Structure and properties of diamond-like carbon nanocomposite films containing copper nanoparticles

Transcription

1 Applied Surface Science 242 (2005) Structure and properties of diamond-like carbon nanocomposite films containing copper nanoparticles Chun-Chin Chen, Franklin Chau-Nan Hong* Department of Chemical Engineering and Center for Micro-Nano Technology, National Cheng Kung University, Tainan 701, Taiwan Received 1 February 2004; received in revised form 14 July 2004; accepted 20 August 2004 Available online 13 October 2004 Abstract Diamond-like carbon (DLC) nanocomposite films, containing copper (Cu) nanocrystallites, were synthesized and studied. Cu bonds very weakly with carbon, and does not form a carbide phase. Therefore, Cu nanoparticles can be easily formed in a DLC matrix by depositing Cu and carbon together. The mechanical properties of DLC films that contain Cu nanoparticles are interesting since the film toughness may be increased by grain matrix interface sliding. Hard, tough and stress-free DLC/Cu films were prepared by a sputtering Cu target in an argon/acetylene atmosphere while biasing the substrate with a radio frequency power supply. The residual stress of the film, calculated by Stoney s equation, was as low as 0.7 GPa. The reduced stress and the increased film toughness increased the critical load from 66 N for a conventional DLC film to 80 N for the DLC/Cu film, as measured in a scratch test. However, the DLC/Cu films were slightly less hard than the DLC films. # 2004 Elsevier B.V. All rights reserved. PACS: Er; w; x; Qp; i Keywords: Diamond-like carbon (DLC); Nanocomposite film; Copper; Nanoparticles; Mechanical properties 1. Introduction Diamond-like carbon (DLC) films have been extensively investigated over the last three decades. DLC films exhibit superb mechanical properties, including high hardness and low friction coefficients. These films are very useful in precision machining and * Corresponding author. Tel.: x62662; fax: address: hong@mail.ncku.edu.tw (F.-N. Hong). manufacturing. The importance of hard and abrasionresistant coatings for machining application is evidenced by the rapidly growing market for cutting tools with wear-resistant coatings. Besides hardness, many film properties including toughness, high adhesion, and low surface energy, are yet to meet the requirements of various applications. DLC films have been synthesized by sputtering, laser ablation, cathodic arc and chemical vapor deposition (CVD) [1 5]. The bombardment of energetic carbon species during deposition is critical for the growth of DLC /$ see front matter # 2004 Elsevier B.V. All rights reserved. doi: /j.apsusc

2 262 C.-C. Chen, F.C.-N. Hong / Applied Surface Science 242 (2005) films. The ion energy is the most important parameter for determining the characteristics of DLC films. However, the ion bombardment tends to result in the highly-dense packing of carbon atoms in the film, yielding a very high compressive stress therein. The internal compressive stress depends strongly on the distortion of the bond length and the angle of sp 3 - bonded carbon. A very high compressive stress tends to detach the film from the substrate, when the film thickness increases above a critical value. The internal stress can be reduced by different mechanisms. Adding a new element to the DLC matrix, which consists mostly of carbon and hydrogen, is a commonly used method. The third incorporated element is normally Si, Ti or W, all of which exhibit bond strongly with the carbon matrix. In this work, soft and ductile metallic copper nanocrystallites were embedded in the hard amorphous DLC matrix to increase the toughness and stress in the film. Copper was also used to prevent the formation of bonds between the nanocrystallite and the carbon matrix, facilitating grain matrix interface sliding, which increases the film s ductility [6,7]. No Cu metal carbide phase has even been formed. Fig. 1. A schematic diagram of the experimental set-up employed in this study.

3 C.-C. Chen, F.C.-N. Hong / Applied Surface Science 242 (2005) Experimental Fig. 1 schematically depicts the experimental setup. A plasma CVD system equipped with a Cu sputtering gun was used to deposit DLC/Cu nanocomposite films on the biased substrate using acetylene/argon mixtures. The substrate holder was biased with a radio frequency (RF) power supply of MHz (Huttinger). Cu was deposited by sputtering from an unbalanced magnetron gun using a negative pulse power supply (100 khz, ENI). The DLC films were deposited by accelerating the acetylene ions to bombard the substrate using RF bias, and Cu nanoparticles were deposited by the sputtering of copper target with Ar ions. The Cu target (99.9% in purity) was cleaned using argon sputtering for 10 min before deposition. Stainless steel (SKD11) substrates (30 mm 7 mm) and Si substrates (20 mm 20 mm 0.38 mm) were used. Before deposition, SKD11 substrates were polished to a mirror finish with 0.3 mm Al 2 O 3 particles; ultrasonically cleaned for 20 min in acetone, and then blown dry in nitrogen. A thin buffer layer about 0.1 mm was deposited using Hexamethldisiloxane (HMDSO, 99%) reactant on the SKD11 substrate before depositing DLC or DLC/Cu films. The buffer layer was grown at Torr, with a substrate bias of 120 V, using an RF power of 110 W and for 20 min. Silicon substrates were cleaned first in acetone and then in HF solution. After the substrate had been loaded, the base pressure of the deposition chamber was pumped down to under 10 5 Torr. The substrates were then cleaned by Ar ion bombardment for 20 min. The pressure was maintained at Torr during deposition. Table 1 lists the deposition parameters. The self-bias normally depends on the geometry of the reactor. The deposition chamber used in this work was 450 mm in internal diameter and 500 mm high. The substrate holder was 300 mm in diameter. The copper sputtering target was 150 mm in diameter. The distance between the target and the substrate was 290 mm. The self-bias varied as the inverse the diameter of the substrate holder. The selfbias of the substrate at an RF power of 400 W was 440 V when the sputtering gun was turned off. Turning up the power of the sputtering gun from 280to320Wonlyslightlyincreasedthesubstrate Table 1 Experimental conditions for incorporating Cu nanocrystals in DLC films Substrate power (RF) 400 W RF self bias (substrate bias) 450 V (with sputtering) 440 V (without sputtering) Sputter power (pulse dc) W Base pressure <10 5 Torr Pre-clean gas Ar Working gas Ar:C 2 H 2 = 1:1 Working pressure Torr Sputtering target Cu Depositing time 30 min bias to 450 V at the same RF power. The pressure and Ar/C 2 H 2 ratio in the gas were maintained at Torr and 1, respectively, during deposition. The self-bias was not influenced by the Ar/C 2 H 2 gas ratio, and was only slightly sensitive to the gas pressure. The self-bias was reduced from 450 to 430 V by increasing the pressure from to Torr. The thickness and chemical composition of the films were analyzed by scanning electron microscopy (SEM) and energy dispersion spectroscopy (EDS), respectively. The structure of the deposited film was characterized using a Raman spectrometer (Dilor) and an Ar ion laser that emits nm radiation as the excitation source. The films were investigated using optical microscopy (OM) and high-resolution transmission electron microscopy (HRTEM). Stoney s equation [8] was used to determine the stress in the film by measuring the its curvature using an a-step profilometer. The strength of the adhesion between the film and the substrate was measured using a scratch tester. The scratch tip was a diamond stylus with a radius of 300 mm. The rate of increase of load was 1 N/s and the speed of the stage was 0.1 mm/s. The hardness of the film was determined by nanoindentation (Digital). A Berkovich diamond indenter with a tip radius of about 50 nm was used to make the measurements. The total included angle on this tip was The sample was indented with maximum loads in the range of mn to keep the indentation depth less than 10% of the film thickness, to minimize the substrate effect. Rockwell indentation was also used to evaluate the adhesion and toughness of the film.

4 264 C.-C. Chen, F.C.-N. Hong / Applied Surface Science 242 (2005) Table 2 The compositions and residual stresses of DLC nanocomposite films containing Cu nanoparticles Sample Sputtering power (W) Composition of Cu (at.%) Residual stresses (GPa) Film thickness (nm) B C C C Table 3 The copper deposition rate without C 2 H 2 Sputtering power (pulse power) (W) Deposition rate (nm/h) Results and discussion 3.1. Film composition and residual stresses Table 2 presents thickness and the residual stress of the film in relation to the Cu content. The Cu content was altered by varying the pulse power of the sputtering gun. Table 3 presents the deposition rate of copper without C 2 H 2, in relation to the pulse power. The analysis by energy dispersive X-ray spectroscopy (EDS) revealed that the Cu content increased monotonically with the pulse power of the sputtering gun. In Fig. 2, the glow discharge spectroscopy (GDS) analysis shows that Cu is uniformly distributed through the thickness of the film. The stress analysis based on Stoney s equation demonstrated that incorporating Cu nanoparticles into the DLC film clearly reduced the residual stress (comparing B1 with C1, C2 and C3). However, the DLC film with the lowest Cu content (C1), rather than that with the highest Cu content, had the lowest residual stress. All depositions had been repeated three times and the results were highly reproducible. As discussed below, characterizations of the film structure by Raman spectroscopy explain results Structure of the DLC films that contain Cu nanoparticles The C1 and C3 films were characterized by HRTEM; Fig. 3 depicts the bright-field images. Evidently, the DLC film was a nanocomposite film that contained Cu nanocrystals of sizes nm. Table 4 summarizes the d-spacings of the C1 and C3 films determined by selective area diffraction. Comparing the spacings of copper and graphite in JCPDS files clearly revealed that only Cu crystallites were formed. The density of the nanoparticles, but not their size, increased with the power of the sputtering gun was increased from 280 to 320 W Raman analysis of the DLC films that contain Cu nanoparticles Fig. 2. The GDS analysis of the C1 film. Cu percentage in the film is plotted vs. the depth of the film from the surface. Fig. 4 presents Raman spectra of the samples, B1, C1 and C3. The crystal structure of copper is facecentered cubic (FCC), and so is centrosymmetric, so Raman spectra are obtained from the DLC matrix only. Table 5 shows the results of fitting the Raman spectra with Gaussian peaks. Comparing C1 with B1

5 C.-C. Chen, F.C.-N. Hong / Applied Surface Science 242 (2005) Fig. 3. TEM bright field images and diffraction patterns for the (a) C1 and (b) C3 films. The inset shows the high resolution image of the Cu particle in the C3 specimen.

6 266 C.-C. Chen, F.C.-N. Hong / Applied Surface Science 242 (2005) Table 4 The diffraction indices of C1 and C3 films compared with Cu (JCPDS# ) and graphite (JCPDS# ) indices Cu (JCPDS# ) Our C1 sample Our C3 sample Graphite (JCPDS# ) Index d-value Index d-value indicates demonstrates that adding 11% Cu to the film shifted the G band and the D band to a higher wavenumber; increased the ratio of intensity of the D band to that of the G band, I D /I G, and reduced the FWHM of the G band. Accordingly, Raman spectra demonstrated that adding 11% Cu in the DLC film Fig. 4. Raman spectra of the DLC films. increased the content of sp 2 carbon. Then, increasing the sputtering power from 280 W (C1 sample) to 320 W (C3 sample) further increased the Cu content from 11 to 23%, shifting the G and the D bands to lower wavenumbers, reducing the intensity ratio, I D / I G, and increasing the FWHM of the G band. Hence, Raman results revealed that increasing Cu content from 11 to 23% reduced the sp 2 carbon content. Increasing the sputtering power from zero (B1 sample) to 320 W (C3 sample) shifted the G band and D band to slightly lower wavenumbers; reduced the I D /I G ratio, and increased the FWHM of the G band, indicating a fall in the sp 2 carbon percentage in the film. This conclusion is drawn from the statement of Ferrari and Robertson [9]. Table 2 indicates that all the films C1, C2 and C3 have similar thickness. The same signal to noise ratios in the Raman spectra also support the claim that all the films have similar thicknesses. The Raman intensity associated with C3 appears to be reduced by the rise of the background. The decline of the luminescence background in the Raman spectra is related to impurities in the films. The extent of the decline in the luminescence background in the Raman spectra indirectly measures the

7 C.-C. Chen, F.C.-N. Hong / Applied Surface Science 242 (2005) Table 5 Raman analysis of the DLC nanocomposite films containing Cu nanoparticles Sample G center (cm 1 ) D center (cm 1 ) FWHM of G (cm 1 ) I D /I G Stress (GPa) B C C hydrogen content in the films. In Fig. 4, the decline of the background of the C3 sample is less than those for the C1 and B1 samples. Hence, the C3 film contains less hydrogen than the C1 or B1 film. The film with 11% Cu (the C1 film) had the lowest stress, probably because it had the most sp 2 carbon in the DLC matrix. The fall in the sp 2 carbon content in the film as the Cu content increased from 11% (C1) to 23% (C3) may have been caused by the increase in the sputtering power from 280 to 320 W. The plasma density and the degree of gas ionization increased with the sputtering power. The fluxes of highly energetic ions that bombarded the growing films thus increased with respect to those of low-energy neutral species, such as hydrocarbon radicals. Accordingly, the mean kinetic energy of all species that struck the film increased, favoring the formation of sp 3 carbon and reducing both the sp 2 carbon content and the hydrogen content in the C3 film. The increase of the ion energy during the film deposition favored the formation of sp 3 carbon in this work. A moderate energetic ion flux that sufficed to penetrate the growing film surface was required to generate a higher fraction of sp 3 -carbon in the diamond-like films. Further energetic ion flux may generate graphite-like films with a larger fraction of sp 2 carbon because of the energy relaxation of a thermal spike. Therefore, the high fraction of sp 3 carbon in the C3 film due to the increase of ion flux reveals that the ion flux was moderate under the deposition conditions in this work. In summary, the lower stresses of the C1 and C3 films than of the B1 film were not related to the increase in the fraction of sp 2 carbon in the film, but to the effect of incorporating Cu nanoparticles in the carbon matrix Mechanical properties of the DLC films that contain Cu nanoparticles Table 6 lists the stresses, hardness and critical loads of the DLC films, including their sp 3 carbon contents and their Cu contents. Incorporating Cu nanoparticles in the DLC matrix clearly reduced the film stress. However, the stress in the C3 film (23% Cu) exceeded that in the C1 film (11% Cu), probably because of the higher sp 3 carbon content. Nanoindentation measurements revealed that the film hardness fell monotonically as the Cu content in the films increased. The hardness of the pure DLC film was 22 GPa. However, as the Cu content in the film was increased to 23%, the hardness of the DLC films fell to 15 GPa. These results are in contrast with those in other reports on superhard nanocomposite films such as TiN/Si 3 N 4 and ZrN/Cu, which revealed a maximum hardness for some compositions of hard nanocrystallites embedded in amorphous regions [6,7,10]. The difference, however, is explained by the absence of strong chemical bonds between Cu nanocrystallites and the DLC matrix [11], and by the ductility of Cu nanocrystals within the amorphous DLC matrix. These two factors enable easy sliding of the grain matrix interface in the DLC/ Cu composites [6], increasing the ductility, and thereby improving toughness [12]. Pure Cu is very soft, with a hardness of under 1 GPa. Notably, copper and graphite will never form an alloy, because the Table 6 Hardness and critical load of DLC nanocomposite films Sample sp 3 carbon content Cu composition (%) Stress (GPa) Hardness (GPa) Critical load (N) B1 Medium C1 Low C3 High

8 268 C.-C. Chen, F.C.-N. Hong / Applied Surface Science 242 (2005) bonding between carbon and copper is very weak. The HRTEM results in Fig. 3 verified the formation of only Cu nanocrystallites (15 30 nm) in the amorphous carbon matrix. DLC films were further deposited on SKD11 steel substrates under B1, C1 and C3 conditions, and the critical load was in each case measured in a scratch test. Critical loads depend strongly on the coating adhesive, the cohesive strength, and the frictional force between the diamond stylus and the coating surface. The adhesion should increase with the critical load. Table 6 reveals that the critical load of pure DLC films was 66 N the lowest value. The critical loads of the C1 and C3 films were higher at 80 and 74 N, Fig. 5. The OM pictures of the DLC nanocomposite films: (a) B1 and (b) C1, after Rockwell indentation tests.

9 C.-C. Chen, F.C.-N. Hong / Applied Surface Science 242 (2005) respectively. Adding Cu nanoparticles to the DLC films reduced the film stress and so increased its toughness, and therefore, the critical load. The effect of the adhesion of Cu nanocrystallites on the steel substrate is unclear. Film toughness was measured by conducting Rockwell indentation tests at a load high enough for significant deformation of both the film and the substrate [7]. Fig. 5(a) is an OM picture of the pure DLC film, B1, following indentation. Cracks were clearly observed on the edge. Pure DLC film was brittle in Rockwell indentation under a load of 150 g. Stress analysis revealed that the pure DLC film exhibited large residual compression stress, which may have accounted for the failure of adhesion. Fig. 5(b) displays an OM picture of the DLC film that contains Cu nanocrystallites, C1, following indentation. No crack in the DLC nanocomposite film is observed. The hardness of the film in Fig. 5(b) was 16 GPa, which is similar to that of hydrogenated diamond-like carbon. The ductility of the DLC nanocomposite film that contained Cu nanoparticles is consistent with the concept of the sliding of the grain matrix interface. 4. Conclusion Copper nanoparticles were embedded in the DLC matrix to increase the film ductility. TEM images demonstrated that the DLC nanocomposite films contained a high concentration of nm Cu nanoparticles. Increasing the Cu content in the DLC matrix slightly reduced the film hardness from 22 GPa for pure DLC to 16 GPa for the DLC nanocomposite film that contained 11% Cu. Embedding copper nanoparticles in the DLC film increased the critical load in the adhesion test from 66 N for pure DLC to 80 N for DLC nanocomposite (11% Cu). Incorporating Cu nanoparticles in the DLC films reduced the film stress (0.7 GPa) and increased the film toughness, increasing of the critical load in the adhesion measurements. However, the film hardness was slightly reduced. Our results demonstrated that the DLC film hardness was about 10 GPa when the film stress was reduced to under 1 GPa. The weak bonding between copper and carbon allows the tough and hard Cu-DLC nanocomposite film to be grown, because of the sliding of the grain matrix interface. Acknowledgement Financial support for this work from the Ministry of Economic Affairs under Contract No. 91-EC-17-A- 07-S and from Center for Micro-Nano Technology, National Cheng Kung University is gratefully acknowledged. References [1] S. Aisenberg, R. Chabot, Ion-beam deposition of thin films of diamond-like carbon, J. Appl. Phys. 42 (1971) [2] D.S. Whitmell, R. Williamson, The deposition of hard surface layers by hydrocarbon cracking in a glow discharge, Thin Solid Films 35 (1976) 255. [3] L. Holland, S.M. Ojha, Deposition of hard and insulating carbonaceous films on an RF target in a butane plasma, Thin Solid Films 38 (1976) 17. [4] J.C. Angus, P. Koidl, S. Domitz, Carbon thin films, in: J. Mort, F. Jansen (Eds.), Plasma Deposited Thin Films, CRC Press, Boca Raton, FL, 1986, p. 89. [5] Y. Lifshitz, Hydrogen-free amorphous carbon films: correlation between growth conditions and properties, Diamond Relat. Mater. 5 (1996) 388. [6] S. Vepřek, J. Vac. Sci. Technol. A17 (5) (1999) [7] A.A. Voevodin, J.J. Hu, J.G. Jones, T.A. Fitz, J.S. Zabinski, Thin Solid Films 401 (2001) 187. [8] X.L. Peng, T.W. Clyne, Thin Solid Films 312 (1998) 207. [9] A.C. Ferrari, J. Robertson, Phys. Rev. B 61 (20) (2000) [10] J. Musil, Surf. Coat. Technol. 125 (2000) 322. [11] Q. Wei, A.K. Sharma, J. Sankar, J. Narayan, Composites: Part B 30 (1999) 675. [12] A.A. Voevodin, J.S. Zabinski, Thin Solid Films 370 (2000) 223w.