October 2007 Tensile Mechanical Properties of Ceramic Matrix Composites 3191 Table Ill. Ultimate Strength Properties of 10/901 Composite It is difficult to compare all of the off-axis UTs data on an in the Literature Tested in Orthogonal and 45/45 Orientations absolute basis since composites vary considerably in fiber vol- ume fractions, in situ strength properties of the fibers, interfacial 0)UTS(0)E45)UTs(45) architecture f=2/o(GPa)(MPa)(GPa)(GPa) on fiber modulus and diameter. However, when comparing the Matrix-dominated composites UTS of [0/90] composites oriented in the [+45]-direction, the Sylramic-MI- 5 HS 0.34268344258200 matrix-dominated SiC composites, whether reinforced by 5 HS 0.34210351169158 Nicalon, Hi-Nicalon, Sylra or Sylramic-ibN SiC fibers NicCⅥISiC2 Weave~0.4265255NA210 (see Tables I and II), exhibited the best off-axis UTS values Nic-CVI SIC 0.36240248183192 (158-242 MPa) compared with the fiber-dominated composites NIC-CAS 0/90 laminate 0.4 136 215 NA (all< 100 MPa) regardless of the fiber type. To estimate the general effect of composite-type and off-axis testing direction on effective fiber strength or load-carrying abil Fiber-dominated composites ity, one can eliminate the fiber tensile strength and volume 0.460320NA80 fraction effects from the strength data by normalizing the 8 HS 0.36 70 252 34 76 off-axis UTS data of Tables I, Ill, and IV using the following C-CVI SiC2 8 HS 0.36873246274 relationship 8 HS 0.361013052170 ALO3 /Mullite-8HS ~04~972105052 Normalized effective fiber strength MI, melt-infiltrated: UTS. ultimate tensile strength. Sr(off-axis)/Sr(o)=(cult off-axis/out o)(o/Jfofr-a 3) where Sr(off-axis) is the average strength of a fiber at failure that are heavily influenced by the elastic properties of the matrix. is oriented at an angle to the loading direction and Sno is the Nonlinearity in the stress-strain curve is caused by transverse average strength of a fiber at failure when oriented in the direc- matrix cracks(cracks perpendicular to the applied load) due te tion of loading. This can be determined by dividing the UTS of terface debonding and fiber sliding resulting in fiber pullout site by the fraction of fibers. ult o and fo refers, re- within the matrix crack and matrix crack opening. Transverse spectively, to the UTS and fraction of fibers when the fibers are matrix cracking occurs for both on-axis and off-axis loading of aligned with the loading direction, i.e., for a [o/90] composite fo half the total volume fraction of fibers g and off-ax For fiber-dominated composites(e.g, porous oxide/oxide refers, respectively, to the UTS and effective load-bearing fra C/C, or C/SiC composites), the matrix carries little load due tion of fibers when the fibers are aligned at an angle to the to the fact that the matrix is porous or is heavily microcracked in loading direction. It can be assumed for the [0/90) composites the as-processed condition, and the mechanical properties are hen loaded symmetric to the primary fiber directions, e.g controlled by the fiber(architecture)response to loading. When [+45] that both fiber axes are carrying load, so that fofr-axis=2 loaded on-axis, elastic moduli are typically only slightly larger fo. For nonsymmetric loading, it can be assumed that the fibers han f(0)x Ef, and there is only minor nonlinearity to the stress in the axis at the greatest angle to the loading direction will strain curve since the matrix is highly porous and/ or a high fracture first, debond, and or pull apart, leaving the other pri damage condition already exists in the matrix. When loaded in ary axis fibers to control UTS, so that fff-axis equals fo of the the off-axis. there is significant nonlinear emaining fibers. For braided composites, specimens tested in behavior beginning near zero stress due to the absence of he hoop direction were compared with specimens allel fibers in the loading direction. More damage is created in same panel tested in the axial direction(C/epoxy- the composite due to shear band formation, fiber-rearran or to [0/90 specimens tested in the 0 direction(Syl-i ment, inter-ply delamination, and fiber"scissoring. This results Table in high ultimate strains as long as the fibers can withstand the Based on these assumptions and Eq ( 3), Fig. ll shows the tensile, bending, and shear forces, but low ultimate strengths due normalized fiber strength results as a function of test angle usi to these combined forces on the fibers. This low off-axis Uts the UtS data from Tables l, Ill, and IV. Again the four matrix- behavior for these fiber-dominated composites may be a limita- dominated Sic-based matrix composites where the elastic mod- tion for certain applications. ulus of the matrix was relatively high exhibit the best relative The Sylramic-iBN composites are an excellent example of retention of fiber strength. For the fiber-dominated or low-mod- matrix-dominated composites and of how off-axis properties ulus matrix composites, C/epoxy, Al2O3/ mullite, Nic/C, C/C can be maximized which may be tant for future applica and C/SiC, there was excellent correlation at an angle of 45, but tions. To demonstrate this, the uts data from the composite poorer off-axis strengths than the matrix-dominated compo tested in this study (see table i) can be con with UTS data ites. The two matrix-dominated glass-CMC were in between the in the literature for a variety of fiber-dominated and matrix- fiber-dominated and Sic-based matrix-dominated composites dominated 2D-woven, braided, and cross-plied ceramic and probably due to the lower modulus of the glass-ceramic matrix polymer composites that were tested in a primary fiber direc- compared with CVI and MI SiC matrices. It is significant that tion as well as in an off-axis direction. Descriptions of these the braided and [0/90] C/epoxy data from different studies over composites and their ultimate properties are listed in Table Ill the range of angles tested correlate with one another, justifying for 0/90 composites tested in the 0 and 45/45 direction and in some of the crude assumptions made above Tables IV for 0/90 and braided composites tested at multiple For fiber orientations 25-45 off the loading axis. the relative ngles. These composites included 2D plain-woven carbon fiber stress-carrying ability of the Syl-iBN fibers in high-modulus SiC- reinforced epoxy, 2D triaxially braided carbon fiber-reinforced based matrices ranged from 50% to 30% of the fiber strength epoxy, 2D eight-harness satin-woven Al2Ox-fiber-reinforced respectively, when oriented in the primary 0 direction. Though porous oxide matrix (mullite) composites, CG Nicalon the relative stress-carrying ability decreased with increasing an Nippon Carbon, Tokyo, Japan, referred to as"Nic")fiber- gle, it was still far superior to the fiber-dominated systems(50% einforced carbon matrix. - CG Nicalon fiber-reinfe greater relative stress at 25 and 100% greater relative stress at matrix composites, CG Nicalon-fiber reinforced matrix c posites, an carbon fiber-reinforced carbo Note that in Eq. (3). the ellipsoid character of off-axis oriented fibers in the pla composites (Nippon Carbon, Japan; referred to as"HN)an fiber-reinforced MI composites.are heavily influenced by the elastic properties of the matrix. Nonlinearity in the stress–strain curve is caused by transverse matrix cracks (cracks perpendicular to the applied load) due to interface debonding and fiber sliding resulting in fiber pullout within the matrix crack and matrix crack opening. Transverse matrix cracking occurs for both on-axis and off-axis loading of composites. For fiber-dominated composites (e.g., porous oxide/oxide, C/C, or C/SiC composites), the matrix carries little load due to the fact that the matrix is porous or is heavily microcracked in the as-processed condition, and the mechanical properties are controlled by the fiber (architecture) response to loading. When loaded on-axis, elastic moduli are typically only slightly larger than f(01) Ef, and there is only minor nonlinearity to the stress strain curve since the matrix is highly porous and/or a high damage condition already exists in the matrix. When loaded in the off-axis, there is significant nonlinearity in the stress–strain behavior beginning near zero stress due to the absence of parallel fibers in the loading direction. More damage is created in the composite due to shear band formation, fiber-rearrangement, inter-ply delamination, and fiber ‘‘scissoring.’’ This results in high ultimate strains as long as the fibers can withstand the tensile, bending, and shear forces, but low ultimate strengths due to these combined forces on the fibers. This low off-axis UTS behavior for these fiber-dominated composites may be a limitation for certain applications. The Sylramic-iBN composites are an excellent example of matrix-dominated composites and of how off-axis properties can be maximized, which may be important for future applications. To demonstrate this, the UTS data from the composites tested in this study (see Table I) can be compared with UTS data in the literature for a variety of fiber-dominated and matrixdominated 2D-woven, braided, and cross-plied ceramic and polymer composites that were tested in a primary fiber direction as well as in an off-axis direction. Descriptions of these composites and their ultimate properties are listed in Table III for 0/90 composites tested in the 0 and 45/45 direction and in Tables IV for 0/90 and braided composites tested at multiple angles. These composites included 2D plain-woven carbon fiberreinforced epoxy,18 2D triaxially braided carbon fiber-reinforced epoxy,19 2D eight-harness satin-woven Al2O3-fiber-reinforced porous oxide matrix (mullite) composites,20 CG Nicalon (Nippon Carbon, Tokyo, Japan, referred to as ‘‘Nic’’) fiberreinforced carbon matrix,2,21 CG Nicalon fiber-reinforced glass matrix composites,2,22 CG Nicalon-fiber reinforced CVI SiC matrix composites,2,21,23 carbon fiber-reinforced carbon matrix composites,17,21 and 2D five-harness satin-woven Hi-Nicalon (Nippon Carbon, Japan; referred to as ‘‘HN’’) and Sylramic fiber-reinforced MI composites.24 It is difficult to compare all of the off-axis UTS data on an absolute basis since composites vary considerably in fiber volume fractions, in situ strength properties of the fibers, interfacial sliding stress, and fiber-bending stiffness, which are dependent on fiber modulus and diameter. However, when comparing the UTS of [0/90] composites oriented in the [745]-direction, the matrix-dominated SiC composites, whether reinforced by Nicalon, Hi-Nicalon, Sylramic, or Sylramic-iBN SiC fibers (see Tables I and II), exhibited the best off-axis UTS values (158–242 MPa) compared with the fiber-dominated composites (allo100 MPa) regardless of the fiber type. To estimate the general effect of composite-type and off-axis testing direction on effective fiber strength or load-carrying ability, one can eliminate the fiber tensile strength and volume fraction effects from the strength data by normalizing the off-axis UTS data of Tables I, III, and IV using the following relationshipJ : Normalized effective fiber strength ¼ Sfðoff-axisÞ=Sfð0Þ ¼ ðsult off-axis=sult 0Þðf0=f off-axisÞ (3) where Sf(offaxis) is the average strength of a fiber at failure that is oriented at an angle to the loading direction and Sf(0) is the average strength of a fiber at failure when oriented in the direction of loading. This can be determined by dividing the UTS of the composite by the fraction of fibers. sult 0 and f0 refers, respectively, to the UTS and fraction of fibers when the fibers are aligned with the loading direction, i.e., for a [0/90] composite f0 is half the total volume fraction of fibers. sult offaxis and f offaxis refers, respectively, to the UTS and effective load-bearing fraction of fibers when the fibers are aligned at an angle to the loading direction. It can be assumed for the [0/90] composites when loaded symmetric to the primary fiber directions, e.g., [745], that both fiber axes are carrying load, so that foffaxis 5 2 f0. For nonsymmetric loading, it can be assumed that the fibers in the axis at the greatest angle to the loading direction will fracture first, debond, and/or pull apart, leaving the other primary axis fibers to control UTS, so that foffaxis equals f0 of the remaining fibers. For braided composites, specimens tested in the hoop direction were compared with specimens from the same panel tested in the axial direction (C/epoxy—Table IV) or to [0/90] specimens tested in the 0 direction (Syl-iBN MI— Table I). Based on these assumptions and Eq. (3), Fig. 11 shows the normalized fiber strength results as a function of test angle using the UTS data from Tables I, III, and IV. Again the four matrixdominated SiC-based matrix composites where the elastic modulus of the matrix was relatively high exhibit the best relative retention of fiber strength. For the fiber-dominated or low-modulus matrix composites, C/epoxy, Al2O3/mullite, Nic/C, C/C, and C/SiC, there was excellent correlation at an angle of 451, but poorer off-axis strengths than the matrix-dominated composites. The two matrix-dominated glass-CMC were in between the fiber-dominated and SiC-based matrix-dominated composites probably due to the lower modulus of the glass–ceramic matrix compared with CVI and MI SiC matrices. It is significant that the braided and [0/90] C/epoxy data from different studies over the range of angles tested correlate with one another, justifying some of the crude assumptions made above. For fiber orientations 251–451 off the loading axis, the relative stress-carrying ability of the Syl-iBN fibers in high-modulus SiCbased matrices ranged from 50% to 30% of the fiber strength, respectively, when oriented in the primary 01 direction. Though the relative stress-carrying ability decreased with increasing angle, it was still far superior to the fiber-dominated systems (50% greater relative stress at 251 and 100% greater relative stress at Table III. Ultimate Strength Properties of [0/90] Composites in the Literature Tested in Orthogonal and 45/45 Orientations Composite Fiber architecture f 5 2f0 E(0) (GPa) UTS(0) (MPa) E(45) (GPa) UTS(45) (GPa) Matrix-dominated composites Sylramic-MI24 5 HS 0.34 268 344 258 200 Hi-Nic-MI24 5 HS 0.34 210 351 169 158 Nic-CVI SiC2 Plain Weave B0.4 265 255 NA 210 Nic-CVI SiC21 8 HS 0.36 240 248 183 192 NiC-CAS2 0/90 laminate B0.4 136 215 NA 95 Fiber-dominated composites Nic-C2 — B0.4 60 320 NA 80 Nic-C21 8 HS 0.36 70 252 34 76 C-CVI SiC21 8 HS 0.36 87 324 62 74 C-C21 8 HS 0.36 101 305 21 70 Al2O3/Mullite20 8 HS B0.4 B97 210 50 52 MI, melt-infiltrated; UTS, ultimate tensile strength. J Note that in Eq. (3), the ellipsoid character of off-axis oriented fibers in the plane of failure was not taken into account. A more exhaustive and complicated analysis would require not only the angle, but the nature of fiber/matrix sliding which is not well understood,2 the matrix crack opening and the degree of fiber straightening. From an engineering and design standpoint, the simple analysis performed here is considered to be more straightforward and useful. October 2007 Tensile Mechanical Properties of Ceramic Matrix Composites 3191