Post-print of: Surface and Coatings TechnologyVolume 206, Issue 7, 25 December 2011, Pages 1913–1920
Identification of the wear mechanism on WC/C nanostructured coatings
S. El Mrabet, M.D. Abad,J.C. Sánchez-López
Instituto de Ciencia de Materiales de Sevilla (CSIC-US), Avda. Americo Vespucio 49, 41092 Sevilla, Spain
Abstract
A series of WC/C nanostructured films with carbon contents ranging from 30 to 70 at.% was deposited on M2 steel substrates by magnetron sputtering of WC and graphite targets in argon. Depending on the amorphous carbon (a-C) incorporated in the coatings, nanocrystalline coating (formed mainly by WC1 − x and W2C phases) or nanocomposite (WC1 − x/a-C) were obtained with tunable mechanical and tribological properties. Ultrahardness values of 36–40 GPa were measured for the nanocrystalline samples whilst values between 16 and 23 GPa were obtained in the nanocomposite ones depending on the a-C content. The tribological properties were studied using a pin-on-disk tester versus steel (100Cr6) balls and 5 N of applied load in dry sliding conditions and the failure modes by scratch adhesion tests. Three different zones were identified according to the observed tribological behavior: I (μ > 0.8; adhesive wear), II (μ: 0.3–0.6; abrasive wear) and III (μ ~ 0.2; self-lubricated). The wear tracks and the ball scars were observed by scanning electron microscopy (SEM) and Raman spectroscopy in order to elucidate the tribochemical reactions appearing at the contact and to determine the wear mechanism present in each type. A correlation among structure, crystalline phases, a-C content and tribomechanical properties could be established for the series of WC/C coatings and extended to understand the trends observed in the literature for similar coatings.
Keywords
WC;Coatings;Amorphous carbon; Structure; Mechanical properties;Tribology
1. Introduction
Tungsten carbide (WC), a well-known refractory material is widely used in the industrial applications because of its high hardness, high elastic modulus, wear resistance and chemical inertness [1], [2], [3], [4], [5], [6] and [7]. One of the main peculiarities of the WC material is the high number of compositional and structural forms that can exist according to the W–C phase diagram [8] and [9]. The deposition of tungsten carbide films have been done by many chemical vapor deposition (CVD) [10], [11], [12] and [13] and physical vapor deposition (PVD) methods [1], [2], [3], [4], [5], [6], [7], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35] and [36]. Gouy-Pailler and Pauleau[3] in 1993 found by X-ray diffraction analysis (XRD) that coatings with carbon contents below 25 at.% contained α-W phase, named W(C) films, whereas cubic WC1 − x phases were detected when carbon content surpassed 30 at.%. Palmquist et al. [14] demonstrated by XRD, transmission electron microscopy (TEM) and electron diffraction (ED) that a hexagonal W2C phase could be formed between the solid solution of C in α-W and the cubic WC1 − x phase for C contents between 20 and 35 at.%. The appearance of amorphous carbon (a-C) surrounding the W2C and WC1 − x crystalline phases was reported for the ranges superior to 30 or 40 at.% of C depending on the author [15], [16], [17], [18], [19] and [20]. The properties of such nanocomposite coatings depend critically on the ability to co-deposit both the nanocrystalline (nc-WC) and amorphous phases (a-C, a-C:H, DLC) controlling its relative amount and distribution inside the nanocomposite.
Focusing on the tribological applications, many groups have synthesized WC/C composites with the aim of combining the benefits of a hard nanocrystalline material with a soft solid lubricant [15], [16], [17], [18], [20], [21], [22], [23] and [24]. Hardness (H), stress and friction coefficient (μ) are usually the most studied parameters to determine the quality of the coatings for this functionality. In Fig. 1 it is shown a data review on hardness, stress and friction coefficient values vs. carbon content of different nanocomposites composed of WC nanocrystals (whichever stoichiometry) embedded in an amorphous carbon matrix (hydrogenated or hydrogen-free) found in the literature. A wide dispersion of the results and a lack of correlation can apparently be inferred. One can observe the existing differences in hardness and friction coefficient values between samples containing similar carbon content but obtained by different groups. This could be partially attributed to the employment of different deposition techniques and synthesis conditions. For instance, Voevodin et al. [15], [16] and [17] prepared WC/DLC nanocomposite coatings using laser ablation of graphite, meanwhile, Czyzniewski et al. [22], [23] and [24] prepared WC/a-C:H with acetylene as a precursor gas. Another property less studied that can provide complementary information on the mechanical behavior is the scratch adhesion [37], [38] and [39]. Hardness and scratch adhesion testing share many features and therefore that scratch testing constitutes a valuable tool to gain understanding of the deformation processes induced by the cumulative action of indentation and friction.
The main motivation of this paper is to investigate the friction and wear mechanisms of nanocomposite WC/a-C coatings using a set of well characterized samples as base material [20]. The determination of the fraction of carbon atoms bonded to tungsten or bonded to carbon (as a-C matrix phase) by X-ray photoelectron spectroscopy (XPS) resulted crucial for understanding the changes in structure and mechanical and tribological properties. In this paper, the acquired knowledge on the phase composition and microstructure together with a detailed analysis by microscopy and Raman techniques on the wear tracks originated by scratch and friction tests are used to explain the observed tribomechanical properties and correlate it with the reported literature data for similar WC/C or WC/C:H compounds.
2. Experimental details
WC/a-C coatings were prepared by Ar sputtering of WC (Kurt J. Lesker, 99.5% purity) and graphite (Goodfellow, 99.5% purity) targets connected to radio frequency (r.f.) and direct current (d.c.) power sources respectively. A series of samples has been prepared by changing the sputtering power ratio, defined as R = PC/PWC, from 0 to 3. The typical power values (PWC) applied to the WC target were 150 and 250 W while those applied to the graphite target (PC) were varied from 0 to 450 W. The obtained films are labeled as R0, R0.1, R0.3, R0.5, R1, R2 and R3. Further experimental details concerning the synthesis conditions can be found in reference [20].
Nanoindentation experiments were performed with a Nanoindenter II (Nano Instruments, Inc., Knoxville, TN) microprobe. All tests were carried out at room temperature with a diamond Berkovich (three-sided pyramid) indenter tip. The load–displacement data obtained were analyzed using the method of Oliver and Pharr [40] to determine the hardness and the elastic modulus as a function of the displacement of the indenter. The maximum load was selected in such a way that the maximum indentation depth did not exceed 10–15% of the coating thickness in order to avoid the influence of the substrate. The film stress was measured by measuring the substrate bending using a profilometer and the Stoney's equation.
The scratch tests were carried out using a TRIBOtechnicMillenium 200 scratch-tester. A Rockwell C diamond tip (200 μm radius) was used as an indenter. During the test, the indenter was drawn over the coated surface for 10 mm-length as the applied normal load increased continuously up to 100 N (loading rate of 50 N/min and a scratching speed of 10 mm/min). The diamond tip was cleaned after each scratch. Acoustic emissions were recorded during the scratching, but optical microscopy was applied to assess the critical normal load values and to indicate the coating fracture and delamination modes. Two critical loads were determined, the lower (LC1) being the start of the cohesive failure within the coating, and the upper (LC2) the onset of an adhesive failure of the coating [39]. Each sample was submitted at least to three scratch experiments to determine the general trends in performance of the coatings. From the collected results, mean values for the critical loads corresponding to specific damage of the films, and experimental variations were determined.
Tribological tests were carried out using 100Cr6 6 mm-diameter steel balls in a pin-on-disk CSM tribometer with a sliding speed of 10 cm/s and 5 N of applied load (maximum initial Herztian contact pressure of 1.12 GPa) in ambient air (30–60% of relative humidity). The sliding distance was 1000 m with typical track radius between 6 and 10 mm. Normalized wear rates (mm3/Nm) were evaluated from cross-sectional profiles taken across the disk-wear track after testing by means of stylus profilometry. Scanning electron microscopy (SEM) data were recorded in a FEG Hitachi S5200 microscope operating at 5 kV. Micro-Raman measurements were performed using a LabRAMJobinYvon spectrometer equipped with a microscope. Laser radiation (λ = 532 nm) was used as excitation source at 5 mW. All measurements were recorded under the same conditions (10 s of integration time and 10 accumulations) using a 100× magnification objective and a 100 μm pinhole.
3. Results
3.1. Hardness, critical loads and tribological properties
In previous works a series of WC/a-C coatings was prepared by magnetron sputtering and exhaustively characterized by XRD, TEM, ED, Raman and XPS techniques [20]. In summary, the results of the chemical and microstructural analysis revealed that nanocrystalline hexagonal W2C phase is the main phase by single sputtering of WC target. Then, the subsequent incorporation of carbon leads to a progressive reduction of the crystalline domain size and the nucleation of the cubic WC1 − x phase. From a total C content of approximately 50 at.% the formation of composite films containing nanocrystallites of cubic WC1 − x phase dispersed in an amorphous carbon matrix is clearly manifested. Further increase of the power applied to the graphite target leads to a progressive increment of the free amorphous carbon content becoming comparable to the crystalline fraction from overall 70 at.% of C.
Table 1 summarizes the mechanical and tribological properties of the set of WC/C coatings prepared varying the R parameter as a function of the total and amorphous free carbon contents. In Fig. 2a the hardness and friction coefficient of the samples are represented as a function of the a-C (at.%). The maximum hardness are obtained for the R0 and R0.1 samples (36 and 40 GPa respectively) with very low of a-C contents (< 10 at.%). The remaining samples exhibited a progressive diminution in hardness and friction coefficient values (μ) by increasing the a-C content to 16–20 GPa and μ ~ 0.2 respectively for the richest carbon samples. In Fig. 2b the values of the ball and film wear rates (K) are represented as a function of the a-C content. It should be mentioned that the Kfilm values for samples R0 and R0.1 are not provided due to the transfer of mating material (steel) to the surface making impossible the estimation of the wear track as it will be explained in the next sub-section. In Fig. 2c both critical load values have been plotted for each coating. It should be mentioned that the first deposition step consisted in switching on only the WC target (for 1 h) and then combined with the graphite one in order to have a graded transition from WC to WC/C layered system. As a result of this procedure a first layer of 100–200 nm of nanocrystalline W2C is located at the substrate interface. The LC1, the starting point of crack formation, was found in a very close range between 12 and 15 N for the coatings containing up to 16 at.% of a-C. For a-C contents higher than 26 at.%, the LC1 value was not observed as they deform mainly plastically as described in the next section. The found values for LC2 exhibited larger differences depending on the amount of the a-C phase. The samples with the lowest a-C showed very high values around between 63 and 69 N, followed by a significant decrease to 30 N when a-C is between 10 and 16 at.%. In the last group, the samples with a-C > 25 at.% the trend is reversed, reaching almost 50 N for the sample with the highest a-C content. In these coatings the critical load corresponded to the point where coating is scrapped off exposing the substrate.
In summary, the study of the dependence of the tribomechanical properties with the phase composition allows to establish three different zones using the amount of atomic carbon present in the a-C phase as main criteria: I (< 10 at.%); II (10–30 at.%) and III (> 30 at.%). This classification will be used in the following sections for the discussion of deformation modes and wear mechanisms.
3.2. Scratch tests and failure modes
A deeper study of the failure mode by means of optical microscopy was carried out in order to understand the differences in the critical load values. In Fig. 3 it is depicted the picture taken at 12.3 N for the coating R0.1 as starting point of the crack formation (conformal cracks). As the indenter moves along, several partial ring cracks are formed along the track and these rings may intercept with each other causing a network of cracks (LC2 = 63.0 N). Conformal cracking occurs by the compressive stress ahead of the moving indenter and this driving force is similar to bucking spallation. The reason for coating cracking without delamination indicates a sufficient interface substrate-film adhesion. This kind of failure response was also observed by scratching nanocrystalline coatings of WC by Czyzniewski [22].
Fig. 3 also shows the failure mode of the coating R0.3 representing the transition from the region I to II. Some diagonal lines are formed when the coatings is deformed by the indenter in its moving direction (LC1 = 13.6 N). These marks are called as “chevron tensile cracks” because they point to the crack origin. This mode of failure also occurs at the trailing end, but the fracture initiates near the two edges of the contact groove, forming a slanted angle to the sliding direction. As can be seen in the Fig. 3 (LC2 = 31.5 N) with further increase of the load, both the density and the length of the cracks increase, and spallation occurs due to the compressive stress generated by the indenter [41]. In Fig. 3 it is likewise shown the typical failure mode of the coating R0.5 (LC1 = 14.1 N). In this case the scratch failure mode corresponds to the formation of “tensile trailing cracks”. The tensile cracks are formed under the dominant effect of frictional traction stress. Besides the stresses caused by frictional pulling, coating bending due to the contact action also contributes to the tensile stress that opens up the crack perpendicular to the sliding direction behind the moving stylus. When the contact load is large severe grooving in the substrate will also cause cracking in the coating alongside the groove edge (LC2 = 27.9 N). The crack lines do not overlap with each other. This results are in good agreement with Czyzniewsky [22] where he observed that increasing carbon content in nanocomposite WC/a-C:H from 48.1 to 56.5 at.% there is an increase of the number of cracks and their propagation in this matrix. Czyzniewski proposed that together with the increasing carbon content, the thickness of the amorphous a-C matrix enclosing WC nanograins is also growing and in consequence number of cracks forms and propagates in this matrix.
In Fig. 3 we can see the wear scars of the R2 and R3 coating taking at LC2 = 23.0 and 45.6 N, respectively. The aspects of the wear tracks reveal that the coatings exhibited surface deformation, which visually appear to be “plastic”. Until above 20 N, it seems that the indenter produces a plastic deformation and just it is possible to observe some marks or longitudinal cracks at the border parallel to the indenter movement indicating material pile-up along the track borders. This behavior can be interpreted as the coatings have sufficient ductility to accommodate plastic deformation. These samples corresponded to the third region described in the previous section whose a-C contents were maximum (around 30 at.%). Voevodin and Zabinski [17] in their early work on WC/DLC coatings explained that this was not true plasticity since dislocation sources were prohibited, but rather the result of WC grain boundary sliding in the a-C matrix. The increase of the applied load led an increased extent of the “plastic” deformation until the point that the coating is scrapped off exposing the substrate.
3.3. Analysis post-tribo by SEM/EDX
In order to have a better insight about the chemical phenomena occurring at the contact area, analysis of the ball and film surfaces after friction tests is carried out by SEM/EDX and Raman on selected samples. R0.1 (3 at.% of a-C), R0.5 (16 at.% of a-C) and R2 (30 at.% of a-C) coatings have been selected as representative of high, medium and low friction regimes respectively. Ball and disk worn surfaces of R0.1 sample are shown in Fig. 4 as representative example of the high friction coefficient region (I) formed by the samples with less a-C content (R0 and R0.1). When sliding against steel, a friction coefficient of about 0.8 was found for these two coatings. This can be understood considering the great difference in hardness properties between film (36–40 GPa) and counterface (5–6 GPa) materials. Under high contact pressure (> 1 GPa), the softer steel is rapidly worn away leading to an increased friction. The more ductile steel is also transferred to the film surface appearing smeared onto the film wear track as can be seen in Fig. 4 in the track disk at higher magnifications. A SEM/EDX line scan of the wear track (depicted as a white mark) revealed that the composition of this adhered layer was mainly iron oxide. The analysis of the debris deposited on the ball scar by EDX was not concluding and so was lately assessed by Raman.
In the second region (R0.3 to R1), friction coefficients varied between 0.6 and 0.35 and the values of wear rates were relatively high for film (~ 10−6) and ball (~ 10−5–10−6 mm3/Nm) respectively. Ball and wear track for R0.5 film with 16 at.% a-C are shown in Fig. 4. The wear track is characterized by surface deformation in the form of longitudinal grooves produced by hard debris particles entrapped within the track. In this case, both high friction and wear values registered could be originated from a high concentration of hard phases in this zone. Therefore the production of the abrasive wear particles during frictional contact contributed to accelerate the coating damage by cracking and spallation achieving high friction and high wear rate.
For the third region corresponding to highest content of a-C (R2 and R3), low friction values in combination with low wear rates were obtained (μ < 0.2 and K ~ 10−7–10−8 mm3/Nm). The SEM pictures of both counterfaces obtained for the sample R2 are depicted in Fig. 4. The wear marks are less visible indicative of the improved tribological behavior in this region. The film track is characterized by shallow grooves and no loose debris appears covering the ball surface. This is in good agreement with the nanocomposite phase composition and the results observed by the scratch test where the high content of amorphous phase allowed to accommodate the shear induced by the tip deforming pseudoplastically.
3.4. Raman analysis of the friction contact regions
Raman spectra from the as-deposited coatings, the wear track surface and the transfer film formed on the ball counterfaces for the selected representative (R0.1, R0.5 and R2) samples are given in Fig. 5. The Raman spectrum of the transfer layer formed on the steel counterpart of sample R0.1 (Fig. 5a) showed the presence of two bands at 720 and 945 cm−1 which can be assigned to a mixture of iron and tungsten oxides or ferritungstite [42]. The strong broadening of these bands suggests that the formed compounds present a high structural disorder, indicating low crystallinity. The peak at 945 cm−1 is also possible to be assigned to the stretching mode of W=O bonds that appear on the boundaries of amorphous or nanostructured tungsten oxides [43]. These results correlate with the strong interaction of the steel counterpart with the film surface for these samples highlighted by the SEM/EDX analysis. In the track, very weak peaks in the range of 1300–1600 cm−1 corresponding to the D and G bands typical of disordered amorphous carbon were also observed with an additional peak at 880 cm−1 from crystalline WO3[16] and [44]. The Raman spectrum of the initial film does not display any features in the region of D and G bands indicating that this carbon Raman signal is only present in the contact after friction tests. The oxidation of the WC nanocrystals in air generates tungsten oxides and free carbon although insufficient or ineffective to lubricate the contact.