J. MATER. CHEM., 1994, 4( 5), 695-701 Co-pyrolysis of Hydrocarbons and SiEt, for the Synthesis of Graduated Si,C, -x Ceramic Thin Films by Chemical Vapour Deposition Jean M. Agullo, Florence Fau-Canillac and Francis Maury* Cristallochimie, Reactivite et Protection des Materiaux, CNRS-URA 445, Ecole Nationale Superieure de Chimie, 118 route de Narbonne, 31077 Toulouse cedex, France The thermal decomposition of hydrocarbons has been investigated under particular conditions in a low-pressure chemical vapour deposition reactor in order to select suitable carbon sources for the preparation of carbon-rich or graduated Si,C, --x layers. Growth of pyrolitic carbon thin films (pyro-C) starts at ca. 1050 K and only above 1273 K using C,H,Pr' and CH,, respectively. The microstructure of the pyro-C layers is more dependent on the deposition temperature than on the nature of hydrocarbons.Their co-pyrolysis with SiEt, used as Sic precursor has been achieved in the temperature range 1050-1250 K. As expected, the film composition does not change significantly at 1173 K using CH, as an additional C source. By contrast, the C content of the films deposited by co-pyrolysis of SiEt4 and C6H5Pri increases continuously from 0.48 to 1 by increasing the mole fraction ratio x(C6H5Pri)/[x(C,H,Pri) +x(SiEt,)] from 0 to 1. Multilayers and compositional gradient layers can be prepared by discrete or continuous changes of the gas- phase composition, respectively. These films were successfully used as interphase in ceramic-ceramic composite materials to weaken the fibre/matrix bond and to improve their ductility. Ceramic-ceramic composite materials such as Sic-Sic are highly resistant to oxidation at high temperature but their brittleness is a limiting feature for many applications.The deposition of a softer material on the Sic fibres, like pyrolitic- carbon thin films (pyro-C), before matrix formation improves significantly their fracture toughness.' This interphase weak- ens the fibre/matrix bond and protects the fibres against microcracks and chemical However, because pyro-C films have a low oxidation resist- ance, alternative solutions are required. From this point of view, since C and Si interphases improve the du~tility~,~ and the oxidation resistance7-' of ceramic composite materials, respectively, multilayers or graduated Si,C1 -x layers, in which the composition changes continuously from C (fibre interface) to Sic (matrix interface), would be suitable interphases.For this purpose, we undertook an investigation of the chemical vapour deposition (CVD) of non-stoichiometric Si,C1 -x films on flat substrates before extending the application of this process to the fibre treatment. Many silicon compounds have been used as molecular precursors for the growth of Sic thin films.g However, prefer- ring a process without chlorine because of the sensitivity of Sic Nicalon fibres to chlorine atmosphere," we have shown previously that SiEt, (Et =C2H5)was a suitable organometal- lic precursor for the deposition of SixC1-, at moderate temperature (700-900 OC)." Furthermore, Si enrichment of these films was achieved by SiH, addition in the gas phase." In the next step of this programme, the preparation of graduated SixC1 -layers requires an increase of the C content from ca.0.5 (Sic stoichiometry) to 1 (pyro-C). The first requirement for the selected hydrocarbons is to be readily decomposed in the conditions used for the pyrolysis of SiEt,, in order to control easily the carbon incorporation in the film by monitoring the gas-phase composition. Secondly, hydrocarbons have to be used also for the deposition of pyro-C films with a highly graphitic structure in order to facilitate the fibre slip in the ceramic matrix. It is assumed, for this objective, that the existence of aromatic rings in the hydrocarbons (sp2 C) will improve the degree of graphitization of the layer.This paper deals successively with the thermal decompo- sition of some hydrocarbons in a low-pressure CVD reactor and their co-pyrolysis with SiEt, in order to select suitable carbon sources for the preparation of either C-rich Si,C1 -x films or compositionally modulated SixC1 -x layers. Experimental Apparatus and Deposition procedure The thin films were deposited using a horizontal hot-wall low-pressure CVD apparatus. This set-up is used for the chemical vapour infiltration of ceramic films for the prep- aration of composite materials and details were rcported previ0us1y.l~ In summary, an electrical furnace with three independently regulated heating zones provides an isothermal length of ca.40 cm. Quartz reactors 90 cm long and 2.5 cm in diameter were used. The total pressure is measured indepen- dently of the gas composition and automatically monitored using a capacitance manometer and a throttle valve control system. The flow rate of gaseous precursors was directly monitored using mass Aowmeters and the partial pressure of the liquid precursors was adjusted both from the flow rate of the carrier gas (He) through their bubblers and their vapour pressure given by the Clausius-Clapeyron equations." Polycrystalline alumina plates and silicon wafers passivated by a thin film of Si3N, were used as substrates. Before being introduced in the reactor, they were degreased for 5 min in hot trichloroethylene and then for 5 min in hot acetone, and were dried under a nitrogen stream.In all experiments, the total pressure and the total flow rate were adjusted in order to keep constant the residence time of the species in the heating zone of the reactor. In co-pyrolysis experiments, the gas-phase composition is defined by the ratio of the mole fractions x(hc)/[x( hc) +x(SiEt,)], where x(hc) means the mole fraction of the hydrocarbon and x(SiEt,) that of the silicon precursor. The typical MOCVD conditions are reported in Table 1. General Instrumentation The degree of crystallinity of the pyro-C and C-rich Si,CI-, films was analysed by transmission electron microscopy (TEM) and selected area electron diffraction (SAED) using a Philips 100 kV microscope.For the TEM analyses, the tilt Table 1 Typical experimental MOCVD conditions used for the depos- ition of C-rich Si,C1-, thin films by co-pyrolysis of SiEt, and hydrocarbons (hc) deposition temperature/K 1050-1250 reactor pressure/Pa 400 total flow rate/sccm 175 carrier gas He reactor diameter/mm 25 residence time of species/s 0.2 mole fraction ratio: x(hc)/[x(hc)+x(SiEt,)] 0-1 angle between the primary electron beam and the surface of the film was 90" unless otherwise specified. The Si content of the films was determined by electron- probe microanalysis (EPMA) using a CAMECA microprobe (accelerating voltage: 5 kV) and an Si wafer as standard.The C content can be satisfactorily deduced by difference. The film composition of most samples was also determined by X-ray photoelectron spectroscopy (XPS) using a spectro-photometer (VG Escalab MK 11) operating with a non-monochromatized A1-Ka,,, X-ray source. After Ar + sputtering (3 kV, 150 PA, 5 min) to remove the surface contamination, the oxygen content was found to be only ca. 5atom% and the Si content was found to be in satisfactory agreement with the EPMA data. The layer thicknesses were measured on fractured cross- sections using a JEOL JSM-25 scanning electron microscope (SEM). Mass spectra of few precursors were recorded using a time-of-flight mass spectrometer (CVC Bendix) at a ioniz- ation potential of 70eV and using direct insertion into the ion source.Si,C, -Deposition from SiEt, Silicon carbide films were deposited by thermal decomposition of SiEt, under reduced pressure in the temperature range 1050-1250 K. The experimental conditions (Table 1) were optimized for the chemical vapour infiltration of Si,C1 -,of porous preform^.'^ In order to prepare C-rich Si,C1 -,films and according to thermodynamic calculations (vide infra), helium was preferred to hydrogen as carrier gas because the C incorporation in the film is facilitated to the detriment of CH, formation. Fig. 1 shows the variation of the growth rate uersus the deposition temperature. As usually observed in CVD pro- TI°C 950 900 850 800 0 0 lo-' 8.0 8.5 9.0 9.5 1O~WT Fig.1 Variation of the growth rate of Si,Cl-, as a function of the reciprocal temperature using SiEt, as a single source. The linear dependence at low temperature reveals an apparent activation energy of 192 kJ mol-' J. MATER. CHEM., 1994, VOL. 4 cesses, the linear dependence at low temperature (T<1123 K) indicates a thermally activated process with an apparent activation energy of 192 kJ mol-'. This is in satisfactory agreement with the value reported for the pyrolysis of this precursor in a cold wall CVD reactor under H, (197 kJ mol-') and He (155 kJ mol-l) atmosphere." This is consistent with a deposition process kinetically limited by the heterogeneous chemical reaction. The stabilization of the growth rate above 1173 K results mainly from the depletion of the reactives along the hot wall reactor but probably the diffusion of the species in the gas phase has a limiting contribution. The silicon content of the Si,C,-, films deposited between 1123 and 1233 K decreases slightly from x=0.5 to 0.4 (Fig.2). This result is different from that obtained for the pyrolysis of SiEt, in a cold-wall CVD reactor since, in the same tempera- ture range, the Si content was found to be 0.6-0.7." In the present process, the efficiency of the C enrichment originates probably from a more complete decomposition of the ethyl ligands. Although the atomic ratio is C :Si =8 : 1 in the starting molecule, the C incorporation in the film is not very high and it is not possible to increase drastically the C content of these Si,C1 -,films only by increasing the pyrolysis temperature.Selection of Hydrocarbons Thermodynamic Calculations in the Si, C, H system The thermodynamic approach frequently provides useful informations for CVD processes.15 Although calculations have been reported previously for the chemical system Si, C, H,16 we have made preliminary calculations using our particular experimental CVD conditions to obtain some clues about the selection of the precursors and the conditions leading to C enrichment of the films. The complex equilibrium calculations were performed using a version of the classical Solgasmix computer program based on the minimization of the Gibbs free energy of the total chemical system.17 The thermodynamic data of both the gaseous species (20 molecules and radicals) and solid phases were found in the usual tables and a data bank." SiEt, was assumed to be fully decomposed and, according to literature data, the C-rich Si,C1-, films were considered to be a mixture of P-Sic and graphitic C.16 Calculations for the system SiEt,-H, show that stoichio- metric P-Sic is obtained under atmospheric pressure at low temperature and high partial pressure of hydrogen. Increasing the pyrolysis temperature and/or decreasing the hydrogen partial pressure leads to C-rich films.This tendency is clearly enhanced under reduced pressure [Fig. 3(a)]. For example, under 0.1 kPa, the C content of the films is theoretically 0.6I I I t E 0.3 8 0.2 i7j 0.1 1 1050 1150 1250 TIK Fig.2 Influence of the substrate temperature on the silicon content x of Si,C1-, films deposited using SiEt, as a single source (0,EPMA data) and a mixture of SiEt, and C,H,Pr' with a mole fraction ratio X(C,H,Pr')/[x(C,H,Pr)+X(SiEt,)] =0.75 (A,EPMA data; .,XPS data) J. MATER. CHEM., 1994, VOL. 4 200ie He H2P gas composition gas composition v CI 1g) (c) 1 c. -500 ie CH4 gas composition gas composition Fig. 3 Thermodynamic calculations of the film composition in the systems SiEt,-H, (a), (b),SiEt4-CH,-H, (c) and SiEt,-CH,-He (d) versus the deposition temperature and the gas-phase composition. The mole fraction of SiEt, was fixed at 5x lop2 and the total pressure was (a) 10, (b)0.1 and (c), (d) 1 kPa higher than 80% in the temperature range 600-1200 K whatever be the hydrogen partial pressure [Fig.3(b)].The equilibrium C(S)+~H~*CH~ (1) where the solid C competes with the C in the gas phase in the form of CH, accounts for these results. The influence of an additional source of carbon has been investigated with the system SiEt,-CH4-H,. The total press- ure was fixed at 1kPa according to the experimental value of the deposition process using SiEt,. The typical results are shown by the Fig. 3(c) and 3(d). As expected, the C incorpor- ation in the films increases by increasing either the partial pressure of methane or the deposition temperature. This tendency is enhanced under an inert atmosphere since H2 facilitates the formation of CH, and is, of course, easily explained by eqn.(1). A carbon content as high as 95% can be expected at about 1273 K using the chemical system SiEt,-CH,-He. According to this result, it was not necessary to perform calculations using more complex hydrocarbons including heavier alkanes and aromatic compounds. Pyrolysis of CH4 Methane is the simplest hydrocarbon and it has frequently been used successfully in CVD processes as the C source; that is supported by the above thermodynamic analysis. Furthermore, its pyrolysis in a LPCVD reactor was reported to start at 973 K.19*20 However, no significant thermal decomposition of CH, was observed in our apparatus using the experimental conditions of Table 1. Even for temperatures higher than 1273 K, the residence time of the species in the reactor had to be increased from cu.0.2 to 200s in order to increase the yield of decomposition and to get acceptable growth rates of the pyro-C layers. Films deposited at 1323 K have a high degree of graphitization. TEM analvsis shows large hexagonal date-shaped crystal- -1 lites which have grown parallel to the surface of the substrate [Fig. 4(u)]. The corresponding SAED pattern contirms the graphitic structure [Fig. 4(b)].According to the literature,21’22 they exhibit two series of $iffraction spots with interplanar spacings of 2.1 1 and 1.21 A assigned to the (1Oz) and (llz) planes of a perfect crystal of graphite, respectively. The radial deformation of some diffraction spots reveals a trandational disorder of the graphitic planes.Furthermore, the multiplicity of the (lOz) spots originates probably from a rotational disorder of the basal planes along the stacking direction. The observation of Moire fringes, both on the bright-field [Fig. 4(u)] and on the dark-field images [Fig. 4(c)], is due to interferences between at least two diffracted beams and con- firms this structural analysis. Pyrolysis of C,H,Pr‘ Since the pyrolysis of CH4 occurs at temperatures higher than the temperature range of our process, less stable hydrocarbons should be found. From this point of view, when the pyrolysis process is initiated by homolytic bond breaking, which is frequently the case with this kind of compo~nd,~.’,~~two precursors are expected to be decomposed in a similar tem- Fig.4 TEM analysis of a pyro-C film obtained by pyrolysis of CH4 at 1323 K under a reduced pressure of 13kPa. Both the bright field micrograph, (a), and the corresponding SAED patterns, (1 I), were obtained with the surface of the sample perpendicular to the primary electron beam. The dark-field image of the same area (c), obtained using a ( 1lz) spot, exhibits the hexagonal plate responsible for the diffraction pattern (b).The arrows show Moire fringes J. MATER. CHEM., 1994, VOL. 4 perature range if they have comparable bond strengths. Table2 reports a comparison between the bond strength of SiEt, and a few hydrocarbons. Isopropylbenzene, C6H5Pr' (or cumene), seems to be a suitable candidate because the lability of the CH, groups is quite similar to that of Et groups of SiEt,.This is confirmed by the fragmentation of these mol- ecules under electronic impact in a mass spectrometer (Table 3). The fragmentation of the cumene is clearly initiated by the loss of a Me group generating the highest peak of the spectrum. The weakening of the C6H5cH(cH3)- (CH,) bond (275.9 kJ mol) results from the inductive effect of the aromatic cycle. Moreover, it was expected that the aromatic ring of this hydrocarbon could facilitate the graphitization of the pyro-C films. The mass spectrum of SiEt, recorded in the same conditions is typical of SiR, molecules;28 the loss of one or two Et groups occurs with a high probability and initiates the main decomposition pathway. Carbon layers were grown from C,H,Pr' between 1050-1273 K using experimental conditions close to those reported in Table 1 (residence time=0.9 s instead of 0.2 s).TEM analyses reveal that the films have a poor crystallinity with a mean size of microcrystallites of only a few nm [Fig. 5(a), 5(d)]. At 1173 K, the SAED pattern of a sample with the surface perpendicular to the electron beam exhibits two diffuse rings assigned to the (102) and (1lz) planes of a graphitic structure [Fig. 5(b)].After tilting the sample by 40" with respect to the electron beam, the SAED pattern does not change significantly indicating a very high disorder of the graphitic structure [Fig. 5(c)].This type of free C can be called amorphous because the basal planes are randomly rotated and they are not stacked in a single direction, for instance parallel to the surface of the film. The microstructure of the pyro-C films at 1273 K is different. SAED patterns give evidence for a turbostratic structure, i.e. the graphitic planes exhibit a rotational disorder but they have a preferential oriention parallel to the surface of the film. When the sample is perpendicular to the electron beam, the SAED pattern shows the two diffraction rings of the (102) and (1lz) planes (rotational disorder) and when it is tilted by 40" degrees diffuse diffraction spots appear with an interplanar spacing of Table 2 Principal bond dissociation energies (EBD) of molecular pre- cursors used in this work precursor bond EBD/kJmol-' ref.SiEt, Et,Si-Et 279.2 25 Et,SiCH, -CH3 372.0 26 CH4 H3C-H 439.7 27 C6H5Me C,jHsCH,-H 355.3 27 CsHs-MMe 418.0 27 C,H5Pr' C6H5CH(CH3)- CH3 275.9 27 C~HS-P? 405.5 26 H-c6H,Pr' 459.8 27 3.37 and 1.68 A corresponding to the (002) and (004) planes, respectively [Fig. 5(e), 501. In order to investigate the influence of the aromatic rings on the microstructure of the pyro-C films, CH,T and C,H,I were tried as C source in the same pyrolysis conditions as for the cumene. The low values of the bond strength between the organic group and the halogen indicates that they should be readily decomposed in these conditions (234 and 267 kJ mol-', respectively).At 1173 K, the films deposited using CH,I are amorphous whereas those grown from C6H,I are highly turbostratic (Table 4). This result argues for a better graphitization of the C layers when aromatic cycles exist in the precursor. Now, from a mechanistic point of view, the question is to know what is the origin of the carbon deposited using cumene? Analyses (gas chromatography, NMR) of the by-products trapped at the liquid-nitrogen temperature at the outlet of the reactor reveal that styrene is the principal product (including traces of C6H6, C6H5Me, C,H,Et). This confirms a preferential breaking of the bond C6H5CH(CH3)-cH3. The radical C6H,CH(CH,)' loses probably a hydrogen atom to form C,H,CH=CH,. The methyl radicals are more reactive and then could lead readily to C deposition and release of H,.This is supported by the fact that, under the same experimental conditions, the growth rate of a pyro-C film deposited using C6H4( Pri)2 is twice that using C,H,Pri indicating that the growth rate is strongly correlated with the number of 'CH, groups (Table 4). In summary, the degree of graphitization increases with the pyrolysis temperature of hydrocarbons (Table 4). For a fixed deposition temperature, aromatic cycles in the precursor induce a higher graphitization of the layers. Methane is probably too stable a hydrocarbon to be used as the carbon source in our process but cumene is an attractive candidate. Although the pyro-C deposited using C6H,Pr' probably orig- inates mainly from the methyl groups, it is not out of the question that the aromatic rings enhance the graphitization of the films.It is difficult to deposit graphite films below 1273 K even with aromatic corn pound^.^^^^^ This was success- fully realized from aromatic precursors by dehydrogenation in a halogen atmosphere31 but for our application, a halogen- free process is required to avoid corrosion of the ceramic fibres. Co-pyrolysis Results and Discussion Co-pyrolysis of SiEt4and CH, Although CH, does not look very promising as an additional carbon source in our Si,Cl-, deposition process since it is quite undecomposed, co-pyrolysis experiments with SiEt, were undertaken because chemical reactions between both precur- sors are possible at these temperatures.Fig. 6 shows that, at 1173K and using the CVD conditions reported in Table 1, the composition of the films does not change significantly when the mole fraction ratio x(CH,)/[x(CH,) +x( SiEt,)] Table3 Principal ions obtained by fragmentation of SiEt, and C6H5Pri under electronic impact in a mass spectrometer using an ionizing potential of 70 eV SiEt, C&,Pr' assignment mlz rel. intensity assignment mlz rel. intensity SiEt,' SiEt,+ HSiEt, + H,SiEt + SiEt+ SiH3+ 144 115 87 59 57 31 25 98 100 43 13 10 25 100 6 12 12 6 J. MATER. CHEM., 1994, VOL. 4 Fig. 5 TEM micrographs and corresponding SAED patterns of pyro-C films obtained by pyrolysis of C6H5Pr at 1173 K, (u)-(c), and 1273 K, (d)-cf), under a reduced pressure of 200Pa.The tilt angle between the electron beam and the surface of the sample was 0" for the SAED patterns (b)and (e) and 40" for (c) and (f). The magnification is the same than that of Fig. 4 Table 4 Some features of pyro-C thin films deposited by thermal decomposition of various hydrocarbons in the same MOCVD reactor using the experimental conditions: total pressure =200 Pa, mol fraction =0.1 1 and He as carrier gas deposition growth rate/ precursor temperature/K nm h-l structure' -CH4 1273 no deposition 1323 - graphitic C,H.+k! 1273 42 amorphous 1343 1866 turbostratic C6H5Pri 1173 132 amorphous 1223 162 turbostratic 1273 222 turbostratic C,H,( Pr' h 1223 312 - CH,I 1223 - amorphous C6H5T 1223 turbostratic 'Determined by TEM analyses.increases from 0 to 0.93. This is of course in agreement with the results on the pyrolysis of CH, and it can be considered that this hydrocarbon is non-decomposed in such pyrolysis conditions. Moreover, addition of methane in the gas phase has a Fig. 6 Variation of the silicon content x of the Si,C,-, films deposited by co-pyrolysis of SiEt, and hydrocarbons as a function of the gas- phase composition: SiEt,-CH, (0,EPMA data), SiEt4-C6H5Pr' (H, EPMA data; A,XPS data). The deposition temperature was 1173 and 11 18 K using CH4 and C,H,Pr', respectively negative effect on the growth rate of the Si,C, -,films. [ndeed, the deposition rate decreases continuousIy when the mole fraction of CH, increases (Fig. 7).This can be partly explained by a decrease of the diffusion coefficient of SiEt, in a He-CH, mixture when the mole fraction of CH, increases. The variation of the diffusion coefficient of SiEt,, calculated using the formula of Fuller et a1.,32as a function of the gas-phase J. MATER. CHEM., 1994, VOL. 4 I I 1 I I1 T -500 I E -400 $ 0).--300 8 -200 .-g u) -100 5 Fig. 7 Variation of the growth rate of SiXC1-, thin films (a)deposited using a mixture SiEt,-CH, as a function of the gas-phase composition. The binary diffusion coefficient of SiEt, is also reported (dotted curve). The mole fraction of SiEt, was fixed at 7 x and the deposition temperature was 1173 K composition supports this assumption (Fig.7). This indicates that the decomposition process of SiEt, is not purely kin- etically controlled above 1173 K but that diffusion phenomena in the gas phase also play a role. However, the variation of the two curves of Fig. 7 is not similar indicating more complex phenomena such as inhibition effects due to radical trapping or other mechanisms. Co-pyrolysisof SiEt, and C,H',Pr Co-pyrolysis experiments between SiEt, and C6H,Pr' were carried out at 1118 K. The gas-phase composition was changed by increasing the mole fraction of cumene and keeping constant that of SiEt,. The Si content of the films decreases continuously from 0.5 to 0, this means the carbon content increases from the Sic stoichiometry to pyro-C by increasing the ratio of the mole fractions x(C,H,Pr')/[x( C,&Pr')+X(SiEt,)] from 0 to 1 (Fig.6). The films have a uniform morphology and thickness. Furthermore, a satisfac- tory agreement is found between XPS (after Ar' sputtering to clean the surface) and EPMA analyses, revealing a constant distribution of the elements through the thickness of the films. Keeping constant the gas-phase composition [x(c,H,Pr')/[x(C,H,Pr') +x(SiEt,)] =0.75, the deposition temperature was increased from 1050 to 1200 K. Fig. 2 shows that the carbon content of the films is higher than in those deposited using SiEt, as a single source whatever the tempera- ture. Furthermore, the Si content decreases by increasing the pyrolysis temperature, that is probably the result of a more efficient decomposition of cumene at high temperatures.By contrast with the chemical system SiEt,-CH,, the growth rate is thermally activated between 1050 and 1200K using the mixture SiEt4-C&Pr' with an apparent activation energy of ca. 109 kJ mol-'. Analysis of the microstructure of these carbon-rich Si,C1-, layers has been recently reported in an abbreviated form.33 Concluding Remarks Cumene is a suitable additional carbon source for the prep- aration of carbon-rich SixC1 --x films by thermal decomposition of SiEt, under reduced pressure. The graphitization of pyro-C layers deposited using various hydrocarbons increases with the temperature but they are generally quite amorphous below ca.1273 K if the decomposition occurs without a halogenated atmosphere, even if the precursor contains aromatic rings. rl600 800 sputtering time/s Fig. 8 Typical SIMS depth profiles of SixC1 -x multilayers deposited at 1118K and under a reduced pressure of 400 Pa using a +mixture SiEt,-C,H,Pr' (Ar sputtering). The C content of the four successive layers was adjusted from the gas-phase ratio x(C,H,Pf)/[x( C,H,Pf)+x(SiEt,)] fixed at (a) 1, (b) 0.85, (c) 0.5 and (d) 0. For clarity, each composition profile has a different intensity scale. H,C'; a, Si+; jl, Sil; +, O+ The pyro-C layers grown using C6H,pr' are effectively amorphous at 1173 K but start to be turbostratic at 1223 K. Using a mixture SiEt,/C&Pr', the carbon incorporation in the films can be increased either by increasing the deposition temperature or by increasing the mole fraction of cumene in the gas phase.This last procedure is more versatile for the CVD process because of the high inertia of the big furnaces used in industrial set-ups. Using computer-monitored mass flowmeters, multilayers and compositional gradient layers can be easily prepared by discrete or continuous changes of the gas-phase composition, respectively. Fig. 8 illustrates Si,C, -, multilayers (four layers) prepared by this process at 1118 K. A pyro-C film is grown first on a silicon substrate and the carbon content of the next layers is adjusted from the gas- phase composition. A slight oxygen contamination is observed in the Si-rich layers as this frequently occurs with this highly reactive element.Details both on the preparation of C-rich Si,C1-, films by this process and on their structural charac- terization is reported in a further paper.34 Furthermore, this deposition process was recently applied to prepare specific interphases in ceramic-ceramic composite materials. It was found that pyro-C and graduated SixC1 -,interphases, grown using cumene and SiEt4--C&Pr', respectively, weaken the fibre-matrix bond and improve the ductility of the cer-amic-ceramic composite materiaL6 The authors would like to thank Professor R. Morancho for the fruitful discussions during this work, Dr. F. Teyssandier for his help concerning the thermodynamic calculations and J.Poujardieu for TEM experiments. This work was partially supported by the SociCtk Europeenne de Propulsion, France, under contract No. 449334. References 1 J. Homeny, W. L. Vaughn and M. K. Ferber, J. Am. Ceram. SOC., 1990,73, 394. 2 T. M. Besmann, R. A. Lowden, D. P. Stinton and T. L. Starr, J. Phys. (Paris), Colloq. CS,1989,50,229. 3 R. Naslain, Proceedings of the 1st International Conference on Functionally Gradient Materials, ed. M. Yamanouchi, M. Koizumi, T. Hirai and M. Shiota, Society of Non-Traditional Technology (FMG Forum), Sendai, 1990, p. 71. 4 J. J. Brennan, Mater. Sci. Eng. A, 1990,126,203. 5 A. J. Caputo, D. P. Stinton, R. A. Lowden and T. M. Besmann, Am. Ceram. SOC.Bull., 1987,66,368. 6 J. M. Agullo, F. Maury and J.M. Jouin, J. Phys. IV (Paris), Colloq. C3, 1993,3, 549. J. MATER. CHEM., 1994, VOL. 4 701 7 R. A. Lowden and D. P. Stinton, Ceram. Eng. Sci. Proc., 1988, 9, 705. 22 A. Oberlin, J. Goma and J. N. Rouzaud, J. Chim. Phys., 1984, 81, 701. 8 9 10 11 12 13 14 15 16 17 18 19 20 21 K. L. Luthra, J. Am. Ceram. SOC., 1986,69, C-231. J. Schlichting, Powder Metal. Int., 1980,12, 141. D. P. Stinton, A. J. Caputo and R. A. Lowden, Am. Ceram. Soc. Bull., 1986,6, 347. F. Maury, A. Mestari and R. Morancho, Muter. Sci. Eng. A, 1989, 106,69. A. Mestari, F. Maury and R. Morancho, J. Phys. (Paris), Colloq. C5, 1989,50, 769. J. M. Agullo, F. Maury and R. Morancho, Thin Solid Films, 1992, 209, 52. Handbook of Chemistry and Physics, ed. R. C. Weast, C.R.C. Press, Boca Raton, FL, 66th edn.1985, D-196. C. Bernard, Y. Deniel, A. Jacquet, P. Vay and M. Ducarroir, J. Less-Common Met., 1975,40, 165. A. 1. Kingon, L. J. Lutz, P. Liaw and R. F. Davis, J. Am. Ceram. SOC.,1983,66, 558. G.Eriksson, Acta Chem. Scund., 1971,25,2651. (a) M. W. Chase Jr., C. A. Davies, J. R. Downey Jr., D. J. Frurip, R. A. McDonald and A. N. Syverud, JANAF Thermochemical Tables, 3rd. Edn., J. Phys. Chem. Ref Data, Suppl. 1, 14, 1985; (b) SGTE databank, Scientific Group Thermodata Europe, BP 66,38402 Saint Martin &Heres, France. P. A. Tesner and 0.G. Shein, Kinet. Katal., 1986,26,1259. E. V. Denisevic and P. A. Tesner, Neftekhimija, 1974,20, 390. G. Schiffmatcher,H. Dexpert and P. Caro, J. Microsc. Spectrosc. Electron., 1980, 5,729. 23 24 25 26 27 28 29 30 31 32 33 34 R. J. Bogaert, T. W. F. Russel, M. T. Klein, R. E. Rochelean and B. N. Baron, J. Electrochem. SOC.,1977,124,790. T. L. Chu, S. S. Chu, S. T. Ang, A. Duong, Y. X. Han and Y. H. Liu, J. Appl. Phys., 1986,60,4268. A. E. Pope and H. A. Skinner, Trans. Faraday SOC.,1964,60,1404. D. Griller, J. M. Kanabus-Kaminska and A. Maccoll, J. Mol. Struct., 1988, 163, 125. Handbook of Chemistry and Physics, ed. R. C.Weast, C.R.C. Press, Boca Raton, FL, 66th edn., 1985, F-185. M. R. Litzow and T. R. Spalding, in Mass Spectiometry of Inorganic and Organometallic Compounds, ed. M. F Lappert, Elsevier, Amsterdam, 1973, p. 233. R. J. Bard, H. R. Baxman, J. P. Bertino and J. A. O’Rourke, Carbon, 1968,6,603. M. L. Lieberman, Proceedings of the 3rd international Conference on Chemical Vapor Deposition, ed. F. A. Glaski, American Nuclear Society, 1972, p. 95. P. H. Chang and M. M. Labes, Chem. Muter., 1989,1,523. E. N. Fuller, K. Ensley and J. C. Giddings, J. Phys. Chem., 1969, 73, 3679. J. M. Agullo, F. Maury, R. Morancho and R. Carles, Muter. Lett., 1991, 11,257. F. Maury and J. M. Agullo, submitted for publication. Paper 3/06761K; Received 1lth Novemher, 1993