Synthesis and characterization of amorphous carbon nitride films

!'! ELSEVIER

ThinSolidFilms290-291 (1996) 94-98

Synthesis and characterization of amorphous carbon nitride films B.C, Holioway a, D.K. Shuh i,, M,A. Kelly a, W. Tong ", J.A, Carlisle c, I. Jimenez ~, D.G.J. Sutherland e, L.J. Terminello c, p. Pianetta a, S. Hagstrom n gtal~ard University, gtm~rd, CA 94305.2205, USA h Lawrence Berkeley National Lab~ratory, Chemical $t'ienl:es Division. Berkeley, CA 94720, USA c Laee~nce Livermore National Laboratory. Livermnre, CA 94551, USA

Abstract We report the high-pressure, 1066 Pa (g Ton-). chemicalvapor deposition (CVD) s)'nthesisof amorphouscarbon nitridc films using a d.c, glow discharge technique. X-ray photoelectron spectroscopy and near edge X.ray absorption fine structure were used to stud)' the film stoiehiomelryand bonding as a function or substrata temperature and flax of hydrogen to the growth surface, Experiments show that the lilm sloichiomeltyis constant with substrata temperatureup to fifi0°C, above which filmgrowth was not observed. The additionof small amounts ( 1,5 at,%) of molecular h),drogencauses poisoning of film growth. Multiple sp2 bonding stales, with little spa bonding, were present in the films.Scanning electron microscopy of the films reveals an unusual filamentarygrowth pha~.

Keywonls: Carbon allrida;

Amorphous fihns; X-ra3, photoelectron spectroscopy; Near edge X.ray absorption fine structure tNEXAFS)

1, Introduction As early as 1979, amorphous carbon nitride films were considered for wear resistant applications [ 1I. Morerecently, due to theoretical calculations by Liu and Cohen which predict a metastable crystalline form of CAN4, interest in CN films h~s been rejuvenated [21. Researchers have been able to synthesize films that eontairt particles which yield diffraction patterns similar to those predicted for crystalline carbon nitride [3,4]. However. to date, no one has been able to produce large enough quantities of these nanoparticles for verification of the actual structure and composition. Results that have come out of the pursuit of crystalline carbon nitride suggest that some of the amorphous films produced could be used as wear resistant coatings [ 5,6]. Most of the amorphous carbon nitride films (hereafter referred to as CN films) are grown at low pressures (0.13-13.33 Pa) with high-energy ion impingement (5-10000 eV) [5,7,8]. Low-pressure growth has led to thin (5-1500 nm) [5,9] CN films which usually have nitrogen to carbon ratios of less than 0,4, unless a directed high energy ion beam of nitrogen is used [ 10,8], The films are generally flat and featureless [5,6]. Most researchers have used solid graphite and N~ gas as the carbon and nitrogen sources, although Song et al. have had success growing CN films using NHs as the nitrogen source [ t 1]. In this paper we report a preliminary investigation of tifick ( 10 Izm) CN films grown at a pressure of 1066 Pa (8 Torr). 0040-6090/96/$15,~© 1996ElsevierScienceS.A,All rightsreserved Pf180040.6090 (96) 09203-6

X-ray photoelectron spectroscopy (XPS) and near edge Xray absorption fine structure (NEXAFS) were used to study the film composition and bonding as a function of substrata temperature and flux of hydrogen, The film morphology was examined using scanning electron microscopy (SEM) and X-ray diffraction (XRD).

2. Experimental description The CN films for this study were grown using a d.c. glow discharge of nitrogen and helium over a graphitic carbon target. Single-crystal
B.C. HoJIo~,ayel al. I Tkin Solid Films 290--2Pl (J996) 9 4 - ~

He/H=Gun

He~ls Ou

I101'eplllla Tll'glll I I

95

binding energy interval values from the li~mtum and an empirically determined peak width [7,13], According to the literature the CN, peak in Fig, 2(a) and (b) mpments stoichiomelric C3Na while the CNy peak represents a carbon-rich plisse of carbon nitride [7]. However. it is evident from Fig. 2(c) and (d) that the spectra also fit well to a deconvolotion of peaks based on non-sps chemical shifts [ 13].

T +

I + Iiq=Oan


Fig. I. Schematicrepresentationoftheupl~ratususedforamoqthouscarbon nltride (CN) fdm I~rowth.Heliumand hydrogenwereintroducedintothe chamberoutsidethe growthapparatusand allowedto diffuseintothe glow di~¢hm'geregion,

¢Na

Table 1 Typlcnlexpedment~conditionsduringgrowthof the CNi11ms;vmqallons in the substmtetemperatureanddepositiontimeare the resullofperametrlc studiesand notcents'ellimitations Pressure(Ton) Substrntetelill~t~tUre(a) Plasmavoltage(V) PlumP,current(mA) Nitrogencontent(%) Heliumconten!(%) Totalflowrate(mndtrd em'~rain- I) Depositiontime(s)

8 70--600 700die. 60 26-28

72-74

BindingEnergy(eV)

i

C~X CNy

220

30-3600

[]oodlgafl tO flood uniformly the sample surface with 0.5 eV electrons. Near edge X-ray absorption fine structure (I~XAFS) measurements were done in the total yield mode at the Stanford Synchrotron Research Laboratory (SSRL) beamline 8-2, The NEXAFS spectrometer is an ultrahigh vacuum (UHV) chamber equipped with a precision manipulator which has a small antechamber that functions as a sample introduction load lock or preparation chamber. NEXAFS spectra were collected by recording the current from the sample while monitoring the photon flux via photoemission from a gold grid with an electron multiplier. The resistivity of the films results in an increased noise level, but does not cause charging-related energy shifts in the NEXAFS spectra. HEXAFS spectra are shown as collected, without background subtraction, A generalized focusing diffractometer (GFD) set-up was used for all XRD measurements. The system is described in detail elsewhere [ 12]. An improvement to the system for this measurement was the addition of a two-dimensional linear detector to increase the signal to noise ratio by allowing the synchronous collection of signal. A Hitachi $2500 scanning electron microscope with a 15 kV electron source and a secondary electron detector was used for all micrographs. To alleviate charging effects the CN films were coul~d with a 10 nm conductive film of gold and palladium for all SEM characterizations.

3. Results

Fig, 2 shows typical high resolution carbon (Is) and nitrogen (Is) core level spectra, The peaks were fitted using the

399 401 4013 BindingEnorBy(eV)

393 39S 39"/

405 4107

(¢)

12

"

~.80 ~

~ 2,/16 288 290 292 s'94 BindingEner~ (eV)

(d) N2

395 397 399 401 403 4~ 40? Binding£ner~v(eV) Fig, 2, Typicalcarbont Is) and nitrogen(Is) corelevel sgectm.For (n) 3 CNy(cm'b~rich)inaccor~tnoe and (b), peaksmelabeledasCN,(sp')and withMemon'sassignments[7]. (c) and(d}showpeakfittingfromnnregnn p]asmaenhencedpolyimidn[ 13], NolethatcomponentsNI andN2are not assignedinspecificnitrogengroupsowingt~the¢omplexilyofthe ni|mgeu functionality[13], 393

96

8, C Hnlloway el el, / Thin $o'id Fil.,gr290-291 (I 996) 94~9~

(a)CJrbon K Edge II I+ I~I i a Io io"

DIt~#d

1.5

l 2; 0 03

-, Z00

i

400

h

~1

6(]O

Substrata Teropemture ('C) Fig, 3. Variation of carbon to nitrogen ratio with temperature. No filmgrewlh was observed above 600 =C. The oxygen content ofth~ film was ignored in calculating the atomic ratio of carbon and nitrogen.

XPS spectra of carbon nitride films grown at different substrata temperatures were used to determine the carbon to nitrogen ratio as a function of temperature. All samples showed the presence of oxygen, although light etching of the substrata by 5 keV argon ions reduced the atomic percentage of oxygen from as high as 15 at,% to 1-2 at,%, TheCN films used in this study were not etched, to prevent potential alterations of the film structure a.d stoichiometry, However, oxygen was not included in all calculations. Fig, 3 shows the variation of the carbon to nitrogen ratio ([C]/[N] ratio) with suhstrate temperature. The [C]/[N] ratio is relatively mvariant with surface temperature over the range 70-585 %; above 585 °C, film growth was not detectable, XPS was also used to obtain a lower hound on the resistivity of the films, A surface charge shift of 72,3 eV was measured from a film with respect to a grounded substrate, From this value a lower bound of the film resistivity was calculated to be 10~ i'~ cm. NEXAFS carbon and nitrogen K-edge total yield spectra from two typical films are shown in Fig. 4, For reference, hexagonal and cubic boron nitride nitrogen K-edge spectra as well as diamond and highly ordered pyrolitic graphite (HOPG) carbon K-edge spectra are included [14], Comparing the CN spectra with the reference spectra shows that less than tO at.% of the bonds in the CN films are sp J. Hydrogen was added to the gas mixture in an attempt to improve the film composition by preferentially etching nonsps bonded carbon. Such preferential etching is observed for diamond growth. The addition of as little as i.5% molecular hydrogen in the growth chamber caused total poisoning of film growth. System limitations did not allow for a lower concentration of hydrogen to be added. XRD performed on two typical films failed to show any crystalline carbon nitride diffraction peaks, although small graphitie crystals were detected, Searches were conducted for a and [J phase crystalline carbon nitride peaks from the structures predicted ', Gun and coworkers [15], However, the morphology of these films, as shown in Fig. 5, is quite dif-

285

290 395 Photon Energy (eV)

300

(b) Nitrogen K-edge

ts I0 ~* It Is Ill e+

[

.

400

405 410 41~ Photon Energy (eV) Fig+ 4. NEXAFS total yield spectre for typical amorphous carbon ninld+ (CN) filrm. (a) is the carbon K edge while (b) is the nitrogen K edge. Hexagonal boron nitdde ( h -BN ), c nbic boron nitride ( c-BN ), diane nd and highty ordered pyrolitie graphite (HOPG) are shown for reference. Temperatures cited am the film substrata growth temperatures.The spc¢Ira were normalized individually to the largest respective feature,

fereat from that reported by other researchers [ 5,6], The films have filamentary structures which are approximately 7,5 p,m i,l length and 0.5 gm in diameter. The reverse taper present is typical of CN fifms deposited by this reactor, Filamentary structures were observed in as little as 60 s of growth.

4. Discussion

q t,e XPS C( I s) and N( I s) high rasolution spectra clearly show that multiple chemical bonds exist between the carbon

K C. ltolloway et al. I Tfiin $olid Fihns 2~-291 (1996) 94-98

Fig,5. Scanningel~tmu mierographof u typiualamorphouscarbonnitfide filra. Filamentsam 03 ~m in dtamemrat thu widestpointand 7.5 ~m in height,Thereverseta~cris typicalof all our filmsstudied. and nitrogen. Marten el el, have developed a criterion to determine ratios of sp3 to sp: bonded material in carbon nitride films [7 ]. They have assigned CN~ as an sp3 phase of carbon nitride based on comparison with carbon in Urotropine. According to Marton's criterion our films should have 50 at.% CN~, as shown in Fig. 2(a) and (b). However, NEXAFS shows that less than 10% of the bonds in the CN films studied are sp3 in nature (CN,~). This suggests that an alternate fitting scheme is needed for these films, The scientific literature contains XP$ spectra with a wide range of chemical shifts for organic carbon and nitrogen groups such as amine (C-N), iminc (C=N), and nitrile (C--N) [13,16,17]. For instance, the XPS high-resolution C(Is) and N(Is) core level spectra for the CN films are also tilted well by a deconvolution based on chemical shifts from Nz plasma treated polyimide surfaces, as shown in Fig. 2(c) and (d) [131. The CH films produced are obviously not polyimide, However, this fitting scheme shows that it is possible to have shifts as large as those Marten attributes to sp3 bonded carbon nitride and. still not have any sp3 bonding present in the carbon nitridc films. It should be clearly noted that the films characterized in this study were grown at a much larger growth rate, higher ambient pressure, and lower ion energy than the CN films Marten studied [ 7]. Because of this, it could be expected that the films produced would be fundamentally different from the films that Marten analyzed, It is possible that Marton's analysis is valid for a certain range of films, although not for all CN films produced. In a system dthis pressure and plasma energy, substrate temperatures on the order of 800 °C are required to create CVD diamond [ 18], suggesting that more energy is required to form an sp3 bond on the film surface.

9'1

Fig. 3 shows that increasing the temperature above 600 °C to try to achieve sp"~bonding is not possible, since no film growth occurred above 600 °C, The high-temperature growth limit is probably due to the formation of volatile CzN2. Cuomo et at. performed high-temperature mass spectrometry on similar films [ I ] and found that the mass 52 (C2Nz) signal peaked at 630 °C, This corresponds to the high-temperature limit determined in our work, A possible role of hydrogen in the system is to expedite the creation of hydrogen cyanide, I-ICN. The ratio of the atomic fiux of carbon leaving the target to the atomic flux of hydrogen entering the system is 30:1. It is reasonable to assume that all the hydrogen is consumed in the creation of HCN, since HCN creation is a thermodynamically favorable reaction. The remaining carbon could easily be accounted for by considering the gas phase creation of cyanogen, CzN2.The creation of cyanogen is also thermodynamically favorable, even over the formation of HCN, Therefore, it is reasonable to assume that some fraction of the carbon sputtered is incorporated into cyanogen, Both cyanlde and cyanogen are volatile at the pressures and temperatures used for growth, This would also be fully consistent with Cuomo's high-temperature mass spectrometry work [ 1], Song ot el, had a similar carbon to hydrogen atomic flux ratio (8:1) yet still achieved growth [ 11 ], However, the hydrogen was at a much higher energy (200 eV or greater) and impinged directly onto the growth surface rather than allowing a large amount of ,goa~ phase reactions to occur. The film morphology is consistent with previously reported work [ 19] and Thornton's model for film growth [20]. It is evident that the film growth in this system is a far Zone 1 type film growth, which is characterized by atomic shadowing and very limited surface diffusion [ 20]. The sb halowing promotes open boundaries and yields the filamentary structure seen. Possible shadowing mechanisms include the following: too high a working gas pressure; uneven surface electric fields; and/or preferential growth sites on the surface. System limitations do not permit more complete parametric variations for verification of these mechanisms.

5, Conclusion

Films of amorphous carbon nitride were grown at a pressure of 1066 Pa (8 Tel'r). X-ray photoelectron spectroscopy (XPS) spectra from the CN films can be dcconvoluted by using chemical shifts from known organic carbon-nitrogen groups, Near edge X-ray absorption fine structure (NEXAFS) spectra show that the films have multiple sp2 bonding slates, with less than I0 at.% sp 3 bonds, which is consistent with the interpretation of the XPS spectra. Varying the temperature of the substrata does not cause an appreciable change in film properties until a high-temperature limit is reached at 600 ~C, above which no film growth is observed. A possible explanation for the high temperature growth limit is the ereation of volatile C2N2. Furthermore, the addition of molecular

98

8,C Hollowuy et aL / Thia Solid Films 290-29I {1996) 94-98

hydrogen causescomple." poisoningof film growth, proba-

bly througha gas phase reactionwhich producesHCN. Scanning electron microscopy shows that the films have a filamentary morphology. The data collected suggest that the films are a polymeric, paracyanogen-likematerialwithexcess carbon incorporatedas graphite.

Aeknnwlodgemenlts The authors thank Bob Jones for his help with the SEM and Glenn Waychunas for the XRD assistance. This work was performed in part at Stanford Synchrotron Radiation Laboratory which is operated by the US Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences.

References [IJ J J, Cuomo, P,A, l..¢aTy,D, Yo, W, Rentcrand M, Frisch,], Vac, Coci, Tech,oL A, 16 { 1979) 299-302, [2] A.Y. Liu and M.L. Cohen, Phys. Roy. E 41 (1990) 10727-10734. [3] K.M, Yu, M,L Cohen, B,E, Huller,W.L, Hanson, A.Y. Liu and I,C. Wo, Phy,~.Roy. B, 49 (1994) 5034-5037. [4] C, NI .. ".,Z, LUand C.M. Lieber, Selance, 261 (1993) 334-337.

[5] EC Culiangco, D, Li, Y,-W, Cllu~ toldC,S. Bhatia,Tribological behaviorof amorphou.s carbon nitrideovercoatsforman.tic lhio-film rigid disks, In P,A, Sutor ted.), SrLD'A$,i./E TRbotoD Con:, Orlando, FL, ASME, 1995, New York, NY~p. L [6] M,Y. Chan, X. Lin. V. P,Dmvld and Y,W, Chang, Surf,Coat. Tech.ol,, 5¢-$5 (1992) 3~0-364. [7] D. Mmlan, KJ, Boyd, A.H, At-Bayati, S,S, T~orov and J.W. Rabalals, Phys. Rev. Lett,, 73 0994) 118-121, (8] K, Ogata, J,Femando D inizChobac[ an.dF, Pujimo[o,J, Appl, Phys,, 76 (1994) 3'791-3796, [9] D. Li, C. Yip-Wah, W. Ming-Show and WD, Sproul,J, Appt, Pl~y,¢., 74 (1993) 219-233. [10] S,S, Todorov, D. Ma.rtan, KJ, Hoyd, A,H, Al-Bayati and J.W. Rabalais, J, Vac, $ci. 7"ec/mot.A, 12 (1994) 3192-31cJ9, [ ! I ] H.W. Song,F.Z.Cui, X.M. He,W.Z. Li andH,D, Li, J, Phy,¢,:Co,dens, Matter, 6 (1994) 6125--6129. 12] P,A, Flinr,and G.A, Way.banns, 1 Vac. Sd. T~cha.l. B, 6 t 1958 ) 1749-1755. []3]J.E. Klem~rg.Sapieha, O,M. Kuttel, L, Martinu and M.R, Wertheimcr,I. Vac. 3ci. Teehnol. A, 9 ( 1991 ) 2975, [ |4] L J. Terminello, A. Chaik~, D.A. Lapiano.Smitk, G.L. Doll and T. Sato, J, Va¢. Sol. TechnoL A, 12 ( 1994} 2, [ 15] Y. Gun and W.A. Goddard Ill,Chem, Phys. L~II,,237 ( 1995 ) 72-'/6, [ 16] F, Rossi et hi,, J, Ma:er. Re:., 9 (1994) 2440-2449, [ 17] S,N, Kumar,F, Gaillard, G, Bouyssouxand A, Santo, Symh, Met,, 36 (1990) 111-127, [ 18l M.A. Kelly, DS. Olson. S Kapoar and S,B, Hagstrom, Appl. Phys. Lelt., 60 ( 1992} 2502-2504. [ 19] S, Kornarand T,L, Tansloy,J. @pl. Pl*ys,, 76 ( [994 ) 4390-4392. [20] J,A,Thorto.,J. V~c, Sci.Teclmol,A, 4 (1986) 3059-3065.