586 © 2012 Society of Vacuum Coaters 505/856-718855th Annual Technical Conference Proceedings, Santa Clara, CA April 28–May 3, 2012 ISSN 0737-5921

Modified Diamond-Like Carbon Multilayer Coatings on Metallic Substrates Produced by Pulsed-DC Hollow Cathode PECVD

R. Birney* and F. Placido, Thin Film Centre, University of the West of Scotland, Paisley, United Kingdom and Scottish Universities Physics Alliance (SUPA)

*SVC Sponsored Student

AbSTRACT

Diamond-like carbons (DLCs) have attracted much interest in recent decades, and have found a wide range of applications, including in optoelectronic devices, solar cells, infrared optics and, notably, as protective coatings for metals in corrosive and/or abrasive environment. These consist of the amorphous hydrogen-carbon alloys (a-C:H), hydrogen-free amorphous carbons (a-C) and tetrahedrally-structured amorphous carbons, (ta-C), which can contain in excess of 90% sp3 bonding. In this paper, modified a-C:H type diamond-like carbon mul-tilayer films have been deposited on a range of metallic substrates: nickel aluminium bronze, copper and 80/20 cu-pronickel alloy by pulsed-DC hollow cathode PECVD. The effects of substrate plasma pretreatment on the structure of both the substrate itself, and the structure and properties of the coating are considered. The protective effect of these coatings has been demonstrated by electrochemical techniques. The coatings are further characterized by Raman spectroscopy (with particular regard to the C-C sp2/sp3 bonding ratio), surface energy determination, SEM, and nanoindentation.

InTRoDUCTIon

Diamond-like carbon (DLC) is a metastable form of amorphous carbon, comprising a network of tetrahedrally- (sp3 hybrid-ized) and trigonally- (sp2 hybridized) bonded carbon atoms, and, often, a significant proportion of hydrogen, both bound and unbound, within the matrix. Modified DLCs, containing a proportion of other elements such as silicon and/or metals have also been synthesised to optimise the performance of such coatings. DLCs have attracted a great deal of interest since they were first synthesised and described by Aisenberg and Chabot [1, 2] and continue to be an active research topic.

Within the DLC film matrix, the bonding can range from being completely graphitic in character (100% sp2) to being very similar to that of diamond; the latter is usually referred to in the literature as tetrahedral amorphous carbon (ta-C) and can have up to 90% sp2 bonding [3].

However, usage of the term ‘DLC’ can also refer to the hydrogen-free amorphous carbons (a-C), and the amorphous carbon-hydrogen alloys (a-C:H) [2-4]. DLC films of these types can contain anywhere from 1% to 50% hydrogen, with a-C:H generally containing less than 50% sp3 fraction and a-C generally containing 85% or more sp3 bonding [5, 6]. (It is worthy of note, however, that in addition to carbon-carbon DLC bonds, carbon-hydrogen bonds can also be in the sp3

configuration. Thus, increased hydrogen content can increase the sp3 content.)

It is generally accepted that many of the attractive properties of DLC are conferred by the sp3 carbon-to-carbon bonds, such as the high density and compact structure, the high level of hardness and wear resistance. Although the addition of C-H sp3 bonds also promotes a tetrahedral structure, these have a detrimental effect on the mechanical properties of the film. (They also have an effect on the optical and electronic properties of the matrix, due to the increase in the quantity of σ-electrons and decrease in the quantity of -electrons) [4].

DLCs are currently utilized in a wide variety of applications, including in optoelectronic devices [7], solar cells, infrared optics [8], as an impermeable gas barrier inside PET drinks bottles [9] and as a protective coating for metals in corrosive and/or abrasive environments [10, 11].

In this paper, the characterization of a-C:H type DLC films prepared by HC-PECVD, on several substrates, is considered, in addition to the effects of substrate pretreatment.

EXPERIMEnTAL DETAILS

5-layer composite films, containing germanium and silicon dopants, capped with a layer of a-C:H type DLC were depos-ited on several metallic substrates; nickel aluminium bronze, 80/20 cupronickel, elemental copper and 316 stainless steel, at 1.5 μm thickness using a hollow cathode pulsed-DC PECVD process. A schematic of the pulsed-DC PECVD apparatus is shown in Figure 1.

587

Figure 1: Schematic of hollow cathode pulsed‑DC PECVD apparatus [10, 12].

An unusual feature of this system is that it is designed to coat the interior of pipelines; the pipe itself acts as both the cathode and deposition chamber. Insertion of a flat aluminium stage facilitates the coating of flat components.

It has been reported elsewhere that plasma pretreatment of metallic substrates such as steel, via an argon etching or sput-tering process, can greatly improve the adhesion of DLC films to such substrates. On bare steel, a clear relation was found between argon sputter cleaning time and the minimisation of coating spallation, with a cleaning step of 15 minutes show-ing a 41.7% delamination area, 30 minutes reducing to 4.0%, and finally a 60 minute argon clean reduced the spallated area to 0.3% [13]. In a later publication, [14] it was pointed out that increasing argon sputter time enhances DLC adhesion, and that although thicker films and multilayer films have also been used to reduce spallation, these type of defects are, “considerably suppressed when the argon cleaning time (is) increased and samples were heat-treated to 100ºC prior to coating”. More recently, it was [15] reported on the effect of argon plasma pretreatment on deposition of diamond-like carbon films, investigating the effects of the process on stain-less steel substrates and on the delamination resistance and structure of the final DLC film. They report increased film adhesion with increasing bias voltage of argon pretreatment, with substantial improvement after argon pretreatment with bias of 600V. Interestingly, as the voltage is increased the DLC film structure changes, becoming more conformal to the substrate’s features. This is related to a change in the surface of the substrate itself, whereby a nano-structured, nano-crystalline surface layer forms, related to relaxation of stress and and reduction of chromium content in the surface layer. While the film adhesion is improved by this process, the DLC chemistry and topography can be affected too; sp3/sp2 ratio is seen to increase, surface roughness reduces and the size and uniformity of film asperities increases. Choi et al.

[16] also report that the adhesion can be strongly dependent on plasma pretreatment, although they also use thin amorphous silicon (a-Si) interlayers for the improvement of adhesion.

These reports led to the conclusion that plasma pretreatment may be of interest in terms of improving film adhesion on copper, cupronickel and nickel aluminium bronze, the latter of which is traditionally noted as being a difficult substrate in terms of coating adhesion, primarily due to the oxide layers at the passivated surface. Indeed, significant adhesion problems were encountered during early experiments, particularly in the case of NAB but also on copper, with films either not adhering to the substrate at all, or delaminating as the sample cooled, or occasionally detaching from the substrate after around 12 hours. One possible explanation for this is that NAB has a coefficient of thermal expansion significantly higher than that of steel, and so contracts markedly upon cooling, post-deposition. Given that the DLC film is inherently stressed, the large thermal contraction may partly have been the cause of such coating failures.

To investigate the plasma pretreatment idea further, samples of each of these metals were subjected to argon cleaning times of 15, 30, 45 and 60 minutes at 400W power, and ultrasonic cleaning.

Samples tested for corrosion resistance were scanned from -0.15V vs. E

OC to 0.5V vs. E

OC with a scan rate of 0.15 mV/s

using a Gamry Instruments Reference 600 Potentiostat / Galvanostat / ZRA unit with the corrosion cell located within a Faraday cage; exposed sample area was 1.1304 cm2. A carbon (graphite) counter electrode was used together with a 3.5M potassium chloride - silver / silver chloride reference electrode. The electrolyte used was simulated seawater, i.e. solution of 3.5% w/v sodium chloride in deionised water (being equivalent to 0.6M NaCl(aq.) Tafel plots were constructed to enable extraction of the corrosion current, i

corr . A corrosion

rate can thereafter be calculated from ,

(1)

where icorr

is the corrosion current density in A.cm-2, M is the molar weight of the metal being corroded in g.mol-1, σ is the volume density of the metal in g.cm-3, F is Faraday’s constant (=9.649x104 C.mol-1) and n is the valence of the corrosion reaction. This gives a corrosion rate in cm.s-1.

Here, a 5-layer system was employed, as summarised in Table 1.

588

Table 1: Summary of 5-layer film deposition.

Layer No.

Reactant Gas Mix (sccm) Duration (min)

1 TMS:TMGe 70:10 12 TMS:TMGe 70:5 13 TMS 40 14 TMS-C

2H

2 20:110 1

5 C2H

2120 3

The structure of the films deposited on each of the substrates was imaged in a Hitachi S-4100 scanning electron microscope (SEM).

The contact angles and surface energy of the DLC films were measured using a KSV Instruments’ CAM200 contact angle goniometer and SFECAM software. Three solvents of known surface tension components; namely deionised water, diiodomethane and ethylene glycol; were used to en-able calculation of surface free energy, using the Fowkes / Owens-Wendt-Kaelble equation:

(2)

Where sv

is liquid-vapour interfacial energy, lv

is solid-vapour interfacial energy, and the d and p superscripts denote dispersive and polar components of the interfacial energies respectively. Thus, at least two measurements with different solvents, of known dispersive and polar surface tension com-ponents, are required in order to solve the system of equations and in doing so, to calculate the surface energy of the solid phase. This approach to the calculation of surface energy is discussed at length elsewhere [17, 18].

The hardness and reduced modulus of the DLC samples deposited on SS316 substrate were measured, as a function of pipe position and cathode power, using a Hysitron Tri-boScope nano-indenter system with a three-sided pyramid (Berkovich) diamond tip. In this method, a small diamond-tipped probe is forced into the film progressively, and the force-displacement curve is plotted. The curve is also mea-sured during the unloading cycle. The pressure under the diamond tip, calculated from the ratio of force to the area of plastic deformation, defines the hardness. The unloading curve is used to calculate the reduced modulus, by drawing the tangent from the maximum load point and extrapolating backward to zero load. The Young’s modulus and hardness can thus be calculated. The maximum load force used in the nanoindentation measurements was 5000 μN.

For this study on a-C:H type samples on SS316, Raman spectra were measured using a Renishaw inVia micro Raman spectrometer system, with 514.5 nm Ar laser excitation source. Each sample was mapped over an area of around 5 mm2, and the spectra averaged over this area; spectra were averaged over 30-40 individual spectra to ensure a truly representative spectrum for each sample. Mapping was carried out on a Leica DM/LM microscope equipped with a Olympus 20x/N.A. 0.4 objective. Cosmic ray spikes were removed using the zap function, and map analysis and fitting carried out using using Renishaw WiRE 3.2 acquisition and fitting software. Line mapping was performed using a Streamline Raman mapping system with the spectrometer centred at 1250 cm-1

and accumulations of 10 seconds per point. The data were fitted using a double Gaussian method to extract the D and G peaks, using a linear background subtraction to remove the slope caused by the photoluminescence background , which in turn can be used to quantify hydrogen content [19]. The sp3

content was then calculated by the three following methods and the mean of the result used to give a final figure for sp3

fraction: extrapolation of the value from the chart of Pos(G) versus sp3 [20]; calculation of the sp3content, again from Pos (G), using the empirical relation proposed in [19]; and extrapolation of the value from the chart of I(D)/I(D) versus sp3 content using the ratio of intensities of the D and G peaks respectively [20].

RESULTS AnD DISCUSSIon

Structure and HardnessFigure 2 shows the hardness and reduced modulus values of the films deposited with varying power, by both deposition methods, and measured by nano-indentation. The values are broadly consistent with those reported elsewhere for a-C:H type films [2, 5], the measured values for hardness and re-duced modulus ranging from approximately 8-11 GPa and 50-80 GPa respectively.

Figure 2: Graph showing the values of hardness and reduced modulus, E

R, for the film samples deposited on SS316, at each of

three cathode power settings. Pipe positions 1‑5 are from the lower part of the pipe, 6‑10 from the upper.

589

No clear relationship emerges between the cathode power and film hardness. However, in general, hardness is slightly lower for samples in positions 6-10, i.e. the samples cut from the upper section of the pipe chamber liner. This likely indicates the existence of a temperature gradient in the pipe between upper and lower sections. Similarly, no clear trends emerge in terms of reduced modulus with respect to cathode power. The moduli seem to be rather low. This may possibly be explained by the nature of the samples themselves. Since they were deposited on 0.075 mm thick stainless steel 316 foil, the inherent stress within the coatings caused the foil to curve significantly. Although great care was taken to cut very small samples from the foil liner, it proved to be difficult to glue the samples to the steel disc sample mounts; this meant that in some areas, there was a small gap between sample and sample mount, which, given the slightly flexible nature of the foil itself, may have led to lower modulus values. (For comparison, the reduced modulus value for the same 5-layer design on rigid silicon wafer substrate was of the order of 100 GPa.)

Plasma PretreatmentFigure 3 shows treated and untreated NAB surface, and the cor-responding coated substrates. Note the delamination evident on the coated untreated sample. The treated substrate shows argon ion damage, and a lower level of surface contamina-tion; the corresponding DLC-coated sample displays nodular structure and significantly better adhesion than the untreated sample. Samples deposited on all substrates displayed similar nodular structure, though these images are omitted for reasons of space. Ultrasonic cleaning followed by 30 minute argon substrate pretreatment is the optimum preparation in order to maximise film adhesion and minimise defects and delamination on NAB. In general, it can be concluded that argon sputtering and ultrasonic cleaning of substrates before deposition im-proves adhesion on difficult substrates considerably, although the length of plasma treatment should be optimised for each substrate and film design. The improvement can be explained by both a reduction in surface contamination (as evidenced even in the case of ultrasonically cleaned substrates), and by the introduction of a nano-structured interface between the coating and bulk substrate, which improves adhesion [15].

Figure 3: Upper left to right: untreated NAB surface, untreated NAB coated with 5‑layer, 1.5 μm DLC. Lower left to right: 30 minute Ar‑sputtered NAB surface, 30 minute Ar‑sputtered NAB surface coated with 5‑layer, 1.5 μm DLC.

590

The latter point can be explained in some cases by an increase of surface area, and by the semi-nodular structure of the sput-tered substrate, which may allow a relaxation of DLC stress by enabling the film to deposit around nano-sized curved areas. This underlines the importance of consideration of substrate preparation in depositing protective carbon coatings.

Raman Spectra: Correlation to other Physical PropertiesAn example of the Raman fit, and a graph showing the varia-tion of Raman spectra with cathode power at the upper mid pipe position, are shown in Figure 4.

Overall, sp3 fractions of 38-46% were measured; the fraction decreasing with increasing cathode power, accompanied by a decrease in H content.

Drawing comparisons between Raman and other DLC param-eters, SFE shows a slight increase with increasing FWHM(G). Since an increasing G peak breadth is linked to an increase in graphitic cluster size and an increase in the degree of disorder in bond length and bond angle, then it may be reasonable to surmise that the increasing SFE is caused by a more disordered a-C:H structure, possibly accompanied by an increase in the concentration of dangling bonds at the film surface.

Mechanical hardness is seen to show a general increasing trend with increasing calculated sp3 content, as expected, and empirically obeys a linear relationship:

Hardness = 6.0557 (sp3 content) + 6.8186 (3)

Although a strong dependence of hardness on sp3 fraction would be expected (since, generally, the greater the sp3 content, the greater the diamond-like character), a greater sp3 content in this case doesn’t necessarily indicate a greater concentra-tion of C-C sp3 bonds, but may be due chiefly to an increase in C-H sp3 bonding, which does not increase the hardness of the film. Conversely, as support for this conclusion, the ratio I(D)/I(G), which, for a-C:H increases with decreasing sp3 content, was seen to increase with increasing deposition power, indicating that the sp3 content is indeed decreasing. However, the simultaneously lowering atomic hydrogen concentration as calculated from the PL background shows that the decrease in sp3 content is due in fact mainly to the loss of C-H bonding.

Corrosion ProtectionFigure 5 shows the potentiodynamic data, used to construct Tafel plots and calculate corrosion rates of cupronickel sub-strates, treated and untreated, coated with 5-layer modified DLC.

Figure 4: Raman spectra of DLC multilayer samples: (left) an example of D- and G- peak fitting; (right) raw average spectra, DLC on SS316 prepared at each of 3 cathode powers.

591

Figure 5: Potentiodynamic scans on bare cupronickel and DLC‑coated cupronickel, and pre‑treated cupronickel.

From the data, a general trend in terms of argon sputtering time of CuNi substrates shows that longer sputtering time before film deposition would seem to lead to lower corrosion current, and hence a lower corrosion rate. This is evidenced by the by the fact that bare CuNi has a higher corrosion current that bare CuNi sputtered in argon for 15 min. Additionally, the multi-layer DLC structure deposited on 15 minute argon sputtered CuNi shows a corrosion current of 924 nA.cm-2. Corrosion current decreases with increasing treatment time, reaching 117 nA.cm-2 for the same film structure on CuNi, sputtered in Ar for 1 hour; the corrosion rate of 0.00123 representing a protective efficiency of 87.50% over bare cupronickel.

This suggests a change in the characteristics of the metal surface due to argon treatment; it is conceivable that the pas-sivation (oxide) layer that forms on cupronickel exposed to oxygen-containing media is removed to some extent by the sputtering process, which may in turn improve film adhesion and thereby lower film porosity and corrosion current.

SURFACE EnERGY / ConTACT AnGLE

DLC is noted for its low surface energy [2].

The SFE data, shown as dispersive, polar and total, are depicted for cathode power 0.144kW in Figure 6. The samples from the middle upper section of the pipe appear, generally and for all cathode powers, to be more hydrophobic than the others, indicated by a higher water contact angle. Average total SFE values may suggest an inverse proportionality to deposition power; the lowest power group having average SFE of 47.7, medium power 46.0 and highest power 45.8mN.m-1 The values are consistent with those expected for a-C:H-type DLC [2].

Figure 6: Graph showing the calculated values of total surface free energy and the separation into its polar and dispersive components, for film samples prepared on SS316 at 0.144 kW power, and taken from lower (1‑5, left‑right) and upper (6‑10, left‑right).

ConCLUSIonS

A series of a-C:H-type multilayer DLC films, incorporat-ing quantities of silicon and germanium in the early layers, were synthesized on SS316, copper, cupronickel and NAB substrates, via pulsed-DC hollow cathode PECVD.

The influence of substrate pretreatment by argon sputtering was determined; argon sputtering and ultrasonic cleaning of substrates before deposition generally improves adhesion and corrosion resistance, although optimum pretreatment conditions can vary between substrates.

The protective efficiency of such films on cupronickel sub-strates was demonstrated, with an approximately eightfold decrease in corrosion current resulting from a 5- layer, 1.5 μm coating on substrate argon pretreated for 60 minutes when compared to untreated cupronickel.

Surface energy values appear to be inversely proportional to deposition power on SS316 substrates.

A relation was presented for mechanical hardness of a-C:H films. Hardness was shown to be proportional to sp3 content, as determined by Raman spectroscopy, although the contribution of C-H bonds to the total sp3 fraction must also be considered in such a determination.

592

REFEREnCES

1. S. Aisenberg, R. Chabot, “Ion-Beam Deposition of Thin Films of Diamondlike Carbon”, J. Appl. Phys. 42 (1971) 2953.

2. J. Robertson, “Diamond-like amorphous carbon”, Mat. Sci. Eng. R. 37 (2002) 129.

3. Y. Lifshitz, “Pitfalls in amorphous carbon studies”, Diamond Rel. Mater. 12 (2003) 13.

4. Y. Lifshitz, “Diamond-like carbon – present status”, Diamond Rel. Mater. 8 (1999) 1659.

5. A. Grill,”Diamond-like carbon: state of the art”, Diamond Rel. Mater. 8 (1999) 428.

6. A. Grill, “Electrical and optical properties of diamond-like carbon”, Thin Solid Films 355-356 (1999) 189.

7. A.M.M. Omer, SudipAdhikari, Sunil Adhikari, M. Rusop, H. Uchida, T. Soga, M. Umeno, “Electrical Conductivity improvement by iodine doping for diamond-like carbon thin-films deposited by microwave surface plasma CVD”, Diamond Rel. Mater. 15 (2006) 645.

8. X. Bian, Q. Chen, Y. Zhang, L. Sang, W. Tang, “Deposi-tion of nano-diamond-like carbon films by an atmospheric pressure plasma gun and diagnostic by optical emission spectrum on the process”, Surf. Coat. Tech. 202 (2008) 5383.

9. A. Ueda, M. Nakachi, S. Goto, H. Yamakoshi, A. Shir-akura, “High Speed and High Gas Barrier Rotary DLC Plasma Coatin System for PET bottles”, Mitsubishi Heavy Industries Ltd., Technical Review 42 (2005) 1.

10. D. Lusk, M. Gore, W. Boardman, T. Casserly, K. Boina-pally, M. Oppus, D. Upadhyaya, A. Tudhope, M. Gupta, Y. Cao, S. Lapp, “Thick DLC films deposited by PECVD on the internal surface of cylindrical substrates”, Diamond Rel. Mater. 17 (2008) 1613.

11. R. Sharma, P.K. Barhai, N. Kumari, “Corrosion resistant behaviour of DLC films”, Thin Solid Films 516 (2008) 539.

12. R. Birney, F. Placido, and J. Kavanagh, “Characterisa-tion and Comparison of Diamond-Like Carbon Thin Films Deposited by RF-PECVD and Pulsed-DC Hollow Cathode PECVD,” 54th Annual Technical Conference Proceedings of the Society of Vacuum Coaters, 2011, ISSN: 0737-5921.

13. S. Podgoric, R. Bulpett, G. Troisi, J. Rumble, “Tribological Behaviour of Diamond-like Carbon Deposited on Epoxy Resin Substrates”, Proceedings of the 2008 Conference for the Engineering Doctorate in Environmental Technol-ogy, 2008.

14. S. Podgoric, B.J. Jones, R. Bulpett, G. Troisi, J. Franks, “Diamond-like carbon / epoxy low-friction coatings to replace electroplated chromium”, Wear 267 (2009) 996.

15. B.J. Jones, L. Anguilano, J.J. Ojeda, “Argon plasma treatment techniques on steel and effects on diamond-like carbon structure and delamination”, Diamond Relat. Mater. 20 (2011) 1030-1035.

16. H.W. Choi, K. Lee, R. Wang, H.W. Oh, “Fracture behavior of diamond-like carbon films on stainless steel under a micro-tensile test condition”, Diamond Relat. Mater. 15 (2006) 38-43.

17. D.Y. Kwok. A.W. Neumann, “Contact angle measurement and contact angle interpretation”, Adv. Colloid Interface Sci. 81 (1999) 202-203.

18. M. Grischke, A. Hieke, F. Morganweck, H. Dimigen,”Variation of the wettability of DLC-coatings by network modification using silicon and oxygen”, Diamond Rel. Mater. 7 (1998) 454.

19. C. Casiraghi, F. Piazza, A.C. Ferrari, D. Grambole, J. Robertson, “Bonding in hydrogenated diamond-like carbon by Raman spectroscopy”, Diamond Relat. Mater 14 (2005) 1098-1102.

20. A.C. Ferrari, J. Robertson, “Raman spectroscopy of amorphous, nanostructured, diamond-like carbon, and nanodiamond”, Phil. Trans. R. Soc. Lond. A 362 (2004) 2477-2512.

21. V.S. Yadav, D.K. Sahu, M. Singh, K. Kumar, “Study of Raman Spectra of Nano-crystalline Diamond-like Car-bon (DLC) films’ Composition (sp2:sp3) with Substrate Temperature”, Proceedings of the World Congress on Engineering and Computer Science 1 (2009) WCECS 2009, October 20-22, San Francisco, USA.