Surface and Coatings Technology 86-87 (1996) 564-568

An investigation of the relationship between graphitization and frictional behavior of DLC coatings

Y. Liu a, A. Erdemir b, E.I. Meletis a,*

a Louisiana State University, Mechanical Engineering Department, Materials Science and Engineering Program, Baton Rouge, LA 70803, USA

b Argonne National Laboratory, Energy Technology Division, Argonne, IL 60439, USA

Abstract

In our recent studies, diamond-like carbon (DLC) films were found to possess low coefficient of friction (fcO.1) and excellent wear resistance. The reduction in f was found to be consistent with wear-induced graphitization of the DLC structure. The purpose of the present work was to study the effect of load and sliding velocity on the frictional behavior and graphitization process occurring in DLC during wear. Pin-on-disc experiments were conducted on DLC-coated Sic substrates at sliding velocities between 0.06 and 1.6 m s-l under 1 and 10 N loading levels using ZrO, balls as the pin material. Analytical transmission electron microscopy was used to characterize the structure and microstructure of the wear debris after testing. The results showed that both sliding velocity and contact load influence the graphitization process. Higher sliding velocities increase the contact frequency and the rate of temperature rise that may facilitate the release of hydrogen atoms from the sp3 structure. Higher loading enhances shear deformation and transformation of the weakened hydrogen-depleted DLC structure into graphite [lo]. The present findings are consistent with our earlier proposed wear-induced graphitization mechanism for these films. An equation was developed to describe the transformation kinetics of DLC into graphite as a function of sliding velocity and applied stress.

Keywords: Diamond-like carbon (DLC); Wear-induced graphitization

1. Background

Diamond-like carbon (DLC) films have attracted considerable interest the last decade owing to their potential as advanced solid lubricant coatings. The state of carbon in these films is mainly amorphous, in which small clusters (< 30 A) of microcrystalline structure with sp3 and sp2 bonding, and amorphous matrix coexist [ 11. Hydrogen atoms are introduced into the DLC lattice during the deposition process when hydrocarbon gases are used. The hydrogen atoms play a crucial role in promoting and stabilizing sp3 tetrahedral bonds and are responsible for the high hardness of DLC films [2]. Since first deposited by Aisenberg and Chabot in 1971 [3], DLC films can now be prepared by different pro- cesses including d.c or r.f. plasma-assisted CVD, sputter- ing and ion-beam deposition (IBD) from a variety of carbon bearing solid or gaseous source materials. DLC films are characterized by high hardness (3000-9000 kg mme2), high elastic modulus [ 1,4] and

* Corresponding author. Tel.: + 1 504 3885806; fax: + 1 504 3885924.

0257-8972/96/$15.00 0 1996 Elsevier Science S.A. All rights reserved

PII SO257-8972(96)03057-5

generally low friction coefficient and wear rate [ 5,6]. The films typically have high optical transparency over a wide spectral range, high electrical resistivity and chemical inertness to both acids and alkalis [ 71. The high hardness and chemical resistance of DLC films make them excellent candidates as wear-resistant coat- ings for a wide variety of engineering components.

Recently, we have performed tribological studies on DLC films prepared by IBD on various metallic sub- strates (M50 steel, Ti-6Al-4V alloy, AISI 440C steel) [ 8,9]. The experimental results in agreement with previ- ous reports showed that DLC films exhibit low values of friction coefficient (f< 0.1) and excellent wear resis- tance (wear rate < 1.6 x lo-’ mm3/N m). Detailed analy- sis and examination of the friction characteristics showed that the tribological behavior of DLC films exhibited three distinct friction regimes and can be described by the following mechanism. Initially a break-in period with continuous decreasing friction coefficient exists, followed by an intermediate constant friction plateau of a relatively short duration which is preceding a low friction (fzO.05) steady-state stage. The initial

Y. Liu et al.lSurface and Coatings Technology 86-87 (1996) 564-568 565

reduction in the friction coefficient during the break-in stage has been attributed to the gradual release of hydrogen from the DLC structure at “hot spots” pro- ducing a low-shear strength layer. The intermediate constant friction plateau was related to the transfer layer formation with possibly some small scale graphit- ization, whereas the steady-state stage was a result of extensive graphitization in the tribolayer [ 91. Recently, we provided direct evidence of wear-induced graphitiza- tion of DLC coatings and a mechanism for the DLC--+graphite transformation was proposed [lo]. According to this mechanism, graphitization is related to the frictional energy and proceeds with a precursor hydrogen atom release stage followed by shear deforma- tion that converts the ( 11 l)nLC into hexagonal ( OO02)oB planes facilitating the nucleation of graphite.

In the present work, wear tests were conducted at different loading levels and sliding velocities in order to assess the effect of these parameters on the tribological response and graphitization process and provide further verification to the proposed wear-induced graphitization mechanism [ lo].

2. Experimental

DLC films of approximately 2 um thickness were deposited on Sic substrate at room temperature by methane IBD using a Kaufmann-type ion source. The deposition was conducted at an accelerating voltage of 750 eV and a current density of approximately 2.5 mA cm-‘. This process produces dense, adherent and low roughness DLC films. The Knoop microhard- ness of the produced films was about 6000 kg mm-’ (0.98 N). Film characterization studies [l] and experi- mental details on the deposition system [8,9] have been reported previously.

Pin-on-disc wear tests were conducted on DLC-coated SIC discs (50 mm in diameter) using ZrO, as the pin material (9.55 mm in diameter) under 1 and 10 N applied load and sliding velocities between 0.06 and 1.6 m s - ‘. A relatively wide range of wear parameters was selected in order to assess the effect of their intensity on the wear-induced graphitization process. The wear tests were conducted in laboratory air (relative humidity -35%) at room temperature (23 f 1 “C) for 10 km distance. The coefficient of friction (j) was monitored with the aid of a linear variable-displacement transducer and was con- tinuously recorded throughout the tests. Wear rate calcu- lations for the ZrO, balls were based on microscopic determination of the diameter (average of measurements in two vertical directions) of the wear scars. The wear of the DLC film was estimated from the traces of surface profiles across the wear tacks (average of four measure- ments) obtained by using a surface profilometer. At least two tests were performed under each wear condi-

tion described above and the reproducibility of the results remained within & 12% margin.

Scanning electron microscopy @EM) was used to observe the morphology of wear tracks. Analytical transmission electron microscopy (TEM) was carried out to study the microstructure of the as-deposited DLC film prior to wear testing and of wear debris collected from the wear tracks after testing. Electron diffraction studies were conducted to examine the structure of DLC prior and after wear testing (debris). Fine particles of as-deposited DLC were scratched from the discs using a sharp blade and transferred onto copper grids for examination. Similarly, wear debris was transferred to copper grids by the following procedure. A drop of distilled water was placed on the wear track and the specimen was ultrasonically vibrated in order to release loosely adherent wear debris on the wear track. Subsequently, a water drop containing fine debris par- ticles from the wear track was placed on a copper grid and let dry.

3. Results and discussion

Table 1 summarizes the testing parameters and experi- mental results of the tribological behavior of ZrOJDLC-coated Sic pair. More specifically, the parameters shown are the initial Hertzian contact pres- sure (a,), the initial and steady-state coefficient of friction (A,, and& respectively), the sliding distance to reach steady-state (S) and the wear rate of the ball (I+‘,,) and disc (Wn,) material. It is noted that in agreement with previous reports [8,9], DLC films exhibit low coefficients of friction and wear rates. Also, it is evident that sliding velocity and applied load have significant effects on their tribological behavior.

The SEM examinations showed that the as-deposited DLC films were smooth (& < 0.03 urn) and featureless. Cross-sectional observations at high magnifications showed that the films were dense and appeared to be free of bulk defects. The TEM observations showed that the as-deposited DLC film has mainly a featureless microstructure, Fig. l(a). Examinations at high magni- fications showed that the films possess a medium-range order with an approximate average domain size of 25 nm. Electron diffraction patterns from as-deposited DLC showed the presence of diffuse rings indicative of the amorphous nature of these films, Fig. 1 (b). Two diffraction rings were mainly observed with d-spacing of approximately 2.1 and 1.2 A, consistent with previous reports [ 1,9,1 I]. The measured d-spacing of the two rings for the as-deposited DLC films is suggesting the presence of a short-ranged diamond-like structure con- sidering that ( 111) and (220) are the strongest diffrac- tions for crystal diamond and correspond to d=2.06 and 1.26 A, respectively.

566

Table 1

Y. Liu et al. ISurface and Coatings Technology 86-87 (1996) 564-568

Tribological parameters of pin-on-disc tests

Test no. Velocity u Load (N) oH (MPa) A:,, f,, S(m) wa, W,’ (m s-‘) (x10-gmm3m-1N-1 ) (x10-‘mm3m-‘N-l) E10-4s-‘)

1 0.06 1 395 0.18 0.18 -* 10.1 ND 2 1.6 1 395 0.10 0.06 8000 3.56 8.36 2.0 3 0.1 10 851 0.13 0.08 9500 1.82 2.48 0.1 4 1.6 10 851 0.08 0.05 4300 0.56 1.99 3.7

uH, initial Hertzian contact pressure; J;-, initial friction coefficient; f,,, steady-state friction coefficient; S, sliding distance required to reach steady- state; ND, nondetectable; R, rate for graphitization.

*Steady-state was not reached under these conditions for a sliding distance of 25 km.

Fig. 1. (a) Transmission electron micrograph showing typical micro- structure of as-deposited DLC film and (b) the corresponding electron diffraction pattern.

Figure 2(a) and Fig. 2(b) present the morphology and microstructure of debris collected from wear scars after testing. The TEM observations revealed that the debris microstructure was distinctly different from that of as-deposited DLC. Electron diffraction analysis showed that graphite was present in all wear debris except for tests under low velocity (0.06 m s-l) and loading level (1 N), Test 1. Electron diffraction pattern analysis from the wear debris showed mainly the (0002)o, sp2-graphite reflection (d% 3.4 A) (Fig. 2(c)), confirming the presence of graphite in agreement with our previous studies [9,10]. Dark field analysis showed that the graphite particles had a size of about 3-4 nm and exhibited an orientation pattern, more than likely coin- ciding with the sliding direction (Fig. 2(d)). Similar

observations regarding the microstructure of the graphi- tized layer were made in our earlier study [lo] and show that, after nucleation, there is limited growth in the graphite particles which maintain a relatively small size (-4 nm). This suggests that when a thin graphitized tribolayer develops, the friction is reduced and no further transformation occurs in the DLC film. The observed low friction can be attributed to the low shear strength between the hexagonal planes in graphite. Under these conditions, the wear process can be envisioned as a low- rate continuous consumption and generation of the graphitized tribofilm.

Thus, the relatively higher coefficient of friction during testing at low velocity and loading condition (f=O. 18, Test 1) is attributed to friction between DLC, rather than graphite, and the ZrO, ball. The coefficient of friction in the rest of the tests reached a low steady- state value (fcO.08) due to the presence of graphite as confirmed by TEM. The higher wear rate of the ball material during Test 1 is consistent with the presence of DLC that has a higher hardness than graphite. Also, the absence of graphite from the wear track in the latter case shows that under a combination of low sliding velocity and loading level, the transformation require- ments for graphitization are not met even after a 10 km sliding distance. In an effort to explore this aspect a little further, an additional experiment was conducted under the same wear parameters for a sliding distance of 25 km. Again no steady-state friction was reached.

A comparison between Tests 2, 3 and 4 shows that both, sliding velocity and loading level have a significant influence on the graphitization process. Utilizing the sliding distance S to reach steady-state (graphitization) as a criterion and for the present experimental parame- ters, the results show that sliding velocity has a greater effect than level of loading. According to the wear- induced graphitization mechanism [lo], reduction in the friction coefficient f during wear involves first, the grad- ual releasing of H from sp3 domains and second, shear- ing of the weakened H depleted layer producing graphite. As has been shown earlier by Dischler [ 121, H begins to evolve from DLC at about 3OO”C, and signifi- cant H release occurs at about 450°C [ 131. Contact

Y. Liu et al./Surface and Coatings Technology 86-87 (1996) 564-568 561

Fig. 2. Bright field transmission electron micrographs showing (a) typi-

cal del bris morphology, (b) graphite particles in the microstructure of

debris (c) Electron diffraction pattern from debris showing mainly the

(00021 I graphite reflection. (d) Dark field image from debris showing

local a arrangement of nucleated graphite particles.

frequ .ency (sliding velocity) is expected to be more effect .ive in increasing temperature at asperity contacts than loading level and thus facilitating the H release

stage of the graphitization process. This is also consistent with early estimates of friction-induced temperature rise, where the temperature rise AT (flash temperatures in excess of 1000°C are predicted for the present system) at the contact region is proportional to sliding velocity and applied load but inversely proportional to the contact area [ 141. Since an increase in the applied load results in an analogous larger contact area, a greater influence on AT by sliding velocity is implied. On the other hand, low rotational frequency results in a long time interval between subsequent contact events, allow- ing heat transfer and thus, reducing asperity temper- ature. However, the present results suggest that under a sufficiently high applied load, graphitization can occur at a low sliding velocity but only after extensive sliding (Test 3).

In an effort to compare the transformation (graphit- ization) rate in the low friction tribolayer under the various loading and sliding velocity values, a parameter, R, was estimated and its value for the various wear tests is shown in Table 1. This parameter was simply calcu- lated from the time required to reach steady-state (graphitization) and is an indicator for the rate of formation of the low friction graphitized tribolayer. For example, a comparison in R between Tests 3 and 4 (a 16 times increase in velocity causes a 37 times increase in R) and Tests 2 and 4 (a 2.15 times increase in CJ” causes a 1.85 times increase in R) indicates again that sliding velocity more than likely exercises a greater effect than the applied load.

As mentioned earlier, the two primary requirements of the proposed wear-induced graphitization mechanism [lo] are H release caused mainly at “hot spots” by friction and shearing of the produced low-strength layer. The wear parameters closely relating to these two pro- cesses can be considered to be sliding velocity and loading level. Furthermore, the present results clearly suggest that these two parameters have a significant effect on the graphitization process. Thus, assuming that the graphitization rate (formation of a graphitized tribolayer) is mainly a function of these two parameters, a first attempt can be made to describe R as

R=Cv"d (1)

where v is the sliding velocity, 0 the applied stress, LY and p power coefficients and C a rate constant. Based on the present and previous experimental data (where steady-state values due to graphitization were also reached) [9] and in a first approximation taking the value of cH as the value of applied stress, the values of the coefficients involved in Eq. (1) were estimated using regression analysis. The analysis showed that C-4 54 x lo-’ so,3 m-1.3 MPa-0.87 ~(~1.3 and fi-0.87hGsing the wear parameters &ring Test 1 (0.06 m s-l and 395 MPa), Eq. ( 1) indicates that graph- itization under these conditions would have occurred

568 Y. Liu et al./Surface and Coatings Technology 86-87 (1996) 564-568

after a sliding distance of more than 28 km which is consistent with the experimental results. Additional tests are conducted at present in our laboratory to provide more validating data for the proposed equation that will allow a closer description of the rate of formation of the graphitized tribolayer that is responsible for the low friction of DLC films.

It is important to note, that the proposed wear- induced graphitization mechanism involving the forma- tion of a low friction graphitized tribolayer is consistent with the existing phenomenology of DLC tribology. For example, an important and controversial issue has been the effect of humidity on the DLC tribological response. DLC films are considered moisture sensitive and in general, their friction coefficient increases with increasing humidity. Recent tribological experiments by Jia et al. [ 151 showed that wear of DLC in humid air (40% humidity) is larger than that in dry air. Also, the coefficient of friction in humid air is initially higher than that in dry air but the steady-state friction coefficient is the same (-0.05) for both cases. These results are consistent with the wear-induced graphitization mecha- nism and can be attributed to the slower rate of forma- tion (smaller value of R) of the graphitized tribolayer in the presence of humidity.

4. Conclusions

The present results showed that the steady-state low coefficient of friction of DLC films was due to wear- induced graphitization, i.e., formation of a low friction graphitized tribolayer. Both, sliding velocity and applied load influence the graphitization process. For the present experimental range of wear parameters, sliding velocity was found to exercise a stronger effect than applied load. This can be attributed to the greater effect of contact frequency to the temperature rise at asperities facilitating hydrogen release from the DLC structure which is the first step of the wear-induced graphitization process. An equation was developed to describe the

wear-induced transformation rate of DLC into graphite and formation of a low friction tribolayer by taking into consideration sliding velocity and applied load effects.

Acknowledgement

Funding for this work was provided by the Louisiana Educational Support Fund under contract LEQSF (94-96)-RD-B07 and the US Department of Energy.

References

[l] V. Palshin, E.I. Meletis, S. Ves and S. Logothetidis, Thin Solid Films, 270 (1995) 165.

[2] H. Tsai and D.B. Bogy, J. tic. Sci. Technol., A 5 (1987) 3287. [3] S. Aisenbcrg and R. Chabot, J. Appl. Phys., 42 (1971) 2953. [4] J.C. Angus, P. Koidl and S. Don&z, in J. Mort and F. Jansen

(eds.), Plasma Deposited Thin Films, CRC Press, Boca Raton, FL, 1986, p. 89.

[5] K. Enke, H. Dimigen and H. Hubsch, Appl. Phys. Left., 36 (1980) 291.

[6] A. Grill, B. Meyerson, V. Patel, J.A. Reimer and M.A. Petrich, J. Appl. Phys., 61 (1987) 2874.

[7] A. Grill, B.S. Meyerson and V. Patel, in A. Feldman and S. Holly (eds.), Diamond Optics, Proc. SPIE, Vol. 969, Belingham, WA, 1989, p. 52.

[8] E.I. Meletis, A. Erdemir and G.R. Fenske, Surf: Coat. Technol., 73 (1995) 39.

[9] Y. Liu, A. Erdemir and E.I. Meletis, Surf Coat. Technol., 1996, in press.

[lo] Y. Liu and E.I. Meletis, J. Mater. Sci. [ 1 l] L.C. Nistor, J. Van Landuyt, V.G. Ralchenko, T.V. Kononenko,

E.D. Obraztsova and V.E. Strelmtsky, J. Appl. Phys., A 58 (1994) 137.

[ 121 B. Dischler, in P. Koidl and P. Oelhafen (eds.), Amorphous Hydro- genated Carbon Films, Proc. Eur. Mater. Res. Sot. 17, Les Edi- tions de Physique, Paris, 1987, p. 189.

[ 131 A.R. Nyaiesh and W.B. Nowak, .I. Vat. Sci. Technol., A I (1983) 308.

[14] E. Rabinowicz, Friction and Wear of Materials, Wiley, New York, 1965.

[ 151 K. Jia, Y.Q. Li, T.E. Fischer and B. Gallois, J. Mater. Res., 10 (1995) 1403.