1. Davies et al. /Journal of the European Ceramic Sociery 25(2005)599-604 may, for example, shed light on the value of t required for the suppression of crack deflection mechanisms in SiC/SIC composites. Thus, the present work will concern itself with a the investigation of mechanical and physical properties within a single fibre bundle for an orthogonal 3-D woven SiC/SiC composite tensile tested at 1 100oC in air. 2. Experimental procedure The composite under investigation, based on the SiC/SiC system, contained Tyranno LoxM Si-THC-O fibres(800 fibres/bundle)surface-modified so as to achieve a 40 nm arbon-rich layer adjacent to the fibre surface. 29 The aim of 12UA21 5 um surface modification was to promote crack deflection at the carbon-rich layer within the fibre itself, rather than at he fibre surface as is the case for most cmcs. The fibre (b) were woven into an orthogonal 3-d orthogonal configura- tion with fibre volume fractions of.19019 and 0.02 in the x, y, and directions, respectively. Matrix densification of the composite was achieved through the repeated poly mer impregnation and pyrolysis(PIP)of a precursor simi- lar to polytitanocarbosilane(PTCs). The use of similar pre cursors for fibre and matrix components was expected to minimise thermal stresses due to any mismatch in the co- efficients of thermal expansion. This composite system eferred to as"NUSK-CMC'' from the initials of the collab- 3 5 um orating partners and has been the subject of recent work by 321 the authors 29.30 Following machining to a suitable test geometry, 29the Fig. 1. Scanning electron micrographs illustrating the observed fracture composite was heated in air to 1 100C at0.75Cs-, loaded urface types in Tyranno" Si-TiC-O fibres:(a) fracture mirror, and(b) fat and featurele under tension(parallel to the y-axis)to failure, and furnace cooled (initial rate of 3.3Cs); the total time spent at 1 100C being on the order of 600s. Further experimental h, and (iii) whether the fibre exhibited a fracture mirror details are available elsewhere 8,29,31 (Fig. 1(a)) or was flat and featureless(Fig. 1 (b)). A total of In contrast to the complex non-linear stress/strain be- 698 fibres(out of the 800 nominally expected)were mea- aviour noted for similar specimens tested at room temper sured with the difference in number being explained by:(1) ature(RT) and 1200C in vacuum. 8, 29,31 the 1100 C/air shielding of fibres by neighbours, and(ii) the presence of specimens stress/strain curve was approximately linear to holes where fibres had pulled out failure with a low tensile strength(55 MPa)compared to In addition to the above parameters, due to the likely re- the rt(381+41 MPa) and 1200oC/vacuum(405+39 MPa) action of oxygen with the fibre surfaces, So and m were in- cases. In fact, the 1100.C/air specimen tensile strength was vestigated at the centre and edge of the fibre bundle by mea- comparable with that of the stress required for propagation Suring the fracture mirror radius, /'m, of individual fibres and suggesting that the specimen may have failed upon initial fibre strength, 5-2 relationship to determine the individual of matrix cracks within transverse fibre bundles (65 MPa) using the followin matrix microcracking(or shortly thereafter) ollowing failure, the specimen fracture surface was nvestigated using a scanning electron microscope(SEM) (Model JSM-6300F, JEOL, Tokyo, Japan)and the following where Am is known as the "mirror constant and was pre viously determined to be 2.50+0.09MPam /2 for the randomly chosen fibre bundle near the specimen centre: (i SH-Ti-C-O fibres in situ the composite 5.36 and close to position within the fibre bundle, (ii) fibre pullout length the value of 2.51 MPam /2 proposed for nominally similar Nicalon Si-C-O fibres values of S. and m were de duced from cumulative failure probability curves of S after National Aerospace Laboratory of Japan, Ube Industries, Ltd, Shik. applying a suitable correction factor. Fibres wi ibo, Ltd, and Kawasaki Heavy Industries, Ltd are known to be effectively tested at a gauge length, Sc600 I.J. Davies et al. / Journal of the European Ceramic Society 25 (2005) 599–604 may, for example, shed light on the value of τ required for the suppression of crack deflection mechanisms in SiC/SiC composites. Thus, the present work will concern itself with the investigation of mechanical and physical properties within a single fibre bundle for an orthogonal 3-D woven SiC/SiC composite tensile tested at 1100 ◦C in air. 2. Experimental procedure The composite under investigation, based on the SiC/SiC system, contained Tyranno® LoxM Si–Ti–C–O fibres (800 fibres/bundle) surface-modified so as to achieve a 40 nm carbon-rich layer adjacent to the fibre surface.29 The aim of the surface modification was to promote crack deflection at the carbon-rich layer within the fibre itself, rather than at the fibre surface as is the case for most CMCs. The fibres were woven into an orthogonal 3-D orthogonal configuration with fibre volume fractions of 0.19, 0.19, and 0.02 in the x, y, and z directions, respectively. Matrix densification of the composite was achieved through the repeated polymer impregnation and pyrolysis (PIP) of a precursor similar to polytitanocarbosilane (PTCS). The use of similar precursors for fibre and matrix components was expected to minimise thermal stresses due to any mismatch in the coefficients of thermal expansion. This composite system is referred to as “NUSK-CMC” from the initials of the collaborating partners† and has been the subject of recent work by the authors.29,30 Following machining to a suitable test geometry,29 the composite was heated in air to 1100 ◦C at 0.75 ◦C s−1, loaded under tension (parallel to the y-axis) to failure, and furnace cooled (initial rate of 3.3 ◦C s−1); the total time spent at 1100 ◦C being on the order of 600 s. Further experimental details are available elsewhere.8,29,31 In contrast to the complex non-linear stress/strain behaviour noted for similar specimens tested at room temperature (RT) and 1200 ◦C in vacuum,8,29,31 the 1100 ◦C/air specimen’s stress/strain curve was approximately linear to failure with a low tensile strength (∼55 MPa) compared to the RT (381±41 MPa) and 1200 ◦C/vacuum (405±39 MPa) cases. In fact, the 1100 ◦C/air specimen tensile strength was comparable with that of the stress required for propagation of matrix cracks within transverse fibre bundles (∼65 MPa),8 suggesting that the specimen may have failed upon initial matrix microcracking (or shortly thereafter). Following failure, the specimen fracture surface was investigated using a scanning electron microscope (SEM) (Model JSM-6300F, JEOL, Tokyo, Japan) and the following parameters measured for each fibre visible within a single randomly chosen fibre bundle near the specimen centre: (i) position within the fibre bundle, (ii) fibre pullout length, † National Aerospace Laboratory of Japan, Ube Industries, Ltd., Shikibo, Ltd., and Kawasaki Heavy Industries, Ltd. Fig. 1. Scanning electron micrographs illustrating the observed fracture surface types in Tyranno® Si–Ti–C–O fibres: (a) fracture mirror, and (b) flat and featureless. h, and (iii) whether the fibre exhibited a fracture mirror (Fig. 1(a)) or was flat and featureless (Fig. 1(b)). A total of 698 fibres (out of the 800 nominally expected) were measured with the difference in number being explained by: (i) shielding of fibres by neighbours, and (ii) the presence of holes where fibres had pulled out. In addition to the above parameters, due to the likely reaction of oxygen with the fibre surfaces, So and m were investigated at the centre and edge of the fibre bundle by measuring the fracture mirror radius, rm, of individual fibres and using the following relationship to determine the individual fibre strength, S: 32–34 S = Am √rm (1) where Am is known as the “mirror constant” and was previously determined to be 2.50 ± 0.09 MPa m1/2 for the Si–Ti–C–O fibres in situ the composite35,36 and close to the value of 2.51 MPa m1/2 proposed for nominally similar Nicalon® Si–C–O fibres37 Values of So and m were deduced from cumulative failure probability curves of S after applying a suitable correction factor.38 Fibres within CMCs are known to be effectively tested at a gauge length, δc