Wear effects on microstructural features of amorphous-carbon thin films

ELSEVIER

Surface and Coatings Technology 94-95 (1997) 519-551

Wear effects on microstructural

features of amorphous-carbon

thin films

Abstract Amorphous-carbon thin films used as protective coatings for magnetic hard disks were subjected to wear by full-sized alumina-titanium carbide sliders. The associated microstructural changes were analyzed by Raman spectroscopy and high-resolution transmission electron microscopy (TEM). Raman spectroscopy did not detect structural differences between worn and unworn regions. However, TEhl micrographs directly showed structural differences between worn and unworn regions, and indicated increased graphitic content. For comparison, unworn carbon samples were annealed and analyzed. The annealed unworn carbon films showed structural changes similar to those of worn samples. This paper presents the experiments and the correlation between the temperature and~n!icrq~~I~~~ure~of th~film=&~and proposes a wear mechanism. 0 1997 Elsevier Science S.A. Keywords:

Wear; Amorphous-carbon

thin films; Micro:;tructure

1. Introduction Higher linear density for data storage requires closer spacing between the disk and recording head. As a result, protective coatings for magnetic recording disks, besides being chemically inert, hard, and resistant to wear, must be thin. Amorphous-carbon thin films provide these properties. Such films can be described as a random network of covalently bonded sp’ (trigonal) clusters linked by sp3 (tetrahedral) bonds [l]. Several researchers have observed the frictional and wear behavior of unlubricated carbon-coated disks and have shown evidence of a wear mechanism that includes surface modifications [2-41 of amorphous carbon. Several studies have identified a wear behavior of carbon that depends on the environment [3-S], but the observatron of microstructural changes in carbon has only recently become an area of interest. The study of wear-induced microstructural changes of carbon is a critical component in the development of wear-resistant coatings. Recently, researchers have postulated that chemical changes occur in carbon thin films during sliding. They have characterized worn carbon films and wear debris by Auger electron spectroscopy [9, IO], Raman spectroscopy

* Corresponding author. Tei.: +I 650 7231102; fax: +l 650 7254031.

0257~5972/97/$17.00 0 1997 b?&r PZZ SO257-8972(97)0016-l-7

Science S.A. All rights reserved

[2,&l 11, and transmission electron microscopy (TEM) 171. These studies reported that carbon films were structurally different after wear ranging from amorphous [7] to graphitic [12], with debris size dependent on the gas environment during sliding [9]. Marchon et al. [4] analyzed carbon that transfers onto the slider during sliding. Using Raman spectroscopy, they observed the films to be graphitic and attributed to them the reduction of friction. The proposed wear mechanism includes tribochemical wear [3-6.131 accompanied by modifications to the sliding surface. Frictional heating has been found to be the cause for structural changes -uJ], almough other mechanisms may be involved as well. Flash-temperature studies of asperities sliding on carbon films indicate temperatures on the order of several hundred degrees Celsius 1141. Calorimetry studies indicate that amorphous carbon transforms to graphite over a large temperature range of700-3000°C [ 151. With experimental observations demonstrating graphitic transformations on a larger scale than that of asperity-tip regions and occurring at slow sliding speeds (6 to 10 cm/s), frictional heating as a sole contributor to graphitization is suspect. Researchers have found that the grdphitization of carbon films can be enhanced by the metal films they are in contact with [16-191. ~The-purpose of this investigation is to study the microstructure of worn amorphous carbon films and to compare it

550

AC. Rumirez, R. Sinclair / Sor$trce and Coorings Trchnnlogy

with thermally annealed films. This study proposes a new approach and focuses directly on the microstructural changes induced by wear. We seek to establish the effect sliding has on carbon films and observe the thermal stability of amorphous carbon in the presence of metal films (magnetic media). In this study, we examine and analyze the microstructure through TEM and Raman spectroscopy to elucidate the wear mechanism involved and to increase understanding of the behavior of carbon films during wear.

2. Experimental

setup

2.1. Scunple preparcdion 2.1.1. Anrlenled samples A 30-nm coating of sputtered amorphous carbon was the final layer deposited onto a Al-ME substrate, coated with a 20 pm Ni-P underlayer, and a 70 nm Co-Pt-Ni magnetic alloy. The disk had a cross-textured surface and was never lubricated. Four 1-inch diameter coupons were cut from the disk and were annealed in a nitrogen-filled furnace for 2 h at 200, 300, 300 and 5OO”C, respectively, then quenched in water for 5 min. We took silicon substrates and deposited on them a trilayer composed of separately sputter-deposited layers of carbon, chromium, and carbon at a thickness of 6.5, 7, and 65 nm, respectively. We made cross-sectional TEM samples for in situ heating experiments from these wafers by a standard technique [20]. It is known that the chromium layer does not influence the carbon until temperatures are above 500°C [21]. 2.1.2. Wear-irnck samples We used the same type of carbon-coated disks for the wear track studies. A full-sized alumina-titanium carbide (A1203-Tic) slider having two rails of 0.3 mm x 0.7 mm

94-95 (1997) 549-554

spaced 2.5 mm apart was the agent for wear. We loaded the slider at 100 mN and monitored the friction. Wear tracks were generated by 5000 slider revolutions and were about 15 nm deep, as verified by atomic force microscopy (AFM) measurements. We cut coupons containing the wear tracks from the disk and analyzed them by Raman spectroscopy and TEM. We prepared cross-sectional TEM samples by a modified procedure: coupons were sputtered with a demarcation chromium layer, which provides easier identification. The coupons were then cut into two 5 x lo-mm rectangles, glued together with the worn films facing inward, and inserted into a copper support tube. Afterwards, we thinned the samples by conventional methods and ionmilled them. 2.2. Characterization 2.2.1. Rnmnn spectroscopy We performed Raman spectroscopy ex situ at room temperature on the annealed disk coupons and wear-track samples. All Raman spectra were excited with the 488 nm line of an Ar’ laser with an incident beam at 70” from normal and scattered light collected normal to the sample. An interference filter was used to eliminate elastically (Rayleigh) scattered light. This system disperses light with a 0.64-m single monochromator and collects it with an imaging photon-counting photomultiplier tube (PMT). It has previously been demonstrated that this system is capable of analyzing thin carbon films [22] and of two-dimensional detection [23]. 2.2.2. Transmission electron microscopy We used a Philips EM430ST microscope operated it at 300 kV at a resolution of 0.2 nm or better. The samples were annealed in situ with a Philips single-tilt heating holder up to 500°C. We controlled the temperature by manually adjusting the voltage at a constant current, while measuring temperature by thermocouple. The annealing steps were 50°C and the annealing holds were 30 min, with a heating rate between temperature steps of 20°C per min.

3. Experimental

results

3.1. Rnrnnn spectrrr for the annealed carbon samples

1800

1700

1600

1500 Raman

1400 Shift

1300

1200

1100

1000

(cm ‘)

Fig. 1. Raman specrra for sputtered carbon-coated magnetic disks that have been annealed at 100°C. 3OO”C, 4OO”C, and 500°C for 2 h in a nitrogen environment and quenched. and for an unannealed disk.

The spectra for the annealed carbon-coated disk samples consist of two broad and overlapping peaks between 1000 and 1800 cm-‘. It has been observed that the Raman spectra for amorphous carbon can be deconvoluted into two Gaussian peaks, the G-peak near 1580 cm-] and the D-peak near 1360 cm-’ [24]. In this work, films were annealed for 2 h in a nitrogen environment. Fig. 1 shows the change in the spectra as a function of annealing temperature. The spectra for the as-

.

Amorphous Carbon Worn/Unworn Regions

.

G position = 1557 cm .’

Inside

.

0

100

200 Anneal

300 Temperature

400 (“C)

500

Track

600

deposited sample and the sample annealed at ZOO’C x2 similar. However, samples heated to 300°C and higher. demonstrate an increase in the D-peak intensity and an up\trard shift of the G-peak position. The G- and D-p2ak.s are apparent at all temperatures. but thzir intensity drops dramatically in the sample annealed at 500°C. This sampl2 became dull copper-colored upon heating, and has a broadel G-peak than the 400°C sampl e. The change in the peaks is similar to the feature chanses obser\,ed by Dillion and et al. [24]. They observed that two parameter5 ch;.mge \\?th annealing: (1) the intensity ratio of the D-peak to the Gpeak (Idloj, which is associated with the cr)fstallitrz size and/or number, increasrs, and (7j the G-peak position. which is associated with graphite content, shifts upv,ard. We find that the intensity ratio as a function of annealing temp2ratur2, bho\vn in Fig. 3,, increases until 500°C at \\.hich temperature the value drops. Fig. 3 sho\vs tan up\\.nrd shift in the G-peak correlated with increased annealing temperature.

Fig. 3. A plot of G-p& position versus annxil tcmpcr2rur2 for spurtcrcd carbon-coated dish> hcn~cd fur 7 h in B nitrogen cm ircwncnt and qwxchcd.

To obtain Ramnn spectra of carbon inside and outside a \vear track. careful positioning of he laser beam \\.ns required. Fig. 1 sho\\.s Raman spectra inside and outklc of a \\\‘c’ar track...The peaks. resol\eJ b! computation. are only diffwxt in height: no shiftin g is apparent. Raman sptxtroscop); is sensiti\,2 to th2 thickness change [35] but not to wear for deformation of this scale.

\iy2ar tracks were identilkd by scanninp the profile of high and low regions separated b>; $opin,o ramps. Fig. 5 shows high-resolution images outside and inside a n’enl track. Outside the \\.eaK track,-[he cgl_?_firuration of the carbon is t>,pical of a amorphous mate@: a randomI\, peppered. ‘gra\:el-like’ looking configuration, u hich is composed of high-and-lo\\, contraht regions 1’61. \kYthin th2 ~venr track, th2 regions look less random and ha\,2 a ‘norm-like’ app2aranw. especialI>;. near the mayxtic ..-. layer. This Ies5 random rqion su,og2& less amorphous carbon and small r2gions of graphitic clust21.~. The dark ‘lvorm-like’ lines x2 consistent with material that has been partialI\, cr!,5tallized. as sren by other researchers for carbon annealed in contact with cobalt [ 161. The close prosimit\, of the magnetic layer makes it difficult to obtain electron diffraction data. but image procejjing (fazt Fourisr transformj of rh2 carbon la\w in the micrographs pro\,ided diffractograms for r2gions outside and inside the Mear track. These diffractograms. depicted in Fig. 5. display a diff2rence in the carbon film and indicate a I2ss random configuration in th2 \\w~r tI:ack. Fig. 6 sh0n.s the in situ TE\l progression of an identical specimen region that is being imaged at all times. while th2 sample is annealed from room temp2rature to 5OO’C. In th2 as-deposited film. the carbon has a random structurt? consisting of high-and-lo\\, contrast regions. However. at

552

A.C. Rumire:

R. Sinclair / Sut$.uce rind Coatings Technology 94-95

anneals 300°C and higher the film contains small polycrystalline regions that are typical of a graphitized structure.

4. Discussion The correlation between the features of the Raman spec-

Fig. 5. Bright-field high-resolution transmission electron micrographs and computed diffractograms of the carbon thin film taken in the region outside (a) and inside (b) the wear track.

(1997) 549-554

tra and the annealing temperature is based on a model proposingdhat the microstructure is made of graphitic clusters (<20 A) connected by sp3 links [27,28]. The data in our study show that the randomness of the graphitic clusters decreases, as seen by TEM, and the number and/or size of these clusters increases, as shown by Raman spectroscopy. There is a marked difference in the microstructure of carbon inside and outside a wear track. The Raman spectra for the annealing series in Fig. 1 indicate rhe onset of increased graphitic-content at temperatures between 200°C and 300°C: the spectra at these temperatures differ by peak shape and position. Researchers have found similar changes in Raman spectra during annealing of amorphous carbon. Dillion et al. [24] annealed ionbeam sputtered films to 950°C and found the graphitic transition between 400°C and 500°C. Their study suggests graphitization is also dependent on film history. In the present study, we see an increase in the intensity ratio IdIG with annealing temperature. This increase has been attributed to the increase in size of microcrystalline graphite or increase in the amount of crystallite boundary [29]. The D-peak, which arises from a relaxation of the Raman selection rules due to grain boundaries, denotes the disorder in the film. However, it increases with increasing annealing temperature. Although not fully understood, an increasing D-peak intensity during annealing is accompanied by increased ordering, as observed by scanning tunneling microscopy @TM) [30]. For the data we present, the size and/or amount of the graphitic component increases with increasing temperature, but drops at 500°C. We attribute this drop to the reaction that occurs between the carbon and the magnetic layer, a reaction that has been verified by differential scanning calorimetry (DSC) for cobalt-carbon trilayers [16]. The G-peak position, shown in Fig. 3, runs from values of 1560 to 1593 cm-’ with increasing annealing temperature. These increasing values also suggest an increasing graphitic content, while the low value for 5OO’C is attributed to a reacted film. In our study, Raman spectroscopy was not able to discern a difference in the carbon inside and outside of the wear track for this scale of deformation. The parameters of peak position and the ratio of IdIG indicate no difference between the worn and unworn regions except in peak amplitude. As a result, this study suggests that Raman scattering is sensitive to changes in film thickness but not in bonding structure. Liu et al. [12] have proposed a model for wear-induced graphitization: the gradual release of hydrogen from sp3 domains gives rise to easily sheared hydrogen-depleted regions of sp’ bonded regions. These easily sheared regions, in turn, reduce friction. As described by Dischler [3 l] for rf plasma deposited films, hydrogen evolves from amorphous carbon at approximately 300°C. Constant sliding imparts energy at asperities and increases their temperature, thus facilitating the release of hydrogen. This hydrogen evolution, accompanied by a change in the coefficient of friction,

Fig. 6. Bright-field high-roolurion transmission electron micrographz oi amorphous carbon anncnlcd at tcmpcraturcs oi (2~ 25T. ib, 300-C, is) 1OO’C and cd) 5OO’C.

has been directly observed [ 131 and is a component of the wear that occurs at the interfxc. The wear-track carbon and the carbvn films annealed at 300°C and higher, shown in Fig. 5b and Fig. 61). rsspectimely, exhibit similar features. These images ha1.e a lessrandom, ‘\vorm-like’ structure \\.hich is indicati\,e of the presence of small graphitic clusters. From thehe TEi\I images we gain information about the temperature> involved in n:ear. The carbon within the wear track ma\’ be experiencing temperatures on the order of 3OO’C or more, as indicated by the similarity between iti microstructure and that of the annealed carbon. Wear is a process in \vhich frictional temperatures at asperity points ha\,e been calculated to be several hundreds of degrees [l-l]. This stud? indicates that heating may not be restricted to these regions. but that a general warming of the entire carbon layer in the Lvear track occurs at temperatures of a fe\v hundred degrees Celsius. This warming of the film may gi1.e r&e to the release of hydrogen and cause graphitic regions to gro\\’ in number and/or size. The TEM micrograph and diffractogram of the region inside of the wear track indicate short range-or&r w,ithin the film, \I,hich buggests the prc’,ence of small graphite clusters. Our findings: like the lindings of other re.,earchers [3 317 -1321, indicate that the carbun is transformed into a more graphitic form during wear. Howe\.er. our micrographs show that this transformation occurs throughout the film, and not just at the friction interface. The ‘lvorm-

like’ structure pre\.ails especialI!, along the contours of the metal (magnetic media). e\‘en at the initial stages of UYXU ijO re\,olutions). These obser\xions suggest that the carbon mq nucleate more easil!, at the cobalt allo!, lay. The ~cvaphitizntion process ma\’ also be aided by the m?talh \\ith \\ hich the carbon is in contact. Researchers ha1.e obsewed that specific metals in contact with carbon act as catalysts for graphitization [ 17.151. Konnu [16.33] and Itoh [21] ha1.e demonstrated that cobalt and chromium mediate the graphitization of sputtered carbon. Their studies indicat? that cxbon transforms to graphite in th? presence of cobalt at 500-600°C. and transforms to graphite in the presence of chromium at SOO’C. In addition, the!, found microcr\,stalline carbides form at about 5OO’C for chromium, and that a thicker metal la),er induced cr) stallization more rapidly and at slightly lower temperatures. Thermal obser\,.ations suggest that thebe metals mediate a graphirk transformation in amorphous carbon films. The! ma!’ beha\.e in a similar fashion in thz \vear process. Our result> indicate that amorphous carbon that has been \x’orn increases in gaphitic content and that this ma!’ be a component of the o\.erall wear mechanism. \i’e ha\,e demonstr;ttecl that the microstructure of \\ orn carbon contains a less-random structure and is simikar to that ui amorphous carbon films that ha\e been amwaled al temperatures higher than 300°C. The contribution of the metal-mediated graphitization to the increase in graphitic content and the o\.erall role of the film modification are not understood. Further research needs to be done, but the results of thih study elucidate the behavior of carbon thin films during \\.ear.

5. Conclusions In thib stud!,. 1I.e ha\.e shown that ilear of amorphous carbon coating induces a structural change. 1L’e used Raman spectroscopy and transmission electron microxop! (TELL) to detect structural changes in carbon. \\‘e ha\‘e found that Raman spectrohcop!’ is not sensitive to the scale of cha~lge within the wear track. althuu$i it can detect changes in thermall> annealed samples. Raman spectra indicate an increase in yaphitic content \\,ith increased temperature. TE\I can detect a difference bet\\ een \x’orn and unworn regions of carbon and indicates short-ran,ae order in the n’om carbon. Through the MC’ of in situ annealing of sunples. 1i.e have directly:-o&ey\.ed the changes in the carbon a5 it increases in graphitic content \\ ith higher annealing temperatures. \iye ha\,e compared the worn films with films that Ivere thermall\, annealed. and ha1.e found similar structure5 at temperatures abo\,e 3OO’C. The comparison of worn films \\.ith thrrmallv annealed films gi\,es insight into the mechanisms involved in u’ear. Further studies need to be done to determine the role of the modified carbon in the \\‘ear process.

554

A.G. Rumire:, R. Sinciuir / Surjke

and Comings Technology 94-95

References [l] J. Robertson. Prog. Solid Sinte Chem., 21’ (1991) 199. [Z] B. Blanpain, M. Franck, H. Mohrbxher, E. Vancoilr, J.P. Celis and JR Roos, in D. Dowson (ed.) n?in Fiirns in Tribology, Elsevier Science Publiahera, p. 623, [3] B. Marchon, MR. Khan, iS. Hciman, P. Pereira and A. Lautie, IEEE Trans. Mqv~., 26 (1990) 2670. [J] B. Marchon, N. H&man and M.R. Khan, IEEE Trcms. h4agn., 26 (1990) 168. [S] B.D. Strom, D.B. Bogy, C.S. Bhatia and B. Bhushan, J. TTiboE., 113 (1991) 689. [6] A.G. Ramirez, M.A. Kelly, B.D. Strom and R.G. Walmsley, Triboi. Trczns., 39 (1996) 710. [7] M. Allouard, L.M. Vardavoulias, A. Thorel and M. Jeandin, Mum. McmllJ: Processes, 10 (1995) 1093. 181 B.D. Strom, D.B. Bogy, R.G. Walmsley. J. Brandt and C.S. Bhatia, J. Appl. Phys., 76 (1994) 465 1. [9] M.T. Dugger, Y.W. Chung, B. Bhusan and W. Rothschild, J. Tribal., 112 (1990) 238. [lo] T. Ych. C. Lin, J.M. Sivertsen, J.H. Judy and G. Chen, IEEE Trms. Mqn,, 29 (1993) 3939. [l I] B.L. Vlcck. B.L. Sargent, J.L. Lauer, S.K. Ganapathi and F.E. Talke. wenr, 170 (1993) 173. 1121 Y. Liu, Erdcmir, A., Meletis, E.I., S!& &a!. T&no%, 82 (1997) 48. 1131 X. Pan and V.J. Novotny, IEEE Tmns. Mngn., 30 (19943 433. [ 141 C.M.M. Ettlcs, AXE Trms., 29 ( 198.5) 32 1. [ 151 IS. McLintock and J.C. Orr, in P.L. Walker and P.A. Thrower (eds.j. Chmis~ry nncl Phpics oj Curbon. !&tier. New York, p. 243.

1161 1171 [IS] [19] [20] [21] 1221 [23] [24] [25] 1261

[27] [38] [29] [30] [31] [32] 1331

/I9971 549-554

T.J. Konno and R. Sinclair, Acre Metdl. Muter., 43 (1995) 471. A. Oya and A. Otani, Cnrbun. I7 (1978) 131. A. Oya and H. Marsh, J. dluter. Sci., 17 (1982) 309. S.R. Herd, P. Chaudhari and M.H. Brodsky, J. h’on-C~>sr. So/ids, 7 (,1972) 309. J.C. Bravman and R. Sinclair, J. Eiectrorz Microsr. Techn., I (1984) 53. T. Itoh, Ph.D. Thesis, Stanford University, 1996. J.W.I. Ager, IEEE Trans. Mngn., 29 (1993) 259. D.K. Veirs. J.W. Agers III, E.T. Loucks and G.M. Rosenblatt, Appl. Opt., 29 (1990) 4969. R.O. Dillion, J.A. Woollam and V. Katkanat, Phys. Rev. B., 29 (1984) 3482. S.S. Varanasi, J.L. Lauer, F.E. Talke, G. Wang and J. Judy, J. Tr& bol.. 119(3) (1997) 471. D,J. Smith, in R. Barer and V.E. Cosslett (eds.), A&ancrs in Opricui rind Electron Microscopy, liol. 2, Academic Press, New York, p. 1. J. Robertson and E.P. O’Reily, Phys. Ret*. B, 35 (1987) 2946. D. Beeman. J. Silverman, R. Lynds and M.R. Anderson, Phys. Rei: B, 30 (1984) 870. F. Tunistra and J.L. Koenig, J. Chem. Phys., 53 (1970) 1126. J.W. Ager. D.K. Veirs, B. Marchon, N. Cho and G. Rosenblatt, SPIE: Appl. Spectrosc. Muter. sci., 1437 (1991) 24. B. Dischler, A. Bubenzer and P. Koidl, Solid S!nre Commun., 48 (1983) 105. Y. Liu. A. Erdetnir and E.I. Meletis, .SL~$ Coat. Technol., 82 (1996) 18. T.J. Konno, Ph.D. Thesis, Stanford University, 1993.