Vacuum 63 (2001) 433–439

Versatile microwave PECVD technique for depositionof DLC and other ordered carbon nanostructures

Sudheer Kumara, C.M.S. Rauthana, P.N. Dixita, K.M.K. Srivatsaa, M.Y. Khanb,R. Bhattacharyyaa,*

aThin Film Technology group, Electronic Materials Division, National Physical Laboratory, Dr. K.S. Krishnan Road,

New Delhi-110 012, IndiabDepartment of Physics, Jamia Millia Islamia University, New Delhi-110 025, India

Accepted 14 March 2001

Abstract

A remote microwave plasma enhanced chemical vapour deposition system was optimized to deposit different types ofdiamond like carbon films, as well as, some ordered carbon structures like carbon nanotubes (CNTs) and

nanodiamonds. These films were obtained at room temperature using C2H2+Ar as the feedstock and at different sets ofparameter space. While some DLC films had the required optical parameters to provide an anti-reflection coating on Sisolar cells (1.9% enhancement of efficiency), others were transparent in the visible region. CNTs and nanodiamond

structures were obtained simply by varying the applied RF bias at an optimized microwave power of 25W. Hardnessand stress of these films were in the range of 9–20 GPa and o1GPa, respectively. The root mean square roughness inDLC films varied from 0.06 to 0.12 nm. Raman spectroscopy was performed to identify different phases present in the

films. The grain size of nanocrystalline diamonds varied from 40 to 120 nm. The CNTs were found to be multiwalledwith their diameters varying from 23 to 30 nm. r 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Diamond like carbon (DLC); Carbon nanotubes (CNTs); MW-PECVD; Atomic force microscopy (AFM)

1. Introduction

Recent years have seen many significant devel-opments around different type of carbon struc-tures such as diamond like carbon (DLC) thinfilms, nanoclusters, buckyball (fullerene) andnanotubes [1–7]. While there are many potentialindustrial applications of DLC [2,3], carbonnanostructures (carbon nanotubes and nanodia-

mond) are expected to find applications in nano-electronics [8–10] and flat panel display devicetechnology [11]. The most preferred technique todeposit such nanodiamond and DLC films isgenerally a suitable plasma assisted chemicalvapour deposition process. Films so grown,however, show a large residual stress and are,therefore, very prone to delaminate. A microwaveplasma-enhanced chemical vapour deposition(MW-PECVD) technique, adopted by Martinuet al. [12], has been found to produce DLC filmswith extremely low stress values. Such low residualstress DLC films could also be reproducibly

*Corresponding author. Tel.: +91-11-5787872; fax: +91-11-

5852678.

E-mail address: rbhat@csnpl.ren.nic.in

(R. Bhattacharyya).

0042-207X/01/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved.

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

produced by a pulsed plasma CVD technique aswell as by the use of a saddle field source [13]. Ingeneral, by decoupling the plasma generation andbias application processes in a remote MW-PECVD system, one is able to independentlycontrol plasma density as well as ion energy. Thepresent authors have developed a similar facility atthe National Physical Laboratory to depositamorphous carbon films. This MW-PECVD sys-tem uses a microwave source at 2.45 GHz and aradio frequency (RF) biasing facility that works at13.56 MHz. By undertaking a detailed optimiza-tion study involving suitable plasma diagnostictechniques, a great deal of tailorability of thematerial being grown and their uniformity over atleast a 75 mm diameter silicon wafers have beenensured.

In this paper we report the observed versatilityof this remote microwave/RF plasma enhancedchemical vapour deposition (PECVD) system todeposit a variety of DLC films and other carbonstructures. This includes transparent DLC films inthe visible region, DLC films suitable for anti-reflecting applications for Si solar cells, nanodia-monds and carbon nanotubes.

2. Experimental details

Ophthalmic glass and silicon substrates used inthe present study were cleaned, following standardcleaning procedures for electronic processing, andwere placed down stream upon an insulatedsubstrate holder which served at the same time asthe RF powered electrode. Before starting theactual deposition process, an argon discharge wasmaintained, at @400 V of RF self-bias for 10 min,to remove the surface impurities from the sub-strates by energetic argon ion bombardment.Subsequently, feedstock (C2H2+Ar gases mixture)is released downstream, mass flow controlled andthe total gas pressure was adjusted using aBaratron (M/s MKS make capacitance man-ometer) and throttle valve controller. We haveused a wide range of parameters where microwavepower was varied from 25 to 250 W along withvariation in RF bias from @100 to @300 V.However, the feedstock ratio (FAr=FC2H2

) was

maintained constant at B0.067 except to deposittransparent DLC films where it was B1.0.

The residual stress in the films was calculated bymeasuring the curvature induced in the substratesdue to the film deposition. The curvature wasmeasured using a laser scanning technique, speci-fically set-up for this purpose in our laboratory.The hardness of these films was measured on aninstrument equipped with a Knoop hardnessindenter under 50 g load (Zwick Gmbh and Co.,Germany, Model Zwick 3212). The surface mor-phology and root mean square (RMS) roughnessof the films were investigated by atomic forcemicroscopy (AFM) on an equipment from DigitalInstruments Inc., USA (Model Nanoscope II) atCEERI, Pilani. This equipment is enclosed in anairtight box to avoid perturbations due to roomair circulation. Scanning for recording topogra-phical details was carried out in a constant forcemode. Images consist of 400 scans of 400 pixelseach. The system was calibrated using a standardthin film mica sample, to monitor thermal driftand piezo-effects prior to the examination ofstructures. Raman scattering spectroscopy wasperformed, at room temperature, by using anonpolarized argon laser (l ¼ 514:5 nm) in analmost back-scattering geometry. Low powerbeams, typically 200 mW at the sample surface,were used to avoid sample heating and anyapparent damage to the films during analysis.

3. Results and discussion

Adhesion of the films on the substrates being themost important consideration for the stability andreliability of any device made out of such films, thevery first measurement that we performed routi-nely, was careful stress measurements on thedeposited films. We were able to choose a set ofprocess parameters to achieve consistently lowstress DLC films in our dual frequency PECVDsystem. This optimization of the process para-meters alone allowed us to grow DLC films withless than 0.5 GPa residual stress, when the appliedmicrowave power is kept fixed around 25 W. Webelieve such observed low stress in these films isprimarily due to the decoupling of plasma genera-

S. Kumar et al. / Vacuum 63 (2001) 433–439434

tion process from the necessity of application ofthe required RF bias to the substrates (to impartoptimal energy to the adatoms). Fig. 1, depicts theresidual stress observed in these films deposited atvarious RF bias values, at an optimized micro-wave power of 25 W. As is evident from the figure,it is difficult to find any particular trend in stressvalues with the increase of RF bias to the subs-trates. Also, as can be seen in this figure, the resi-dual stress values in these films increase as theapplied negative RF self-bias increases, except at acertain RF bias (@200 V) where it shows a dip.One also observes such behaviour in RF self-biasdeposited DLC films and this may as well be anindication of onset of graphatization in these films.It is found that the hardness increases with theincrease in RF bias, from about 9 GPa to 20 GPa.Subsequent studies presented later show that themaximum hardness of 20 GPa was observed forthe films that consist of nanocrystalline diamondcrystallites.

3.1. Deposition rate

The variation of deposition rate (rd) with RFbias, at the optimized microwave power of 25 W, ispresented in Fig. 2. It is evident from the figurethat, the deposition rate increases with the increaseof substrate bias up to @150 V and beyond that itbecomes almost constant. This is understandablebecause in the microwave PECVD process, the

increase of negative RF bias will directly raisethe energy of the film forming precursors (ions)in the plasma impinging on the substrates as alsothe abundance ratio of the argon ions there. Theobserved constant deposition rate, as is seen in thefigure, may have resulted from the competitionbetween the process of deposition and etching ofthe growing films. The high energy argon ions willsputter etch the deposited film. The typical rd valuewas found to vary from 17 to 30 (A/s.

3.2. Optical properties

3.2.1. Optical transmission and highly transparentDLC films in the visible

Fig. 3 shows the optical transmittance of atransparent DLC film. This film was deposited atoptimized parameters for this purpose: a micro-wave power of 120 W and a RF self-bias of@100 V with the feedstock ratio (FAr=FC2H2

) ofB1.0. It is evident that optical properties of theDLC film could be suitably tailored so that verygood optical transmission is obtained in the visibleregion.

3.2.2. Anti-reflection coatings for silicon solar cellsTo act as an efficient anti-reflection coating, we

need to tailor the optical properties of the DLCfilms by correctly monitoring the thickness of theDLC film as also ensuring that films are of therequired refractive index and low absorption.These required films were deposited at a micro-

Fig. 1. Residual stress in the films deposited at various values

of RF bias at an optimized microwave power of 25W.

Fig. 2. Variation of deposition rate (rd) with RF bias at an

optimized microwave power of 25W.

S. Kumar et al. / Vacuum 63 (2001) 433–439 435

wave power of 120 W and a RF self-bias of@100 V. The thickness of the film was B650 (Aand corresponding refractive index value is 1.9. Asshown in Fig. 4, the short circuit current Isc of thechosen c-Si solar cells increased by almost 30%and correspondingly the efficiency, Z, increased byB1.9% on application of an optimized anti-reflecting DLC coating on these cells. Further,we observed the entire area of the 75 mm diasilicon solar cells to be uniformly blue in reflection.

3.3. AFM study

AFM micrographs shown in Fig. 5(a)–(d) pre-sent an excellent view of the surface morphologyas observed in different DLC and carbon films.The AFM characterization, at different stages ofRF bias variation (other process parametersoptimized and kept constant like applied micro-wave power B25 W and required dilution ofC2H2/Ar), reveals that the growing film assumesa variety of different structures. The film depositedat a RF self-bias of @100 V is a DLC film andexhibits a high degree of smoothness, without anymicrostructure, as shown in Fig. 5(a). Similarmorphologies were also observed for other DLCfilms that were obtained by suitably changing theprocess parameters. However, the variation in thesurface roughness across the deposited DLC films,scanned over a 1 mm� 1 mm area was found to be

different. The typical values for RMS roughness inthese DLC films varied from 0.06 to 0.12 nm.Shown in Fig. 5(a) is the micrograph for aparticular chosen DLC film that shows the highestdegree of smoothness observed during our inves-tigations.

The morphologies of the films deposited withhigher RF biases are remarkably different from theone that is generally observed for the conventionalDLC films as shown in AFM micrographsFig. 5(b)–(d). Unlike what is seen for DLC films,these AFM pictures show some interesting nanos-tructures. To confirm that what has been observedis not simply an artifact, AFM measurements wererepeated in two ways: first, at various places on thesame sample and second, on different samplesprepared under similar process parameters andstudied similarly. However, in all these experi-ments we achieved very similar results. It alsoensured that we have accomplished the desiredrepeatability of the process.

At low RF bias of @150 V some channel like(lane) patterns were observed (Fig. 5(b)), ran-domly distributed all over the film area withdifferent lengths, which we consider to be the onsetof CNT formation. It is clear from Fig. 5(c) thaton increasing the RF bias to @200 V, tube likestructures clearly emerge out at many places. Onecan also reasonably estimate the extent of curva-ture of these tubular structures. It is worthmentioning here that increasing RF bias to@250 V, leads to formation of nanodiamond

Fig. 3. Optical transmittance spectra of a transparent DLC film

deposited at a MW power of 120W and a RF self-bias of

@100V.

Fig. 4. Effect of anti-reflecting DLC film (deposited at a RF

bias of @100 at an optimized microwave power of 120W) on

c-Si solar cell.

S. Kumar et al. / Vacuum 63 (2001) 433–439436

films with grain size from 40 to 100 nm, as shownin Fig. 5(d). The AFM picture in Fig. 5(c) clearlyshows the presence of CNTs in the particular filmbeing studied. Here, the diameters of CNTs,marked as (a) and (b) in the micrograph werefound to be 23 and 30 nm, respectively, whichsuggests that they are multi-wall CNTs. It alsoindicates that the CNTs marked as (a) and (b) aremore than 1 mm in length. Fig. 6 is the typical 3DAFM image of the CNTs presented in Fig. 5(c).Looking minutely into this micrograph, we ob-

serve that many such CNTs, with relatively smallerdiameters, are present in this sample. However, wefeel that a more stringent control of the processparameters may yield CNTs of nearly equaldiameters.

We identify and attribute the structures inFig. 5(d) as nanocrystalline diamond grains in apolycrystalline matrix phase. The film showspronounced nanocrystalline structures of well-defined habits appearing mostly in hexagonalshapes. Similar structures were observed in dia-

Fig. 5. AFM micrograph of the films deposited with at an optimized microwave power of 25W but at different RF bias: (a) a DLC film

at a RF of @100V, (b) a DLC film with onset of nanotube formation at a RF of @150V, (c) a amorphous carbon film with

nanotubes at a RF of @200V and (d) a nanodiamond film at a RF of @250V.

S. Kumar et al. / Vacuum 63 (2001) 433–439 437

mond films synthesized by microwave plasma at300–7001C by Kamo et al. [14] and have beenidentified those as cubo-octahedral and octahedraldiamond crystals. The size of the diamondnanocrystals in the samples varied from 60 to100 nm. In the AFM micrograph, some irregularshaped structures can also be seen in this film. Webelieve that these structures may be non-diamondphases, graphite grains, or any others cross-linkingamorphous/polymeric carbon structures betweenthe diamond nanocrystallites.

3.4. Raman study

The typical Raman spectra, recorded at roomtemperature for a DLC film along with that of ananodiamond film, are presented in Fig. 7. TheRaman spectrum of DLC film in Fig. 7(a) showstwo peaks: a small peak around 1340 cm@1 and abroad peak at 1530 cm@1. The peak at 1340 cm@1

may be due to the sp3-hybridized carbon contentsin the film, whereas, the other peak at 1530 cm@1

may be due to disordered sp2 hybridization. Thecrystalline graphite exhibits a Raman peak at1580 cm@1 [15]. This peak position has beenobserved to shift due to disordered sp2 carbonand has been reported to occur anywhere from1521 to 1600 cm@1 [16]. The Raman spectrum ofthe nanodiamond film in Fig. 7(b) is characterizedby a sharp peak centred at 1314 cm@1 and otherpeaks centred at 1479, 1523 and 1579 cm@1 sitting

on a broad hill. The two peaks centred at 1523 and1579 cm@1 could be due to the presence of smallordered crystallites or amorphous graphitic carbon[17]. The origin of the other peak centred at1479 cm@1 on the broad hill is not known. Someother workers have also reported a similar peak at1475 cm@1 [15,18–20], and have attributed it dueto the diamond forming precursors. We attributethe very intense and sharp peak centred at1314 cm@1, to the presence of hexagonal nano-crystalline polymorph diamond structures. Thisstructure was also confirmed by glancing angleX-ray diffraction study which will be reportedseparately. The completely sp3-hybridized poly-morph of diamond has been reported [15,21] toshow a Raman shift of 1315–1326 cm@1. In ourfilm, the value of line-width was found to be5–7 cm@1. The higher line width as compared to2–3 cm@1 observed for single crystal diamond [21]is, perhaps, due to the occurrence of strain andcrystalline imperfections/defects in the films.

4. Conclusions

In conclusion we have set up and optimized adual microwave (MW)/radio frequency (RF)PECVD system and could reproducibly obtain a

Fig. 6. 3D-AFM micrograph of the CNTs in Fig. 5(c). Fig. 7. Raman spectra of two films deposited at an optimized

microwave power of 25W but at different RF bias: (a) a DLC

film at a RF of @100V and (b) a nanodiamond film at a RF of

@250V.

S. Kumar et al. / Vacuum 63 (2001) 433–439438

variety of DLC films whose properties could betailored to a great extent for some well identifiedapplications. Deviating from these rather well triedout applications, we could explore the parameterspace and identify where, in the same dualMW/RF PECVD system, one can reproduciblygrow a variety of ordered carbon structures likenanodiamond films and carbon nanotubes.

This versatility of a decoupled microwavePECVD system (i.e. plasma generation and bias-ing the substrates independently) has been veryclearly demonstrated for depositing a range ofDLC films and some more ordered carbonstructures. Further work on the other applicationsof such films is under investigation. Just tomention one, we have observed some extendedvoltage tunability of the capacitance of nanodia-mond films for possible varactor application,which will be reported in a separate article shortly.

Acknowledgements

The authors are grateful to the Director NPL,New Delhi for his permission to publish this paper.We wish also to acknowledge Dr. R. Chopra andMr. T.K. Bhattacharyya both NPL, New Delhi fortheir help and Dr. R.P. Gupta, CEERI, Pilani forfruitful discussion. Financial support from theDepartment of Science and Technology, Govern-ment of India to carry out this work is gratefullyacknowledged.

References

[1] Lifshitz Y. Diamond Related Mater 1999;8:1659.

[2] Grill A. Diamond Related Mater 1999;8:428.

[3] Bubenzer A, Dischler B, Brandt G, Koidl P. J Appl Phys

1983;54:4590.

[4] Iijima Sumio. Nature 1991;354:56.

[5] Wang XK, Lin XW, Dravid VP, Ketterson JB, Chang

RPH. Appl Phys Lett 1993;62:1881.

[6] McEuen Paul L. Nature 1998;393:16.

[7] Odom Teri Wang, Huang Jin-Lin, Kim Philip, Lieber

Charles M. J Phys Chem B 2000;104:2794.

[8] Collins Philip G, Bando Hiroshi, Zettl A. Nanotechnology

1998;9:153.

[9] Cheung CH, Hafner JH, Odom TW, Kim K, Lieber CM.

Appl Phys Lett 2000;76:3136.

[10] Saito Y, Hamaguchi K, Hata K, Uchida K, Tasaka Y,

Ikazaki F, Yummura M, Katsuya A, Nishina Y. Nature

1997;389:554.

[11] Gulyaev YuV, Chernozatonskii LA, Kosakovskaya ZYa,

Sinitsyn NI, Torgashov GV, Zakharchenko YuF. J Vac Sci

Technol B 1995;13:435.

[12] Raveh A, Klemberg-Saphieha JE, Martinu L, Wertheimer

MR. J Vac Sci Technol A 1992;10:1723.

[13] Kumar S, Dixit PN, Sarangi D, Bhattacharyya R. J Appl

Phys 1999;85:3866.

[14] Kamo Mutsukazu, Sato Youchiro, Matsumoto Seiichiro,

Setaka Nobuo. J Crystal Growth 1983;62:642.

[15] Bachnann Peter K, Weichert DU. In: Clausing RE,

Horton LL, Angus JC, Koidl P. editors. Diamond and

diamond like films and coatings, Nato ASI Series B

Physics vol. 266, p. 677. Plenum Press, New York.

[16] Nemarich RJ, Sorin SA. Phy Rev B 1979;20:392.

[17] Dillon RO, Woollam JA, Katkanant V. Phy Rev B

1984;29:3482.

[18] Nemarich RJ, Glass JT, Lucovsky G, Shroder RE. J Vac

Sci Technol A 1988;6:1783.

[19] Jou Shyankay, Doerr Hans J, Bunshah Rointan F. Thin

Solid Films 1996;280:256.

[20] Varbrough WA, Roy R. In: Badzian A, Geis M, GJ.

editors. Diamond and diamond like materials, Extended

abstracts vol EA-15. Material Research Society, Pittus-

burg; 1988. p. 433.

[21] Knight DS, White WB. J Mater Res 1989;4:389.

S. Kumar et al. / Vacuum 63 (2001) 433–439 439