- Email: [email protected]
Influence of lubricant formulation on rolling contact fatigue of gears – interaction of lubricant additives with fatigue cracks
Author’s Accepted Manuscript INFLUENCE OF LUBRICANT FORMULATION ON ROLLING CONTACT FATIGUE OF GEARS – INTERACTION OF LUBRICANT ADDITIVES WITH FATIGUE CRACKS Benoit L’Hostis, Clotilde Minfray, Marion Frégonèse, Catherine Verdu, Benoit TerOvanessian, Béatrice Vacher, Thierry Le Mogne, Frédéric Jarnias, Alder Da-Costa D’Ambros
PII: DOI: Reference:
www.elsevier.com/locate/wear
S0043-1648(16)30729-3 http://dx.doi.org/10.1016/j.wear.2017.04.025 WEA102154
To appear in: Wear Received date: 7 December 2016 Revised date: 25 April 2017 Accepted date: 26 April 2017 Cite this article as: Benoit L’Hostis, Clotilde Minfray, Marion Frégonèse, Catherine Verdu, Benoit Ter-Ovanessian, Béatrice Vacher, Thierry Le Mogne, Frédéric Jarnias and Alder Da-Costa D’Ambros, INFLUENCE OF LUBRICANT FORMULATION ON ROLLING CONTACT FATIGUE OF GEARS – INTERACTION OF LUBRICANT ADDITIVES WITH FATIGUE CRACKS, Wear, http://dx.doi.org/10.1016/j.wear.2017.04.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
INFLUENCE OF LUBRICANT FORMULATION ON ROLLING CONTACT FATIGUE OF GEARS – INTERACTION OF LUBRICANT ADDITIVES WITH FATIGUE CRACKS Benoit L’Hostisa,b,c, Clotilde Minfraya, Marion Frégonèseb, Catherine Verdub, Benoit Ter-Ovanessianb, Béatrice Vachera, Thierry Le Mognea, Frédéric Jarniasc, Alder DaCosta D’Ambrosc a
LTDS, Ecole Centrale de Lyon, UMR 5513 MATEIS, INSA-Lyon, UMR5510 c TOTAL Marketing Services – Centre de Recherches de Solaize b
Corresponding author: Clotilde Minfray email address: [email protected]
Abstract The influence of lubricant additives on rolling contact fatigue crack propagation and the mechanisms responsible for the resulting micro-scale damage, was studied via experiments conducted on complete transmissions in a test cell. Bench-scale tribological tests and the exposure of steel surfaces to two different formulated lubricants were also carried out. Scanning and Transmission Electronic Microscopy observations, Electron Dispersive Energy and X-ray Photoelectron Spectroscopy analyses indicated that the sulphur present in the extreme pressure (EP) additives has a positive impact on limiting damage propagation. Thanks to TEM observations of cracks, it was demonstrated that a sulphur rich film is formed at the crack tip. This film can act as both a barrier film towards hydrogen permeation within the metal and / or as an inhibitor of oil decomposition. The latter is associated with the nascent surface’s ability to limit hydrogen generation. Without such hydrogen embrittlement, crack propagation is slowed down.
Highlights: - Lubricant additives are found to influence Rolling Contact Fatigue of gears - Elements from lubricant additives are detected within cracks - Sulphur is found to form a film at the crack tip and to have a positive effect on RCF - The sulphide film formed at the crack tip seems to inhibit hydrogen embrittlement
Keywords Rolling contact fatigue, Gears, Steel, Lubricant additives, Electron microscopy, XPS, Hydrogen Embrittlement
1
Introduction Rolling Contact Fatigue (RCF) is a typical failure mode affecting bearings, gears, cams and rails. It refers to the surface damage process that leads to the initiation and propagation of cracks on the surface or in the subsurface loaded by repeated Hertzian contact pressure. RCF can cause detrimental vibrations to mechanical systems and, in extreme cases, breakage of components. The first studies dealing with RCF appear in the 1940’s in response to a demand for better knowledge and understanding of bearing operation [1]. Depending on the components studied (bearing, gears or rails) different terms are sometimes used to describe similar types of RCF damage. Olver et al. were the ones of the first to classify types of fatigue for rail and bearing communities [2]. Unfortunately, even within a single community, there is no common definition of RCF damage. Some define gear spalling and pitting as synonyms [3] while others consider pitting and spalling to be surface and subsurface initiated defects respectively [4]. The term micropitting is even more controversial [5]. From propositions made by Littman and Olver [2,4], Santus distinguishes four RCF mechanisms [6]; (i) case crushing: a type of surface fatigue in surface hardened components; (ii) spalling: a type of subsurface originated RCF; (iii) pitting: a type of RCF initiated at the surface; (iv) micropitting: shallow pitting. The present study uses the latter definitions. Despite many studies, the understanding of mechanisms involved in pitting damage is still incomplete. This is partly due to large number of influencing factors that must be taken into account when studying RCF. Indeed, literature underlines the impact of tribological parameters (loading, contact conditions and lubricant viscosity [7]) together with material parameters (steel composition, thermo-chemical treatment, surface roughness and residual stresses [2,7,8]) and environmental parameters (temperature, humidity and lubricant chemistry [9–11]). Amongst these parameters it is now wellknown that lubrication has a significant influence on RCF and on pitting in particular [1,2,11]. As far as initiation is concerned, after Way observed that the presence of lubricant was required for pitting initiation [1], several studies have suggested that lubricant additives may promote crack initiation by creating corrosion pits on steel surfaces [12,13]. The formation of a tribofilm from Zinc Dialkyl Dithio Phosphate (ZDDP) anti-wear additive can also promote crack initiation by preventing surfaces roughness reduction during running-in [14].
2
Hypotheses concerning the role of the lubricant in mechanical mechanisms of RCF crack propagation were proposed by Bower [15] whereby only the mechanical influence of lubricant pressure on mode I crack propagation is considered: (i) fluid pressurization: because of the high pressure occurring in the contact, the fluid flows into the crack and causes it to open (Figure1a), (ii) fluid entrapment: the closure of the mouth seals fluid into the crack and contact motion generates fluid pressure at the crack tip (Figure 1b). Olver et al. proved by directly observing a micropit in an elastohydrodynamic contact that lubricant penetrates into RCF cracks [16] which allowed the mechanical influence of lubricant to be modeled when crack propagation mixed-modes are involved [17–19]. Moreover, the presence of elements from additives (such as calcium, phosphorus and sulphur) were detected inside the RCF cracks after rolling sliding tests [20]. Interestingly, the distribution of additive elements inside the crack is not homogeneous: calcium and phosphorus are present all along the crack whilst sulphur is mainly found at the crack tip. However the action of lubricant additives that penetrate inside pitting cracks remains unclear with regards to crack propagation. The role of lubricant additives in terms of chemical reactions with the crack surfaces during propagation must be considered when studying the influence of the lubricant on RCF.
Figure 1 – Hypotheses of Bower concerning the mechanical influence of lubricant on crack propagation: a) fluid pressurization, b) fluid entrapment. Furthermore, hydrogen assisted fatigue was shown to accelerate classic RCF mechanisms [21]. Metal embrittlement associated with hydrogen assisted fatigue is the result of hydrogen uptake, permeation and eventual trapping at metal lattice defects which act as hydrogen traps (intermetallics, interphases, grain boundaries, etc.). Bond scission and the formation of lattice defects is the result of the heat created by friction, shear forces and material deformation. The generation of hydrogen in a tribological contact was shown to be the result of the decomposition of the base oil molecules catalyzed by a nascent metal surface generated locally in severe mixed friction contacts [22,23]. The nature of the tribofilm that is formed on nascent surfaces, itself dependent of the nature of the lubricant additives,
3
was demonstrated to directly influence the generation of hydrogen [21–23]. More precisely, the formation of metallic sulfide containing films tends to deactivate the nascent surface [22,23], therefore hindering the catalyzed decomposition of the lubricant since the nascent metal surface is protected by the tribofilm. The tribofilm can also act as a physical barrier towards hydrogen uptake [21]. Yet, whereas the penetration of lubricant additives within the crack was proven [20,24], these assessments concern only external surface tribochemical reactivity, and no attempt has yet been made to extend these conclusions to the films formed at the crack tip and on the faces of the RCF cracks. In this context, the present study focuses on the influence of lubricant additives inside the crack during crack propagation and on the mechanisms responsible for RCF damage at the crack scale. The RCF resistance of passenger car manual transmission gears was chosen for this study. This kind of damage is the primary factor limiting gearbox durability because severe adhesive wear such as scuffing or scoring is, in general, prevented via transmission lubricants and the additives within these lubricants, particularly EP additives. This situation is not expected to change in the near future because the fuel economy driven trend leading to lower lubricant viscosities and higher average torque for gearboxes is expected to continue unchanged. In order to generate used gears with representative wear and mechanical / chemical degradation, endurance tests were performed using complete transmissions mounted in a test cell. Two lubricants were then tested in this transmission test with a severe endurance method in order to investigate the role of sulphur specifically. The two lubricants tested were deliberately formulated with poor to average performance in order to keep the test duration short. After the endurance test, the nature of damage due to RCF on gear teeth with both lubricants was characterized via optical microscopy and Secondary Electron Microscopy (SEM). A systemic Transmission Electron Microscopy (TEM) characterization of films formed within the crack is also carried out. Additional tests were carried out on samples with the same grade of steel and subjected to the same heat treatment as the gears: (i) a period of exposure of metallic surfaces in the two different lubricants followed by Energy Dispersive X-Ray Spectrometry (EDS) analyses (ii) bench scale friction tests followed by X-Ray Photoelectron Spectroscopy (XPS) analyses of wear tracks. These results are used to discuss lubricant reactivity with internal crack surfaces and its potential role in hydrogen embrittlement at the crack tip. No specific experiments on hydrogen embrittlement were carried out to confirm the proposed mechanism.
4
Experimental Two fully formulated SAE J306 75W lubricants (for manual transmissions) were tested on a full-scale transmission stand. The final reducer, pinion and axle ring gear, is permanently loaded whichever gear ratio is selected in the gearbox, so we have focused our damage analysis on this secondary shaft pinion, characterized through macroscopic, microscopic and metallurgical observations. To confirm or not the presence of additive elements inside the cracks, Scanning Electron Microscopy (SEM) together with Energy Dispersive Spectroscopy (EDS) analyses were conducted on pinion flank crosssections. After this first set of characterizations, Focused Ion Beam (FIB) cross-sections containing one representative crack from each pinion were prepared and analyzed using TEM/EDS. Gears The studied gears were made of 23MnCrMo5 carburized steel, after forging, low pressure carburizing and shot-peening (Table 1). The surface microstructure consisted of martensite and retained austenite while bainite formed the core microstructure. From Electron Back Scattered Diffraction (EBSD) analysis, the retained austenite content at the surface was estimated to be about 15% vol. on the tooth flank. Hardness, measured from the surface to a depth of 400 µm, varied from 720 to 600 HV. The average roughness for all samples was Ra = 0.7 ± 0.1 μm. Element wt%
C 0.21
Si 0.26
Mn 1.35
Ni 0.24
Cr 1.28
Mo 0.23
Table 1 – Chemical composition of 23MnCrMo5 grade
Lubricants In order to evaluate the influence of additive packages on pitting, two fully formulated manual gearbox lubricants (A and B) were subjected to a severe operating cycle in a complete transmission on a transmission test stand. Elementary chemical compositions together with the viscosities of formulations A and B are presented in Table 2. The same base stock, composed of polyalphaolefin (PAO) mixed with ester, is used for lubricant A and B. Anti-Wear (AW), detergent (Det) and extremepressure (EP) additives were part of the additive packages. However the chemistries and amounts of these additives in formulation A were different from those in formulation B. Since this study focuses on the influence of lubricant chemistry on crack propagation, the lubricants were formulated to have final
5
viscosities as close as possible to each other. In this way the potential impact of lubricant viscosity can be discounted in any analysis of observed differences in gear fatigue between lubricant A or B.
Additives Package A B
Base stock
S (wt ppm)
Ca (wt ppm)
P (wt ppm)
PAO + ester
3470 7660
680 3051
660 942
η 40°C (cSt) 30.29 30.18
η 100°C (cSt) 6.65 6.68
VI 184 188
Table 2 – Information about lubricants Transmission test stand procedure
A complete transmission was mounted on a test stand (Figure 2) and was used to evaluate the tendency of lubricants A and B to promote or resist RCF on the gears. The gearboxes were tested under high torque, using an electrically driven test rig with two asynchronous induction motors, one working in driving mode and the other providing the load via resistance; the gearbox lubricant temperature was maintained at below 105°C; input rotation speed was set to 3000 rpm. Load and velocities distributions (dimensionless) along the line of action of teeth are presented on Figure 2.c. The test was stopped automatically when one of the following conditions was met (i) vibrations were detected; (2) gear damage was identified; (3) after 200 cycles if no damage was detected. In one test cycle, all gear ratios were tested with relative duration relevant vs. service field, for several hours, as described in Table 3. After testing, the gearbox was unmounted from the test stand and the secondary shaft pinion was removed (one pignon for each lubricant). Endurance life expectancy is defined by car
manufacturer fatigue life proprietary model. The duration target and the number of repeats may vary according the gearbox component considered. In this study we have considered lubricants which failed before the expected target of fifty percent of endurance completed without any damage.
The center was drilled out of the pinion and cuts were then made along each bottom land, separating each gear tooth for individual analysis. The final preparation steps consisted of cutting tooth transversally providing cross-sections and metallographic preparation in order to perform SEM observations of the cracks. The morphologies of cracks along the tooth surface (length and orientation of the cracks) were studied on five teeth for each lubricant. Only representative conclusions will be reported in the results part.
6
Air blowers
b)
a)
Brake
c)
Figure 2 : a) General view of the transmission test stand b) schematic of manual gearbox : secondary shaft pinion evaluated c) pinion/wheel teeth contact : load and velocities distribution along the line of action. The dash line represents the position on the length of action of teeth where the cracks were selected for TEM observations. The stars represent the velocities and load conditions at this selected position.
Gear ratio 1 2 3 4 5
Torque (N.m) 56% torque 94% torque Full torque Full torque Full torque
Unit cycle proportional weight 5% 27% 17% 17% 53%
Table 3 – Mechanical conditions during one cycle. The full test consists of 200 repetitions of this cycle.
7
Complementary tests were performed to help understand additive-crack interactions: Chemical reactivity of lubricants with surfaces In order to evaluate the reactivity of lubricant additives with nascent steel surfaces, freshly polished specimens were exposed to 15 ml of the lubricant at 120°C in a glass beaker. Specimens were made of the same steel grade as the tested gearbox pinions and were subjected to similar low pressure carburization process. No difference in micro-hardness was observed between these specimens and the tested pinion surfaces. The final steps of specimen preparation consisted of silicon carbide polishing and were performed whilst immersed in PAO in order to minimize the formation of iron oxides. This was confirmed by X-ray Photoelectron Spectroscopy (XPS) analysis. After exposure to the lubricants, EDS and XPS analyses were performed. Friction tests Friction tests were carried out using a custom designed tribometer employing a reciprocating ball-onflat configuration. Balls used were made of AISI 52100 steel (hardness of about 800 HV). The same type of specimen with the same preparation as those prepared for the lubricant exposure test was used as the flat. The roughness (Ra) of the plates after the last polishing step (Grit 1200 whilst immersed in PAO) was about 30 nm. Tests were performed at 120°C with lubricant A or B. Two friction tests were performed for each set of conditions. Parameters (frequency, load) and the calculated maximal Hertzian pressures are presented in Table 4. For each test, the friction coefficient was calculated and XPS analysis of the wear scars was performed. Test
Frequency (Hz)
1 2
5 5
Stroke length (mm) 3 3
Load (N) 10 15
Mean PHertz (MPa) 595 683
Max PHertz (MPa) 895 1024
Table 4 – Parameters of friction tests Microstructural characterization Micrographs were made using a Zeiss Supra VP55 scanning electronic microscope with a Field Emission Gun (FEG) using high current mode with a tension from 5 to 15 kV. The FIB was used to prepare TEM foils containing a pre-selected pitting crack. To compare lubricant effect, the cracks were selected on pinion teeth at the same position on the length of action. For each lubricant tested, the studied cracks were so obtained for identical contact conditions. These samples were observed using
8
a JEOL 2010F FEG TEM. SEM and TEM devices were equipped with an Oxford Instrument Xmax Silicon Drift Detector for EDS analysis. An EBSD (Electron Backscattered Diffraction Spectroscopy) system installed on the SEM was also used to investigate the gear teeth microstructure evolution during RCF tests. Film formation after either friction tests or exposure of the steel to a lubricant was investigated by XRay Photoelectron Spectrometry (XPS) with an ULVAC-PHI VersaProbe II spectrometer using a monochromatized AlKα X-ray source (1486.6 eV). The binding energy reference was taken for C1s adventitious contribution at 284.8 eV.
9
Results Damage was characterized for both lubricants, A and B, for the most stressed part of the gear, i.e. the pinion, after transmission tests. A macroscopic evaluation is first carried out prior to microscopic and microstructural observations. Automatic interruption of the transmission test (complete transmission on a test stand) occurred after 18 cycles for lubricant A and 66 cycles for lubricant B (Table 5), which corresponds to 9% and 33% of the maximum expected test duration (200 cycles) respectively. Damage development occurs therefore earlier and/or faster for lubricant A. In both cases, the macroscopic observation of pinion tooth surfaces reveals the same pattern of damage: RCF damage is observed on the whole contact surface and the pitted area is located mainly below the pitch circle. Large pits are also observed in the tooth flank on some teeth (Figure 3). These large pits propagate to the tip of the flank, towards the pitch circle, and sometimes lead to tooth breaking. Lubricant A
Testing advance 18 cycles
B
66 cycles
Reason to end the test Automatically stopped due to vibrations
Observations - Pitting on seven teeth - 2 chipped teeth - Pitting on eight teeth - 4 chipped teeth
Table 5 – Results of full-transmission stand
a)
b) Figure 3 – Macroscopic observation of pinions after full scale transmission tests: a) Pinion tested with lubricant A. b) Pinion tested with lubricant B
Further SEM observations of the teeth flanks and teeth cross-sections prove that all micro-cracks initiate from the surface; 1 µm diameter MnS inclusions are detected by EDS whereas no subsurface initiated cracks are observed. As expected, the statistical analysis of micro-crack orientation shows that the mean angle between the propagation direction and surface is higher at the dedendum bottom and addendum top and decreases with an inversion of crack orientation direction at pitch line. Classic micro-crack orientations are reported in Figure 4. Further, no significant difference (>5°) in crack
10
orientation is observed between pinions tested with lubricant A or lubricant B (Figure 5) for a given position on the tooth (the chosen position is shown in figure 2.c : dashed line). SEM backscattered electron imaging of the tooth surface cross section exhibits grain refinement between the micro-cracks and the surface (Figure 6). EBSD mapping of these areas confirms and clarifies this observation (Figure 7). Indeed, areas away from the micro-cracks present the lenticular (needle-shaped) microstructure characteristic of the martensitic crystalline structure together with residual austenite whereas the areas between the micro-cracks and the external surface exhibit a nano-scale microstructure free of residual austenite. These refined microstructure areas are only observed conjointly with micro-cracks.
a)
b)
Figure 4 – SEM images of pinion teeth after full scale test: a) image of tooth flank b) tooth cross-section
11
Figure 5 – SEM images of micro-cracks obtained at the same position on the teeth after full scale test for lubricant A and B.
2 µm
Figure 6 – Backscattered electron mode (SEM) image showing the refined microstructure between cracks and surface.
12
a)
b)
Ferrite Residual austenite Carbides
c)
10 µm Figure 7 – a-b) SEM and EBSD images showing the refined microstructure between cracks and the surface and the disappearance of retained austenite c) EBSD image of initial steel microstructure.
13
Scanning electron micrographs reveal that the crack faces are covered with a film (Figure 8). In order to permit higher resolution analysis by TEM, FIB is used to cut very thin slices of the gear teeth which contain micro-cracks. To allow a comparison between lubricants A and B, the micro-cracks for analysis were selected such that comparisons could be made between areas that had been subjected stress conditions as similar as possible to each other (same rolling/sliding ratio). For this reason cracks selected for further observation are systematically selected and compared at the same position on the pinion tooth flank (i.e. same distance from pitch line and both dedendum or both addendum) with morphologies as similar as possible (direction of and length of propagation).
Figure 8 – SEM micrograph of tooth flank cross-section. A film is present along the micro-crack faces.
14
TEM observations confirm the presence of a solid film along internal crack faces for both lubricants (Figures 9 and 10). This film is composed of amorphous areas and crystallites. The thickness of the observed films fluctuated from few nanometers (20 nm) close to the crack tip and up to 200 nm at the crack mouth; differences in film thickness were observed between lubricants A and B. Whereas the progression in film thickness, from crack to tip, is relatively uniform with lubricant A, a more variable progression is observed along the crack length with lubricant B (Figure 10). It must also be noticed that higher film thickness is observed when the crack is oriented parallel to the surface. Further TEM/EDS analysis shows a clear distinction between the composition of films induced by lubricants A and B. In both cases oxygen, phosphorus and calcium are found in the film all along the crack (α and β positions in Figures 9 and 10), however sulphur is not detected in the case of lubricant A, neither on the crack faces (α and β positions in Figure 9) nor at the crack tip (γ position in Figure 9). On the other hand, sulphur is detected at the crack tip for lubricant B (γ’ position in Figure 10) and all along the crack near the metal / film interface (γ position in Figure 10). The EDS spectra of film obtained on crack faces with lubricant B (Figure 10) demonstrates that the relative concentrations of oxygen, phosphorus, sulfur and calcium vary with film depth, sulphur being predominant close to the metal / film interface whilst oxygen, phosphorus and calcium predominate near the film surface.
15
Intensity (arb. units) Intensity (arb. units)
Figure 9 – TEM images and EDS analyses of selected crack for lubricant A
Figure 10 – TEM images and EDS analyses of selected crack for lubricant B
16
Discussion The SEM observations of the pinions after the transmission tests have enabled us to set up RCF characteristics of the case studied here. The morphologies of RCF cracks are in agreement with literature [17] resulting in a decrease in the angle between the crack direction and the external surface from about 45° at the tooth filet to about 10° close to the pitch diameter for both lubricants. Propagation direction can be linked to the slide-to-roll ratio, decreasing from the tooth filet to the pitch diameter. Therefore all the samples, regardless of the tested lubricant, encountered the same mechanical contact conditions. However the duration of the protection from damage of the pinions by lubricant A is 70% lower than that of lubricant B. Consequently, we can conclude that the nature of the additives plays a role in the initiation and / or the propagation of RCF damage. Moreover, no subsurface crack initiation is observed; Cracks seems to initiate from the external surface although crack initiation from inclusions can’t be completely excluded. According to the different types of RCF damage defined by Santus [6], the types of damage encountered here is pitting and micropitting and the influence of a lubricant on crack initiation and propagation must be considered. Full-scale transmission tests show a clear influence of lubricant formulation on gear resistance to RCF. The pitting that leads to the breakage of the pinion teeth in the case of the present study is clearly delayed when lubricant B is used. As stated previously, different hypotheses may be introduced regarding the lubricant effect on RCF; the lubricant can accelerate or delay crack initiation: (i) through a mechanical-chemical action if additives initiate corrosion pits [12,13] or (ii) through a tribochemical reaction forming a tribofilm which modifies the surface roughness during run-in [25,26]. In addition, the tribofilm formed on the gear tooth surface can affect the coefficient of friction between sliding parts which can modify the stress field seen by the surface. This study has not been designed to allow us to discrimate against these three phenomena. The lubricant can also exert an influence on the propagation of cracks through different effects: (i) the mechanical effect of lubricant: the pressure transmitted by the lubricant which has penetrated (or which is trapped inside the crack) increases the crack propagation rate; it can also modify lubrication conditions between crack faces [27]; (ii) the effect of the tribofilm formed on the gear tooth surface at the crack mouth can prevent the penetration of the lubricant inside the crack [24], (iii) the effect of
17
some lubricant additives which react with nascent steel inside the crack and form film inside the crack which can affect positively or negatively crack propagation [20]. Since the studied lubricants have similar viscosities and, in terms of the base oil, have similar formulations, the mechanical effect of lubricant on propagation shall not be considered here. Moreover crack mouth obstruction by the tribofilm was not observed during this study. Further discussion is therefore focused on the effects of lubricant additive components. EDS/SEM and EDS/TEM analyses confirm the observations of Méheux et al. [20]. Indeed calcium, phosphorus and sulphur, deriving from anti-wear and extreme-pressure additives, as well as oxygen can be detected inside the cracks. A difference between lubricant A and lubricant B is nevertheless observed. No sulphur is detected for lubricant A whereas in the case of lubricant B sulphur is detected at the crack tip and on the crack faces close to the metal. The results of observations and chemical analyses are summarized in Figure 11, they once again confirm the penetration of lubricant additives inside the crack. From this crack characterization it appears that the lubricant reacts in two different ways inside the crack: (i) at the crack tip where only a thin film is observed and where, in the case of lubricant A, only oxygen is detected and the case of lubricant B only sulphur is detected; (ii) along the crack faces where the lubricants form tribofilms of heterogeneous thickness, composed of calcium, phosphorus and oxygen. In the case of lubricant B, sulphur is also found close to the metal surface.
Figure 11 – Overview of cracks chemistries from TEM observations and EDX analyses
18
In order to better understand the formation of the observed films into cracks, complementary experiments were set up. To simulate the reaction of additives with nascent steel surfaces at the crack tip during the propagation freshly polished steel samples were exposed to the lubricants. Friction tests were also carried out to simulate the friction between the crack faces. After exposure to lubricant B nascent steel surfaces exhibit a dark blue color for which the intensity depends on exposure time and temperature. No visible surface alteration is observed after exposure to lubricant A. SEM observations of exposed samples confirm the presence of a film on the steel surface in the case of lubricant B (Figure 12). Film thickness varies from 200 nm to 1 µm depending on exposure time and temperature. EDS analysis shows that the formed film is composed of iron and sulphur. Complementary XPS analysis shows that the energy values of the main XPS S2p3/2 peaks are 161.7 eV and 162.5 eV (Figure 13), which corresponds to FeS and FeS2 respectively [28,29]. Steel surfaces exposed to lubricant A exhibit a film that contains no sulphur. The growth of an iron sulfide containing film is therefore only obtained when the steel samples are exposed to lubricant B which is in accordance with observations made during the characterization of the crack. This tends to prove that the presence of sulphur (lubricant B) observed at the crack tip is due to the lubricant additive’s reaction to freshly formed steel surfaces.
19
a)
b)
Figure 12 – a) SEM image of sample cross-section after exposure to lubricant B (12 h – 120°C) b) Spectrum of EDS analysis of sample surface
FeS
FeS2 Adsorbed additive
Figure 13 – XPS analysis of steel surface after exposure to lubricant B during 12h, S2p peak
20
XPS survey scans and XPS depth profiles of tribofilms from the complementary friction tests show the presence of calcium, oxygen and phosphorus but no sulphur regardless of the lubricant tested (Figures 14 and 15). Sulphur is therefore not detected at the surface of either tribofilm (Figure 15) as was the case for both lubricants films formed on the crack faces (α and β positions in Figures 9 and 10). Yet, for the tribofilm formed during friction tests with lubricant A, sulphur is not detected at the iron / tribofilm interface, nor in the whole depth of the tribofilm. It was however detected for the film formed on crack faces with lubricant B (γ’ position in Figure 10). It may consequently be suggested that iron sulfide forms at the crack tip when a nascent steel surface is in contact with a sulphur reach lubricant (i.e. lubricant B in our study); the crack then propagates due to RCF (probably via mode II) and a Ca and P rich tribofilm is formed on crack faces due to friction. The friction conditions in the friction tests and the ones encountered in real cracks are certainly different, it is however reasonable to assume that the film observed on the crack faces is a tribofilm i.e. obtained by friction, as already demonstrated by Aldana et al [24]. In this later study the observation of the transformation of WS2 nanoparticles (used as an additive in the lubricant) into a layered and planar structure allowed the authors to describe the film formed within the crack as a tribofilm.
21
Figure 14 – XPS survey scans of wear tracks obtained with lubricant A and B
22
Lubricant B
Lubricant A
Fe
Fe Tribofilm
Substrate Atomic concentration (at%)
Atomic concentration (at%)
Tribofilm
O
Ca P
Etching time (min)
Substrate
O
Ca P Etching time (min)
Figure 15 – XPS depth profiles of tribofilms obtained after friction tests with lubricant A and B. In the present case, the occurrence of a tribofilm within the crack is also in agreement with the observation of grain refinement between the external surface and the cracks highlighted by EBSD analysis (Figure 7). Both phenomena actually result from severe friction conditions between crack faces. Frequent displacements of this upper part (between the external surface and the cracks) causing by sliding can be responsible for high strain hardening leading to severe refinement of the microstructure [17,19]. Based on these observations, some mechanisms for crack propagation can be discussed. The modification of the friction coefficient between crack faces due to the formation of the tribofilm may firstly be considered. Several studies have considered the role of the friction coefficient between the crack faces when modeling RCF crack propagation [17,19]. More generally speaking, this is a parameter known to have an influence on mode II or III fatigue driven crack propagation [19]. In actual fact, by reducing the coefficient of friction, crack face sliding is less energy consuming and a higher propagation rate is expected. Numerous different models and hypotheses are used for RCF crack propagation and the conclusions of different works differ regarding the importance of the crack face friction coefficient. However differences in the propagation rate are generally observed in models for a difference of friction coefficients of at least 0.2 [19]. If the assumption is made that tribofilms formed during friction tests and inside cracks are similar then the difference in friction coefficient between
23
lubricant A and B (0.10 versus 0.12) appears to be too weak to have a significant influence here (Figure 16). Moreover, damage observed on the pinions is greater when lubricant A is employed (i.e. the crack propagation rate is higher in the case of lubricant A) while its friction coefficient is the highest. This contradicts the better resistance to RCF observed with lubricant B. The modification of the friction coefficient between crack faces due to the formation of the tribofilm is therefore not a first order mechanism for explaining RCF crack propagation.
Figure 16 – Coefficient of friction obtained from reciprocating ball-on-flat tests Another mechanism by which the lubricant could impact the crack propagation is proposed by Aldana et al. [24]. They noticed that tribofilms formed inside cracks could prevent the penetration of lubricant inside the crack and thus avoid fluid pressurization or entrapment. In the present study, no evidence of the presence of a film closing the crack was observed. On the other hand, the influence of hydrogen on RCF has been widely cited in literature. Studies have proved that the presence of diffusible or trapped hydrogen in concentration above 1 or 2 wt. ppm reduces resistance to RCF [30–32]. However, before considering hydrogen embrittlement mechanisms, the origin of the hydrogen must be discussed. Indeed the intrinsic hydrogen content of steel used in gear manufacturing (less than 0.3 wt ppm) cannot explain such an embrittlement mechanism. Water contamination of the lubricant could be considered as an explanation; however the amount of contamination needed to lead to 1 wt ppm hydrogen concentration cannot be reached with
24
our testing conditions. Thus hydrogen generation from lubricant base oil decomposition is a much more promising explanation. It has been demonstrated experimentally that lubricant components can chemically decompose at nascent steel surfaces allowing local hydrogen chemisorption and then diffusion into the bulk steel [22,23,30,33–35]. This has been demonstrated for mineral and poly-alphaolefin (PAO) base oils and PAO is the key base oil used in this study. The presence of ester in the base oil mixture has been demonstrated to increase wear by fatigue which is believed to be due to a relative increase in the liberation of hydrogen compared to other base oil types [34]. The lubricants in this study contained ester base oils along with the PAO base oils and our findings are coherent with those noted above. As far as the influence of lubricant additives is concerned, it is known that corrosion inhibitors and AW/EP additives can deactivate the catalytic action of nascent steel surfaces by rapidly forming passivating films, such as antiwear phosphate and sulfide films [21,22]. In the present case the iron sulfide film formed at the crack tip when lubricant B is used could prevent or decrease the adsorption of hydrocarbon compounds leading to a reduced release of hydrogen into the steel. On the contrary, no sulfide film is observed for lubricant A. The moment when the iron oxide film was formed is therefore questionable: during the full-scale transmission test or after FIB sample preparation? In the latter case, no barrier exists and base oil hydrocarbon compounds adsorb and decompose on native iron surfaces which increase the amount of hydrogen at the crack tip. Hydrogen penetration consequently occurs and leads to an increase in the crack propagation rate by hydrogen embrittlement. If oxide formation occurs during crack propagation then a barrier effect has to be considered as demonstrated by Niste et al. [21] who show that hydrogen permeation is better prevented by films containing than iron oxide films. Finally, the resulting hydrogen assisted RCF could be described by classical mechanisms considered for hydrogen embrittlement in high strength steels [36]: (i) hydrogen enhanced decohesion (HEDE): local hydrogen accumulation in hydrostatic tensile regions at crack tips weakens the cohesive bonds between metal atoms and increases the crack propagation rate; (ii) hydrogen enhanced localized plasticity (HELP): local hydrogen favors dislocation mobility which results in localized softening, therefore enhancing localized plasticity.
25
CONCLUSION In this study, full scale transmission tests together with detailed sample characterizations by TEM and XPS have been used to study the role of lubricant additives in RCF crack propagation mechanisms. Damage was identified as surface initiated pitting and micropitting. The mechanical influence of the lubricant on RCF was excluded and the discussion focused on the chemical reactivity of the addditives. Indeed EDS analyses of crack cross-sections demonstrate the presence of elements from lubricant additives on crack faces and down to crack tips, confirming the penetration of the lubricant within the crack. Sulphur was found to play a major role in the formation of films at the crack tip. Two types of metal/additive interactions within the crack were identified: (i) at the crack tip: depending on the lubricant formulation additives may react with nascent steel surfaces leading to iron sulfide film formation; (ii) on crack faces: crack face friction leads to tribofilm formation. The modification of the friction coefficient on crack faces and the obstruction at the crack by formation of the tribofilm were considered to be second order processes compared to hydrogen embrittlement effects. Base oil degradation has been demonstrated in the literature to be at the origin of hydrogen generation. Sulphur contained in the lubricant additives leads to the formation of iron sulfide at the crack tip that can act as both (i) an inhibitor of lubricant decomposition at the nascent surface since access to the surface is physically blocked thereby reducing the catalytic degradation of the lubricant and therefore limiting hydrogen generation and/or, (ii) once hydrogen is formed, a barrier film to hydrogen permeation. The rate of crack propagation is therefore reduced.
Acknowledgements The authors would like to thank ANRT for support.
26
BIBLIOGRAPHY [1] [2] [3] [4] [5] [6]
[7] [8]
[9] [10]
[11]
[12] [13] [14]
[15] [16] [17] [18] [19] [20]
[21] [22]
[23]
[24]
[25]
S. Way, Pitting due to rolling contact, ASME Trans. J. Appl Mech. (1935) A49–A58. A.V. Olver, The mechanism of rolling contact fatigue: An update, Proc. Inst. Mech. Eng. Part J J. Eng. Tribol. 219 (2005) 313–330. doi:10.1243/135065005X9808. Y. Ding, N.F. Rieger, Spalling formation mechanism for gears, Wear. 254 (2003) 1307–1317. doi:10.1016/S0043-1648(03)00126-1. W.E. Littmann, The mechanism of contact fatigue, in: Interdiscip. Approach Lubr. Conc. Contacts, P. M. Ku, Troy, NY, 1969: pp. 309–378. T.E. Tallian, Y.P. Chiu, A. Van, Prediction of traction and microgeometry effects on rolling contact fatigue life., TRANS ASME J LUBR TECHNOL. 100 (1978) 156–166. C. Santus, M. Beghini, I. Bartilotta, M. Facchini, Surface and subsurface rolling contact fatigue characteristic depths and proposal of stress indexes, Int. J. Fatigue. 45 (2012) 71–81. doi:10.1016/j.ijfatigue.2012.06.012. D. Nélias, M.L. Dumont, F. Champiot, A. Vincent, L. Flamand, Role of inclusions, surface roughness and operating conditions on rolling contact fatigue, J. Tribol. 121 (1999) 240–250. A.C. Batista, A.M. Dias, J.L. Lebrun, F. Le, G. Inglebert, Contact fatigue of automotive gears: Evolution and effects of residual stresses introduced by surface treatments, Fatigue Fract. Eng. Mater. Struct. 23 (2000) 217–228. doi:10.1046/j.1460-2695.2000.00268.x. B.-R. Höhn, K. Michaelis, Influence of oil temperature on gear failures, Tribol. Int. 37 (2004) 103– 109. doi:10.1016/S0301-679X(03)00047-1. W. Tuszynski, R. Michalczewski, W. Piekoszewski, M. Szczerek, Effect of ageing automotive gear oils on scuffing and pitting, Tribol. Int. 41 (2008) 875–888. doi:10.1016/j.triboint.2007.12.010. W.E. Littmann, B.W. Kelley, W.J. Anderson, R.S. Fein, E.E. Klaus, L.B. Sibley, W.O. Winer, Chemical effects of lubrication in contact fatigue - 3. load-life exponent, life scatter, and overall analysis., J Lubr Technol Trans ASME. 98 Ser F (1976) 308–318. A.B. Cardis, Gear oil micropitting evaluation, Gear Technol. 17 (2000) 30–35. A.A. Torrance, J.E. Morgan, G.T.Y. Wan, An additive’s influence on the pitting and wear of ball bearing steel, Wear. 192 (1996) 66–73. E. Lainé, A.V. Olver, M.F. Lekstrom, B.A. Shollock, T.A. Beveridge, D.Y. Hua, The effect of a friction modifier additive on micropitting, Tribol. Trans. 52 (2009) 526–533. doi:10.1080/10402000902745507. A.F. Bower, Influence of crack face friction and trapped fluid on surface initiated rolling contact fatigue cracks, in: 1988. A.V. Olver, L.K. Tiew, S. Medina, J.W. Choo, Direct observations of a micropit in an elastohydrodynamic contact, Wear. 256 (2004) 168–175. doi:10.1016/S0043-1648(03)00374-0. M. Kaneta, Y. Murakami, Effects of oil hydraulic pressure on surface crack growth in rolling/sliding contact, Tribol. Int. 20 (1987) 210–217. doi:10.1016/0301-679X(87)90076-4. Y. Murakami, S. Hamada, A new method for the measurement of mode II fatigue threshold stress intensity factor range ΔKτth, Fatigue Fract. Eng. Mater. Struct. 20 (1997) 863–870. J.W. Ringsberg, A. Bergkvist, On propagation of short rolling contact fatigue cracks, Fatigue Fract. Eng. Mater. Struct. 26 (2003) 969–983. doi:10.1046/j.1460-2695.2003.00657.x. M. Meheux, C. Minfray, F. Ville, T.L. Mogne, A.A. Lubrecht, J.M. Martin, H.P. Lieurade, G. Thoquenne, Effect of lubricant additives in rolling contact fatigue, Proc. Inst. Mech. Eng. Part J J. Eng. Tribol. 224 (2010) 947–955. doi:10.1243/13506501JET719. V.B. Niste, H. Tanaka, M. Ratoi, J. Sugimura, WS2 nanoadditized lubricant for applications affected by hydrogen embrittlement, RSC Adv. 5 (2015) 40678–40687. doi:10.1039/c5ra03127c. R. Lu, S. Mori, T. Kubo, H. Nanao, Effect of sulfur-containing additive on the decomposition of multialkylated cyclopentane oil on the nascent steel surface, Wear. 267 (2009) 1430–1435. doi:10.1016/j.wear.2009.03.028. R. Lu, H. Nanao, K. Kobayashi, T. Kubo, S. Mori, Effect of lubricant additives on tribochemical decomposition of hydrocarbon oil on nascent steel surfaces, J. Jpn. Pet. Inst. 53 (2010) 55–60. doi:10.1627/jpi.53.55. P.U. Aldana, F. Dassenoy, B. Vacher, T. Le Mogne, B. Thiebaut, A. Bouffet, Antispalling Effect of WS2 Nanoparticles on the Lubrication of Automotive Gearboxes, Tribol. Trans. 59 (2016) 178–188. doi:10.1080/10402004.2015.1061080. L.J. Taylor, H.A. Spikes, Friction-enhancing properties of ZDDP antiwear additive: Part I Friction and morphology of ZDDP reaction films, Tribol. Trans. 46 (2003) 303–309.
27
[26] L.J. Taylor, H.A. Spikes, Friction-enhancing properties of ZDDP antiwear additive: Part II Influence of ZDDP reaction films on EHD lubrication, Tribol. Trans. 46 (2003) 310–314. [27] A.F. Bower, Influence of crack face friction and trapped fluid on surface initiated rolling contact fatigue cracks, J. Tribol. 110 (1988) 704–711. [28] C. Minfray, G. Fromentin, A. Bierla, J.-M. Martin, T. Le Mogne, The effect of an organic pentasulfide EP additive in turning and milling operations, Wear. (n.d.). doi:10.1016/j.wear.2014.05.009. [29] M. Descostes, F. Mercier, N. Thromat, C. Beaucaire, M. Gautier-Soyer, Use of XPS in the determination of chemical environment and oxidation state of iron and sulfur samples: Constitution of a data basis in binding energies for Fe and S reference compounds and applications to the evidence of surface species of an oxidized pyrite in a carbonate medium, Appl. Surf. Sci. 165 (2000) 288–302. doi:10.1016/S0169-4332(00)00443-8. [30] N. Kino, K. Otani, The influence of hydrogen on rolling contact fatigue life and its improvement, JSAE Rev. 24 (2003) 289–294. doi:10.1016/S0389-4304(03)00035-3. [31] H. Uyama, The Mechanism of White Structure Flaking In Rolling Bearings, (2011). http://www.nrel.gov/wind/pdfs/day2_sessioniv_1_nsk_uyama.pdf. [32] A. Ruellan, F. Ville, X. Kleber, A. Arnaudon, D. Girodin, Understanding white etching cracks in rolling element bearings: The effect of hydrogen charging on the formation mechanisms, Proc. Inst. Mech. Eng. Part J J. Eng. Tribol. 228 (2014) 1252–1265. doi:10.1177/1350650114522452. [33] M. Kohara, T. Kawamura, M. Egami, Study on Mechanism of Hydrogen Generation from Lubricants, Tribol. Trans. 49 (2006) 53–60. doi:10.1080/05698190500486324. [34] P. Nevasmaa, A. Laukkanen, Assessment of hydrogen cracking risk in multipass weld metal of 2.25Cr-1Mo-0.25V-TiB (T24) boiler steel, Weld. World. 49 (2005) 45–58. [35] R. Lu, H. Nanao, K. Takiwatari, S. Mori, Y. Fukushima, Y. Murakami, S. Ikejima, T. Konno, The Effect of the Chemical Structures of Synthetic Hydrocarbon Oils on Their Tribochemical Decomposition, Tribol. Lett. 60 (2015). doi:10.1007/s11249-015-0606-2. [36] W. Gerberich, Modeling hydrogen induced damage mechanisms in metals, in: Gaseous Hydrog. Embrittlement Mater. Energy Technol. Mech. Model. Future Dev., 2012: pp. 209–246.
28
PII: DOI: Reference:
www.elsevier.com/locate/wear
S0043-1648(16)30729-3 http://dx.doi.org/10.1016/j.wear.2017.04.025 WEA102154
To appear in: Wear Received date: 7 December 2016 Revised date: 25 April 2017 Accepted date: 26 April 2017 Cite this article as: Benoit L’Hostis, Clotilde Minfray, Marion Frégonèse, Catherine Verdu, Benoit Ter-Ovanessian, Béatrice Vacher, Thierry Le Mogne, Frédéric Jarnias and Alder Da-Costa D’Ambros, INFLUENCE OF LUBRICANT FORMULATION ON ROLLING CONTACT FATIGUE OF GEARS – INTERACTION OF LUBRICANT ADDITIVES WITH FATIGUE CRACKS, Wear, http://dx.doi.org/10.1016/j.wear.2017.04.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
INFLUENCE OF LUBRICANT FORMULATION ON ROLLING CONTACT FATIGUE OF GEARS – INTERACTION OF LUBRICANT ADDITIVES WITH FATIGUE CRACKS Benoit L’Hostisa,b,c, Clotilde Minfraya, Marion Frégonèseb, Catherine Verdub, Benoit Ter-Ovanessianb, Béatrice Vachera, Thierry Le Mognea, Frédéric Jarniasc, Alder DaCosta D’Ambrosc a
LTDS, Ecole Centrale de Lyon, UMR 5513 MATEIS, INSA-Lyon, UMR5510 c TOTAL Marketing Services – Centre de Recherches de Solaize b
Corresponding author: Clotilde Minfray email address: [email protected]
Abstract The influence of lubricant additives on rolling contact fatigue crack propagation and the mechanisms responsible for the resulting micro-scale damage, was studied via experiments conducted on complete transmissions in a test cell. Bench-scale tribological tests and the exposure of steel surfaces to two different formulated lubricants were also carried out. Scanning and Transmission Electronic Microscopy observations, Electron Dispersive Energy and X-ray Photoelectron Spectroscopy analyses indicated that the sulphur present in the extreme pressure (EP) additives has a positive impact on limiting damage propagation. Thanks to TEM observations of cracks, it was demonstrated that a sulphur rich film is formed at the crack tip. This film can act as both a barrier film towards hydrogen permeation within the metal and / or as an inhibitor of oil decomposition. The latter is associated with the nascent surface’s ability to limit hydrogen generation. Without such hydrogen embrittlement, crack propagation is slowed down.
Highlights: - Lubricant additives are found to influence Rolling Contact Fatigue of gears - Elements from lubricant additives are detected within cracks - Sulphur is found to form a film at the crack tip and to have a positive effect on RCF - The sulphide film formed at the crack tip seems to inhibit hydrogen embrittlement
Keywords Rolling contact fatigue, Gears, Steel, Lubricant additives, Electron microscopy, XPS, Hydrogen Embrittlement
1
Introduction Rolling Contact Fatigue (RCF) is a typical failure mode affecting bearings, gears, cams and rails. It refers to the surface damage process that leads to the initiation and propagation of cracks on the surface or in the subsurface loaded by repeated Hertzian contact pressure. RCF can cause detrimental vibrations to mechanical systems and, in extreme cases, breakage of components. The first studies dealing with RCF appear in the 1940’s in response to a demand for better knowledge and understanding of bearing operation [1]. Depending on the components studied (bearing, gears or rails) different terms are sometimes used to describe similar types of RCF damage. Olver et al. were the ones of the first to classify types of fatigue for rail and bearing communities [2]. Unfortunately, even within a single community, there is no common definition of RCF damage. Some define gear spalling and pitting as synonyms [3] while others consider pitting and spalling to be surface and subsurface initiated defects respectively [4]. The term micropitting is even more controversial [5]. From propositions made by Littman and Olver [2,4], Santus distinguishes four RCF mechanisms [6]; (i) case crushing: a type of surface fatigue in surface hardened components; (ii) spalling: a type of subsurface originated RCF; (iii) pitting: a type of RCF initiated at the surface; (iv) micropitting: shallow pitting. The present study uses the latter definitions. Despite many studies, the understanding of mechanisms involved in pitting damage is still incomplete. This is partly due to large number of influencing factors that must be taken into account when studying RCF. Indeed, literature underlines the impact of tribological parameters (loading, contact conditions and lubricant viscosity [7]) together with material parameters (steel composition, thermo-chemical treatment, surface roughness and residual stresses [2,7,8]) and environmental parameters (temperature, humidity and lubricant chemistry [9–11]). Amongst these parameters it is now wellknown that lubrication has a significant influence on RCF and on pitting in particular [1,2,11]. As far as initiation is concerned, after Way observed that the presence of lubricant was required for pitting initiation [1], several studies have suggested that lubricant additives may promote crack initiation by creating corrosion pits on steel surfaces [12,13]. The formation of a tribofilm from Zinc Dialkyl Dithio Phosphate (ZDDP) anti-wear additive can also promote crack initiation by preventing surfaces roughness reduction during running-in [14].
2
Hypotheses concerning the role of the lubricant in mechanical mechanisms of RCF crack propagation were proposed by Bower [15] whereby only the mechanical influence of lubricant pressure on mode I crack propagation is considered: (i) fluid pressurization: because of the high pressure occurring in the contact, the fluid flows into the crack and causes it to open (Figure1a), (ii) fluid entrapment: the closure of the mouth seals fluid into the crack and contact motion generates fluid pressure at the crack tip (Figure 1b). Olver et al. proved by directly observing a micropit in an elastohydrodynamic contact that lubricant penetrates into RCF cracks [16] which allowed the mechanical influence of lubricant to be modeled when crack propagation mixed-modes are involved [17–19]. Moreover, the presence of elements from additives (such as calcium, phosphorus and sulphur) were detected inside the RCF cracks after rolling sliding tests [20]. Interestingly, the distribution of additive elements inside the crack is not homogeneous: calcium and phosphorus are present all along the crack whilst sulphur is mainly found at the crack tip. However the action of lubricant additives that penetrate inside pitting cracks remains unclear with regards to crack propagation. The role of lubricant additives in terms of chemical reactions with the crack surfaces during propagation must be considered when studying the influence of the lubricant on RCF.
Figure 1 – Hypotheses of Bower concerning the mechanical influence of lubricant on crack propagation: a) fluid pressurization, b) fluid entrapment. Furthermore, hydrogen assisted fatigue was shown to accelerate classic RCF mechanisms [21]. Metal embrittlement associated with hydrogen assisted fatigue is the result of hydrogen uptake, permeation and eventual trapping at metal lattice defects which act as hydrogen traps (intermetallics, interphases, grain boundaries, etc.). Bond scission and the formation of lattice defects is the result of the heat created by friction, shear forces and material deformation. The generation of hydrogen in a tribological contact was shown to be the result of the decomposition of the base oil molecules catalyzed by a nascent metal surface generated locally in severe mixed friction contacts [22,23]. The nature of the tribofilm that is formed on nascent surfaces, itself dependent of the nature of the lubricant additives,
3
was demonstrated to directly influence the generation of hydrogen [21–23]. More precisely, the formation of metallic sulfide containing films tends to deactivate the nascent surface [22,23], therefore hindering the catalyzed decomposition of the lubricant since the nascent metal surface is protected by the tribofilm. The tribofilm can also act as a physical barrier towards hydrogen uptake [21]. Yet, whereas the penetration of lubricant additives within the crack was proven [20,24], these assessments concern only external surface tribochemical reactivity, and no attempt has yet been made to extend these conclusions to the films formed at the crack tip and on the faces of the RCF cracks. In this context, the present study focuses on the influence of lubricant additives inside the crack during crack propagation and on the mechanisms responsible for RCF damage at the crack scale. The RCF resistance of passenger car manual transmission gears was chosen for this study. This kind of damage is the primary factor limiting gearbox durability because severe adhesive wear such as scuffing or scoring is, in general, prevented via transmission lubricants and the additives within these lubricants, particularly EP additives. This situation is not expected to change in the near future because the fuel economy driven trend leading to lower lubricant viscosities and higher average torque for gearboxes is expected to continue unchanged. In order to generate used gears with representative wear and mechanical / chemical degradation, endurance tests were performed using complete transmissions mounted in a test cell. Two lubricants were then tested in this transmission test with a severe endurance method in order to investigate the role of sulphur specifically. The two lubricants tested were deliberately formulated with poor to average performance in order to keep the test duration short. After the endurance test, the nature of damage due to RCF on gear teeth with both lubricants was characterized via optical microscopy and Secondary Electron Microscopy (SEM). A systemic Transmission Electron Microscopy (TEM) characterization of films formed within the crack is also carried out. Additional tests were carried out on samples with the same grade of steel and subjected to the same heat treatment as the gears: (i) a period of exposure of metallic surfaces in the two different lubricants followed by Energy Dispersive X-Ray Spectrometry (EDS) analyses (ii) bench scale friction tests followed by X-Ray Photoelectron Spectroscopy (XPS) analyses of wear tracks. These results are used to discuss lubricant reactivity with internal crack surfaces and its potential role in hydrogen embrittlement at the crack tip. No specific experiments on hydrogen embrittlement were carried out to confirm the proposed mechanism.
4
Experimental Two fully formulated SAE J306 75W lubricants (for manual transmissions) were tested on a full-scale transmission stand. The final reducer, pinion and axle ring gear, is permanently loaded whichever gear ratio is selected in the gearbox, so we have focused our damage analysis on this secondary shaft pinion, characterized through macroscopic, microscopic and metallurgical observations. To confirm or not the presence of additive elements inside the cracks, Scanning Electron Microscopy (SEM) together with Energy Dispersive Spectroscopy (EDS) analyses were conducted on pinion flank crosssections. After this first set of characterizations, Focused Ion Beam (FIB) cross-sections containing one representative crack from each pinion were prepared and analyzed using TEM/EDS. Gears The studied gears were made of 23MnCrMo5 carburized steel, after forging, low pressure carburizing and shot-peening (Table 1). The surface microstructure consisted of martensite and retained austenite while bainite formed the core microstructure. From Electron Back Scattered Diffraction (EBSD) analysis, the retained austenite content at the surface was estimated to be about 15% vol. on the tooth flank. Hardness, measured from the surface to a depth of 400 µm, varied from 720 to 600 HV. The average roughness for all samples was Ra = 0.7 ± 0.1 μm. Element wt%
C 0.21
Si 0.26
Mn 1.35
Ni 0.24
Cr 1.28
Mo 0.23
Table 1 – Chemical composition of 23MnCrMo5 grade
Lubricants In order to evaluate the influence of additive packages on pitting, two fully formulated manual gearbox lubricants (A and B) were subjected to a severe operating cycle in a complete transmission on a transmission test stand. Elementary chemical compositions together with the viscosities of formulations A and B are presented in Table 2. The same base stock, composed of polyalphaolefin (PAO) mixed with ester, is used for lubricant A and B. Anti-Wear (AW), detergent (Det) and extremepressure (EP) additives were part of the additive packages. However the chemistries and amounts of these additives in formulation A were different from those in formulation B. Since this study focuses on the influence of lubricant chemistry on crack propagation, the lubricants were formulated to have final
5
viscosities as close as possible to each other. In this way the potential impact of lubricant viscosity can be discounted in any analysis of observed differences in gear fatigue between lubricant A or B.
Additives Package A B
Base stock
S (wt ppm)
Ca (wt ppm)
P (wt ppm)
PAO + ester
3470 7660
680 3051
660 942
η 40°C (cSt) 30.29 30.18
η 100°C (cSt) 6.65 6.68
VI 184 188
Table 2 – Information about lubricants Transmission test stand procedure
A complete transmission was mounted on a test stand (Figure 2) and was used to evaluate the tendency of lubricants A and B to promote or resist RCF on the gears. The gearboxes were tested under high torque, using an electrically driven test rig with two asynchronous induction motors, one working in driving mode and the other providing the load via resistance; the gearbox lubricant temperature was maintained at below 105°C; input rotation speed was set to 3000 rpm. Load and velocities distributions (dimensionless) along the line of action of teeth are presented on Figure 2.c. The test was stopped automatically when one of the following conditions was met (i) vibrations were detected; (2) gear damage was identified; (3) after 200 cycles if no damage was detected. In one test cycle, all gear ratios were tested with relative duration relevant vs. service field, for several hours, as described in Table 3. After testing, the gearbox was unmounted from the test stand and the secondary shaft pinion was removed (one pignon for each lubricant). Endurance life expectancy is defined by car
manufacturer fatigue life proprietary model. The duration target and the number of repeats may vary according the gearbox component considered. In this study we have considered lubricants which failed before the expected target of fifty percent of endurance completed without any damage.
The center was drilled out of the pinion and cuts were then made along each bottom land, separating each gear tooth for individual analysis. The final preparation steps consisted of cutting tooth transversally providing cross-sections and metallographic preparation in order to perform SEM observations of the cracks. The morphologies of cracks along the tooth surface (length and orientation of the cracks) were studied on five teeth for each lubricant. Only representative conclusions will be reported in the results part.
6
Air blowers
b)
a)
Brake
c)
Figure 2 : a) General view of the transmission test stand b) schematic of manual gearbox : secondary shaft pinion evaluated c) pinion/wheel teeth contact : load and velocities distribution along the line of action. The dash line represents the position on the length of action of teeth where the cracks were selected for TEM observations. The stars represent the velocities and load conditions at this selected position.
Gear ratio 1 2 3 4 5
Torque (N.m) 56% torque 94% torque Full torque Full torque Full torque
Unit cycle proportional weight 5% 27% 17% 17% 53%
Table 3 – Mechanical conditions during one cycle. The full test consists of 200 repetitions of this cycle.
7
Complementary tests were performed to help understand additive-crack interactions: Chemical reactivity of lubricants with surfaces In order to evaluate the reactivity of lubricant additives with nascent steel surfaces, freshly polished specimens were exposed to 15 ml of the lubricant at 120°C in a glass beaker. Specimens were made of the same steel grade as the tested gearbox pinions and were subjected to similar low pressure carburization process. No difference in micro-hardness was observed between these specimens and the tested pinion surfaces. The final steps of specimen preparation consisted of silicon carbide polishing and were performed whilst immersed in PAO in order to minimize the formation of iron oxides. This was confirmed by X-ray Photoelectron Spectroscopy (XPS) analysis. After exposure to the lubricants, EDS and XPS analyses were performed. Friction tests Friction tests were carried out using a custom designed tribometer employing a reciprocating ball-onflat configuration. Balls used were made of AISI 52100 steel (hardness of about 800 HV). The same type of specimen with the same preparation as those prepared for the lubricant exposure test was used as the flat. The roughness (Ra) of the plates after the last polishing step (Grit 1200 whilst immersed in PAO) was about 30 nm. Tests were performed at 120°C with lubricant A or B. Two friction tests were performed for each set of conditions. Parameters (frequency, load) and the calculated maximal Hertzian pressures are presented in Table 4. For each test, the friction coefficient was calculated and XPS analysis of the wear scars was performed. Test
Frequency (Hz)
1 2
5 5
Stroke length (mm) 3 3
Load (N) 10 15
Mean PHertz (MPa) 595 683
Max PHertz (MPa) 895 1024
Table 4 – Parameters of friction tests Microstructural characterization Micrographs were made using a Zeiss Supra VP55 scanning electronic microscope with a Field Emission Gun (FEG) using high current mode with a tension from 5 to 15 kV. The FIB was used to prepare TEM foils containing a pre-selected pitting crack. To compare lubricant effect, the cracks were selected on pinion teeth at the same position on the length of action. For each lubricant tested, the studied cracks were so obtained for identical contact conditions. These samples were observed using
8
a JEOL 2010F FEG TEM. SEM and TEM devices were equipped with an Oxford Instrument Xmax Silicon Drift Detector for EDS analysis. An EBSD (Electron Backscattered Diffraction Spectroscopy) system installed on the SEM was also used to investigate the gear teeth microstructure evolution during RCF tests. Film formation after either friction tests or exposure of the steel to a lubricant was investigated by XRay Photoelectron Spectrometry (XPS) with an ULVAC-PHI VersaProbe II spectrometer using a monochromatized AlKα X-ray source (1486.6 eV). The binding energy reference was taken for C1s adventitious contribution at 284.8 eV.
9
Results Damage was characterized for both lubricants, A and B, for the most stressed part of the gear, i.e. the pinion, after transmission tests. A macroscopic evaluation is first carried out prior to microscopic and microstructural observations. Automatic interruption of the transmission test (complete transmission on a test stand) occurred after 18 cycles for lubricant A and 66 cycles for lubricant B (Table 5), which corresponds to 9% and 33% of the maximum expected test duration (200 cycles) respectively. Damage development occurs therefore earlier and/or faster for lubricant A. In both cases, the macroscopic observation of pinion tooth surfaces reveals the same pattern of damage: RCF damage is observed on the whole contact surface and the pitted area is located mainly below the pitch circle. Large pits are also observed in the tooth flank on some teeth (Figure 3). These large pits propagate to the tip of the flank, towards the pitch circle, and sometimes lead to tooth breaking. Lubricant A
Testing advance 18 cycles
B
66 cycles
Reason to end the test Automatically stopped due to vibrations
Observations - Pitting on seven teeth - 2 chipped teeth - Pitting on eight teeth - 4 chipped teeth
Table 5 – Results of full-transmission stand
a)
b) Figure 3 – Macroscopic observation of pinions after full scale transmission tests: a) Pinion tested with lubricant A. b) Pinion tested with lubricant B
Further SEM observations of the teeth flanks and teeth cross-sections prove that all micro-cracks initiate from the surface; 1 µm diameter MnS inclusions are detected by EDS whereas no subsurface initiated cracks are observed. As expected, the statistical analysis of micro-crack orientation shows that the mean angle between the propagation direction and surface is higher at the dedendum bottom and addendum top and decreases with an inversion of crack orientation direction at pitch line. Classic micro-crack orientations are reported in Figure 4. Further, no significant difference (>5°) in crack
10
orientation is observed between pinions tested with lubricant A or lubricant B (Figure 5) for a given position on the tooth (the chosen position is shown in figure 2.c : dashed line). SEM backscattered electron imaging of the tooth surface cross section exhibits grain refinement between the micro-cracks and the surface (Figure 6). EBSD mapping of these areas confirms and clarifies this observation (Figure 7). Indeed, areas away from the micro-cracks present the lenticular (needle-shaped) microstructure characteristic of the martensitic crystalline structure together with residual austenite whereas the areas between the micro-cracks and the external surface exhibit a nano-scale microstructure free of residual austenite. These refined microstructure areas are only observed conjointly with micro-cracks.
a)
b)
Figure 4 – SEM images of pinion teeth after full scale test: a) image of tooth flank b) tooth cross-section
11
Figure 5 – SEM images of micro-cracks obtained at the same position on the teeth after full scale test for lubricant A and B.
2 µm
Figure 6 – Backscattered electron mode (SEM) image showing the refined microstructure between cracks and surface.
12
a)
b)
Ferrite Residual austenite Carbides
c)
10 µm Figure 7 – a-b) SEM and EBSD images showing the refined microstructure between cracks and the surface and the disappearance of retained austenite c) EBSD image of initial steel microstructure.
13
Scanning electron micrographs reveal that the crack faces are covered with a film (Figure 8). In order to permit higher resolution analysis by TEM, FIB is used to cut very thin slices of the gear teeth which contain micro-cracks. To allow a comparison between lubricants A and B, the micro-cracks for analysis were selected such that comparisons could be made between areas that had been subjected stress conditions as similar as possible to each other (same rolling/sliding ratio). For this reason cracks selected for further observation are systematically selected and compared at the same position on the pinion tooth flank (i.e. same distance from pitch line and both dedendum or both addendum) with morphologies as similar as possible (direction of and length of propagation).
Figure 8 – SEM micrograph of tooth flank cross-section. A film is present along the micro-crack faces.
14
TEM observations confirm the presence of a solid film along internal crack faces for both lubricants (Figures 9 and 10). This film is composed of amorphous areas and crystallites. The thickness of the observed films fluctuated from few nanometers (20 nm) close to the crack tip and up to 200 nm at the crack mouth; differences in film thickness were observed between lubricants A and B. Whereas the progression in film thickness, from crack to tip, is relatively uniform with lubricant A, a more variable progression is observed along the crack length with lubricant B (Figure 10). It must also be noticed that higher film thickness is observed when the crack is oriented parallel to the surface. Further TEM/EDS analysis shows a clear distinction between the composition of films induced by lubricants A and B. In both cases oxygen, phosphorus and calcium are found in the film all along the crack (α and β positions in Figures 9 and 10), however sulphur is not detected in the case of lubricant A, neither on the crack faces (α and β positions in Figure 9) nor at the crack tip (γ position in Figure 9). On the other hand, sulphur is detected at the crack tip for lubricant B (γ’ position in Figure 10) and all along the crack near the metal / film interface (γ position in Figure 10). The EDS spectra of film obtained on crack faces with lubricant B (Figure 10) demonstrates that the relative concentrations of oxygen, phosphorus, sulfur and calcium vary with film depth, sulphur being predominant close to the metal / film interface whilst oxygen, phosphorus and calcium predominate near the film surface.
15
Intensity (arb. units) Intensity (arb. units)
Figure 9 – TEM images and EDS analyses of selected crack for lubricant A
Figure 10 – TEM images and EDS analyses of selected crack for lubricant B
16
Discussion The SEM observations of the pinions after the transmission tests have enabled us to set up RCF characteristics of the case studied here. The morphologies of RCF cracks are in agreement with literature [17] resulting in a decrease in the angle between the crack direction and the external surface from about 45° at the tooth filet to about 10° close to the pitch diameter for both lubricants. Propagation direction can be linked to the slide-to-roll ratio, decreasing from the tooth filet to the pitch diameter. Therefore all the samples, regardless of the tested lubricant, encountered the same mechanical contact conditions. However the duration of the protection from damage of the pinions by lubricant A is 70% lower than that of lubricant B. Consequently, we can conclude that the nature of the additives plays a role in the initiation and / or the propagation of RCF damage. Moreover, no subsurface crack initiation is observed; Cracks seems to initiate from the external surface although crack initiation from inclusions can’t be completely excluded. According to the different types of RCF damage defined by Santus [6], the types of damage encountered here is pitting and micropitting and the influence of a lubricant on crack initiation and propagation must be considered. Full-scale transmission tests show a clear influence of lubricant formulation on gear resistance to RCF. The pitting that leads to the breakage of the pinion teeth in the case of the present study is clearly delayed when lubricant B is used. As stated previously, different hypotheses may be introduced regarding the lubricant effect on RCF; the lubricant can accelerate or delay crack initiation: (i) through a mechanical-chemical action if additives initiate corrosion pits [12,13] or (ii) through a tribochemical reaction forming a tribofilm which modifies the surface roughness during run-in [25,26]. In addition, the tribofilm formed on the gear tooth surface can affect the coefficient of friction between sliding parts which can modify the stress field seen by the surface. This study has not been designed to allow us to discrimate against these three phenomena. The lubricant can also exert an influence on the propagation of cracks through different effects: (i) the mechanical effect of lubricant: the pressure transmitted by the lubricant which has penetrated (or which is trapped inside the crack) increases the crack propagation rate; it can also modify lubrication conditions between crack faces [27]; (ii) the effect of the tribofilm formed on the gear tooth surface at the crack mouth can prevent the penetration of the lubricant inside the crack [24], (iii) the effect of
17
some lubricant additives which react with nascent steel inside the crack and form film inside the crack which can affect positively or negatively crack propagation [20]. Since the studied lubricants have similar viscosities and, in terms of the base oil, have similar formulations, the mechanical effect of lubricant on propagation shall not be considered here. Moreover crack mouth obstruction by the tribofilm was not observed during this study. Further discussion is therefore focused on the effects of lubricant additive components. EDS/SEM and EDS/TEM analyses confirm the observations of Méheux et al. [20]. Indeed calcium, phosphorus and sulphur, deriving from anti-wear and extreme-pressure additives, as well as oxygen can be detected inside the cracks. A difference between lubricant A and lubricant B is nevertheless observed. No sulphur is detected for lubricant A whereas in the case of lubricant B sulphur is detected at the crack tip and on the crack faces close to the metal. The results of observations and chemical analyses are summarized in Figure 11, they once again confirm the penetration of lubricant additives inside the crack. From this crack characterization it appears that the lubricant reacts in two different ways inside the crack: (i) at the crack tip where only a thin film is observed and where, in the case of lubricant A, only oxygen is detected and the case of lubricant B only sulphur is detected; (ii) along the crack faces where the lubricants form tribofilms of heterogeneous thickness, composed of calcium, phosphorus and oxygen. In the case of lubricant B, sulphur is also found close to the metal surface.
Figure 11 – Overview of cracks chemistries from TEM observations and EDX analyses
18
In order to better understand the formation of the observed films into cracks, complementary experiments were set up. To simulate the reaction of additives with nascent steel surfaces at the crack tip during the propagation freshly polished steel samples were exposed to the lubricants. Friction tests were also carried out to simulate the friction between the crack faces. After exposure to lubricant B nascent steel surfaces exhibit a dark blue color for which the intensity depends on exposure time and temperature. No visible surface alteration is observed after exposure to lubricant A. SEM observations of exposed samples confirm the presence of a film on the steel surface in the case of lubricant B (Figure 12). Film thickness varies from 200 nm to 1 µm depending on exposure time and temperature. EDS analysis shows that the formed film is composed of iron and sulphur. Complementary XPS analysis shows that the energy values of the main XPS S2p3/2 peaks are 161.7 eV and 162.5 eV (Figure 13), which corresponds to FeS and FeS2 respectively [28,29]. Steel surfaces exposed to lubricant A exhibit a film that contains no sulphur. The growth of an iron sulfide containing film is therefore only obtained when the steel samples are exposed to lubricant B which is in accordance with observations made during the characterization of the crack. This tends to prove that the presence of sulphur (lubricant B) observed at the crack tip is due to the lubricant additive’s reaction to freshly formed steel surfaces.
19
a)
b)
Figure 12 – a) SEM image of sample cross-section after exposure to lubricant B (12 h – 120°C) b) Spectrum of EDS analysis of sample surface
FeS
FeS2 Adsorbed additive
Figure 13 – XPS analysis of steel surface after exposure to lubricant B during 12h, S2p peak
20
XPS survey scans and XPS depth profiles of tribofilms from the complementary friction tests show the presence of calcium, oxygen and phosphorus but no sulphur regardless of the lubricant tested (Figures 14 and 15). Sulphur is therefore not detected at the surface of either tribofilm (Figure 15) as was the case for both lubricants films formed on the crack faces (α and β positions in Figures 9 and 10). Yet, for the tribofilm formed during friction tests with lubricant A, sulphur is not detected at the iron / tribofilm interface, nor in the whole depth of the tribofilm. It was however detected for the film formed on crack faces with lubricant B (γ’ position in Figure 10). It may consequently be suggested that iron sulfide forms at the crack tip when a nascent steel surface is in contact with a sulphur reach lubricant (i.e. lubricant B in our study); the crack then propagates due to RCF (probably via mode II) and a Ca and P rich tribofilm is formed on crack faces due to friction. The friction conditions in the friction tests and the ones encountered in real cracks are certainly different, it is however reasonable to assume that the film observed on the crack faces is a tribofilm i.e. obtained by friction, as already demonstrated by Aldana et al [24]. In this later study the observation of the transformation of WS2 nanoparticles (used as an additive in the lubricant) into a layered and planar structure allowed the authors to describe the film formed within the crack as a tribofilm.
21
Figure 14 – XPS survey scans of wear tracks obtained with lubricant A and B
22
Lubricant B
Lubricant A
Fe
Fe Tribofilm
Substrate Atomic concentration (at%)
Atomic concentration (at%)
Tribofilm
O
Ca P
Etching time (min)
Substrate
O
Ca P Etching time (min)
Figure 15 – XPS depth profiles of tribofilms obtained after friction tests with lubricant A and B. In the present case, the occurrence of a tribofilm within the crack is also in agreement with the observation of grain refinement between the external surface and the cracks highlighted by EBSD analysis (Figure 7). Both phenomena actually result from severe friction conditions between crack faces. Frequent displacements of this upper part (between the external surface and the cracks) causing by sliding can be responsible for high strain hardening leading to severe refinement of the microstructure [17,19]. Based on these observations, some mechanisms for crack propagation can be discussed. The modification of the friction coefficient between crack faces due to the formation of the tribofilm may firstly be considered. Several studies have considered the role of the friction coefficient between the crack faces when modeling RCF crack propagation [17,19]. More generally speaking, this is a parameter known to have an influence on mode II or III fatigue driven crack propagation [19]. In actual fact, by reducing the coefficient of friction, crack face sliding is less energy consuming and a higher propagation rate is expected. Numerous different models and hypotheses are used for RCF crack propagation and the conclusions of different works differ regarding the importance of the crack face friction coefficient. However differences in the propagation rate are generally observed in models for a difference of friction coefficients of at least 0.2 [19]. If the assumption is made that tribofilms formed during friction tests and inside cracks are similar then the difference in friction coefficient between
23
lubricant A and B (0.10 versus 0.12) appears to be too weak to have a significant influence here (Figure 16). Moreover, damage observed on the pinions is greater when lubricant A is employed (i.e. the crack propagation rate is higher in the case of lubricant A) while its friction coefficient is the highest. This contradicts the better resistance to RCF observed with lubricant B. The modification of the friction coefficient between crack faces due to the formation of the tribofilm is therefore not a first order mechanism for explaining RCF crack propagation.
Figure 16 – Coefficient of friction obtained from reciprocating ball-on-flat tests Another mechanism by which the lubricant could impact the crack propagation is proposed by Aldana et al. [24]. They noticed that tribofilms formed inside cracks could prevent the penetration of lubricant inside the crack and thus avoid fluid pressurization or entrapment. In the present study, no evidence of the presence of a film closing the crack was observed. On the other hand, the influence of hydrogen on RCF has been widely cited in literature. Studies have proved that the presence of diffusible or trapped hydrogen in concentration above 1 or 2 wt. ppm reduces resistance to RCF [30–32]. However, before considering hydrogen embrittlement mechanisms, the origin of the hydrogen must be discussed. Indeed the intrinsic hydrogen content of steel used in gear manufacturing (less than 0.3 wt ppm) cannot explain such an embrittlement mechanism. Water contamination of the lubricant could be considered as an explanation; however the amount of contamination needed to lead to 1 wt ppm hydrogen concentration cannot be reached with
24
our testing conditions. Thus hydrogen generation from lubricant base oil decomposition is a much more promising explanation. It has been demonstrated experimentally that lubricant components can chemically decompose at nascent steel surfaces allowing local hydrogen chemisorption and then diffusion into the bulk steel [22,23,30,33–35]. This has been demonstrated for mineral and poly-alphaolefin (PAO) base oils and PAO is the key base oil used in this study. The presence of ester in the base oil mixture has been demonstrated to increase wear by fatigue which is believed to be due to a relative increase in the liberation of hydrogen compared to other base oil types [34]. The lubricants in this study contained ester base oils along with the PAO base oils and our findings are coherent with those noted above. As far as the influence of lubricant additives is concerned, it is known that corrosion inhibitors and AW/EP additives can deactivate the catalytic action of nascent steel surfaces by rapidly forming passivating films, such as antiwear phosphate and sulfide films [21,22]. In the present case the iron sulfide film formed at the crack tip when lubricant B is used could prevent or decrease the adsorption of hydrocarbon compounds leading to a reduced release of hydrogen into the steel. On the contrary, no sulfide film is observed for lubricant A. The moment when the iron oxide film was formed is therefore questionable: during the full-scale transmission test or after FIB sample preparation? In the latter case, no barrier exists and base oil hydrocarbon compounds adsorb and decompose on native iron surfaces which increase the amount of hydrogen at the crack tip. Hydrogen penetration consequently occurs and leads to an increase in the crack propagation rate by hydrogen embrittlement. If oxide formation occurs during crack propagation then a barrier effect has to be considered as demonstrated by Niste et al. [21] who show that hydrogen permeation is better prevented by films containing than iron oxide films. Finally, the resulting hydrogen assisted RCF could be described by classical mechanisms considered for hydrogen embrittlement in high strength steels [36]: (i) hydrogen enhanced decohesion (HEDE): local hydrogen accumulation in hydrostatic tensile regions at crack tips weakens the cohesive bonds between metal atoms and increases the crack propagation rate; (ii) hydrogen enhanced localized plasticity (HELP): local hydrogen favors dislocation mobility which results in localized softening, therefore enhancing localized plasticity.
25
CONCLUSION In this study, full scale transmission tests together with detailed sample characterizations by TEM and XPS have been used to study the role of lubricant additives in RCF crack propagation mechanisms. Damage was identified as surface initiated pitting and micropitting. The mechanical influence of the lubricant on RCF was excluded and the discussion focused on the chemical reactivity of the addditives. Indeed EDS analyses of crack cross-sections demonstrate the presence of elements from lubricant additives on crack faces and down to crack tips, confirming the penetration of the lubricant within the crack. Sulphur was found to play a major role in the formation of films at the crack tip. Two types of metal/additive interactions within the crack were identified: (i) at the crack tip: depending on the lubricant formulation additives may react with nascent steel surfaces leading to iron sulfide film formation; (ii) on crack faces: crack face friction leads to tribofilm formation. The modification of the friction coefficient on crack faces and the obstruction at the crack by formation of the tribofilm were considered to be second order processes compared to hydrogen embrittlement effects. Base oil degradation has been demonstrated in the literature to be at the origin of hydrogen generation. Sulphur contained in the lubricant additives leads to the formation of iron sulfide at the crack tip that can act as both (i) an inhibitor of lubricant decomposition at the nascent surface since access to the surface is physically blocked thereby reducing the catalytic degradation of the lubricant and therefore limiting hydrogen generation and/or, (ii) once hydrogen is formed, a barrier film to hydrogen permeation. The rate of crack propagation is therefore reduced.
Acknowledgements The authors would like to thank ANRT for support.
26
BIBLIOGRAPHY [1] [2] [3] [4] [5] [6]
[7] [8]
[9] [10]
[11]
[12] [13] [14]
[15] [16] [17] [18] [19] [20]
[21] [22]
[23]
[24]
[25]
S. Way, Pitting due to rolling contact, ASME Trans. J. Appl Mech. (1935) A49–A58. A.V. Olver, The mechanism of rolling contact fatigue: An update, Proc. Inst. Mech. Eng. Part J J. Eng. Tribol. 219 (2005) 313–330. doi:10.1243/135065005X9808. Y. Ding, N.F. Rieger, Spalling formation mechanism for gears, Wear. 254 (2003) 1307–1317. doi:10.1016/S0043-1648(03)00126-1. W.E. Littmann, The mechanism of contact fatigue, in: Interdiscip. Approach Lubr. Conc. Contacts, P. M. Ku, Troy, NY, 1969: pp. 309–378. T.E. Tallian, Y.P. Chiu, A. Van, Prediction of traction and microgeometry effects on rolling contact fatigue life., TRANS ASME J LUBR TECHNOL. 100 (1978) 156–166. C. Santus, M. Beghini, I. Bartilotta, M. Facchini, Surface and subsurface rolling contact fatigue characteristic depths and proposal of stress indexes, Int. J. Fatigue. 45 (2012) 71–81. doi:10.1016/j.ijfatigue.2012.06.012. D. Nélias, M.L. Dumont, F. Champiot, A. Vincent, L. Flamand, Role of inclusions, surface roughness and operating conditions on rolling contact fatigue, J. Tribol. 121 (1999) 240–250. A.C. Batista, A.M. Dias, J.L. Lebrun, F. Le, G. Inglebert, Contact fatigue of automotive gears: Evolution and effects of residual stresses introduced by surface treatments, Fatigue Fract. Eng. Mater. Struct. 23 (2000) 217–228. doi:10.1046/j.1460-2695.2000.00268.x. B.-R. Höhn, K. Michaelis, Influence of oil temperature on gear failures, Tribol. Int. 37 (2004) 103– 109. doi:10.1016/S0301-679X(03)00047-1. W. Tuszynski, R. Michalczewski, W. Piekoszewski, M. Szczerek, Effect of ageing automotive gear oils on scuffing and pitting, Tribol. Int. 41 (2008) 875–888. doi:10.1016/j.triboint.2007.12.010. W.E. Littmann, B.W. Kelley, W.J. Anderson, R.S. Fein, E.E. Klaus, L.B. Sibley, W.O. Winer, Chemical effects of lubrication in contact fatigue - 3. load-life exponent, life scatter, and overall analysis., J Lubr Technol Trans ASME. 98 Ser F (1976) 308–318. A.B. Cardis, Gear oil micropitting evaluation, Gear Technol. 17 (2000) 30–35. A.A. Torrance, J.E. Morgan, G.T.Y. Wan, An additive’s influence on the pitting and wear of ball bearing steel, Wear. 192 (1996) 66–73. E. Lainé, A.V. Olver, M.F. Lekstrom, B.A. Shollock, T.A. Beveridge, D.Y. Hua, The effect of a friction modifier additive on micropitting, Tribol. Trans. 52 (2009) 526–533. doi:10.1080/10402000902745507. A.F. Bower, Influence of crack face friction and trapped fluid on surface initiated rolling contact fatigue cracks, in: 1988. A.V. Olver, L.K. Tiew, S. Medina, J.W. Choo, Direct observations of a micropit in an elastohydrodynamic contact, Wear. 256 (2004) 168–175. doi:10.1016/S0043-1648(03)00374-0. M. Kaneta, Y. Murakami, Effects of oil hydraulic pressure on surface crack growth in rolling/sliding contact, Tribol. Int. 20 (1987) 210–217. doi:10.1016/0301-679X(87)90076-4. Y. Murakami, S. Hamada, A new method for the measurement of mode II fatigue threshold stress intensity factor range ΔKτth, Fatigue Fract. Eng. Mater. Struct. 20 (1997) 863–870. J.W. Ringsberg, A. Bergkvist, On propagation of short rolling contact fatigue cracks, Fatigue Fract. Eng. Mater. Struct. 26 (2003) 969–983. doi:10.1046/j.1460-2695.2003.00657.x. M. Meheux, C. Minfray, F. Ville, T.L. Mogne, A.A. Lubrecht, J.M. Martin, H.P. Lieurade, G. Thoquenne, Effect of lubricant additives in rolling contact fatigue, Proc. Inst. Mech. Eng. Part J J. Eng. Tribol. 224 (2010) 947–955. doi:10.1243/13506501JET719. V.B. Niste, H. Tanaka, M. Ratoi, J. Sugimura, WS2 nanoadditized lubricant for applications affected by hydrogen embrittlement, RSC Adv. 5 (2015) 40678–40687. doi:10.1039/c5ra03127c. R. Lu, S. Mori, T. Kubo, H. Nanao, Effect of sulfur-containing additive on the decomposition of multialkylated cyclopentane oil on the nascent steel surface, Wear. 267 (2009) 1430–1435. doi:10.1016/j.wear.2009.03.028. R. Lu, H. Nanao, K. Kobayashi, T. Kubo, S. Mori, Effect of lubricant additives on tribochemical decomposition of hydrocarbon oil on nascent steel surfaces, J. Jpn. Pet. Inst. 53 (2010) 55–60. doi:10.1627/jpi.53.55. P.U. Aldana, F. Dassenoy, B. Vacher, T. Le Mogne, B. Thiebaut, A. Bouffet, Antispalling Effect of WS2 Nanoparticles on the Lubrication of Automotive Gearboxes, Tribol. Trans. 59 (2016) 178–188. doi:10.1080/10402004.2015.1061080. L.J. Taylor, H.A. Spikes, Friction-enhancing properties of ZDDP antiwear additive: Part I Friction and morphology of ZDDP reaction films, Tribol. Trans. 46 (2003) 303–309.
27
[26] L.J. Taylor, H.A. Spikes, Friction-enhancing properties of ZDDP antiwear additive: Part II Influence of ZDDP reaction films on EHD lubrication, Tribol. Trans. 46 (2003) 310–314. [27] A.F. Bower, Influence of crack face friction and trapped fluid on surface initiated rolling contact fatigue cracks, J. Tribol. 110 (1988) 704–711. [28] C. Minfray, G. Fromentin, A. Bierla, J.-M. Martin, T. Le Mogne, The effect of an organic pentasulfide EP additive in turning and milling operations, Wear. (n.d.). doi:10.1016/j.wear.2014.05.009. [29] M. Descostes, F. Mercier, N. Thromat, C. Beaucaire, M. Gautier-Soyer, Use of XPS in the determination of chemical environment and oxidation state of iron and sulfur samples: Constitution of a data basis in binding energies for Fe and S reference compounds and applications to the evidence of surface species of an oxidized pyrite in a carbonate medium, Appl. Surf. Sci. 165 (2000) 288–302. doi:10.1016/S0169-4332(00)00443-8. [30] N. Kino, K. Otani, The influence of hydrogen on rolling contact fatigue life and its improvement, JSAE Rev. 24 (2003) 289–294. doi:10.1016/S0389-4304(03)00035-3. [31] H. Uyama, The Mechanism of White Structure Flaking In Rolling Bearings, (2011). http://www.nrel.gov/wind/pdfs/day2_sessioniv_1_nsk_uyama.pdf. [32] A. Ruellan, F. Ville, X. Kleber, A. Arnaudon, D. Girodin, Understanding white etching cracks in rolling element bearings: The effect of hydrogen charging on the formation mechanisms, Proc. Inst. Mech. Eng. Part J J. Eng. Tribol. 228 (2014) 1252–1265. doi:10.1177/1350650114522452. [33] M. Kohara, T. Kawamura, M. Egami, Study on Mechanism of Hydrogen Generation from Lubricants, Tribol. Trans. 49 (2006) 53–60. doi:10.1080/05698190500486324. [34] P. Nevasmaa, A. Laukkanen, Assessment of hydrogen cracking risk in multipass weld metal of 2.25Cr-1Mo-0.25V-TiB (T24) boiler steel, Weld. World. 49 (2005) 45–58. [35] R. Lu, H. Nanao, K. Takiwatari, S. Mori, Y. Fukushima, Y. Murakami, S. Ikejima, T. Konno, The Effect of the Chemical Structures of Synthetic Hydrocarbon Oils on Their Tribochemical Decomposition, Tribol. Lett. 60 (2015). doi:10.1007/s11249-015-0606-2. [36] W. Gerberich, Modeling hydrogen induced damage mechanisms in metals, in: Gaseous Hydrog. Embrittlement Mater. Energy Technol. Mech. Model. Future Dev., 2012: pp. 209–246.
28