artikel 82005 buinsters SCRATCH Computer

Thin Solid Films 474 (2005) 186 – 196 www.elsevier.com/locate/tsf Adhesion analysis of polycrystalline diamond films on molybdenum by means of scratch, indentation and sand abrasion testing J.G Buijnstersa, P Shankarb, W.J.P van Enckevortc, J.J Schermerd, J.J ter Meulena,* a Applied Physics, IMM, Department of Applied Physics, Radboud University Nijmegen, Toernooiveld 1, 6525 ED Nijmegen, The Netherlands b Metallurgy and Materials Group, Indira Gandhi Centre for Atomic Research, Kalpakkam-603 102, India c Solid State Chemistry, IMM, Radboud University Nijmegen, Toernooiveld 1, 6525 ED Nijmegen, The Netherlands d Experimental Solid State Physics III, IMM, Radboud University Nijmegen, Toernooiveld 1, 6525 ED Nijmegen, The Netherlands Received 27 November 2003; received in revised form 10 September 2004; accepted 10 September 2004 Available online 28 October 2004 Abstract Diamond films have been grown by hot-filament chemical vapour deposition (CVD) on molybdenum substrates under different growth conditions The films grown with increasing substrate temperatures show a higher interconnection of diamond grains, whereas increasing methane concentrations in the 0.5–4.0% range lead to a transition from micro- towards nanocrystalline films X-ray diffraction analysis shows Mo2C interlayer formation Indentation, scratch and sand erosion tests are used to evaluate the adhesion strength of the diamond films Using steel ball indenters (F 750 Am), indentation and scratch adhesion tests are performed up to final loads of 200 N Upon indentation, the load values at which diamond film failure such as flaking and detachment is first observed, increase for increasing temperatures in the deposition temperature range of 450–850 8C The scratch adhesion tests show critical load values in the range of 16–40 N normal load for films grown for h In contrast, diamond films grown for 24 h at a methane concentration of 0.5% not show any failure at all upon scratching up to 75 N Film failure upon indenting and scratching is also found to decrease for increasing methane concentration in the CVD gas mixture The sand abrasion tests show significant differences in coating failure for films grown at varying CH4/H2 ratios In contrast to the other tests, here best coating performance is observed for the films deposited with a methane concentration of 4% D 2004 Elsevier B.V All rights reserved PACS: 68.35.Gy; 68.55.-a Keywords: Chemical vapour deposition; Diamond; Adhesion; Molybdenum Introduction The development of chemical vapour deposited (CVD) diamond thin films has led to numerous applications Benefiting from the excellent material properties of diamond, polycrystalline diamond films are used in cutting tools, as protective coatings, composite additives, infrared windows and by virtue of the wide band gap and the possibility of doping, diamond films are ideal for high temperature and high power electronic devices [1,2] Additionally, the negative electron affinity makes diamond an ideal candidate for field emission display applications [3] The electronic * Corresponding author Tel.: +31 24 3653022; fax: +31 24 3653311 E-mail address: htmeulen@sci.kun.nl (J.J ter Meulen) 0040-6090/$ - see front matter D 2004 Elsevier B.V All rights reserved doi:10.1016/j.tsf.2004.09.021 properties of diamond are also suited for field effect transistors, piezoelectric effect devices, radiation detectors and ultraviolet photodetectors Till date, there is only a limited number of substrates onto which polycrystalline diamond films can be grown successfully without application of interlayers These include ceramics like WC, Si3N4 and SiC, metallic substrates like Mo and W and the semiconductor Si The adhesion of the grown diamond films plays a crucial role in the performance of the final product Therefore, many studies have been directed towards the improvement of the adhesion by careful optimization of the various substrate pretreatments as well as diamond growth conditions For example, substrate roughening by manual or ultrasonic scratching is known to enhance the mechanical locking of J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 the grown diamond film [4] The use of a bias-enhanced nucleation step during the early stages of the diamond CVD process also leads to an increased adhesion of the deposited film [5] The most important factor influencing the adhesion is the interlayer formation during the diamond growth process The final adhesion of the diamond film with the underlying substrate is determined by the number density of interfacial bonds between atoms from the interface and carbon atoms from the nucleated diamond together with the corresponding chemical bond strength These properties are strongly influenced by the substrate surface phases and the growth conditions, respectively In earlier work, molybdenum has been studied as a model substrate because the use of molybdenum generally leads to strongly adhering diamond layers [6] Trava-Airoldi et al [7] reported on the surface nitridation of molybdenum by ion sub-implantation They found that the formation of a thin molybdenum nitride surface layer reduced the indiffusion of carbon and hydrogen from the vapour phase thereby leading to increased adhesion of the diamond films Experimental results have shown that the field emission from diamond films deposited on molybdenum substrates is much stronger as compared to uncoated Mo, thereby emphasizing the importance of diamond film growth on molybdenum for cold cathode applications [6] It is known that diamond deposition at high temperatures causes fragilization of the molybdenum structure, which is unfavourable for most applications [7] Therefore, the diamond deposition has to be performed in the lower temperature range, though the adhesive strength is known to decrease with decreasing deposition temperatures In the present work, the adhesive properties of hot-filament CVD grown diamond films on pure molybdenum substrates are studied for varying growth conditions from 475 up to 850 8C Indentation and scratch adhesion as well as sand abrasion tests are performed and the outcomes are discussed with respect to the applied deposition conditions Experimental details The diamond films are grown in a conventional hotfilament-assisted CVD reactor on square, pure molybdenum substrates with dimensions of 12Â12Â0.5 mm3 Diamond deposition is carried out utilizing CH4/H2 volume ratios of 0.5–4% at substrate temperatures up to 850 8C For all deposition runs, the total pressure is kept at about 50 mbar and the total flow rate at 300 standard cm3 minÀ1 The TaC filament temperature as measured with an optical pyrometer is kept constant at 2150F20 8C A fixed filament-tosubstrate distance of 6–9 mm is used in all deposition runs To determine the substrate temperature, a K-type thermocouple is placed inside the substrate holder, approximately mm below the actual deposition surface In a separate study, the temperature profile on the substrate surface perpendicular to the filament axis was evaluated using infrared 187 pyrometry (spot size ~3 mm), for a fixed filament-tosubstrate distance of 10 mm [8] A maximum difference of about 50 8C is measured for samples with dimensions up to 12Â12 mm2 Prior to diamond deposition, the molybdenum substrates are manually scratched in a slurry of diamond powder (1–2 Am) in glycerol, ultrasonically abraded in a suspension of diamond powder (1–2 Am) in isopropanol followed by ultrasonic and manual cleaning in isopropanol The mean surface roughness after the substrate pretreatment procedure is in the order of R a=0.8 Am, as measured by a Perthen Perthometer M4P Field emission scanning electron microscopy (JEOL JSM 6330 F) is employed to study the diamond film morphology and film response after carrying out the different tests The diamond film quality is studied by means of micro-Raman spectroscopy using an Ar ion laser (514.5 nm) with an output power of 20 mW and a focused laser beam diameter of about Am (Renishaw System 1000) The Raman spectra are taken in the 1055–1920 cmÀ1 range The interlayer phases formed upon diamond CVD are studied by X-ray diffraction utilizing a Bruker AXS D5005 diffractometer using CuKa radiation (k=1.5418 2) in the h–2h geometry The indentation and scratch adhesion tests are performed in the 0–200 N normal load range using a scratch tester (Revetest, CSM instruments) with steel ball indenters (750 Am in diameter) The acoustic emission signals are recorded using a resonant detector (Vallen-System type SE150-M) set at 125 dB gain in the 20–500 kHz range The penetration depth is measured by a depth sensor, which is positioned inbetween the indenter head and the force transducer (Interface Force Measurements, Model SSB-AJ-50) On each specimen, a series of indentations is made applying normal loads of 25–200 N Constant indenter loading rates of 50 N minÀ1 are used for the b100 N indents and 100 N minÀ1 for the z100 N indents The steel ball is replaced after each indentation series Scratch tests are performed on the diamond coated molybdenum samples using the steel ball with a loading rate of 80 N minÀ1, a track length of mm and a scan speed of mmd minÀ1 In the indentation tests, the penetration depth, acoustic emission signal and normal load are simultaneously recorded, while in the scratch tests the lateral displacement, penetration depth, acoustic emission signal, normal load and tangential force are recorded Sand abrasion tests are performed on the diamond coated samples by a slightly modified standard rubber wheel test (ASTM G65) using sand particles of 250 Am in diameter, a flux of 15 g minÀ1, a sliding speed of 0.1 m sÀ1 and a normal force of 50 N All tests are performed at room temperature Results 3.1 Diamond film characterization Fig shows the scanning electron microscopy (SEM) micrographs of the diamond film surfaces for varying 188 J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 Fig SEM micrographs of the diamond film surfaces obtained at varying deposition conditions, i.e at a substrate temperature of (a) 475 8C, (b) 550 8C, (c) 650 8C, (d) 750 8C and (e) 850 8C at a fixed CH4/H2 ratio of 0.5% and (f) using a CH4/H2 ratio of 4.0% at a fixed substrate temperature of 750 8C The deposition time for all films is h deposition conditions Fig 1a–e shows the surface morphologies of the grown diamond layers as obtained at different substrate temperatures It is apparent that at 475 8C the grain size is very small and that the interconnection of the individual grains is poor With increasing substrate temperature, the average grain size as well as the interconnection increases However, at temperatures higher than 650 8C, secondary nucleation and twinning lead to the formation of diamond films exhibiting grain sizes, which are varying between about Am and several tens of nanometers Fig 1d and f shows the effect of different methane-to-hydrogen ratios on the surface morphology of the grown diamond layers at a fixed substrate temperature of 750 8C Using CH4/H2 ratios of 0.5%, microcrystalline films are formed (Fig 1d), whereas for 4.0% only nanocrystalline structures are observed (Fig 1f) For intermediate values, the film morphology gradually changes from microcrystalline towards nanocrystalline with increasing methane-to-hydrogen ratio It has to be mentioned that the average grain size for the films grown for h at a CH4/H2 ratio of 0.5% is smaller as compared to that of the films grown under the same conditions for 24 h (not shown) For the latter films, the average surface grain size is about Am It is known that the growth of grains within a continuous polycrystalline film leads to the formation of columnar grain structures Due to the mutual competition between these structures [9], the number of grains present at the surface reduces as the film thickness increases As a result, the grains observed at the growth surface become larger for longer deposition times The thickness of the films grown for 24 h is in the order of 20 Am, whereas it varies from about to Am for the films grown for h, depending on the applied growth conditions J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 In Fig 2, the micro-Raman spectra of the diamond layers are displayed Although no remarkable difference in the diamond film quality is seen as a function of the deposition temperature from the micro-Raman peaks, the maximum diamond peak intensity is found for the films grown at 650 8C The peak position is shifted from 1333.0 cmÀ1 for the films grown at 475 8C to 1336.5 cmÀ1 for those deposited at 850 8C A broad band centred around 1500 cmÀ1, which can be ascribed to amorphous carbon, is detected for all films and its intensity is increasing with increasing temperature For temperatures less than 650 8C, a low intensity band at about 1130 cmÀ1 is observed as well This is commonly attributed to nanocrystalline diamond [10], but recently Ferrari and Robertson [11] assigned it to transpolyacetylene segments at grain boundaries In contrast to the temperature series, the micro-Raman spectra show more significant changes for varying methane concentrations In Fig 2b, the spectra of the layers grown at 750 8C using varying methane concentrations are displayed From this figure, it is clear that the quality of the deposited layers is decreasing with increasing methane concentration The diamond peak intensity is drastically decreasing and is fully dominated by the broad D-band of graphitic carbon (centred at ~1360 cmÀ1) for the film grown at CH4/H2=4% For all samples 189 Fig X-ray diffraction pattern of a diamond coated molybdenum sample The applied growth conditions are Tsub=750 8C, CH4/H2=0.5% and t=4 h grown at CH4/H2z1.0%, the amorphous carbon band (~1500 cmÀ1), G-band of graphite (~1580 cmÀ1) and 1150 cmÀ1 band are clearly distinguished No significant change in the diamond peak position (~1336 cmÀ1) is observed with respect to the applied methane concentration In Fig 3, the X-ray diffraction pattern of a diamond coated molybdenum sample is shown Diffraction peaks from diamond, molybdenum and molybdenum carbide, Mo2C, are apparent The formation of a Mo2C intermediate layer between the diamond film and molybdenum substrate is the result of the carbide-forming tendency of the molybdenum and is common for the applied deposition conditions [12] 3.2 Indentation and scratch adhesion tests Fig Micro-Raman spectra of the diamond layers as a function of (a) substrate temperature and (b) CH4/H2 ratio The deposition time for all films is h For reasons of clarity, the subsequent spectra are shifted vertically with respect to each other 3.2.1 General description One of the most frequently used methods to quantify coating failure is based on indentation by a well-defined indenter tip Though in previous work, the radial crack length has been used as a measure for the interfacial cracking resistance of the film [13], the formation of such radial and/or circumferential cracks is common to hard and brittle films and is not a measure for adhesive failure The critical load for adhesive failure should only be taken as the load at which failure of the interface occurs, thereby leading to film delamination Another commonly applied tool to evaluate the adhesion of coatings is the scratch test It consists of sliding an indenter in a single scratch across the coating surface with increasing normal load A critical load P cr at which stripping of the coating occurs, is estimated and used as a measure of the adhesion [14] In this work, we have chosen for a blunt indenter type, i.e a spherical steel ball, for two main considerations The first was to avoid damage of the commonly applied and expensive diamond indenters Further, it has been shown that, to suppress non-cohesive failure and to ensure only adhesive failure of the films, it is necessary to use an indenter with a large radius Particularly for hard coatings 190 J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 on soft substrate materials, the Rockwell C indenter may not have a radius sufficiently large enough to reliably assess the adhesive failure of the interface [15] The commercial instrument used in this study couples an acoustic transducer with the indenter so that the critical load might be identified by the acoustic emission signal from the coating fracture However, it appears extremely difficult to determine the critical load by detection of the acoustic signal from the steel ball indenter The scratching of the ball on the diamond layers results in the production of a very strong acoustic background, thereby overlapping the signal from the coating detachment Therefore, in this study, the critical load is derived from the plot of the tangential force ( F t) versus track length, which shows an abrupt change at the position of coating stripping 3.2.2 Indentation test In Fig 4, the SEM micrographs of the indented regions on diamond coated Mo substrates obtained at varying substrate temperatures are shown Careful examination of all indented regions reveals the presence of concentric ring cracks within the indents combined with radial cracks running perpendicular outwards from the indents Upon 200 N normal loading, diamond films grown at 550 8C show a tendency for multiple, partial and concentric flaking at the exterior of the indent At 650 8C, the flaking results in the delamination of a large single piece, reducing the number of delamination events drastically At 750 and 850 8C, no delamination or coating flaking is observed at all The difference in film flaking behaviour of the 550 and 650 8C specimens can be explained by the difference in the interconnection of the individual grains within the coating As can be seen from Fig 1, the interconnection increases for increasing substrate temperature and, therefore, the flaking results in larger film pieces being detached from the substrate for increasing temperatures For lower temperatures, the connection between the individual grains is poorer and delamination results in the detachment of smaller film parts On all samples, a series of indents is made with final, normal loads of 25–200 N, which are stepwise increasing by 25 N For the diamond layer grown at 550 8C, film delamination is first seen at 75 N normal load, whereas it is observed at 150 N for the layer deposited at a substrate temperature of 650 8C Upon close examination of the indented regions, only interior concentric ring cracks and exterior radial cracks are seen for temperatures of 750 and 850 8C As there is no delamination at all, the adhesion strength of the diamond layers grown at these temperatures is higher than the stress field which is applied by the steel ball at 200 N normal loading Thus, an increasing diamond film adhesion for increasing substrate temperatures is concluded from Fig In order to study the effect of the applied methane concentration on the adhesion of the diamond films upon indentation, 200 N indents are made on a series of samples grown under varying methane concentrations, i.e from 0.5% to 4.0%, at a fixed substrate temperature of 750 8C Fig SEM micrographs of the indented regions on diamond coated Mo substrates for different substrate temperatures during deposition, as indicated at the upper left-hand side of each micrograph The applied normal load is 200 N The deposition time for all films is h and a CH4/H2 ratio of 0.5% is applied J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 For the films grown with 0.5% and 1.0% CH4/H2 gas ratios (Figs 4c and 5a), no delamination is observed at all This implies that the adhesion strength of these films is larger than the stress field generated by the indentation However, for 2.5% and 4.0% methane gas mixtures (Fig 5b,c), the films show significant film detachment within the indented regions Critical load values of 125 and 100 N are derived for film detachment at methane concentrations of 2.5% and 191 4.0%, respectively At regions where the concentric ring crack density is highest, the detachment is strongest None of the samples shows film delamination outside the indents In virtually all samples studied, radial cracks are formed upon indentation at loads in the 50–200 N range For these relatively low loads, the radial crack length is of about the same order of magnitude as the indent radius 3.2.3 Scratch test In Fig 6, the applied normal load and resulting tangential force are displayed as a function of the scratch track length for three specimens grown at 750 8C In all plots, the first part shows a linear behaviour of the tangential force with respect to the applied load The abrupt change in tangential force is detected at the moment at which the diamond layer starts to flake and gets detached from the molybdenum substrate, i.e at the critical load At CH4/H2=0.5%, the critical load is 38.5 N, whereas it is 16.5 N at CH4/ H2=4.0% It is clear that the critical load decreases for increasing CH4/H2 gas ratio SEM analysis at the positions at which coating stripping and flaking is first observed shows that at the exterior of the scratch channel the formation of large radial cracks leads to the flaking of the diamond film Within the scratch channel, the diamond film is stripped off by the steel ball indenter for slightly higher loads From the obtained scratch data, the coefficient of friction for the steel-diamond sliding contact can be deduced from the linear part of the plots prior to coating failure The ratio F t/F n (=l) is constant for loads bP cr and varies from about 0.15 to 0.19 depending on the applied growth conditions After coating failure, the coefficient of friction is not fully constant anymore, but values varying from 0.28 to 0.34 are obtained for these sliding contacts The scratch data for the film deposited for 24 h at 750 8C with a CH4/H2 gas ratio of 0.5% not reveal any abrupt change in the tangential force at all As the F t/F n ratio (~0.20) is nearly constant over the entire load range (1–75 N), it can be concluded that the critical load for coating delamination for this sample exceeds 75 N This is also confirmed by SEM analysis of the produced scratch track Only the deposition of steel ball indenter material onto the diamond-coated surface is seen and no stripping or coating failure of the film at all 3.3 Sand abrasion wear test Fig SEM micrographs of the indented diamond film surfaces at 200 N normal loading The films are grown at various CH4/H2 ratios, as indicated at the upper left-hand side of each micrograph The deposition time for all films is h and a substrate temperature of 750 8C is applied The experimental setup of the sand abrasion test is shown schematically in Fig 7a The sample is pressed against a rubber wheel and due to the elastic deformation of the rubber, a nearly rectangular contact is formed During the test, a constant flux of dry sand particles is introduced in between the sample and the rubber wheel Due to partial elastic embedding in the rubber, the abrasive material (sand particles) reaches the contact area between the sample and the rubber wheel and abrades the stationary sample In standard experiments, the weight 192 J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 bond strength, failure occurs most commonly by delamination at the interface During each sand abrasion test, the sample surface is investigated every minute by optical microscopy in order to detect failure/detachment of the diamond coating In this way, lifetimes of the various films are obtained, which correspond to the moments at which the diamond layers are detached from the molybdenum substrates and abrasion wear tracks are first observed In other words, the sand abrasion wear test is used as a tool for evaluating the diamond film adhesion, contrary to its application as a method to obtain the wear resistance by probing the weight loss In Fig 7b, the lifetimes of the diamond coated samples as obtained with varying CH4/H2 ratios are displayed After 10 of exposure to the abrasive sand particles, the diamond layer grown at CH4/H2=0.5% detaches from the molybdenum substrate at the contact area For CH4/H2 ratios of 1.5% and 2.5%, the lifetimes are and min, respectively Surprisingly, the diamond film grown at CH4/H2=4.0% only detaches after 25 In Fig 8, the SEM micrographs of two tested samples are displayed Fig 8a shows the wear track on the sample Fig Tangential and normal load as a function of scratch length The deposition time is h and a substrate temperature of 750 8C is applied for all samples The CH4/H2 gas ratio is stated at the upper left-hand side of each graph loss of the sample is a direct measure for the wear resistance of the sample However, in the case of diamond coated samples, since the diamond is much harder than the abrasive used, it only leads to abrupt adhesion failure of the diamond films The sliding force at the surface of the film results in a shear force component at the interface Due to differential deformation between the substrate and the film, coupled with the lower interfacial Fig A schematic representation of the sand abrasion test (a) and the lifetime of the diamond coated molybdenum samples with respect to the applied CH4/H2 ratio (b) All diamond layers are grown at 750 8C using a deposition time of h J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 193 Fig The SEM micrographs of the sample surfaces exposed to the sand abrasion wear test: (a) overall view of the wear track on the sample grown at CH4/ H2=0.5%; (b) the boundary between the wear track and the diamond film (CH4/H2=0.5%); (c) the diamond layer morphology at ~200 Am from the wear track (CH4/H2=0.5%); (d) view of the wear track on the sample grown at CH4/H2=4.0%; (e) the boundary between the wear track and the diamond film (CH4/ H2=4.0%); and (f) the diamond layer morphology at ~100 Am from the wear track (CH4/H2=4.0%) The white arrows indicate the sliding direction of the abrading sand particles grown at CH4/H2=0.5% It is about mm in length and 750 Am wide A magnified view of the boundary between the wear track and the adherent diamond film is shown in Fig 8b At the wear track, the diamond layer is fully removed and the molybdenum substrate is exposed As molybdenum is less wear resistant than diamond, the removal of substrate material at the wear track is significant However, outside the wear track the diamond film is still adherent and only little wear of the diamond grains is observed (see Fig 8c) Due to the abrading effect of the sand particles, the diamond film is planarized, i.e only the tops of the grains which are sticking out from the surface are worn down For the sample grown at CH4/H2=4.0%, coating detachment is first observed after 25 The corresponding wear track, which is about 500 Am in length and 180 Am in width, is shown in Fig 8d The boundary between the wear track and the adherent diamond film is shown in Fig 8e As can be seen from this figure as well as from the magnified view of the dballasT diamond layer outside the detached area (Fig 8f), only the tops of the round structures are affected by the abrasive sand particles The lower regions within the dballasT diamond layer still exhibit the nanocrystalline features Another striking feature is the presence of openings behind the round dballasT structures, which indicates that these ball-shaped structures behave like one single cluster upon abrasive contact Surprisingly, apart from these openings, no microcracks are observed at the abraded regions of all the tested films 194 J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 Discussion 4.1 Interfacial structure Molybdenum is one of the strong carbide forming materials The X-ray diffraction pattern in Fig clearly indicates the formation of a Mo2C interlayer So, the bonding structure of diamond films on molybdenum substrates undergoes a transition from fully metallic (M– M) in the molybdenum substrate via covalent metal carbidic (M–C) in the interlayer region towards covalent carbon bonds (C–C) in the diamond layer Detailed SEM investigation of the delaminated areas upon indentation and scratching in the present work shows that delamination of the diamond films takes place at the carbide–diamond interface This agrees well with the findings of Bahr et al [12] who analysed the bonding structures within the molybdenum carbide interlayer in fractured regions after indentation using X-ray photoelectron spectroscopy and Auger spectroscopy They detected the presence of carbon– carbon bonding and suggested that the extra carbon is most likely present along the grain boundaries of the Mo2C grains, thereby impacting the adhesion behaviour of the film Apart from applying optimum deposition temperatures and methane concentrations, it is widely known that the surface roughness of the substrate has a strong influence on the adhesion In this work, we have chosen for a substrate pretreatment based on a combination of manual and ultrasonic scratching The manual scratching leads to an increased surface roughness due to the formation of fine scratches, thereby supplying good diamond nucleation sites Additionally, the use of diamond powder in both the manual and ultrasonic scratching results in the presence of small diamond seeds enhancing the nucleation rate and density After the pretreatment procedure, the overall roughness of the molybdenum substrate surfaces is about R a=0.8 Am and the nucleation density of the films grown at 750 8C and CH4/H2=0.5% is in the order of 109 cmÀ2 The presence of the brittle molybdenum carbide interlayer will also have a strong influence on the mechanical behaviour of the grown diamond films when investigated by indentation and scratch testing From the present as well as previous work [12], it is seen that diamond growth on molybdenum can lead to carbide layers of even several microns The thickness of the carbide interlayer is strongly determined by the applied substrate temperature and methane concentration A systematic study of the carbide interlayer thickness and composition as a function of the applied growth conditions together with their effect on the adhesion and other mechanical properties is under progress 4.2 Influence of composition and morphology on adhesive strength It can be concluded that the effect of the substrate temperature on the carbon constitution of the grown diamond films is relatively small (Fig 2a) Considering that the scattering cross-section for the sp2-bonded carbon structures is even about 233 times larger than that of sp3-bonded structures for the argon ion 514.5 nm radiation [16], the grown films consist of more than 99% sp3-bonded carbon On the contrary, the CH4/H2 ratio applied during the hotfilament CVD process strongly affects the film morphology and composition, as can be concluded from Figs and An increasing CH4/H2 ratio leads to the formation of smaller grains and a higher percentage of non-sp3 bonded carbon structures By increasing the methane gas concentration in the stock gas slightly, a gradual transition from microtowards nanocrystalline diamond films is observed Though the diamond peak is fully dominated by the graphitic Dband for films grown at CH4/H2z2.5%, the percentage of sp3-bonded carbon will still be ample The actual sp3/sp2carbon fraction within the diamond films strongly determines the mechanical properties of these films For example, the hardness, the elastic modulus, the cracking behaviour (initiation and propagation) and the coefficient of friction in many tribological systems will vary as a function of the sp3/sp2 bonding ratio within the diamond films [17,18] Especially, the distribution of graphitic phases is determinant for the cracking behaviour As no additional nucleation step is applied in the present work, increasing methane concentrations will not only lead to lower sp3/sp2carbon fractions within the film, but also to the same at the nucleation side Consequently, the higher concentrations of non-diamond phases at the nucleation side lead to higher propensity for film delamination, which is clearly demonstrated in the indentation and scratch adhesion tests as the critical load for coating delamination decreases for increasing methane concentration The detrimental effect of increasing methane concentration on the adhesion seems to be expressed in the decreasing lifetime for sand abrasion as well Up to methane concentrations of 4.0%, the lifetime decreases significantly The higher resistance against sand abrasion of the film grown at CH4/H2=4.0% cannot only be explained by the difference in the percentages of the various carbon structures within the diamond film and at the carbide interface, as this will only gradually change as a function of the methane concentration For increasing methane concentrations, a gradual transition from sub-micron diamond towards dballasT diamond is observed At CH4/H2=4.0%, the deposited layer consists of dballasT diamond exhibiting ballshaped clusters of nano-grained crystallites During the initial stages of the sand abrasion test, only the higher ballshaped clusters are in contact with the abrasive sand particles and, therefore, the impact of the abrasive material is only taking place at these topographically higher sites An exact explanation for the better coating performance as obtained with CH4/H2=4.0% cannot be given However, it is known that apart from the reduced contact areas the formation of thin, low shear strength films on hard coatings or on the asperity tips of these coatings results in low J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 friction behaviour as well [19] The higher sp2-ratio within the diamond layer results in a higher propensity for the formation of such nanofilms, ultimately leading to a lower friction Additionally, the slightly larger coating thickness will also contribute to an increased coating lifetime Cleavage and crack propagation along crystallographic planes is—due to the presence of a widespread network of nano-twins—expected to be lower than in single-crystalline diamond grains as well These hypotheses will be verified in the near future by means of atomic force microscopy roughness measurements of diamond layers grown in a large range of methane concentrations prior to and at several intervals during the sand abrasion test 4.3 Residual stress and coating thickness The presence of residual stress in grown films generally has a strong effect on the adhesion For example, in the range of 1.0–2.0 GPa residual, compressive stress the cracking resistance and adhesion of plasma grown diamond films was found to increase [20] It is also known that the residual stress acting on a diamond film is not homogeneously distributed along the film thickness Especially at the carbide–diamond interface the stress can be much larger than at the surface layer of the diamond film The higher stress value measured for the thin film (4 h) indicates a high stress close to the interface Though the interfacial stress cannot be measured directly for the thicker diamond layer (24 h), it is believed to differ only slightly from that of the thin film A difference in interfacial stress will then not explain the difference in adhesion of the two films as measured by scratching More likely, it is originating from the difference in the load bearing capacity of the coatings If the diamond coating is thicker, it can, because of its stiffness, carry part of the load and the deformation of the molybdenum substrate will be smaller The frictional situation is more favourable as compared to the thin diamond layer because ploughing or hysteresis effects due to substrate deformation will be relatively minor [19] Compressive stress values varying from about 0.3 to 1.8 GPa are derived for the films grown at 475–850 8C For the films grown at varying CH4/H2 ratios, the diamond peak position seems to be almost unaffected indicating little effect of the applied methane concentration on the residual stress As the adhesion increases for increasing temperatures and decreases for increasing methane concentrations, a direct correlation between the residual stress and the adhesion can be ruled out Certainly, the differences in carbon structures and bonding densities at the carbide–diamond interface have a much stronger effect on the adhesion of the diamond layers Conclusions In this work, the effect of substrate temperature and methane concentration on the adhesion of polycrystalline 195 diamond films grown by hot-filament CVD on molybdenum substrates is investigated by means of indentation, scratch and sand abrasion tests Increasing substrate temperatures lead to a higher interconnection of the individual diamond grains and increasing methane concentrations in the 0.5– 4.0% range result in a transition from micro-towards nanocrystalline films Micro-Raman spectra taken from the various films show an increasing level of non-diamond phases for increasing methane concentrations, whereas only a slight change in film composition is seen for increasing deposition temperatures X-ray diffraction analysis shows that the diamond film growth is preceded by the formation of a Mo2C interlayer From the indentation and scratch adhesion tests, it is concluded that higher deposition temperatures lead to stronger adhesion, whereas increasing methane concentrations result in a decrease of the adhesion However, for sand abrasion, the lifetime of films grown at a methane concentration of 4.0% is about three to eight times higher than that of films grown for lower methane concentrations This work shows that, though the scratch, indentation and sand abrasion tests differ largely, the coating performance or, more particularly, the film failure as visualized by flaking, stripping and/or detachment in all three tests, enables to class diamond films deposited from different gas mixtures and at different substrate temperatures according to their adhesion strength Acknowledgements The authors wish to thank Dr Ir Bert Huis in ’t Veld and Cor Lossie for performing the sand abrasion tests and Leander Gerritsen for his technical support This work was performed as part of the research program of the Netherlands Technology Foundation (STW) with financial support from the Netherlands Organization for Scientific Research (NWO) References [1] W.D Man, J.H Wang, C.X Wang, Z.B Ma, New Carbon Mater 17 (2002) 62 [2] I.S Forbes, J.R Rabeau, J.I.B Wilson, P John, Mater Sci Technol 19 (2003) 553 [3] J.L Davidson, W.P Kang, A Wisitsora-At, Diamond Relat Mater 12 (2003) 429 [4] Y Avigal, A Hoffman, Diamond Relat Mater (1999) 127 [5] S.D Wolter, F Okuzumi, J.T Prater, Z Sitar, Thin Solid Films 440 (2003) 145 [6] W.L Wang, K.J Liao, L Fang, J Esteve, M.C Polo, Diamond Relat Mater 10 (2001) 383 [7] V.J Trava-Airoldi, E.J Corat, L.V Santos, A.V Diniz, J.R Moro, N.F Leite, Diamond Relat Mater 11 (2002) 532 [8] G Janssen, Homoepitaxial Diamond Synthesized by CVD Processes PhD Thesis, University of Nijmegen, Nijmegen, 1994 [9] A van der Drift, Philips Res Rep 22 (1967) 267 196 J.G Buijnsters et al / Thin Solid Films 474 (2005) 186–196 [10] P.K Bachmann, H.J Hagemann, H Lade, D Leers, D.U Wiechert, H Wilson, D Fournier, K Plamann, Diamond Relat Mater (1995) 820 [11] A.C Ferrari, J Robertson, Phys Rev., B 63 (2001) 121405 [12] D.F Bahr, D.V Bucci, L.S Schadler, J.A Last, J Heberlein, E Pfender, W.W Gerberich, Diamond Relat Mater (1996) 1462 [13] C.R Lin, C.T Kuo, R.M Chang, Diamond Relat Mater (1998) 1628 [14] P.K Mehrotra, D.T Quinto, J Vac Sci Technol., A (1985) 2401 [15] Y Xie, H.M Hawthorne, Surf Coat Technol 155 (2002) 121 [16] S.R Sails, D.J Gardiner, M Bowden, J Savage, D Rodway, Diamond Relat Mater (1996) 589 [17] N Savvides, T.J Bell, J Appl Phys 72 (1992) 2791 [18] N Savvides, T.J Bell, Thin Solid Films 228 (1993) 289 [19] D Dowson, Coatings Tribology: Properties Techniques and Applications in Surface Engineering, Elsevier, Amsterdam, 1994 [20] C.T Kuo, C.R Lin, H.M Lien, Thin Solid Films 290 (1996) 254  ... based on a combination of manual and ultrasonic scratching The manual scratching leads to an increased surface roughness due to the formation of fine scratches, thereby supplying good diamond nucleation... commonly applied tool to evaluate the adhesion of coatings is the scratch test It consists of sliding an indenter in a single scratch across the coating surface with increasing normal load A critical... of magnitude as the indent radius 3.2.3 Scratch test In Fig 6, the applied normal load and resulting tangential force are displayed as a function of the scratch track length for three specimens