Thermal gas effusion from diamond-like carbon films

Diamond and Related Materials 6 ( 1997) 1830-1835

Thermal gas effusion from diamond-like carbon films Manju Malhotra,

Satyendra

Kumar

*

Dr partmerit of’Physics. II lim Instiiute of Technology, Kunpur-208016, India Received 5 F. *oust 1996; accepted 4 June 1997

Abstract Good-quality diamond-like carbon films ( - 6 at.% H,, 2400 kg f/mm2 microhardness, 2.7 eV bandgap, highly insulating) have been obtained by the DC glow discharge decomposition. of acetylene. Mass spectroscopic thermal effusion measurements were carried out on the films deposited under different deposition conditions. Analyses of hydrogen in conjunction with hydrocarbon effusing species yield information on the microstructure and nature of C-H bonding configurations. It is shown to be a useful analytical too; to study hydrogenated amorphous carbon fihns of different microstructures varying from polymer-like to diamondlike. 0 1997 Elsevier Science S.A. Keywords: DC plasma CVD; Diamond-like

carbon; Microstructure;

1. Introduction

Amorphous carbon films prepared by plasma decompoof hydrocarbon gasses exhibit physical properties that may range from soft polymer-like to hard diamondlike, These films contain considcrablc amounts of bonded hydrogen (up to -_ 50 at.%) [I]. Soft carbon&d films may even have an H:C ratio greater than I 1:2].Hydrogen may bc prcscnt in the fihns in isolated or clustered form

sition

resulting from rnonohydridc or polyhydride type C-I-1 bonding. Hence, properties of the films critically depend upon the structure of the films, hydrogen content and the nature of C-C and C-H bonds. Thermal gas effusion is a sensitive technique to investigate the mode of hydrogen incorporation and microstructural features in the films. This characterization technique has been widely applied to hydrogenated amorphous silicon (a-Si:H ) [ 3,4]. However, unlike a-Si:H, the multiplicity of C-C and C-H bonding configurations leads to complex microstructure and related propertics in thin films of hydrogenated amorphous carbon (a-C:H ). In particular, a variety of hydra.carbons along with hydrogen are known to be evolved whcll a-C% films are heated above 350 “C [5-81. ElTcsion studies performed on RF-plasma-deposited a-C;H samples under different bias [7,8] and pressu,re conditions [7] suggested the role of microvoids in the -* Corresponding author. 09%9635/97/$17.00 0 1997Elsevier Science S.A. All rights reserved. Pit 509X-9635(97)00146-5

Thermal effusion

films in controlling their effusion characteristics. Hydrogen evolution experiments combined with infrared ( IR) absorption measurements have been shown to yield useful information on the microstructure and hydrogen incorporation in a-C:H films [9, IO]. We have carried OUKthermal gas e&ion studies on DC glow discharge dcpositcd diamond-like carbon (DLC) films in order to obtain information regarding microstructure and hydrogen bonding configuration in these films. The results are correlated with hardness and IR absorption measurements on the same samples. The activation energies for hydrogen evolution are obtained by the thermal analysis technique and are interpreted in terms of C-H and C-C bonding in the films.

2. Experimental details

Thin films of a-C:H were deposited on c-Si and Corning 7059 substrates by the DC glow discharge decomposition of acetylene in a reactor using parallel Flare diode configuration. Controlled deposition parameters include the hydrogen dilution, gas pressure, anode voltage, DC power and substrate temperature. For the sake of clarity in thermal effusion spectra, results are reported here on well-characterized representative samples having different microstructures. Type A samples were prepared using pure acetylene under high gas pressure conditions (p = 10.0 mbar, substrate tem-

M. hlcrllrotr~, S. Kunw ! Diatnond otd RrMrd

perature Tsab= 250 “C, cathode voltage V,= - 710 V ). On the cthc: hand, type B samples were prepared at relatively lower pressures ( y = 0.6 mbar) using hydrogen diluted acetylene (CzH2/C,Hz + H, =0.5) under optimized substrate bias (Vi = - 250V ) conditions ( Tsub= 250 “C, V,= - 7 10 V ). Table 1 shows the deposition parameters along with some microstructural properties for samples presented in this study. Optical transmission and reflection data on these samples were recorded in the wavelength region of 250-900 nm. Taut optical gap (&) and refractive index (n) were calculated from the optical data and are also given in Table 1. A computer-controlled mass spectroscopic thermal effusion system was designed to detect the effusion of gasses from thin films. Design considerations for higher sensitivity involve low volume ( - 230 cm3) of the sample holder attachment and low leak rate. Hydrogen effusion from a polycrystalline diamond film containing < 1 at.% hydrogen could be detected using this thermal effusion set-up. The sample was placed in a quartz tube and the system was evacuated using a turbomolecular pump. The quartz tube was baked at 600 “C and cooled prior to each experiment. The sample and thermocouple were made in similar contact with quartz tube for temperature measurements. The samples were normally heated to 1000 “C at a linear heating rate of 20 “C/min. The outdiffusing species, up to 12 in number. were monitored as a function of temperature using a quadrupole mass analyzer ( Ametek, Dycor Model MA200). Thermal effusion from the samples is governed by the detrapping and recombination of mobile species followed by the diffusion of these molecules through the sample volume. Therefore, heating rates and sample thickness are the external parameters controlling the nature of emusion spectra. In particular, at lower i>eating rates, more time is available for the evolution process leading to lowering in the peak position in the effusion rate versus temperature spectrum [S]. Different heating rates (/II’) of 2, 10, 20 and 50 C/min were employed for exploring the mechanism of hydrogen effusion from a-C:H samples. In addition, similar samples of different thicknesses were studied at fixed heating rates. The sample temperature and gas partial pressures were continuously recorded during heating. The partial pressures of effusing species at the pumping port represent a measure of the emusion rate (dN/dt) as the pumping

Materials 6 ( 1997) 1~3&1&?4

1831

speed of turbomolecular pump over a wide range of pressure is constant. The IR transmission was measured from 400 to 4000 cm-’ using a Nicolet FTIR spectrophotometer (Model 510 P). A Vickers microindentor was used to measure the film hardness. Special care is required to interpret the microhardness data on thin Gms. Composite Vickers hardnesses, as a function of applied load, were measured on - 1.5 pm thick samples deposited on crystalline silicon substrates. The substrate microhardness was independently measured. Film hardness was calculated using the method developed by Jonsson and Hogmark [ 111. Hardness numbers reported in this work are at a load of 10 gf. In the case of films having hardness lower than that of the silicon substrate ( - 1100 kgf/mm’), no reliable hardness number was obtained and we refer to them as soft films.

3. Results and discussion Fig. 1 shows the complete Fourier transform IR spectra in the range 400-4000 cm-’ on two representative samples A and Bl. Strong absorption is observed in sample A at - 3400 and - 1700 cm- I, corresponding to OH and C-O respectively. These bands are absent in the IR spectrum of sample BI. The absorption coefficient was calculated from transmission data using measured film thickness and standard IR data analyses procedures [ 121. The deconvoluted ai>sorption spectra in the stretching mode corresponding t&.Fig. 1 are shown in Fig. 2. The C-H bonding configurations of the films have been identified from the gaussian peaks fitted to the experimental spectrum after background correction. For the type B film. the stretching band is dominated by contributions from CH and Cl-l1 groups. i.e. absorption peaks at 2850. 2910 and 3000crn-‘. T~EX are assigned to sp” CH2/CH (symmetric), spJ Cl-l (asymmetric) and sp” CH (Olef) type of bonding respectively [ 13,141. Small contributions due to sp3 Cl-I3 (2960 cm-‘) and sp” CH2 (2936 cm ‘) are also seen in Fig. 2b. On the other hand, there is a large contribution from peaks due to sp3 CH3 and sp3 CH;! (2870, 2963 and 2927 cm - ‘), and sp’ CH? ( 3025 cm ’) type Of bonding in sar;lple A (Fig. 2a). The relative contributions of mono and polyhydride contributions are given

Table I Prq~,~l;~liull wndiliwls

and propertics of ;I-c’:H tilms of type\ A and H. Thcrc arc: IWO samples OC1~17~’ B diawd.

percentages of CH. CH, and CIi, Sample

_P Eg (eV 1

Preparation conditions

(type)

DLC53 (A)

namely BI ;md 61. Rcla~lw

are also shown

DLCSI(BI)

,T= 10.0 mbar, Ig= j;. C2HZ/(C2HI t Hz)= 1.0. T,=‘5() C /I-0.6mbar. b,= -3OV. C2H2i(C2H,+H2)=0.5. T,=?sfl

DLC40(BI)

p=O.6 mbar. Cg= vr. C2H2/(C2H,+H2)=0.5.

r,=z50

C

C

1.45 2.67 2.12

II

I.6 2.1 2.1

C&

Cl-i,

(‘H

(kgf,mm”)

CH (Ul.%)

(%)

(%)

(Q ) --

Soli 2420 2100

39.0 5.6 i 3.i)

71 13 2’)

‘4 II 3

Micro Hardness

5 76 46 -

n

/

__ I I

0,

I 30 0

s

L_

-

C-H

I1 3400

I

II 2800

.I

C-C

Stretching

I

I 2200

Wavenumbers

I

I

I 1600

Stretching

I

I

II 1000

t

41

(cm-‘)

Fig. I. Infrared spectra of typical (a) polymeric (p- 10.0 mbar; Tgu,,-250 ‘C; V,-710 V; pure C,H,); and (b) diamond-like carbon 1:s710 v: L’b5 - 250 v: (p-0.6mbar; T,,,,- 250 c; C2HL/C2H, + H, _ 0.5) films.

Wavenunber

(cm”)

Fig. 2. Deconvolution of the infrared strc:ching hand of a-C:H films from the spectra in Fig. I.

in Table 1. Furthermore, hydrogen concentration in the films was estimated using integrated intensity of absorption coefficient (Fig. 2) and a constant multiplier as reported in Ref. [IS]. The hydrogen contents of 39 and 5.6 at.% were obtained from samples A and B respectively. The IR analysis is found to give hydrogen content estimates in good agreement with elastic recoil detection analysis ( ERDA) [ lo]. In type B films, a higher percentage of C-C crosslinking is indicated by a microstructure having more of sp3 CH bonding as compared with that of sp3 CH2 and q3 CH,. In addition, enhanced sp2 CH bonding is also a chaiscteristic of hard a-C:H films [ 14,161. Moreover, high optica! bandgap along with high refractive index (Table 1) also suggest a diamond-like character of the films prepared under type B conditions. In contrast, there is a large proportion of hydrogen bonds in sp’ CH3, sp3 CH, and sp’ CH2 configurations in type A films. T>e CH, complex is normally a bond terminator and its abundance leads to the enhanced void structure in the films. The sp’ CH, bonding constitute the polymeric chain-like component. Higher hydrogen content primarily bonded in CH, and CH2 groups as well as C=C double bonds are a signature of polymeric nature of the sample A. The C=C bond absorbs water when exposed to the atmosphere; the polymeric sample therefore shows H,O absorption peaks at 3400 and 1700cm-’ [17]. L ower values of hardness (< 1100 kgf/mm2), along with a high optical bandgap but low refractive index (Table 1). are in accordance with the above porous and polymeric microstructure of this sample. Fig. 3 shows the thermal elusion spectra of the same two samples corresponding to Fig. I. The thermal eflusion spectra of soft a-C:Ii film (type A) in Fig. 3 show contributions from hydrocarbons CH3, &HZ, C2HJ, C3H3 and C4H3 along with H,. In contrast, effusion measurements yield peaks only due to hydrogen and CH, in the case of the hard DLC film (type n). It may be pointed out that the nature of thermal effusion spectra from CH,, CH,, and CH4 was found to be identical indicating the same origin for the methyl group radicals. We show only the CI-I, spectra for clarity. The maximum in evolution temperatures (T,,) of the CJ-I,, molecules ranges from -450 to 550 “C. Similar effusing species have been observed by other researchers from a-C:H films [5,7,9]. Fig. 4 summarizes the relative concentrations of evolved hydrocarbons and hydrogen molecules for these two typical satnples or” approximately same volumes. The hard film having lower total hydrogen content desorbs less amount of hydrc:arbons. This figure also shows that the estimates of total hydrogen fzontent using either hydrogen effusion alone 3r total pressures may be highly misleading. In order to understand the mechanism of hydrogen

-____.

--~_-

(a)

Diamondlike

:n : I !

50-

i -

40-

t-h+ . . . . . . . . . C&+ i i

i

30-

Polymeric

Ptiiymeric

!*

i ;

:*!.. j j: :

- .._..

C-H;

----

C3H3+

-.-.-

C2 H;

1

‘$‘,

_______ Qf_g -

-

=? Hz

I (b)

Diamond-like

CH3

C,%

Cz”,

C3”3

CL&

Fig. 4. Relative concentration ofdesorbed hydrogen and hydracarbons as calculated from the thermal effusion spectra in Fig. 2.

I? is the order of reaction; and R is the gas constant. Assuming a first-order desorption kinetics, the activation energy may be determined from T,,, by the relationship [ 191: \10exp(-E,/kT,,,)=E,/?/kTi,

(2)

where /I is the heating rate; \lOh 1013s- *; and k is Boltzmann’s constant. E, for CH3 was obtained as -2.3 eV for /I=O.33 K/s. The origin of this activation energy may be explained by simultaneous detrapping and H followed by the recombination lo form U-1, 171. The net energy required for the process is: E,,=E(C-H)+E(C-CN,)-E(ta-C

Temperature

(“C 1

Fig. 3. Thermal gas effusion spectra as a function ol‘heating tempcrHture for lilms corresponding to Fig. I.

and hydrocarbon release, we first focus our attention on the CHJ peak which is common to all effusion spectra. It is interesting to note that the position of CH3 peak CT,.,,)and its shape was found to be almost independent of the type and microstructure of a large number of DC glow discharge deposited a-C:H films in our study. 7’, was found to be 520 + I5 “C fcr samples with different microhardness. In addition, no shift in ?‘, wag found for identical samples of different thickness. This implies bond breaking or thermal desorption to be the rate limiting step in case of CH, effusion. The free energy of r;Qvation, Ea. is determined from surface desorption kinetics [ 181: dN/dt=A(l-N)“exp(-EJRT),

(1)

where N is the number of gas atoms evo’lved by overcoming the free energy barrier E,; A is the reaction constant;

(3)

Taking the values of E(C-44). E(C-CH3) and E( 1II--CH3) from the literature as 3.68, 3.0 and 4.5 eV, respectively 1201. E, is determined to be 2. I8 cV. This is in good agreement with the experimentally obtained value of 2.3 eV. Another possibility is the rupture of bonds from two bonded carbon two C-CH3 C& CH, atoms >c--C< thereby converting C-C to C-C. Energy is gained in this conversion, which is the difference of C-C binding energy (4.33 eV ) and that of C--C ( 2.55 eV ). Therefore. activation energy for desorption in this process [2Ea =2E(C-C 3)-1.78] is 2.11 elf. Concentrating on Hz evolution in Fig. 3. its threshold temperature for evolution is about 400 ‘C. Two major peaks are observed around 600 and -750 at. In additier?, there is a small peak appearing at about 425 “C. Thus peak disappears in the samples prepared at substrate temperatures higher than 400 “C. peaks may be labeled as low-temperature (LT) and high-temperature (MT) peaks. The LT peak occurs near hydrocarbon peak temperatures. It shifts to higher temperatures with the increase in hardness of the samples.

No variation in peak positions has been observed for samples of varying thickness. This implies thermal desorption as the ra tc-‘limiting step of LT hydrogen. For accurate estimation of desorption activation energies of hydrogen, thermal analysis method [1X] was applied. By differentiating the equation obtained from Eq. (1) for maximum evolution rate (dN/dT) at temperature T,, we get: d( In /?/Ti)/d(

l/T,)=

- EJR

(4)

regardless of the order of reaction. The thermal analysis of LT as well as HT peaks using Eq. (4) at a number of heating rates fi is used to determine the activation energy Ea. A plot of In /3/T: versus l/T, is shown in Fig. 5 for LT and HT peaks obtained using different heating rates for polymeric and DLC films. Activation energies of 2.4 and 2.7 eV are obtained from the slope of the LT curve for a soft and a hard film respectively. The observed activation energy for our samples can be understood if the desorption process is assumed to involve the simultaneous rupture of two nearby C-H bonds followed by the recombination of two H atoms to form Hz as: E, = 2E(C-I-I) - E(H-H).

(5)

The energy for rupture of neighboring C-H bonds is lowered by an H-H binding energy of 4.5 cV giving an activation energy E,,, of 2.86 eV. This is in close agreement with cxperimcntal value of 2.7 eV. LT hydrogc~~ may also be coming from wcakencd double

bonds Pc’--“c<) [6]. If hydrogen leaves such sites, the C=C bond will be restored, which is indeed observed at 1600 cm-’ in the samples heated to 600 “C. It may be pointed out that our IR absorption measurements after step heating a typical polymeric sample (Fig. la) to 600 “C show that about 25% of the bonded hydrogen .till ren:ains in the sample. Furthermole, ERDA shows this to be uniformly distributed across the thickness of the sample (not shown). In higher hardness DLC film (Fig. lb), most of the hydrogen is evolved at higher temperatures. Interestingly, p dependence of T, can still be fitted using Eq. (4) as seen in Fig. 5, indicating desorption to be the rate-limiting step. Activation energies obtained are higher for the NT peak and these are found to be 2.9 and 3.2 eV for soft and hard films, respectively. It is difficult to ascribe a definite mechanism to these energies as structural changes (bond reconstruction to form C=C bonds) are also known to take place at these temperatures. However, it is likely that HT hydrogen is released from uniformly distributed hydrogen in C-H configurations in the solid network. Additional energy and/or time lag may be required for a detrapped H atom to hop to the nearest available H site to form a mobile Hz molecule. Wild and Koidl carried out the thermal effusion experiments on bilaycrs of a-C:H and a-C:D and arrived at the conclusion that the diffusion of larger hydrocarbon molecules requires a porous microstructure of the samples [8]. They also showed that hydrogen molecults arc formed inside the material after getting detrapped from its hydrocarbon configuration. Results shown in Fig. 4, ;~long with the IR absorption data ( Fig. 2 ) and lower hardness ( -c I 100 kgfjmm”) of sample A arc consistent and show a porous microstructure of this sample. In contrasi, high microhardness ( -- 2400 kgf/mm% low hydrogen content and thermal elyusion profile (Fig. 3) of sample B show its compact microstructure. These features can be correlated with the growth parameter of DC glow discharge plasma [21].

4. Conclusions

-14L 0.6 Fig. 5. Dctcrmimllion polymtic

0.8 1.0 1.2 (I/Tm 1x '10e3

1.4

of actiwtion cmxgies of hydrogen evolution for

(microhatxhcss -CIf00 kgf;mml) and diamond-like carbon

(microhardncss - 2100 kgf,mm’) films of thicknesscs I .6 and rcspcctivcly.

I .1 pm.

Amorphous DLC films have been produced by DC glow discharge of acctylenc. By comparing the thermal gas effusion profiles with the IR absorption spectm, the bonding configurations responsible for different microstructures of diamond-like to polymeric a-C:H films have been identified. Two major peaks (LT and HT) for hydrogen evolution from the samples are obtained. The desorption kinetics of hydrogen effusion from a DLC film gives activation energies of 2.7 and 3.2 eV for LT and HT peaks, respectively. The LT peak could be due to weakly bonded hydrogen or hydrogen bonded to

the internal surfaces of microvoids in a c’~stered form. In higher-hardness DLC films, most of the hydrogen is evolved at higher temperatures. This is likely to be released from the uniformly distributed hydrogen in the solid network.

Acknowledgement We thank Dr B.D. Malhotra of National Physical Laboratory, New Delhi for FTIR measurements. We are grateful to Dr V.N. Kulkarni and T. Som for elastic recoil detection analyses. The experimental help of A. Joshi is duly acknowledged.

References [I] P. Koidl. Ch. Wild. B. Dischler. J. Wagner, M. Ramsteiner. Mater. Sci. Forum 5253 i 1989) 4’ [2] D. Boutard, B.M.U. Scherzer. ‘..‘. Moller, J. Appl. Phys. 65 ( 1989) 3833. [3] W. Beyer, Physica B 170 ( 1991) I05 and references therein [4] W.B. Jackson, NH. Nickel. N.M. Johnson, F. Pardo. P.V. Santos. Mater. Res. Sot. Proc. 336 (1994) 31 I. (51 I. Watanabe, T. Okurnura. Jap. J. Appl. Phys. 25 (1986) 1851.

[6] AL Kosarev, M.Sh Abdulvagabov. Ju.M. Baikov, N.S. Zhdanovich, V.F. Cvetkov, Surf. Coat. Technol. 50 ( 1992) 209. [7] X. Jiang, W. Beyer, K. Reichelt, J. Appl. Phys. 68 (1990) 1378. [S] Ch. Wild. P. Koidl, Appl. Phys. Lett. 51 ( 1987) 1506. [9] R. Stief, J. Schafer, J. Ristein. L. Key, W. Beyer, Proceedings of the 16th international Conference on Amorphous Semiconductors, rapan, 1995. J. Non. Cryst. Solids 198-200 (1996) 636. [IO] M. Malhotra. T. Som, V.N. Kulkami. S. Kuraar, Vacuum 47 (1996) 1265. [I I ] B. Jonsson, S. Hogmark, Thin Solid Films 1I4 ( 1984) 257. [I21 M.H. Brodsky, M. Cardona. J.J. Cuomo. Phys. Rev B !6 (1977) 3556. [ 131 B. Dischler, A. Bubenzer, P Koidl. Solid State Commun. 48 (1983) 105. [ 141 B. Dishler, in: P. Koidl and P. Oelhafen (Eds.). E-MRS Symposium Proceedings, vol 17, Les Editions de Physique, Paris, 1987. p. 189. [ 151 F. Fujimcto, A. Ootuka. K. Komaki. Y, Iwata. I. Yamane, H. Yamashita, Y. Hashimoto. Y. Tawada. K. Nishimura. H. Okamoto, Y. Hamakawa. Jap. J. Appl. Phys. 23 (1984) g10. [16] E.H. Dekempcneer. R. Jacob, J. Smeets. J. Meneve. L. Eersels. B. Blanpain, J. Roos, D.J. Oostra. Thin Solid Films217 ( 1992) 56. [ 171 W. Dworschak, R. Kleber, A. l‘uchs. B. Scheppat. G. Keller. K. Jung. H. Ehrhardt, Thin Solid Films IS97 ( 1990) 257. [ix] H.E Kissinger. Analyt. Chrm. 29 (1957) 1702. [ 191 P Koidl. Ch. Wild, B. Dischler, J. Wagner, M. Ramsteiner. Miter. Sci. Forum 52 (1989) 41. [20] K.1’. Hubet. in: D.E. Gray (Ed.). AIP Handbook of Physics. McGraw Hill, New York, 1972, pp. 7-168. [2l] $I. Malhoir,t. Ph D. thesis, Indian Institute of Technology. Konpur. India 1996.