Printed in norie m irela atl ihs served MICROSTRUCTURE AND MECHANICAL PROPERTIES OF HI-NiCALON/BN/a-SILICON-NITRIDE CERAMIC-MATRIX COMPOSITES Monssef Drissi-Habti, *t Kikuo nakano The National Industrial Research Institute of Nagoya, 1-1 Hirate-cho, Kita-ku, Nagoya 462, Japan (Received 19 March 1996; revised 5 March 1997; accepted 21 March 1997) Abstract interface with carbon or boron nitride viz. mechanical Unidirectional ceramic-matrix composites consisting of tests performed at high temperatures along with post BN-coated Hi-Nicalon (Sic) fibre in a silicon nitride test observation under a microscope have proved that matrix were fabricated by a slurry-impregnation/hot- limited success has been achieved to date. Starting pressing route. The microstructures of the composites from the considerations addressed above, the scope of vere investigated by scanning and transmission electron this work is to provide an alternative to available microscopies and atomic force microscopy, while the CMCs by associating the a silicon nitride ceramic mechanical parameters of the composites were eval- matrix with Hi-Nicalon ceramic fibers. uated by flexure and micro-indentation tests. The effect Silicon nitride has excellent characteristics for use in of the processing route on the constituents is evaluated. high-temperature structural materials owing to its high Owing to the small size of the specimens required tability up to 1400C and its low density. However, it latter technique seems to be suitable for evaluation of is brittle as monolithic ceramics. The major concern in the in situ mechanical properties of the matrix and utilizing silicon nitride is to improve its R-curve fibers of a given composite. C 1997 Elsevier Science behavior. This could be achieved by using continuous Limited ramic fibers and this work presents the first results obtained by combining a silicon nitride with Hi- Keywords: A. ceramic-matrix composites; Hi-Nicalon Nicalon fibers(<0 5% oxygen content, *30 mol% fibres; B. mechanical properties; B. microstructure free carbon). This composite was manufactured by the polymer pyrolysis route, chosen on the basis of cost effectiveness considerations. 1 INTRODUCTION 2 MATERIALS AND EXPERIMENTS Continuous-fiber ceramic-matrix composites(CMCs) are considered to be promising candidates for high The material investigated is made of Hi-Nicalon fibers temperature application in structural parts, mainly in with 14 um mean diameter, 500 filaments/yarn, 2.74 g aerospace and military applications. However, despite cmdensity, Si: 63. 7%, C: 35-8%, O: 0 5% in weight their attractiveness and the great effort expended with a boron nitride coating of 0 5 um thickness. The during the last decade towards the improvement of longitudinal Youngs modulus of the virgin fibers is their processing routes and the prediction of their Er=250 GPa and the volume fraction of fibers is mechanical behavior, CMCs are still not totally V=0-48. The reinforcement is imprcgnated with convincing for large-scale manufacturing. This is slurry of a Si3n4 containing 5%Y2O3 and 5% Al2O3 mainly due to the numerous problems which remain to as sintering additives (referred to as the filler) be solved. In addition to the high cost of available uniaxial fiber prepreg was made by filament winding processing routes, the oxidation sensitivity of com- The slurry consisted of the filler, polysilazane(Chisso lonly used fibers as a consequence of their high NCP 201)and toluene. The prepreg was cut into oxygen and/or free carbon content is one of the main coupons of equal size to form a post-form such that limiting parameters. Even the relevant attempts the fibers were all arranged in the same direction. In directed towards the protection of the fiber/matrix order to make post-for To whom correspondence should addressed, at: 19 Avenue pyrolyzed(850C in nitrogen)and hot pressed(1550oC Horatio Smith. 14000 Caen. france in nitrogen). To allow a comparison to be made t On leave from LERMAT, URA CNRS 1317. ISMRa between the composite behavior and that of the Caen. france constituents, monolithic silicon nitride matrix material
Composites Science and Technology Sl(l997) 1483-1489 0 1997 Elsevier Science Limited Printed in Northern Ireland. All rights reserved ELSEVIER PII: SO266-3538(97)00070-S 0266-3538/97/$17.00 MICROSTRUCTURE AND MECHANICAL PROPERTIES OF HI-NICALONlBNh-SILICON-NITRIDE CERAMIC-MATRIX COMPOSITES Monssef Drissi-Habti,*t & Kikuo Nakano The National Industrial Research Institute of Nagoya, I -I Hirate-cho, Kita-ku, Nagoya 462, Japan (Received 19 March 1996; revised 5 March 1997; accepted 21 March 1997) Abstract Unidirectional ceramic-matrix composites consisting of BN-coated Hi-Nicalon (Sic) fibre in a silicon nitride matrix were fabricated by a slurry-impregnationlhotpressing route. The microstructures of the composites were investigated by scanning and transmission electron microscopies and atomic force microscopy, while the mechanical parameters of the composites were evaluated by flexure and micro-indentation tests. The effect of the processing route on the constituents is evaluated. Owing to the small size of the specimens required, the latter technique seems to be suitable for evaluation of the in situ mechanical properties of the matrix and fibers of a given composite. 0 I997 Elsevier Science Limited Keywords: A. ceramic-matrix composites; Hi-Nicalon fibres; B. mechanical properties; B. microstructure 1 INTRODUCTION Continuous-fiber ceramic-matrix composites (CMCs) are considered to be promising candidates for hightemperature application in structural parts, mainly in aerospace and military applications. However, despite their attractiveness and the great effort expended during the last decade towards the improvement of their processing routes and the prediction of their mechanical behavior, CMCs are still not totally convincing for large-scale manufacturing. This is mainly due to the numerous problems which remain to be solved. In addition to the high cost of available processing routes, the oxidation sensitivity of commonly used fibers as a consequence of their high oxygen and/or free carbon content is one of the main limiting parameters. Even the relevant attempts directed towards the protection of the fiber/matrix * To whom correspondence should addressed, at: 19 Avenue Horatio Smith, 14000 Caen, France. ’ On leave from LERMAT, URA CNRS 1317, ISMRa, Caen, France. interface with carbon or boron nitride, viz. mechanical tests performed at high temperatures along with posttest observation under a microscope, have proved that limited success has been achieved to date.’ Starting from the considerations addressed above, the scope of this work is to provide an alternative to available CMCs by associating the CY silicon nitride ceramic matrix with Hi-Nicalon ceramic fibers. Silicon nitride has excellent characteristics for use in high-temperature structural materials owing to its high stability up to 1400°C and its low density. However, it is brittle as monolithic ceramics. The major concern in utilizing silicon nitride is to improve its R-curve behavior. This could be achieved by using continuous ceramic fibers and this work presents the first results obtained by combining (Y silicon nitride with HiNicalon fibers ( < 05% oxygen content, = 30 mol% free carbon). This composite was manufactured by the polymer pyrolysis route, chosen on the basis of costeffectiveness considerations.* 2 MATERIALS AND EXPERIMENTS The material investigated is made of Hi-Nicalon fibers with 14 pm mean diameter, 500 filaments/yarn, 2.74 g cmp3 density, Si: 63.7%, C: 358%, 0: 0.5% in weight, with a boron nitride coating of 0.5 pm thickness. The longitudinal Young’s modulus of the virgin fibers is3 Ef = 250 GPa and the volume fraction of fibers is V,= 0.48. The reinforcement is impregnated with a slurry of (Y S&N, containing 5% Y,O, and 5% Al,O, as sintering additives (referred to as the filler). A uniaxial fiber prepreg was made by filament winding. The slurry consisted of the filler, polysilazane (Chisso NCP 201) and toluene. The prepreg was cut into coupons of equal size to form a post-form such that the fibers were all arranged in the same direction. In order to make a composite, the post-form was pyrolyzed (850°C in nitrogen) and hot pressed (1550°C in nitrogen). To allow a comparison to be made between the composite behavior and that of the constituents, monolithic silicon nitride matrix material 1483
1484 M. Drissi-Habti, K. Nakano was prepared under the same experimental conditions virgin and BN-coated fibers were provided The microstructure of the as-processed material was ivestigated by scanning electron microscopy (SEM) and transmission electron microscopy(TEM)(Hitachi H9000 NAR). The TEM specimens were prepared by mechanical and ion thinning of a slice cut perpen dicular to the fiber axis of the composite. The topographies of both virgin and coated fibers were investigated by means of atomic force microscopy 季 (AFM) Mechanical characterization of the composite was carried out in an Instron testing machine( model 4204) Fig. 2. TEM micrograph of a fiber/BN-interphase/matrix with a 5 kN load cell. Three-and four-point bend tests region in a Hi-Nicalon(BN)/Si N4 composite material( scale wert rformed both unnotched and notched bar is 0-1 um) specimens. The configuration of both flexural tests, together with the dimensions of the specimens, are close to 2-9%. With the aid of an optical microscope, llustrated in Fig. I Tests were displacement-controlled the fiber volume fraction was estimated as being 0.48 and conducted at a crosshead speed of 0 5 mm min Figure 2 shows a TEM micrograph of the interfacial Micro-indentation tests were performed on a force- fiber/matrix region in the composite. It is conjectured driven, static, measuring ultra micro-indentation that the BN interphase has a layer crystal structure by instrument(UMIS 2000). The force system was varied its electron diffraction patterns shown in the inserts. between 50 and 1000 mN, since the depth measure- The Sic fiber had a structure consisting of fine B Sic ment system had a maximum value of 40 um with 20 crystallites. 2 The lattice image around a boundary nm resolution. The system was equipped with a video region between the BN interphase and the matrix, microscope and a positioning stage with a field of Fig. 3, shows that the (0001) planes of the BN travel of 20 mm and a minimum step size under motor interphase are aligned nearly parallel to the boundary, ontrol of 2: 5 um. A spherical diamond indenter wa which suggests that these planes form concentric indenter was equipped with a personal computer with circles around the fiber two spare 8-and 16-bit expansion slots. Tests were 3.2 Mechanical characterization conducted on a fine polished surface of the specimens 3.2.1 Notched specimens (up to 1 um diamond paste)which was previously Figure 4 shows a four-point notch bend test for two embedded in a resin specimens: monolithic silicon nitride and the icalon/BN/silicon-nitride composite with the same 3 RESULTS AND DISCUSSION notch depth of 1 mm. As noticed, the conjunction of the fibers with the Si3N4 leads to a significant non 3.1 Microstructure linear behavior, thus suggesting that the load is The first results obtained were related to the sintering characteristics and microstructure of the as-fabricated 0.01 um composite. The composite was sintered at 1550.C.It showed a bulk density of 2 81 g cm(corresponding to 94. 3% of theoretical density) and an open porosity №⊥LN4】 一一一一 Fig. 1. Configuration of the tests with the corre sponding parameters. L is the of the specimen, h is Fig 3. Lattice image around a region located between the the notch depth, /, and /2 are th ances between the inner n interphase and matrix in the composite(scale bar is and oute 001m)
1484 M. Drissi-Habti, K. Nakano was prepared under the same experimental conditions as those used for the composite. In addition, both virgin and BN-coated fibers were provided. The microstructure of the as-processed material was investigated by scanning electron microscopy (SEM) and transmission electron microscopy (TEM) (Hitachi, H 9000 NAR). The TEM specimens were prepared by mechanical and ion thinning of a slice cut perpendicular to the fiber axis of the composite. The topographies of both virgin and coated fibers were investigated by means of atomic force microscopy (AFM). Mechanical characterization of the composite was carried out in an Instron testing machine (model 4204) with a 5 kN load cell. Three- and four-point bend tests were performed on both unnotched and notched specimens. The configuration of both flexural tests, together with the dimensions of the specimens, are illustrated in Fig. 1. Tests were displacement-controlled and conducted at a crosshead speed of 0.5 mm min-‘. Micro-indentation tests were performed on a forcedriven, static, measuring ultra micro-indentation instrument (UMIS 2000). The force system was varied between 50 and 1000 mN, since the depth measurement system had a maximum value of 40 pm with 20 nm resolution. The system was equipped with a video microscope and a positioning stage with a field of travel of 20 mm and a minimum step size under motor control of 2.5 pm. A spherical diamond indenter was used for this study. For data analysis, the microindenter was equipped with a personal computer with two spare 8- and 16-bit expansion slots. Tests were conducted on a fine polished surface of the specimens (up to 1 ,um diamond paste) which was previously embedded in a resin. 3 RESULTS AND DISCUSSION 3.1 Microstructure The first results obtained were related to the sintering characteristics and microstructure of the as-fabricated composite. The composite was sintered at 1550°C. It showed a bulk density of 2.81 g cmP3 (corresponding to 94.3% of theoretical density) and an open porosity B w c, /L_--_-_ ,1=30lWl f--b L=50mm,B=4mm,W=3mm L I2=10mm 4-w B h=llIlm w rt” J-- --- l1=30mm f---b .=50mm,B=3mm,W=4mm Fig. 1. Configuration of the flexure tests with the corresponding parameters. L is the length of the specimen, h is the notch depth, 1, and I2 are the distances between the inner and outer spans. Fig. 2. TEM micrograph of a fiber/BN-interphase/matrix region in a Hi-Nicalon (BN)/S&N, composite material (scale bar is 0.1 pm). close to 2.9%. With the aid of an optical microscope, the fiber volume fraction was estimated as being 0.48. Figure 2 shows a TEM micrograph of the interfacial fiber/matrix region in the composite. It is conjectured that the BN interphase has a layer crystal structure by its electron diffraction patterns shown in the inserts. The SIC fiber had a structure consisting of fine p SIC crystallites.2 The lattice image around a boundary region between the BN interphase and the matrix, Fig. 3, shows that the (0001) planes of the BN interphase are aligned nearly parallel to the boundary, which suggests that these planes form concentric circles around the fiber.’ 3.2 Mechanical characterization 3.2.1 Notched specimens Figure 4 shows a four-point notch bend test for two specimens: monolithic silicon nitride and the HiNicalon/BN/silicon-nitride composite with the same notch depth of 1 mm. As noticed, the conjunction of the fibers with the Si3N4 leads to a significant nonlinear behavior, thus suggesting that the load is Fig. 3. Lattice image around a region located between the BN interphase and matrix in the composite (scale bar is 0.01 pm)
Hi-Nicalon/BN/a-Si,N, ceramic-matrix composites 1485 that(1)the ultimate tensile stress, ou, and the ultimate bending stress of the composite are equal and(2) to the ultimate stress of the composite, all the fibers are still bearing the load, then the ultimate tensile Load (N) stress of the fibers in the composite, o can be calculated by using eqn(4). The value so calculated can be viewed as an upper limit of the ultimate stress of the fiber tested under tensile mode: 0000050.100.15020025 With u=b=672 MPa, o, is 1400 MPa. This value is much lower than that measured by Nippon Carbon for ngn0ithic silicon nitride and the hi-nicalon /silicon-nitride virgin Hi-Nicalon fibers(a =1940 MPa)2,and this composite tested under four-point flexure point must be emphasized Otherwise, from eqn(2), the longitudinal Youngs correctly transferred at the fiber/matrix interface. modulus is E=206 GPa and 164 GPa for the Observation after flexural tests shows that the crack monolithic matrix and the composite, respectively. As starts propagating from the notch tip and its noted, the modulus of the monolithic silicon nitride is progression is tortuous owing to the extensive fiber ligher than that of the composite. This surprising bridging mechanism. In the case of linear elastic result would suggest that there is no enhancement in behavior, the matrix fracture energy could be the stiffness of the monolithic matrix by the estimated by assuming that the matrix crack starts incorporation of these high-tnodulus fibers(E=250 propagating at the end of the proportional limit GPa). In fact, when inserting the values of the characterized by the proportional load, PI(corre- longitudinal Young s modulus and the volume fraction ponding to a displacement ui), by using the following of the matrix [Em =193 GPa(see the next paragraph relationship G=Pu,2B(W-h) where B and w are the dimensions of the specimens and h the height of the notch( Fig. 1). The calculated matrix fracture energy valucs arc similar for both thc monolithic silicon nitride and the composite: G=345J m. Such a value is reasonable compared with others reported in the literature. Otherwise, since this energy value corresponds to initiation of the matrix crack. it Hi-Nicaion/BN/SI3N4 would be interesting, as a further investigation, to calculate the increasing crack growth resistance values B-curve). Because of a lack of larger specimens (compact tension ones, for example), such a study has not been conducted vet deflection(um) Fig. 5(a)Three-point flexure test of unnotched monolithic 3.2.2 Unnotched specimens silicon nitride and Hi-Nicalon/BN/silicon-nitride composite Figure 5 shows the load/displacement relationshi tained in a three- point bend flexure test performed on two unnotched specimens of monolithic silicon xy一 from the follow classical beam theory PeO E=P/48Bw3 where P is the applied load and 8 the deflection from these relationships, the ultimate bending stress is Tb=538 and 672 MPa for the matrix and the Three-point flexure test of unnotched monolithic composite, respectively. Making the crude assumptions itride and Hi-Nicalon/BN/silicon-nitride composit
Hi-Nicalon/BN/c-u-Si,N, ceramic-matrix composites 1485 0 SiN4(M) l Hi-NicalodSiiN4 I I I I I 1 I ! , I I I 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 040 Displacement (mm) Fig. 4. Load/displacement curves of notched specimens of monolithic silicon nitride and the Hi-Nicalonkilicon-nitride composite tested under four-point flexure. correctly transferred at the fiber/matrix interface. Observation after flexural tests shows that the crack starts propagating from the notch tip and its progression is tortuous owing to the extensive fiberbridging mechanism. In the case of linear elastic behavior, the matrix fracture energy could be estimated by assuming that the matrix crack starts propagating at the end of the proportional limit, characterized by the proportional load, Pi (corresponding to a displacement CL,), by using the following relationship: G = P,uiI[2B(W - h)] (1) where B and W are the dimensions of the specimens and h the height of the notch (Fig. 1). The calculated matrix fracture energy values are similar for both the monolithic silicon nitride and the composite: G = 34.5 J m -2. Such a value is reasonable compared with others reported in the literature.4 Otherwise, since this energy value corresponds to initiation of the matrix crack, it would be interesting, as a further investigation, to calculate the increasing crack growth resistance values (R-curve). Because of a lack of larger specimens (compact tension ones, for example), such a study has not been conducted yet. 3.2.2 Urmotched specimens Figure 5 shows the load/displacement relationships obtained in a three-point bend flexure test performed on two unnotched specimens of monolithic silicon nitride and Hi-Nicalon/BN/silicon-nitride composite. The ultimate bending stress, (TV, and the longitudinal Young’s modulus, E, of these materials were calculated from the following relationships derived from the classical beam theory: ub = 3P112BW2 (2) and E = P13/4SBW’ (3) where P is the applied load and 6 the deflection. From these relationships, the ultimate bending stress is (rb=538 and 672 MPa for the matrix and the composite, respectively. Making the crude assumptions that (1) the ultimate tensile stress, (T,,, and the ultimate bending stress of the composite are equal and (2) up to the ultimate stress of the composite, all the fibers are still bearing the load, then the ultimate tensile stress of the fibers in the composite, rf, can be calculated by using eqn (4). The value so calculated can be viewed as an upper limit of the ultimate stress of the fiber tested under tensile mode: (Tf = U”/V, (4) With g’, = (To = 672 MPa, gf is 1400 MPa. This value is much lower than that measured by Nippon Carbon for virgin Hi-Nicalon fibers ((TV = 1940 MPa)“, and this point must be emphasized. Otherwise, from eqn (2), the longitudinal Young’s modulus is E =206 GPa and 164 GPa for the monolithic matrix and the composite, respectively. As noted, the modulus of the monolithic silicon nitride is higher than that of the composite. This surprising result would suggest that there is no enhancement in the stiffness of the monolithic matrix by the incorporation of these high-modulus fibers3 (Ef = 250 GPa). In fact, when inserting the values of the longitudinal Young’s modulus and the volume fraction of the matrix [E, = 193 GPa (see the next paragraph) 6oq HkNicalonlBNISi3N4 ---+-- S13N4(M) 00 0.1 0.2 0.3 0.4 deflection (mm) Fig. 5(a) Three-point flexure test of unnotched monolithic silicon nitride and Hi-Nicalon/BN/silicon-nitride composite, 0 FSF, a P>P, OF=0 h, Fig. 5(b) Three-point flexure test of unnotched monolithic silicon nitride and Hi-Nicalon/BN/silicon-nitride composite
1486 M. Drissi-Habti, K. Nakano and Vm=0 49 and that of the fibers(Er=250 GPa and Vr=0-48)into the rule of mixtures E=EV+ emv the as-calculated longitudinal Youngs modulus of the 一Es3N4(m) omposite is 215 GPa, Such a value would be expected a 一日 Ecomp tional Hi-Nicalon/BN/Silicon-nitride composite. How- ever, the value of E calculated from eqn (5)is greater than that measured in the three-point bend test. Since the reverse would be expected. such a difference may be due to a loss in stiffness of the hi-Nicalon fibers during the bn coating process at 1500C, or when hot 100200300400500600 pressing during the sintering stage. This will be partially elucidated by micro-indentation tests and Fig. 6. Illustration of the parameters used when interpreting atomic force microscope observations. indentation tests (after Puttock and Thwaite) 3.2.3 Micro-indentation tests Micro-indentation tests were performed on the fibers Analysis of the partial unload data provides for the and the matrix(both on monolithic Si3 N4 and on extraction of Eat each step. The longitudinal Young's matrix-rich regions in the composite), to evaluate the modulus of the material is then determined if the local Youngs modulus of the matrix and the fibers longitudinal Youngs modulus and poissons ratio of This was accomplished by a multiple partial unloading the indenter(v; and Ei) and the poissons ratio of the procedure. Such tests are particularly suitable for a aterial (vm) are known, according to the earlier spherical indentation and give reliable measures of the analysis of Ref. 6: hardness and the longitudinal Young s modulus as a 1/E′=(1-v)E+(1-vm)E function of the depth of elastic/plastic penetration for An important parameter to control is the contact materials whose properties vary with distance from the force. To prevent impact damage, the contact force surface. To determine the hardness from an indenta must be properly chosen. The above relationships have tion test, it is necessary to determine the penetration already been integrated into a computer program and of the indenter below the perimeter of contact, the this makes the exploitation of experimental results remainder of the total measured depth being elastic straightforward. However, the procedure described depression of the surrounding material. The elastic above implies that no damage appears when loading contribution obtained from unloading data is required In this case the test must be rejected, since measured to make these calculations. In the multiple partial valucs are dependent on the amount of damage unloading(20 in our case), a single indentation is Figure 7 shows micro-indentation curves recorded partially unloaded in one increment at each step by an for the monolithic matrix and the matrix-rich region amount of 50% of the total step. Each force step in the composite, using 250 mN maximum appl provides two pieces of information: the total depth of elastic/plastic penetration and a measure of the recovery from that load. For a spherical indenter, the elastic recovery, d, and the longitudinal Youngs modulus of both the material and the indenter e, are given by the following relationships, derived from the work of Puttock and Thwaite(Fig. 6) E′=F(98)(D-1Dn)2d-32 where D(=2R)is the diameter of the spherical indenter, D is the diameter of the residual spherical 鲁鲁鲁鲁一你 depression and F the total load. If h,, ho and h, are the total(corresponding to the total force, F), plastic and residual depths, the elastic recovery may be obtained neration be low depth(nm) Fig. 7. Variation of the mean longitudinal Youngs modulus values with the penetration below contact when indenting he monolithic silicon nitride [ESi,N,(m) and the matrix rich regions in the Hi-Nicalon/BN/silicon-nitride compo (Ecomp). The maximum applied force is 250 mN
1486 M. Drissi-Habti, K. Nakano and V, = 0.491 and that of the fibers (E, = 2.50 GPa and V, = 0.48) into the rule of mixtures: 225 1 E = EfVf + E,V, (5) the as-calculated longitudinal Young’s modulus of the composite is 215 GPa. Such a value would be expected when testing uniaxial tensile specimens of unidirectional Hi-Nicalon/BN/silicon-nitride composite. However, the value of E calculated from eqn (5) is greater than that measured in the three-point bend test. Since the reverse would be expected, such a difference may be due to a loss in stiffness of the Hi-Nicalon fibers during the BN coating process at 15OO”C, or when hot pressing during the sintering stage. This will be partially elucidated by micro-indentation tests and atomic force microscope observations. 220 I 7 215 '\ i i L 0 210 ~. % I '\ iG 205 -: + 200 t i ‘; ‘1\ 195 4 \ b-_- i-, ---I)---------o ,go -., : ;.-~-~-~-._+__, 0 100 200 300 400 500 600 Penetration below depth (nm) Fig. 6. Illustration of the parameters used when interpreting indentation tests (after Puttock and Thwaite)‘. ----)- E33N4(m) -*-komp 3.2.3 Micro-indentation tests Micro-indentation tests were performed on the fibers and the matrix (both on monolithic S&N, and on matrix-rich regions in the composite), to evaluate the local Young’s modulus of the matrix and the fibers. This was accomplished by a multiple partial unloading procedure. Such tests are particularly suitable for a spherical indentation and give reliable measures of the hardness and the longitudinal Young’s modulus as a function of the depth of elastic/plastic penetration for materials whose properties vary with distance from the surface. To determine the hardness from an indentation test, it is necessary to determine the penetration of the indenter below the perimeter of contact, the remainder of the total measured depth being elastic depression of the surrounding material. The elastic contribution obtained from unloading data is required to make these calculations. In the multiple partial unloading (20 in our case), a single indentation is partially unloaded in one increment at each step by an amount of 50% of the total step. Each force step provides two pieces of information: the total depth of elastic/plastic penetration and a measure of the recovery from that load. For a spherical indenter, the elastic recovery, d, and the longitudinal Young’s modulus of both the material and the indenter, E’, are given by the following relationships, derived from the work of Puttock and Thwaite’ (Fig. 6): Analysis of the partial unload data provides for the extraction of E’ at each step. The longitudinal Young’s modulus of the material is then determined if the longitudinal Young’s modulus and Poisson’s ratio of the indenter (Yi and Ei) and the Poisson’s ratio of the material (Y,) are known, according to the earlier analysis of Ref. 6: l/E ’ = (1 - v;)lE + (1 - I&/E,,, (10) An important parameter to control is the contact force. To prevent impact damage, the contact force must be properly chosen. The above relationships have already been integrated into a computer program and this makes the exploitation of experimental results straightforward. However, the procedure described above implies that no damage appears when loading. In this case the test must be rejected, since measured values are dependent on the amount of damage.’ Figure 7 shows micro-indentation curves recorded for the monolithic matrix and the matrix-rich regions in the composite, using 250 mN maximum applied d = (9/8)1’3(1/E’)2’3(1/D - 1/D,)“3F2’3 (6) E’ = F(9/8)1’2(1/D - 1/D,)“2d-3’2 (7) where D (=2R) is the diameter of the spherical indenter, D, is the diameter of the residual spherical depression and F the total load. If h,, h, and h, are the total (corresponding to the total force, F), plastic and residual depths, the elastic recovery may be obtained from: d = h, - h, (8) ,o(J ,/,,i,iiLtLY,i ,,,, l.iAl 0 100 200 300 400 500 Penetration below depth (nm) where Fig. 7. Variation of the mean longitudinal Young’s modulus values with the penetration below contact when indenting the monolithic silicon nitride [ES&N,(m)] and the matrixrich regions in the Hi-NicalorUBNMicon-nitride composite h, = h, - d/2 (9) (Ecomp). The maximum applied force is 250 mN
Hi-Nicalon/BN/a-Si,N, ceramic-matrix composites 1487 Topography of a virgin fiber [2 um] 1454nm 93953nn 96977nm 969.77nm Topography of a BN coated fiber[2 um] o nm 230.9nm 200881nm 10044nm Fig. 8. Variation of the longitudinal Youngs modulus with the penetration below contact in BN-coated Hi-Nicalon fibers from micro-indentation tests. The maximum applied force is 250 mN force and a contact force of 1 mN. As noticed local the sintering and bn coating processes, it seems a longitudinal Youngs modulus values plotted as a priori that the former parameter has less influence for function of the depth below penetration show a similar the following reasons: (1)the hot pressing was plateau value close to 193 GPa, obtained after a performed in a controled atmosphere(N2),(2)at a penetration of about 100 um(the Poisson's ratio of relatively low pressure (=30 MPa) and (3)at a the matrix being Vm=0-28). Despite the low difterence, temperature (1550"C)where the fibers exhibit the this value is in good agreement with the one measured same longitudinal Youngs modulus as at room during flexure tests. Since this result reflects the mean temperature. Moore" has also shown that, depending values of 55 indentation tests, carried out on various on the BN coating temperature range, the ultimate regions of the specimen, it could be therefore tensile strength and the longitudinal Youngs modulus concluded that the longitudinal Youngs modulus of of SiC Nicalon fibers could be affected by the coating the fiber is Fm=193 GPa. The corresponding hardness process is 19. 8 GPa at a maximum depth of 392 nm For the above reasons, an AFm examination was Figure 8 shows the variations of the longitudinal carried out to check the surface of the virgin Hi- Young's modulus as a function of penetration depth, Nicalon fibers(without coating) and BN-coated fibers as derived from indentation tests performed on high both as-received from Nippon Carbon. An example of Nicalon fiber he composite. The values are mean these two fiber topographies is given in Fig 9 Figure 9 in the case of 2 um and 0.5 um total lengt maximum applied force( the contact force is I mN). scale, respectively. It can be seen that the BN-coated As noticed, a plateau value of 106 GPa in the Young's fiber shows a very rough surface, whereas the virgin modulus is reached aftcr a penetration of about 180 fiber displays much smaller roughness amplitude. The nm. The corresponding hardness is 14 GPa at roughness amplitude for BN-coated fibers is nearly maximum depth of 410 nm, A comparison between double that of uncoated fibers when analyzing the this value and that reported in the literature(250 highest magnification micrograph GPa) leads to the conclusion that the fibers had lost To explain this difference, let us analyze the BN more than 50%of their initial stiffness. With regard to coating process. In fact, boron nitride is deposited on
Hi-NicalonlBN/cY-Si,N, ceramic-matrix composites Topography of a virgin fiber [2 pm] 1487 Topography of a BN coated fiber [2 pm] Fig. 8. Variation of the longitudinal Young’s modulus with the penetration below contact in BN-coated Hi-Nicalon fibers from micro-indentation tests. The maximum applied force is 2.50 mN. force and a contact force of 1 mN. As noticed, local longitudinal Young’s modulus values plotted as a function of the depth below penetration show a similar plateau value close to 193 GPa, obtained after a penetration of about 100 ,um (the Poisson’s ratio of the matrix being Y, = O-28). Despite the low difference, this value is in good agreement with the one measured during flexure tests. Since this result reflects the mean values of 55 indentation tests, carried out on various regions of the specimen, it could be therefore concluded that the longitudinal Young’s modulus of the fiber is E,= 193 GPa. The corresponding hardness is 198 GPa at a maximum depth of 392 nm. Figure 8 shows the variations of the longitudinal Young’s modulus as a function of penetration depth, as derived from indentation tests performed on high Nicalon fibers in the composite. The values are mean ones obtained by testing 65 fibers up to 250 mN maximum applied force (the contact force is 1 mN). As noticed, a plateau value of 106 GPa in the Young’s modulus is reached after a penetration of about 180 nm. The corresponding hardness is 14 GPa at a maximum depth of 410 nm. A comparison between this value and that reported in the literature3 (250 GPa) leads to the conclusion that the fibers had lost more than 50% of their initial stiffness. With regard to the sintering and BN coating processes, it seems a priori that the former parameter has less influence for the following reasons: (1) the hot pressing was performed in a controled atmosphere (NJ, (2) at a relatively low pressure (= 30 MPa) and (3) at a temperature (1550°C) where the fibers exhibit the same longitudinal Young’s modulus as at room temperature? Moore8 has also shown that, depending on the BN coating temperature range, the ultimate tensile strength and the longitudinal Young’s modulus of Sic Nicalon fibers could be affected by the coating process. For the above reasons, an AFM examination was carried out to check the surface of the virgin HiNicalon fibers (without coating) and BN-coated fibers, both as-received from Nippon Carbon. An example of these two fiber topographies is given in Fig. 9 Figure 9 in the case of 2 pm and 0.5 pm total length scale, respectively. It can be seen that the BN-coated fiber shows a very rough surface, whereas the virgin fiber displays much smaller roughness amplitude. The roughness amplitude for BN-coated fibers is nearly double that of uncoated fibers when analyzing the highest magnification micrograph. To explain this difference, let us analyze the BN coating process. In fact, boron nitride is deposited on
1488 M. Drissi-Habti, K. Nakano Topography of a virgin fiber [0.5 uml 48487nm 48487nm 24244m Topography of a BN coated fiber (0. 5 umI 24244n nm l nm Fig. 9. AFM image(with a total length scale of 2 um), showing the topography of the virgin and coated fibers. As noticed, the coated fiber shows greater roughness amplitude the fibers at a temperature of 1500 C via the following the mechanical properties of Hi-Nicalon fibers can reaction only be achieved by means of uniaxial tensile tests on BCl、+NH3→BN+3HCl (11) both virgin and BN-coated fibers. Such a study is now In progress. Therefore, two explanations could be involved. The Moreover, in addition to the bn coating process, it first one is related to the atmosphere in which the as not been excluded that the combined effects of hot reaction is conducted. Indeed, if the atmosphere is not pressing and the high content of sintering additives controlled sufficiently, the excess fiber roughness could (5% alumina+5% yttria) can have a detrimental originate simply from an increase in the mean grain effect on the mechanical properties of the fibers Such size of the Hi-Nicalon fibers. Such explanation is in a case may involve a diffusion mechanism of accordance with previous results. The second explana- aluminum(coming from the sintering additives)into tion could be that an undesirable reaction between the the fibers. This point is also of major importance and fibers and the reactants occurs at 1500C which leads must be verified. to the evaporation of a gaseous product. In this case the mechanical properties of the fiber bulk could altered. This could explain the drastic decrease in the 4 CONCLUSIONS longitudinal Young's modulus of the fibers. In any The microstructure and mechanical propcrties case, on the basis of these preliminary results and Nicalon/ BN/silicon-nitride composites have observations, the BN coating conditions seem to have investigated. Results show that the longitudinal an effect in the increasing the roughness amplitude of Youngs modulus of the fibers in the composite the Hi-Nicalon fibers. It is worth noting, however. that decreases drastically. Preliminary results and observa final clarification of the effects of the bn coating on tions tend to incriminate the bn coating stage
1488 Act. Drissi-Habti, K. Nakano virgin fiber IO.5 pm] I Fig. 9. AFM image (with a total length scale of 2 pm), showing the topography of the virgin and coated fibers. As noticed, coated fiber shows greater roughness amplitude. the the fibers at a temperature of 1500°C via the following reaction: BC13 + NH3 + BN + 3HCl (11) Therefore, two explanations could be involved. The first one is related to the atmosphere in which the reaction is conducted. Indeed, if the atmosphere is not controlled sufficiently, the excess fiber roughness could originate simply from an increase in the mean grain size of the Hi-Nicalon fibers. Such explanation is in accordance with previous results.’ The second explanation could be that an undesirable reaction between the fibers and the reactants occurs at 1500°C which leads to the evaporation of a gaseous product. In this case, the mechanical properties of the fiber bulk could altered. This could explain the drastic decrease in the longitudinal Young’s modulus of the fibers. In any case, on the basis of these preliminary results and observations, the BN coating conditions seem to have an effect in the increasing the roughness amplitude of the Hi-Nicalon fibers. It is worth noting, however, that final clarification of the effects of the BN coating on the mechanical properties of Hi-Nicalon fibers can only be achieved by means of uniaxial tensile tests on both virgin and BN-coated fibers. Such a study is now in progress. Moreover, in addition to the BN coating process, it has not been excluded that the combined effects of hot pressing and the high content of sintering additives (5% alumina +5% yttria) can have a detrimental effect on the mechanical properties of the fibers. Such a case may involve a diffusion mechanism of aluminum (coming from the sintering additives) into the fibers. This point is also of major importance and must be verified. 4 CONCLUSIONS The microstructure and mechanical properties of HiNicalon/BN/silicon-nitride composites have been investigated. Results show that the longitudinal Young’s modulus of the fibers in the composite decreases drastically. Preliminary results and observations tend to incriminate the BN coating stage
Hi-Nicalon/BN/a-Si,N, ceramic-matrix composites not excluded that the processing REFERENCES route of the composite could also have an undesirable 1. Lamicq, P. and Boury, D, Ceramic matrix composite effect owing mainly to the diffusion of some elements parts design. In Proceedings of AGArD Workshop on such as aluminum. which could come from the tructural sintering additives. However, the composite showed Composites, Antalya, Turkey, 21-22 April 1993, pp. 12-1 extensive non-linear behavior, thus demonstrating that to12-12. he load transfer at the fiber/matrix interface is 2. Nakano. K. Kume, S, Sasaki d Saka, H Microstructure and mechanical properties of Hi-Nicalon acceptable fibers reinforced silicon nitride composite, Ceram. Eng One possible way to overcome the sensitivity of Hi Sci. Proc., in press Nicalon fibers to the Bn coating process is to replace 3. Ichikawa, T, Recent developments of the Sic fiber BCI by BF3 in reaction (11), the latter requiring a Nicalon and its composites, including properties of the lower temperature range(close to 1200C) C fiber Hi-Nicalon for ultra high temperature. Compos. Sci. TechnoL., 1994, 51, 135-144 Instrumented micro-indentation tests perforned on 4. Drissi-HabtL, M, Damage development and modt the matrix and fibers in the composite are very reduction in a unidirectional SiC-MAS L composite suitable and could provide an accurate estimate of the aterial tested under uniaxial tensile loading. Scripta mechanical properties of the composite. These pieces Metall.Maer,i995,33,967-973 of information could be helpful in evaluating the 5. Puttock, M.J. and Thwaite, E. G. Elastic compression of spheres and cylinders at point and line contact. Internal effects of the sintering process on the composite Report, CSIRO, Melbourne, 1969 constituents. In addition, only small samples are 6. Sneddon, L The relation between load and required and, at a laboratory scale, this is not a penetration in the axisymetric Boussinesq problem. Int negligible argument J. Eng, Sci,1965,1,47-52 7. Parlier, M, Passilly, B, and Sudre, O csented at AGARD Workshop on Introduction mics into Aerospace Structural Composites, Turkey, ACKNOWLEDGEMIENT 21-22 April 1993. AGARD Report 795, November 199 5-1to5-8. The authors wish to thank Dr miki from the national 8. Moore,A w, Facilities for continuous Cvd coating of Industrial Research Institute of Nagoya for his ceramIc fibers. In Materials Research Society Symposium ssistance in manipulating the atomic force Proceedings, VoL, 250. Materials Research Society, Pittsburg, Pennsylvania, 1992. microscope. 9. Matsuda. T, J Mater. Sci., 1989, 24, 2353
Hi-Nicalon/BN/cy-Si,N, ceramic-matrix composites 1489 Nevertheless, it is not excluded that the processing route of the composite could also have an undesirable effect owing mainly to the diffusion of some elements such as aluminum, which could come from the sintering additives. However, the composite showed extensive non-linear behavior, thus demonstrating that the load transfer at the fiber/matrix interface is acceptable. One possible way to overcome the sensitivity of HiNicalon fibers to the BN coating process is to replace BC13 by BF3 in reaction (ll), the latter requiring a lower temperature range (close to 1200°C). Instrumented micro-indentation tests performed on the matrix and fibers in the composite are very suitable and could provide an accurate estimate of the mechanical properties of the composite. These pieces of information could be helpful in evaluating the effects of the sintering process on the composite constituents. In addition, only small samples are required and, at a laboratory scale, this is not a negligible argument. ACKNOWLEDGEMENT The authors wish to thank Dr Miki from the National Industrial Research Institute of Nagoya for his assistance in manipulating the atomic force microscope. REFERENCES 1. 2. 3. 4. 5. 6. Lamicq, P. and Boury, D., Ceramic matrix composite parts design. In Proceedings of AGARD Workshop on Introduction of Ceramics into Aerospace Structural Composites, Antalya, Turkey, 21-22 April 1993, pp. 12-1 to 12-12. Nakano, K., Kume, S., Sasakai, H. and Saka, H., Microstructure and mechanical properties of Hi-Nicalon fibers reinforced silicon nitride composite. Ceram. Eng. Sci. Proc., in press. lchikawa, T., Recent developments of the SIC fiber Nicalon and its composites, including properties of the SIC fiber Hi-Nicalon for ultra high temperature. Compos. Sci. Technol., 1994,51,135-144. Drissi-Habti, M., Damage development and moduli reduction in a unidirectional SIC-MASL composite material tested under uniaxial tensile loading. Scripta Metall. Mater., 1995, 33,967-973. Puttock, M. J. and Thwaite, E. G., Elastic compression of spheres and cylinders at point and line contact. Internal Report. CSIRO, Melbourne, 1969. Sneddon, L. N., The relation between load and penetration in the axisymetric Boussinesq problem. Int. 1. Eng. Sci., 1965, 1,47-52. Parlier, M., Passilly, B. and Sudre, O., paper presented at AGARD Workshop on Introduction of Ceramics into Aerospace Structural Composites, Antalya, Turkey, 21-22 April 1993. AGARD Report 795, November 1993, pp. 5-l to 5-8. Moore, A. W., Facilities for continuous CVD coating of ceramic fibers. In Materials Research Society Symposium Proceedings, Vol. 250. Materials Research Society, Pittsburg, Pennsylvania, 1992. Matsuda, T., J. Mater Sci., 1989,24,2353