Journal of Materials Processing Technology 89–90 (1999) 556–560

Nitrogenated carbon layer on magnetic recording disks

Sam Zhang a,*, M.J. Tan b, P. Hing b, H. Xie a, H.L. Wong c, W.L. Ng c

a Gintic Institute of Manufacturing Technology, 71 Nanyang Dri6e, Singapore 638075, Singaporeb Nanyang Technological Uni6ersity, Nanyang A6enue, Singapore 639798, Singaporec StorMedia International (S) Ltd., 9 Tuas A6enue 5, Singapore 639335, Singapore

Received 7 September 1998

Abstract

Hydrogenated carbon films have been used in the magnetic recording industry as a protective overcoat for memory disks formany years. In recent years, with shortening of the gap between the head and the disk (as in proximity heads), the inclusion ofnitrogen in the films to form hydro/nitrogenated films has also aroused tremendous interest. In this study, nitrogenatedamorphous carbon films (without hydrogen) (a-C:N) of thickness ranging from 5 to 20 nm were deposited on rigid magnetic disksvia the magnetron sputtering of a graphite target. The deposition was conducted in an argon/nitrogen atmosphere under RF(radio frequency) substrate bias power. A scanning probe microscope (SPM) with a contact mode of atomic force microscopy(AFM) mode was used in the roughness determination. A Raman spectroscope was employed to study the structural evolutionof the coatings with varying nitrogen/argon proportion and flow rate, radio frequency substrate bias and target power density.The surface contact angle between the a-C:N coating and deionized water was measured using a Rame–Hart goinometer. Underthe deposition conditions, the carbon thickness decreased with gas flow rate, but increases with nitrogen content. An increase incontact angle with bias power and decrease with flow rate were also observed. The roughness (Ra) of the coatings did not changebecause the coating was too thin to change the original substrate roughness. Also studied were the Raman peak positions, andthe D-band and G-band intensity ratio and their relationship with the processing parameters. © 1999 Elsevier Science S.A. Allrights reserved.

Keywords: Nitrogenated carbon; Magnetic disc; Sputtering; a-C:N; Carbon overcoat; Lubricant; Raman

1. Introduction

Amorphous hydrogenated carbon films (a-C:H) havebeen the subject of research for the last two decadesand have been used in the magnetic recording industryas a protective overcoat for memory disks for manyyears [1–3]. In recent years, with shortening of the gapbetween the head and the disk (as in proximity heads),inclusion of nitrogen in the films to form hydrogenated/nitrogenated films has also aroused tremendous interest[4]. Very recently, the use of nitrogenated amorphouscarbon (a-C:N) without hydrogen has been explored[5]. Currently, overcoats are typically on the order of15–20 nm thick and must possess high durability, lowfriction coefficient and chemical inertness. The perfor-mance of the thin film disks depends largely upon the

film microstructures as well as upon their mechanicalproperties. In this study, nitrogenated amorphous car-bon films (without hydrogen) (a-C:N) of thicknessranging from 5 to 20 nm were deposited on rigidmagnetic disks via unbalanced magnetron sputtering ofgraphite target in an argon–nitrogen atmosphere underRF (radio frequency) substrate bias power. The rela-tionship between the microstructures and the depositionconditions has been explored.

2. Experimental procedure

2.1. Deposition

Carbon overcoats of various thicknesses were de-posited in a d.c. magnetron sputtering system withpremixed argon–nitrogen gas (90% Ar–10% N2) in anindustrial production chamber; the graphite target to

* Corresponding author. Tel. +1-65-799-6091; fax.+65-792-2779.E-mail address: szhang@gintic.gov.sg (S. Zhang)

0924-0136/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved.

PII: S 0 9 2 4 - 0 1 3 6 ( 9 9 ) 0 0 0 3 5 - 7

S. Zhang et al. / Journal of Materials Processing Technology 89–90 (1999) 556–560 557

Fig. 1. Typical structure of the thin film medium.

bias power was also applied to the substrate to studysubstrate bias effects. The RF-induced bias voltage onsubstrate was varied from 0 to negative 160 V and thedeposition time was varied to give different filmthicknesses.

2.2. Physical and mechanical characterization

The coating roughness was determined using a scan-ning probe microscope (NanoScope III-A) from DigitalInstruments Inc. In operation, the contact atomic modeof force microscopy (AFM) mode was used at a setpoint of 2.5 V, and a scan rate of 0.9988 Hz over a scanarea of 1 mm×1 mm. A Raman spectroscope (Ren-nishaw RamaScope Model 127) using a He–Ne laserbeam of 632.8 nm as the excitation source was em-ployed to study the structural evolution of the coatingswith varying nitrogen/argon ratios and flow rates, radiofrequency substrate biases and target power densities.

The surface contact angle between the a-C:N coatingand the deionized water was measured using a Rame–Hart goinometer. To characterize the bonding betweenthe lubricant and the coating surface, the disks weretaken from the sputtering chamber, buffed, and lubri-cated via a dip coating process in a Fomblin Z-dolsolution at the same withdrawal rate. Then, the diskswere immersed in fresh PF5060 delube solution threetimes for 10 s each to dissolve the physically bondedlubricant. After delubrication, the remaining lubricantthickness was measured by the Fourier transform in-frared (FTIR) method. The average thickness of theremaining lubricant layer is termed the residual lubri-cant thickness. Since the delubrication process removesthe physically bonded layers but retains the chemicallybonded layers, the residual lubricant thickness is there-fore a measure of the chemical bonding of the lubricantonto the coating surface. An abrasion test was con-ducted on samples with different flow rates and deposi-tion power densities using a Selket Profile-AbraderSystem 1000 under the standard load. The abrasion was

disks distance being 25 mm. For easy control of the gascomposition, pure argon and nitrogen were used. Thegas composition was controlled by the ratio of the flowrates in a laboratory chamber (Teer Coating UDP 550)measuring 500 mm internal diameter by 500 mm high.Textured and non-textured 95 mm aluminum alloydisks were mounted on the sample holder parallel to330 by 133 mm rectangular graphite targets (99.9%)mounted on the chamber walls. The distance betweenthe disk surface and the target was set at 100 mm. Apair of facing targets was used for the deposition.Non-textured disks were included to provide a way toassess the coating morphology and roughness changebefore and after deposition. The substrate used was thetypical Al–Mg alloy disk [6]: on top of the Al–Mgalloy is a thick layer (10–25 mm) of NiP followed by aCr layer and the Co (magnetic) layer. The carbon filmis applied onto the Co layer, and finally, outermost, isa lubrication layer. A post-deposition and lubricationdisk is illustrated in Fig. 1. During the deposition, thebackground pressure in the sputtering chamber wasbelow 5×10−5 torr (in the case of the industrial cham-ber, 5×10−8 torr), and the working pressure wasbetween 3 and 11 mtorr during deposition. For somesamples, a standard radio frequency (RF, 13.56 MHz)

Fig. 3. Carbon film optical reflectance increase with substrate biaspower.

Fig. 2. Film thickness decrease and carbon reflectance increase withdeposition gas flow rate.

S. Zhang et al. / Journal of Materials Processing Technology 89–90 (1999) 556–560558

Fig. 4. Variation of film thickness with bias power: after a ‘thresh-hold’, the substrate bias generates back-sputtering, which reduces thecoating thickness.

suggested that a likely explanation is that the atomicmass of nitrogen is closer to that of carbon than argon,thereby resulting in better momentum transfer duringsputtering and hence a higher sputter rate and thereforea higher deposition rate.

3.1.2. Surface roughness and contact angleThe surface roughness of the film is another impor-

tant physical property, especially on magnetic recordingdisks. However, since the coating is very thin (up to 20nm), the roughness does not seem to change much withthe deposition parameters. Fig. 5 illustrates the rough-ness change with bias power.

The water contact angle, on the other hand, is moresensitive to deposition parameters. Fig. 6 plots thecontact angle as a function of the substrate bias powerfor non-textured and textured disks at different nitro-gen concentrations. A slight increase of contact anglewas observed with bias power but little effect was seenmeasured as the change in percentage of fiber optic

sensor output in terms of voltage after 60 s:

Wear rate=Vf−V0

V0

×100% (1)

where V0 is the initial voltage and Vf the voltage after60 s. A Keithley ohmmeter was configured to performthe resistance measurements on the textured surface ofthe coating at a voltage of 0.35 V.

3. Results and discussion

3.1. Physical properties

3.1.1. Film thicknessThe gas flow rate has a significant influence on the

film thickness. The deposition rate was found to de-crease with increasing flow rate, as plotted in Fig. 2 assolid circles. This was confirmed by the carbon filmoptical reflectance which decreases with increasingthickness (open circles) shown also in Fig. 2. Themeasurements were done on a CIE-1931 ChromaMeter.

The negative dependence of the deposition rate onthe gas flow rate, and hence on the chamber pressure,can be attributed to reduced gas species collision withthe substrates according, to kinetic theory.

Fig. 3 shows the dependence of the reflectance onsubstrate bias at various Ar/N2 mixture ratios. Increas-ing bias generates more bombardment on the substrateand the film up to a point where further increase resultsin loss of the film already deposited due to back-sput-tering. This is best seen in Fig. 4.

Also seen in Fig. 3 is the effect of the nitrogencontent: higher concentration of nitrogen gives lowerreflectance, or a thicker coating, since a-C:N has afaster deposition rate than a-C or a-C:H. Chen et al. [7]

Fig. 5. Showing the constant roughness with bias power.

Fig. 6. Showing contact angle increase with substrate bias, but alsolittle effect of nitrogen concentration, or whether the disks weretextured.

S. Zhang et al. / Journal of Materials Processing Technology 89–90 (1999) 556–560 559

Fig. 7. Contact angle as a function of flow rate during sputteringdeposition for both textured and non-textured Al–Mg disks for a gascomposition of 90% Ar with 10% N2.

The surface energy of the coating is related to thesurface contact angle through Young’s equation:

gf=gw cos u+gw–f

where g is the surface energy, subscript f is for film andw for water. From Fig. 6 it is seen that the surfaceenergy of the a-C:N film decreased with bias power.The increase in gas flow rate, however, brought about adecrease in the contact angle, as shown in Fig. 7.

3.2. Raman spectroscopic and wear characteristics

The Raman spectra of the samples were fitted intotwo Gaussian peaks; one centered around 1350 cm−1

(the D-band) and the other around 1560 cm−1 (theG-band). Plotting the G-band peak position (right-hand ordinate) and the band intensity ratio Id/Ig (left-hand ordinate) against substrate bias gives Fig. 8.According to theoretical work by Beeman et al. [8], ashift of the Raman G-peak to a lower frequency indi-cates a greater sp3/sp2 bonding ratio: this is seen in Fig.8. Decreasing Id/Ig indicates reduction in the total num-ber and/or size of graphitic micro-domains, or increasein the number of four-fold coordinated carbon atoms(sp3/sp2), as demonstrated by Lee and co-workers [2],Marchon and co-workers [1] and Ager III [9] in hydro-genated amorphous carbon films with increasing hydro-gen content. The influence of the bias power is believedto originate from the increased degree of bombardmentas the bias power is increased.

However, with increasing target power density(through reduction of the number of targets from threepairs to two pairs and then to one pair), an increase inId/Ig (open circles) and G-Peak wave number (solidtriangles) were observed (Fig. 9). This means that in-creased target power density hindered sp3 formation orpromoted sp2 bonding, as also indicated by the in-creased wear rate (the solid points Fig. 9).

Fig. 8. Band intensity ratio (Id/Ig) and G-band position decrease withsubstrate bias, indicating increased four-fold carbon coordination(sp3 bonding).

Fig. 9. Effects of sputtering target power density on the Raman shiftand the wear properties.

Fig. 10. Effects of the gas flow rate on the Raman shift and the wearproperties: no significant trend is observed.

with nitrogen concentration, or whether the disks weretextured.

S. Zhang et al. / Journal of Materials Processing Technology 89–90 (1999) 556–560560

Fig. 11. The average residual lubricant thickness as a function of thegas flow and the substrate bias voltage.

4. Conclusions

Magnetron sputtering of graphite target in the pres-ence of nitrogen was employed to deposit a-C:N film onAl–Mg rigid magnetic disks. The effects of the process-ing parameters on the growth, the contact angle and theRaman spectra of the carbon overcoat were studied.The following gives the main points again:1. Increasing flow rate (chamber pressure) or bias

power suppresses the film growth.2. The surface energy of the film increases with flow

rate and decreases with bias.3. Increasing substrate bias decreases the G-band peak

position and the band intensity ratio (Id/Ig), indicat-ing favored carbon four-fold coordination.

4. In the range of thickness coated, no obvious correla-tion is found between the coating roughness and thedeposition parameters (flow rate, substrate bias,etc.).

Acknowledgements

This work was funded by the Singapore NationalScience and Technology Board via Gintic upstreamproject U96-P-059 in collaboration with the School ofApplied Science, Nanyang Technological University.Technical assistance from E. Fong and Y.C. Liu isgratefully acknowledged.

References

[1] B. Marchon, P.N. Vo, M.R. Khan, J.W. Ager III, IEEE Trans.Magnet. 27 (6) (1991) 5160–5162.

[2] J.K. Lee, M. Smallen, J. Enguero, H.J. Lee, A. Chao, IEEETrans. Magnet. 29 (1) (1993) 276–281.

[3] E. Teng, C. Jiaa, A. Eltoukhy, Surf. Coat. Technol. 68/69 (1994)632–637.

[4] F.L. Freir Jr., C.A. Achete, G. Mariotto, R. Canteri, J. Vac. Sci.Technol. A12 (6) (1994) 3048–3053.

[5] B.B. Lal, M.M. Yang, J. Chao, M.A. Russak, IEEE Trans.Magnet. 32 (5) (1996) 3774–3776.

[6] H. Tanaka, K. Gomi, Y. Miyake, IEEE Trans. Magnet. 29 (1)(1993) 270–275.

[7] M.Y. Chen, X. Lin, V.P. Dravid, et al., Surf. Coat. Technol.54/55 (1992) 360–364.

[8] D. Beeman, R. Lynds, M.R. Anderson, Phys. Rev. B 30 (1988)870.

[9] J.W. Ager III, IEEE Trans. Magnet. 29 (1) (1993) 259–263.

Similarly, Fig. 10 plots the same properties as afunction of the flow rate. In general, there were nosignificant trends observed. This means that the forma-tion of sp3 or sp2 bonding had little to do with the rateof gas flow (i.e. the chamber pressure). However, it wasalso interesting to note that within the fluctuation,samples having greater Id/Ig values also exhibited higherwear when put under the abrasion test, and vice versa.This again supports the observation that decreasingId/Ig is associated with increasing sp3 coordination.

3.3. A6erage residual lubricant thickness

After debonding, the average residual lubricantthickness data were plotted, in Fig. 11, as a function ofthe gas flow rate and RF induced substrate biasvoltage. It is seen that the flow rate (or chamberpressure) does have some effects on bonding. The num-ber of dangling (unbonded) carbon atoms decreases athigher pressure because there is a higher probability forthem to bond with one another. Therefore, the totalnumber of dangling bonds on the surface decreaseswith increase in chamber pressure or gas flow rate,resulting in a decrease in the average residual lubricantthickness. On the other hand, bias power seems topromote chemical bonding (the top x-axis). Althoughthe mechanism is unclear, the increase in chemicalbonding or the number of dangling bonds are associ-ated with increased bombardment and sp3 bonding (c.f.Fig. 8).

.