《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_TOUGHNESS AND MICROSTRUCTURAL DEGRADATION AT HIGH TEMPERATURE IN SIC FIBER-REINFORCED CERAMICS

Acta mater. Vol. 46. No. 7, pp. 2441-2453, 1998 blished by Elsevier Science Ltd. All rights reserved PI:Sl359-645497)00402 359645498s1900+0.00 TOUGHNESS AND MICROSTRUCTURAL DEGRADATION AT HIGH TEMPERATURE IN SIC FIBER-REINFORCED CERAMICS J LLORCA, M. ELICES and J. A. CELEMIN Department of Materials Science, Polytechnic University of Madrid, ETS de Ingenieros de Caminos. 28040 Madrid, spain Abstract- The fracture behavior of three different SiC fiber-reinforced ceramics at emperature in air is studied. The fracture properties were obtained by Flexure tests on notched heams and approach. The microstructi terfacial and fiher degradation, were analyzed in each material and related to the ms and to the degradation in toughness and fracture resistance. Finally, the prove the high temperature toughness of fiber-reinforced ceramics are briefly discussed 1 INTRODUCTION systems result in exceptionally high levels of tough mh ness and strength, their properties drop very quickl h combine the properti ramics with a above 800C in oxidizing environments due to damage tolerant, ductile behavior has been one of matrix softening and fiber/matrix reactions [5, 6] the most active research topics in materials science in the last two decades. The driving forces for these stable ceramic matrices, and several processing tech e the potential gains in efficien niques(chemical vapor infiltration direct-metal oxi- and in power output of thermal engines at higher dation, polymer pyrolysis, melt infiltration, etc. [] s,and the interest in tough amic ma- were developed to introduce the ceramic matrix terials increased as further improvements in the into the fiber preform These efforts were rewarded with a number of working temperature of Ni-based superalloys were fiber-reinforced ceramics(FRC) with excellent frac (around 1250C). In principle, ceramics are the ture resistance and non-linear stress-strain curve at ideal candidates to substitute Ni-based superalloys mation and fracture in these materials were an as high temperature structural materials. They exhi tail,and para bit very high melting points, excellent chemical micromechanical models which related the macro- stability and wear and creep resistance as well scopic behavior t he microstructural low density. Their main drawback in structural ap- parameters (8,9). However, most of this work was plications lies in their reduced ductility and fracture carried out in the ambient temperature regime, toughness, which makes Ic components prone where FRC are unlikely to be used. The amount of to catastrophic failure. The addition of fibers to cer- work on their elevated temperature performance is amics has been known for many years to be one still limited, and this is especially true for the frac- way of improving these properties. The develop- ture toughness and fracture resistance. Only a hand- ment of fiber- reinforced cements Instance, is ful of investigations on these topics are available in undoubtedly a good example of this approach the open literature [6, 10-16 It was demonstrated at the beginning of the 70s It should also be noted that the fracture beh hat high performance C bers could be successfully of FRC at elevated temperature can be considered incorporated into ceramic glasses to achieve high a litmus test because the presence of a pre-existing strength, tough composite materials [1-4]. These notch or crack facilitates the entry of oxygen into early developments were not carried further due to the composite and accelerates any degradation pro the oxidative instability of C fibers at high tempera- cess which might occur. In addition, it is necessary ture. As the new oxidation-resistant SiC bers de- to find whether the present materials are able to rived by pyrolysis of organometallic precursors achieve the minimum toughness values of became available at the end of this decade, a new 15 MPa. m at 1350C laid down by the industry to generation appeared of glass and glass-ceramic introduce these new materials as structural com- matrices reinforced with SiC fibers [5]. while these ponents in gas turbines [17]. The present knowledge

Pergamon Acia mater. Vol. 46, No. I, pp. 2441-2453, 1998 0 1998 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain PII: S1359-6454(97)00402-3 1359-6454/98 $19.00 + 0.00 TOUGHNESS AND MICROSTRUCTURAL DEGRADATION AT HIGH TEMPERATURE IN SIC FIBER-REINFORCED CERAMICS J. LLORCA, M. ELICES and J. A. CELEMiN Department of Materials Science, Polytechnic University of Madrid, ETS de Ingenieros de Caminos, 28040 Madrid, Spain Abstract-The fracture behavior of three different Sic fiber-reinforced ceramics at ambient and elevated temperature in air is studied. The fracture properties were obtained by flexure tests on notched beams and the increase in fracture resistance with crack length was determined using an equivalent elastic crack approach. The microstructural changes during elevated temperature exposure, which could be divided into interfacial and fiber degradation, were analyzed in each material and related to the fracture micromechan￾isms and to the degradation in toughness and fracture resistance. Finally, the different strategies to improve the high temperature toughness of fiber-reinforced ceramics are briefly discussed. 0 1998 Acta Metallurgica Inc 1. INTRODUCTION The development of new ceramic-based materials which combine the properties of ceramics with a damage tolerant, ductile behavior has been one of the most active research topics in materials science in the last two decades. The driving forces for these investigations were the potential gains in efficiency and in power output of thermal engines at higher temperatures, and the interest in tough ceramic ma￾terials increased as further improvements in the working temperature of Ni-based superalloys were hindered by the proximity to their melting point (around 1250°C). In principle, ceramics are the ideal candidates to substitute Ni-based superalloys as high temperature structural materials. They exhi￾bit very high melting points, excellent chemical stability and wear and creep resistance as well as low density. Their main drawback in structural ap￾plications lies in their reduced ductility and fracture toughness, which makes ceramic components prone to catastrophic failure. The addition of fibers to cer￾amics has been known for many years to be one way of improving these properties. The develop￾ment of fiber-reinforced cements, for instance, is undoubtedly a good example of this approach. systems result in exceptionally high levels of tough￾ness and strength, their properties drop very quickly above 800°C in oxidizing environments due to matrix softening and fiber/matrix reactions [5,6]. The emphasis was then focussed on using more stable ceramic matrices, and several processing tech￾niques (chemical vapor infiltration, direct-metal oxi￾dation, polymer pyrolysis, melt infiltration, etc. [7]) were developed to introduce the ceramic matrix into the fiber preform. These efforts were rewarded with a number of fiber-reinforced ceramics (FRC) with excellent frac￾ture resistance and non-linear stress-strain curve at ambient temperature. The mechanisms of defor￾mation and fracture in these materials were ana￾lyzed in detail, and parallel investigations provided micromechanical models which related the macro￾scopic behavior to the microstructural parameters [8,9]. However, most of this work was carried out in the ambient temperature regime, where FRC are unlikely to be used. The amount of work on their elevated temperature performance is still limited, and this is especially true for the frac￾ture toughness and fracture resistance. Only a hand￾ful of investigations on these topics are available in the open literature [6,10-161. It was demonstrated at the beginning of the 70s It should also be noted that the fracture behavior that high performance C fibers could be successfully of FRC at elevated temperature can be considered incorporated into ceramic glasses to achieve high a litmus test because the presence of a pre-existing strength, tough composite materials [ 141. These notch or crack facilitates the entry of oxygen into early developments were not carried further due to the composite and accelerates any degradation pro￾the oxidative instability of C fibers at high tempera- cess which might occur. In addition, it is necessary ture. As the new oxidation-resistant SIC fibers de- to find whether the present materials are able to rived by pyrolysis of organometallic precursors achieve the minimum toughness values of became available at the end of this decade, a new 15 MPa.,/E at 1350°C laid down by the industry to generation appeared of glass and glass-ceramic introduce these new materials as structural com￾matrices reinforced with SIC fibers [5]. While these ponents in gas turbines [17]. The present knowledge 2441

LORCA et al. SiC FIBER-REINFORCED CERAMICS FRC is outlined in this paper. Firstly, the fracture 25 MPa.m from room temperature to 1400cer of the elevated temperature fracture behavior of posite samples exhibited Kic values of toughness and the fracture resistance data at high More recently, Xu et al. [ 15] tested Si3N2-matrix temperature available in the literature are reviewed. SiCt composites in N2 atmosphere. To prepare the These results are completed with new experimental samples, Si3N4 powders were mixed with sintering data obtained in our laboratory on three different additives (Y2O3, Al2O3 and MgO). Layers of pow FRC, and special emphasis is laid on examining the ders and SiC SCS-6 monofilaments were stacked causes of the high temperature degradation and alternatively and hot pressed in a N2 atmosphere their influence on the fracture micromechanisms. Fracture tests were performed by fiexure of notched Finally, the current strategies to improve the high beams, with the notch plane perpendicular to the temperature fracture behavior of FRC are briefly reinforcing fibers. The unidirectional composite discussed exhibited high Kic values from ambient temperature to 1200C; about 36 MPa. m and 50 MPa/m,re- 2. SUMMARY OF PREVIOUS RESULTS spectively for samples with 14% and 29% fraction of fibers when measured as vEJe.Direct Few experimental results have been published on measurement of Kic gave much lower results, 16 the high temperature toughness and crack growth and 10 MPa vm respectively. Xu et al. [15]sus- resistance of FRC, although other mechanical prop- pected that the direct measurements underestimated erties such as load-displacement curves, flexural the composite toughness. Nevertheless, the point in strength or creep behavior have been reported since question is not the absolute values but the lack of the eighties. Such fracture tests are difficult to per- degradation inside the testing temperature interval form and even more difficult to interpret; perform- regardless of the technique used to determine the ing stable tests at high temperatures with toughness precracked samples and extracting from them The results of fracture tests performed in inert meaningful fracture parameters when anisotropy, environments are summarized in Fig. 1. It is nonlinear behavior and inelastic load-displacement remarkable that no detectable degradation of frac- culties, fracture toughness figures and R-curves peratures was loulle n the interval of testing tem- are present is not an easy task. Despite these diffi- ture toughness withi based on linear elastic fracture analysis have beer Fracture tests at high temperatures in air are published. In this respect, the data reported in lit- more recent, and contrary to tests in inert environ- erature must be considered with care. In fact they ments, they showed a slow fracture toughness are usually presented more as a simple way to com- degradation with increasing temperature, Nair and pare homologous materials or to optimize proces- Wang [12] tested a bidirectional woven SiC/Sic sing conditions than as Intrinsic material composite in air. The composite was processed by chemical vapor infiltration of Sic into a woven Farly reported fracture toughness tests at high Nicalon SiC fiber preform. The fiber content was temperatures with SiC-fiber/ ceramic-matrix compo 40% in volume. Fracture tests were performed on sites were performed in inert environments, and compact tension specimens with a notch depth to showed almost constant values from room tempera- sample length of 0.5. KIc values ranged from about ture up to more than 1000.C, Brennan and Prewo [6] measured the fracture properties of sic. fiber glass ceramic composites in Ar atmosphere. TESTS IN INERT ENVIRONMENTS The SiC fiber( Nippon Carbon Co )diameters ran- sI2N,siC [15] The glass matrix was essentially the Corning Glass commercial 9608 lithium-alumino-silicate (LAS). E except that the TiOz nucleating agent laced SI, N, /SIC [15] by a Z O2 nucleating agent. The volume fraction of 2 32887888222445 point bending The 0 /90 cross-plied samples exhi 1 bit Kic values of over 10 MPa,m from ambient 1000°C ernhart et al. [10]and LaImicy et ul. [ll] tested SiC/SiCr composites in an Ar/H2 atmosphere. The LAS Glass/SIC [6 directional flat laminates were manufactured by acking the Nicalon SiC fiber fabrics and then 200 1600 introducing the SiC matrix by a chemical vapor infiltration proccss. Fracture tests also per- Fig. 1. Apparent toughness at initiation as a function formed by flexure of notched beams and the com- temperature in various FRC tested in inert environt

2442 LLORCA et al.: Sic FIBER-REINFORCED CERAMICS of the elevated temperature fracture behavior of FRC is outlined in this paper. Firstly, the fracture toughness and the fracture resistance data at high temperature available in the literature are reviewed. These results are completed with new experimental data obtained in our laboratory on three different FRC, and special emphasis is laid on examining the causes of the high temperature degradation and their influence on the fracture microm~hanisms. Finally, the current strategies to improve the high temperature fracture behavior of FRC are briefly discussed. 2. SUMMARY OF PREVIOUS RESULTS Few experimental results have been published on the high temperature toughness and crack growth resistance of FRC, although other mechanical prop￾erties such as load-displacement curves, flexural strength or creep behavior have been reported since the eighties. Such fracture tests are difficult to per￾form and even more difficult to interpret; perform￾ing stable tests at high temperatures with precracked samples and extracting from them meaningful fracture parameters when anisotropy, nonlinear behavior and inelastic load-displacement are present is not an easy task. Despite these diffi￾culties, fracture toughness figures and R-curves based on linear elastic fracture analysis have been published. In this respect, the data reported in lit￾erature must be considered with care. In fact they are usually presented more as a simple way to com￾pare homologous materials or to optimize proces￾sing conditions than as intrinsic material characteristics. Early reported fracture toughness tests at high temperatures with Sic-fiber/ceramic-matrix compo￾sites were performed in inert environments, and showed almost constant values from room tempera￾ture up to more than 1000°C. Brennan and Prewo [6] measured the fracture properties of SiC￾fiber/glass ceramic composites in Ar atmosphere. The SIC fiber (Nippon Carbon Co.) diameters ran￾ged from 5 to 60 pm with an average size of 12 pm. The glass matrix was essentially the Corning Glass commercial 9608 lithium-alumina-silicate (LAS), except that the TiOz nucleating agent was reptaced by a ZrOz nucleating agent. The volume fraction of Sic fibers was about 50%. The fracture tests were carried out on notched beams subjected to three￾point bending. The 0”/90” cross-plied samples exhi￾bit Kit values of over 10 MPa.&ii from ambient temperature to 1000°C. Bernhart ef al. [lo] and Lamicq et al. [l 11 tested SiC/SiCr composites in an Ar/Hz atmosphere. The bidirectional flat laminates were manufactured by stacking the Nicalon SIC fiber fabrics and then introducing the Sic matrix by a chemical vapor infiltration process. Fracture tests were also per￾formed by flexure of notched beams and the com￾posite samples exhibited KIc values of over 25 MPa.6 from room temperature to 1400°C. More recently, Xu et nl. [IS] tested Signs-rnat~x~ SiCr composites in Nz atmosphere. To prepare the samples, S&N4 powders were mixed with sintering additives (YzOs, Al203 and MgO). Layers of pow￾ders and SIC SCS-6 monofilaments were stacked alternatively and hot pressed in a N2 atmosphere. Fracture tests were performed by flexure of notched beams, with the notch plane perpendicular to the reinforcing fibers. The unidirectional composite exhibited high KIc values from ambient temperature to 1200°C; about 36 MPa.&ii and 50 MPa.,/iii, re￾spectively for samples with 14% and 29% volume fraction of fibers when measured as m. Direct measurement of KIc gave much lower results, 16 and 10 MPa.& respectively. Xu et al. [15] sus￾pected that the direct measurements underestimated the composite toughness. Nevertheless, the point in question is not the absolute values but the lack of degradation inside the testing temperature interval regardless of the technique used to determine the toughness. The results of fracture tests performed in inert environments are summarized in Fig. 1. It is remarkable that no detectable degradation of frac￾ture toughness within the interval of testing tem￾peratures was found. Fracture tests at high temperatures in air are more recent, and contrary to tests in inert environ￾ments, they showed a slow fracture toughness degradation with increasing temperature. Nair and Wang [12] tested a bidirectional woven SiC/SiCr composite in air. The composite was processed by chemical vapor infiltration of Sic into a woven Nicalon SIC fiber preform. The fiber content was 40% in volume. Fracture tests were performed on compact tension specimens with a notch depth to sample length of 0.5. Z&c values ranged from about 0 400 600 1200 1600 Temperature (‘C) Fig. 1. Apparent toughness at initiation as a function of temperature in various FRC tested in inert environments

LLORCA et al.: SiC FIBER-REINFORCED CERAMICS TESTS IN AIR ufactured by stacking several of nicalon (0. 1 um)of pyrolytic C was deposited on the fiber surface, and the sic matrix was introduced into the SIC/SIC [13] preform by the chemical vapor infiltration form of prismatic bars of 10 mm x 3 mm cross-sec tIon Porosity was around 7-8% The second material was an AlyO3 matrix bidirec tionally reinforced with 37 vol. Nicalon SiC fibers. The preform was manufactured by stacking several layers of Nicalon harness satin weave fabric SIC/SIC [12] The fibers were coated sition with a thin layer of BN (20.3 um) and after wards with a thicker layer of Sic (in the range 3- Temperature('C into the preform by a direct metal oxidation process [19]. The composite was received in the Fig. 2. Apparent toughness at initiation as a function of form of prismatic bars of 10 mm x 3 mm cross-sec temperature in various FRC tested in air. 15 MPa m at ambient temperature to 11 MPa: m The third composite was formed by a zrSio4 at 1200.C. Another bidirectional woven composite matrix uniaxially reinforced with 25 vol %SCS-6 produced by chemical vapor infiltration of SiC into SiC monofilaments made by chemical vapor depo- the Nicalon Sic fiber preform was tested by sition of Sic on a C core. Afterwards they were Gomina and Chermant[13]. The volume fraction of coated with a carbon layer of around 3 um in thick- Sic fibers was about 38%. Fracture tests were car- ness, attaining a final diameter of 142 um. The com- ried out by three-point flexure of notched beams. posite was manufactured by aligning SiC filaments Kic values ranged from about 28 MPa m at ambi. and then incorporating the matrix powder around Fareed et al. [14 measured the toughness of out by hot-pressing the matrix and the fibers in a bidirectional woven Al2O3/SiCr composites in air. flowing nitrogen atmosphere [20]. The composite he material was manufactured by direct metal oxi- plates were fully dense and prismatic bars of dation of molten al which was introduced into the 4 mm x. mm cross-section were machined from fiber preform. The volume fraction of Nicalon Sic the plates fibers was about 35%. Fracture tests were per determined by three-point bend tests on notched beams. Kic values ranged from 28 MPa- m at specimens. Notches were machined in the bars with ambient temperature to 15 MPa. m at 1400.C. a thin diamond wire, leading to a notch radius of Figure 2 summarizes all these results of fract 150 Am. The notch length was around 2 mm for the tests in air. It is seen that a gradual degradation first two composites and around 1. 2 mm for the more noticeable from 700C pears as tempera- third one. The notched hars were tested in a cer- ture increases. It is evident. however. that the amic three-point bend testing fixture with either amount of experimental data currently available is 50 mm(for SiC/SiC and Al203/SiC)or 40mm(for y limited, and as a result, there is not conclusive the zrsiO4/SiC) loading span and the specimen evidence of the mechanisms responsible for the placed edgewise on the fixture degradation of the fracture properties of FRC in air The specimen and the fixture, placed in the hi at elevated temperature. This is the subject of the temperature furnace, were loaded through two following sections, where the role of the alumina rods connected to the actuator and to the and fiber degradation in lowering fracture ad cell, respectively, of a servo-mechanical testing ness with temperature is studied in three machine. The external end of one rod was water- FRC cooled to avoid overheating the load cell. The heat- ing rate was 12 C per min and the specimen was 3. MATERIALS AND EXPERIMENTAI test temperature for at least 30 min TECHNIQUES prior to testing. All the tests were performed in air under stroke control, with a cross-head speed of This investigation of the fracture behavior at el- 50 um per min The load (P)and the cross- head dis- evated temperature was performed in three different placement of the testing machine relative to the composites. The first material was made up of a frame( O)were recorded continuously during the SiC matrix bidirectionally (0-90) reinforced with Lests, the latter through a linear-variable differential 35 voL. Nicalon SiC fibers. The preform was man- transducer placed outside the furnace

LLORCA et al.: SiC FIBER-REINFORCED CERAMICS 2443 TESTS IN AIR 40/ 0”” ” ” “‘1 0 A00 600 1200 1600 Temperature (‘C) Fig. 2. Apparent toughness at initiation as a function of temperature in various FRC tested in air. 15 MPa.Jiii at ambient temperature to 11 MPa.6 at 1200°C. Another bidirectional woven composite produced by chemical vapor infiltration of SIC into the Nicalon SIC fiber preform was tested by Gomina and Chermant [13]. The volume fraction of SIC fibers was about 38%. Fracture tests were car￾ried out by three-point flexure of notched beams. Km values ranged from about 28 MPa.,/iii at ambi￾ent temperature to 10 MPa.,/iii at 1000°C. Fareed et al. [14] measured the toughness of bidirectional woven AlzOs/SiCr composites in air. The material was manufactured by direct metal oxi￾dation of molten Al which was introduced into the fiber preform. The volume fraction of Nicalon SIC fibers was about 35%. Fracture tests were per￾formed using the chevron notch technique on beams. Krc values ranged from 28 MPa.,/iii at ambient temperature to 15 MPa.6 at 1400°C. Figure 2 summarizes all these results of fracture tests in air. It is seen that a gradual degradation - more noticeable from 700°C - appears as tempera￾ture increases. It is evident, however, that the amount of experimental data currently available is very limited, and as a result, there is not conclusive evidence of the mechanisms responsible for the degradation of the fracture properties of FRC in air at elevated temperature. This is the subject of the following sections, where the role of the interface and fiber degradation in lowering fracture tough￾ness with temperature is studied in three different FRC. 3. MATERIALS AND EXPERIMENTAL TECHNIQUES This investigation of the fracture behavior at el￾evated temperature was performed in three different composites. The first material was made up of a SIC matrix bidirectionally (O”-90”) reinforced with 35 vol.% Nicalon Sic fibers. The preform was man￾ufactured by stacking several layers of Nicalon plain satin weave fabric. A very thin layer (~0.1 pm) of pyrolytic C was deposited on the fiber surface, and the SIC matrix was introduced into the preform by the chemical vapor infiltration process [18]. The composite was received in the form of prismatic bars of 10 mm x 3 mm cross-sec￾tion. Porosity was around 7-8%. The second material was an A1203 matrix bidirec￾tionally reinforced with 37 vol.% Nicalon SIC fibers. The preform was manufactured by stacking several layers of Nicalon harness satin weave fabric. The fibers were coated by chemical vapor depo￾sition with a thin layer of BN (20.3 pm) and after￾wards with a thicker layer of SiC (in the range 3- 4 pm) onto the BN. The matrix was then introduced into the preform by a direct metal oxidation process [19]. The composite was received in the form of prismatic bars of 10 mm x 3 mm cross-sec￾tion. Porosity was around 7-8%. The third composite was formed by a ZrSiOA￾matrix uniaxially reinforced with 2.5 vol.% SCS-6 Sic monofilaments made by chemical vapor depo￾sition of SIC on a C core. Afterwards they were coated with a carbon layer of around 3 pm in thick￾ness, attaining a final diameter of 142 pm. The com￾posite was manufactured by aligning Sic filaments and then incorporating the matrix powder around the uniaxial preform. The consolidation was carried out by hot-pressing the matrix and the fibers in a flowing nitrogen atmosphere [20]. The composite plates were fully dense and prismatic bars of 4 mm x 1.3 mm cross-section were machined from the plates. The fracture behavior of the composites was determined by three-point bend tests on notched specimens. Notches were machined in the bars with a thin diamond wire, leading to a notch radius of 150 pm. The notch length was around 2 mm for the first two composites and around 1.2 mm for the third one. The notched bars were tested in a cer￾amic three-point bend testing fixture with either 50 mm (for Sic/Sic and A120s/SiC) or 40 mm (for the ZrSiO,/SiC) loading span and the specimen placed edgewise on the fixture. The specimen and the fixture, placed in the high temperature furnace, were loaded through two alumina rods connected to the actuator and to the load cell, respectively, of a servo-mechanical testing machine. The external end of one rod was water￾cooled to avoid overheating the load cell. The heat￾ing rate was 12°C per min and the specimen was held at the test temperature for at least 30 min prior to testing. All the tests were performed in air under stroke control, with a cross-head speed of 50 pm per min. The load (P) and the cross-head dis￾placement of the testing machine relative to the frame (6) were recorded continuously during the tests, the latter through a linear-variable differential transducer placed outside the furnace

LLORCA et al. SiC FIBER-REINFORCED CERAMICS The f avior was obtained from the P-8 have fracture energies around 20 J/ m2, so 5 J/m2 curve using an equivalent elastic crack approach. can be considered an upper limit for the debond This method assumes that the energy release rate of energy. This condition is achieved by coating the the composite during quasi-static crack propagation fibers with thin layers of either C, bn or Mo is equal to that of a linear elastic material which which avoid chemical reactions between the matrix presents the same P-8 curve. The increase in frac- and the fibers during processing, and provide the ture resistance with crack length, R, for the linear weak interfaces with very low fracture energies elastic material can be computed as Experimental observations on different FRC have p2 ac shown that crack deflection and fiber/matrix sliding R (1) give rise to the redistribution of stresses(and thus he dissipation of energy) around a notch or other where B is the specimen thickness and C the speci- strain concentration site by two fundamental mech- men compliance, which is obtained directly from anisms: distributed matrix cracking and fiber fa the P-a curve in an elastic material because there involving pull-out 9]. In the first case, multiple are no residual displacements at zero load. It should mode I matrix cracks grow from the notch, and be noted that this method underestimates the frac- may even extend across the net section prior te ure resistance because the non-linear mechanisms fiber failure If the second mechanism is dominant of energy dissipation are not included. Various the specimen fracture occurs by the propagation of recent studies have shown, however, that the energy a dominant mode I crack from the notch, with fiber release rate and the toughness values provided by failures occurring as the crack extends. Stress redis he equivalent elastic crack approach are compar- tribution is provided by the tractions exerted on the able to those obtained with more sophisticated crack by the failed fibers as they are pulled out methods [21-23]. The accuracy of this approach from the matrix. It should be noted that a third was considered to be sufficient for the purpose of mechanism of energy dissipation by shear damage this investigation, which was to establish the re- in the matrix was found in several C matrix ationship between the microstructural changes composites [9]. However, this damage mode is not occurring during high temperature testing and the analyzed here because these materials are not suit- fracture response of the composites able for high temperature application in oxidizing Once broken, the composites were examined atmospheres sing scanning electron microscopy and energy-dis- Both mechanisms of energy dissipation lead to pensive X-ray microanalysis to determine the micro- the development of a fibrous fracture surface, where structural changes during high temperature the fibers protrude from the matrix. This mor- exposure and the associated failure mechanisms. phology was observed in the fracture surfaces of the The fracture surfaces were first analyzed and the SI C/SiC composite tested at room temperature, as pecimens were then sliced in the longitudinal direc- depicted in Fig. 3(a). a closer examination tion (perpendicular to the fracture surface)with a (Fig 3(b)showed the thin layer of pyrolytic Con low speed diamond saw. The surfaces were polished the lateral fiber surface, indicating that fiber/ matrix successively on diamond cloths of 40, 9, 3 and 1 um debonding took place between the C coating and grain size and finally on alumina of 0. 3 um grain the SiC matrix. The panorama changed completely size. They were cleaned for 30 min by ultrasound in when the material was tested at 1200"C, as is show acetone to remove the alumina from polishing, and in Fig. 3(a). The fracture surfaces at 1200 C wer sputtered with Au-Pd for three minutes before predominantly flat, and the fibers were fractured in being observed in the scanning electron microscope, the crack plane as the matrix crack propagated into them(Fig. 3(C), indicating that fiber/ matrix decohe sion did not take place. This behavior is typical of 4. MICROSTRUCTURAL CHIANGES AND C and Mo coatings which are not stable in oxidiz. ACTURE RESISTANCE ing atmospheres above 700C, and form volatile oxides. The elimination of the coating creates a gap 4.. Interface degi at the interface, which is eventually filled by the for The critical factor to obtain a tough FRC lies in mation of a glassy phase by either matrix or fiber the nature of the fiber matrix interface. If the fibers oxidation, which bonds the fiber to the matrix are strongly bonded to the matrix, a crack Under such conditions, crack deflection cannot take nucleated in the matrix breaks the fibers as it pro- place and the composite fails in a brittle galEs,and the composite is as brittle as the fashion [13, 16 matrix On the contrary, crack deflection and fiber/ The differences in the fracture behavior between matrix sliding at the interface occur when the fibers ambient and elevated temperature are shown in ire weakly bonded to the matrix and this debond Fig. 4, where the fracture resistance is plotted as energy is lower than approximately one fourth of function of the crack length increment, Aa(normal- the fibcr fracture cncrgy [24]. Otherwise the crack izcd by the specimen width, w). The apparent propagates through the fiber, Most ceramic fibers toughness at initiation at 20C was three times

2444 LLORCA et al.: Sic FIBER-REINFORCED CERAMICS The fracture behavior was obtained from the P-6 curve using an equivalent elastic crack approach. This method assumes that the energy release rate of the composite during quasi-static crack propagation is equal to that of a linear elastic material which presents the same P-6 curve. The increase in frac￾ture resistance with crack length, R, for the linear elastic material can be computed as RX: (1) where B is the specimen thickness and C the speci￾men compliance, which is obtained directly from the P-6 curve in an elastic material because there are no residual displacements at zero load. It should be noted that this method underestimates the frac￾ture resistance because the non-linear mechanisms of energy dissipation are not included. Various recent studies have shown, however, that the energy release rate and the toughness values provided by the equivalent elastic crack approach are compar￾able to those obtained with more sophisticated methods [21-231. The accuracy of this approach was considered to be sufficient for the purpose of this investigation, which was to establish the re￾lationship between the microstructural changes occurring during high temperature testing and the fracture response of the composites. Once broken, the composites were examined using scanning electron microscopy and energy-dis￾persive X-ray microanalysis to determine the micro￾structural changes during high temperature exposure and the associated failure mechanisms. The fracture surfaces were first analyzed and the specimens were then sliced in the longitudinal direc￾tion (perpendicular to the fracture surface) with a low speed diamond saw. The surfaces were polished successively on diamond cloths of 40, 9, 3 and 1 pm grain size and finally on alumina of 0.3 pm grain size. They were cleaned for 30 min by ultrasound in acetone to remove the alumina from polishing, and sputtered with Au-Pd for three minutes before being observed in the scanning electron microscope. 4. MICROSTRUCTURAL CHANGES AND FRACTURE RESISTANCE 4.1. Interface degradation The critical factor to obtain a tough FRC lies in the nature of the fiber/matrix interface. If the fibers are strongly bonded to the matrix, a crack nucleated in the matrix breaks the fibers as it pro￾pagates, and the composite is as brittle as the matrix. On the contrary, crack deflection and fiber/ matrix sliding at the interface occur when the fibers are weakly bonded to the matrix and this debond energy is lower than approximately one fourth of the fiber fracture energy [24]. Otherwise the crack propagates through the fiber. Most ceramic fibers have fracture energies around 20 J/m*, so 5 J/m2 can be considered an upper limit for the debond energy. This condition is achieved by coating the fibers with thin layers of either C, BN or MO, which avoid chemical reactions between the matrix and the fibers during processing, and provide the weak interfaces with very low fracture energies. Experimental observations on different FRC have shown that crack deflection and fiber/matrix sliding give rise to the redistribution of stresses (and thus the dissipation of energy) around a notch or other strain concentration site by two fundamental mech￾anisms: distributed matrix cracking and fiber failure involving pull-out [9]. In the first case, multiple mode I matrix cracks grow from the notch, and may even extend across the net section prior to fiber failure. If the second mechanism is dominant, the specimen fracture occurs by the propagation of a dominant mode I crack from the notch, with fiber failures occurring as the crack extends. Stress redis￾tribution is provided by the tractions exerted on the crack by the failed fibers as they are pulled out from the matrix. It should be noted that a third mechanism of energy dissipation by shear damage in the matrix was found in several C matrix composites [9]. However, this damage mode is not analyzed here because these materials are not suit￾able for high temperature application in oxidizing atmospheres. Both mechanisms of energy dissipation lead to the development of a fibrous fracture surface, where the fibers protrude from the matrix. This mor￾phology was observed in the fracture surfaces of the Sic/Sic composite tested at room temperature, as depicted in Fig. 3(a). A closer examination (Fig. 3(b)) showed the thin layer of pyrolytic C on the lateral fiber surface, indicating that fiber/matrix debonding took place between the C coating and the Sic matrix. The panorama changed completely when the material was tested at 12OO”C, as is shown in Fig. 3(a). The fracture surfaces at 1200°C were predominantly flat, and the fibers were fractured in the crack plane as the matrix crack propagated into them (Fig. 3(c)), indicating that fiber/matrix decohe￾sion did not take place. This behavior is typical of C and MO coatings, which are not stable in oxidiz￾ing atmospheres above =7OO”C, and form volatile oxides. The elimination of the coating creates a gap at the interface, which is eventually filled by the for￾mation of a glassy phase by either matrix or fiber oxidation, which bonds the fiber to the matrix. Under such conditions, crack deflection cannot take place and the composite fails in a brittle fashion [13,16]. The differences in the fracture behavior between ambient and elevated temperature are shown in Fig. 4, where the fracture resistance is plotted as function of the crack length increment, Aa (normal￾ized by the specimen width, I+‘). The apparent toughness at initiation at 20°C was three times

LlORCA et al. siC FIBER-REINFORCED CERAMICS 2445 Fig 3. (a)Fracture surfaces of the SiC/SiCr composite at 20C (left)and 1200 c (right).(b) SiC fiber pulled out from the matrix. The pyrolytic C coating on the fiber surface is marked with an arrow.(c) SiC fibers bonded to the matrix and broken in the crack plane during fracture at 1200.c higher than at 1200C, and in addition, the interface produced an almost fat R-curve, the frac resistance increased rapidly with crack ture resistance similar to that of the matrix ambient temperature, as more energy is di dependent of the crack in the crack wake through fiber pull-out The degradation mechanism presented above contrary,the high temperature embrittlement be partially suppressed by using coatings with better induced by the development of a strong fiber/ matrix oxidation resistance, such as BN. The oxidation of

higher ambie in the contra induce LLORCA et al.: Sic FIBER-REINFORCED CERAMICS 2445 Fig. 3. (a) Fracture surfaces of the SiC/SiCr composite at 20°C (left) and 1200°C (right). (b) Sic fiber pulled out from the matrix. The pyrolytic C coating on the fiber surface is marked with an arrow. (c) Sic fibers bonded to the matrix and broken in the crack plane during fracture at 1200°C. than at 12OO”C, and in addition, the fracture interface produced an almost flat R-curve, the frac￾nce increased rapidly with crack length at ture resistance being similar to that of the matrix nt temperature, as more energy is dissipated and almost independent of the crack length. crack wake through fiber pull-out. On the The degradation mechanism presented above can .ry, the high temperature embrittlement be partially suppressed by using coatings with better :d by the development of a strong fiber/matrix oxidation resistance, such as BN. The oxidation of

LLORCA er al. SiC FIBER-REINFORCED CERAMIC SIC/SIC BN also leads to the formation of B,Os, a glassy phase which could act as a microcrack sealant, lim iting the ingress of oxygen to the interface. Another E strateg egy to avoid interfacial oxidation is the deposit- g of multilayer coatings with different mechanical coatings (where the bn provides a weak fiber/ matrix interface while the SiC outer coating pro- tects the BN from oxidation) were resistant to gross 苏巴 chemical degradation during exposure to 1100- 1200"C for extended periods of time [16 25].An example of this behavior is presented in the follow- vent oxidation was to apply an additional Sic surface coating to the whole specimen prior to ting [26] 0.3 0.4 Nevertheless, none of the methods mentioned in Aa/w the previous paragraph can completely exclude the Fig. 4. Fracture resistance of the SiC/SiCr composite at local oxidation of the interface or its reaction with 20C and 1200.C as a function of the crack length in either the matrix or the fibers. These changes may not hinder fiber/matrix debonding but give rise to ZrsiO4 SIC ZrSIo Fig. 5. Fracture surfaces of the Zrsio4/SiCr composite showing fiber pull-out. (a)20"C, The C coating at the interface is clearly visible. (b)1400 C. The C coating has disappeared

2446 LLORCA et al.: Sic FIBER-REINFORCED CERAMICS SE I Sic OL, ” ” ” ” ” ” ” 1 0 0.1 0.2 0.3 0.4 As/W Fig. 4. Fracture resistance of the SiC/SiCf composite at 20°C and 1200°C as a function of the crack length incre￾ment, Au (normalized by the specimen width, IV). BN also leads to the formation of B,Os, a glassy phase which could act as a microcrack sealant, lim￾iting the ingress of oxygen to the interface. Another strategy to avoid interfacial oxidation is the deposit￾ing of multilayer coatings with different mechanical and chemical properties. For instance, dual SiC/BN coatings (where the BN provides a weak fiber/ matrix interface while the Sic outer coating pro￾tects the BN from oxidation) were resistant to gross chemical degradation during exposure to 1 lOO& 1200°C for extended periods of time [16,25]. An example of this behavior is presented in the follow￾ing section. Finally, another method used to pre￾vent oxidation was to apply an additional SIC surface coating to the whole specimen prior to testing [26]. Nevertheless, none of the methods mentioned in the previous paragraph can completely exclude the local oxidation of the interface or its reaction with either the matrix or the fibers. These changes may not hinder fiber/matrix debonding but give rise to Fig. 5. Fracture surfaces of the ZrSiO.$SiCf composite showing fiber pull-out. (a) 20°C. The C coating at the interface is clearly visible. (b) 1400°C. The C coating has disappeared

LLORCA et al. SiC FIBER-REINFORCED CERAMICS 20%c 200 ENERGY ( Kev) ENERGY (KeV Fig. 6. Longitudinal sections of the ZrSio4/SiCr composite after testing at different temperatures. (a) 20°C(b)1000°C.(c)1400C.The sive substitution of the C layer by sioz at the interface is clearly seen. Energy spectra of the interface obtained by X-ray microanalysis after testing at 20C and

LLORCA et al.: Sic FIBER-REINFORCED CERAMICS 2447 4OOr- 300- z f zoo- 8 loo￾o- 0 2ooc :(W 0.5 1 1.5 2 2.5 ENERQY (KeV) 0 0 0.5 E’NEFldf (Ket) 2.5 3 Fig. 6. Longitudinal sections of the ZrSi04/SiCr composite after testing at different temperatures. (a) 20°C (b) 1000°C. (c) 1400°C. The progressive substitution of the C layer by Si02 at the interface is clearly seen. Energy spectra of the interface obtained by X-ray microanalysis after testing at 20°C and 1400°C are shown together with the corresponding micrographs

LLORCA el al. SiC FIBER-REINFORCED CERAMICS important variations in interface roughness and fric ZrSiO./SiC tion coefficient which modify substantially the inter- facial sliding resistance, t. Reliable data on the 20°c interfacial sliding resistance at high temperature are they are usually based on indirect observations, but a they seem to indicate that t increases with tempera 00C ture due to these mechanisms. For instance, Singh 8 et al. [26] considering the average pull-out lengths, computed an increase in from 13 MPa at ambient 30 mperature to 18 MPa at 1300C in a SiC/sic g composite. p °··s The influence of the changes in the interfacial gness can be und developed [8, 9). For vcry large specimens, the frac ture resistance of FRC increases with crack length of the 0.3 to a maximum,R。, men geometry [27]. Curtin [8] obtained an ex- pression for roo as of the temperature on the crack growth resistand Roo=R(m)fo LoP/(om+//r%)/om+1) ZrSiO4 matrix composite reinforced with %o sCs-6 SiC monofilament The gradual changes at the interfac ce were here f is the fiber volume fraction, t the fiber/ observed in the fracture resistance curves plotted in matrix interfacial sliding resistance, r the fiber Fig. 7. Both the initial toughness and the increase radius, and do and m are, respectively, the charac- in fracture resistance with crack length decreased as teristic strength and the Weibull modulus of the the temperature increased. This progressive degr fiber strength distribution. Lo is a constant with dation sccms to bc associatcd with the Sio2 layer dimensions of length associated to do, and AR is a growth at the interface, which did not prevent fiber non-dimensional function of the Weibull modulus matrix debonding from taking place but changed of the fibers. This expression shows that the the fiber/matrix interaction during sliding. In fact, fracture resistance scales with the interfacial sliding fibers pulled out from the matrix were observed resistance according to t-m)(+ m). The Weibull even in the fracture surfaces of the specimen tested modulus of the fibers is usually in the range at 1400C(Fig. 5(b)) 3-626, 28 and thus Roo decreases as It should also be noted that changes in t with temperature may be induced even when no chemical A good example of this progressive degradation reactions occur at the interface. Residual stresses of the interface with tcmpcrature was found in the arise upon cooling from the processing temperature ZrSiO4/SiC material. The C coating of around 3 um as a result of the thermal expansion mismatch in thickness acted as the"mechanical fuse which between the matrix (m) and the fibers (ad If induced crack deflection and relative fiber/matrix m-f>0, the residual stresses at the interface are sliding at ambient temperature, as is shown in the compressive at ambient temperature. They are fracture surface of the specimens tested at ambient relieved as the temperature increases, reducing the temperature (Fig. 5(a). This coating has disap- normal stresses at the interface and thus t. The peared, however, in the specimens tested at 1400 c opposite trend is expected when am-a<0, and in although fiber pull-out was still observed on the fact, both behaviors were experimentally confirmed fracture surfaces( Fig. 5(b)). The longitudinal sec- Celemin et al. [29] estimated from average matrix tions of specimens tested at different temperatures crack spacing measurements a decrease in t from showed a progressive substitution of the C coating 24 MPa to 18 MPa in an Al2O3/SiCr composite (Fig. 6(a)) by another layer, which started to grow from 20 to 1200"C, while Morscher al.[30 from the matrix at temperatures above 800%C. This obtained from fiber push-out experiments 1000 c(Fig. 6(b))and had replaced it completely to 12-32 MPa at 1300"C in a SiaN/SiCr composit beyond 1300 C(Fig. 6(c)). The energy spectra of the interface at 20 C and 1400@C obtained by X-ray 4.2. Fiber degradation microanalysis are also shown in Figs 6(a)and (c), Tensile tests of the available oxide and non-oxide nd they indicate that the new layer formed during fibers have shown that the degradation of their high tcmpcraturc cxposurc was mainly made up of mcchanical properti starts ab Sio Polycrystalline oxide fibers showed evidence of soft

2448 LLORCA ef al.: Sic FIBER-REINFORCED CERAMICS important variations in interface roughness and fric￾tion coefficient which modify substantially the inter￾facial sliding resistance, r. Reliable data on the interfacial sliding resistance at high temperature are scarce because of the experimental difficulties, and they are usually based on indirect observations, but they seem to indicate that r increases with tempera￾ture due to these mechanisms. For instance, Singh et al. [26] considering the average pull-out lengths, computed an increase in r from 13 MPa at ambient temperature to 18 MPa at 1300°C in a Sic/Sic composite. The influence of the changes in the interfacial sliding resistance on the toughness can be under￾stood through the micromechanical models recently developed [8,9]. For very large specimens, the frac￾ture resistance of FRC increases with crack length up to a maximum, R,, independent of the speci￾men geometry [27]. Curtin [8] obtained an ex￾pression for R, as cm-l)/(m+l) R, = ~&z)J[~;;L,,]~‘@“+‘) (2) where f is the fiber volume fraction, t the fiber/ matrix interfacial sliding resistance, Y the fiber radius, and rro and m are, respectively, the charac￾teristic strength and the Weibull modulus of the fiber strength distribution. Lo is a constant with dimensions of length associated to go, and /IR is a non-dimensional function of the Weibull modulus of the fibers. This expression shows that the fracture resistance scales with the interfacial sliding resistance according to r(’ -@‘(’ + m). The Weibull modulus of the fibers is usually in the range 336 [26,28] and thus R, decreases as t increases. A good example of this progressive degradation of the interface with temperature was found in the ZrSiO,/SiC material. The C coating of around 3 pm in thickness acted as the “mechanical fuse” which induced crack deflection and relative fiber/matrix sliding at ambient temperature, as is shown in the fracture surface of the specimens tested at ambient temperature (Fig. 5(a)). This coating has disap￾peared, however, in the specimens tested at 1400°C although fiber pull-out was still observed on the fracture surfaces (Fig. 5(b)). The longitudinal sec￾tions of specimens tested at different temperatures showed a progressive substitution of the C coating (Fig. 6(a)) by another layer, which started to grow from the matrix at temperatures above 800°C. This new layer was of similar thickness to the C layer at 1000°C (Fig. 6(b)) and had replaced it completely beyond 1300°C (Fig. 6(c)). The energy spectra of the interface at 20°C and 1400°C obtained by X-ray microanalysis are also shown in Figs 6(a) and (c), and they indicate that the new layer formed during high temperature exposure was mainly made up of SiOz. 60 ZrSiO, / Sic I I, 1 I 1 1 ’ 20% 0 1000% 7 1200%? p 1400% cl I - . v I 1 O- 0 0.1 0.2 0.3 0.4 As/W Fig. 7. Influence of the temperature on the crack growth resistance of a ZrSiOb-matrix composite reinforced with 25 vol.% SCS-6 Sic monofilaments. The gradual changes at the interface were also observed in the fracture resistance curves plotted in Fig. 7. Both the initial toughness and the increase in fracture resistance with crack length decreased as the temperature increased. This progressive degra￾dation seems to be associated with the SiOz layer growth at the interface, which did not prevent fiber/ matrix debonding from taking place but changed the fiber/matrix interaction during sliding. In fact, fibers pulled out from the matrix were observed even in the fracture surfaces of the specimen tested at 1400°C (Fig. 5(b)). It should also be noted that changes in r with temperature may be induced even when no chemical reactions occur at the interface. Residual stresses arise upon cooling from the processing temperature as a result of the thermal expansion mismatch between the matrix (cI,) and the fibers (cq). If cI,--cLf >O, the residual stresses at the interface are compressive at ambient temperature. They are relieved as the temperature increases, reducing the normal stresses at the interface and thus r. The opposite trend is expected when cI,--c(r< 0, and in fact, both behaviors were experimentally confirmed. Celemin et al. [29] estimated from average matrix crack spacing measurements a decrease in r from 24 MPa to 18 MPa in an A1203/SiCr composite from 20 to 12OO”C, while Morscher et al. [30] obtained from fiber push-out experiments an increase in z from 5-18 MPa at ambient temperature to 12-32 MPa at 1300°C in a Si3N4/SiCf composite. 4.2. Fiber degradation Tensile tests of the available oxide and non-oxide fibers have shown that the degradation of their mechanical properties starts above 800°C. Polycrystalline oxide fibers showed evidence of soft-

LlORCA et al.: SiC FIBER-REINFORCED CERAMICS oum touman Fig 8. Fracture mechanisms of the Al,O3/SiCr composite tested at 1200".(a)Fiber/matrix interaction cracks,(b)Fracture surfaces showing fiber pull-out and fiber/ matrix decohesion between the Bn and ening at the grain boundary glassy phase and inter- fiber/matrix interface was observed on the longi granular fracture at this temperature, which were tudinal sections of the specimens tested at 1200C. accompanied by a rapid drop in strength beyond (Fig. 8(a). The fractographic study showed fiber 1000C[31]. The degradation of non-stoichiometric pull-out and indicated that the slippage between the Sic fibers (such as Nicalon) beyond 1000C was matrix and the fibers at elevated temperature took bserved in air and Ar atmospheres [32] and it was place between the BN layer and the Sic outer coat due to the weakening caused by chemical reactions ing(Fig. 8(h). The excellent hehavior of the dual within the thermodynamically unstable fibers and coating interface was finally demonstrated by the from the nucleation and growth of defects at the matrix crack spacing measurements, which indi fiber surface by oxidation (31, 321 cated that the interfacial sliding resistance at The degradation in the fracture resistance of the 1200 C was lower than at ambient temperature Al,O,/Sic composite at 1200.. was attributed to the reasons mentioned at the end of the previous this latter mechanism. The use of a dual BN SiC section [29] ating on the fibers was effective in providing aa weak fiber/ matrix interface and a reduction in weak fiber matrix interface even at elevated tem- the interfacial sliding resistance should lead to an perature, where crack deflection and arrest at the increase in toughness according to the current

LLORCA et al.: SK FIBER-REINFORCED CERAMICS 2449 Fig. S. Fracture mechanisms of the A120$SiCr composite tested at 1200°C. (a) Fiber~matrix interaction showing crack deflection and arrest at the fiber/matrix interface as well as fibers bridging the matrix cracks. (b) Fracture surfaces showing fiber pull-out and fiber/matrix decohesion between the BN and the Sic coatings. ening at the grain boundary glassy phase and inter￾granular fracture at this temperature, which were accompanied by a rapid drop in strength beyond 1000°C 13 11. The degradation of non-stoichiometvic SIC fibers (such as Nicalon) beyond 1000°C was observed in air and Ar atmospheres [32] and it was due to the weakening caused by chemical reactions within the thermodynamically unstable fibers and from the nucleation and growth of defects at the fiber surface by oxidation [31,32]. The degradation in the fracture resistance of the A1203/SiCf composite at 1200°C was attributed to this latter mechanism. The use of a dual BN/SiC coating on the fibers was effective in providing a weak fiber~matrix interface even at elevated tem￾perature, where crack deflection and arrest at the fiber/matrix interface was observed on the longi￾tudinal sections of the specimens tested at 1200°C (Fig. S(a)). The fra~tographic study showed fiber pull-out and indicated that the slippage between the matrix and the fibers at elevated temperature took place between the BN layer and the SIC outer coat￾ing (Fig. 8(b)). The excellent behavior of the dual coating interface was finally demonstrated by the matrix crack spacing measurements, which indi￾cated that the interfacial sliding resistance at 1200°C was lower than at ambient temperature for the reasons mentioned at the end of the previous section [29]. A weak fiber/matrix interface and a reduction in the interfacial sliding resistance should lead to an increase in toughness according to the current

LLORCA et al.: SiC FIBER-REINFORCED CERAMICS ALO./SiC suming that the fibers behaved in a fashion, and that the fiber strength was dictate °c the largest defect in the fiber, the strength of an in dividual fiber can be expressed as K e na F(a/c, a/R) where Kc is the fracture toughness of the fiber which is close to 2 MPa. m [33] and was assumed constant within the temperature range studied. The results of Astiz [34] for a semielliptical surface crack in a circular bar when a c, while the expressions given by Newman and Raju [35] for an embedded elliptical crack were used for a>c. The fracture probability, P, for a fiber subjected to a stress S is 0. 4 plotted in Fig. 1l(b) for the three testing conditions Aa/w The parameter oo can be obtained from this plot as the stress which gives a fracture probability equal 1200 C as a function of the crack length incre- to 63%o, leading to 2.81 GPa and 1.78 GPa for the samples tested, respectively, at 20C and 1200C. In addition, the fiber fracture probability according to micromechanical models. The mechanical test the Weibull statistics is given by with these predictions and the fracture resistance was lower at P=1 Attention was then focussed on the fibers. and sec- and m can be obtained by the least squares fitting tions of the composite were cut far away from the of equation (4) to the experimental results plotted fracture surfaces and polished to be observed in the in Fig. 11(b). The corresponding values of m were canning electron microscope. Two typical micro- 3. 3 and 3.2 for the samples tested, respectively, at graphs are shown in Fig. 10, where the effect of ex- 20C and 1200"C. These results show that the posure at 1200 C on the nucleation and growth of characteristic fiber strength, do, was significantly defects on the fibers is evident. Most of the defects degraded at elevated temperature while the Weibull were nucleated on the fiber surfaces. and their modulus did not change appreciably. According to depth and width were sometimes comparable to the equation (1), the fracture resistance scales with the fiber radius fiber properties according to The size and shape of the fiber defects was stu- auction in the characteristic strength of the fibers died by quantitative microscopy. The defects were was responsible for the 50% drop in fracture resist assimilated to semiellipse and characterized by ance observed between 20C and 1200"C in this their axes a and c, a being the defect depth and 2c composite. It is also worth noting that the results of Simon and Bunsell [32] on isolated fibers and of the semiellipse axes a and c for the longest defect in Singh et al. [26]( who estimated the in situ strength for each temperature. The cumulative probability of of the fracture mirrors)reached similar conclusions a fiber having a defect whose depth was shorter the Weibull modulus did not change substantially than a is plotted in Fig. 11(a). A similar analysis between 20"C and 1300C but the characteristic was carried out on the defect aspect ratio, a/ c, dis- strength was reduced ribution but the results are not presented here for It should finally be noted that the SCS-6 Sic the sake of brevity [29). Forty percent of the fibers monofilaments are reported to exhibit better in the as-received material presented defects shorter strength at elevated temperature than the Nicalon than 0.5 um. This was reduced to 10% after high SiC counterparts [36]. This is due to the quasi-stoi temperature exposure. In addition, the fraction of chiometric composition of the monofilaments pro the fibers exhibiting defects over I um increased duced by chemical vapor deposition which from 20 to 40%after the thermal treatment and improves their oxidation resistance and structural the depth of the longest defect also grew from 2 up stability at high temperature. Their large diameter to 4.6 um. It should be indicated that the average (142 um while the other fibers are in the range 10- 20 um)is, however, a serious practical limitation The results 11(a)can be used to estimate bccause thcy cannot be woven to form bidirectional the fiber

2450 LLORCA et al.: Sic FIBER-REINFORCED CERAMICS 60 - 60 E Xi A&O, / Sic :yq ’ ’ ’ ’ ’ oLoA”’ : 80° .df O0 O0 8 0 00 0 0 0 .’ 0 0.1 0.2 0.3 0.4 M/W Fig. 9. Fracture resistance of the A120,/SiCr composite at 20°C and 1200°C as a function of the crack length incre￾ment, Au (normalized by the specimen width, I+‘). micromechanical models. The mechanical test results were not in agreement, however, with these predictions and the fracture resistance was lower at 1200°C than at ambient temperature (Fig. 9). Attention was then focussed on the fibers, and sec￾tions of the composite were cut far away from the fracture surfaces and polished to be observed in the scanning electron microscope. Two typical micro￾graphs are shown in Fig. 10, where the effect of ex￾posure at 1200°C on the nucleation and growth of defects on the fibers is evident. Most of the defects were nucleated on the fiber surfaces, and their depth and width were sometimes comparable to the fiber radius. The size and shape of the fiber defects was stu￾died by quantitative microscopy. The defects were assimilated to semiellipses and characterized by their axes a and c, a being the defect depth and 2c the surface length. The fiber radius, R, as well as the semiellipse axes a and c for the longest defect in each fiber were measured on more than 100 fibers for each temperature. The cumulative probability of a fiber having a defect whose depth was shorter than a is plotted in Fig. 11(a). A similar analysis was carried out on the defect aspect ratio, a/c, dis￾tribution but the results are not presented here for the sake of brevity [29]. Forty percent of the fibers in the as-received material presented defects shorter than 0.5 pm. This was reduced to 10% after high temperature exposure. In addition, the fraction of the fibers exhibiting defects over 1 pm increased from 20 to 40% after the thermal treatment, and the depth of the longest defect also grew from 2 up to 4.6 pm. It should be indicated that the average fiber radius, R, was 7.2 pm. The results in Fig. 11(a) can be used to estimate the fiber strength at elevated temperature. Assuming that the fibers behaved in a brittle fashion, and that the fiber strength was dictated by the largest defect in the fiber, the strength of an in￾dividual fiber can be expressed as ‘= $kF(a/c f a/R) where Kc is the fracture toughness of the fiber, which is close to 2 MPa.fi [33] and was assumed constant within the temperature range studied. The shape function F was evaluated from the numerical results of Astiz [34] for a semielliptical surface crack in a circular bar when a c. The fracture probability, P, for a fiber subjected to a stress S is plotted in Fig. 1 l(b) for the three testing conditions. The parameter co can be obtained from this plot as the stress which gives a fracture probability equal to 63%, leading to 2.81 GPa and 1.78 GPa for the samples tested, respectively, at 20°C and 1200°C. In addition, the fiber fracture probability according to the Weibull statistics is given by I r ~-imi P=l-exp{-121 1 and m can be obtained by the least squares fitting of equation (4) to the experimental results plotted in Fig. 11(b). The corresponding values of m were 3.3 and 3.2 for the samples tested, respectively, at 20°C and 1200°C. These results show that the characteristic fiber strength, 00, was significantly degraded at elevated temperature while the Weibull modulus did not change appreciably. According to equation (l), the fracture resistance scales with the fiber properties according to c$‘(~ + ‘) and the re￾duction in the characteristic strength of the fibers was responsible for the 50% drop in fracture resist￾ance observed between 20°C and 1200°C in this composite. It is also worth noting that the results of Simon and Bunsell [32] on isolated fibers and of Singh et al. [26] (who estimated the in situ strength of the fibers in Sic/Sic composite through the size of the fracture mirrors) reached similar conclusions: the Weibull modulus did not change substantially between 20°C and 1300°C but the characteristic strength was reduced. It should finally be noted that the SCS-6 Sic monofilaments are reported to exhibit better strength at elevated temperature than the Nicalon Sic counterparts [36]. This is due to the quasi-stoi￾chiometric composition of the monofilaments pro￾duced by chemical vapor deposition which improves their oxidation resistance and structural stability at high temperature. Their large diameter (142 pm while the other fibers are in the range lo- 20 pm) is, however, a serious practical limitation because they cannot be woven to form bidirectional fabrics