G. Ziegler et al./Composites: Part 4 30(1999)411-417 Fig. 2. Comparison of the microstructure of the(a)C/SiCN and (b) Hi-Nicalon/SiCN composites. density change as a function of the number of infiltration the matrix has already reached its saturation crack density and pyrolysis cycles. upon cooling due to the thermal expansion mismatch of fib After three infiltration and pyrolysis cycles, a density of and matrix(a F(axial)0, aM=3.0X 10K) 1. 4 g/cm is reached in the C/SiCN system with an open The residual stresses upon cooling due to thermal expa porosity of 28%, which is further reduced to 16% and sion mismatch between fiber and matrix were calculated 12% after five and seven cycles, respectively. The final using a simple model, as described elsewhere [8], assuming geometrical density is approximately 1.7 g/em,, bulk a stress-free state at the pyrolysis temperature. The calcu- density 1.9 g/cm. A closed porosity of about 10% cannot lated stresses axial and radial between fiber and matrix(C/ be eliminated and is present already after three cycles. Fig. 2 SiCN) were determined to be 780 and 90 MPa, respectively shows the comparison of the microstruture of the C/Sicn Upon cooling, the fiber should therefore shrink away from and Hi-Nicalon/SiCN composites the matrix, which was observed. The matrix is also under Microstructural evolution and conclusions regarding the tension and with a matrix yield strength(as has been experi stress-strain behavior of the C/SiCn composites were mentally determined on monolithic material), an array of monitored on fracture surfaces of tested composite materials regular matrix cracks will develop, as has been confirmed P3-P7(F1g.3) by SEM-micrographs on polished cross-sections As is evident, with an increasing number of infiltration Different results are obtained when working with C cycles, a more continuous matrix develops, but also the coated Hi-Nicalon fibers. Fig. 5 shows the stress-strain fracture characteristics are changing. Fig. 3(b)shows the diagrams of these composites after five and seven infiltra microstructure of a composite after five cycles, clearly tion cycles. The high strength values are obviously due to ith pull-out lengths >100 um. However, with increasing matrix density, the Polished cross-sections of a Hi-Nicalon/SiCn material fracture behavior becomes more brittle, as is seen in were used to get some insight into the matrix formation Fig. 3(c). This can be attributed to the development of (Fig. 6). Various points can be made. First, different infiltra local residual compressive stresses at the fiber interface, tion cycles(marked as I and 2 in Fig. 6) can be distin- caused by the large shrinkage(30% linear) of the polymer guished, appearing as lighter and darker areas. Second, upon pyrolysis. In addition, stresses caused by the thermal looking at the crack patterns in matrix areas adjacent to expansion mismatch between fiber and matrix can develop, the fibers(lst cycle)it is obvious that the matrix may shrink although, as calculations have shown, the interface should onto the fibers (3) but also away from them(4), depending be in tension in this system upon cooling from the proces- on the local geometric conditions. The arrow marks a condi sing temperature tion, where the fiber applies a load on the surrounding The corresponding stress-strain curves are shown in Fig. matrix material, with subsequent crack formation at the 4. Comparing composite materials P5 and P7, the more center of the matrix area. Similar observations were also brittle behavior of composite P7 is evident made in the C/SiCN system, indicating that this phenomena The overall strength level, however, is low, compared to is of general meaning the strength of the C fibers. Evaluation of the fiber surface after annealing at 1400"C in N2, showed substantial kink 3.2. Coeficient of thermal expansion formation at the surface, thus degrading the strength. No further matrix cracking is observed during loading, since Further characterization of the cision material includeddensity change as a function of the number of infiltration and pyrolysis cycles. After three infiltration and pyrolysis cycles, a density of 1.4 g/cm3 is reached in the C/SiCN system with an open porosity of 28%, which is further reduced to 16% and 12% after five and seven cycles, respectively. The final geometrical density is approximately 1.7 g/cm3 , bulk density 1.9 g/cm3 . A closed porosity of about 10% cannot be eliminated and is present already after three cycles. Fig. 2 shows the comparison of the microstruture of the C/SiCN and Hi-Nicalon/SiCN composites. Microstructural evolution and conclusions regarding the stress–strain behavior of the C/SiCN composites were monitored on fracture surfaces of tested composite materials P3–P7 (Fig. 3). As is evident, with an increasing number of infiltration cycles, a more continuous matrix develops, but also the fracture characteristics are changing. Fig. 3(b) shows the microstructure of a composite after five cycles, clearly demonstrating pull-out behavior with pull-out lengths .100 mm. However, with increasing matrix density, the fracture behavior becomes more brittle, as is seen in Fig. 3(c). This can be attributed to the development of local residual compressive stresses at the fiber interface, caused by the large shrinkage (30% linear) of the polymer upon pyrolysis. In addition, stresses caused by the thermal expansion mismatch between fiber and matrix can develop, although, as calculations have shown, the interface should be in tension in this system upon cooling from the proces￾sing temperature. The corresponding stress–strain curves are shown in Fig. 4. Comparing composite materials P5 and P7, the more brittle behavior of composite P7 is evident. The overall strength level, however, is low, compared to the strength of the C fibers. Evaluation of the fiber surface, after annealing at 14008C in N2, showed substantial kink formation at the surface, thus degrading the strength. No further matrix cracking is observed during loading, since the matrix has already reached its saturation crack density upon cooling due to the thermal expansion mismatch of fiber and matrix (aF(axial) < 0, a M < 3.0 × 1026 K21 ). The residual stresses upon cooling due to thermal expan￾sion mismatch between fiber and matrix were calculated using a simple model, as described elsewhere [8], assuming a stress-free state at the pyrolysis temperature. The calcu￾lated stresses axial and radial between fiber and matrix (C/ SiCN) were determined to be 780 and 90 MPa, respectively. Upon cooling, the fiber should therefore shrink away from the matrix, which was observed. The matrix is also under tension and with a matrix yield strength (as has been experi￾mentally determined on monolithic material), an array of regular matrix cracks will develop, as has been confirmed by SEM-micrographs on polished cross-sections. Different results are obtained when working with C￾coated Hi-Nicalon fibers. Fig. 5 shows the stress–strain diagrams of these composites after five and seven infiltra￾tion cycles. The high strength values are obviously due to the undamaged SiC fibers and the coating. Polished cross-sections of a Hi-Nicalon/SiCN material were used to get some insight into the matrix formation (Fig. 6). Various points can be made. First, different infiltra￾tion cycles (marked as 1 and 2 in Fig. 6) can be distin￾guished, appearing as lighter and darker areas. Second, looking at the crack patterns in matrix areas adjacent to the fibers (1st cycle) it is obvious that the matrix may shrink onto the fibers (3) but also away from them (4), depending on the local geometric conditions. The arrow marks a condi￾tion, where the fiber applies a load on the surrounding matrix material, with subsequent crack formation at the center of the matrix area. Similar observations were also made in the C/SiCN system, indicating that this phenomena is of general meaning. 3.2. Coefficient of thermal expansion Further characterization of the C/SiCN material included G. Ziegler et al. / Composites: Part A 30 (1999) 411–417 413 Fig. 2. Comparison of the microstructure of the (a) C/SiCN and (b) Hi-Nicalon/SiCN composites