S. Bertrand et al. /Journal of the European Ceramic Society 20(2000)1-13 Fibre Interphase Matrix Fibre Fibre Matrix Fibre un 200nm Fig. 11. TEM ction of a minicomposite: as-received Hi-Nicalon/(PyC3/SiC3ohlo/SiC. Note the sharp interfaces controlled by the poor Sic crystallisation obtained by P-CVE (a) low magnification and (b) higher magnification on the interfacial sequence. reated Hi-Nicalon fibre. The residual carbon layer evi- denced by aEs is visible but with a thickness, usually thinner than 50 nm. It is in fact a bilayer. A rather dense carbon(less than 10 fringes)is lying directly on the surface of the fibre. Then a poorly organised carbon, at least 30 nm-thick, is observed. In Fig. 13, it can be seen than the first pyrocarbon sublayer (PyC1) is deposited on this poorly organised carbon with a more densely packed The two fibre surfaces are thus different, their rough ness meanwhile looking equivalent. In short, depending on the occurrence of a pretreatment or not, the first interface is either a SiC/PyC interface or a free-C/PyC 4.1.3. Nanostructure of the PyC sublayers The nanostructure of the Pyc-sublayers has been udied by TEM and optical microscopy. Values of the extinction angle(Ae),obtained for the PyC deposited by P-CVl, were all around 18-19. Each PyC sublayer grows onto surfaces made of well nanocrystallised SiC +C which exhibit some roughness at the nm-scale epted for the first sublayer). As shown in Figs. 12 13, the PyC first fills the concave parts of the Sic based substrate, at a distance the Pyc aromatic layers tend to deposit parallel to the mean surface of the Fig 12. Interface of an untreated hi-Nicalon fibre reinforced sic coated fibre and exhibit a pronounced anisotropy composite, with a(PyC3/SiC3o)o interphasetreated Hi-Nicalon ®bre. The residual carbon layer evi￾denced by AES is visible but with a thickness, usually, thinner than 50 nm. It is in fact a bilayer. A rather dense carbon (less than 10 fringes) is lying directly on the surface of the ®bre. Then a poorly organised carbon, at least 30 nm-thick, is observed. In Fig. 13, it can be seen than the ®rst pyrocarbon sublayer (PyC1) is deposited on this poorly organised carbon with a more densely packed stacking. The two ®bre surfaces are thus di€erent, their rough￾ness meanwhile looking equivalent. In short, depending on the occurrence of a pretreatment or not, the ®rst interface is either a SiC/PyC1 interface or a free-C/PyC1 interface. 4.1.3. Nanostructure of the PyC sublayers The nanostructure of the PyC-sublayers has been studied by TEM and optical microscopy. Values of the extinction angle (Ae),21 obtained for the PyC deposited by P-CVI, were all around 18±19. Each PyC sublayer grows onto surfaces made of well nanocrystallised SiC+C which exhibit some roughness at the nm-scale (excepted for the ®rst sublayer). As shown in Figs. 12 and 13, the PyC ®rst ®lls the concave parts of the SiC￾based substrate, at a distance, the PyC aromatic layers tend to deposit parallel to the mean surface of the coated ®bre and exhibit a pronounced anisotropy. Fig. 11. TEM cross-section of a minicomposite: as-received Hi-Nicalon/(PyC3/SiC30)10/SiC. Note the sharp interfaces controlled by the poor SiC crystallisation obtained by P-CVI: (a) low magni®cation and (b) higher magni®cation on the interfacial sequence. Fig. 12. Interface of an untreated Hi-Nicalon ®bre reinforced SiC composite, with a (PyC3/SiC30)10 interphase. S. Bertrand et al. / Journal of the European Ceramic Society 20 (2000) 1±13 9