You are on page 1of 6

Thin Solid Films 446 (2004) 8590

Mechanical properties evaluation of fluor-doped diamond-like carbon coatings by nanoindentation


S.C. Trippea, R.D. Mansanoa, F.M. Costab, R.F. Silvac,*
a

Integrated System Laboratory, Electronics Systems Department, Polytechnic School University of Sao Paulo, Av. Prof. Luciano Gualberto, trav.3, 158, Sao Paulo City, SP 05508-900, Brazil b Physics Department, University of Aveiro, Aveiro 3810-193, Portugal c Glass and Ceramic Engineering Department, CICECO, University of Aveiro, Aveiro 3810-193, Portugal Received 28 February 2003; received in revised form 13 August 2003; accepted 28 August 2003

Abstract Fluor-doped diamond-like carbon (DLC) films were grown by RF sputtering, varying the CF4 partial pressure and the total gas pressure. Nanoindentation is required to evaluate the mechanical properties of such very thin films. The Vickers nanohardness and the Youngs modulus values fall in the 5570 GPa and 235326 GPa ranges, respectively, the lowest values being observed for the highest degree of fluorine content in the film (80% CF4 in the gas mixture) and for the highest processing pressure (1.5 kPa). 2003 Elsevier B.V. All rights reserved.
Keywords: Diamond-like carbon; Hardness; Elastic properties; Fluorine

1. Introduction Diamond-like carbon (DLC) is the given name of many materials that have amorphous carbon in its structure. This coating material contains carbon bonds, mainly sp3 and sp2 hybridizations, and can incorporate hydrogen in its structure. These recombine in order to minimize the total energy, making the DLC properties close to those presented by diamond w1,2x. The DLC presents many technological applications due to a vast range of unique characteristics. Among these, the transparency at visible and infrared region makes optical antireflective applications possible w2x. It is also used as a protective layer on hard disks and magnetic tapes once it is very hard w1,2x. By including fluor in DLC films, changes occur on the materials characteristics. By varying the fluor concentration in the film, the electrical, optical and mechanical characteristics can be tailored. In fact, the amorphous carbon fluorinated materials have as main characteristic a low value of dielectric constant
*Corresponding author. Tel.: q351-2344-370-243; fax: q351-234425-300. E-mail address: rsilva@cv.ua.pt (R.F. Silva). 0040-6090/04/$ - see front matter doi:10.1016/j.tsf.2003.08.069

(approx. 2.0) w35x, giving place to applications like passivation layers, dielectric gates and as an isolation material for integration into ULSI (Ultra Large Scale Integrated) devices, allowing circuits better efficiency w3,68x. The thermal stability up to 400 8C of F-doped DLC films is improved when compared to undoped ones w9x. In what regards to gains in the optical characteristics, F-doping increases the refractive index and the optical band gap w9x. This, together with the high abrasion resistance, makes its application as anti-reflection coating for infrared optics possible, as a result of the wider absorption band of CFx up to 9 mm compared to ones of CHn in the range of 33.5 mm w10x. Fluor incorporation in DLC is also advantageous as a fluorinated lubricant layer to reduce film wear of computer hard disks w3,10x. The mechanical properties of DLC films are determinant for the above-referred applications. The superior hardness of DLC coatings results not only from the presence of sp3 bonds but also depends on the films structure, namely the number and size of graphite clusters and the imperfections and absorbed gas occurrence at the cluster grain boundaries w11x. Such variety of the DLC structure characteristics, added to the wide

2003 Elsevier B.V. All rights reserved.

86

S.C. Trippe et al. / Thin Solid Films 446 (2004) 8590

range of hydrogen content in the films, leads to a very broad range of hardness and Youngs modulus values. Moreover, the dependence of these properties on the applied load used for their determination also contributes to this scattering. Indentation is the most used technique for such evaluation in thin films, but it forces to take much care due to the very small film thickness and the consequent influence of the substrate. This imposes the use of nanoindentation. For DLC hard films an additional care in the interpretation of the results must be take into account due to cracking events and delamination phenomena w12x. In the recent review of J. Robertson w1x, one can find update values for DLC mechanical properties depending on the film processing technique and composition. The maximum nanohardness and Youngs modulus of the a-C:H films are 17 GPa and 300 GPa, respectively, while for ta-C (8590% sp3) these properties can attain 88 GPa and 822.9 GPa, respectively. As far as the fluorinated DLC coatings are concerned, their mechanical characterization by nanoindentation has never been evaluated. F.L. Freire Jr. et al. w3x studied the dependence of hardness on the PECVD self-bias and CF4 partial pressure, but they used microhardness (5 gf of applied load) resulting in underestimated hardness values, in the range of 5.515 GPa, due to the influence of the substrate. Thus, in the present work, nanoindentation technique was used to estimate the mechanical properties, namely hardness and Youngs modulus, of thin DLC fluorinated films grown by CF4 yCH4 assisted reactive sputtering method. The CF4 concentration on the gas mixture varied from 0 to 80% at three distinct process pressures of 0.5 kPa, 1.0 kPa and 1.5 kPa. The F-doped DLC mechanical properties were correlated with the structural characteristics of the films that were assessed by FTIR and Raman spectroscopies. 2. Experimental The equipment used for the growth of the F-doped DLC films is a RF magnetron sputtering system home made at LSI, University of Sao Paulo, Brazil, as described elsewhere w13x. A rotaryyturbo-molecular vacuum pump system allowed a working pressure of 9.3=10y5 kPa. The RF (13.56 MHz) power was fixed at 150 W. The sputtering process was done with a 99.9999% pure graphite target, located at 10 mm of the Si p-type {100} substrate wafers (7.62 cm diameter and 381"50 mm thickness), with an electric resistivity of 10 V cm. Before the deposition, the substrates were submitted to a Piranha clean, followed by diluted HF dip before film deposition w13x. The films were produced with three different process pressures (0.5 kPa, 1.0 kPa and 1.5 kPa). A CF4qCH4 gas mixture was used during the sputtering process and the CF4 flow varied from 0 to 90% of the total volume flow.

The infrared absorption spectra were acquired in nitrogen ambient using a Digilab FTS-40 FTIR spectrometer. The CFx vibration modes peaks were identified in the range 10001700 cmy1 w14x. The associated absorption bands were the following: 1100 cmy1 -CF2 symmetric mode, 1200 cmy1 CF2 asymmetric mode, 1340 cmy1 C-F stretch mode and 17001800 cmy1 F2CsC group w8x. Micro-Raman spectroscopy was carried out using a Jobin Yvon T64000 with three monocromators, the two first ones in subtraction mode. The measurements were performed in air using a line laser 514.4 nm. The detected bands were the G band, centered at approximately 1550 cmy1 and associated with vibration mode sp2, and the D band, centered at approximately 1350 cmy1 resulting from the presence of ordered graphite clusters w15,16x. Nanoindentation experiments were used to evaluate the hardness and the Youngs modulus of the F-doped DLC films. These experiments were carried out on a dynamic ultra-microhardness tester (Shimadzu DUH200) with Vickers diamond indenter. Loadingyunloading cycles with holding times at maximum load of 1 s were performed to evaluate the hardness from the penetration depth. A very small indentation load of 5 mN was applied in order to avoid the influence of the substrate deformation on the films mechanical characterization. Taking into account the film thickness (t) and the maximum penetration depth (hmax), a load of 5 mN guaranteed a hmax yt ratio in the range 0.080.20 below the critical value of 0.22 recently appraised for DLC coatings on steel w12x. The load (P) vs. displacement (h) data were analysed using the consistent method of Oliver and Pharr w17x. The first third portion of the unloading curves was fitted to a power-law to calculate the contact stiffness (dPydh) at the peak load (Pmax), which corresponds to the maximum displacement of the indenter (hmax). The residual depth impression value (hP) was extrapolated at zero load from dPydh. The average of exponents of power law relationship between P and h took the value of 1.93"0.14, which is comparable to ceramics materials w17x. The Vickers hardness (H) and the Youngs modulus (E) were calculated using the following equations: Hs Pmax 26.43h2 P 1 2hp p y 24.5 dP dh (1)

Ers

(2)

2 1 1yn2 1ynD s q Er E ED

(3)

where Er is the reduced modulus of elasticity of the

S.C. Trippe et al. / Thin Solid Films 446 (2004) 8590 Table 1 Processing conditions and thickness of the F-doped DLC samples. Sample CF0-P5 CF60-P5 CF80-P5 CF0-P10 CF30-P10 CF50-P10 CF70-P10 CF0-P15 CF10-P15 CF30-P15 CF70-P15 CF4 (vol.%) 0 60 80 0 30 50 70 0 10 30 70 Total pressure (kPa) 0.5 0.5 0.5 1.0 1.0 1.0 1.0 1.5 1.5 1.5 1.5 Thickness (A) 3289"30 2538"50 3038"50 8294"13 8729"11 2075"25 16731"73 13097"200 13022"140 12836"110 19168"150

87

combined system diamond indenteryDLC film, n and nD are the Poissons coefficients of the DLC film and diamond, respectively, and ED the Youngs modulus of diamond. Typical values of ns0.12, nDs0.07 and EDs 1144 GPa were considered w1x. The dynamic hardness (DH) was calculated at peak load according to the equation: Pmax h2 max

Fig. 1. A typical nanoindentation loadyunload displacement curve of a F-doped DLC.

3. Results and discussion Table 1 presents the processing conditions and the thickness of the fluorinated DLC films on silicon. Fig. 1 is a typical plot of a loadingyunloading nanoindentation curve of F-doped DLC coatings. Under a load up to the very small value of 5 mN, the displacement behaviour does not give a completely regular and smooth line, namely at the final section of the unloading curve. This is due to the combination of two effects: (i) the brittle nature of the coatingysubstrate system, that leads to discrete fracture events originating small steps

DHs37.838

(4)

The indentation impressions were observed by scanning electron microscopy (Hitachi S-4100) in order to analyse the indentation pattern.

Fig. 2. Representative indentation morphology of samples (a) CF0-P5 (0%CF4; 0.5 kPa), (b) CF60-P5 (60%CF4 ; 0.5 kPa), (c) CF80-P5 (80%CF4; 0.5 kPa) and (d) CF30-P15 (30%CF4; 1.5 kPa).

88

S.C. Trippe et al. / Thin Solid Films 446 (2004) 8590

Fig. 3. Dynamic hardness (DH) as a function of the CF4 amount in the admission gas mixture.

in the curve; (ii) the accuracy of the depth measuring unit in such a small indentation load. The shapes of the indentations of the F-doped DLC coatings can be grouped in four distinct patterns that are illustrated in Fig. 2. For F-free DLC coated materials, Fig. 2a the mechanical response is essentially plastic and the film remains adherent to the substrate after the indenter removal, although four small diagonal cracks appear at the Vickers impression corners. For pure DLC coatings, Xu et al. w12x also observed that the indentations were accommodated by plastic deformation below the critical depth hcrit that corresponds to maximum penetration without influence of the substrate. For intermediate fluor doping amounts up to 60% of CF4 in the gas admission, the indentation morphology reveals a brittle nature with film spalling at the central indented region from which long diagonal cracks are developing, Fig. 2b. For the highest F-doping amounts corresponding to 80% CF4, the impression is surrounded by extensive microcracking of the film, Fig. 2c. This behaviour also occurs for lower CF4 amounts but with high processing pressures, as can be seen in Fig. 2d for Ps1.5 kPa and 30%CF4. In these conditions, film microcrackings tend to appear under the indenter contact region. The variation of the dynamic hardness (DH) with the CF4 amount is plotted in Fig. 3. As dynamic hardness is measured at the maximum load that corresponds to the point when the indenter penetrates the most, this parameter includes not only the plastic deformation of the film but also the elastic component. This parameter was taken into account to compare the mechanical strength of the different samples instead of the Vickers nanohardness, whose values are altered by the brittle nature of the F-doped DLC coatings that originate cracking followed by film delamination, mainly during unloading w18x. For the lowest processing pressure, 0.5 kPa, dynamic hardness is almost invariant with CF4 addition until 60% followed by a significant decrease for 80%, Fig. 3. The film softening at 80% CF4 is due

to the substitution of strong CsC (bond energy of 610 630 kJ moly1 w19x) by weaker CF bonds (bond energy of 442450 kJ moly1 w20x). The fluorine is always a termination radical in the CC networks and consequently decreases cross-linking and film density, leading to a new more open structural arrangement w3,10x. F.L. Freire Jr. et al. w3x also observed a hardness decreasing with CF4 partial pressure. The most intense vibration mode measured by FTIR analysis for CFx was the CF2 asymmetric mode (1200 cmy1). The amount of CF2 chains, estimated for each sample from integrated area of this peak, is represented as a function of CF4 percentage in Fig. 4. The increase of the film fluorine content with the CF4 incorporation in the reactive gas was already demonstrated by other techniques like ion beam analysis (IBA) and X-ray photoelectron spectroscopy (XPS) w3,10x. For the processing pressure of 1.0 kPa, an increment of dynamic hardness takes place for intermediate fluor doping of the film, Fig. 3. This increase comes from the development of a significant amount of ordered graphite clusters w15x when fluor is incorporated in such scale in the DLC film, as the ID yIG ratio taken from Raman spectra proves, Fig. 5. The enhancement of the ID yIG ratio with the CF4 partial pressure was also found by other authors w3,10x. The effect on DH is counteracted for high CF4 amounts by the CF2 chains increment, Fig. 4, leading to a remarkable DH fall, Fig. 3. For the highest processing pressure used in this study, Ps1.5 kPa, the early formation of CF2 chains for low CF4 incorporation in the gas mixture, Fig. 4, leads to a notable drop off of the hardness values (Fig. 3). The evolution of DH and the film structure characteristics as a function of the processing parameters is in agreement with the morphology of indentations presented in Fig. 2. The undoped films reveal a major plastic nature in their indentation impression therefore they are relatively soft. The enhancement of dense incoherent

Fig. 4. Variation of the amount of CF2 chains with the CF4 percentage in the admission gas mixture.

S.C. Trippe et al. / Thin Solid Films 446 (2004) 8590

89

Fig. 5. Intensity ratio between D and G graphite bands in the Raman spectra of samples processed at different total gas pressures as a function of the with the CF4 amount in the admission gas mixture.

Fig. 6. Vickers hardness (HV ) and Youngs modulus (E) as a function of the CF4 amount in the admission gas mixture.

boundaries between graphite clusters w11x leads to a higher rigidity and so to a brittleness increment, as shown in Fig. 2b, together with a small hardness increase for intermediate CF4 contents and low to medium (0.5 1.0 kPa) gas pressure. Under these conditions, but for high CF4 concentrations, above 60%, the CsC replacement by CF2, Fig. 4, originates extensive microcracking, Fig. 2c. This is due to a softening mechanism, Fig. 3, similar to what can be observed in ceramic materials w21x. The mechanical behaviour of samples processed at Ps1.5 kPa has a distinct dependence on the CF4 amount. This processing pressure reveals excessive to the film strength once softening by microcracking, Fig. 2d, soon controls the mechanical response, similar to the characteristics of the films processed at lower pressures with higher F contents. The formation of CF2 chains is strongly activated at 1.5 kPa, as evidenced in Fig. 4. The above referred lack of uniformity of the most part of the loadingyunloading indentation curves with 5 mN of applied load, only makes the calculation of the Vickers nanohardness and Youngs modulus of samples CF70-P10, CF10-P15 and CF70-P15 possible. These data are plotted in Fig. 6. The Vickers nanohardness values of these samples range from 55 to 70 GPa, which are comparable values to those reported for DLC coatings with high sp3 content w1,22x. The dependence of nanohardness on the CF4 content and total pressure corroborates the trend observed for the dynamic hardness, Fig. 3. For F-doped DLC coatings produced with high total pressures (Ps1.5 kPa), the hardness does not change with the F content being lower than the values observed for Ps1.0 kPa. The low rigidity of 1.5 kPa samples due to the high CF2 replacement of CsC bonds is corroborated by their Youngs modulus that present values in the range 235242 GPa, contrasting to 326 GPa for the 1.0 kPa sample.

4. Conclusions The mechanical evaluation of thin and hard DLC films requires the use of the nanoindentation technique with very small indentation loads. Despite the brittle nature of the films that often conducted to irregular loadingyunloading displacement curves, it was possible to measure the hardness and the Youngs modulus of fluorinated DLC films as a function of the CF4 percentage in the reactant gas mixture and total processing pressure. SEM observations of the indentation impression patterns allowed the classification of the mechanical response according to the plasticity degree andyor film cracking and delamination extent. Undoped films are essentially plastic while highly F-doped ones reveal an extensive microcracking around the indentation. The last behaviour is also observed in the coatings processed at the highest pressure of 1.5 kPa. As a consequence, the highest hardness values were achieved for low to medium CF4 amounts (3050%) and intermediate processing pressures (1.0 kPa). These results come from the combination of two opposite structural phenomena contributing to hardness, revealed by FTIR and Raman spectroscopies: the formation of graphite clusters that increases the hardness and the replacement of CsC by CF2 chains that weakens the films structure. The Vickers nanohardness and the Youngs modulus values estimated by the depth sensing indentation method with 5 mN load for F-doped DLC fall in the 5570 GPa and 235326 GPa ranges, respectively. References
w1x J. Robertson, Mater. Sci. Eng. R 37 (2002) 129. w2x A. Grill, Diamond Relat. Mater. 8 (1999) 428. w3x F.L. Freire Jr, M.E.H. Maia da Costa, L.G. Jacobsohn, D.F. Franceschini, Diamond Relat. Mater. 10 (2001) 125.

90

S.C. Trippe et al. / Thin Solid Films 446 (2004) 8590 w13x S.C. Trippe, R.D. Mansano, Modern Phys. Lett. 16 (2002) 577. w14x N.B. Colthup, L.H. Daly, S.E. Wiberly, Introduction to Infrared and Raman Spectroscopy, Academic Press, New York, 1964. w15x G. Adamopoulos, K.W.R. Gilkes, J. Robertson, N.M.J. Conway, B.Y. Kleinsorge, A. Buckley, D.N. Batchelder, Diamond Relat. Mater. 8 (1999) 541. w16x J.L. Andujar, F.J. Pino, M.C. Polo, A. Pinyol, C. Corbella, E. Bertran, Diamond Relat. Mater. 11 (2002) 1005. w17x W.C. Oliver, G.M. Pharr, J. Mater. Res. 7 (1992) 1564. w18x A.G. Evans, T.R. Wilshaw, Acta Metall. 24 (1976) 939. w19x M.B. Smith, J. March, Marchs Advanced Organic Chemistry, John Wiley and Sons, Inc, New York, 2001. w20x W.H. Brown, C.S. Foote, Organic Chemistry, Saunders College Publishing, Fort Worth, 1998. w21x M.G.S. Naylor, T.F. Page, J. Microscopy 130 (1983) 345. w22x E. Martnez, J.L. Andujar, M.C. Polo, J. Esteve, J. Robertson, W.I. Milne, Diamond Relat. Mater. 10 (2001) 145.

w4x T.W. Mountsier, J.A. Samuels, Thin Solid Films 332 (1998) 362. w5x C.P. Lungu, A.M. Lungu, Y. Sakay, H. Sugawara, M. Tabata, M. Akazawa, M. Miyamoto, Vacuum 59 (2000) 210. w6x J.W. Yi, Y.H. Lee, B. Farouk, Thin Solid Films 374 (2000) 103. w7x A. Grill, Diamond Relat. Mater. 10 (2001) 234. w8x X. Wang, H.R. Harris, K. Bouldin, H. Temkin, S. Gangopadhyay, M.D. Strathman, M. West, J. Appl. Phys. 87 (2000) 621. w9x S. Liu, S. Gangopadhyay, G. Sreenivas, S.S. Ang, H.A. Naseem, Phys. Rev. B 55 (1997) 13 020. w10x M.E.H. Maia da Costa, F.L. Freire Jr, L.G. Jacobsohn, D.F. Franceschini, G. Mariotto, I.R.J. Baumvol, Diamond Relat. Mater. 10 (2001) 910. w11x V. Kulikovsky, P. Bohac, F. Franc, A. Deineka, V. Vorlicek, L. Jastrabik, Diamond Relat. Mater. 10 (2001) 1076. w12x Z. Xu, D. Rowcliffe, Surf. Coat. Technol. 161 (2002) 44.

You might also like