Table Il. Summary of Tow and Composite Properties o(MPa) Var(MPa UTS(MPa) N720/porous oxide 0.164±0.015 750±125 124±30 138±9 ncoated-after HT N720/porous oxide 0.164±0.015 470±67 fugitive--after HT N720/CAS--as-processed 0.325±0.02 750±1 22.4±3 N7200.04m 0.325±0.02 477±10 1主45 C/CAS-as-processed itive-after ht 0.325±0.025 477±102 156±45 750±125 245±60 “CCAs- as-processed Fugitive--after HT 0.325±0.02 750±125 245±60 0.23 200 coated 150 田1150°c/100hair △1150°C/500hair Fig. 8. Composite data normalized with respect to tow strength for Nextel 720 porous matrix composites. or is based on sample dimensions, umber of fabric layers, and either the manufacturer's (uncoated)or measured(CVD"C )strength data. Results show little difference between uncoated and"C" composites. 0.5mm measured moduli for the control composites ("C ) are wi graph of a typical Nextel 720 porous matrix predicted range. The moduli of the fugitive("C-removed)com- posites are less than the "C composites, but they are still 15% of the predicted rule-of-mixtures value, at 130-145 GPa. If the modulus was fully dominated by either fibers or the matrix, then the composite modulus would have upper bounds of only 90 these composites. For a composite GPa and 64 GPa, respectively (L 0 or V=0 in Eq(1). perpendicular to the stress axis(ag Clearly, significant load transfer is present even after the"C"layer removed. note. however that the modulus does decrease as the ugitive"C layer thickness(thus the gap width) is increased from E=Eol 1-2 0.02 to 0.04 um. The optimal fugitive layer thickness, therefore, is (3) likely to be <0. I um for this system, because load transfer controls ultimate tensile strength and modulus This is in the same where Eo is the modulus of the dense matrix, and p the porosity range as the expected roughness amplitude, if it is taken to scale For a volume fraction of 0.3-0.35, Eq (3 )predicts the transverse with the grain size of the crystallites in the fiber, which are on the modulus to be 32-37 GPa. The"C"and fugitive composites have order of o1 a modulus between this and the value predicted by Eq(2)for a The modulus of a composite in the off-axis(+45)orientation well-bonded interface. This is understandable, because, with in- is more difficult to calculate, however, the lower bound is the creasing strain, the gap between the fiber and matrix tends to close transverse modulus(90%), E,, expressed by the following approx- up perpendicular to the stress axis and eventually leads to load imate rule: 36 transfer between the fiber and matrix The modulus results of the +45 composites indicated that the off-axis load transfer was not as efficient as in the oo direction Et Er Off-axis loading of the +45 samples would be expected to improve the mechanical interlocking(due to shear components), The above equation applies only to composites where the fiber and yet the interfacial gap would provide little transverse reinforce- matrix are strongly bonded, and displacements are continuous ment. The latter effect appeared to dominate in this case. A across the interface. From the values in Table I, the transverse fabric, if used to reinforce a fugitive dense matrix com modulus is calculated to be 122.5+ 2.5 GPa. The modulus of the would increase the mechanical interlocking and subsequent uncoated composite(Fig. 3(a)) is within 15% of this predicted load transfer between the fiber and the matrix. However unidirec value. However the"C and fugitive composites have much lower tional strength would decrease because of the reduction in fiber moduli, as might be expected from the weak interface strengths in volume percentage in the loading directionmeasured moduli for the control composites (“C”) are within this predicted range. The moduli of the fugitive (“C”-removed) com￾posites are less than the “C” composites, but they are still within 15% of the predicted rule-of-mixtures value, at 130–145 GPa. If the modulus was fully dominated by either fibers or the matrix, then the composite modulus would have upper bounds of only 90 GPa and 64 GPa, respectively (Vm 5 0 or Vf 5 0 in Eq. (1)). Clearly, significant load transfer is present even after the “C” layer is removed. Note, however, that the modulus does decrease as the fugitive “C” layer thickness (thus the gap width) is increased from 0.02 to 0.04 mm. The optimal fugitive layer thickness, therefore, is likely to be ,0.1 mm for this system, because load transfer controls ultimate tensile strength and modulus. This is in the same range as the expected roughness amplitude, if it is taken to scale with the grain size of the crystallites in the fiber, which are on the order of 0.1 mm.35 The modulus of a composite in the off-axis (645°) orientation is more difficult to calculate; however, the lower bound is the transverse modulus (90°), Et , expressed by the following approx￾imate rule:36 1 Et 5 Vf Ef 1 Vm Em (2) The above equation applies only to composites where the fiber and matrix are strongly bonded, and displacements are continuous across the interface. From the values in Table I, the transverse modulus is calculated to be 122.5 6 2.5 GPa. The modulus of the uncoated composite (Fig. 3(a)) is within 15% of this predicted value. However the “C” and fugitive composites have much lower moduli, as might be expected from the weak interface strengths in these composites. For a composite with cylindrical holes aligned perpendicular to the stress axis (again 90°), Rice37 gives the following relationship: E 5 EOF 1 2 2S P pD 1/ 2G (3) where EO is the modulus of the dense matrix, and P the porosity. For a volume fraction of 0.3–0.35, Eq. (3) predicts the transverse modulus to be 32–37 GPa. The “C” and fugitive composites have a modulus between this and the value predicted by Eq. (2) for a well-bonded interface. This is understandable, because, with in￾creasing strain, the gap between the fiber and matrix tends to close up perpendicular to the stress axis and eventually leads to load transfer between the fiber and matrix. The modulus results of the 645° composites indicated that the off-axis load transfer was not as efficient as in the 0° direction. Off-axis loading of the 645° samples would be expected to improve the mechanical interlocking (due to shear components), yet the interfacial gap would provide little transverse reinforce￾ment. The latter effect appeared to dominate in this case. A woven fabric, if used to reinforce a fugitive dense matrix composite, would increase the mechanical interlocking and subsequently the load transfer between the fiber and the matrix. However, unidirec￾tional strength would decrease because of the reduction in fiber volume percentage in the loading direction. Table II. Summary of Tow and Composite Properties Material Vf † sf ‡ (MPa) Vfsf (MPa) UTS (MPa) UTS/ Vfsf N720/porous oxide uncoated—after HT 0.164 6 0.015 750 6 125 124 6 30 138 6 9 1.11 N720/porous oxide fugitive—after HT 0.164 6 0.015 470 6 67 77 6 20 102 6 23 1.32 N720/CAS—as-processed 0.325 6 0.025 750 6 125 245 6 60 22.4 6 3 0.09 N720/0.04 mm “C”/CAS—as-processed 0.325 6 0.025 477 6 102 156 6 45 122.5 6 40 0.785 Fugitive—after HT 0.325 6 0.025 477 6 102 156 6 45 92 6 42 0.59 N720/0.02 mm “C”/CAS—as-processed 0.325 6 0.025 750 6 125 245 6 60 84 6 51 0.34 Fugitive—after HT 0.325 6 0.025 750 6 125 245 6 60 56 6 15 0.23 † Volume of fibers in loading direction; ‡ Average measured tow strength. Fig. 8. Composite data normalized with respect to tow strength for Nextel 720 porous matrix composites. Vf sf is based on sample dimensions, number of fabric layers, and either the manufacturer’s (uncoated) or measured (CVD “C”) strength data. Results show little difference between uncoated and “C” composites. Fig. 9. Optical micrograph of a typical Nextel 720 porous matrix composite fracture surface. 334 Journal of the American Ceramic Society—Keller et al. Vol. 83, No. 2