JOURNAL OF MATERIALS SCIENCE LETTERS 7 (1988) 885-890

Characterization of some pyrolysed polycarbosilanes by transmission electron microscopy J. AYACHE, S. B O N N A M Y , X. BOURRAT, A. DEURBERGUE, Y. MANIETTE, A. OBERLIN Laboratoire Marcel Mathieu, UA 1205 CNRS-Universit~, 2, avenue du Prdsident P. Angot, F64000 Pau, France E. BACQUE, M. BIROT, J. DUNOGUES, J-P. PILLOT Laboratoire de Chimie Organique du Silicium et de I'Etain, Universit6 de Bordeaux I, Chimie Organique, L. A. no. 35 CNRS, 351, cours de la Lib6ration, F33405 Talence Cedex, France

The exploration, either complete or radial, of reciprocal space has long been used by one of us [1-3] to characterize poorly organized materials, such as turbostratic carbons, or amorphous ones [4]. Particles the order of magnitude of resolution, i.e. 0.5 nm in size, could also be identified in a mixture of phases [5]. These techniques of transmission electron microscopy (TEM) use one of all the scattered beams emitted by the specimen, selected by a suitable objective aperture displaced relative to the selected-area electron diffraction (SAD) pattern. The regions emitting a given beam or portion of a diffraction ring appear in the image as bright domains in a dark field (DF). After having successfully applied these techniques to commercial SiC fibres [6], the present letter will present results obtained from pyrolysed polycarbosilanes (PCS). Four different polymers (Table I) were pyrolysed up to 1000°C in an alumina boat, under an argon flow. A heating rate of 4 ° C rain I without residence time at maximum temperature was used. For TEM examination, specimens were dusted on an amorphous carbon film, as thinly as possible. After heat-treatment, samples were mostly SiC. Polymer A~ is a commercial PCS purchased from Nippon Carbon Co. A: is a Yajima's Mark I PCS [7], obtained by thermolysis of polydimethylsilane (PDMS) in an autoclave at 470 ° C. B1 was prepared by thermolysis (450 ° C, autoclave) of dodecamethylcyclohexasilane, according to Yajima's procedure [8]. The polyearbosilane B: was prepared by thermolysis of a copolymer obtained from Me2 SIC12 and C1HMeSiCH2 SiCH: SiMeHCI [9]. A fl-SiC crystal yields 111 (0.251nm), 220 (0.154 nm), 3 1 1 (0.131 nm) and possibly 2 0 0 scattered beams (0.217 nm), the most intense of which is 1 1 1. In a fragment of heat-treated PCS, if not amorphous, SiC crystals are known to be small and to have all orientations. Correspondingly a Debye-Scherrer pattern is obtained. The smaller the crystals, the more diffuse and faint are the h k l rings (Fig. l a). If the specimen is amorphous, the distribution of atom pairs is random with a higher probability of tetrahedral bond corresponding to the (1 1 1) spacing. Any phase other than SiC forming particles larger than the resolution of the microscope appears as a 0261-8028/88 $03.00 + .12 © 1988 Chapman and Hall Ltd.

bright domain if one of its own scattered beams passes through the aperture. Then this beam can be physically selected from all other scattered beams, the "foreign" particle is the only one illuminated. In this work, the aperturewas chosen small enough to separate aromatic carbon and silica from SiC (Fig. lb). The 2.4nm -1 aperture size allows a resolution of 0.5nm. It lets through a scattered beam within an azimuthal and radial tolerance given by its diameter. Correspondingly, it permits azimuthal twists of particles relative to the centred position. After centring on 2.4, 4.2 or 6.1 nm -~, the azimuthal twists permitted are, respectively, _+25° , -t- 17.5° and + 12°. The radial tolerance of the aperture admits beams corresponding to interplanar distance ranges of, respectively, 0.28 to 0.90 nm, 0.19 to 0.33 nm and 0.14 to 0.21 nm. These data are gathered in Fig. 1 where the various radial positions of the aperture used are indicated. In summary, in position 1 turbostratic carbons and various forms of silica can be imaged in DF; SiC is excluded. In position 2, a crystalline SiC 1 1 1 DF image is formed, superimposed to a very weak carbon 10 DF image. In position 3, only SiC crystals can be imaged. If the material is amorphous [4], very small and faintly bright dots appear for any aperture centring and any focusing value. They correspond to the statistical distribution of couples of atoms within the specimen. In this case, free carbon or silica cannot be distinguished from SiC. TEM techniques were first applied to transverse thin sections of commercial Nicalon fibres [6].

T A B L E I Experimental data for characterization Sample Silica Carbon

SiC

Poorly organized

Well organized

Amorphous Crystallized Strongly heterogeneous Yes Yes Strongly heterogeneous Yes Yes Very homogeneous Very small crystals Heterogenous Yes Yes

A1

Yes

Yes

No

A2

Yes

Yes

No

B2

No

No

Yes

B~

No

Yes

No

885

Figure l (a) SAD pattern of microcrystalline SiC. (b) The 2.4nm -~ aperture is represented, superimposed on the pattern for positions 1, 2 and 3.

The following questions could be answered. Is the sample homogeneous or not? Does it contain amorphous, or crystallized, SiC? What is the size, distribution and, if any, orientation in space of the crystals? Does it contain phases other than SiC? Virtually all types of data could be obtained from commercial fibres. Micrographs (Figs 2 to 5) illustrate the same various features found in pyrolysed samples in the present work. Table I summarizes the results obtained from each material. With one exception, all samples are heterogeneous. Depending on the particle studied, the major SiC phase is found to be made of crystals of various sizes, associated, or not, with amorphous material. Fig. 2 corresponds to aperture position 2, whereas Fig. 3 corresponds to position 3. A mixture of SiC crystals of various sizes is observed. One of the largest crystals

(about 5 nm) is circled in Fig. 2. On the other hand, the homogeneous sample B2 (Table I and Fig. 4) is made up only of very small domains. These are crystalline, because they appear bright only for position 2 or 3 of the aperture. In all other samples (Table I), some more or less extended regions do not show bright domains, either in position 2 or in position 3 (arrow in Fig. 5a). Their only contribution to the image is a faint and homogeneous illumination, due to inelastic scattering. In position 1, these regions are seen to be full of bright dots smaller than 1 nm (Fig. 5b). Such dots also occupy the dark regions of Fig. 5a. In Fig. 5b (arrows) a higher density of bright dots is seen along the particle edge. It disappears when the aperture is displaced along a circle. In some cases (Fig. 6a), edges are decorated by a very brilliant rim (arrow), also disappearing when the aperture is rotated (Fig. 6b). The

Figure 2 Dark-field image of commercial precursor A~ for position 2 (SiC 1 1 1 DF). A large SiC crystal is circled.

886

Figure 3 Dark-field image of the same particle as in Fig. 2 but for position 3 (sic 220 DF).

complete disappearance of the particle edge decoration when the aperture is rotated 90 ° is a general phenomenon. These data suggest that free aromatic carbon coexists with SiC crystals about 2 nm in size. In some regions carbon alone is present. Carbon also tends to form oriented rims along particle edges. The fact that different regions show bright domains for all three position of the aperture proves that at least two different phases may coexist in one particl e . The material imaged in position I is either free carbon or silica, whereas that imaged in positions 2 and 3 is SiC. The fact that brilliant rims or areas of high bright

dot density along particle edges disappear when the aperture is rotated along a circle in position 1 implies a preferred orientation of either carbon or silica. SiC crystals are known to be able to produce easily turbostratic carbon shells [10, 11]. The occurrence of crystallized silica could not be proved. Therefore, we are inclined to attribute the accumulation of the unknown phase on particle edges to very small aromatic layers or layer stacks issued from SiC. In some micrographs, the direction of lattice planes either C or SiO2 - is indicated by a bar (Figs 5b and 6). If they represented turbostratic carbon, they would correspond to aromatic layers lying fiat on the particle lateral surface.

Figure 4 Dark-field image of Be for position 2 (SiC I I 1 DF).

887

Figure 5 Dark-field image of commercial precursor At. (a) Position 2 (SiC 1 1 1 DF), only SiC appears bright. (b) Position 1 (C or silica DF), only carbon or silica appear bright.

888

Figure 6 Dark-field images o f B2. (a) Position 1 (C or silica DF). (b) Position 1 (aperture rotated 90 ° relative to the pattern).

889

References 1.

2. 3. 4. 5. 6.

890

A. O B E R L I N , G. T E R R I E R E , B. D U R A N D and C. C L I N A R D , in "Advances in Organic Geochemistry 1971" (Pergamon, Oxford, 1972) p. 577. A. O B E R L I N and G. T E R R I E R E , J. Microseopie 14 (1972) 1. A. O B E R L I N , G. T E R R I E R E and J. L. B O U L M I E R , ibid. 21 (1974) 301. A. O B E R L I N , M. O B E R L I N and M. M A U B O I S , Phil. Mag. 32 (1975) 833. A. O B E R L I N , M. ENDO and T. K O Y A M A , J. Crystal Growth 32 (1976) 335. M. G U I G O N and A. O B E R L I N , Contract report D R E T , No. 84087 (1986).

7. 8. 9. 10. 11.

S. Y A J I M A , Y. H A S E G A W A , J. H A Y A S H I and M. I I M U R A , J. Mater. Sci. 13 (1978) 2569. S. Y A J I M A , J. H A Y A S H I and M. OMOR1, Chem. Lett. (1975) 931. J.-P. PILLOT, E. BACQUE, J. D U N O G U E S and P. OLRY, French Patent, D e m a n d No. 8607 899 (1986). D. V. B A D A M I , Carbon 3 (1965) 53. A. O B E R L I N , M. O B E R L I N and J . R . CONTET R O T E T , J. Microsc. Spectrosc. Electron. ! (1976) 391.

Received 22 January and accepted 23 March 1988