Plasma-enhanced chemical vapor deposition of SiO2 from a Si(CH3)3Cl precursor and mixtures Ar/O2 as plasma gas

Plasma-enhanced chemical vapor deposition of SiO2from aSi„CH3…3Cl precursor and mixtures ArÕO2as plasma gas A. Barranco Instituto de Ciencia de Materiales de Sevilla (CSIC-Universidad de Sevilla) and Dpto de Quı ´mica Inorga ´nica, Avda, Ame ´rico Vespucio s/n, 41092 Sevilla, Spain J. Cotrino Dpto de Fı ´sica Ato ´mica, Molecular y Nuclear, Facultad de Fı ´sica, Universidad de Sevilla, Avda. Reina Mercedes s/n, Sevilla, Spain F. Yubero Instituto de Ciencia de Materiales de Sevilla (CSIC-Universidad de Sevilla) and Dpto de Quı ´mica Inorga ´nica, Avda, Ame ´rico Vespucio s/n, 41092 Sevilla, Spain T. Girardeau and S. Camelio Laboratoire de Me ´tallurgie-Physique Ba ˆt SP2M, Bd. Pierre et Marie Curie, 86962 Futuroscope Chasseneuil Cedex, France C. Clerc Centre de Spectrome ´trie Nucle ´aire et de Spectrome ´trie de Masse, IN2P3-CNRS, Bat. 104-108, F-91405 Orsay, France A. R. Gonzalez-Elipea) Instituto de Ciencia de Materiales de Sevilla (CSIC-Universidad de Sevilla) and Dpto de Quı ´mica Inorga ´nica, Avda. Ame ´rico Vespucio s/n, 41092 Sevilla, Spain 共Received 3 January 2003; accepted 31 March 2003; published 16 May 2003兲 Silicon dioxide thin films have been prepared at room temperature by remote plasma-enhanced chemical vapor deposition in a downstream reactor by using Si(CH3)3Cl as a volatile precursor and a microwave electron cyclotron resonance external source. Experiments are done at constant pressure by changing the relative amount of Ar species Rin the plasma gas. The aim was to obtain thin films with low density and, therefore, low refractive index. Characterization of the species of the plasma is carried out by optical emission spectroscopy. The changes of the plasma conditions are correlated with the growing rate and microstructure of the films, the latter determined by atomic force microscopy and infrared spectroscopy. It is found that the growing rate of the films decreases and their roughness increases as Rincreases. The optical properties of SiO2thin films are analyzed by optical ellipsometry. A decrease in the refractive index is found for the films grown with high values of R. The possible routes for activation of the precursor and the formation of the SiO2thin films are discussed. © 2003 American Vacuum Society. 关DOI: 10.1116/1.1577134兴 I. INTRODUCTION Silicon dioxide thin films are prepared by a large variety of methods including plasma-enhanced chemical vapor deposition 共PECVD兲. This technique is widely used for this purpose because it permits a low deposition temperature, high deposition rate, etc.1Traditionally, SiO2thin films have been deposited from SiH4/O2or SiH4/N2O plasmas.1,2 Other silicon precursors used are SiCl4,1,3 tetraethoxysilane 关(TEOS) Si(O(C2H5)4)兴,4hexamethyldisilox- ane 关(HDMSO) O(Si(CH3)3)2兴,5and more recently, tetramethylsilane 关(TMS) Si(CH3)4兴6,7 and clorotrimethylsilane 关(TMSCl) Si(CH3)3Cl兴.8These precursors have been utilized in pure form or mixed with oxygen or N2O. There are some advantages in using organosilicon precursors, such as their safer handling, the fact that they yield a better step coverage, and the possibility to prepare inorganic SiO2as well as polymeric SiOxCyHzthin films, depending on the dilution degree of the precursor in the oxidizing gas.9,10 More recently, the use of TMS for preparing SiO2thin films by PECVD has been introduced because the films deposited with TMS⫹O2mixtures produce better water repellent layers than thin films deposited by using an ethoxide-like tetramethoxysilane⫹O2.11 Other authors have also shown that by using TMS⫹O2, it is possible to deposit hydrophobic SiOxthin films in a corona discharge plasma at atmospheric pressure.6In addition, since TMS and TMSCl do not contain oxygen, it is possible to deposit SiC and polymeric SiCxHythin films12,13 without adding to the plasma CH4or other hydrocarbons, as has to be done when using SiH4. However, in contrast with the ample literature existing about the use of SiH4, TEOS, or HDMSO, much less is known about the use of TMS or TMSCl. In previous works, we have employed the precursor TMSCl to obtain stoichiometric SiO2or SiOxCyHzthin films in a remote microwave electron cyclotron resonance 共ECR兲-PECVD reactor working in a downstream configuration.8,14 Moreover, we showed that under our experimental configuration, the existence of a SiuCl a兲Author to whom correspondence should be addressed; electronic mail: [email protected] 900 900J. Vac. Sci. Technol. A 21„4…, JulÕAug 2003 0734-2101Õ2003Õ21„4…Õ900Õ6Õ$19.00 ©2003 American Vacuum Society bond in TMSCl makes this precursor more susceptible than TMS to chemical attack by the excited species of the plasma.8 The use of SiO2thin films as an antireflective coating in, for example, photovoltaic cells,15 requires the fabrication of low refractive index porous coatings of this material. This can be done by increasing the total pressure of the plasma chamber during deposition. This parameter usually favors the condensation processes in the gas phase and, therefore, leads to an enhancement of porosity. However, as a drawback, this procedure usually has detrimental effects with respect to the mechanical stability of the layers because of the poor adhesion between the large aggregates formed in the gas phase. The present article reports the preparation at room temperature of SiO2thin films by using the TMSCl precursor and mixtures Ar/O2as plasma gas in a remotely coupled PECVD system. It has been found that, for a constant working pressure, the roughness, growing rate, and other microstructural characteristics of the films depend on the ratio between the plasma gases and precursor. From this study, some clues are deduced about the reaction mechanism of the TM- SCl precursor and the deposition processes leading to the formation of the SiO2thin film. The dependence of the optical properties of the films on their microstructural characteristics is also examined and discussed in relation with the parameters of the deposition process. II. EXPERIMENT Silicon dioxide thin films have been prepared at room temperature by a remote PECVD procedure. Si(CH3)3Cl has been used as volatile precursor of Si. The plasma reactor has a ‘‘downstream’’ configuration with an external microwave ECR source 共SLAN兲operated at 400 W. The source was separated from the reactor by a metallic grid preventing the microwave irradiation of the substrates. This grid allows easy control of the plasma ignition and the separation of the deposition area from the plasma region. Also, owing to the presence of the grid, it is expected that most activated species in the deposition zone are neutral, although it cannot be ruled out that some electrons and positively charged plasma species arrive at the sample position. The distance from the substrate and grid was 10 cm. Pieces of one side polished Si共100兲wafers of 3 cm2were placed in the reactor, while the precursor was dosed on top of their surface through a dispersal ring placed 6 cm from the substrate to ensure its homogeneous spatial distribution. The base pressure of the system is lower than 10⫺6Torr. A fully detailed description of the experimental setup has been presented elsewhere.8,16 Precursor and plasma gases were dosed by suitable mass flow controllers. Thus, fixed flow rates of 10 sccm for the oxygen and 5 sscm for the precursor were used, while the flow rate of Ar was varied in the range 0–5 sccm. Pure oxygen or mixtures Ar/O2were supplied to the source. A total pressure of ⬃2⫻10⫺2Torr was maintained constantly in the chamber during deposition by adjusting the pumping capacity of the system with a throttle valve. We have shown previously that this pressure corresponds to the maximum in the deposition rate when pure oxygen is used as plasma gas.8The main parameter followed in this work is the ratio Rbetween the Ar flow and the total flow of gases supplied to the reaction chamber 共i.e., Ris the mass flow of Ar divided by the addition of the mass flows of Ar, O2, and TMSCl兲. This relative estimation of the Ar added to the plasma gas provides a more appropriate description of the process than the mere Ar/O2ratio. Optical emission spectroscopy was carried out during thin film deposition. The light emitted by the plasma was collected by an optical fiber which was positioned 4 cm downstream from the source center through a hole in the SLAN air cooling ring. The fiber was joined to a monochromator Jobin–Yvon HR250 with a 1200 grooves/mm grating and equipped with a R928 Hamamatsu photomultiplier. The growth rate was measured with a quartz-crystal monitor placed beside the substrate. The typical thickness of the studied samples was about 500 nm. Atomic force microscopy 共AFM兲images were obtained in air by using a Topometrix Explorer microscope working in the noncontact mode. The cantilever 共supplied by Topometrix兲with a spring constant of 42.6 Nm⫺1was oscillated at 130 kHz. Amplitude reduction was monitored while the surface is scanned at a typical rate of one line per second. The scanner was calibrated in the X,Y, and Zdirections with gratings provided with the equipment. Both topographic 共constant force兲and direct signal 共constant height兲images were acquired. Fourier transform infrared 共FT-IR兲spectra were collected in transmission mode at normal geometry in a Nicolet 510 spectrometer for samples deposited on Si共100兲wafers. The thicknesses of the analyzed samples have been chosen to avoid the saturation of the infrared signal, thus enabling the comparison of the relative intensity of the bands. Rutherford backscattering spectroscopy 共RBS兲measurements of the films were performed in the Ion Accelerator ARAMIS 共Orsay, France兲using 1.5 MeV ␣ particles. The energy resolution of the detector used in the RBS measurements was 15 keV. The spectra were simulated using the RUMP code. The samples have been deposited on Si共100兲and channeling was used to improve the sensitivity of the technique for the SiO2thin films. Spectroscopic ellipsometry experiments were performed using a SOPRA commercially available system. The measurements of ellipsometric parameters were made within a wavelength range from 0.21 ␮ mto1.2 ␮ m at different angles of incidence 共65°, 70°, and 75°兲. These are the most sensitive angles, since they are close to the Brewster angle of a silicon substrate. For the studied films, the first step of the analysis procedure consists of a regression using a Cauchy law to simulate the refractive index n(␭), the extinction coefficient k(␭), and the thickness of the layer. In a second step, the layer thickness was fixed as determined herein and a point-by-point extraction of the complex refractive index was performed at each wavelength. 901 Barranco et al. : Plasma enhanced chemical vapor deposition of SiO2901 JVST A - Vacuum, Surfaces, and Films III. RESULTS AND DISCUSSION A. Plasma characterization and deposition process Optical emission spectra of the plasma were recorded for increasing values of R. The spectra of oxygen/argon plasmas 共without precursor兲are dominated by sharp lines due to emission from atomic oxygen or argon and by several molecular bands resulting from transitions between low vibrational levels of the O2 ⫹species. The addition of the precursor to the reactor increases the emission from the lines of atomic hydrogen in detriment to those of oxygen or argon. Thus, the optical emission spectra of the Ar/O2/TMSCl mixture, apart from the lines due to oxygen and argon, exhibit strong emission lines due to H*共lines of the Balmer series兲and other excited species such as CH*,CO *,CO ⫹,OH *, etc., the latter very likely coming from the fragmentation and oxidation of the precursor molecule. A systematic analysis of the variation in intensity of some selected lines can be used to understand some of the reaction pathways of species in the plasma. Figure 1 shows the variation of the intensity of emission lines/bands (Ar*at 750 nm, O*at 777 nm, H*at 486 nm, OH*at 306.4 nm, CH*at 431.4 nm, and CO*at 507.1 nm兲with R. The assignments of these signals have been done in agreement with the Refs. 17 and 18. The plot of Fig. 1 shows a net increase in the intensity of the Ar*line as R increases, while the O*emission increases only very slightly and the H*emission decreases. On the other hand, the emissions corresponding to CH*and OH*bands increase as R increases. The intensity of the CO*remains practically constant when Ar is added to the discharge. As a first approximation, the intensity of H*lines can be taken as a trace of the degree of fragmentation of the precursor molecule in the plasma 共note, however, that this is not rigorously true because the electron density and/or temperature may also contribute to changes in the intensity of the H* emission兲. Figure 1 clearly shows that the yield of this process decreases as the Ar content in the plasma increases. It is also important to mention that, for a given value of R, the intensity of all emission lines increases with the microwave power. The growth rate of the SiO2thin film was also determined as a function of R. Figure 2 shows that the deposition rate of the film diminishes almost linearly with R. We would also like to stress that no deposition was observed when only pure Ar was used as plasma gas or when TMS was used as a silicon precursor molecule instead of TMSCl. In the case of TMS, no deposition is observed either when O2or mixtures Ar/O2are used as plasma gas,8while for TMSCl, the growth rate increases for increasing precursor flow. B. Thin-film characterization The thin-film samples prepared in this work were analyzed by RBS to determine their composition. All of them had a silicon dioxide stoichiometry 共i.e., O/Si ratio equal to 2兲and no traces of Cl or C were detected. Therefore, the samples, at least up to the sensitivity limit of the RBS spectroscopy 共⭐5 at.% for C and ⭐1 at.% for Cl兲, can be considered as free from these elements present in the TMSCl precursor. The microstructure of the deposited SiO2thin films was investigated by AFM and FT-IR. Figure 3 shows as an example two AFM images of the surface of SiO2thin films corresponding to R⫽0关i.e., Fig. 3共a兲兴 and R⫽0.6 关i.e., Fig. 3共b兲兴. These images clearly show that the roughness of the films increases with the relative concentration of Ar. A more quantitative evaluation of the roughness of the films is provided by the root-mean-square 共rms兲surface roughness values. Figure 4 shows the variation of the rms surface roughness for SiO2thin films prepared with increasing relative FIG. 1. Intensity of the indicated emission lines excited in the plasma reactor as a function of R. FIG. 2. Deposition rate of SiO2thin films as a function of R. The line is plotted to guide the eyes. 902 Barranco et al. : Plasma enhanced chemical vapor deposition of SiO2902 J. Vac. Sci. Technol. A, Vol. 21, No. 4, JulÕAug 2003 amounts of Ar in the plasma gas. It is found that the rms surface roughness increases as the Ar concentration in the plasma increases. Figure 5 shows FT-IR spectra of SiO2samples prepared with increasing concentrations of Ar. It is important to note that no signals from organic groups8are detected in the FT-IR spectra for any of the deposited thin films, indicating that even if the investigated process is carried out at room temperature, the complete removal of organic groups takes place. The different peaks and shoulders in the 400–1500 cm⫺1region have been attributed to the vibrational modes of elemental SiO4tetrahedral forming the structure of the SiO2 thin films.14,19–21 They can be assigned to a transversal optical (TO1) rocking mode 共460 cm⫺1兲,aTO 2bending mode 共800 cm⫺1兲, an asymmetric stretching mode 共the big band at ⬃1070 cm⫺1兲,aTO 4asymmetric mode 共1170 cm⫺1兲, and its corresponding longitudinal mode LO4共1200 cm⫺1兲. These two latter bands give rise to the well defined shoulder at the right-hand side of the most intense band at ⬃1090 cm⫺1. SiuOH and SiuO⫺rocking 共578 cm⫺1兲and stretching 共960 cm⫺1兲modes can be also detected in the spectra. The 960 cm⫺1peak increases with the value of Rin agreement with an increase of the incorporation of OH⫺groups in the less compact SiO2thin films. The relative intensity of the shoulder at 1200 cm⫺1共due to the contribution of several elemental bands as has been mentioned herein兲has sometimes been taken as a measure of the degree of porosity of the films 共see Ref. 14 and references therein and Ref. 19兲. Although some other factors besides compactness may contribute to this intensity, the results obtained here show a parallel tendency of shoulder intensity, rms surface roughness values and, as it will be shown in the next subsection, decrease in the density of the films. C. Optical properties The optical properties of the SiO2thin films prepared under different conditions were determined by spectroscopic ellipsometry. Figure 6 shows the refractive index n(␭) and the extinction coefficient k(␭) as a function of the wavelength ␭, for thin films prepared with different values of R. The plot shows that the refractive index of the films decreases as the relative amount of Ar in the plasma gas increases, for the entire wavelength range. FIG. 3. AFM images of SiO2thin films prepared with R⫽0共a兲and 0.6 共b兲. FIG. 4. rms surface roughness of SiO2thin films as a function of R. The line is plotted to guide the eyes. FIG. 5. FT-IR spectra of SiO2thin films prepared with different values of R. 903 Barranco et al. : Plasma enhanced chemical vapor deposition of SiO2903 JVST A - Vacuum, Surfaces, and Films It is possible to estimate changes in the density ␳ of the SiO2thin films from the variation of their refractive indices. It has been reported22,23 that in the case of SiO2,⌬ ␳ ⫽5 ⫻⌬ngcm⫺3. In the present case, a maximum change of 0.023 is found in the value of nfor practically all of the visible region. This difference in ngives a maximum difference in density ⌬ ␳ ⫽0.11 gcm⫺3, between the most and least compact SiO2thin films reported here. The less compact films are those prepared at high values of R. This tendency agrees with the higher degree of roughness and porosity deduced from the AFM and FT-IR characterization analysis of the samples 共cf. Figs. 3–5兲. It is also worth noting that all of the samples were totally transparent 共kvalues equal to zero on the whole spectral range兲with independence of the Rvalue, and that no light scattering was detected in any of the deposited films, as it would happen if the size of the aggregates were large enough to scatter visible light. The high transparency of the thin films supports the possible use of the less compact thin films as antireflective coatings for photovoltaic or similar applications.15 D. Plasma-enhanced chemical vapor deposition thin-film formation with TMSCl and mixtures ArÕO2 The choice of TMSCl as volatile precursor of Si rather than TMS has been imposed, under our experimental conditions, by the fact that no thin-film deposition is found in this latter case either by using pure O2or mixtures O2/Ar as plasma gas. This results contrasts with those reported by other authors showing the formation of SiO2thin films by using TMS in plasma microwave reactors with downstream configurations.13,24 Jauberteau et al.25 have reported that an afterglow Ar plasma may yield a large fragmentation of the TMS molecule, mainly via SiuC bond breaking. This process principally generates Si(CH3) fragments. Radicals formed through the dissociation of CuH bonds 关i.e., SiCxHy radicals (y⬍3x)] may be also produced, though with smaller efficiency. In our system, a process similar to the first one does not seem to occur because there is no deposition with TMS under any condition or with TMSCl with an Ar plasma. However, we believe that the polymerization of intermediate radicals of silicon produced through an alternative mechanism could contribute to the formation of the film 共see next兲. The different behavior found in our experiment by using TMS or TMSCl precursors8points to that a SiuCl bond in the precursor molecule favors its decomposition by the plasma to produce SiO2thin films. The fact that no chlorine is detected in the films supports that this atom is removed during the first stages of the reaction mechanism and that no, or very few, fragments containing this atom arrive to the growing film surface. Thus, it is very likely that the first or one of the initial steps in the deposition mechanism of the films is the attack of the SiuCl bond by some activated species of the plasma and/or free electrons. Since in our experimental configuration the concentration of free electrons and ionic species in the deposition zone must be rather small, we assume that oxygen species are most important for the activation of the precursor molecule. Thus, such activation could be written schematically as Si共CHx兲yCl⫹O→Si共CHx兲yO⫹Cl*. With this process, we propose that activated oxygen species intervene in the initial activation steps of the precursor molecule leading to the removal of chlorine and the formation of Si-containing radicals, either in the gas phase or adsorbed on the surface. It has to be mentioned that the intensity of the O*emission line at 777 nm remains practically constant independently of the value of the parameter R, a feature that points to a complex nature of the whole deposition process and stresses the need of considering other parameters, like concentration and energy of free electrons, other radicals, the role of Ar, etc., which are not accessible in our experiment. Although, we do not have enough evidence to propose a complete mechanism to account for the formation of the SiO2thin films, in line with the previous results by Jauberteau25 and other authors,9,10 it is reasonable to assume that the formation of the oxide film from TMSCl is a process that involves the adsorption on the substrate of either the precursor molecule or some reactive fragments resulting from its activation and eventual fragmentation. On the surface, the competition between the polymerization of these fragments to yield SiOxCyHzpolymeric chains and the oxidation or etching of organic fragments of these films leading to the removal of organic components will control the formation of SiuO bonds. In fact, previous results of our laboratory using TMSCl and pure oxygen plasmas8showed that it is possible to deposit SiOxCyHzthin films when the flow of oxygen is lower than that of the precursor flow. In the present investigation, the increase with Rof the intensity of the emis- FIG. 6. Refractive index nand extinction coefficient kfor SiO2thin films deposited with different values of R. 904 Barranco et al. : Plasma enhanced chemical vapor deposition of SiO2904 J. Vac. Sci. Technol. A, Vol. 21, No. 4, JulÕAug 2003 sion lines of OH*and CH*species 共cf. Fig. 1兲might indicate that the Ar addition favors the etching and oxidation at the surface of the organosilicon fragments, a process that has been reported previously by other authors.25,26,22 Within this phenomenological description, it is possible to reconcile the high roughness with the low growing rate as well as the SiO2composition of the prepared thin films by assuming that intermediate siliconelike layers and/or fragments are being produced during the deposition of the films. Then, this polymeric material deposited on the surface is further oxidized by the plasma to yield gases (CO2and H2O) that would be desorbed whereas SiCxOyHzlayers or fragments are being incorporated onto the surface. The release of volatile compounds is likely to yield films with a large roughness and porosity as has been actually observed here by AFM and FT-IR characterization. IV. CONCLUSIONS It has been shown that TMSCl is a suitable precursor for the deposition of SiO2thin films by PECVD using mixtures Ar/O2as plasma gas and a microwave ECR plasma system in a downstream configuration. Under these conditions, no deposition was found by using TMS, thus indicating that the existence of a CuCl bond favors the plasma activation of the precursor molecule. SiO2with no traces of Cl was obtained in all cases. This supports that Cl is being removed during the first steps of activation of the precursor molecule. It has been found that the thin-film microstructure and growth rate depend on the plasma conditions. Increasing the relative concentration of Ar in the plasma gas at constant total pressure leads to a decrease in the growing rate by a factor of 0.6 and, unexpectedly, to an increase in the film roughness and porosity. As a consequence, the refractive index of the films can be tailored by controlling the relative concentration of Ar in the plasma gas. This technique could be very useful for the synthesis of antireflective SiO2thin films of interest in photovoltaic and similar applications. ACKNOWLEDGMENTS The authors thank the Ministerio de Ciencia y Tecnologı ´a 共Project No. MAT2001-2820兲and the EU 共Project No. ERK6-CT-1999-00015兲for financial support. 1A. Grill, Cold Plasma in Materials Fabrication 共IEEE, New York, 1994兲. 2C. Charles, G. Giroult-Matlakowski, R. W. Goswell, A. Goullet, G. Turban, and C. Cardinaud, J. Vac. Sci. Technol. A 11, 2954 共1993兲. 3A. Ortiz, S. Lo ´pez, C. Falcony, M. Farias, L. C. Araiza, and G. Soto, J. Electron. Mater. 19, 1411 共1990兲. 4A. Granier, C. Valle ´e, A. Goullet, K.Aumaille, and G. Turban, J. Vac. Sci. Technol. A 17, 2470 共1999兲. 5J. Schwarz, M. Schmidt, and A. Ohl, Surf. Coat. Technol. 98,859共1998兲. 6R. Thyen, A. Weber, and C. P. Klages, Surf. Coat. Technol. 97,426 共1997兲. 7Y. Inoue, H. Sugimura, and O. Takai, Thin Solid Films 345,90共1999兲. 8A. Barranco, J. Cotrino, F. Yubero, J. P. Espino ´s, J. Benı ´tez, C. Clerc, and A. R. Gonza ´lez-Elipe, Thin Solid Films 401, 150 共2001兲. 9Plasma Processing of Polymers, edited by R. d’Agostino, P. Favia, and F. Fracassi 共Kluwer, Dordrecht, 1997兲, Chap. 5 共and references therein兲. 10C. Valle ´e, A. Goullet, A. Granier, A. van der Lee, J. Durand, and C. Marlie `re, J. Non-Cryst. Solids 272, 163 共2000兲. 11A. Hozumi and O. Takai, Thin Solid Films 334,54共1998兲. 12D. S. Kim and Y. H. Lee, Thin Solid Films 283, 109 共1996兲. 13A. Grill and V. Patel, J. Appl. Phys. 85, 3314 共1999兲. 14A. Barranco, F. Yubero, J. Cotrino, J. P. Espino ´s, J. Benitez, T. C. Rojas, J. Allain, T. Girardeau, J. P. Rivie `re, and A. R. Gonza ´lez-Elipe, Thin Solid Films 396,9共2001兲. 15H. Nagel, A. Metz, and R. Hezel, Sol. Energy Mater. Sol. Cells 65,71 共2001兲. 16J. Cotrino, A. Palmero, V. Rico, A. Barranco, and A. R. Gonza ´lez-Elipe, J. Vac. Sci. Technol. B 19, 410 共2001兲. 17National Institute of Standards and Technology, http://physics.nist.gov 共2002兲. 18R. W. B. Pearse and A. G. Gayden, The Identification of the Molecular Spectra 共Chapman and Hall, London, 1976兲. 19C. T. Kirk, Phys. Rev. B 38, 1255 共1988兲. 20R. M. Almeida, T. A. Guiton, and C. G. Pantano, J. Non-Cryst. Solids 121, 193 共1990兲. 21P. Lange, J. Appl. Phys. 66, 2001 共1989兲. 22L. Zajı ´ckova ´, J. Janca, and V. Perina, Thin Solid Films 338,49共1999兲. 23R. A. B. Devine, Trans. Mater. Res. Soc. Jpn. 8, 165 共1992兲. 24A. M. Wro ´bel and Y. Hatanaka, in Plasma Processing of Polymers, edited by R. d’Agostino, P. Favia, and F. Fracassi 共Kluwer, Dordrecht, 1997兲, p. 334. 25J. L. Jauberteau, I. Jauberteau, and J. Aubreton, Chem. Phys. Lett. 327, 351 共2000兲. 26C. E. Viana, N. I. Morimoto, and O. Bonnaud, Microelectron. Reliab. 40, 613 共2000兲. 905 Barranco et al. : Plasma enhanced chemical vapor deposition of SiO2905 JVST A - Vacuum, Surfaces, and Films