G. Ziegler et al./Composites: Part A 30(1999)411-417 b) K Fig 3. Fracture surfaces of a C/SICN composite after(a)three, (b)five and (c) seven infiltration and pyrolysis cycles, and schematic model for the formation of esidual stresses the determination of coefficients of thermal expansion adjusted by varying the fiber orientation. The changes after (CTE)parallel and perpendicular to the fiber axis of UD- fferent infiltration and pyrolysis cycles are small laminates(Fig. 7). As expected, the thermal expansion of In corporation with the TU Munich, which performed unidirectional reinforced C/SiCN composites perpendicular FEM calculations with respect to an optimal prepreg config to the fiber orientation is higher than parallel. This is due to uration in order to minimize the 2D anisotropy, C fiber the anisotropy of the Cte of the carbon fiber which(accord- laminates with varying configurations(0%, 30., 45, etc. ing to the manufacturers data) varies by two orders of were prepared ( Table I magnitude perpendicular versus parallel to the fiber axis As is shown in Fig 8(a), isotropic properties in the x-and C fibers embedded in a SiCN matrix, however, exhibit a y-direction were achieved already for a simple 0/90 CTE-ratio of only 2-4(perpendicular/parallel). The prepreg(configuration 0) with eight symmetrical arranged measured average CTE values of C/SiCn (20-1000C) layers. Fig. 8(b) shows configuration 6(0/90/45), with are between1.5×10-6K-l( parallel) and45×10-6K basically the same result (-direction is perpendicular to (perpendicular). Between these two bounds, the Cte can be the x-y-area)the determination of coefficients of thermal expansion (CTE) parallel and perpendicular to the fiber axis of UD￾laminates (Fig. 7). As expected, the thermal expansion of unidirectional reinforced C/SiCN composites perpendicular to the fiber orientation is higher than parallel. This is due to the anisotropy of the CTE of the carbon fiber which (accord￾ing to the manufacturer’s data) varies by two orders of magnitude perpendicular versus parallel to the fiber axis. C fibers embedded in a SiCN matrix, however, exhibit a CTE-ratio of only 2–4 (perpendicular/parallel). The measured average CTE values of C/SiCN (20–10008C) are between 1.5 × 1026 K21 (parallel) and 4.5 × 1026 K21 (perpendicular). Between these two bounds, the CTE can be adjusted by varying the fiber orientation. The changes after different infiltration and pyrolysis cycles are small. In corporation with the TU Munich, which performed FEM calculations with respect to an optimal prepreg config￾uration in order to minimize the 2D anisotropy, C fiber laminates with varying configurations (08, 308, 458, etc.) were prepared (Table 1). As is shown in Fig. 8(a), isotropic properties in the x- and y-direction were achieved already for a simple 08/908- prepreg (configuration 0) with eight symmetrical arranged layers. Fig. 8(b) shows configuration 6 (08/908/458), with basically the same result (z-direction is perpendicular to the x–y-area). 414 G. Ziegler et al. / Composites: Part A 30 (1999) 411–417 Fig. 3. Fracture surfaces of a C/SiCN composite after (a) three, (b) five and (c) seven infiltration and pyrolysis cycles, and schematic model for the formation of residual stresses