Mater. Sci. Eng. A 303 (2001) 150–157.