《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_c-g-4

ournal of the European Ceramic Society 19(1999)317-327 C)1999 Elsevier Science Limited Printed in Great Britain. All rights reserved PII:S0955-2219(98)00081-8 0955-2219/99/S--see front matter Carbon Fiber-reinforced (YMAS Glass-Ceramic Matrix Composites III. Interfacial Aspects Valerie bianchi. a paul goursat a* Wharton Sinkler b Marc Monthioux b and erik menessierd "LMCTS, URA CNRS 320, Faculte des Sciences, 123, Avenue Albert Thomas, 87060 Limoges Cedex, France bCEMES UPR A-8011 CNRS B.P. 4347 31055 Toulouse Cedex france cCeramiques et Composites, B.P. 7, 65460 Bazet, france Received 12 May 1997; accepted 25 February 1998) Abstract 1 Introduction Unidirectional continuous carbon fiber-reinforced The use of ceramics at high temperature and in glass-ceramic matrix composites have been fabricated corrosive atmospheres for thermostructural for dry sliding applications. Different fracture beha- applications is always limited by their low tough viors have been observed(three-point bend test and ness. Fiber-reinforced ceramic matrix composite fracture surfaces observation). Besides, the distance have been particularly promoted mainly for the between next microcracks, which have appeared in the aerospace industries. Indeed, damage in ceramic matrix on cooling after hot-pressing of the compo- matrix composites(CMC) is characterized at the sites, changes with the sintering temperature, sug- beginning by a progressive microcracking in the gesting fiber-matrix reactions. The nature of the matrix. According to discontinuities in the stresses fiber-matrix interface is observed in Transmission at the matrix-cracks and the fiber-matrix inter- Electronic Microscopy and the interfacial shear stress faces, both the fibers and matrix must be able to is determined by push-in tests. C 1998 Elsevier Sci- move in order to fit the local stress field. It implies ence Limited. All rights reserved that the interfacial bond must be just weak enough to allow the debonding of the interface and the Resume sliding of the fibers Glass and glass-ceramic matrix composites ha Des composites unidirectionnels a fibres longues de been investigated by numerous works because they carbone et matrice vitroceramique ont ete elabores pour present the advantage, among others, of being des applications en frottement sec. Differents compor- fabricated at low temperature. 2- For wear appli tements d la rupture ont ete observes (essai en flex- cations, carbon fiber-reinforced matrix competent ion trois points et observation des fractures). Par can be expected to present a low friction coefficient ailleurs, la distance separant deux fissures voisines, due to the possible lubricant properties of carbon apparues dans la matrice lors du refroidissement apres fibers in some conditions. Carbon fiber-reinforced le frittage sous charge, change avec la temperature de YMaS glass-ceramic matrix composites have been frittage, suggerant une reaction entre la fibre et la fabricated for dry sliding applications. In a first matrice. La nature de l'interface fibre-Imatrice est step, it was necessary to control the micro- observee en Microscopie Electronique en Transmis- structural changes with temperature in the compo- ion et la contrainte de cisaillement interfacial est sites, the mechanical strength and crack determine lors d'essais de micro-indentation des propagation resistance, before beginning the tribo- fibre logical study. Indeed the behavior of materials in dry friction depends on many parameters like the Q Keywords: mechanical properties, glass ceramics, test conditions and the geometry of the contacts. It mposites, carbon fibres, interfaces depends also on the mechanical strength of the materials, on mechanical and physico-chemical To whom correspondence should be addressed. reactions at the interface between both the parts 17

Carbon Fiber-reinforced (YMAS) Glass-Ceramic Matrix Composites. III. Interfacial Aspects ValeÂrie Bianchi,a Paul Goursat,a * Wharton Sinkler,b Marc Monthiouxb and Erik MeÂnessierc a LMCTS, URA CNRS 320, Faculte des Sciences, 123, Avenue Albert Thomas, 87060 Limoges Cedex, France b CEMES, UPR A-8011 CNRS, B.P. 4347, 31055 Toulouse Cedex, France c CeÂramiques et Composites, B.P. 7, 65460 Bazet, France (Received 12 May 1997; accepted 25 February 1998) Abstract Unidirectional continuous carbon ®ber-reinforced glass-ceramic matrix composites have been fabricated for dry sliding applications. Di€erent fracture beha￾viors have been observed (three-point bend test and fracture surfaces observation). Besides, the distance between next microcracks, which have appeared in the matrix on cooling after hot-pressing of the compo￾sites, changes with the sintering temperature, sug￾gesting ®ber±matrix reactions. The nature of the ®ber±matrix interface is observed in Transmission Electronic Microscopy and the interfacial shear stress is determined by push-in tests. # 1998 Elsevier Sci￾ence Limited. All rights reserved ReÂsume Des composites unidirectionnels aÁ ®bres longues de carbone et matrice vitroceÂramique ont eÂte eÂlaboreÂs pour des applications en frottement sec. Di€eÂrents compor￾tements aÁ la rupture ont eÂte observeÂs (essai en ¯ex￾ion trois points et observation des fractures). Par ailleurs, la distance seÂparant deux ®ssures voisines, apparues dans la matrice lors du refroidissement apreÂs le frittage sous charge, change avec la tempeÂrature de frittage, suggeÂrant une reÂaction entre la ®bre et la matrice. La nature de l'interface ®bre-matrice est observee en Microscopie Electronique en Transmis￾sion et la contrainte de cisaillement interfaciale est deÂterminee lors d'essais de micro-indentation des ®bres. Keywords: mechanical properties, glass ceramics, composites, carbon ®bres, interfaces. 1 Introduction The use of ceramics at high temperature and in corrosive atmospheres for thermostructural applications is always limited by their low tough￾ness. Fiber-reinforced ceramic matrix composites have been particularly promoted mainly for the aerospace industries. Indeed, damage in ceramic matrix composites (CMC) is characterized at the beginning by a progressive microcracking in the matrix.1 According to discontinuities in the stresses at the matrix-cracks and the ®ber±matrix inter￾faces, both the ®bers and matrix must be able to move in order to ®t the local stress ®eld. It implies that the interfacial bond must be just weak enough to allow the debonding of the interface and the sliding of the ®bers. Glass and glass-ceramic matrix composites have been investigated by numerous works because they present the advantage, among others, of being fabricated at low temperature.2±4 For wear appli￾cations, carbon ®ber-reinforced matrix composites can be expected to present a low friction coecient due to the possible lubricant properties of carbon ®bers in some conditions.5 Carbon ®ber-reinforced YMAS glass-ceramic matrix composites have been fabricated for dry sliding applications.6 In a ®rst step, it was necessary to control the micro￾structural changes with temperature in the compo￾sites, the mechanical strength and crack propagation resistance, before beginning the tribo￾logical study. Indeed the behavior of materials in dry friction depends on many parameters like the test conditions and the geometry of the contacts. It depends also on the mechanical strength of the materials, on mechanical and physico-chemical reactions at the interface between both the parts in Journal of the European Ceramic Society 19 (1999) 317±327 # 1999 Elsevier Science Limited Printed in Great Britain. All rights reserved PII: S0955-2219(98)00081-8 0955-2219/99/$Ðsee front matter 317 *To whom correspondence should be addressed

318 V. Bianchi et al contact on the formation and the evolution of a mine the interfacial shear strength and relate it to third body in the contact. This work 6, 8 has led us the different mechanical behaviors observed.The o study in C/YMAS composites different beha- fiber-matrix interface of some composites is inves- viors according to the used carbon fiber and the tigated by Transmission Electronic Microscopy hermal treatment applied for their processing. It (TEM) before mechanical testin has been shown that, as is well known nowadays for CMC, these behaviors are closely related to interface properties. The required debonding 2 Experimental Procedure energy and the average reloading stress--reloading of the matrix on both sides of the crack results 2.1 Composites from the stress transfer from fibers to the matrix, The YMAS(Y2O3, Mgo, Al2O3, Sio2) matrix is hich is assumed to occur by shearing along the reinforced by unidirectional fibers. Two grades interface-are key parameters for CMc damage which differ from their precursor, pitch or PAN resistance. According to the magnitude of the and their properties, are used: the pitch-based shear stress t, the interface will be named strong carbon fiber P25(AMOCO) and the PAN-based or weak. The stress t depends at the same time on carbon fiber T400H TORAYCA). Their main he friction coefficient thermal stresses due to the characteristics are summarized in table 1. their hermal expansion mismatch between the fibers microstructure and texture have been widely and the matrix, and on stresses induced by the described previously. Cross-sections of Pitch Poissons effects during mechanical tests. So, if the based carbon fibers exhibit a 'Pan-am'texture coefficient of thermal expansion(CTE)of the fiber with a conjunction of isotropic and anisotropic (ar) is lower than the Cte of the matrix(am), areas and a highly anisotropic but incomplete the fiber is submitted during the fabrication of polyaromatic carbon skin. The cross-sections of the composites to a radial compression on cooling, the T400H fiber are characterized, like the PAN and the fiber-matrix sliding is difficult. By contrast, based T300 grade from TORAYCA, by a coarsely if am-af <0, the interface is submitted to tension isotropic texture with a local anisotropic radial which facilitates the extraction of the fibers texture in a ring at about one half or one third of For SiC/LAS composites, 0-14 it appears that the the radius. Due to the dissimilar texture of T400H presence of a carbonaceous interphase contributes and P25 fibers, their coefficients of thermal expan to composite strengthening by making easier slid- sion are different. 6 ing between the fiber and the matrix. The reactivity The glass transition and crystallization tempera between a carbon fiber and silica, which is the pre- ture intervals of the YMAS matrix are, respec dominant phase in the various glass matrix used tively, 815-855C and 945-1055C. for carbon fiber composites, has been studied by Pre-preg sheets are prepared by using a slurry Benson et al. At 1400C, the contact between the infiltration process of fiber tows. The impregnated carbon fiber and the matrix can lead to the forma- fibers are wound on an hexa gonal mandrel. The tion of CO, CO2 and Sio gases and solid reaction volumic fraction of fibers is nearly equal to 35% products, oxide or oxycarbide. Very adhesive After drying, cutting and debinding, the tapes are interfaces with high shear stresses can be expected densified in argon by hot-pressing in the 950 to if the external mechanical pressure is sufficient to 1250C temperature interval with a pressure of maintain the contact. Thermodynamic calculations 10 MPa(LPA-DVM Goliath), 6 show that the reaction between the carbon fiber Up to 1000C, X-ray diffraction patterns 7 and the matrix depends strongly on the presence of show that the matrix is mainly vitreous. The CO and CO2 gases. Therefore, Tredway et al. 4 MgAl2O4 spinel precipitates above 1050 C, as well suggest the introduction of oxides in the matrix, as the hexagonal high temperature cordierite and like Nb2Os or MoO3, which can be reduced in the a and B Y2Si2O7 yttrium silicates. The SiO2- carbides for very high pressures, in order to limit Y203 system is very complex with many poly the reaction between silica and carbon morphic phases in the case of Y2Si2O7(a, Bby, 8) The aim of this paper is to understand the role played by the fiber-matrix interface on the fracture Table 1. Characteristics of the fibers(given by the manu behavior of C/YMAS composites. Two grades of facturers fibers are used. so that the influence of their differ- fib P25T400H ent properties, textures and chemical reactivity will Longitudinal tensile strength(MPal 14004500 have to be considered. This work had to precede Longitudinal tensile modulus(GPa) the study of the tribological behavior composites, Tensile fracture elongation ( % in order that it could take into account this influ Density (g cm-3) 1-901-80 ential parameter. Push-in tests are used to deter- Diameter of the filaments (um)

contact, on the formation and the evolution of a third body in the contact.7 This work6,8 has led us to study in C/YMAS composites di€erent beha￾viors according to the used carbon ®ber and the thermal treatment applied for their processing. It has been shown that, as is well known nowadays for CMC,9 these behaviors are closely related to interface properties. The required debonding energy and the average reloading stressÐreloading of the matrix on both sides of the crack results from the stress transfer from ®bers to the matrix, which is assumed to occur by shearing along the interfaceÐare key parameters for CMC damage resistance. According to the magnitude of the shear stress , the interface will be named strong or weak. The stress  depends at the same time on the friction coecient, thermal stresses due to the thermal expansion mismatch between the ®bers and the matrix, and on stresses induced by the Poisson's e€ects during mechanical tests. So, if the coecient of thermal expansion (CTE) of the ®ber (f) is lower than the CTE of the matrix (m), the ®ber is submitted during the fabrication of the composites to a radial compression on cooling, and the ®ber±matrix sliding is dicult. By contrast, if m ÿ f < 0, the interface is submitted to tension which facilitates the extraction of the ®bers. For SiC/LAS composites,10±14 it appears that the presence of a carbonaceous interphase contributes to composite strengthening by making easier slid￾ing between the ®ber and the matrix. The reactivity between a carbon ®ber and silica, which is the pre￾dominant phase in the various glass matrix used for carbon ®ber composites, has been studied by Benson et al.15 At 1400C, the contact between the carbon ®ber and the matrix can lead to the forma￾tion of CO, CO2 and SiO gases and solid reaction products, oxide or oxycarbide. Very adhesive interfaces with high shear stresses can be expected if the external mechanical pressure is sucient to maintain the contact. Thermodynamic calculations show that the reaction between the carbon ®ber and the matrix depends strongly on the presence of CO and CO2 gases. Therefore, Tredway et al.14 suggest the introduction of oxides in the matrix, like Nb2O5 or MoO3, which can be reduced in carbides for very high pressures, in order to limit the reaction between silica and carbon. The aim of this paper is to understand the role played by the ®ber±matrix interface on the fracture behavior of C/YMAS composites. Two grades of ®bers are used, so that the in¯uence of their di€er￾ent properties, textures and chemical reactivity will have to be considered. This work had to precede the study of the tribological behavior composites, in order that it could take into account this in¯u￾ential parameter. Push-in tests are used to deter￾mine the interfacial shear strength and relate it to the di€erent mechanical behaviors observed. The ®ber±matrix interface of some composites is inves￾tigated by Transmission Electronic Microscopy (TEM) before mechanical testing. 2 Experimental Procedure 2.1 Composites The YMAS (Y2O3, MgO, Al2O3, SiO2) matrix is reinforced by unidirectional ®bers. Two grades which di€er from their precursor, pitch or PAN, and their properties, are used : the pitch-based carbon ®ber P25 (AMOCO) and the PAN-based carbon ®ber T400H (TORAYCA). Their main characteristics are summarized in Table 1. Their microstructure and texture have been widely described previously.8 Cross-sections of Pitch￾based carbon ®bers exhibit a `Pan-am' texture, with a conjunction of isotropic and anisotropic areas and a highly anisotropic but incomplete polyaromatic carbon skin. The cross-sections of the T400H ®ber are characterized, like the PAN￾based T300 grade from TORAYCA, by a coarsely isotropic texture with a local anisotropic radial texture in a ring at about one half or one third of the radius. Due to the dissimilar texture of T400H and P25 ®bers, their coecients of thermal expan￾sion are di€erent.6 The glass transition and crystallization tempera￾ture intervals of the YMAS matrix are, respec￾tively, 815±855C and 945±1055C.8,17 Pre-preg sheets are prepared by using a slurry in®ltration process of ®ber tows. The impregnated ®bers are wound on an hexagonal mandrel. The volumic fraction of ®bers is nearly equal to 35%. After drying, cutting and debinding, the tapes are densi®ed in argon by hot-pressing in the 950 to 1250C temperature interval with a pressure of 10MPa (LPA-DVM Goliath).8,16 Up to 1000C, X-ray di€raction patterns8,17 show that the matrix is mainly vitreous. The MgAl2O4 spinel precipitates above 1050C, as well as the hexagonal high temperature cordierite and the  and  Y2Si2O7 yttrium silicates. The SiO2- Y2O3 system is very complex with many poly￾morphic phases in the case of Y2Si2O7 (, b , ). Table 1. Characteristics of the ®bers (given by the manu￾facturers) Fiber P25 T400H Longitudinal tensile strength (MPa) 1400 4500 Longitudinal tensile modulus (GPa) 160 250 Tensile fracture elongation (%) 0.9 1.8 Density (g cmÿ3 ) 1.90 1.80 Diameter of the ®laments (m) 11 7 318 V. Bianchi et al

Carbon fiber-reinforced glass-ceramic matrix composites. III 319 At 1250C, the a-Y2Si2O7 turns almost completely 3 Results in the B form. 3.1 Fracture behavior 2.2 Mechanical tests Pitch-based carbon fiber-reinforced YMAS Three-point bend tests are performed on compo- composites have been hot-pressed. 8, 7 Due to the sites bars(34x6x1.5mm )at room temperature, thermal expansion mismatch between the carbon with a span 30 mm long and a loading rate of fibers(axial CTE N 4.5.10-bC- radial CTE A 0.2 mm min.8, I6 A displacement cell, set on the 15.10-6C-) and the matrix (CTE N tension surface of the sample, in the middle, has 6.10-0C-), thermal stresses are induced in the allowed to obtain load-deflection curves composites on cooling making a network of microcracks. Whatever the sintering temperature, 2.3 Push-in test the average distance between two microcracks is The Vickers indentation equipment used is built on the same(about 250 um), except for composites the base of an optical microscope equipped with sintered at the highest temperature(1250C) for two lenses and the indentation cell. 8 The sample is which the distance can reach I mm. fracture sur- polished with a diamond powder suspension(grain faces have shown fiber extractions about I mm size about 1 um) and set up on a force cell which long for composites sintered at 1250 C whereas, for allows forces up to in to be measured with an the others, they are only about 100 um long. The accuracy of 0.5mN. Indentation have been realised fracture of the composites is always controlled. in the middle of two crack phases and where the The mechanical behavior of T400HYMAS fiber distribution is homogeneous. The motoriza- composites is modified for each sintering condi tion of the microscope focusing permits to reach tion. The ultimate strength varies from 300 to loading speeds varying between 2 and 50 mN S 1100 MPa for a 35% fiber volumic fraction. The The minimum speed is used in order not to intro- average distance between microcracks, which have duce any damage in surrounding fibers. The dis- appeared oling in the composites due to the placement of the indentor relative to the sample thermal expansion mismatch between the fibers surface is measured with a capacitive cell, which xial CTE≈4.l 10-6C-1: radial CtE≈ gives a resolution of 10 nm for a 300 um cell-sam- 10-5C-) and the matrix, increases with sintering ple surface distance. The load-indentor displacement temperature. Up to 1150C, the ultimate strength curves are recorded by a computer(5 points per s) increases but the fracture remains brittle, except for the composite T1. In the case of brittle fractures, 2.4 Transmission electron microscopy fibe tions are about 100 um long. For non The micro- and nanotexture was studied with a brittle fractures (T1), some bundles of fibers are transmission electron microscope(Philips CM12 easily pulled-out, due to their bad impregnation by TEM), using a 120 k V high voltage and a super-twin the matrix at this sintering temperature. For tem- objective stage. The microscope was also equipped peratures higher than 1150C, the distance between with an energy dispersive analysis of X-ray system cracks is increased again, the fracture strength (EDAX). Prior to TEM investigations, small pieces decreases at the same time and the fracture tends to of composites were mechanically thinned on both become less brittle. Fibers extractions are still sample faces using a dimple grinder(Gatan 656) about 100 um long, but in delamination planes with a diamond powder suspension (grains sizes some bundles of fibers are easily and largely were in the 2-4 um range). Hence, ion-thinning was extracted performed(Gatan Duomill 600) using a 4 kV high In table 2. the nomenclature used is summarized voltage and incidence angles from 20 to 120 for the different composites with the hot-pressing Because of decohesion effects between fibers and conditions, the average distance between micro- matrices, embedding the composite samples in cracks and the average fracture strength for a 35% epoxy resin was often necessary to maintain the fiber volumic fraction. sample integrity during the thinning steps. The epoxy resin used ( TAAB or POLarBED 812)was 3.2 Push-in results characterized by some content in chlorine. Any A huge number of studies and models are devoted misinterpretation was then avoided by using the to the characterisation and properties of the fib EDAX analysis matrix interfaces21-24 in order to explain the frac- a glassy ceramic matrix being an electrical insu- ture behaviour of ceramic-ceramic composites lator, the quality of the TEM images are not so good The thermochemical analysis is based on debond- as with C/SiC composites. 9,20 Micrographs were then ing of the interface and the relative sliding of the sometimes improved by using a very thin(< 5nm) fibers and the matrix which can be defined by the carbon coating on the sample specimens for TEM shear stress T In the case of C/YMAS composites

At 1250C, the -Y2Si2O7 turns almost completely in the  form. 2.2 Mechanical tests Three-point bend tests are performed on compo￾sites bars (3461.5 mm3 ) at room temperature, with a span 30 mm long and a loading rate of 0.2 mm min.8,16 A displacement cell, set on the tension surface of the sample, in the middle, has allowed to obtain load-de¯ection curves. 2.3 Push-in test The Vickers indentation equipment used is built on the base of an optical microscope equipped with two lenses and the indentation cell.18 The sample is polished with a diamond powder suspension (grain size about 1m) and set up on a force cell which allows forces up to 1 N to be measured with an accuracy of 0.5 mN. Indentation have been realised in the middle of two crack phases and where the ®ber distribution is homogeneous. The motoriza￾tion of the microscope focusing permits to reach loading speeds varying between 2 and 50 mN sÿ1 . The minimum speed is used in order not to intro￾duce any damage in surrounding ®bers. The dis￾placement of the indentor relative to the sample surface is measured with a capacitive cell, which gives a resolution of 10 nm for a 300m cell-sam￾ple surface distance. The load-indentor displacement curves are recorded by a computer (5 points per s). 2.4 Transmission electron microscopy The micro- and nanotexture was studied with a transmission electron microscope (Philips CM12 TEM), using a 120 kV high voltage and a super-twin objective stage. The microscope was also equipped with an energy dispersive analysis of X-ray system (EDAX). Prior to TEM investigations, small pieces of composites were mechanically thinned on both sample faces using a dimple grinder (Gatan 656) with a diamond powder suspension (grains sizes were in the 2±4m range). Hence, ion-thinning was performed (Gatan Duomill 600) using a 4 kV high voltage and incidence angles from 20 to 12. Because of decohesion e€ects between ®bers and matrices, embedding the composite samples in epoxy resin was often necessary to maintain the sample integrity during the thinning steps. The epoxy resin used (TAAB or POLARBED 812) was characterized by some content in chlorine. Any misinterpretation was then avoided by using the EDAX analysis. A glassy ceramic matrix being an electrical insu￾lator, the quality of the TEM images are not so good as with C/SiC composites.19,20 Micrographs were then sometimes improved by using a very thin (< 5 nm) carbon coating on the sample specimens for TEM. 3 Results 3.1 Fracture behavior Pitch-based carbon ®ber-reinforced YMAS composites have been hot-pressed.8,17 Due to the thermal expansion mismatch between the carbon ®bers (axial CTE & 4.5.10ÿ6Cÿ1 ; radial CTE & 15.10ÿ6Cÿ1 ) 6 and the matrix (CTE & 6.10ÿ6Cÿ1 ),8 thermal stresses are induced in the composites on cooling making a network of microcracks. Whatever the sintering temperature, the average distance between two microcracks is the same (about 250m), except for composites sintered at the highest temperature (1250C) for which the distance can reach 1 mm. Fracture sur￾faces have shown ®ber extractions about 1 mm long for composites sintered at 1250C whereas, for the others, they are only about 100m long. The fracture of the composites is always controlled. The mechanical behavior of T400H/YMAS composites is modi®ed for each sintering condi￾tion. The ultimate strength varies from 300 to 1100MPa for a 35% ®ber volumic fraction. The average distance between microcracks, which have appeared on cooling in the composites due to the thermal expansion mismatch between the ®bers (axial CTE & 4.10ÿ6Cÿ1 ; radial CTE & 10ÿ5Cÿ1 ) 6 and the matrix, increases with sintering temperature. Up to 1150C, the ultimate strength increases but the fracture remains brittle, except for the composite T1. In the case of brittle fractures, ®ber extractions are about 100m long. For non￾brittle fractures (T1), some bundles of ®bers are easily pulled-out, due to their bad impregnation by the matrix at this sintering temperature.8 For tem￾peratures higher than 1150C, the distance between cracks is increased again, the fracture strength decreases at the same time and the fracture tends to become less brittle. Fibers extractions are still about 100m long, but in delamination planes some bundles of ®bers are easily and largely extracted. In Table 2, the nomenclature used is summarized for the di€erent composites with the hot-pressing conditions, the average distance between micro￾cracks and the average fracture strength for a 35% ®ber volumic fraction.8 3.2 Push-in results A huge number of studies and models are devoted to the characterisation and properties of the ®ber￾matrix interfaces21±24 in order to explain the frac￾ture behaviour of ceramic±ceramic composites. The thermochemical analysis is based on debond￾ing of the interface and the relative sliding of the ®bers and the matrix which can be de®ned by the shear stress . In the case of C/YMAS composites, Carbon ®ber-reinforced glass-ceramic matrix composites. III 319

320 V Bianchi et al Table 2. Main characteristics of P25/YMAS and T400H/YMAS composites, interfacial debonding(od) and shear(t)stresses(for T400H fiber composites, the number of pushed fibers is indicated) Ref. Densification Crystallisation Distance between Fracture Fracture T(MPa) conditions conditions microcracks strength (MPa)(Number of pushed fibers) °C-h °C-h) (MPa) PI 950-1 P2 970 P3 1000-1 P41050-0.5 PS 1050-1 245±65 455±73 17-8±48 P6 1050-1 1050-1-5 440±60 Controlled P7 1050-6 P8 1100-1 1150-1 P101050-0.5 1250-0.5 600±180 Pll 1050-1 1250-1-5 1000±270 59±33 1-0±07 1000-1 95±25 300±45 Controlled 108±337[15] 1025-1 100±30 520±70 Less controlled1137±167 363±80阿6 670±35 1100-1 330±8 770±105 1150-1 450±1201100±300 Brittle 769±285 167±88[0 T6 400±110 760±65 1050-1 1250-1-5 550±150 570±35 Less brittle 846±162 20.4±11-19 as is shown later, many parameters (thermal If all these parameters are considered, the expansion, mismatch, matrix crystallisation, shear stress must not be supposed constant along oxidation of the carbon fibers.) influence the the interface and the analysis becomes more fiber-matrix interface, so we have performed complex microindentation tests for a comparison and a The displacement cell measures the distance d quantitative study of the samples. In the case of between the indentor and the sample surface. It continuous fiber composites, Marshall> was the corresponds to the elastoplastic strain plus the first using a Vickers indentor to apply a load on a fiber-matrix sliding, for forces higher than the fiber(Nicalon)being perpendicular to the surface debonding force Fa. Therefore, it is necessary, for of a polished material. If the fiber-matrix bond is valuing the shear stress t, to determine, and extra not too strong, the applied load is high enough to plate for forces higher than Fd, the elastoplastic induce the sliding of the fiber in its matrix sheath. strain according to the law h=a F. Then the The first part of an indentation curve corre- fiber-matrix sliding u is obtained by subtracting ponds to the elastoplastic response of the fiber, the elastoplastic strain h to the measured displace- then the change of slope is related to debonding of ment d, and the shear stress can be extracted from the interface for the Fa force. If R is the radius of the following relation(2) the fiber the critical debonding stress can be expressed by the following equation F--Fd 4.x2.R3·E F R Experimental load-indentor displacement curves with the corresponding elastoplastic curves are The second part of the curve describes the relative represented in Fig. I where different debonding fiber-matrix sliding. If the sliding is assumed and sliding behaviors are shown. For the P5 com- purely frictional without shrinking of the interface, posite, the force required for debonding is higher and if the deformation of the fiber under the load is than for Pll for which afterwards sliding is easier neglected, the fiber-matrix reloading is carried out These experiments have been reproduced six times for a constant shear stress t 21 for each composite. Statistical debonding and To obtain an absolute value of the shear stress, it sliding stresses are summarized in Table 2. They is necessary to take into account in this analysis confirm that the strength of the interface is very other parameters such as the influence of the Pois- low for the composite Pll, which agrees with son's effect under the indentation load which previous observations on the pull-out lengths on modifies the interfacial radial stress the influence surface fractures and the increase of the distance of the residual axial stress the influence of the between microcracks. the distance between anisotropy of the fiber and the influence of the fiber cracks is inversely proportional with the inte roughness facial shear stress. A low interfacial shear stress

as is shown later, many parameters (thermal expansion, mismatch, matrix crystallisation, oxidation of the carbon ®bers...) in¯uence the ®ber±matrix interface, so we have performed microindentation tests for a comparison and a quantitative study of the samples. In the case of continuous ®ber composites, Marshall25 was the ®rst using a Vickers indentor to apply a load on a ®ber (Nicalon) being perpendicular to the surface of a polished material. If the ®ber±matrix bond is not too strong, the applied load is high enough to induce the sliding of the ®ber in its matrix sheath. The ®rst part of an indentation curve corre￾sponds to the elastoplastic response of the ®ber, then the change of slope is related to debonding of the interface for the Fd force. If R is the radius of the ®ber, the critical debonding stress can be expressed by the following equation: d ˆ Fd 
R2 …1† The second part of the curve describes the relative ®ber±matrix sliding. If the sliding is assumed purely frictional without shrinking of the interface, and if the deformation of the ®ber under the load is neglected, the ®ber±matrix reloading is carried out for a constant shear stress . 21 To obtain an absolute value of the shear stress, it is necessary to take into account in this analysis other parameters such as the in¯uence of the Pois￾son's e€ect under the indentation load which modi®es the interfacial radial stress, the in¯uence of the residual axial stress, the in¯uence of the anisotropy of the ®ber and the in¯uence of the ®ber roughness.22±26 If all these parameters are considered, the shear stress must not be supposed constant along the interface and the analysis becomes more complex. The displacement cell measures the distance d between the indentor and the sample surface. It corresponds to the elastoplastic strain plus the ®ber±matrix sliding, for forces higher than the debonding force Fd. Therefore, it is necessary, for valuing the shear stress , to determine, and extra￾polate for forces higher than Fd, the elastoplastic strain according to the law h ˆ 
F. Then the ®ber±matrix sliding u is obtained by subtracting the elastoplastic strain h to the measured displace￾ment d, and the shear stress can be extracted from the following relation(2): 8F  Fd;  ˆ F2 ÿ F2d 4
2
R3
Ef
u …2† Experimental load-indentor displacement curves with the corresponding elastoplastic curves are represented in Fig. 1 where di€erent debonding and sliding behaviors are shown. For the P5 com￾posite, the force required for debonding is higher than for P11, for which afterwards sliding is easier. These experiments have been reproduced six times for each composite. Statistical debonding and sliding stresses are summarized in Table 2. They con®rm that the strength of the interface is very low for the composite P11, which agrees with previous observations on the pull-out lengths on surface fractures and the increase of the distance between microcracks. The distance between cracks is inversely proportional with the inter￾facial shear stress.1 A low interfacial shear stress Table 2. Main characteristics of P25/YMAS and T400H/YMAS composites,8 interfacial debonding (sd) and shear (t) stresses (for T400H ®ber composites, the number of pushed ®bers is indicated). Ref. Densi®cation conditions (C-h) Crystallisation conditions (C-h) Distance between microcracks (m) Fracture strength (MPa) Fracture type d (MPa)  (MPa) (Number of pushed ®bers) P1 950±1 P2 970±1 P3 1000±1 P4 1050±0.5 P5 1050±1 245‹65 455‹73 17.8‹4.8 P6 1050±1 1050±1.5 440‹60 Controlled P7 1050±1 1050±6 P8 1100±1 P9 1150±1 P10 1050±0.5 1250±0.5 600‹180 P11 1050±1 1250±1.5 1000‹270 59‹33 1.0‹0.7 T1 1000±1 95‹25 300‹45 Controlled 1322‹258 108‹33.7 [15] T2 1025±1 100‹30 520‹70 Less controlled 1137‹167 36.3‹8.0 [6] T3 1050±1 190‹35 670‹35 T4 1100±1 330‹85 770‹105 T5 1150±1 450‹120 1100‹300 Brittle 769‹285 16.7‹8.8 [20] T6 1200±1 400‹110 760‹65 T7 1050±1 1250±1.5 550‹150 570‹35 Less brittle 846‹162 20.4‹11.1 [9] 320 V. Bianchi et al

Carbon fiber-reinforced glass-ceramic matrix composites. III 321 8000 cally a lack of control of the phase distribution within the precursor, which is a two-phase petro- leum pitch The poorly organised carbon in the isotropic 4000 areas exhibits a higher reactivity regarding the matrix compared to the graphite-like fakes lying flat on the fiber. Indeed. it has been shown that the edge atoms in graphitic structures are more reac- tive than basal atoms and the study of carbon 02004006008001000120014001600 reactivity is now grounded on the active site con Indentor displacement (nm cept. 32 Depending on the time/temperature condi- Fig. 1. Experimental load-indentor displacer nd asso- tions, the reactivity may induce chemical bondings, ciated elastoplastic curves for P5 and when limited, or carbon gasification, when exten- sive. Partial oxidation of some fibers during the corresponds to a weak interface which allows the composite preparation could thus lead to some sliding of the fibers in matrix blocks and their alteration of the isotropic areas close to the fiber easier extraction surface, as observed. On the contrary the outer In the same way, indentations are realised on graphite-like skin, when any, could prevent the composites containing T400H fibers. Average fiber from oxidation by the matrix to some extent debonding and sliding stresses are summarized on Whatever the sintering conditions are, the Table 2 with the number of indented fibers. These debondings are very often localized between results allow a general trend to be defined. The od stiff carbon layers of the graphite-like flakes and t stresses decrease when the sintering ter the rest of the fiber(Figs 3 and 4). This feature perature is raised up to 1150C, which is in agree- often links the fiber and the matrix one to each ment with the changes in ultimate strengths and distances between microcracks but also with tem observations of the fiber-matrix interface. as will be explained later. For T5 and T7 composites, MATRⅨX debonding and sliding stresses are nearly similar, in he same way as fiber extraction lengths except in delamination planes. Thus the lower fracture strength and the more controlled fracture for T7 would be due to this delamination. 3.3 Observation of the fiber-matrix interface and discussion The P25 fiber presents statistically a 'Pan-Am cross-sectional texture with a highly anisotropic but uncomplete graphite-like carbon skin, which 00 nm does not adhere well to the fiber: such a skin. if b any, is made of polyaromatic layer stacks, exhibit- ing few structural defects, and often lying fat on the fiber surface. The fiber surface may appear ATRIX either smooth or corrugated(Fig. 2) Highly corrugated fibers [Fig. 2(b)] are asso- ciated with high concentrations of isotropic areas in the outer zone of the fibers. These areas (light areas in the picture) are often found altered(oxi- dized), and not associated with graphite-like flakes FIBER nor interfacial separations. On the contrary, for the smooth fibers [Fig. 2(a)], debondings are more obvious, while pieces of graphite-like flakes do not adhere to the fiber and are often found on the matrix side of the debonding, sometimes embedded Fig. 2. Composite P5. Bright field TEM images (a)and (b) are within the matrix. Such discrepancies in outer iso- Light areas within fibers are isotropic part tropic zone concentrations are assumed to two examples of P25 fibers with various corrugated surfaces. The bright rim at the interface indicates a fiber-matrix originate from the preparation conditions, specifi

corresponds to a weak interface which allows the sliding of the ®bers in matrix blocks and their easier extraction. In the same way, indentations are realised on composites containing T400H ®bers. Average debonding and sliding stresses are summarized on Table 2 with the number of indented ®bers. These results allow a general trend to be de®ned. The d and  stresses decrease when the sintering tem￾perature is raised up to 1150C, which is in agree￾ment with the changes in ultimate strengths and distances between microcracks but also with TEM observations of the ®ber±matrix interface, as will be explained later. For T5 and T7 composites, debonding and sliding stresses are nearly similar, in the same way as ®ber extraction lengths except in delamination planes. Thus the lower fracture strength and the more controlled fracture for T7 would be due to this delamination.8 3.3 Observation of the ®ber±matrix interface and discussion The P25 ®ber presents statistically a `Pan-Am' cross-sectional texture with a highly anisotropic but uncomplete graphite-like carbon skin, which does not adhere well to the ®ber; such a skin, if any, is made of polyaromatic layer stacks, exhibit￾ing few structural defects, and often lying ¯at on the ®ber surface. The ®ber surface may appear either smooth or corrugated (Fig. 2).8 Highly corrugated ®bers [Fig. 2(b)] are asso￾ciated with high concentrations of isotropic areas in the outer zone of the ®bers. These areas (light areas in the picture) are often found altered (oxi￾dized), and not associated with graphite-like ¯akes nor interfacial separations. On the contrary, for the smooth ®bers [Fig. 2(a)], debondings are more obvious, while pieces of graphite-like ¯akes do not adhere to the ®ber and are often found on the matrix side of the debonding, sometimes embedded within the matrix. Such discrepancies in outer iso￾tropic zone concentrations are assumed to originate from the preparation conditions, speci®- cally a lack of control of the phase distribution within the precursor, which is a two-phase petro￾leum pitch. The poorly organised carbon in the isotropic areas exhibits a higher reactivity regarding the matrix compared to the graphite-like ¯akes lying ¯at on the ®ber. Indeed, it has been shown that the edge atoms in graphitic structures are more reac￾tive than basal atoms31 and the study of carbon reactivity is now grounded on the active site con￾cept.32 Depending on the time/temperature condi￾tions, the reactivity may induce chemical bondings, when limited, or carbon gasi®cation, when exten￾sive. Partial oxidation of some ®bers during the composite preparation could thus lead to some alteration of the isotropic areas close to the ®ber surface, as observed. On the contrary the outer graphite-like skin, when any, could prevent the ®ber from oxidation by the matrix to some extent. Whatever the sintering conditions are, the debondings are very often localized between the sti€ carbon layers of the graphite-like ¯akes and the rest of the ®ber (Figs 3 and 4). This feature often links the ®ber and the matrix one to each Fig. 1. Experimental load-indentor displacement and asso￾ciated elastoplastic curves for P5 and P11 composites. Fig. 2. Composite P5. Bright ®eld TEM images, cross-section. Light areas within ®bers are isotropic parts.8 (a) and (b) are two examples of P25 ®bers with various corrugated surfaces. The bright rim at the interface indicates a ®ber±matrix decohesion. Carbon ®ber-reinforced glass-ceramic matrix composites. III 321

32 other through the graphite-like flakes(Fig 4). In mainly due to the thermal expansion mismatch some cases [P3 or P5, Fig. 2(a)], the outlines of the between both constituents. On the other hand, for matrix may follow rather well those of the fiber, other composites, the outlines on both sides of the which indicates specifically that the debondings are debonding are not parallel, which suggests that debonding is not due to the thermal expansion could be a IBER h only. Other wettability of the fiber surface by the glassy matrix when the hot-pressing too short (P10 Fig. 5) gasification due to oxydo-red tion reactions with the matrix when an annealing is performed at high temperature(P10 or Pll)for a long time(P7, Fig. 6). Indeed, the amorphous aspect of several fiber-matrix contact zones sup- ports a possible matrix reduction effect, as far as the partial oxidation of polyaromatic carbon can MATRIX lead to the destruction of its structural organiza tion and to a polyaromatic carbon phase without any evident texture. 3 It is noteworthy that the high anisotropy of the graphite-like carbon flakes prevents them from being oxidised by the matrix 100nm FIBER and that oxidized carbon is found behind the flakes bulk fiber(Fig. 7) Fig. 3. Composite P3. Coo2 dark field TEM image, long from the microcrack examination the indenta itudinal section, Bright areas are due to graphenes layers tion experiments(Table 2), and the TEM study, it (arrow) into the matrix is a graphite-like flake sur- is possible to state that interfacial debonding can rounded then ed by the glas issue from four origins: (1)the thermal expansion mismatch between the fiber and the matrix, (2) the matrix shrinkage due to structural changes such glass/crystal transformation or phase changes, (3) the poor wettability due to the high viscosity of the glass, and(4) the oxidation of carbon fibers by the FIBER FIBER FIBER MATRIX 00 nm 100 Fig. 5. Composite P10. Bright field TEM image, cross-section Fig.4. Composite Pll. Bright field TEM image, longitudinal Example of the poor wettability of the fibers by the matrix due phite-like flake(arrow)and the fiber, as in Figs 3and 2 Q. section. The fiber-matrix decohesion occurs between the to the high viscosity of the glass associated with a short densification time

other through the graphite-like ¯akes (Fig. 4). In some cases [P3 or P5, Fig. 2(a)], the outlines of the matrix may follow rather well those of the ®ber, which indicates speci®cally that the debondings are mainly due to the thermal expansion mismatch between both constituents. On the other hand, for other composites, the outlines on both sides of the debonding are not parallel, which suggests that debonding is not due to the thermal expansion mismatch only. Other reasons could be a poor wettability of the ®ber surface by the glassy matrix when the hot-pressing time is too short (P10, Fig. 5), or some gasi®cation due to oxydo-reduc￾tion reactions with the matrix when an annealing is performed at high temperature (P10 or P11) for a long time (P7, Fig. 6). Indeed, the amorphous aspect of several ®ber±matrix contact zones sup￾ports a possible matrix reduction e€ect, as far as the partial oxidation of polyaromatic carbon can lead to the destruction of its structural organiza￾tion and to a polyaromatic carbon phase without any evident texture.33 It is noteworthy that the high anisotropy of the graphite-like carbon ¯akes prevents them from being oxidised by the matrix, and that oxidized carbon is found behind the ¯akes, i.e. between the ¯akes and the bulk ®ber (Fig. 7). From the microcrack examination, the indenta￾tion experiments (Table 2), and the TEM study, it is possible to state that interfacial debonding can issue from four origins : (1) the thermal expansion mismatch between the ®ber and the matrix, (2) the matrix shrinkage due to structural changes such as glass/crystal transformation or phase changes, (3) the poor wettability due to the high viscosity of the glass, and (4) the oxidation of carbon ®bers by the Fig. 3. Composite P3. C002 dark ®eld TEM image, long￾itudinal section. Bright areas are due to graphenes layers oriented parallel to the double bar sign. The bright rim (arrow) into the matrix is part of a graphite-like ¯ake sur￾rounded then embedded by the glass. Fig. 4. Composite P11. Bright ®eld TEM image, longitudinal section. The ®ber±matrix decohesion occurs between the gra￾phite-like ¯ake (arrow) and the ®ber, as in Figs 3 and 7. Fig. 5. Composite P10. Bright ®eld TEM image, cross-section. Example of the poor wettability of the ®bers by the matrix due to the high viscosity of the glass associated with a short densi®cation time. 322 V. Bianchi et al

Carbon fiber-reinforced glass-ceramic matrix composites. III 323 matrix(without formation of any carbide). The due to the thermal expansion mismatch between relative importance of the four reasons depends on the fibers and the matrix(parallel outlines). How the sintering conditions. Effect of reason (1) ever some contact zones are detected (thermal expansion mismatch) increases with the In the interfacial contact zones both for ti or densification plus crystallization temperature, T5, the carbon is highly disordered(Fig. 10). Thi effect of reason(2)(matrix shrinkage)is high from amorphous-like carbon interphase, about 20 nm the crystallization temperature (1100 C) but thick, may be due to the oxidation of the fiber more limited afterwards, effect of reason(3)(poor surface by the matrix but can also originate from wettability) is high for short densification times, the pyrolysis and carbonization of sizing. Sizing is effect of reason(4)(carbon gasification) increases about 30 nm thick4and its carbonization can have with densification plus crystallization temperature led to a 15 nm thick carbonaceous residue For the t6 composite, the pressure is applied at TI composite was the one exhibiting the highest 1200oC, i.e. at a temperature where the matrix porosity (15%), making it very friable so that viscosity has become too high for a good wett- cross-sections were not possible to prepare. There- ability(Fig. 11)of the fibers by the matrix due to fore, the quality of the fiber-matrix interface was the progress of the crystallisation as shown by not able to be accurately characterized. However, XRD experiments from TEM observations(Fig 8), it is unlikely that 30 nm wide is observed on a cross-section of the t5 composite(Fig. 9). The debonding is essentially FIBER FIBER MATRIX 10 nm 100nm Fig. 7. Composite Pll. Lattice fringe TEM image, cross-sec- Fig. 6. Compos Bright field TEM ion. Again, the decohesion occurs between the graphite-like The matrix surface outline does not match well the fiber sur- flake(arrow )and the fiber. The actual decohesion width is 25- face, possibly due to the partial oxidation of the fiber 30 nm. The fiber surface is altered by oxidation( the graphene enhanced by the high temperature/time densification condi- fringe cannot be image) while the graphite-like flake(arrow)is not, though at the very contact to the matrix

matrix (without formation of any carbide). The relative importance of the four reasons depends on the sintering conditions. E€ect of reason (1) (thermal expansion mismatch) increases with the densi®cation plus crystallization temperature, e€ect of reason (2) (matrix shrinkage) is high from the crystallization temperature (1100C)17 but more limited afterwards, e€ect of reason (3) (poor wettability) is high for short densi®cation times, e€ect of reason (4) (carbon gasi®cation) increases with densi®cation plus crystallization temperature or time. T1 composite was the one exhibiting the highest porosity (15%), making it very friable so that cross-sections were not possible to prepare. There￾fore, the quality of the ®ber±matrix interface was not able to be accurately characterized. However, from TEM observations (Fig. 8), it is unlikely that the porosity is mainly located in interfacial zones. On the contrary, an interfacial separation about 30 nm wide is observed on a cross-section of the T5 composite (Fig. 9). The debonding is essentially due to the thermal expansion mismatch between the ®bers and the matrix (parallel outlines). How￾ever some contact zones are detected. In the interfacial contact zones, both for T1 or T5, the carbon is highly disordered (Fig. 10). This amorphous-like carbon interphase, about 20 nm thick, may be due to the oxidation of the ®ber surface by the matrix but can also originate from the pyrolysis and carbonization of sizing. Sizing is about 30 nm thick34 and its carbonization can have led to a 15 nm thick carbonaceous residue. For the T6 composite, the pressure is applied at 1200C, i.e. at a temperature where the matrix viscosity has become too high for a good wett￾ability (Fig. 11) of the ®bers by the matrix due to the progress of the crystallisation as shown by XRD experiments.8 Fig. 6. Composite P7. Bright ®eld TEM image, cross-section. The matrix surface outline does not match well the ®ber sur￾face, possibly due to the partial oxidation of the ®ber enhanced by the high temperature/time densi®cation condi￾tions. Fig. 7. Composite P11. Lattice fringe TEM image, cross-sec￾tion. Again, the decohesion occurs between the graphite-like ¯ake (arrow) and the ®ber. The actual decohesion width is 25± 30 nm. The ®ber surface is altered by oxidation (the graphene fringe cannot be image) while the graphite-like ¯ake (arrow) is not, though at the very contact to the matrix. Carbon ®ber-reinforced glass-ceramic matrix composites. III 323

324 V. Bianchi et al For the T7 composite, the interfacial separation is slightly wider, on average 100 nm. The outlines of the matrix are very often different from those of the fiber(Fig. 12). Nevertheless, some contact zones, where carbon is amorphous-like, revealing a strong alteration, always subsist The same four reasons already discussed above for the P25-reinforced composites are able to account for the differences in interfacial features within the T400H-reinforced composites. The thermal expansion mismatch, though effective, less important because the radial Cte of the FIBER MATRIX IBER 100nm 8. Composite T1. Bright field TEM n. Since not d image, longitudinal evidence interfacial decohesion. however. it is doubtful that the 15% porosity are located at interfacial areas MATRIX MATRIX m nmm Fig. 10. Composite T5 Lattice fringe TEM image, cross-sec- Fig 9. Composite T5. Bright field TEM image, cross-section. tion Occurrence of an amorphous carbon zone(arrow) at the Decohesions occur at the fiber-matrix interfaces, though some points of fiber-matrix contact, either due to the oxidation of contact zones are found. Fiber and matrix outlines match the fiber surface or possibly to the carbonisation of the poly rather well mer sIzing

For the T7 composite, the interfacial separation is slightly wider, on average 100 nm. The outlines of the matrix are very often di€erent from those of the ®ber (Fig. 12). Nevertheless, some contact zones, where carbon is amorphous-like, revealing a strong alteration, always subsist. The same four reasons already discussed above for the P25-reinforced composites are able to account for the di€erences in interfacial features within the T400H-reinforced composites. The thermal expansion mismatch, though e€ective, is less important because the radial CTE of the Fig. 8. Composite T1. Bright ®eld TEM image, longitudinal section. Since not diametral, such a section is not suitable to evidence interfacial decohesion, however, it is doubtful that the 15% porosity are located at interfacial areas. Fig. 9. Composite T5. Bright ®eld TEM image, cross-section. Decohesions occur at the ®ber±matrix interfaces, though some contact zones are found. Fiber and matrix outlines match rather well. Fig. 10. Composite T5. Lattice fringe TEM image, cross-sec￾tion. Occurrence of an amorphous carbon zone (arrow) at the points of ®ber±matrix contact, either due to the oxidation of the ®ber surface or possibly to the carbonisation of the poly￾mer sizing. 324 V. Bianchi et al

Carbon fiber-reinforced glass-ceramic matrix composites. III T400H fiber is lower than that of P25 fibers(see features similar to those of Tl, i.e. a low fracture Section 3-1).Actually, interfacial decohesion are less strength and a controlled fracture. The fact that important for all of the T400H-reinforced compo- the sintering temperature range(1050oC)was cor- sites compared to P25-reinforced composites. This responding to the largest N2 release together with a has helped in maintaining fiber-matrix contact good fluidity of the matrix, and also the fact that areas in all of the T-type composites, which are the the high crystallization temperatures(1250C)was sites of fiber-matrix chemical interactions. able to alter the carbon interfacial zones could Depending on the hot-pressing conditions, the lead to local great fiber-matrix debondings. More contact zones are more or less numerous, and generally, moderate t values have been measured depending on the ultimate temperature used, the and comparatively short extractions observed chemical interactions may have a positive(bond- except in delamination planes ing)or negative(gasification effect)on the interface Finally, t5 composite appears as the convenien strength. Finally, two main differences are that, compromise, with a hot-pressing during the thermal treatment ( densification or (1150 C)sufficiently low not to reach the complete crystallization), gaseous nitrogen(originating from matrix crystallization, sufficiently high to allow the pan precursor)is released from T400H fibers, most of the gaseous N2 to be released before and that the sizing of the fibers is likely to carbo- nize into a thin amorphous-like carbon phase. Thus for T-400 H-reinforced composites, T1 composite exhibits a low fracture strength with a high interfacial shear stress but a controlled frac ture, which is possibly the resulting effect of both a relatively strong fiber-matrix bonding through a poorly organized interphase(which induces high t values and a low fracture strength) together with the high porosity mainly due to the N2 release (which leads to a controlled fracture because of a bad impregnation of some bundles of fibers). From TI to T5, the N2 release is less and less important at the hot-pressing step therefore less and less dis- turbing for the fiber-matrix contact, while the fiber-matrix chemical interactions are closer to an alteration process, making the interfacial shear MATRIX stress weaker. T7 composite exhibits mechanical MATRIX FIBER FIBER m 200nm Fig. 11. Composite T6. Bright field mage, cross-section Example of the poor wettability of the fibers by the matrix, Fig 12. Composite T7 Bright field TEM image, cross-section due to the high viscosity of the matrix associated with a high The fiber surface outline does not match well the matrix out- densification temperature, which enhances the recrystallisation line. due conjunction of events such as high gaseous itrogen release and crystallisation of the matrix

T400H ®ber is lower than that of P25 ®bers (see Section 3.1). Actually, interfacial decohesions are less important for all of the T400H-reinforced compo￾sites compared to P25-reinforced composites. This has helped in maintaining ®ber±matrix contact areas in all of the T-type composites, which are the sites of ®ber±matrix chemical interactions. Depending on the hot-pressing conditions, the contact zones are more or less numerous, and depending on the ultimate temperature used, the chemical interactions may have a positive (bond￾ing) or negative (gasi®cation e€ect) on the interface strength. Finally, two main di€erences are that, during the thermal treatment (densi®cation or crystallization), gaseous nitrogen (originating from the PAN precursor) is released from T400H ®bers, and that the sizing of the ®bers is likely to carbo￾nize into a thin amorphous-like carbon phase. Thus for T-400 H-reinforced composites, T1 composite exhibits a low fracture strength with a high interfacial shear stress but a controlled frac￾ture, which is possibly the resulting e€ect of both a relatively strong ®ber±matrix bonding through a poorly organized interphase (which induces high  values and a low fracture strength) together with the high porosity mainly due to the N2 release (which leads to a controlled fracture because of a bad impregnation of some bundles of ®bers). From T1 to T5, the N2 release is less and less important at the hot-pressing step, therefore less and less dis￾turbing for the ®ber±matrix contact, while the ®ber±matrix chemical interactions are closer to an alteration process, making the interfacial shear stress weaker. T7 composite exhibits mechanical features similar to those of T1, i.e. a low fracture strength and a controlled fracture. The fact that the sintering temperature range (1050C) was cor￾responding to the largest N2 release together with a good ¯uidity of the matrix, and also the fact that the high crystallization temperatures (1250C) was able to alter the carbon interfacial zones, could lead to local great ®ber±matrix debondings. More generally, moderate  values have been measured and comparatively short extractions observed, except in delamination planes. Finally, T5 composite appears as the convenient compromise, with a hot-pressing temperature (1150C) suciently low not to reach the complete matrix crystallization, suciently high to allow most of the gaseous N2 to be released before Fig. 11. Composite T6. Bright ®eld TEM image, cross-section. Example of the poor wettability of the ®bers by the matrix, due to the high viscosity of the matrix associated with a high densi®cation temperature, which enhances the recrystallisation progress. Fig. 12. Composite T7. Bright ®eld TEM image, cross-section. The ®ber surface outline does not match well the matrix out￾line, due to a conjunction of events such as high gaseous nitrogen release and crystallisation of the matrix. Carbon ®ber-reinforced glass-ceramic matrix composites. III 325

326 V. Bianchi et al applying the pressure, and sufficiently moderate to Refe rences allow non-drastic carbon oxidation reactions to occur.Correspondingly, the fracture strength is 1. Aveston, J, Cooper, G. A and Kelly, A, The properties highest with a rather brittle fracture of fibre composites: strength and toughness in fibre reinforced ceramics In Conference Proceedings, National Physics Laboratory, Teddington, 1971, pp 63-73 2. Sambell, R. A. J, Briggs, A, Phillips, D. C. and Bowen, 4 Conclusion D. H, Carbon fibers composites with ceramic and glass matrices, Part 2. Continuous fibers. J. Mater. Sci., 1972, 7 The chemical reactivity of carbon fiber-reinforced 67681 3. Prewo, K. M. and Brennan, J. J, High strength silicon YMAS composites depends strongly on the carbon carbide fiber-reinforced glass matrix composites.J. Mater fibers used and the hot-pressing conditions which Sci,1980,15,463-68. influence the nature and the strength of the fiber- Seraudie, C, Elaboration et proprietes thermomecaniques matrix interface de composites a fibres de carbure de silicium et matrices vitroceramiques. Ph. D. thesis, Limoges University, 1995 The P25 pitch-based fibers are characterized by a 5. Hutchings, I. M, Tribology Friction and Wear of Engi- highly anisotropic but uncomplete graphite-like neering Materials. Edward Arnold. London. Melbourne carbon skin which does not adhere well to the 6. Bianchi, V, Composites a fibres de carbone et matrice fibers, so that the debondings are very often loca- YMAS elaboration, microstructure, comportement lized between the stiff carbon layers and the rest of mecanique et tribologique. Ph. D. thesis, Limoges the fibers. Due to this interfacial separation, the 7. Berthier, Y, Tribologie, Science Carrefour. In Journee interfacial shear stress is always relatively low and Europeenne du Freinage, JEF 92 GFC 92 Lille, 1992 the sliding is then made possible. Consequently, th 8. Bianchi. V. Sinkler. Goursat. P. W. Monthioux. M. and fracture is always controlled and the ultimate Menessier, E, Carbon fiber-reinforced (YMAs) glass strength of all the composites in the same order of ceramic matrix composites. I. preparation, structure and fracture strength. J. Europ Ceram. Soc., accer magnitude. However, the preparation of compo 9. Kerans, R.J., Hay, R.S., Pagano, N. J. and Parthasar- sites at high temperature(1250C)seems to lead to athy, T.A., The role of the fiber-matrix interface in cera a predominant effect of the oxidation of the carbon mic composites. Am. Cera. Soc. Bull., 1989, 68, 429- fibers by the matrix with some residual amor- 10. Brennan, J J, Interfacial characterization of glass and phous-like contact zones glass-ceramic matrix/Nicalon SiC fiber composites. Tai The mechanical behaviour of PAN-based carbon loring multiphase and composite ceramics. Mater. Sci fiber-reinforced composites is more sensible to hot Pantano and r. e. Newham Plenum Press New york pressing conditions, due to several mechanisms 986,pp.549-560 which can have opposite effects. The strength of leay, s M, Scott, v. D, Harris, B, Cooke, R. G. and the fiber-matrix bonds depends on the quality of Habib. F.A. Interface characterization and fracture of calcium aluminosilicate glass-ceramic reinforced with the amorphous-like carbon interphase, which may Nicalon fibres. J. Mater. Sci. 1992. 27. 2811-282 proceed from the carbonization of sizing or may be 12. Prewo, K M, Carbon fibre reinforced glass matrix com- due to the oxidation of the fiber surface by the 23.2745-2752. matrix. The interfacial shear stress is high for sin- 13. Benson, P. M, Spear, K. E and Pantano, C.G., Inter- tering at low temperature because the interphase is facial characterization of glass matrix/Nicalon SiC fiber not altered and the bond is strong. For higher hot composites: a thermodynamic approach. Ceram. Eng. Sci. Proc.,1988,9,663-669 pressing temperatures, the shear stress decreases, 14. Tredway, WK,Prewo, K. Pantano, C.G. llowing dissipation of much energy by debondin matrix interfacial effects in fiber-reinforced and sliding, and thus the improvement of the frac atrix composites. Carbon 27,717-727 15. Benson, P. M, Spear, K.E. and Pantano, C. G,Ther ture strength. A high crystallisation of the matrix mechanical analyses of interface reactions in carbon prevents the fibers from a good wetting, but also lber-reinforced glass matrix composites. Mater. Sci Res accentuates the effect of oxidation which. added to change N2 releases, leads to local debondings. The 16. Bianchi, V, Goursat, P,Menessier, E,Sinkler,W.and Monthioux, M, C/YMAS composites--effects of the ultimate strength is then lowered but the fracture is interface and the residual stresses on the rupture behavior more controlled Advanced Structural Fiber Composites, Advances The aim of this work was to relate, precisely, the 17. Sinkler W. Monthioux. M. Bianchi. V. Goursat P and Menessier, E, Carbon fiber-reinforced YMAS glass-cera- macroscopic behavior of composites. The impor mic matrix composites. II: Structural changes in the of the thermal expansion mismatch between constituents has not yet been entirely dis- 18. Parlier, M. Passilly, B and Sudre O, Caracterisation nd is the subject of a separate pal micromecanique dees composites a matrice ceramique a when a synthesis between the role played by the Taide de la technique de microindentation instrumented thermal residual stresses and the chemical reactiv In Work shop ll: Introduction of Ceramics Structural Composites, 76th Meeting of the Structures and ity of both constituents is realised Materials Panels, Anthalya, Turkey, 1993

applying the pressure, and suciently moderate to allow non-drastic carbon oxidation reactions to occur. Correspondingly, the fracture strength is highest with a rather brittle fracture. 4 Conclusion The chemical reactivity of carbon ®ber-reinforced YMAS composites depends strongly on the carbon ®bers used and the hot-pressing conditions which in¯uence the nature and the strength of the ®ber± matrix interface. The P25 pitch-based ®bers are characterized by a highly anisotropic but uncomplete graphite-like carbon skin, which does not adhere well to the ®bers, so that the debondings are very often loca￾lized between the sti€ carbon layers and the rest of the ®bers. Due to this interfacial separation, the interfacial shear stress is always relatively low and the sliding is then made possible. Consequently, the fracture is always controlled and the ultimate strength of all the composites in the same order of magnitude. However, the preparation of compo￾sites at high temperature (1250C) seems to lead to a predominant e€ect of the oxidation of the carbon ®bers by the matrix with some residual amor￾phous-like contact zones. The mechanical behaviour of PAN-based carbon ®ber-reinforced composites is more sensible to hot￾pressing conditions, due to several mechanisms which can have opposite e€ects. The strength of the ®ber±matrix bonds depends on the quality of the amorphous-like carbon interphase, which may proceed from the carbonization of sizing or may be due to the oxidation of the ®ber surface by the matrix. The interfacial shear stress is high for sin￾tering at low temperature because the interphase is not altered and the bond is strong. For higher hot￾pressing temperatures, the shear stress decreases, allowing dissipation of much energy by debonding and sliding, and thus the improvement of the frac￾ture strength. A high crystallisation of the matrix prevents the ®bers from a good wetting, but also accentuates the e€ect of oxidation which, added to change N2 releases, leads to local debondings. The ultimate strength is then lowered but the fracture is more controlled. The aim of this work was to relate, precisely, the nature and the strength of the interface to the macroscopic behavior of composites. The impor￾tance of the thermal expansion mismatch between both constituents has not yet been entirely dis￾cussed and is the subject of a separate paper35 when a synthesis between the role played by the thermal residual stresses and the chemical reactiv￾ity of both constituents is realised. References 1. Aveston, J., Cooper, G. A. and Kelly, A., The properties of ®bre composites: strength and toughness in ®bre reinforced ceramics. In Conference Proceedings, National Physics Laboratory, Teddington, 1971, pp. 63±73. 2. Sambell, R. A. J., Briggs, A., Phillips, D. C. and Bowen, D. H., Carbon ®bers composites with ceramic and glass matrices, Part 2, Continuous ®bers. J. Mater. Sci., 1972, 7, 676±81. 3. Prewo, K. M. and Brennan, J. J., High strength silicon carbide ®ber-reinforced glass matrix composites. J. Mater. Sci., 1980, 15, 463±68. 4. Seraudie, C., Elaboration et proprieÂteÂs thermomeÂcaniques de composites aÁ ®bres de carbure de silicium et matrices vitroceÂramiques. Ph.D. thesis, Limoges University, 1995. 5. Hutchings, I. M., Tribology : Friction and Wear of Engi￾neering Materials. Edward Arnold, London, Melbourne, Auckland, 1992. 6. Bianchi, V., Composites aÁ ®bres de carbone et matrice YMAS : elaboration, microstructure, comportements meÂcanique et tribologique. Ph.D. thesis, Limoges University, 1995. 7. Berthier, Y., Tribologie, Science Carrefour. In Journee EuropeÂenne du Freinage, JEF 92, GFC 92 Lille, 1992. 8. Bianchi, V., Sinkler, Goursat, P. W., Monthioux, M. and MeÂnessier, E., Carbon ®ber-reinforced (YMAS) glass￾ceramic matrix composites. I. preparation, structure and fracture strength. J. Europ. Ceram. Soc., accepted. 9. Kerans, R. J., Hay, R. S., Pagano, N. J. and Parthasar￾athy, T. A., The role of the ®ber-matrix interface in cera￾mic composites. Am. Ceram. Soc. Bull., 1989, 68, 429± 442. 10. Brennan, J. J., Interfacial characterization of glass and glass-ceramic matrix/Nicalon SiC ®ber composites. Tai￾loring multiphase and composite ceramics. Mater. Sci. Res., Vol. 20 ed. R. E. Tressler, G. L. Messing, C. G. Pantano and R. E. Newham. Plenum Press, New York, 1986, pp. 549±560. 11. Bleay, S. M., Scott, V. D., Harris, B., Cooke, R. G. and Habib, F. A., Interface characterization and fracture of calcium aluminosilicate glass-ceramic reinforced with Nicalon ®bres. J. Mater. Sci., 1992, 27, 2811±2822. 12. Prewo, K. M., Carbon ®bre reinforced glass matrix com￾posites tension and ¯exure properties. J. Mater. Sci., 1988, 23, 2745±2752. 13. Benson, P. M., Spear, K. E. and Pantano, C. G., Inter￾facial characterization of glass matrix/Nicalon SiC ®ber composites : a thermodynamic approach. Ceram. Eng. Sci. Proc., 1988, 9, 663±669. 14. Tredway, W. K., Prewo, K. M. and Pantano, C. G., Fiber matrix interfacial e€ects in carbon ®ber-reinforced glass matrix composites. Carbon, 1989, 27, 717±727. 15. Benson, P. M., Spear, K. E. and Pantano, C. G., Ther￾momechanical analyses of interface reactions in carbon- ®ber-reinforced glass matrix composites. Mater. Sci. Res., 1987, 21, 415±425. 16. Bianchi, V., Goursat, P., MeÂnessier, E., Sinkler, W. and Monthioux, M., C/YMAS compositesÐe€ects of the interface and the residual stresses on the rupture behavior. In Advanced Structural Fiber Composites, Advances in Science and Technology, Vol. 7, 1994, pp. 695±702. 17. Sinkler, W., Monthioux, M., Bianchi, V., Goursat, P. and MeÂnessier, E., Carbon ®ber-reinforced YMAS glass-cera￾mic matrix composites. II: Structural changes in the matrix with temperature. J. Europ. Ceram. Soc., 1995, submitted. 18. Parlier, M., Passilly, B. and Sudre O., CaracteÂrisation micromeÂcanique dees composites aÁ matrice ceÂramique aÁ l'aide de la technique de microindentation instrumenteeÂ. In Workshop II: Introduction of Ceramics in Aerospace Structural Composites, 76th Meeting of the Structures and Materials Panels, Anthalya, Turkey, 1993. 326 V. Bianchi et al