Journal of the European Cerami so ieis vier scien le limited Printed in Great Britain. All rights reserved PII:s0955-2219(97)00007 0955-221997s1700 Sol-Gel Control of Matrix Net-Shape Sintering in 3D Fibre reinforced Ceramic Matrix Composites Ph Colomban*& m. Wey ONERA, Direction dcs Matcriaux, BP 72, 92322 Chatillon Cedex france (Received 4 October 1996; revised version rcccivcd 26 November 1996 accepted 2 December 1996) Abstract matrix precursor must be incorporated through the voids between fibres (a few microns or less The origin of thermochemical degradation of 3D size). This is only possible by infiltration of liquid reinforced ceramic matrix composites has been or gaseous precursors for which the ceramic yield analysed hy dilatometry, pore-size distribution and is necessarily low, which implies porosity and electron microscopy and correlated to the cracking cracks. The latter method is used for the synthesis arising from the matrix shrinkage in an invariant of covalent bonded ceramics of a rather simple 3D fibre network. The matrix shrinkage of an alu- composition(e.g. C, SiC, BN, .)2 whereas the mina matrix composite has been delayed up to a former is more versatile(for SiC, Si, N4, various higher temperature (+200C) by post-infiltration multicomponent oxides, ...).3.4 The liquid pre with pure alkoxide which was subsequently in situ cursor is a slurry of submicron powder or a hydrolysed, polycondensed and pyrolysed. Mechan- metal-organic reagent(alkoxide, ester, polysil- ical strength of the composites has been increased oxane, polysilazane ng after pyrolysis four times by optimizing alkoxides. The tensile and to oxide, carbide nitride .. In both cases, there flexural mechanical strengths of these alumina is considerable shrinkage after pyrolysis which matrix composites are similar to those obtained produces new voids; however, the presence of the from SiC matrix composites prepared by the CVI woven fabrics can efficiently inhibit the coherent process using the same 3d carbon preform. c 1997 shrinkage of the matrix. In the case of one-or Elsevier Science Limited. two-dimensional reinforcements this dilemma coated fabrics.0 To solve this problem is more 1 Introduction complicated when a three-dimensional (3D)rcin- forcement is used The 3D interconnected array of The great weakness of lithic ceramics fibres segments the matrix, which cracks when trinsic inability to te mechanica sintering takes place. Thus, the methods based without brittle rupture because of their polycrys- on the slow condensation of gaseous precursor talline state and the nature of the chemical bonds Chemical Vapour Infiltration(CVi) is commonly existing in these compounds. The use of long, used. Ihis paper will demonstrate how the post woven ceramic fibres embedded in a refractory infiltration of a specified ceramic precursor--and ceramic matrix can result in a metastable compo- its in situ pyrolysis-between the primary infiI- site material exhibiting greater toughness through trated powder particles by pressure-assisted infil- a specific micromechanism at the fibre-matrix inter- tration can improve the mechanical strength of 3D face: the matrix cracking can be deflected dissociated fibre reinforced ceramic matrix composites and and even arrested when a propagating crack meets that liquid precursors offer an alternative method the reinforcing fibres and thus composite materials of preparing 3D reinforced ceramic matrix com can exhibit a fibrous, pseudo-plastic fracture posites(CMCs) One of the main problems in the preparation of Various sol gel precursors leading to inert ceramic matrix composites is to achieve a low refractory non-reacting interphase(Zr-i-propoxide open porosity in the matrix. The fibres must be as zirconia precursor, Al-butoxide as alumina pre- thoroughly embedded in the matrix and the cursor), reacting interphase (Ti-i- propoxide as rutile precursor, tetraethoxysilane (tEoS)as silica * Corresponding author, also at CNRS, LASIR, 2 rue Henri precursor)or bloating interphase(Al-Si ester, as Dunant. 94320 Thiais. france aluminosilicate precursor) have been considered 1475
Journal of the European Ceramic Society 17 (1997) 1475-1483 0 1997 Elsevier Science Limited Printed in Great Britain. All rights reserved PII: SO955-2219(97)00007-l 0955-2219/97/$17.00 Sol-Gel Control of Matrix Net-Shape Sintering in 3D Fibre Rei.nforced Ceramic Matrix Composites Ph. Colomban* & ML Wey ONERA, Direction des MattSriaux, BP 72, 92322 Chatillon Cedex, France (Received 4 October 1996; revised version received 26 November 1996; accepted 2 December 1996) Abstract The origin of thermochemical degradation of 30 reinforced ceramic matri.x composites has been analysed by dilatometry, pore-size distribution and electron microscopy and correlated to the cracking arising from the matrix shrinkage in an invariant 30 jibre network. The matrix shrinkage of an alumina matrix composite has been delayed up to a higher temperature (+2OO”C) by post-injiltration with pure alkoxide which was subsequently in situ hydrolysed, polycondensed and pyrolysed. Mechanical strength of the composites has been increased four times by optimizing alkoxides. The tensile and Jlexural mechanical strengths of these alumina matrix composites are similar to those obtained from Sic matrix composites prepared by the CVI process using the same 30 carbon preform. 0 1997 Elsevier Science Limited. 1 Introduction The great weakness of mo:nolithic ceramics is their intrinsic inability to tolerate mechanical damage without brittle rupture because of their polycrystalline state and the naturle of the chemical bonds existing in these compounds. The use of long, woven ceramic fibres embedded in a refractory ceramic matrix can result in a metastable composite material exhibiting greater toughness through a specific micromechanism at the fibre-matrix interface: the matrix cracking can be deflected, dissociated and even arrested when a propagating crack meets the reinforcing fibres and thus composite materials can exhibit a fibrous, pseudo-plastic fracture.’ One of the main problems in the preparation of ceramic matrix composites is to achieve a low open porosity in the matrix. The fibres must be thoroughly embedded in the matrix and the *Corresponding author, also at CNRS, LASIR, 2 rue Hem-i Dunant, 94320 Thiais, France. matrix precursor must be incorporated through the voids between fibres (a few microns or less, in size). This is only possible by infiltration of liquid or gaseous precursors for which the ceramic yield is necessarily low, which implies porosity and cracks. The latter method is used for the synthesis of covalent bonded ceramics of a rather simple composition (e.g. C, Sic, BN, ...)2 whereas the former is more versatile (for Sic, Si,N,, various multicomponent oxides, . . .).3,4 The liquid precursor is a slurry of submicron powder or a metal-organic reagent (alkoxide, ester, polysiloxane, polysilazane, . . .)3,4 leading after pyrolysis to oxide, carbide, nitride, . . . . In both cases, there is considerable shrinkage after pyrolysis which produces new voids; however, the presence of the woven fabrics can efficiently inhibit the coherent shrinkage of the matrix. In the case of one- or two-dimensional reinforcements, this dilemma is solved by the hot pressing of impregnated and coated fabricP To solve this problem is more complicated when a three-dimensional (3D) reinforcement is used. The 3D interconnected array of fibres segments the matrix, which cracks when sintering takes place. Thus, the methods based on the slow condensation of gaseous precursor (Chemical Vapour Infiltration (CVI) is commonly used. This paper will demonstrate how the postinfiltration of a specified ceramic precursor-and its in situ pyrolysis-between the primary infiltrated powder particles by pressure-assisted infiltration can improve the mechanical strength of 3D fibre reinforced ceramic matrix composites and that liquid precursors offer an alternative method of preparing 3D reinforced ceramic matrix composites (CMCs). Various sol-gel precursors leading to inert refractory non-reacting interphase (Zr-i-propoxide as zirconia precursor, Al-butoxide as alumina precursor), reacting interphase (Ti-i-propoxide as rutile precursor, tetraethoxysilane (TEOS) as silica precursor) or bloating interphase (Al-Si ester, as aluminosilicate precursor) have been considered in 1475
Ph Colomban, M. We association with a primary matrix made of infil- with the same kind of fibre(Novoltex" from SEP, trated alumina powder. Le hailan, 33165 Saint Medard en Jalles, france). The idea of using liquid ceramic precursors for This linkage is obtained by passing fibre thread the consolidation of a fibre preform originates fro through the sheets to form stitches. The fibre vol the processing route used for C/C composites ume fraction is -28%. The fibre diameter ranges (infiltration of pitch or of phenolic resin). Similar between 7 and 10 um. The 8-mm thick preform is routes have been tested for the preparation of Sic cut to form a 40 mm X 120 mm plate matrix composite using polysilane precursors. .8 Preparation of composite materials by infiltration 2.2 Powder and slurry of oxide porous bodies with organic metallic A submicron powder (AKP50 from Sumitomo) reagents has been recently discussed by Honeyman- consisting of spherical, well-crystallized a-alumina Calvin and Lange, but the case of oxide matrix particles, of mean diameter 0- 1-0-3 um and 3D fibre reinforced bodies has not been consid- specific area 10 m"g, is used as the matrix starting ered except for the experimental details given material. A 300 g/litre alumina slurry is prepared our patent. by dispersion in water with addition of nitric acid (pH=4), with the help of mechanical and ultra sonic stirring for 1 h. Aggregates larger than 2 Experimental Procedure 0.6 um are eliminated by overnight sedimentation Figure 1 shows a schematic of the two-step infil- Ultrasonic mixing is carried out for 15 min just tration process. The first step(stages 1 and 2)con- before use. The aggregate size distribution is sists in the preparation of a powder compact checked using a HORIBA CAPA 700 size analyser within the fibre preform by slip-cast infiltration (stage 1), the specimen being dried and the 2.3 Monoliths and composites strengthened by heating(stage 2). The second step The slurry is slip-cast under 2 MPa air pressure consists in the infiltration of the strengthened through the fibre preform plate using the tech- body with a liquid precursor(stage 3) that is nique previously described by Jamet et al converted into a gel by reaction with water (or A filter paper under the preform plate is used to diols)(stage 4) and then into a refractory phase retain the thinner particles inside the preform during pyrolysis(stage 2). The second step can be (Fig. 1). a drilled aluminium plate acts as a repeated many times mechanical support. Nevertheless, some of the first particles with mean diameter below 0-05 2.1 3D preform are not retained in the preform. a vacuum The carbon fibre preform consists of sheets of felted applied under the filter paper support to promote fibres which are sewn along the third dimension rapid slip-casting (<2 h). Then, the infiltrated 3 R江BE分587 TED IMMERSION Alkoxide(s) 2 PYROLYSS 2MPa Aluminum support 1000°C 3-10 Cycles 10m日 3D FIBER PREFORM Fig. 1. Flow diagram of two-step infiltration process, first using a pressure-assisted slip-casting infiltration device and second by in situ hydrolysis-polycondensation of a specihc interphase precursor
1476 Ph. Colomban, M. Wey association with a primary matrix made of infiltrated alumina powder. The idea of using liquid ceramic precursors for the consolidation of a fibre preform originates from the processing route used for C/C composites (infiltration of pitch or of phenolic resin). Similar routes have been tested for the preparation of Sic matrix composite using polysilane precursors.7,8 Preparation of composite materials by infiltration of oxide porous bodies with organic metallic reagents has been recently discussed by HoneymanCalvin and Lange, 9 but the case of oxide matrix 3D fibre reinforced bodies has not been considered except for the experimental details given in our patent.‘O 2 Experimental Procedure Figure 1 shows a schematic of the two-step infiltration process. The first step (stages 1 and 2) consists in the preparation of a powder compact within the fibre preform by slip-cast infiltration (stage l), the specimen being dried and then strengthened by heating (stage 2). The second step consists in the infiltration of the strengthened body with a liquid precursor (stage 3) that is converted into a gel by reaction with water (or diols) (stage 4) and then into a refractory phase during pyrolysis (stage 2). The second step can be repeated many times. 2.1 3D preform The carbon fibre preform consists of sheets of felted fibres which are sewn along the third dimension, 1 PRESSUREASSISTED m.TRA TION with the same kind of fibre (Novoltex@ from SEP, Le Haillan, 33165 Saint Medard en Jalles, France). This linkage is obtained by passing fibre thread through the sheets to form stitches. The fibre volume fraction is -28%. The fibre diameter ranges between 7 and 10 pm. The &mm thick preform is cut to form a 40 mm X 120 mm plate. 2.2 Powder and slurry A submicron powder (AKPSO from Sumitomo) consisting of spherical, well-crystallized a-alumina particles, of mean diameter 0.1-0.3 pm and specific area 10 m2/g, is used as the matrix starting material. A 300 g/litre alumina slurry is prepared by dispersion in water with addition of nitric acid (pH = 4), with the help of mechanical and ultrasonic stirring for 1 h. Aggregates larger than 0.6 pm are eliminated by overnight sedimentation. Ultrasonic mixing is carried out for 15 min just before use. The aggregate size distribution is checked using a HORIBA CAPA 700 size analyser. 2.3 Monoliths and composites The slurry is slip-cast under 2 MPa air pressure through the fibre preform plate using the technique previously described by Jamet et al.” A filter paper under the preform plate is used to retain the thinner particles inside the preform (Fig. 1). A drilled aluminium plate acts as a mechanical support. Nevertheless, some of the first particles with mean diameter below 0.05 pm are not retained in the preform. A vacuum is applied under the filter paper support to promote rapid slip-casting (~2 h). Then, the infiltrated 3 MMZRSION Akoxide (s) Gaz 2MPa +G __. ,., 1 [Filter _,.__” 3-N Cycles l~m$lAzkktY 3D FIBERPREFORM HYDROLYSISPOLYCONDElVSATION Fig. 1. Flow diagram of two-step infiltration process, first using a pressure-assisted slip-casting infiltration device and second by in situ hydrolysis-polycondensation of a specific interphase precursor
Sol-gel control of the matrix net-shape sintering preform is dried stepwise in a controlled humidity invariant fibre preform promotes the segmentation tmosphere at selected temperatures between 40 of the matrix. a typical example is shown in Fig. 2. and 120oC. The drying time is 15 h This phenomenon lowers the mechanical strength Monoliths are processed without the fibre pre- The improvement to the thermomechanical prop form plate, and the powder particles are retained erties can be achieved if (i) net-shape shrinkage is by the sole filter paper, Dried monoliths with 60% obtained, (ii) the porosity is decreased and (iii)the of the theoretical density for a-alumina can be mechanical strength of the bulk matrix is increased without the formation of a strong fibre- The infiltrated preform and monoliths are the matrix interface mally treated at selected temperatures between At temperatures ranging from 1000 to 1400C 1000 and 1400C in a reducing argon atmosphere the sintering of an alumina submicron powder takes for I h in a carbon resistance furnace place by solid-state diffusion at the particle con- tacts. The interposition of an inert second phase 2. 4 Microstructure and mechanical between particles would prevent, or reduce, the formation of the contact for solid-state diffusion Monolith shrinkage was recorded versus tempera- The interparticle voids could also affect shortening ture using an Adamel Lhomargy DI24 apparatus of the interparticle distance during sintering (Instrument SA, 91 Longjumeau, France) with an The high green density of the monolith sample alumina rod and support(heating/cooling rate:(60% of the theoretical density) indicates a highly 5C/min). The pore size and distribution are optimized packing of the AKP50 particles. The meritics Pore Sizer 9,, infiltration using a Micro- open porosity of monoliths thermally treated at mcasured by merc according to the astM 1000oC ranges from 34 to 36% and that of vol. 1201, C699-1983 notice. The validity of the posites with the 3D preform(carbon fibre volume measurement has been verified by calibration of fraction: 28%)ranges from 35 to 38%. This the mercury/composite surface angle by compari- cates that the aK P50 particle packing inside the son with the pore volume determined using N2 preform is rather similar to that in the monolith adsorption/desorption plots carried out with a Figure 3(a) compares the pore volume distribution Micromeritics ASAP2000 instrument Comparable in the monolith alumina sample with that in the results are obtained for a surface angle equal to composite heated at various temperatures. only 145, this value being intermediate between the one family of pores(mean diameter -005 um) is angle values commonly used for Hg/C(155%)and observed in the monolith alumina sample. this Hg/oxide(130%) interfaces. The open porosity has pore range corresponds well to the voids between been calculated according to Archimedes'method. adjacent AKP50 particles. It is noted that the pore The flexural strength was recorded by a three- volume ranges up to 0-01 um for the monolith point bending test using specimens of 40 mm X8 alumina sample heated below 1200oC, according mm x 2 mm at a cross-head speed of 0. 1 mm/min to its shrinkage On the other hand, three kinds of at room temperature and at 1200oc(1300c) porosity are observed in the composite under argon, with I h stabilization at the tempera- ture of measurement. The tensile strength was ( The first family, centred near 0-05-0.1 um, recorded at room temperature on a dumbbell is assigned to the interparticle voids as ecimen machined in a 120 mm x 40 mm observed in the monolith alumina sample mm composite using a 10 um deformation gaug The small shift toward high values indi cates that the packing is slightly disturbed by the 3d fibre network 3 Results and Discussion (ii a broad distribution from 0.2 to 10 um can be assigned to the voids between fibres 3.1 Matrix shrinkage and its influence on (infiltrated powder can be lacking fror mechanical properties bundles, as evidenced by microscopy) and Dilatometric traces of the slip cast monoliths from to voids between fibres and matrix particles AKP50 suspensions show that sintering starts the latter voids increasing with increase in at 1000 C and ends at 1450 C 2 (The bending thermal treatment temperature ultimate strength can reach 400 MPa for the (iia bimodal distribution between 10 and 20 monolith specimen sintered at 1400C). The nil um Is assigned to cracks originating from dilatation/shrinkage point is observed at1l00°C constricted shrinkage, as shown in Fig. 2 Consequently, microcracking and a fbre-matrix Increasing the sintering temperature led to nar gap are observed in the composite heated above rowing of the pore size in the monolith, resulting 1100C, and the presence of the 3d geometric from particle centre shortening and pore collapsing
Sol-gel control of the matrix net-shape sintering 1477 preform is dried stepwise in a controlled humidity atmosphere at selected temperatures between 40 and 120°C. The drying time is 15 h. Monoliths are processed without the fibre preform plate, and the powder particles are retained by the sole filter paper. Dried monoliths with 60% of the theoretical density for a-alumina can be achieved. The infiltrated preform and monoliths are thermally treated at selected temperatures between 1000 and 1400°C in a reducing argon atmosphere for 1 h in a carbon resistance furnace. 2.4 Microstructure and mechanical characterization Monolith shrinkage was recorded versus temperature using an Adamel Lhomargy D124 apparatus (Instrument SA, 91 Longjumeau, France) with an alumina rod and support (heating/cooling rate: SUmin). The pore size and distribution are measured by mercury infiltration using a Micromeritics Pore Sizer 9310 according to the ASTM vol. 1201, C699-1983 not:ice. The validity of the measurement has been verified by calibration of the mercury/composite surface angle by comparison with the pore volume determined using N, adsorption/desorption plaits carried out with a Micromeritics ASAP2000 instrument. Comparable results are obtained for a surface angle equal to 145”, this value being intermediate between the angle values commonly us.ed for Hg/C (155’) and Hg/oxide (130”) interfaces.. The open porosity has been calculated according to Archimedes’ method. The flexural strength was recorded by a threepoint bending test using specimens of 40 mm X 8 mm X 2 mm at a cross-hecad speed of 0.1 mm/min at room temperature and at 1200°C (1300°C) under argon, with 1 h stabilization at the temperature of measurement. The tensile strength was recorded at room temperature on a dumbbell specimen machined in a 120 mm X 40 mm X 8 mm composite using a 10 ,um deformation gauge. 3 Results and Discussion 3.1 Matrix shrinkage and its influence on mechanical properties Dilatometric traces of the slip cast monoliths from AKPSO suspensions show that sintering starts at 1000°C and ends at l,450°C.i2 (The bending ultimate strength can reach 400 MPa for the monolith specimen sintered at 1400°C). The nil dilatation/shrinkage point is observed at 1100°C. Consequently, microcracking and a fibre-matrix gap are observed in the composite heated above llOO”C, and the presence of the 3D geometric invariant fibre preform promotes the segmentation of the matrix. A typical example is shown in Fig. 2. This phenomenon lowers the mechanical strength. The improvement to the thermomechanical properties can be achieved if (i) net-shape shrinkage is obtained, (ii) the porosity is decreased and (iii) the mechanical strength of the bulk matrix is increased without the formation of a strong fibrematrix interface. At temperatures ranging from 1000 to 14OO”C, the sintering of an alumina submicron powder takes place by solid-state diffusion at the particle contacts. The interposition of an inert second phase between particles would prevent, or reduce, the formation of the contact for solid-state diffusion. The interparticle voids could also affect shortening of the interparticle distance during sintering. The high green density of the monolith sample (60% of the theoretical density) indicates a highly optimized packing of the AKPSO particles. The open porosity of monoliths thermally treated at 1000°C ranges from 34 to 36% and that of composites with the 3D preform (carbon fibre volume fraction: 28%) ranges from 35 to 38%. This indicates that the AKPSO particle packing inside the preform is rather similar to that in the monolith. Figure 3(a) compares the pore volume distribution in the monolith alumina sample with that in the composite heated at various temperatures. Only one family of pores (mean diameter -0.05 pm) is observed in the monolith alumina sample. This pore range corresponds well to the voids between adjacent AKPSO particles. It is noted that the pore volume ranges up to 0.01 pm for the monolith alumina sample heated below 12OO”C, according to its shrinkage. On the other hand, three kinds of porosity are observed in the composite: (i) The first family, centred near 0.0550.1 pm, is assigned to the interparticle voids as observed in the monolith alumina sample. The small shift toward high values indicates that the packing is slightly disturbed by the 3D fibre network. (ii) A broad distribution from 0.2 to 10 pm can be assigned to the voids between fibres (infiltrated powder can be lacking from bundles, as evidenced by microscopy) and to voids between fibres and matrix particles, the latter voids increasing with increase in thermal treatment temperature. (iii) A bimodal distribution between 10 and 20 pm is assigned to cracks originating from constricted shrinkage, as shown in Fig. 2. Increasing the sintering temperature led to narrowing of the pore size in the monolith, resulting from particle centre shortening and pore collapsing
which took place after sintering at 1300oC. The The low viscosity(0.5-1 poise)of the liquid same phenomenon occurred for the composite polymeric precursors makes it possible to infiltrate matrix but cracks dominated after sintering at them in porous ceramics and to convert them into 1300C due to the matrix segmentation imposed ceramics by 'polymerization'and then pyrolysis by the invariable fibre array: the matrix densifi- Hydrolytic polycondensation of alkoxides is well cation crcatcd cracks documented for the preparation of oxides 3-16 FIBER FIBER MATRIX MATRIX Sinte Inter-particle Fig. 2. Microphotographs of a 1200oC sir composite (a, bar: 20 um; a', bar: 100 um ). The microcracking induced by the competition between the matrix shrinkage he invariant 3D fibre network is sketched in(b); detail of the dense packing of primary oxide particles inter-particle voids available for alkoxide infiltration is shown 1300c 00010.01 P。 re diameter(μm Pore diameter【μm Fig. 3. Pore distribution in an alumina monolith (a)and in a 3D reinforced composite(b), versus sintering temperature
1478 Ph. Colomban, M. Wey which took place after sintering at 1300°C. The The low viscosity (-0.5-l poise) of the liquid same phenomenon occurred for the composite polymeric precursors” makes it possible to infiltrate matrix but cracks dominated after sintering at them in porous ceramics and to convert them into 1300°C due to the matrix segmentation imposed ceramics by ‘polymerization’ and then pyrolysis. by the invariable fibre array: the matrix densifi- Hydrolytic polycondensation of alkoxides is well cation created cracks. documented for the preparation of oxides.‘3m’6 Inteqmticle voids Fig. 2. Microphotographs of a 1200°C sintered composite (a, bar: 20 pm; a’, bar: 100 pm). The microcracking induced by the competition between the matrix shrinkage and the invariant 3D fibre network is sketched in (b); detail of the dense packing of primary oxide particles with inter-particle voids available for alkoxide infiltration is shown. Pore diametwfpm) Pore diameter lpm) Fig. 3. Pore distribution in an alumina monolith (a) and in a 3D reinforced composite (b), versus sintering temperature
Sol-gel control of the matrix net-shape sintering Among commercially available reagents, we have at 150oC leads to gels with the following typical selected titanium butoxide (Ti(OCHs)4), aluminium formula: TiO,5 (OH).5H,O, ZrO198(OHDo04 4H2O S-butoxide(Al(oC4H53, TEOS (Si(OC, Hs)4), alu- and Al,O2(OH)0-6.5H2O minium-silicon ester (OC4H5)x-Al-O-Si(oC3 H,)3) and zirconium i-propoxide(zr(oC3 H,)4); the high 3.2 Alkoxide infiltration and pyrolysis melting temperature of the corresponding oxides The infiltration has been detailed in our patent may exclude a strong reaction with alumina below Samples(i.e. the strengthened powder compacts, 1400C. Furthermore, their viscosity can be adjusted stage 2 of Fig. 1) are carefully dried at 200oC, at about 0.7 poise by heating between 40 and then immersed in liquid alkoxides(under atmo 80C. 0 Hydrolysis-polycondensation and drying spheric pressure or I MPa)for 30 min with the help of ult immersed in boiling water at 120C for 3 h(stage 4) then finally dried at -120C and pyrolysed at various temperatures from 1000 to 1400C (stage 2). Typically 2-3% of the porosity is filled by oxide after one cycle. 2 Figure 4 shows an example of the deposition on the fibre, which was obtained from the alkoxide hydrolysis-polycon densation inside the composite. The small agglom erates(mean diameter so.5 um) after pyrolysis are visible. Analysis of the zirconium concentration across the monolith thickness after four cycles of infiltration and pyrolysis reveals a higher concen tration zone in the vicinity of the monolith surface. This gives a laminar behaviour to the monolith(Fig. 5). The location of high zirconia P 4. Zirconia deposit at surface of a C fibre( diameter -g content regions after infiltration seems to be posite after eight infiltration cycles or -0ooc heat-treated com- )observed in the fracture of a 120 related to the pore heterogeneity resulting fr rom slip-casting Fig. 5. Back-scattered electron micrograph of an alumina monolith after four cycles of zirconium i-propoxide infiltration-hydrolysis firing and final heat treatment at 1400 C in air(a, bar: 10 mm). Detail at larger magnitude is given in(b, bar: 20 um)and(c 600 nm). a SEM micrograph of the frontier between the dense core and the porous contour is given in(d, bar: 4oAn'bar *This reagent must be handled in a glove box(s10 pp. m. H2O)and careful drying of the porous materials to be infiltrated is required
Sol-gel control of the matrix net-shape sintering Among commercially available reagents, we have selected titanium butoxide (Ti(OC4H5)J, aluminium s-butoxide (Al(OC,H& *, TEOS (Si(OC,H,),), aluminium-silicon ester ((OC,+H,)2-Al-O-Si(OC3H5)3) and zirconium i-propoxide (Zr(OC,H,),); the high melting temperature of the corresponding oxides may exclude a strong reaction with alumina below 1400°C. Furthermore, their viscosity can be adjusted at about 0.7 poise by heating between 40 and 80”C.‘” Hydrolysis-polycondensation and drying Fig. 4. Zirconia deposit at surface of a C fibre (diameter -9 pm) observed in the fracture of a 1200°C heat-treated composite after eight infiltration cycles of zirconium i-propoxide. at 150°C leads to gels with formula: TiO,.,(OH).SH,O, and Al,0,.,(OH)o.,.5H20.17~18 1479 the following typical Z~,.,8(OH)o.,.4H20 3.2 Alkoxide infiltration and pyrolysis The infiltration has been detailed in our patent.” Samples (i.e. the strengthened powder compacts, stage 2 of Fig. 1) are carefully dried at 2OO”C, then immersed in liquid alkoxides (under atmospheric pressure or 1 MPa) for 30 min with the help of ultrasonic stirring (stage 3); next they are immersed in boiling water at 120°C for 3 h (stage 4) then finally dried at -120°C and pyrolysed at various temperatures from 1000 to 1400°C (stage 2). Typically 2-3% of the porosity is filled by oxide after one cycle.‘* Figure 4 shows an example of the deposition on the fibre, which was obtained from the alkoxide hydrolysis-polycondensation inside the composite. The small agglomerates (mean diameter 10.5 pm) after pyrolysis are visible. Analysis of the zirconium concentration across the monolith thickness after four cycles of infiltration and pyrolysis reveals a higher concentration zone in the vicinity of the monolith surface. This gives a laminar behaviour to the monolith (Fig. 5). The location of high zirconia content regions after infiltration seems to be related to the pore heterogeneity resulting from slip-casting. Fig. 5. Back-scattered electron mbcrograph of an alumina monolith after four cycles of zirconium i-propoxide infiltration-hydrolysis firing and final heat treatment at 1400°C in air (a, bar: 10 mm). Detail at larger magnitude is given in (b, bar: 20 pm) and (c, bar: 500 nm). A SEM micrograph of the frontier between the dense core and the porous contour is given in (d, bar: 200 pm). *This reagent must be handled in a glove box (510 p.p.m. H,O) and careful drying of the porous materials to be infiltrated is required
Ph Colomban, M. Wey Examination of the monolith shows a poor infiltrated with a zirconia gel first increases up to egion below the upper surface. After 10 cycles of 1200C and then decreases(Fig. 6(a)). This agrees infiltration-thermal treatment at 1000 C, the open with the lack of shrinkage up to 1300C. the infil porosity is typically 24% using simple capillarity tration yield is a maximum for materials infil- intrusion and 20% using pressure assisted infiltra- trated with aluminium s-butoxide pyrolysed at tion. Full densification is obtained after a 1400c 1200 C ( Fig. 6(b)). On the other hand, the highest thermal treatment for 1 h. Figure 5(c) shows that yield is observed for materials infiltrated with elting from the gel pyrolysis zirconium propoxide pyrolysed below 1100.C are homogeneously dispersed and form particles This phenomenon can be related to the high reac with 0.2 um mean diameter which corresponds tivity of aluminium s-butoxide versus water. The well to the size of interparticle voids. Consequently surface of alumina porous bodies heated below the flexural strength measured at room tempera- 1200oC easily retains water molecules. Uncon ture reaches 550 MPa trolled hydrolysis leads to pore clogging and hence The evolution of the pore volume distribution as decreases the infiltration yield. Comparison of the function of size for various thermal treatments dilatometric traces of the same monolith shows shows that the mean pore size of a monolith post that the linear shrinkage after 1400oC thermal treatment is reduced from 12 to 5% after only one 1300120 cycle of infiltration with zirconium i-propoxide btained 1300C. Figure 7 gives a comparison between the pore volume distribution measured for a compos- ite without and after eight cycles of zirconium 1400c i-propoxide infiltration-hydrolysis-polycondensa tions and 1200 C firing The lack of voids gener ated by matrix cracking in the post-infil pore diameter (pm) o1 alumina matrix composite is straightforward in the pore diameter distribution(Fig. 7) 3.3 Optimization of the post-infiltrated interphase Post- infiltration with a liquid precursor of a very refractory composition allows the prevention of 4.00 interparticle diffusion and hence matrix shrinkage by the formation of an interphase acting as a 200 diffusion barrier. However, this barrier cannot continue because of the low yield of ceramization 1200 Fig. 4 shows cracks between zirconia aggregates Pyrolysis temperature") (0.05 mm)and between scales of aggregates Additional infiltration must be made with a pre- cursor which strengthens the interparticle bond Attempts have been made with titanium alkoxide infiltration. However, titanium ethoxide(propox ide, butoxide,.)infiltration activates the sintering of an alumina monolith and segmentation of the matrix occurs in the composite. Consequently the mechanical properties remain poor. The net-shape temperature (nil expansion/shrinkage)on the dilatometric trace occurs at 1050C and the final linear shrinkage of 1400oC heat-treated rutile pre 1000 13001400 cursor infiltrated monolith reaches 14% instead of PYROLYSIS TEM PERATURElC) 2% for a pure AKP50 monolith. The origin of Fig. 6. Evolution of pore diameter distribution as a function this sintering activation can be found in the easy of sintering temperature for a monolith after four cycles diffusion of titanium in alumina zirconium i-propoxide post-infiltration and in situ hydrolysis- Figure 8 shows the dilatometric trace of alumino- polycondensation (a); cumulative porosity variation for an alumina monolith post-infiltrated with aluminum s-butoxide silicate gel prepared by hydrolysis-polycondensa- (b) and zirconium propoxide()and pyrolysed at various tion of aluminium-silicon ester in vacuo. Marked temperatures( 1: first cycle, 2: second cycle, 3 third cycle, expansion occurs above 1200.C. This phenomenon
1480 Ph. Colomban, A4. Wey Examination of the monolith shows a porous region below the upper surface. After 10 cycles of infiltration-thermal treatment at lOOO”C, the open porosity is typically 24% using simple capillarity intrusion and 20% using pressure-assisted infiltration. Full densification is obtained after a 1400°C thermal treatment for 1 h. Figure 5(c) shows that the oxide particles resulting from the gel pyrolysis are homogeneously dispersed and form particles with O-2 pm mean diameter which corresponds well to the size of interparticle voids. Consequently the flexural strength measured at room temperature reaches 550 MPa. The evolution of the pore volume distribution as a function of size for various thermal treatments shows that the mean pore size of a monolith postPore diameter (pm) o+--- ’ 1 I I # I 1000 1200 1400 Pyrolysis temperature?C) C) ‘T Q 1000 1100 1200 1300 1400 PYROLYSIS TEMPERATUREW) Fig. 6. Evolution of pore diameter distribution as a function of sintering temperature for a monolith after four cycles of zirconium i-propoxide post-infiltration and in situ hydrolysispolycondensation (a); cumulative porosity variation for an alumina monolith post-infiltrated with aluminum s-butoxide (b) and zirconium propoxide (c) and pyrolysed at various temperatures (1: first cycle, 2: second cycle, 3: third cycle, . ..). infiltrated with a zirconia gel first increases up to 1200°C and then decreases (Fig. 6(a)). This agrees with the lack of shrinkage up to 1300°C. The infiltration yield is a maximum for materials infiltrated with aluminium s-butoxide pyrolysed at 1200°C (Fig. 6(b)). On the other hand, the highest yield is observed for materials infiltrated with zirconium propoxide pyrolysed below 1100°C. This phenomenon can be related to the high reactivity of aluminium s-butoxide versus water. The surface of alumina porous bodies heated below 1200°C easily retains water molecules. Uncontrolled hydrolysis leads to pore clogging and hence decreases the infiltration yield. Comparison of the dilatometric traces of the same monolith shows that the linear shrinkage after 1400°C thermal treatment is reduced from 12 to 5% after only one cycle of infiltration with zirconium i-propoxide. A net-shape consolidation is obtained up to 1300°C. Figure 7 gives a comparison between the pore volume distribution measured for a composite without and after eight cycles of zirconium i-propoxide infiltration-hydrolysis-polycondensations and 1200°C firing. The lack of voids generated by matrix cracking in the post-infiltrated alumina matrix composite is straightforward in the pore diameter distribution (Fig. 7). 3.3 Optimization of the post-infiltrated interphase precursor Post-infiltration with a liquid precursor of a very refractory composition allows the prevention of interparticle diffusion and hence matrix shrinkage by the formation of an interphase acting as a diffusion barrier. However, this barrier cannot continue because of the low yield of ceramization: Fig. 4 shows cracks between zirconia aggregates (co.05 mm) and between scales of aggregates. Additional infiltration must be made with a precursor which strengthens the interparticle bond. Attempts have been made with titanium alkoxide infiltration. However, titanium ethoxide (propoxide, butoxide, . ..) infiltration activates the sintering of an alumina monolith and segmentation of the matrix occurs in the composite. Consequently the mechanical properties remain poor. The net-shape temperature (nil expansion/shrinkage) on the dilatometric trace occurs at 1050°C and the final linear shrinkage of 1400°C heat-treated rutile precursor infiltrated monolith reaches 14% instead of 12% for a pure AKPSO monolith. The origin of this sintering activation can be found in the easy diffusion of titanium in alumina. Figure 8 shows the dilatometric trace of aluminosilicate gel prepared by hydrolysis-polycondensation of aluminium-silicon ester in vacua. Marked expansion occurs above 1200°C. This phenomenon
Sol-gel control of the matrix net-shape sintering 1481 (22%) cracks inter-fibre and libre particle pores m6A/mh个 Fig. 7. Comparison of pore diameter distribution for a composite without(dotted line) and with eight cycles of zircon on (solid line). Thermal treatment was mac 1200C. Assignment of the different kinds of pore is given 4 Zr+Alosi 8Z c/A203 0 500 1500 T(°c) d 100lb) 1200 Fig. 8. Shrinkage expansion of aluminosilicate xerogel from aluminium-silicon ester heated in vacuum reducing atmosph 1300 which is related to gas evolution(H,O and Co owing to the final dehydroxylation and to the oxi- dation of carbon residues, is observed when 4 Zr +AlOSi simultaneously, the viscosity is lowered. 9 post- infiltration with this precursor may promote good , interparticle bonding without shrinkage Strain(%) 3. 4 Mechanical properties Fig 9. Room temperature flexural stress-strain traces(a) for Figure 9 shows typical fiexural stress strain plot a CAlO,cor recorded for composites before and after post- polycondensation (HP) and 1000oC firing(8Zr)and after four infiltration. a significant increase in the strength is cycles of zirconium-propoxide post-infiltration and HP ane observed at temperatures up to 1300C. The 000C firing with a subsequent infiltration cycle using alu ultimate flexural strengths measured at room tem minium-silicon ester(4Zr+AlOSi. The stress-strain traces recorded for this last sample sintered at various temperatures perature,1200° c and I300° C are compared in
Sol-gel control of the matrix net-shape sintering 1481 $ interparticle I: (35%) I\ : : ; : : 1 j j i micropores 1 it J ! : i : 1200°C grains I cracks l l :: ( ‘: : [ i : : : Pore diameter (pm) Fig. 7. Comparison of pore diameter distribution for a composite without (dotted line) and with eight cycles of zirconium i-propoxide post-infiltration and subsequent in situ hydrolysis-polycondensation (solid line). Thermal treatment was made at 1200°C. Assignment of the different kinds of pore is given. 500 1000 1500 T (“C 1 Fig. 8. Shrinkage expansion of aluminosilicate xerogel from ahtminium-silicon ester heated in vacuum reducing atmosphere. which is related to gas evolution (H,O and CO) owing to the final dehydroxylation and to the oxidation of carbon residues, is observed when, simultaneously, the viscosity is lowered.19 Postinfiltration with this precursor may promote good interparticle bonding without shrinkage. 3.4 Mechanical properties Figure 9 shows typical flexural stress-strain plots recorded for composites before and after postinfiltration. A significant increase in the strength is observed at temperatures up to 1300°C. The ultimate flexural strengths measured at room temperature, 1200°C and 1300°C are compared in looa) dzr+AIOSi ..-: . . . . . . . ..” ,.- : 1 2 St rain(%) Fig. 9. Room temperature flexural stress-strain traces (a) for a C/A&O3 composite before and after eight cycles of zirconium i-propoxide post-infiltration, in situ hydrolysispolycondensation (HP) and 1000°C firing (8Zr) and after four cycles of zirconium-propoxide post-infiltration and HP and 1000°C firing with a subsequent infiltration cycle using aluminium-silicon ester (4Zr+AlOSi). The stress-strain traces recorded for this last sample sintered at various temperatures are given in (b)
1482 Ph. Colomban, M. Wey Fig. 10(a), and the use of four zirconium propox- might not be sensitive to the thermal cycling, but ide infiltrations followed by one infiltration with the low rigidity of their matrix prevents their use aluminium-silicon ester appeared an effective under tensile-compressive cycling as established combination both for the fabrication of the com for the SiC matrix composit posites(handling in air)and for strength. It is post-infiltrated oxide matrix composite and CVi noticed that the strength at room temperature is Sic matrix composite exhibit the same mechanical ower than that at high temperatures. Self-healing strengths could occur at high temperature between the aluminosilicate interphases derived from the(alu minium) silicon alkoxide precursor at high tem- 4 Conclusions peratures. A typical tensile stress-strain plot recorded at room temperature is given in This novel processing route using the liquid infil- b)and the hysteresis indicates the load tration cycles, i.e. first using a suspension of a ing-unloading cycle. The same phenomenon has submicron alumina powder, then using an alk- been observed for SiC matrix composites prepared oxide (or any liquid precursor) offers several using the pyrolysis of infiltrated polysilazane pre- advantages: controlling the matrix shrinkage cursor in the 3D composite. B, I9 On the other hand, during sintering to avoid matrix segmentation in the Sic matrix composites processed using the 3D preform, controlling the interparticle chemical CVI route do not exhibit such hysteresis. The diff- bonding to strengthen the interparticle bonding erence can be related to the meso/microporosity of without shrinkage and simplifying the process the matrix in the composites derived from pyrol posites have a ysed metal-organic reagent from which the solids higher strength at high temperatures is believed to yield is low(25-30% for alkoxides). The homo- be the reaction between the aluminosilicate inter geneity of this meso/micro-porosity can help to phase derived from the alkoxide by the last post ccommodate, by microcracking, the thermal infiltration and the primary alumina matrix, the stress from thermal expansion mismatch but low- matrix shrinkage being simultaneously hindered ers the Youngs modulus. For instance, the tensile by the initial post-infiltration of a zirconia inter Young's modulus of post-infiltrated alumina granular pha matrix composites, prepared from the same 3D fibre preform, ranges between 20 and 30 GPa whereas that of the SiC matrix composite pro- Acknowledgement cessed by the CVI route is 75 GPa. Thus, the mechanical properties of CVI processed compo- Thanks are due to the Societe Europeenne de sites are significantly sensitive to thermal cycling. Propulsion(SEP), Le Haillan, 33165 Saint Medard Composites prepared from liquid oxide precursor en Jalles, for supporting M. Wey financially D 4Al+4Zr 4 Zr+ AlOSi 4 Zr +AlOS 3⊙ 60}1000 8o/ 4 p4zr+TEoS. 401 A293 1000 1200 T Fig. 10. Room-temperature flexural strength versus deformation of a composite before( CAlo, )and after post-infiltration and in situ hydrolysis-polycondensation and subsequent sintering at various temperatures: four infiltrations with zirconium i-propoxide (4Zr), with additional TEOS (4Zr+TEOS) or aluminium-silicon ester (4Zr+AlOSi) post-infiltration or alternatively, zirconium i- propoxide and aluminium (4A1+47r)post -infiltrations have been suc ively made. Black circles correspond recorded at the temperatur y the abscissa, in argon atmosphere, after one hour stabilization at high temperature for 4Zr+AlOSi post-infiltrated es. The corresponding room-temperature tensile strain-stress plot is given for a 4Zr+ AlOSi 1000.C(three loading/unloading cycles have been made before the rupture)
1482 Ph. Colomban, A4. Wey Fig. 10(a), and the use of four zirconium propoxide infiltrations followed by one infiltration with aluminium-silicon ester appeared an effective combination both for the fabrication of the composites (handling in air) and for strength. It is noticed that the strength at room temperature is lower than that at high temperatures. Self-healing could occur at high temperature between the aluminosilicate interphases derived from the (aluminium) silicon alkoxide precursor at high temperatures. A typical tensile stress-strain plot recorded at room temperature is given in Fig. 10(b) and the hysteresis indicates the loading-unloading cycle. The same phenomenon has been observed for SIC matrix composites prepared using the pyrolysis of infiltrated polysilazane precursor in the 3D composite.8,‘g On the other hand, the Sic matrix composites processed using the CVI route do not exhibit such hysteresis. The difference can be related to the meso/microporosity of the matrix in the composites derived from pyrolysed metal-organic reagent from which the solids yield is low (-25-30% for alkoxides). The homogeneity of this mesa/micro-porosity can help to accommodate, by microcracking, the thermal stress from thermal expansion mismatch but lowers the Young’s modulus. For instance, the tensile Young’s modulus of post-infiltrated alumina matrix composites, prepared from the same 3D fibre preform, ranges between 20 and 30 GPa, whereas that of the SIC matrix composite processed by the CVI route is 75 GPa.20 Thus, the mechanical properties of CVI processed composites are significantly sensitive to thermal cycling. Composites prepared from liquid oxide precursor 160 F 0 4A1+4Zr a 720. : ; 60 1000 1200 VW might not be sensitive to the thermal cycling, but the low rigidity of their matrix prevents their use under tensile-compressive cycling as established for the Sic matrix composites.20 Nevertheless, post-infiltrated oxide matrix composite and CVI SIC matrix composite exhibit the same mechanical strengths. 4 Conclusions This novel processing route using the liquid infiltration cycles, i.e. first using a suspension of a submicron alumina powder, then using an alkoxide (or any liquid precursor) offers several advantages: controlling the matrix shrinkage during sintering to avoid matrix segmentation in 3D preform, controlling the interparticle chemical bonding to strengthen the interparticle bonding without shrinkage and simplifying the process. The mechanism whereby composites have a higher strength at high temperatures is believed to be the reaction between the aluminosilicate interphase derived from the alkoxide by the last postinfiltration and the primary alumina matrix, the matrix shrinkage being simultaneously hindered by the initial post-infiltration of a zirconia intergranular phase. Acknowledgement Thanks are due to the SociCtC Europeenne de Propulsion (SEP), Le Haillan, 33165 Saint MCdard en Jalles, for supporting M. Wey financially. 80 b 4 Zr +AIOSi GO_ tooo”c .2 .4 .6 Strain(o& Fig. 10. Room-temperature flexural strength versus deformation of a composite before (CAl,Os) and after post-infiltration and in situ hydrolysis-polycondensation and subsequent sintering at various temperatures: four infiltrations with zirconium i-propoxide (4Zr), with additional TEOS (4Zr+TEOS) or ahuninium-silicon ester (4Zr+AlOSi) post-infiltration or alternatively, zirconium ipropoxide and aluminium s-butoxide (4A1+4Zr) post-infiltrations have been successively made. Black circles correspond to data recorded at the temperature given by the abscissa, in argon atmosphere, after one hour stabilization at high temperature for 4Zr+AlOSi post-infiltrated composites. The corresponding room-temperature tensile strain-stress plot is given for a 4Zr+AlOSi composite smtered at 1000°C (three loading/unloading cycles have been made before the rupture)
Sol-gel control of the matrix net-shape sintering 1483 References ening via compositional grading, grain size control and ransformation toughening. J. Am. Ceram. Soc., 199 l. Aveston,J, Cooper, G. A. and Kelly, A, In Proc. Conf 79(7,1810-181 on the Properties of Fiber Composites, National Physical 10. Colomban ey, M. and Parlier, M, Procede Lab, 4 November 1971. London. IPC Sci. and Tech d'elaboration d'un material ceramique par infiltration Press Ltd, Guilford, Surrey, 1971, pp. 15-26 d'un precurser dans preux ceramique 2. Naslain, R. and langlais, F, cvd processing of ONERA French Patent No. 2713222(9/6/1996 ceramic-ceramic composite materials In Tailoring Multi- 11. Jamet, J, Demange, D. and Loubeau, J, Nouveau phase and Composite Ceramics, Mat Sci. Res, Vol. 20 materiaux composites alumine-alumine a rupture forte- 145-164 Bernhart, G. A, Dauchier, M ment et leur preparation. oNERa French Patent No.2526785(l8/111983) SiC/SiC composite ceramics. Am. Ceram. 12. Wey, M. and Colomban, Ph, Densifier sans retrait:une Bw,1986,65(2),336338 necessite pour Optimisation des proprietes mecaniques de 4. Colomban, Ph, Process for fabricating a ceramic matrix composites ceramique 3D satisfaits par la polymerisation composite incorporating woven bers and materials with in situ. Proc. JNC10, /Demes Journees Nationales sur le different compositions and properties in the same com- Composites, Paris, 29-31 1996,ed.D. and hnOl.,1995,105/6,93-96 J. Vautrin. AMAC. Pa 5. Colomban, Ph. Bruneton. E. ,vol.3,pp.1133-1142 agrange, J. L. and 13. Klein, L. C.(ed ) Sol-gel Publ. Park Sol-gel mullite matrix- SiC and -mullite Ridge, N.J., 1988 2D woven fabric composites with or without zirconia 14. Colomban, Ph, Gel technology in ceramics, glass-ceramics containing interphase. Elaboration and properties, J. Eur. and ceramic-ceramic composites. Ceramics Int, 1989, 15 Ceramic Soc,1996,16(2),301-314. 23-50. 6. Mouchon, E and Colomban, Ph, Oxide ceramic matrix- 15. Pierre, A. C, Introduction aux Procedes Sol-Gel Editions ative fracture behavior. Composites, 1995, 26, 175-1022a oxide fibers woven fabric composites exhibiting diss Septima, Paris, 1992 16. Segal, D, Chemical Synthesis of Advanced Ceramic 7. Parlier, M, bouillon, E, Muller, C, Bloch, B Materials. Cambridge University Press, Cambridge, 1989 Noireaux, P. and Jamet, J, Procede d'elaboration d'un 17. Bruneton, E and Colomban, Ph,, Influence of hydrolysis materia composite ceramique fibres-matrice et materia conditions on crystallisation, phase transition and sinter composite ob par ce procede. oNEra French ing of zirconia prepared by alkoxide hydrolysis. J. No Patent No.8916918(20/12/1989) rystalline Solid, 1992, 147&148, 201-205 8. Sudre, O, Parlier, M. and Bouillon, E, Comparative 18. Colomban, Ph and Vendange, V, Sintering of alumina mechanical evaluation of two 2.50 C/SiC composites and mullite prepared by slow hydrolysis of alkoxides: the role of the protonic species and of pore topology J. Non- powder/polymer injection route. Proc. 2nd Int. Conj Crystalline Solid, 1992, 147&148, 245-250 on High-Temperature Ceramic Matrix Compost 19. Bruneton, E, Bigarre, J, Michel, D and Colomban, Ph HT-CMCIL, Santa-Barbara, 21-24 Aug. 1995, Vol. I Heterogeneity, nucleation, shrinkage and bloating in Ceram. Trans. 58. ed,R. Naslain and A. G. Evans. solgel glass-ceramics. J. Mater. Sci., 32, in press Am. Ceram Soc., Westerville, 1995, pp. 13-18 20. Parlier, M. and Colomban, Ph, Composites a matrice 9. Honeyman-Calvin, P. and Lange, F. F, Infiltration of ramique pour applications thermostructurales, La porous alumina bodies with solution precursors: strength Recherche aerospatiale, 1996, 5/6, 457-469
Sol-gel control of the matrix net-shape sintering 1483 References 1. Aveston, J., Cooper, G. A. #and Kelly, A., In Proc. Conf on the Properties of Fiber Composites, National Physical Lab., 4 November 1971, London. IPC Sci. and Tech. Press Ltd, Guilford, Surrey, 1971, pp. 15-26. 2. Naslain, R. and Langlais, F., CVD processing of ceramic-ceramic composite materials. In Tailoring Muftiphase and Composite Ceramics, Mat. Sci. Res., Vol. 20. Plenum Press, New York, 1!)8$, pp. 145-164. 3. Lamicq, P. J., Bernhart, G. A., Dauchier, M. M. and Mace, J. G., Sic/SIC composite ceramics. Am. Ceram. Bull., 1986, 65(2), 336338. 4. Colomban, Ph., Process for fabricating a ceramic matrix composite incorporating woven fibers and materials with different compositions and properties in the same composite. Muter. Technof., 1995, 10(5/6), 93-96. 5. Colomban, Ph., Bruneton: E., Lagrange, J. L. and Mouchon, E., Sol-gel mullite matrix-Sic and -mullite 2D woven fabric composites with or without zirconia containing interphase. Elaboration and properties, J. Eur. Ceramic Sot., 1996, 16(2), 301-314. 6. Mouchon, E. and Colomban, Ph., Oxide ceramic matrixoxide fibers woven fabric composites exhibiting dissipative fracture behavior. Composites, 1995, 26, 175-182. 7. Parlier, M., Bouillon, E., Muller, C., Bloch, B., Noireaux, P. and Jamet, J., Pro&de d’elaboration dun materiau composite ceramique fibres-matrice et materiau composite obtenu par ce pro&de. ONERA French Patent No. 8916918 (200211989). 8. Sudre, O., Parlier, M. and1 Bouillon, E., Comparative mechanical evaluation of two 2.50 C/Sic composites processed via chemical vapor infiltration and powder/polymer injection route. Proc. 2nd Znt. Co@ on High-Temperature Ceramic Matrix Composites HT-CMCZZ, Santa-Barbara, 21-24 Aug. 1995, Vol. I, Ceram. Trans. 58, ed. R. Naslain and A. G. Evans. Am. Ceram. Sot., Westerville, 1995, pp. 13-18. 9. Honeyman-Calvin, P. and Lange, F. F., Infiltration of porous alumina bodies with solution precursors: strengthening via compositional grading, grain size control and transformation toughening. J. Am. Ceram. Sot., 1996, 79(7), 1810-1814. 10. Colomban, Ph., Wey, M. and Parlier, M., ProddC d’elaboration d’un mat&au ceramique par infiltration d’un precurseur dans un support poreux dramique. ONERA French Patent No. 2713222 (9/6/1996). 11. Jamet, J., Demange, D. and Loubeau, J., Nouveaux materiaux composites alumine-alumine a rupture fortement et leur preparation. ONERA French Patent No. 2526785 (18/l l/1983). 12. Wey, M. and Colomban, Ph., Densifier sans retrait: une necessite pour l’optimisation des proprietes mecaniques de composites dramiques 3D satisfaits par la polymerisation in situ. Proc. JNCIO. 1Oemes Journees Nationales sur les Composites, Paris, 29-31 Cktobre 1996, ed. D. Baptiste and J. Vautrin. AMAC, Paris, 1996, Vol. 3, pp. 1133-1142. 13. Klein, L .C. (ed.), Sol-gel Technology. Noyes Publ., Park Ridge, N.J., 1988. 14. Colomban, Ph., Gel technology in ceramics, glassceramics and ceramic-ceramic composites. Ceramics Znt., 1989, 15, 23-50. 15. Pierre, A. C., Introduction aux Procedes Sol-Gel. Editions Septima, Paris, 1992. 16. Segal, D., Chemical Synthesis of Advanced Ceramic Materials. Cambridge University Press, Cambridge, 1989. 17. Bruneton, E. and Colomban, Ph., Influence of hydrolysis conditions on crystallisation, phase transition and sintering of zirconia prepared by alkoxide hydrolysis. J. NonCrystalline Solid, 1992, 147&148, 201-205. 18. Colomban, Ph. and Vendange, V., Sintering of alumina and mullite prepared by slow hydrolysis of alkoxides: the role of the protonic species and of pore topology. J. Noncrystalline Solid, 1992, 1478~148, 245-250. 19. Bruneton, E., Bigarre, J., Michel, D. and Colomban, Ph., Heterogeneity, nucleation, shrinkage and bloating in sol-gel glass-ceramics. J. Muter. Sci., 32, in press. 20. Parlier, M. and Colomban, Ph., Composites a matrice dramique pour applications thermostructurales. La Recherche Aerospatiale, 1996, 516, 457469