《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_porous-sIC

Availableonlineatwww.sciencedirect.com DIRECT E噩≈3S SEVIER Journal of the European Ceramic Society 25(2005)589-597 www.elsevier.com/locate/jeurceramsoc Mechanical properties and mechanical behaviour of Sic dense-porous laminates C. Reynaud a F. Thevenot a T. Chartier b, *.J.-L. Besson b Dept. Ceramiques Speciales, E.N.S. des Mines de Saint-Etienne, 42023 Saint-Etienne, france b SPCTS, U.M.R. C.N.R.S. 6638, E.N.S. de Ceramiques Industrielles, 87065 Limoges, france Received 3 December 2003; received in revised form 23 February 2004; accepted 28 February 2004 Available online 15 June 2004 Abstract Porous laminar materials and alternate laminates of silicon carbide dense and porous layers have been elaborated by tape casting and liquid phase sintering processing. Porosity was introduced by incorporation of pore forming agents(corn starch or graphite platelets) in the slurry. Homogeneous distributions of porosity have been obtained for both monolithic and composite laminates. The microstructure of the SiC matrix was equiaxed and was not affected by the porosity. The porosity(P) dependence of Young s modulus(e), modulus of rupture(or), toughness (Kic)and fracture energy(Gic)was found to be well described on the entire range of porosity by relations of the form Xo(1-P)x proposed by Wagh et al. from a model that takes into account the tortuosity of the porosity In the case of our materials, mE= 2.7, mog Kic= mE +0.5 and mGIc =mE+ 1. All the ex-corn starch composites behaved in a brittle manner, even those having weak interlayers with a porosity content higher than the critical value of about 0. 4 predicted by the model developed by Blanks et al. A non-purely brittle behaviour started to be obtained with ex-graphite laminar composites in which the pores are elongated and oriented parallel to the interfaces Keywords: Composites; Porosity; Mechanical properties; SiC; Laminates 1. Introduction As a necessary condition for the interfacial material that it must be chemically compatible with the material of Among the strategies developed to improve the flaw tol- the strong layers, Clegg and co-workers studied ceramic erance of non-transforming ceramics, the designing of lami- laminates with alternating dense and porous layers made nar structures with weak interfaces or interphases promoting of silicon carbide or alumina. This approach has also the crack deflection mechanisms have proved to be a successful advantage of avoiding the building up of any internal stresses way to increase fracture energy. The weak interfaces most due to difference in thermal expansion coefficients which mmonly used are either graphite or boron nitride. 2,3 For can be the source of delamination during cooling. 9 instance, in the case of graphite interfaces, Clegg et al. have According to theoretical analysis by He and Hutchinson obtained an apparent toughness and a fracture energy re- of the kinking of a crack out of an interface, 0 crack deflec- spectively 5 and 200 times higher than the typical values of tion is to occur when the ratio between the fracture energy monolithic a-SiC. However, since graphite and boron ni- of the homogeneous weak interphase, Gi, and the fracture tride have a low oxidation resistance and cannot be used at energy of the strong layer, Gs, is lower than 0.57. However high temperature in oxidising atmosphere without a protec- in the case of dense-porous laminates, the porous layer is tive barrier, other systems have been investigated, specially inhomogeneous, and the fracture energy to be considered is with oxidation resistant weak interphases such as monazite, 4 that of the ligament of matter between the crack and the pore fluorophlogopite or MoSi2+ Mo2 B5.6 ahead of it, Glig?. Than the criterion for crack deflection in the weak interphase becomes Glig/Gs <0.57. Though these ligaments of matter are made of the same material as Corresponding author the dense layers, and might be expected to have the same E-mail address: t chartier(aensci fr (T. Chartier). fracture energy, theoretical analyses show that if the tip of a 0955-2219/s-see front matter o 2004 Published by Elsevier Ltd doi: 10.1016/j jeurceramsoc 2004.02.009

Journal of the European Ceramic Society 25 (2005) 589–597 Mechanical properties and mechanical behaviour of SiC dense-porous laminates C. Reynaud a, F. Thévenot a, T. Chartier b,∗, J.-L. Besson b a Dept. Céramiques Spéciales, E.N.S. des Mines de Saint-Etienne, 42023 Saint-Etienne, France b S.P.C.T.S., U.M.R. C.N.R.S. 6638, E.N.S. de Céramiques Industrielles, 87065 Limoges, France Received 3 December 2003; received in revised form 23 February 2004; accepted 28 February 2004 Available online 15 June 2004 Abstract Porous laminar materials and alternate laminates of silicon carbide dense and porous layers have been elaborated by tape casting and liquid phase sintering processing. Porosity was introduced by incorporation of pore forming agents (corn starch or graphite platelets) in the slurry. Homogeneous distributions of porosity have been obtained for both monolithic and composite laminates. The microstructure of the SiC matrix was equiaxed and was not affected by the porosity. The porosity (P) dependence of Young’s modulus (E), modulus of rupture (σR), toughness (K1C) and fracture energy (G1C) was found to be well described on the entire range of porosity by relations of the form X0(1 − P)mX proposed by Wagh et al. from a model that takes into account the tortuosity of the porosity. In the case of our materials, mE = 2.7, mσR = mK1C = mE + 0.5 and mG1C = mE + 1. All the ex-corn starch composites behaved in a brittle manner, even those having weak interlayers with a porosity content higher than the critical value of about 0.4 predicted by the model developed by Blanks et al. A non-purely brittle behaviour started to be obtained with ex-graphite laminar composites in which the pores are elongated and oriented parallel to the interfaces. © 2004 Published by Elsevier Ltd. Keywords: Composites; Porosity; Mechanical properties; SiC; Laminates 1. Introduction Among the strategies developed to improve the flaw tol￾erance of non-transforming ceramics, the designing of lami￾nar structures with weak interfaces or interphases promoting crack deflection mechanisms have proved to be a successful way to increase fracture energy. The weak interfaces most commonly used are either graphite1 or boron nitride.2,3 For instance, in the case of graphite interfaces, Clegg et al. have obtained an apparent toughness and a fracture energy re￾spectively 5 and 200 times higher than the typical values of monolithic
-SiC.1 However, since graphite and boron ni￾tride have a low oxidation resistance and cannot be used at high temperature in oxidising atmosphere without a protec￾tive barrier, other systems have been investigated, specially with oxidation resistant weak interphases such as monazite,4 fluorophlogopite5 or MoSi2 + Mo2B5. 6 ∗ Corresponding author. E-mail address: t.chartier@ensci.fr (T. Chartier). As a necessary condition for the interfacial material is that it must be chemically compatible with the material of the strong layers, Clegg and co-workers studied ceramic laminates with alternating dense and porous layers made of silicon carbide7 or alumina.8 This approach has also the advantage of avoiding the building up of any internal stresses due to difference in thermal expansion coefficients which can be the source of delamination during cooling.9 According to theoretical analysis by He and Hutchinson of the kinking of a crack out of an interface,10 crack deflec￾tion is to occur when the ratio between the fracture energy of the homogeneous weak interphase, Gi, and the fracture energy of the strong layer, GS, is lower than 0.57. However, in the case of dense-porous laminates, the porous layer is inhomogeneous, and the fracture energy to be considered is that of the ligament of matter between the crack and the pore ahead of it, Glig7. Than the criterion for crack deflection in the weak interphase becomes Glig/GS < 0.57. Though these ligaments of matter are made of the same material as the dense layers, and might be expected to have the same fracture energy, theoretical analyses show that if the tip of a 0955-2219/$ – see front matter © 2004 Published by Elsevier Ltd. doi:10.1016/j.jeurceramsoc.2004.02.009

C. Reynaud et al. /Joumal of the European Ceramic Sociery 25 (2005)589-592 growing crack comes sufficiently close to a microcrack or De tip of th tic parameters of the two PFAs used increased and then, the apparent fracture energy of the liga- ment is reduced This reduction varies with the relative den- GP Blanks et al derived a minimum level of porosity equal to 37% to ensure Corn starch Graphite platelet ds0=14μm 8m×8pmx3um crack deflection, in good agreement with their experimental Aspect ratio 67 results for silicon carbide. Assuming that the fracture energy of the dense ligament is related to that of the porous layer GP, by the relation cosity and the rheological behaviour of the slurry and com- plement one another to produce handleable ceramic tapes Glig 1-P (1) The optimum concentrations for the binder and the plasti cizer were determined to be 8wt% for both additives on where P is the volume fraction of porosity, Davis et al. 8ex- the base of the ceramic powder. The third step corresponds pressed the criterion for continued crack deflection in terms to the incorporation, into the suspension, of the pore form- of easily measurable variables ing agent(PFA): corn starch(Roquette-France)or graphite platelets (Union Carbide-US)(Table 1). The PFA volume 0.57(1-P) (2) fraction, referred to the sum of the ceramic volume and pFA G volume, was 45 and 50% for graphite platelets and varied The experimental results obtained by Clegg's group with from 5 to 55% for corn starch. We were unable to fabricate ther alumina or silicon carbide were consistent with re- sound parts when larger PFA volume fraction were incor lation(2), crack deflection being observed when the ratio porated. A mixing time of 3.5 h leads to an uniform distri Gp/Gs falls under the line 0.57(1-P), that occurs for a bution of the PFA. Finally the slurry was de-aired at a low lume fraction of porosity of 0.37 rotation speed during 24 h The aim of the work presented in this paper was to deter Then the slurries were tape cast onto a Mylar" film using mine the porosity dependence of the mechanical properties a moving double blade device on a laboratory tape casting of a SiC material densified by liquid phase sintering on a bench(Elmeceram-France ). The tapes had smooth surfaces wide range of porosity, and to compare the mechanical be- and uniform thicknesses, which could be varied between haviour of dense-porous laminates to that obtained by Blanks 100 and 150 um. Sheets were punched in the green tapes et al. for silicon carbide densified by solid state sintering and stacked on each other using two types of stacking se- quence: (i) stacking of identical layers to obtain monolithic dense or porous specimens, and(ii) alternate stacking of 2. Processing and materials microstructure PFA free and PFA containing layers to obtain symmetrical dense/porous laminar composites. The outer layers were al 2. 1. Processing ways dense layers. The number of tapes was chosen to lead to a final thickness of about 3-4 mm after sintering. The The materials were made by stacking layers obtained by thickness of the layers was varied by stacking several tapes tape casting and co-sintering. The ceramic powder must of the same nature. This led to different architectures that al- ave a narrow particle size distribution and a rather low lowed to study the influence of the dense to porous thickness mean diameter dso was 1 um for the starting a-SiC powder ratio (Sika Tech FCP13, Norton-Norway) and the specific sur- After stacking, the specimens were pressed under 60 MPa face area(BEt)was 13 m/g. The densification aids(5 wt. at 65C. The burnout of organics was performed by heat- of the total ceramic content) for the liquid phase sinter ing very slowly(6 C/h)up to 550oC in air. Sintering was ing of SiC were Y203(Rhone-Poulenc-France)and Al203 conducted in a graphite furnace (VAS, France)under ar- (CR15 Baikowski-France)in a ratio corresponding to the gon at atmospheric pressure during I h at 1950C. The rel- YAG-Al2O3 eutectic composition(40 wt. Y203-60 wt ative density of the PFA free monolithic specimens was AlO3) 98% The first step of the elaboration of the suspensions for tape-casting consists in the dispersion of the ceramic pow- 2. 2. Microstructure ders in a solvent containing a dispersant. This was performed by planetary milling with alumina balls during 4 h in the a detailed description of the sintered microstructures of MEK -ethanol azeotrope containing 0.6wt. of a phosphate the monolithic and laminated Sic materials have been al ester(CP213, Cerampilot-France). Then, an acrylic binder ready published. So, only the main results will be briefly (Degalan" LP51701, Rohm and Haas-US) and a phtalate eported here plasticizer (DPB, Prolabo-France)were added to the suspen The microstructures were homogeneous. and in the case sion and mixed during 16h. They allow to optimise the vis- of the laminar composites, the layers were parallel with an

590 C. Reynaud et al. / Journal of the European Ceramic Society 25 (2005) 589–597 growing crack comes sufficiently close to a microcrack11 or a pore,12 the stress intensity factor at the tip of the crack is increased and then, the apparent fracture energy of the liga￾ment is reduced. This reduction varies with the relative den￾sity, and for a cubic array of spherical pores, Blanks et al.7 derived a minimum level of porosity equal to 37% to ensure crack deflection, in good agreement with their experimental results for silicon carbide. Assuming that the fracture energy of the dense ligament is related to that of the porous layer, GP, by the relation: Glig = GP 1 − P (1) where P is the volume fraction of porosity, Davis et al.8 ex￾pressed the criterion for continued crack deflection in terms of easily measurable variables: GP GS < 0.57(1 − P) (2) The experimental results obtained by Clegg’s group with either alumina or silicon carbide were consistent with re￾lation (2), crack deflection being observed when the ratio GP/GS falls under the line 0.57(1 − P), that occurs for a volume fraction of porosity of 0.37. The aim of the work presented in this paper was to deter￾mine the porosity dependence of the mechanical properties of a SiC material densified by liquid phase sintering on a wide range of porosity, and to compare the mechanical be￾haviour of dense-porous laminates to that obtained by Blanks et al.7 for silicon carbide densified by solid state sintering. 2. Processing and materials microstructure 2.1. Processing The materials were made by stacking layers obtained by tape casting and co-sintering. The ceramic powder must have a narrow particle size distribution and a rather low mean diameter d50 was 1m for the starting
-SiC powder (Sika Tech FCP13, Norton-Norway) and the specific sur￾face area (BET) was 13 m2/g. The densification aids (5 wt.% of the total ceramic content) for the liquid phase sinter￾ing of SiC were Y2O3 (Rhone-Poulenc-France) and Al2O3 (CR15 Ba¨ıkowski-France) in a ratio corresponding to the YAG–Al2O3 eutectic composition (40 wt.% Y2O3–60 wt.% Al2O3). The first step of the elaboration of the suspensions for tape-casting consists in the dispersion of the ceramic pow￾ders in a solvent containing a dispersant. This was performed by planetary milling with alumina balls during 4 h in the MEK-ethanol azeotrope containing 0.6 wt.% of a phosphate ester (CP213, Cerampilot-France). Then, an acrylic binder (Degalan® LP51/01, Röhm and Haas-US) and a phtalate plasticizer (DPB, Prolabo-France) were added to the suspen￾sion and mixed during 16 h. They allow to optimise the vis￾Table 1 Characteristic parameters of the two PFAs used PFA CS GP Chemical nature Corn starch Graphite platelet Dimension d50 = 14 m 8m × 8m × 3m Aspect ratio 1.2 2.67 cosity and the rheological behaviour of the slurry and com￾plement one another to produce handleable ceramic tapes. The optimum concentrations for the binder and the plasti￾cizer were determined to be 8 wt.% for both additives on the base of the ceramic powder. The third step corresponds to the incorporation, into the suspension, of the pore form￾ing agent (PFA): corn starch (Roquette-France) or graphite platelets (Union Carbide-US) (Table 1). The PFA volume fraction, referred to the sum of the ceramic volume and PFA volume, was 45 and 50% for graphite platelets and varied from 5 to 55% for corn starch. We were unable to fabricate sound parts when larger PFA volume fraction were incor￾porated. A mixing time of 3.5 h leads to an uniform distri￾bution of the PFA. Finally the slurry was de-aired at a low rotation speed during 24 h. Then the slurries were tape cast onto a Mylar® film using a moving double blade device on a laboratory tape casting bench (Elmeceram-France). The tapes had smooth surfaces and uniform thicknesses, which could be varied between 100 and 150m. Sheets were punched in the green tapes and stacked on each other using two types of stacking se￾quence: (i) stacking of identical layers to obtain monolithic dense or porous specimens, and (ii) alternate stacking of PFA free and PFA containing layers to obtain symmetrical dense/porous laminar composites. The outer layers were al￾ways dense layers. The number of tapes was chosen to lead to a final thickness of about 3–4 mm after sintering. The thickness of the layers was varied by stacking several tapes of the same nature. This led to different architectures that al￾lowed to study the influence of the dense to porous thickness ratio. After stacking, the specimens were pressed under 60 MPa at 65 ◦C. The burnout of organics was performed by heat￾ing very slowly (6 ◦C/h) up to 550 ◦C in air. Sintering was conducted in a graphite furnace (VAS, France) under ar￾gon at atmospheric pressure during 1 h at 1950 ◦C. The rel￾ative density of the PFA free monolithic specimens was 98%. 2.2. Microstructure A detailed description of the sintered microstructures of the monolithic and laminated SiC materials have been al￾ready published.13 So, only the main results will be briefly reported here. The microstructures were homogeneous, and in the case of the laminar composites, the layers were parallel with an

C. Reynaud et al. /Journal of the European Ceramic Society 25(2005)589-597 50 um 50 um Fig. 2. Optical micrographs showing the morphology of the porosity.(a) Monolithic laminate, (b) laminar composite Fig. 1 SEM micrographs; grain microstructure revealed by plasma etcl (a) Dense SiC,(b) porous SiC obtained by the incorporation of 45 tween the porous monoliths and the corresponding porous layers in the laminates remained low, that allowed the prop- erties of the porous layers to be obtained from tests on their monolithic counterparts without introducing a noticeable uniform thickness. The thickness was about 70 um for the error dense layers, 80-90 um for the ex-corn starch porous layers and 20-30 um for the porous layers where graphite platelets were used. No migration of pores from porous to dense lay ers was observed. Dense layers and skeletons in the porous layers were constituted by the same homogeneous, fine and 20um equiaxed microstructure with a mean equivalent diameter of about 1.25 um(Fig. 1). The pore characteristics were the same in the porous layers of the laminates and in their monolithic counterparts(Fig. 2). Whereas the pores intro- duced by corn starch had a low aspect ratio, those coming from graphite platelets were elongated and mostly oriented rallel to the interfaces(Fig. 3). The porosity of the mono- liths reached a maximum and levelled off for corn starch amounts higher than 40%. When the porous layers were co-sintered with dense layers, the amount of final porosity followed the law established by Slamovich and Lange+for thermodynamically stable pores shrinking the same amount as the matrix(Fig. 4). The lower shrinkage of the dense lay ers introduced tensile internal stresses in the highest porous Fig. 3. Optical micrograph showing the morphology of the porosity re- layers. However, the difference in volumetric shrinkage be- sulting from the incorporation of 50 vol. of graphite platelets

C. Reynaud et al. / Journal of the European Ceramic Society 25 (2005) 589–597 591 Fig. 1. SEM micrographs; grain microstructure revealed by plasma etching. (a) Dense SiC, (b) porous SiC obtained by the incorporation of 45 vol.% corn starch. uniform thickness. The thickness was about 70 m for the dense layers, 80–90 m for the ex-corn starch porous layers and 20–30m for the porous layers where graphite platelets were used. No migration of pores from porous to dense lay￾ers was observed. Dense layers and skeletons in the porous layers were constituted by the same homogeneous, fine and equiaxed microstructure with a mean equivalent diameter of about 1.25 m (Fig. 1). The pore characteristics were the same in the porous layers of the laminates and in their monolithic counterparts (Fig. 2). Whereas the pores intro￾duced by corn starch had a low aspect ratio, those coming from graphite platelets were elongated and mostly oriented parallel to the interfaces (Fig. 3). The porosity of the mono￾liths reached a maximum and levelled off for corn starch amounts higher than 40%. When the porous layers were co-sintered with dense layers, the amount of final porosity followed the law established by Slamovich and Lange14 for thermodynamically stable pores shrinking the same amount as the matrix (Fig. 4). The lower shrinkage of the dense lay￾ers introduced tensile internal stresses in the highest porous layers. However, the difference in volumetric shrinkage be￾Fig. 2. Optical micrographs showing the morphology of the porosity. (a) Monolithic laminate, (b) laminar composite. tween the porous monoliths and the corresponding porous layers in the laminates remained low, that allowed the prop￾erties of the porous layers to be obtained from tests on their monolithic counterparts without introducing a noticeable error. Fig. 3. Optical micrograph showing the morphology of the porosity re￾sulting from the incorporation of 50 vol.% of graphite platelets.

C. Reynaud et al. /Joumal of the European Ceramic Sociery 25 (2005)589-592 E 000 , 0102030405060 Pore Forming Agent (%o) 5101520 Fig. 4. Volume fraction of porosity after sintering vs PFA content. Mono- Square root of the notch radius(um) lithic laminates:(O)corn starch; laminar composites: (O)corn starch, Fig. 5. Dependence of toughness measured by the SENB-s method (A)graphite platelets. The dotted line corresponds to the equation from the square root of the notch radius for dense monolithic laminates.The Slamovich and Lange. 4 The solid line corresponds to the critical porosity doted curve corresponds to Eq (5)with Y= 1. 12 and Sa= 10 um for crack deflection according to Refs. 7, 8 ness, Kic, is related to the measured toughness by the rela 3. Experimental procedures tanh 2) Youngs modulus, E, was calculated from the measure- ment of the velocity of longitudinal, VL, and shear, Vs ultrasonic waves using a pulse echo overlap technique in where y is a geometric correction factor and da is the size the approximation of an homogeneous isotropic infinite finite of the small defect at the notch-tip, under the influence of the stress interaction field of the notch In order to assess if it was possible to use Eq. (5)for 3V-4v the correction of our experimental data, the toughness of VS 22 (3) Sic dense monolith was measured for three different root radii (100, 300 and 1000 um) and fitted by Eq. (5). In the where p is the density case of this dense, fine grained material, the defect at the The modulus of rupture, OR, was measured in 3-point notch-tip was assumed to result from machining, i.e., an edge bending at a cross-head speed of 0. 2 mm/min crack configuration(r= 1. 12)and machining scratches, The experimental method for the toughness determina- &a, of 10 um were adopted. 6 Fig. 5 shows that a goo tion must be applicable to highly porous materials, that agreement was obtained, giving an extrapolated value atr eliminates indentation based methods. So, the single edge 0 of 3.82 MPam/, which falls nicely in the values between notch beam-saw cut method(SENB-S)was selected. An- 3 and 4 MPam/2 usually obtained for a-SiC with this kind other reason for this choice was that, analysing the results of microstructure In the case of the porous monoliths, the of an European round robin test, Damani et al. 6 have con- crack defects at the notch-tip were assumed to have also an cluded that this method seemed to deliver the most repro- edge crack configuration and their size was taken equal to ducible results. The toughness was calculated from the mea- 10 um or to the mean average Feret diameter of the porosity surement of the modulus of rupture in 3-point bending and when it was greater than 10 um. 8 the depth of the initial notch, a, according to the following The fracture energy was calculated from the measure- ments of the toughness and Young's modulus from the re- lation k器=[4(= IC here w is the height of the sample and A; are coefficients given by: Ao=1.9+0.0075L/; A1=-339+0.08L/w, 4. Mechanical properties A2=154-0.2175Lhw;A3=-26.24+0.1825Lh;A4= 2638-0. 145L/ with L being the span The experimental data for dense and porous monoliths are However, this method leads to an overestimated value of reported in Fig. 6. Seven specimens were used for the de- the toughness when the notch root value, r, is larger than a termination of the modulus of rupture and of the toughness critical value of the order of a few microns. Damani et al. 16 The data were fitted by a(1- Py form, since Blanks suggested and verified experimentally that the true tough- et al. have obtained for SiC a good description of Youngs

592 C. Reynaud et al. / Journal of the European Ceramic Society 25 (2005) 589–597 0 10 20 30 40 50 0 10 20 30 40 50 60 Pore Forming Agent (%) Porosity (%) Fig. 4. Volume fraction of porosity after sintering vs. PFA content. Mono￾lithic laminates: (
) corn starch; laminar composites: () corn starch, () graphite platelets. The dotted line corresponds to the equation from Slamovich and Lange.14 The solid line corresponds to the critical porosity for crack deflection according to Refs. 7,8. 3. Experimental procedures Young’s modulus, E, was calculated from the measure￾ment of the velocity of longitudinal, VL, and shear, VS, ultrasonic waves using a pulse echo overlap technique in the approximation of an homogeneous isotropic infinite medium:15 E = ρV2 S
3V2 L − 4V2 S V2 L − V2 S (3) where ρ is the density. The modulus of rupture, σR, was measured in 3-point bending at a cross-head speed of 0.2 mm/min. The experimental method for the toughness determina￾tion must be applicable to highly porous materials, that eliminates indentation based methods. So, the single edge notch beam-saw cut method (SENB-S) was selected. An￾other reason for this choice was that, analysing the results of an European round robin test, Damani et al.16 have con￾cluded that this method seemed to deliver the most repro￾ducible results. The toughness was calculated from the mea￾surement of the modulus of rupture in 3-point bending and the depth of the initial notch, a, according to the following equation:17 KSENB 1C = σR √a 4 i=0  Ai  a w i  (4) where w is the height of the sample and Ai are coefficients given by: A0 = 1.9 + 0.0075L/w; A1 = −3.39 + 0.08L/w; A2 = 15.4−0.2175L/w; A3 = −26.24+0.1825L/w; A4 = 26.38 − 0.145L/w with L being the span. However, this method leads to an overestimated value of the toughness when the notch root value, r, is larger than a critical value of the order of a few microns. Damani et al.16 suggested and verified experimentally that the true tough- 0 2 4 6 8 10 12 14 0 5 10 15 20 25 Square root of the notch radius (µm1/2) Toughness (MPa.m1/2 ) Fig. 5. Dependence of toughness measured by the SENB-S method on the square root of the notch radius for dense monolithic laminates. The doted curve corresponds to Eq. (5) with Y = 1.12 and a = 10 m. ness, KT 1C, is related to the measured toughness by the rela￾tion: KT 1C = KSENB 1C tanh  2Y a r (5) where Y is a geometric correction factor and a is the size of the small defect at the notch-tip, under the influence of the stress interaction field of the notch. In order to assess if it was possible to use Eq. (5) for the correction of our experimental data, the toughness of SiC dense monolith was measured for three different root radii (100, 300 and 1000 m) and fitted by Eq. (5). In the case of this dense, fine grained material, the defect at the notch-tip was assumed to result from machining, i.e., an edge crack configuration (Y = 1.12) and machining scratches, a, of 10 m were adopted.16 Fig. 5 shows that a good agreement was obtained, giving an extrapolated value at r = 0 of 3.82 MPa m1/2, which falls nicely in the values between 3 and 4 MPa m1/2 usually obtained for
-SiC with this kind of microstructure. In the case of the porous monoliths, the crack defects at the notch-tip were assumed to have also an edge crack configuration and their size was taken equal to 10m or to the mean average Féret diameter of the porosity ˆ when it was greater than 10 m.18 The fracture energy was calculated from the measure￾ments of the toughness and Young’s modulus from the re￾lation: G1C = K2 1C E (6) 4. Mechanical properties The experimental data for dense and porous monoliths are reported in Fig. 6. Seven specimens were used for the de￾termination of the modulus of rupture and of the toughness. The data were fitted by a (1 − P) m form, since Blanks et al.7 have obtained for SiC a good description of Young’s

C. Reynaud et al. /Journal of the European Ceramic Society 25(2005)589-597 50 051015202530354045 051015202530354045 Porosity(vol%) Porosity(vol%) 3.5 30 3221100 051015202530354045 051015202530354045 Porosity (vol%) Porosity(vol%) Fig. 6. Dependence on porosity of the mechanical properties of monolithic laminates(PFA= CS).(a) Youngs modulus; (b)modulus of rupture;(c) toughness;(d) fracture energy. The solid lines correspond to the best fit with Wagh's equations modulus with Ep= Eo(1-P)2. Such an equation was also Wagh et al. this value of 2 is characteristic of ceramics established by Wagh et al. 19 for a model derived from an densified without additives or applied pressure. However, previous analytical one developed by Wong et al.20 in ef- for materials hot-pressed or densified using sintering aids forts to explain charge and mass transport through the ran- leading to an intergranular glassy phase, mE takes value dom pore structure of rocks. The assumed ceramic structure reater than 2 is a three-dimensional. intertwined continuous network of Wagh et al. 22 derived, for the fracture properties, power are interconnected by small channels oore i A 3 material chains and open-pore channels. This assumption is laws with different exponents connected together by based on the fact that it is possible to fabricate ceramics with a very high porosity, as high as 93%, and that oper mog=mIC =mE+0.5 and mGIc =mE+I exist even at very low porosity(6%). This is consistent with The results from fits done with the exponents put at their the mainly open nature of the porosity in the present mate- neoretical values are reported in Table 2. The correlation rials where the pores introduced by the factors are fairly good The model developed by Wagh et al. 9. that takes into The fitting parameters obtained using a classical mean account the tortuosity of the porosity, describes the whole square method, are reported in Table 2 set of our experimental data of the mechanical properties The mE parameter(2.68), is greater than the value of over the entire porosity range(from P=0%to P=42%) 2 obtained experimentally by Blanks et al. 7 According to and appears to be well adapted to the present porosity mor- Table 2 Fi ers for the mechanical properties using Wagh's model Property Correlation 5. Fracture behaviour 2.68±0.08 5.1. Experimental results 590±15 0.98 4.34±0.09 The fracture behaviour of the laminar composite (LC) 46.4±2.0 was studied in 3-point bending for unnotched and notche

C. Reynaud et al. / Journal of the European Ceramic Society 25 (2005) 589–597 593 Fig. 6. Dependence on porosity of the mechanical properties of monolithic laminates (PFA = CS). (a) Young’s modulus; (b) modulus of rupture; (c) toughness; (d) fracture energy. The solid lines correspond to the best fit with Wagh’s equations. modulus with EP = E0(1 − P)2. Such an equation was also established by Wagh et al.19 for a model derived from an previous analytical one developed by Wong et al.20 in ef￾forts to explain charge and mass transport through the ran￾dom pore structure of rocks. The assumed ceramic structure is a three-dimensional, intertwined, continuous network of material chains and open-pore channels. This assumption is based on the fact that it is possible to fabricate ceramics with a very high porosity, as high as 93%,21 and that open pores exist even at very low porosity (6%). This is consistent with the mainly open nature of the porosity in the present mate￾rials where the pores introduced by the pore forming agents are interconnected by small channels.13 The fitting parameters obtained using a classical mean square method, are reported in Table 2. The mE parameter (2.68), is greater than the value of 2 obtained experimentally by Blanks et al.7 According to Table 2 Fitting parameters for the mechanical properties using Wagh’s model Property Extrapolated at P = 0 m Correlation factor E (GPa) 407 ± 7 2.68 ± 0.08 0.995 σR (MPa) 590 ± 15 3.18 0.98 K1C (MPa m1/2) 4.34 ± 0.09 3.18 0.99 G1C (J/m2) 46.4 ± 2.0 3.68 0.96 Wagh et al. this value of 2 is characteristic of ceramics densified without additives or applied pressure. However, for materials hot-pressed or densified using sintering aids leading to an intergranular glassy phase, mE takes value greater than 2. Wagh et al.22 derived, for the fracture properties, power laws with different exponents connected together by: mσR = mK1C = mE + 0.5 and mG1C = mE + 1 The results from fits done with the exponents put at their theoretical values are reported in Table 2. The correlation factors are fairly good. The model developed by Wagh et al.19,22 that takes into account the tortuosity of the porosity, describes the whole set of our experimental data of the mechanical properties over the entire porosity range (from P = 0% to P = 42%) and appears to be well adapted to the present porosity mor￾phology. 5. Fracture behaviour 5.1. Experimental results The fracture behaviour of the laminar composite (LC) was studied in 3-point bending for unnotched and notched

C. Reynaud et al. /Joumal of the European Ceramic Sociery 25 (2005)589-592 Table 3 Laminar composites Pore forming agent PFA content(vol. % Porosity(vol %. Laminar composite sequences Corn starch 47(42户 l/-2/1-1/2 l/-2/-3/1-2/1-1/2 0545 43.5(40)3 Graphite platelets l/-2/-1/2 ce, the numbers refer to the number of tapes stacked to make one layer and indicate the relative thickness of the dense/porous layers. The first number concems the dense layers, the second one the porous ones(see Section 2.2) The values into brackets correspond to the porous monoliths whereas the others are the values in the porous interlayers of the composites samples. The cross-head displacement rate was varied from mens where the porous layers have a porosity equal to 40 0.5 to 0.004 mm/min to test a possible effect of the solici 43.5and47% tation rate. The role of the pore morphology was illustrated However, in the case of CSLC, whatever the porosity by comparing the behaviour of composites obtained using level, the cross-head speed or the architecture, the rupture corn starch(CSLC) and graphite platelets(GPLC) as pore was catastrophic with no increase in the work of fracture forming agent. The different materials and architectures are SEM micrographs of the side surfaces are shown in Fig. 9 listed in Table 3 For porosity levels up to 30%, the surface of rupture is flat Typical load-deflection curves for CSLC and GPLC are and is not affected by the alternation of dense/porous layer reported in Fig. 7. For all the materials, the load remains For larger porosity levels, whereas the crack propagates per- linear up to the peak load, at which point a crack initiates pendicularly through the dense layers, its direction changes at some surface defects and grows quickly in the through progressively in the porous layers, but, the lengths of de thickness direction flection are extremely short, of the order of 100 um or less, According to Clegg's group, 7.8 crack deflection should be except for the architecture with a dense to porous thickness observed when relation 2 is fulfilled, i.e., when Gp/Gs(1-P) ratio of 3 where deflections of 300 um are observed. Then, becomes lower than 0.57, and extensive deflection must oc- the crack kinks out of the porous interphase when it reaches cur when this ratio is lower than 0. 4, that corresponds to the porous/dense interface without inducing delamination porosity higher than 37 and 44%, respectively(see Fig 8 in In the case of GPLC, where the pores are elongated and Ref. 7). Using the parameters from Wagh's model (Table 2), lie roughly parallel to the interfaces, the occurrence of a sec- the corresponding values of porosity for the present materi- ondary load peak(see curves B in Fig. 7)indicates that the als are 19 and 33%, respectively. Then, from Fig. 8, where penetrating crack was effectively arrested, that allowed the Gp/Gs(1-P), calculated from the experimental data re- measured load to rise again until a new crack formed. The de- orted in Section 4, is plotted versus the porosity of the flection in curve Bl corresponds to the cross-head displace- porous layers, it should be the case at least for the speci- ment whereas the deflection in curve B2 corresponds to the splacement of the outer tensile surface(see the schematic illustration in Fig. 7). The difference between the deflection 18 B2 12 Deflection(mm) Porosity(vol%) res in 3-point bend test uence 1/1, P=43.5%); B: GPLC Fig. 8. GP/Gs(1-P)vS. porosity of the porous layers(corn starch). The ection for Bl corresponds to the solid line corresponds to Gp/Gs(1-P)=0.57(Eq (2). The shadowed splacement; the corresponds to the displacement of band divides the graph in an upper region where crack does not deflect tensile layer(see explanation at the end of Section 5.1). and a lower region where crack did deflect according to Blanks et al. 7

594 C. Reynaud et al. / Journal of the European Ceramic Society 25 (2005) 589–597 Table 3 Laminar composites Pore forming agent PFA content (vol.%) Porosity (vol.%) Laminar composite sequencesa Corn starch 55 47 (42)b 1/1–2/1–1/2 50 43.5 (40)b 1/1–2/2–3/1–2/1–1/2 45 40 1/1 34 28 1/1 25 21 1/1 Graphite platelets 50 41 1/1–2/1–1/2 a For a given sequence, the numbers refer to the number of tapes stacked to make one layer and indicate the relative thickness of the dense/porous layers. The first number concerns the dense layers, the second one the porous ones (see Section 2.2). b The values into brackets correspond to the porous monoliths whereas the others are the values in the porous interlayers of the composites. samples. The cross-head displacement rate was varied from 0.5 to 0.004 mm/min to test a possible effect of the solici￾tation rate. The role of the pore morphology was illustrated by comparing the behaviour of composites obtained using corn starch (CSLC) and graphite platelets (GPLC) as pore forming agent. The different materials and architectures are listed in Table 3. Typical load–deflection curves for CSLC and GPLC are reported in Fig. 7. For all the materials, the load remains linear up to the peak load, at which point a crack initiates at some surface defects and grows quickly in the through thickness direction. According to Clegg’s group,7,8 crack deflection should be observed when relation 2 is fulfilled, i.e., when GP/GS(1−P) becomes lower than 0.57, and extensive deflection must oc￾cur when this ratio is lower than 0.4, that corresponds to porosity higher than 37 and 44%, respectively (see Fig. 8 in Ref. 7). Using the parameters from Wagh’s model (Table 2), the corresponding values of porosity for the present materi￾als are 19 and 33%, respectively. Then, from Fig. 8, where GP/GS(1 − P), calculated from the experimental data re￾ported in Section 4, is plotted versus the porosity of the porous layers, it should be the case at least for the speci- 0 20 40 60 80 100 120 140 0 0.1 0.2 0.3 0.4 0.5 Deflection (mm) Load (N) A B1 B2 B1 B2 Fig. 7. Typical load–displacement curves in 3-point bend test of laminar composites. A: CSLC (sequence 1/1, P = 43.5%); B: GPLC (sequence 1/1, P = 41%). The deflection for B1 corresponds to the cross-head displacement; that for B2 corresponds to the displacement of the outer tensile layer (see explanation at the end of Section 5.1). mens where the porous layers have a porosity equal to 40, 43.5 and 47%. However, in the case of CSLC, whatever the porosity level, the cross-head speed or the architecture, the rupture was catastrophic with no increase in the work of fracture. SEM micrographs of the side surfaces are shown in Fig. 9. For porosity levels up to 30%, the surface of rupture is flat and is not affected by the alternation of dense/porous layers. For larger porosity levels, whereas the crack propagates per￾pendicularly through the dense layers, its direction changes progressively in the porous layers, but, the lengths of de- flection are extremely short, of the order of 100 m or less, except for the architecture with a dense to porous thickness ratio of 3 where deflections of 300 m are observed. Then, the crack kinks out of the porous interphase when it reaches the porous/dense interface without inducing delamination. In the case of GPLC, where the pores are elongated and lie roughly parallel to the interfaces, the occurrence of a sec￾ondary load peak (see curves B in Fig. 7) indicates that the penetrating crack was effectively arrested, that allowed the measured load to rise again until a new crack formed. The de- flection in curve B1 corresponds to the cross-head displace￾ment whereas the deflection in curve B2 corresponds to the displacement of the outer tensile surface (see the schematic illustration in Fig. 7). The difference between the deflection Fig. 8. GP/GS(1 − P) vs. porosity of the porous layers (corn starch). The solid line corresponds to GP/GS(1 − P) = 0.57 (Eq. (2)). The shadowed band divides the graph in an upper region where crack does not deflect and a lower region where crack did deflect according to Blanks et al.7

C. Reynaud et al. /Journal of the European Ceramic Society 25(2005)589-597 200um Fig. 10. SEM micrograph of typical surface of fracture showing that the rupture is intergranular Monolithic laminate, PFA= CS, porosity= inates studied by Vandeperre and Van Der Biest24 can also shed light on our results. These authors have fabricated, by electrophoretic deposition, laminar composites with alter nating layers of dense B-Sic and B-SiC containing different amounts of graphite(21, 34, 52 and 76 vol %C, respectively labelled types B, C, D and E). The mechanical behaviour tested in 3-point bending, showed no crack deflection for the 公 100um two lowest graphite contents(types B and C)but only slight deviation from the straight-through path. In the case of the composite with layers containing 52 vol. of graphite, some Fig 9. SEM micrographs of side surfaces of representative broken laminar specimens did not show a completely brittle failure but ex omposites( CSLC).(a)P=47vol % sequence 1/2; (b)P=43.5 vol %, hibited a load-deflection curve similar to that observed for sequence 3/1. Though the crack changes direction, it kinks out of the porous interlayer giving no useful toughening the GPlC specimens of the present work. Crack deflection occurred reliably only for type E material (i.e, the highest graphite content). The fracture energy of CSLC is compared with that of the materials studied by Vandeperre and Van in the two curves reflects the opening of a crack in a plane Der Biest in Fig. 11. The fracture energy of their type D perpendicular to the applied load and corresponds to a de- interlayers(52 vol%C)is the same as that of our highest lamination porous CslC samples and the behaviour of the correspond 5.2. Discussion Contrary to was expected, deflection was not observed for specimens that meet relation 2. As the only obvious difference between our samples and those of Blanks et al. 7 is that Blanks' ones were densified by solid state sintering 30505 whereas ours are densified by liquid phase sintering, the lack of crack deflection in our specimens might be linked to the presence of an intergranular amorphous phase. An argument that might support this hypothesis is the fact that, whereas in solid state sintered SiC crack propagation is transgranular, 3 the propagation in the present case is intergranular(Fig. 10) that suggests an influence of the intergranular amorphous SiC content(vol%) films on the mechanical behaviour Fig. 11. Fracture energy of porous SiC layers() in CSLC as a function However, the comparison of the behaviour of the present of the volume fraction of Sic compared to that of SiC+graphite interfaces laminar composites with those of the SiC/SiC-graphite lam-(A)in SiC/SiC-graphite laminar composites

C. Reynaud et al. / Journal of the European Ceramic Society 25 (2005) 589–597 595 Fig. 9. SEM micrographs of side surfaces of representative broken laminar composites (CSLC). (a) P = 47 vol.%, sequence 1/2; (b) P = 43.5 vol.%, sequence 3/1. Though the crack changes direction, it kinks out of the porous interlayer giving no useful toughening. in the two curves reflects the opening of a crack in a plane perpendicular to the applied load and corresponds to a de￾lamination. 5.2. Discussion Contrary to was expected, deflection was not observed for specimens that meet relation 2. As the only obvious difference between our samples and those of Blanks et al.7 is that Blanks’ ones were densified by solid state sintering whereas ours are densified by liquid phase sintering, the lack of crack deflection in our specimens might be linked to the presence of an intergranular amorphous phase. An argument that might support this hypothesis is the fact that, whereas in solid state sintered SiC crack propagation is transgranular,23 the propagation in the present case is intergranular (Fig. 10) that suggests an influence of the intergranular amorphous films on the mechanical behaviour. However, the comparison of the behaviour of the present laminar composites with those of the SiC/SiC–graphite lam￾Fig. 10. SEM micrograph of typical surface of fracture showing that the rupture is intergranular. Monolithic laminate, PFA = CS, porosity = 15 vol.%. inates studied by Vandeperre and Van Der Biest24 can also shed light on our results. These authors have fabricated, by electrophoretic deposition, laminar composites with alter￾nating layers of dense -SiC and -SiC containing different amounts of graphite (21, 34, 52 and 76 vol.% C, respectively labelled types B, C, D and E). The mechanical behaviour, tested in 3-point bending, showed no crack deflection for the two lowest graphite contents (types B and C) but only slight deviation from the straight-through path. In the case of the composite with layers containing 52 vol.% of graphite, some specimens did not show a completely brittle failure but ex￾hibited a load–deflection curve similar to that observed for the GPLC specimens of the present work. Crack deflection occurred reliably only for type E material (i.e., the highest graphite content). The fracture energy of CSLC is compared with that of the materials studied by Vandeperre and Van Der Biest in Fig. 11. The fracture energy of their type D interlayers (52 vol.% C) is the same as that of our highest porous CSLC samples and the behaviour of the correspond￾Fig. 11. Fracture energy of porous SiC layers () in CSLC as a function of the volume fraction of SiC compared to that of SiC+graphite interfaces () in SiC/SiC–graphite laminar composites.24

C. Reynaud et al. /Joumal of the European Ceramic Sociery 25 (2005)589-59 ing laminar composites mimics either that of CslC or that cant delaminations were observed. However, the behaviour of GPlc of the present composites has been shown to be consis- Let us now compare our materials with laminar com- tent with the behaviour observed for SiC/SiC-graphite4 and posites in the Si3N4 system (i.e, materials densified by Si3 N4/Si3N4-BN3 composites. Our opinion is that poros iquid phase sintering through the addition of 6 vol. of ity levels higher than the critical value proposed by Blanks tria)that were studied by Kovar et al. Five grades of et al. 7 are needed for crack deflection to occur reliably in weak layers were obtained from mixtures of Si3N4 and dense/porous laminar composites hexagonal BN: 20, 50, 75. 90 and 100 vol. of boron ni- tride. The load-deflection curves showed brittle failure up to 50 vol%bn. For this last bn content the delamina- tions are reported to be extremely short, less than 100 um References Substantial crack deflection and delamination started to be observed for bn content in the weak interlayers equal to 1. Clegg, W.J., Kendall, K, Alford, N. M, Birchall, D and Button, 75 voL% Due to the low cohesive force between graphite and Sic 455-457. or between BN and Si3N4 grains, it could be assumed that 2. Liu, H. and S. M. fracture behavior of multilayer silicon nitride/boron nitride ceramics. Am. Ceram. Soc. 1996.79. 2452- the graphite or BN platelets play a role similar to the pores resulting from the pyrolysis of the graphite platelets in the 3. Kovar, D, Thouless, M. D and Halloran, J. W, Crack deflection and GPLC specimens. In this respect, it appears that crack de- propagation in layered silicon nitride/boron nitride ceramics.J.Am. flection reliably occurs only for volume content significantly Ceram.Soc.1998,81,1004-1012. greater than the 37 vol % derived by Blanks et al. Unfortu- Mawdsley, J.R., Kovar, D. and Halloran, J. W, Fracture behavior nately, we were not able to process layers with a porosity f alumina/monazite multiplayer laminates. J. Am. Ceram. Soc. 2000 83,802-808 higher than 47% because the quantity of gas evolving from T. T. and Cooper, R. F, Ambient-temperature mechanical re- he burn out of the pore forming agent was too large and laminates. J. Am. Ceram. Soc. 1994 led to cracking and swelling of the materials. However. in 77,16991705 the light of the above comparison with SiC/SiC-graphite 6. Zhang, G.J., Yue, X. M. and Watanabe, T, High-temperature mul- and Si3N4/Si3N4-BN composites, our opinion is that it is bilayer composites with superplastic interlayers. J. Am. Ceram. Soc. 999,82,3257-3259 unlikely that the difference in the densification mechanism 7. Blanks, K.S., Kristoffersson, A, Carlstrom, E and Clegg, W.J could be the reason why Blanks' specimens showed crack Crack deflection in ceramic laminates using porous interlayers. J. Eur. deflection and not ours. Unfortunately, a more in-depth com- Ceran.Soc.1998,18,1945-1951 parison of the two sets of dense/porous laminates could not 8. Davis, J. B, Kristoffersson, A, Carlstrom, E. and Clegg, w. J be performed because of the lack of information on the mi- Fabrication and crack deflection in ceramic laminates with porous crostructure of the materials fabricated by Blanks and his interlayers. J 4m. Ceram. Soc. 2000, 83, 2369-2374 9. Howard, S.J., Stewart, R. A. and Clegg, w. J, Delamination of co-workers ceramic laminates due to residual thermal stresses In Key Engineering Materials, Vols 116-117, ed. T. w. Clyne. Trans. Tech Publications Switzerland, 1996, pp. 331-350 6. Conclusion 10. He, M -Y. and Hutchinson, J. W, Kinking of a crack out of interface.J.App. Mech. 1989, 56, 270-278 I1 Gong, S.-X. and Horil, H, General solution to the problem of mi- SiC materials have been fabricated by stacking tape cast crocracks near the tip of a main crack. J. Mech. Plns. Solids 1989 sheets and densification of the stack by liquid phase sin- 37(1),27-46 tering(YAG-alumina eutectic). Controlled porosity was in- 12. Alford, N M, Birchall, J.D. and Kendall,K, High strengthe troduced by the incorporation of pore forming agent in the slurry. Two types of samples have been prepared: mono- 13. Reynaud, C, Thevenot, F. and Chartier, T, Processing and m laminar composites. Int. Refract. Met. Hard Mater lithic blocks to determine the mechanical properties and al- ternate dense/porous laminates to test the ability to increase 14. Slamovich, E. B. an weak interlay pture by promoting crack deflection in the of ultrasonic e dependence on porosity of the mechanical properties solids in the frequency range 100 to 500 MHZ. J. Acoust. Soc. Am (Youngs modulus, modulus of rupture, toughness and frac 16. Damani, R. J, Gstrein, R. and Danzer, R, Critical notch-root radius ture energy)has been found to be well fitted on the entire effect in SENB-S fracture toughness testing. J. Eur: Ceram Soc. 1996, range of porosity(0-42%)by the set of equations proposed 16.695-70 by Wagh et al. 19,22 from a model that takes into account the 17. Jordan, Y, Elaboration et caracterisation de composites Dispersoides tortuosity of the porosity ENSM-SE/INSA Lyon, Saint-Etienne, France, 1991 Though laminates with weak interlayers having a poros- ity higher than the critical level required for crack deflec en Carbure de silicium Ph. D. thesis, ENSM-SE/INPG, Saint-Etien tion according to Clegg's group were fabricated, no signifi-

596 C. Reynaud et al. / Journal of the European Ceramic Society 25 (2005) 589–597 ing laminar composites mimics either that of CSLC or that of GPLC. Let us now compare our materials with laminar com￾posites in the Si3N4 system (i.e., materials densified by liquid phase sintering through the addition of 6 vol.% of yttria) that were studied by Kovar et al.3 Five grades of weak layers were obtained from mixtures of Si3N4 and hexagonal BN: 20, 50, 75, 90 and 100 vol.% of boron ni￾tride. The load–deflection curves showed brittle failure up to 50 vol.% BN. For this last BN content the delamina￾tions are reported to be extremely short, less than 100 m. Substantial crack deflection and delamination started to be observed for BN content in the weak interlayers equal to 75 vol.%. Due to the low cohesive force between graphite and SiC or between BN and Si3N4 grains, it could be assumed that the graphite or BN platelets play a role similar to the pores resulting from the pyrolysis of the graphite platelets in the GPLC specimens. In this respect, it appears that crack de- flection reliably occurs only for volume content significantly greater than the 37 vol.% derived by Blanks et al. Unfortu￾nately, we were not able to process layers with a porosity higher than 47% because the quantity of gas evolving from the burn out of the pore forming agent was too large and led to cracking and swelling of the materials. However, in the light of the above comparison with SiC/SiC–graphite and Si3N4/Si3N4–BN composites, our opinion is that it is unlikely that the difference in the densification mechanism could be the reason why Blanks’ specimens showed crack deflection and not ours. Unfortunately, a more in-depth com￾parison of the two sets of dense/porous laminates could not be performed because of the lack of information on the mi￾crostructure of the materials fabricated by Blanks and his co-workers. 6. Conclusion SiC materials have been fabricated by stacking tape cast sheets and densification of the stack by liquid phase sin￾tering (YAG-alumina eutectic). Controlled porosity was in￾troduced by the incorporation of pore forming agent in the slurry. Two types of samples have been prepared: mono￾lithic blocks to determine the mechanical properties and al￾ternate dense/porous laminates to test the ability to increase the work of rupture by promoting crack deflection in the weak interlayers. The dependence on porosity of the mechanical properties (Young’s modulus, modulus of rupture, toughness and frac￾ture energy) has been found to be well fitted on the entire range of porosity (0–42%) by the set of equations proposed by Wagh et al.19,22 from a model that takes into account the tortuosity of the porosity. Though laminates with weak interlayers having a poros￾ity higher than the critical level required for crack deflec￾tion according to Clegg’s group were fabricated, no signifi- cant delaminations were observed. However, the behaviour of the present composites has been shown to be consis￾tent with the behaviour observed for SiC/SiC–graphite24 and Si3N4/Si3N4–BN3 composites. Our opinion is that poros￾ity levels higher than the critical value proposed by Blanks et al.7 are needed for crack deflection to occur reliably in dense/porous laminar composites. References 1. Clegg, W. J., Kendall, K., Alford, N. M., Birchall, D. and Button, T. W., A simple way to make tough ceramics. Nature 1990, 347, 455–457. 2. Liu, H. and Hsu, S. M., Fracture behavior of multilayer silicon nitride/boron nitride ceramics. J. Am. Ceram. Soc. 1996, 79, 2452– 2457. 3. Kovar, D., Thouless, M. D. and Halloran, J. W., Crack deflection and propagation in layered silicon nitride/boron nitride ceramics. J. Am. Ceram. Soc. 1998, 81, 1004–1012. 4. Mawdsley, J. R., Kovar, D. and Halloran, J. W., Fracture behavior of alumina/monazite multiplayer laminates. J. Am. Ceram. Soc. 2000, 83, 802–808. 5. King, T. T. and Cooper, R. F., Ambient-temperature mechanical re￾sponse of alumina-fluoromica laminates. J. Am. Ceram. Soc. 1994, 77, 1699–1705. 6. Zhang, G.-J., Yue, X. M. and Watanabe, T., High-temperature mul￾tilayer composites with superplastic interlayers. J. Am. Ceram. Soc. 1999, 82, 3257–3259. 7. Blanks, K. S., Kristoffersson, A., Carlström, E. and Clegg, W. J., Crack deflection in ceramic laminates using porous interlayers. J. Eur. Ceram. Soc. 1998, 18, 1945–1951. 8. Davis, J. B., Kristoffersson, A., Carlström, E. and Clegg, W. J., Fabrication and crack deflection in ceramic laminates with porous interlayers. J. Am. Ceram. Soc. 2000, 83, 2369–2374. 9. Howard, S. J., Stewart, R. A. and Clegg, W. J., Delamination of ceramic laminates due to residual thermal stresses. In Key Engineering Materials, Vols 116–117, ed. T. W. Clyne. Trans. Tech. Publications, Switzerland, 1996, pp. 331–350. 10. He, M.-Y. and Hutchinson, J. W., Kinking of a crack out of an interface. J. Appl. Mech. 1989, 56, 270–278. 11. Gong, S.-X. and Horii, H., General solution to the problem of mi￾crocracks near the tip of a main crack. J. Mech. Phys. Solids 1989, 37(1), 27–46. 12. Alford, N. M., Birchall, J. D. and Kendall, K., High strength ceramics through colloidal control to remove defects. Nature 1987, 330, 51–53. 13. Reynaud, C., Thevenot, F. and Chartier, T., Processing and microstruc￾ture of SiC laminar composites. Int. J. Refract. Met. Hard Mater. 2000, 19, 425–435. 14. Slamovich, E. B. and Lange, F. F., Densification of large pores: I. Experiments. J. Am. Ceram. Soc. 1992, 75, 2498–2508. 15. Mc Skimmin, H. J., Measurement of ultrasonic wave velocities for solids in the frequency range 100 to 500 MHz. J. Acoust. Soc. Am. 1960, 34, 404–409. 16. Damani, R. J., Gstrein, R. and Danzer, R., Critical notch-root radius effect in SENB-S fracture toughness testing. J. Eur. Ceram. Soc. 1996, 16, 695–702. 17. Jordan, Y., Elaboration et Caractérisation de Composites Disperso¨ıdes Base Alumine-zircone à Vocation Thermodynamique. Ph.D. thesis, ENSM-SE/INSA Lyon, Saint-Etienne, France, 1991. 18. Reynaud, C., Céramiques Lamellaires Monolithiques et Composites en Carbure de Silicium. Ph.D. thesis, ENSM-SE/INPG, Saint-Etienne, France, 2002.

C. Reynaud et al. /Journal of the European Ceramic Society 25(2005)589-597 19. Wagh, A.S., Poeppel, R. B and Singh, J. P, Open pore description A.S., Singh, J. P. and F R. B, Dependence of ceramic of mechanical properties of ceramics. J. Mater. Sci. 1991, 26, 3862- 3868 roR Pheno penology of fracture in sinte 20. Wong, P, Koplik, J and Tomanic, J. P, Conductivity and permeability carbide.. Mater. Sci. 1984. 19. 2111-2120 of rocks. Phs. Rev B 1984. 30. 6606-6614 24. Vandeperre, L. and Van Der Biest, O, Composite SiC-graphite in 21. Ashby, M. F, The mechanical properties of cellular solids. Metall. terlayers for crack deflection in ceramic laminates. Silic. Ind. 1998. trans.A1983.14.1755-1768 63,3943

C. Reynaud et al. / Journal of the European Ceramic Society 25 (2005) 589–597 597 19. Wagh, A. S., Poeppel, R. B. and Singh, J. P., Open pore description of mechanical properties of ceramics. J. Mater. Sci. 1991, 26, 3862– 3868. 20. Wong, P., Koplik, J. and Tomanic, J. P., Conductivity and permeability of rocks. Phys. Rev. B 1984, 30, 6606–6614. 21. Ashby, M. F., The mechanical properties of cellular solids. Metall. Trans. A 1983, 14, 1755–1768. 22. Wagh, A. S., Singh, J. P. and Poeppel, R. B., Dependence of ceramic fracture properties on porosity. J. Mater. Sci. 1993, 28, 3589–3593. 23. Govila, R. K., Phenomenology of fracture in sintered alpha silicon carbide. J. Mater. Sci. 1984, 19, 2111–2120. 24. Vandeperre, L. and Van Der Biest, O., Composite SiC–graphite in￾terlayers for crack deflection in ceramic laminates. Silic. Ind. 1998, 63, 39–43