1. Davies et al. /Journal of the European Ceramic Society 25(2005)599-604 independent of the composite specimen gauge length and to have the form: 1.38 4(h) rSo λ(m) 1380 oC/vacuum where r is the fibre radius(4.03 um20),(h)is the mean fibre 1200°/ai pullout length, and A(m)is a function only of m. In order 亏04 compare values of So obtained under different conditions, So was normalised to a gauge length, Lo, of 10-m using the Weibull scaling relationship, i.e. 0.0 So(Lo =10-m)=So Fibre pullout length(oqm) Values of t within the fibre bundle were estimated using the Fig 3. Fibre pullout length distributions for an orthogonal 3-D woven following rearrangement of Eq ( 2): 1.38 SiC/SiC composite tested under various conditions 2,20 ri(m)So (4) has been presented in Fig. 2(b) which indicates a strong correlation between fibres with low h values and the ex- istence of flat fracture surfaces(Fig. 1(b); in contrast to 3. Results and discussion fracture mirrors(Fig. 1(a)) which typically occur when crack deflection mechanisms are present. That the major 3.1. Fibre pullout length ity of fibres with significant pullout lengths and fracture mirrors were concentrated towards the centre of the fibre bundle was consistent with previous work2 that suggested maximum oxidation damage to have occurred at the fibre length within the fibre bundle with it being clear that the bundle perimeters. A likely mechanism for this phenomenon majority of fibres, particularly those adjacent to the fibre would be the propagation of matrix cracks in the transverse bundle perimeter, exhibited either zero or negligible fibre fibre bundles, allowing oxygen to access the perimeters of pullout and indicative of brittle failure, i.e. suppression of crack deflection mechanisms at the fibre/matrix interface 14 longitudinal (-axis) fibre bundles due to t being excessively high. Further evidence for this Although the largest h values were observed at the fibre bundle centre. these values were still considerable smaller when compared to those of specimens tested at rT or in vac e in the absence of oxidati age. Fig. 3 com pares fibre pullout length distributions for specimens tested under a variety of conditions.20 with(h)for the 1100.C/air 10, pecimen being an order of magnitude lower compared to that of the RT case(810 um). With reference to Eq(4),it would thus be expected that T for the 1 C/air specimen even in the fibre bundle centre where oxidation damage was minimal, would still be significantly larger compared to the values of 5-10 MPa previously measured in non-oxidised E8E3E 3. 2. Fibre strength parameters Fibre strength distributions(normalised to a gauge length of 10-3m) at the centre and edge2 of individual fibre bun dles in RT and 1 C/air specimens have been presented in Fig. 3 with data for the respective values of So and m being given in Table 1. Whereas the centre and edge fibre strength distributions were similar for the rt case(Fig 4(a)) Fig. 2. Positional dependence of properties within a single fibre bundle in an orthogonal 3-D SiC/SiC composite tested at 1100C in air:(a) 2 For the 1100C/air specimen, fibre strengths fibre pullout length, and(b)existence of flat surface or fracture mirror. could not be measured due to the lack of fracture mirrors(Fig. 2(b). Data Note that fibre pullout lengths in(a) have been shown using a logarithmic for edge fibres was thus calculated using fibres adjacent to the embrittled scaleI.J. Davies et al. / Journal of the European Ceramic Society 25 (2005) 599–604 601 independent of the composite specimen gauge length and to have the form:1,38 δc = 4 h λ(m) = rSo τ (2) where r is the fibre radius (4.03
m20), h is the mean fibre pullout length, and (m) is a function only of m. In order to compare values of So obtained under different conditions, So was normalised to a gauge length, Lo, of 10−3 m using the Weibull scaling relationship, i.e. So(Lo = 10−3m) = So
δc 10−3 1/m (3) Values of τ within the fibre bundle were estimated using the following rearrangement of Eq. (2): 1,38 τ = rλ(m)So 4 h (4) 3. Results and discussion 3.1. Fibre pullout length Fig. 2(a) illustrates the spatial dependence of pullout length within the fibre bundle with it being clear that the majority of fibres, particularly those adjacent to the fibre bundle perimeter, exhibited either zero or negligible fibre pullout and indicative of brittle failure, i.e. suppression of crack deflection mechanisms at the fibre/matrix interface14 due to τ being excessively high. Further evidence for this Distance along bundle width ( ∝m) 0 50 100 150 Flat surface Fracture mirror Distance along bundle length (∝m) 0 150 300 450 600 Distance along bundle width ( ∝m) 0 50 100 150 0 ∝m 180 ∝m (a) (b) Fig. 2. Positional dependence of properties within a single fibre bundle in an orthogonal 3-D woven SiC/SiC composite tested at 1100 ◦C in air: (a) fibre pullout length, and (b) existence of flat surface or fracture mirror. Note that fibre pullout lengths in (a) have been shown using a logarithmic scale. 1 10 100 1000 Cumulative failure 0.0 0.2 0.4 0.6 0.8 1.0 Room temperature 1200 oC/vacuum 1300 oC/vacuum 1350 oC/vacuum 1380 oC/vacuum 1100 oC/air 1200 oC/air Fibre pullout length (∝m) Fig. 3. Fibre pullout length distributions for an orthogonal 3-D woven SiC/SiC composite tested under various conditions 2,20. has been presented in Fig. 2(b) which indicates a strong correlation between fibres with low h values and the ex￾istence of flat fracture surfaces (Fig. 1(b)); in contrast to fracture mirrors (Fig. 1(a)) which typically occur when crack deflection mechanisms are present. That the major￾ity of fibres with significant pullout lengths and fracture mirrors were concentrated towards the centre of the fibre bundle was consistent with previous work2 that suggested maximum oxidation damage to have occurred at the fibre bundle perimeters. A likely mechanism for this phenomenon would be the propagation of matrix cracks in the transverse fibre bundles,8 allowing oxygen to access the perimeters of longitudinal (y-axis) fibre bundles. Although the largest h values were observed at the fibre bundle centre, these values were still considerable smaller when compared to those of specimens tested at RT or in vac￾uum, i.e. in the absence of oxidation damage. Fig. 3 com￾pares fibre pullout length distributions for specimens tested under a variety of conditions2,20 with h for the 1100 ◦C/air specimen being an order of magnitude lower compared to that of the RT case (810
m3). With reference to Eq. (4), it would thus be expected that for the 1100 ◦C/air specimen, even in the fibre bundle centre where oxidation damage was minimal, would still be significantly larger compared to the values of 5–10 MPa previously measured in non-oxidised specimens.2 3.2. Fibre strength parameters Fibre strength distributions (normalised to a gauge length of 10−3 m) at the centre and edge2 of individual fibre bun￾dles in RT and 1100 ◦C/air specimens have been presented in Fig. 3 with data for the respective values of So and m being given in Table 1. Whereas the centre and edge fibre strength distributions were similar for the RT case (Fig. 4(a)), 2 For the 1100 ◦C/air specimen, fibre strengths close to the perimeter could not be measured due to the lack of fracture mirrors (Fig. 2(b)). Data for edge fibres was thus calculated using fibres adjacent to the embrittled region.