Thin Solid Films 519 (2011) 2043–2048

Contents lists available at ScienceDirect

Thin Solid Films

j ourna l homepage: www.e lsev ie r.com/ locate / ts f

Effects of duty cycle andwater immersion on the composition and friction performanceof diamond-like carbon films prepared by the pulsed-DC plasma technique

F. Zhao a,b, H.X. Li a, L. Ji a, Y.F. Mo c, W.L. Quan a,b, Y.J. Wang a,b, J.M. Chen a,⁎, H.D. Zhou a

a State Key Laboratory of Solid Lubrication, Lanzhou Institute of Chemical Physics, Chinese Academy of Sciences, Lanzhou 730000, PR Chinab Graduate University of Chinese Academy of Sciences, Beijing 100049, PR Chinac School of Chemistry and Chemical Engineering, Guangxi University, Nanning 530004, PR China

⁎ Corresponding author. Tel.: +86 931 4968018; fax:E-mail address: chenjm@licp.cas.cn (J.M. Chen).

0040-6090/$ – see front matter © 2010 Elsevier B.V. Aldoi:10.1016/j.tsf.2010.10.042

a b s t r a c t

a r t i c l e i n f o

Article history:Received 5 December 2009Received in revised form 20 October 2010Accepted 27 October 2010Available online 3 November 2010

Keywords:Diamond-like carbonPulsed DC PlasmaRaman SpectroscopyInfrared SpectroscopyFrictionX-ray photoelectron Spectroscopy

Diamond-like carbon (DLC) films were prepared in a pulsed-DC discharged CH4/Ar plasma. Effects of duty cycle([ton/(ton+ toff)]×100%) on the composition and properties of DLC films were investigated. In general, theincreased duty cycle led to an up-shift of the G peak position, an increase in the ID/IG and sp2/sp3 ratio, and areduction of the number of C–Hbonds and thefilmhardness, revealing a graphitization tendencywith increasingduty cycle. Tribologically, ultralow and steady friction coefficients (0.005 and 0.008) in dry nitrogen atmospherewere obtained for the films prepared under a duty cycle of 50% and 65%. The unique mechanical property andchemical nature brought by the moderate sp2/sp3 ratio and the proper H content were considered to beresponsible as the films deposited in this duty cycle range could simultaneously provide the high chemicalinertness and the ultrasmooth sliding surfaces required for achieving ultralow friction. In addition, the structurewas less vulnerable to watermolecules in the case of stewing. The diamond-like nature and the ultralow frictionperformance were hardly affected even experiencing a 4-month immersion in water.

+86 931 8277088.

l rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Diamond-like carbon (DLC) films prepared from hydrogen-rich CH4

source gas plasmas have been reported to exhibit an ultralow frictioncoefficient of 0.001–0.003 and awear rate of 10−9 to 10−10 mm3/Nm invacuum or dry nitrogen environments [1]. Excellent tribologicalperformance was also shown by our hydrogenated DLC films preparedby the pulse bias-assisted radiofrequency plasma enhanced chemicalvapor deposition (RF-PECVD) method. They displayed an ultralowfriction coefficient of~0.006 in both dry N2 and CO2 atmospheres [2].Nevertheless, from the perspective of industrial application, the RF-PECVD method may not be a candidate as competitive as the pulseddirect current (DC) plasma technique. Unlike the RF technique, thepulsed-DC technique could use higher power levels (while simulta-neously keeping the average power in an usual range [3]) and does notrequire matching networks, thus leading to an easy implementation inindustrial plasma processing systems and a reduction of the productioncost [4–7]. In addition, the pulsed-DC deposition method has been usedto deposit DLC films on surfaces with special geometry, e.g., to coat DLCinside the inner wall surface of model dies with holes of 2 and 0.9 mmin diameter and 20 mm in depth [8]. Moreover, DLC films producedfrom the pulsed-DC PECVD possessed improved mechanical andtribological properties (high adherence andwear resistance, low stress,

low roughness, and low friction coefficient), which made them analternative for use as protective coatings on magnetic storage devicesand sliding surfaces [5].

Among others, one key parameter in the pulsed-DC depositionprocess is the duty cycle (here defined as ([(pulse-on time)/(pulse-ontime+pulse-off time)]×100%), or ([(ton/(ton+ toff)]×100%)). It is asignificant and effective factor in controlling the composition (sp3/sp2

ratio) and properties (especially the hardness and stress) of the DLCfilms. After systematically investigating the effect of pulse parameterson the properties of DLC films, Kumar et al. [9] found that DLC filmswith both low stress and high hardness could be obtained and threefactors were responsible: (i) relaxation of adions/adatoms, (ii) controlof substrate temperature, and (iii) creation of a hard/soft multilayerstructure [9]. All these three factors intimately correlated with thecooling and relaxing effect during the pulse-off period of the pulseplasma discharge (but this effect was absent in the continuousdischarge techniques). Similarly, Anders [3] proposed that the pulsedprocessing was able to produce amorphous carbon film with high sp3

content by using a suitably low duty cycle since it could keep a smallaverage heat load on the substrate surface and, accordingly, controlthe substrate heating [10] without affecting the kinetic energy of ions.Effects of the duty cycle on the film structure and properties were alsonoted in our previous work [11]. It was found that the low pulse dutycycle promoted the formation of fullerene-like structure in thehydrogenated carbon films produced by the pulse bias-assistedRF-PECVD method. In that case, however, the role of the duty cyclewas not fully reflected due to the coexistence of the RF power. Hence,

Table 1Deposition parameters for DLC films.

Parameter Value

Working pressure, Pa 1.5Mass flow rate of CH4, sccm 200Mass flow rate of Ar, sccm 75Pulse frequency, kHz 40Pulse voltage, V −1000DC bias, V −200Duration, min 60Duty cycle, % 20; 35; 50; 65; 80 (film-1 to film-5)

2044 F. Zhao et al. / Thin Solid Films 519 (2011) 2043–2048

in present work, we synthesized DLC films by using the pulsed-DCmethod alone and investigated the effect of the duty cycle on the filmcomposition and friction performance in dry nitrogen atmosphere.In addition, in view of the significant effect of water on thehydrogenated structure and friction performance of DLC films [12],4-month immersion of the sample (with the best lubricationperformance) in water was performed, followed by the Raman andfriction characterization.

2. Experimental details

DLC films were deposited on polished stainless steel (1Cr18Mo8-Ni5N) substrates and n-type Si (1 0 0) wafers by the pulsed-DC PECVDmethod. The schematic illustration of the deposition system has beengiven previously [13]. Driven by a 40-kHz mid-frequency unipolarpulsed-DC power, the cathode (substrate holder) was negativelycharged to generate plasma. A schematic diagram of the rectangularwaveform of the pulsed-DC power supply is shown in Fig. 1. The dutycycle (ton/(ton+ toff)×100%) could be adjusted according to theprocess requirement. A mixture of methane and argon was used asthe source gas, and the pressure was controlled to be constant at1.5 Pa during deposition. The deposition duration was 60 min, and thethickness for all samples was measured to be 485±10 nm. Detaileddeposition parameters are described in Table 1. Prior to deposition,substrates were firstly etched with Ar+ plasma at a bias voltage of−1000 V for 20 min in order to remove the native oxide layer on thesample surface. Then, a Si adhesive layer of about 200-nm thickness,with a graded intermixed Si–Fe interface, was deposited in order toenhance the film-to-substrate adhesion strength. The detailed Si-interlayer deposition process can be found in Ref. [14].

All samples were analyzed using a Bruker IFS 66v/S Fouriertransform infrared (FT-IR) spectrometer. The transmission spectrawere taken between 400 and 4000 cm−1 with a resolution of0.23 cm−1. A LabRAM HR800 (HORIBA Jobin Yvon, France) micro-Raman spectrometer operating with 532 nm Ar+ laser as theexcitation source was used to characterize the film structure. Thelaser beam was focused onto the sample surface using an opticalmicroscope with a magnification of 100×(laser spot size~1 μm). Theacquired Raman spectra were fitted based on Gaussian curve shapeswith the curve-fitting software. A PHI-5702 X-ray photoelectronspectroscope (XPS) operating with monochromated Al-Ka irradiationat a pass energy of 29.35 eV was employed to analyze the chemicalcomposition and chemical bond states of the film surfaces, with thebinding energy of Au (Au4f: 84.0 eV) as the reference.

The film hardness and elastic modulus values were determinedusing an Nano-Hardness Tester (MTS Nano Indenter XP), where themaximum indent depth was controlled to be 45 nm (less than 1/10 offilm thickness) so as to minimize the effect of the substrate [15], andfive repeated indentations were made for each sample. Friction andwear properties of the DLC films were evaluated on a reciprocating-type ball-on-flat CSM tribometer, which was equipped with achamber where the relative humidity (RH) and gaseous environmentcould be controlled. Sliding tests were performed in dry nitrogen

Fig. 1. Schematic diagram of waveform of pulse power supply. Duty cycle=[ton/(ton+ toff)]×100%.

environment (RHb5%) at an average sliding velocity of 80 mm/s atroom temperature. DLC-coated steel balls (12.0 mm in diameter)were used as the counterparts. Tap water was used to immerse theselected DLC film.

3. Results and discussion

Fig. 2 gives the FT-IR spectra of five samples. The strong peakscentered at ~2851 cm−1 and ~2923 cm−1 correspond to the sym-metric and asymmetric stretching mode of sp3 CH2, respectively. Theweak peaks present at ~2874 cm−1 and ~2954 cm−1 are assigned tothe symmetric and asymmetric stretching mode of sp3 CH3,respectively. As seen, the peak intensity generally decreases withincreasing duty cycle, suggesting a reduction in the number of C–Hbonds. Variation of the H content could also be reflected by the Ramanspectra since high amounts of H usually lead to a high sp3 content anda down-shift of the G peak [16], as discussed below.

Raman spectroscopy is a standard non-destructive technique forthe characterization of carbon-based materials, and the characteristicRaman spectra can be used to study the structural arrangements of thecarbon atoms [17]. Visible Raman spectra of DLC films are dominatedby scattering of the sp2 sites, since the Raman cross section of sp2 sitesis 50–230 times larger than that of the sp3 sites [16,18]. Usually, theRaman spectra of DLC films are characterized by a G peak around1550 cm−1 and aD shoulder around 1360 cm−1 [17]. TheGpeak is thestretching vibration of sp2 sites in both rings and chains, while the Dpeak is due to the breathing mode of those sp2 sites only in aromaticrings [16]. Although the visible Raman spectra depend formally on theconfiguration of the sp2 sites and only indirectly on the sp3 content,they can still be used to derive the information about the sp2/sp3 ratiofrom ID/IG (intensity ratio of the D peak to the G peak) [16–18].Generally, if the FWHM (full width half maximum) of the G peakexceeds 50 cm−1, the in-plane correlation length La (or the diameter ofgraphitic cluster) would be below 1 nm. And for La below 2 nm, thefollowing relationship could be used: ID/IG=cLa

2, where c is a constant

Fig. 2. FT-IR spectra of DLC films.

Fig. 3. (a) Raman spectra of DLC films. (b) Typical peak deconvolution of Ramanspectrum. (c) ID/IG ratio as a function of duty cycle.

Fig. 4. (a) C1s XPS spectra and (b) fitting parameters [sp2(C=C) fraction and sp3(C–C+C–H) fraction] of DLC films as a function of duty cycle. The inserted picture in panel (b)gives the typical fitting curves of C1s XPS spectrum.

2045F. Zhao et al. / Thin Solid Films 519 (2011) 2043–2048

[16,18]. Raman spectra of the five samples are shown in Fig. 3(a). It canbe seen that all films take on typical diamond-like characteristic,with the D peak and the G peak present at ~1350 cm−1 and 1553–1564 cm−1, respectively. In addition, the Raman scattering intensityincreases with increasing duty cycle (both the D and the G peakbecome more pronounced). Another phenomenon worth noting isthat the G peak maximum shifts from ~1553 cm−1 to ~1564 cm−1

with increasing duty cycle. The simultaneous enhancement of the Dpeak intensity and the up-shift of the G peak position with increasingduty cycle indicate a tendency towards graphitic clustering, asillustrated in Ref. [16]. The graphitization trend is further supportedby the increased ID/IG with increasing duty cycle, Fig. 3(b) and (c),since the ID/IG ratio is proportional to the size of graphitic cluster. Thus,

it could be inferred from the Raman spectra that increased duty cyclepromoted the formation and growth of graphitic clusters in carbonmatrix.

This deduction can be further supported by the C1s XPS spectracomparisonand the peak-fitting result in Fig. 4. XPS, a usefulmethod toassess the sp3 and sp2 components, has been considered to be veryuseful in supplementing the optical characterization (Raman) ofamorphous carbon films [19]. The C1s peak position in graphite anddiamond is 284.15 eV and 285.50 eV, respectively. Thus, the differentC1s core-level binding energies of the sp2 and sp3 hybrids, and thedifferent sp2/sp3 ratio, in DLC would make the C1s peak position varyfrom one sample to another. As shown in Fig. 4(a), the C1s peakposition shifts from ~284.35 eV to ~284.14 eV with increasing dutycycle, indicating an enhanced sp2 contribution. To further reveal thesp2/sp3 ratio in different samples, peak fitting was conducted and partof the result is given in Fig. 4(b). As shown in the insert picture, thedeconvoluted subpeaks at 284.2±0.1 eV, 285.1±0.1 eV, 286.5±0.1 eV, and 288.5 ± 0.1 eV are assigned to C=C, C–C/C–H, C–O, andO=C–O, respectively [20,21]. Clearly, most of the carbon atoms aresp2-hybridized and the sp2 (C=C) fraction increases while the sp3

(C–H+C–C) fraction decreases with increasing duty cycle. This resultis consistent with the Raman spectra analysis and confirms thepromoted graphitization with increasing duty cycle.

Three factors may account for the graphitization trend. Firstly, thebombardment of the substrates by the energetic ions became morecontinuous with increasing duty cycle and would accelerate thesubstrate heating, especially in such a high pulse voltage of −1000 Vcombined with a DC bias of −200 V. The heating effect was able topromote the graphitization process, as occurred in the annealed DLC

Fig. 6. Friction behavior of different DLC films sliding against self-coated steel balls indry nitrogen. The inserted picture is given for eye guidance.

2046 F. Zhao et al. / Thin Solid Films 519 (2011) 2043–2048

films [22,23]. Secondly, high duty cycle might lead to a high degree ofdissociation of methane and hydrocarbon fragments with less boundH would participate in the film growth. It was also possible that amore energetic ion bombardment led to the preferential hydrogensputtering because of the weaker strength of C–H bonds than the C–Cbonds [7]. As a result, the H incorporation effect was reduced and thegrowth of graphitic clusters was favored. Otherwise, the DLC filmwould contain a significant content of H and hence lead to a high sp3

content and produce very small graphitic clusters [16]. Thirdly, underthe high accelerating voltage (−1000 V for pulse and−200 V for DC),there was still a likelihood that the thermodynamically metastablesp3–C hybrid state converted into stable sp2–C hybrid state after beingbombarded by the Ar+ ions with high dynamic energy and theconsequent relaxation during pulse-off time.

Fig. 5 illustrates the variation of the nanohardness as a function ofthe duty cycle. Duty cycle has been reported to strongly affect thehardness and sp3 content of the DLC coatings (the sp3 fraction andhardness usually increased with increasing duty cycle) [9,24].However, in present case, the hardness generally decreases withincreasing duty cycle (15.9±0.1 GPa to 14.1±0.1 GPa). Superficially,this result appears to be in contrast with previous studies [9,24].Actually, however, it is well consistent with them if the factorscontributing to the film hardness are considered. Generally, themechanical hardness of carbon film is directly proportional to its sp3

fraction [16,25]. Nevertheless, in the present case, the increased dutycycle led to an increase in the sp2 fraction and the growth of graphiticclusters, rather than an enhancement of the sp3 fraction, thusresulting in a general reduction in hardness. However, it is worthnoting in Fig. 5 that the hardness of film-1 (20 %) is aberrantly lowerregardless of its highest sp3 fraction. This is probably due to the highercontent of incorporated H, as shown in Fig. 2. Because although theincorporated H atoms were able to increase the sp3–C content byconverting the sp2–C sites (C=C bonds) into sp3=CH2 and≡CH sites[16], these generated sp3 C–H bonds made no contribution to thehardness or elastic modulus [25] (only those C–C bonds contributedsignificantly to hardness [25,26] but their number did not increase atall). On the contrary, the numerous H atoms rendered a developmentof the polymer-like chains [16,27] of which the H-terminated C–Hbonds would break the interlinked carbon network, reduce the filmcohesive strength, and accordingly, lower the film hardness [28].

Fig. 6 summarizes the friction behavior of different films slidingagainst the self-coated steel balls in dry nitrogen environment.Superlow friction coefficients approaching 0.001 had been reportedwhen the a-C:H slid against itself in dry nitrogen environment [1,29].The superlubricious performance was attributed to the partially di-hydrated sliding DLC surfaces because the abundant H atoms on thetop surfaces were able to saturate the dangling σ-bonds, prevent theπ–π⁎ interaction and provide better surface passivation [1,29].

Fig. 5. Variation of nanohardness with duty cycle.

According to this model, the friction coefficient should increase withincreasing duty cycle since the content of the di-hydrated groups (sp3–CH2) decreaseswith increasingduty cycle, as revealedby the IR spectrain Fig. 2. However, Fig. 6 displays a quite different variation trend. Asseen, the friction coefficient first descends from 0.016 to 0.005 whenthe duty cycle increases from 20% to 50% and then ascends back to0.016 as the duty cycle continually increases to 80%. The appearance ofsuch an irregular variation trend was not occasional. It is the typicalfeature of DLC films. Since the friction behavior of DLC is stronglyaffected by many factors and the controlling factors may changesignificantly from one type of DLC to another [30]. Generally, thesefactors can be divided into two categories: intrinsic and extrinsic. (Theintrinsic or film-specific factors include the degree of sp2 versus sp3

bonding, the relative amounts of hydrogen in the structure or on thesliding surfaces, film thickness, surface roughness, and a range ofmechanical properties (including hardness, elastic modulus, andviscoelasticity) [30]. The extrinsic or tribotest-condition-specificfactors involve the nature of the substrate and counterface materials,the contact pressure, the nature of motion, the speed, the chemicalnature of the test environment, etc. [1]). The key requirements forachieving ultra- to superlow friction in DLC films are given in Ref. [30].Generally, apart from the high degree of chemical inertness, a smoothsurface finish is also preferred for achieving low friction [30]. Thechemical inertness would not function without a smooth slidingsurface. This may account for the abnormally high friction coefficientof thefilms deposited at lower duty cycle, especially forfilm-1, becausethe films with higher H content were more likely to destruct the filmcontinuity and lead to serious delamination at the initial stage of thesliding tests, as observed previously [31]. Then extra energy would beconsumed to press the delaminated spalls into fine particles, thusincreasing the friction coefficient. In addition, the plowing effectcaused by the fine particles trapped in the wear track might also beresponsible. Similarly, debris could be generated and trapped aswell inthe contact area of the films prepared at a high duty cycle due to theirlower hardness and load-carrying ability. Actually, a large amount ofdebris has been observed aside thewear tracks of bothfilm-1 and film-5. In contrast, almost no debris could be detected inside or outside thewear track of film-3 and the sliding surface was so smooth that noasperity higher than 30 nmcould be detected. Thismay result from thesynergic effect of many factors, including the continuous interlinkingstructure, the high cohesive strength, the high enough load-carryingability, the proper plastic deformation ability, and the well-passivatedsliding surfaces (by both H [1] and N2 [2]) brought by the propercontent of H and the moderate sp2/sp3 ratio at a moderate duty cycle.Other factors including the practical adhesion strength and thethermal stability may be also responsible and further investigation isstill needed.

Fig. 7. (a) Raman spectra comparison between water-immersed and non-immersedDLC films. Friction behavior comparison between the two samples in dry nitrogenunder a normal load of (b) 5 N and (c) 10 N.

2047F. Zhao et al. / Thin Solid Films 519 (2011) 2043–2048

To examine the effect of water molecules on the structure andlubrication performance of the DLC films prepared by the pulsed-DCtechnology, we immersed the film-3 (with the best friction perfor-mance) in water for 4 months. It was found that even experiencing along-term immersion, the film was still strongly adhesive to the steelsubstrate and no peeling-off could be clearly observed. Fig. 7 showsthe comparison of the Raman spectra and the friction behaviorbetween the water-immersed DLC and the non-immersed one (i.e.,exposed to air rather than immersed in water). As shown in Fig. 7 (a),the water-immersed film not only retained the typical diamond-likecharacteristic but also displayed a higher Raman scattering intensitythan the air-exposed one. It seems as if the water immersion

treatment delayed the air-exposure-induced structural deteriorationof the DLC films. Moreover, the water-immersed film gave lowerfriction coefficients under a normal load of both 5 N and 10 N, Fig. 7(b)and (c). In particular, the steady-state friction coefficient of the water-immersed film was still able to reach an ultralow value of 0.005 andendure 30,000 laps without failure under the identical test conditionin Fig. 6. Hence, it can be deduced that the water immersion behaviorhardly affected the structure, the film-to-substrate adhesion, or theultralow friction behavior of the DLC films prepared by the pulsed-DCPECVD technique.

4. Conclusions

A series of DLC films were prepared in the pulsed-DC dischargedCH4/Ar plasma with the duty cycle varying from 20% to 80%, and thencharacterized by IR, Raman, XPS, nanoindentation, and friction tests. Itwas found that the duty cycle strongly affected the composition andstructure of the DLC films. Generally, increased duty cycle led to anup-shift of the G peak position, an increase in the ID/IG and sp2/sp3

ratio, a reduction in the number of C–H bonds, and a deterioration ofthe film hardness. All these results consistently indicated a graphi-tization trend (or the growth of graphitic clusters) with increasingduty cycle. Tribologically, the friction coefficient first descended andthen ascended with increasing duty cycle. The worse frictionperformance at either high duty cycle or low duty cycle was attributedto the lack of ultrasmooth surfaces during sliding tests caused by thelow adhesion/cohesion strength or the poor load-carrying ability. Incontrast, the ultralow friction performance (0.005 to 0.008) presentedby the films prepared at the duty cycle of 50% to 65% was more likelydue to the coexisted ultrasmooth sliding surfaces and high chemicalinertness. Further investigation revealed that even if the film with thebest friction performance was immersed in water for 4 months, itsdiamond-like nature, film-to-substrate adhesion, and ultralow frictionperformance in dry nitrogen were hardly affected.

Acknowledgments

The authors thank the National Natural Science Foundation ofChina (grant no. 50705093 and no. 50575217), the Innovative GroupFoundation from NSFC (grant no. 50421502), and the National 973Project (no. 2007 CB607601) for the financial support.

References

[1] A. Erdemir, C. Donnet, J. Phys. D Appl. Phys. 39 (2006) 311.[2] L. Ji, H. Li, F. Zhao, W. Quan, J. Chen, H. Zhou, J. Phys. D Appl. Phys. 42 (2009)

135301.[3] A. Anders, Surf. Coat. Technol. 183 (2004) 301.[4] J.L. Andujar, M. Vives, C. Corbella, E. Bertran, Diamond Relat. Mater. 12 (2003) 98.[5] C. Corbella, E. Pascual, M.A. Gomez, M.C. Polo, J. Garcia-Cespedes, J.L. Andujar, E.

Bertran, Diamond Relat. Mater. 14 (2005) 1062.[6] C. Corbella, M. Vives, G. Oncins, C. Canal, J.L. Andujar, E. Bertran, Diamond Relat.

Mater. 13 (2004) 1494.[7] V.J. Trava-Airoldi, L.F. Bonetti, G. Capote, J.A. Fernandes, E. Blando, R. Hubler, P.A.

Radi, L.V. Santos, E.J. Corat, Thin Solid Films 516 (2007) 272.[8] X.D. Yang, T. Saito, Y. Nakamura, Y. Kondo, N. Ohtake, Diamond Relat. Mater. 13

(2004) 1984.[9] S. Kumar, P.N. Dixit, D. Sarangi, R. Bhattacharyya, J. Appl. Phys. 85 (1999) 3866.

[10] C.V. Budtz-Jørgensen, J. Bøttiger, P. Kringhøj, Surf. Coat. Technol. 137 (2001) 104.[11] L. Ji, H. Li, F. Zhao, W. Quan, J. Chen, H. Zhou, J. Appl. Phys. 105 (2009) 106113.[12] K. Holmberg, H. Ronkainen, A. Matthews, Ceram. Int. 26 (2000) 787.[13] L. Ji, H. Li, F. Zhao, J. Chen, H. Zhou, Diamond Relat. Mater. 17 (2008) 1949.[14] L. Ji, H. Li, F. Zhao, J. Chen, H. Zhou, Key Eng. Mater. 373–374 (2008) 151.[15] J. Robertson, Diamond Relat. Mater. 1 (1992) 397.[16] J. Robertson, Mater. Sci. Eng., R 37 (2002) 129.[17] G. Irmer, A. Dorner-Reisel, Adv. Eng. Mater. 7 (2005) 694.[18] A.C. Ferrari, J. Robertson, Phys. Rev. B 61 (2000) 14095.[19] P.K. Chu, L. Li, Mater. Chem. Phys. 96 (2006) 253.[20] G.J. Wan, P. Yang, R.K.Y. Fu, Y.F. Mei, T. Qiu, S.C.H. Kwok, J.P.Y. Ho, N. Huang, X.L.

Wu, P.K. Chu, Diamond Relat. Mater. 15 (2006) 1276.[21] L. Zajíková, K. Veltruská, N. Tsud, D. Franta, Vacuum 61 (2001) 269.[22] S. Kumar, Appl. Phys. Lett. 58 (1991) 1836.

2048 F. Zhao et al. / Thin Solid Films 519 (2011) 2043–2048

[23] M. Chhowalla, A.C. Ferrari, J. Robertson, G.A.J. Amaratunga, Appl. Phys. Lett. 76(2000) 1419.

[24] J.-B. Wu, J.-J. Chang, M.-Y. Li, M.-S. Leu, A.-K. Li, Thin Solid Films 516 (2007) 243.[25] J. Robertson, Tribol. Int. 36 (2003) 405.[26] S. Neuville, A. Matthews, Thin Solid Films 515 (2007) 6619.[27] J.F. Zhao, P. Lemoine, Z.H. Liu, J.P. Quinn, P. Maguire, J.A. McLaughlin, Diamond

Relat. Mater. 10 (2001) 1070.

[28] S. NEUVILLE, JP J. Solids Struct. 3 (2009) 33.[29] A. Erdemir, Tribol. Int. 37 (2004) 577.[30] A. Erdemir, J. Fontaine, C. Donnet, in: C. Donnet, A. Erdemir (Eds.), Tribology of

Diamond-Like Carbon Films: Fundamentals and Applications, Springer, New York,2008, p. 237.

[31] F. Zhao, H.X. Li, L. Ji, Y.F. Mo, W.L. Quan, H.D. Zhou, J.M. Chen, J. Phys. D Appl. Phys.42 (2009) 165407.